Building and decoding ubiquitin chains for mitophagyMitochondria
play a central role in the life of cells and organisms by serving
as a centre for energy gener ation, in the form of ATP via
oxidative phosphoryl ation, and by organizing metabolic
machineries, such as the citric acid cycle, that drive many
cellular functions. The spatial organization of mitochondria is key
to their function, wherein oxidative phosphorylation occurs via the
respir atory chain in the mitochondrial inner mem- brane, and other
metabolic reactions are organized in the mitochondrial
matrix.
Mitochondrial proteins are encoded by the nuclear and mitochondrial
genomes and require the precise orchestration of protein import
through the outer and inner mitochondrial membranes, folding and
assembly into protein complexes to ultimately form a functional
spatially organized organelle. Errors in these processes can result
in damaged mitochondria that are detri- mental to cellular
physiology. For example, defects in respir atory chain function
promote the production of damaging reactive oxygen species and the
loss of mem- brane potential that is crucial for mitochondrial
func- tion. Moreover, defects in protein folding in the matrix
promote the mitochondrial unfolded protein response (mtUPR), which
controls both protein synthesis in the mitochondrial matrix and
production of mitochondrial chaperones1–4. Although homeostatic
mechanisms such as the mtUPR may be sufficient to repair
mitochondria when damage is transient, prolonged or
unrepairable
damage can lead to elimination of mitochondria via a process known
as mitophagy5–7. Mitophagy is a form of selective autophagy, during
which mitochondria are decorated with polyubi quitin chains,
engulfed by autophagosomes and degraded following lysosomal
fusion (FIG. 1).
Mitophagy was first visualized in electron micro- graphs of
cultured cells8, and work over the past two decades has revealed
the fundamental biochemical steps involved in targeting of
mitochondria to the auto- phagy system through both
ubiquitin-dependent and ubiquitin- independent pathways. Our
understanding of ubiquitin-dependent mitophagy has been driven
largely through analysis of two genes that were found to be mutated
in familial forms of Parkinson disease — the
E3 ubiquitin-protein ligase parkin (PRKN)9 and its activat-
ing kinase PTEN-induced putative kinase 1 (PINK1), which is present
on damaged mitochondria10–12. Early studies in Drosophila
melanogaster revealed a genetic pathway in which PINK1 functioned
upstream of parkin and overexpression of parkin could bypass
defects in PINK1 (REFS 13,14). We now know that these proteins
operate in a common pathway to catalyse the assembly of ubiquitin
chains on mitochondrial outer membrane (MOM) proteins13–15. These
ubiquitin chains bind auto- phagic cargo receptors such as
optineurin (OPTN) and sequestosome 1 (SQSTM1, also known as
p62), which act in concert with the general
autophagy machinery to capture
Department of Cell Biology, Harvard Medical School, Boston,
Massachusetts, USA.
Correspondence to J.W.H. wade_harper@ hms.harvard.edu
doi:10.1038/nrm.2017.129 Published online 23 Jan 2018
Ubiquitin A 76-amino-acid protein that can be covalently conjugated
to lysine residues in other proteins to specify several protein
fates. Poly-ubiquitin chains can be generated using seven internal
lysine residues in ubiquitin or its first methionine. Lys11-linked
or Lys48-linked chains usually target proteins for degradation,
whereas other chains, such as Lys63-linked or Met1-linked chains,
have signalling roles.
Building and decoding ubiquitin chains for mitophagy J. Wade
Harper, Alban Ordureau and JinMi Heo
Abstract | Mitochondria produce energy in the form of ATP via
oxidative phosphorylation. As defects in oxidative
phosphorylation can generate harmful reactive oxygen species, it is
important that damaged mitochondria are efficiently removed via a
selective form of autophagy known as mitophagy. Owing to a
combination of cell biological, structural and proteomic
approaches, we are beginning to understand the mechanisms by which
ubiquitin-dependent signals mark damaged mitochondria for
mitophagy. This Review discusses the biochemical steps and
regulatory mechanisms that promote the conjugation of ubiquitin to
damaged mitochondria via the PTEN-induced putative kinase 1 (PINK1)
and the E3 ubiquitin-protein ligase parkin and how ubiquitin chains
promote autophagosomal capture. Recently discovered roles for
parkin and PINK1 in the suppression of mitochondrial antigen
presentation provide alternative models for how this pathway
promotes the survival of neurons. A deeper understanding of these
processes has major implications for neurodegenerative diseases,
including Parkinson disease, where defects in mitophagy and other
forms of selective autophagy are prominent.
R E V I E W S
NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 19 | FEBRUARY 2018 |
93
M i to c h o n d r i a
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
mailto:
[email protected]
mailto:
[email protected]
http://dx.doi.org/10.1038/nrm.2017.129
Parkinson disease A long-term disease of the central nervous system
that primarily affects motor functions as a result of loss
of dopaminergic neurons.
E3 ubiquitin-protein ligase A protein or protein complex that can
facilitate the transfer of ubiquitin from an E2 conjugating enzyme
to a substrate.
General autophagy machinery Composed of protein and lipid kinases
that coordinate the formation of autophagic membranes and the ATG8
conjugation machinery, which is involved in maturation of
autophagosomal membranes and fusion with lysosomes.
Amyotrophic lateral sclerosis (ALS). A progressive and fatal motor
neuron disorder that affects the function of voluntary muscles,
leading to an inability to move, swallow, speak and breathe.
Translocase of the outer membrane (TOM). The TOM complex is a
multi-protein channel that functions to facilitate import
of nuclear-encoded but mitochondrial-localized proteins into
all intra- mitochondrial compartments. The only proteins that do
not pass through the TOM complex during import are single-pass
mitochondrial outer membrane proteins.
N-end rule ubiquitin ligase A subfamily of RING E3 ubiquitin
ligases, including UBR1, UBR2 and UBR3, that use their N-terminal
UBR domain to bind to substrates containing hydrophobic or arginine
residues at their N-terminus.
damaged mitochondria in the autophagosomal double- membrane
(FIG. 1). Fusion with lyso somes facilitates degradation of
mitochondria via lysosomal hydrolases.
OPTN and SQSTM1, as well as their associated
serine/threonine-protein kinase TBK1, have recently been
genetically linked to sporadic and familial forms of amyotrophic
lateral sclerosis (ALS)16–19. This genetic link has two major
implications: first, it suggests that selective forms of autophagy
underlie particular forms of ALS, and second, it suggests that
understanding the biochemical mechanisms involved in marking auto-
phagic cargo, including mitochondria, with ubiquitin and its
capture by cargo receptors could pave the way for new approaches
for therapeutic intervention in neuro- degenerative diseases. In
addition to these implications in disease, recent studies have
shown that some forms of parkin-mediated mitophagy that remove
apparently healthy mitochondria are important for several physio-
logical processes, such as self-renewal of angiopoietin 1
receptor (TIE2)-positive haematopoietic stem cells, and paternal
mitochondrial clearance during fertiliza- tion20–25 (BOX 1).
Moreover, parkin has been implicated in the xenophagic removal of
Mycobacterium tubercu- losis, which may have parallels with
mitophagy, but the underlying mechanisms are poorly
understood26.
In this Review, we focus on the biochemical mech- anisms that drive
the assembly of ubiquitin chains on damaged mitochondria and on how
ubiquitylation is decoded by the autophagy machinery to capture
damaged organelles. Given that mitochondria are con- stantly
engaged in fusion–fission cycles that may mix healthy and damaged
organelles, it is important that mitophagy pathways are rapid and
robust in their ability to selectively mark and degrade only
damaged organ- elles. This selectivity is achieved through the use
of two distinct and sequential feedforward loops, driven by the
kinase PINK1 and ubiquitin on the surface of mitochondria that are
predicted to create a switch-like behaviour for the detection and
capture of only damaged organelles. We describe a generalizable
framework for ubiquitin- dependent forms of mitophagy and discuss
the substantial gaps that exist in our understanding of these
crucial pathways.
Overview of the PINK1–parkin pathway It is thought that the
assembly of ubiquitin chains on mitochondria is necessary for the
recruitment of this autophagy machinery and the removal of damaged
mitochondria by mitophagy6,7. In its simplest form, the ubiquitin
chain assembly pathway can be described as containing three
positively acting elements: a mito- chondrial damage sensor
(PINK1), a signal amplifier (parkin) and a signal effector
(ubiquitin chains), which determines which mitochondria should be
captured by the autophagy machinery (FIG. 1).
PINK1 contains an N-terminal mitochondrial target- ing sequence and
binds to the translocase of the outer membrane (TOM) complex10,27.
When the mito chondrial targeting sequence and transmembrane
segment of PINK1 reach the translocase of the inner membrane (TIM)
complex and are laterally transferred to the inner
membrane, the transmembrane segment is proteo- lytically
cleaved by the inner membrane- localized protease PARL
(presenilins-associated rhomboid-like protein, mitochondrial).
Cleavage results in a 52 kDa protein fragment containing the kinase
domain that is probably still associated with the
TOM complex28–30.
When mitochondria are healthy, this 52 kDa PINK1 fragment is
released into the cytosol and is rapidly ubi- quitylated by an
N-end rule ubiquitin ligase, which targets it for degradation by
the proteasome30. Thus, PINK1 levels are low in cells with healthy
mitochondria. However, when mitochondria are damaged, PINK1
translocation and processing is blocked, leading to the
accumulation of active PINK1 on the MOM31, where it can activate
the E3 ubiquitin ligase activity of parkin via a multistep feed
forward mechanism, as detailed below. Parkin-dependent ubiquitin
chain assembly on the MOM then promotes recruitment of ubiquitin-
binding mitophagy receptors to promote capture by the
autophagosome32–35 (FIG. 1).
As with most signal transduction pathways, neg ative regulators
help to ensure that damage signals are suffi- ciently strong such
that healthy mitochondria are not inappropriately degraded. In
this mitophagy path- way, deubiquitinating enzymes (DUBs),
including the mitochondrially localized ubiquitin carboxyl-terminal
hydrolase 30 (USP30), seem to antagonize the pathway by removing
ubiquitin chains from mitochondria until parkin activation is
sufficient to outpace ubiquitin chain removal by USP30
(REF. 36). Elucidating the biochemical mechanisms that
determine the activation state of parkin has been a major focus of
this research field.
Mechanism of parkin activation by PINK1 In cells with healthy
mitochondria, parkin is localized diffusely in the cytoplasm in an
autoinhibited form37–40. However, following mitochondrial damage
and stabiliza- tion of PINK1 on the MOM, parkin can undergo a
series of modifications (including phosphorylation, multiple
conformational changes and association with Ser65-phosphorylated
ubiquitin (pSer65-Ub)) initiated by PINK1 that promote its stable
association with the MOM and activation of its E3 ubiquitin ligase
activity. For simplicity, the two major steps of parkin activation
— direct activation by phosphorylation of the parkin ubiquitin-like
(UBL) domain and activation by bind- ing to pSer65-Ub — are
described separately, followed by a description of how these events
work together on the MOM to generate a feedforward ubiquitylation
process that promotes mitophagy. It should be noted that in many
studies described below, the experi mental approach has been to
overexpress parkin or PINK1 in model cell systems, which, given the
involvement of feedback loops, could have unknown effects, and the
results could be more difficult to interpret.
Intrinsic parkin activation by PINK1dependent UBL phosphorylation.
The ubiquitin system employs a series of ubiquitin charging and
transfer events culminat- ing in the transfer of the C-terminal
Gly76 residue of ubiquitin to a lysine residue on the substrate
(primary
R E V I E W S
94 | FEBRUARY 2018 | VOLUME 19 www.nature.com/nrm
R E V I E W S
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
Isopeptide bond An amide bond formed between the amino group of a
lysine side chain on a protein (substrate) and the C-terminus of
another protein (ubiquitin).
Figure 1 | Overview of parkin-dependent mitophagy. In cells with
healthy mitochondria, the PTEN-induced putative kinase 1 (PINK1) is
rapidly degraded, and the E3 ubiquitin-protein ligase parkin is in
an autoinhibited form in the cytoplasm. Upon mitochondrial damage,
PINK1 is stabilized on the mitochondrial outer membrane (MOM) and
can activate parkin through a feedforward mechanism involving
parkin and ubiquitin phosphorylation. Parkin then assembles
ubiquitin chains on numerous MOM proteins, which can recruit
ubiquitin-binding autophagy receptors. In the canonical model
of autophagy, ubiquitin-binding autophagy receptors function to
recruit the ATG8-positive phagophore, which ultimately encases the
damaged mitochondria and allows fusion with lysosomes, thereby
promoting degradation of damaged mitochondria. The canonical
autophagosome assembly pathway is composed of three major
arms136–138: the phosphatidylinositol 3-kinase catalytic subunit
type 3 (VPS34) arm responsible for production of
phosphatidylinositol-3-phosphate (PtdIns3P) on donor membranes
(step 1), the serine/threonine- protein kinase ULK1 arm that
regulates phagophore initiation and expansion (step 2) and the ATG8
conjugation pathway involving ATG7 (E1), ATG3 (E2) and the
ATG5/ATG12–ATG16 (E3) complex (where ‘/’ indicates an isopeptide
bond). The conjugation pathway attaches ATG8 proteins to
phosphatidylethanolamine (PE) on the growing autophagosomal
membrane (step 3). ATG8 proteins are thought to function by
interacting with cargo receptors and other regulators
of the pathway. However, there is evidence of
ATG8-conjugation-independent forms of autophagosome formation
through non-canonical pathways101,102, which may also function in
mitophagy100, possibly with reduced efficiency. AMBRA1,
activating molecule in BECN1-regulated autophagy protein 1;
AMPK, 5-AMP-activated protein kinase; BECN1, beclin 1; DFCP1,
zinc-finger FYVE domain-containing protein 1; ER, endoplasmic
reticulum; FIP200, RB1-inducible coiled-coil protein 1; TBK1,
serine/threonine-protein kinase TBK1; ULK2,
serine/threonine-protein kinase ULK2; VPS15, phosphoinositide
3-kinase regulatory subunit 4; WIPI2, WD repeat domain
phosphoinositide-interacting protein 2.
Nature Reviews | Molecular Cell Biology
ER
Omegasome
ATG4 ATG7 E1
NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 19 | FEBRUARY 2018 |
95
M i to c h o n d r i a
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
ubiquitylation) or to lysine residues on ubiquitin itself to extend
a ubiquitin chain. Ubiquitin is activated by an E1
ubiquitin-activating enzyme in the presence of ATP and is
transferred to the active site cysteine residue in one of several
E2 ubiquitin-conjugating enzymes to form a thioester bond. Such
E2 enzymes interact with RING finger or HECT domains in E3
ubiquitin ligases to facilitate the transfer of ubiquitin to the
substrate. Parkin is a RING-between-RING (RBR) E3 ubiquitin ligase;
it contains an N-terminal UBL domain, a central RING1 domain that
binds an E2 enzyme, an in-between RING (IBR) domain and a
C-terminal RING2 domain containing the catalytic cysteine residue
(FIG. 2a).
RBR E3s are analogous to HECT domain-containing proteins41 and
use their RING1 domain to catalyse the transfer of ubiquitin from a
charged E2 enzyme to the catalytic cysteine residue in
RING2 (REF. 42). As with HECT E3s, this ubiquitin thioester is
sub- sequently discharged onto a substrate lysine residue
(FIG. 2a, left panel). Three mechanisms are involved in parkin
auto inhibition and must be overcome to convert parkin into an
active enzyme37–40,43 (FIG. 2a, right panel). First, the UBL
domain of parkin rests against one of two core α-helical elements
within RING1, block- ing E2 access (FIG. 2b). Second, the
repressive (REP) element docks with both the UBL domain and
RING1,
Box 1 | Mitophagy during development and via parkin-independent
mechanisms
There is an increasing appreciation that large alterations in the
abundance of mitochondria are necessary during particular cellular
transitions that occur during development and in normal physiology,
in addition to the removal of defective mitochondria (see the
figure).
Loss of paternal mitochondrial DNA during fertilization occurs
through the process of mitophagy in Caenorhabditis elegans,
Drosophila melanogaster and vertebrates89,121–124. In mice, the E3
ubiquitin-protein ligase parkin and the mitochondrial ubiquitin
ligase activator of NFKB 1 (MUL1) function redundantly and in
combination with sequestosome 1 (SQSTM1) and PTEN-induced putative
kinase 1 (PINK1), indicating that some elements of the established
PINK1–parkin pathway are used in this case89. By contrast, the
analogous process in D. melanogaster requires autophagy
machinery and SQSTM1 but not the parkin orthologue121.
For removal of mitochondria during reticulocyte maturation,
autophagosome assembly for mitochondrial capture is orchestrated by
an autophagy receptor called BCL2/adenovirus E1B 19 kDa
protein-interacting protein 3-like (NIX; also known as BNIP3L),
which is located on the mitochondrial outer membrane (MOM),
contains LC3-interacting region (LIR) motifs used to associated
with ATG8 proteins and is required for efficient mitochondrial
clearance20,125. Recent data suggest that phosphorylation of NIX
near the LIR motif increases binding to ATG8 and increases
autophagosomal recruitment to mitochondria126. Additional MOM
proteins containing LIR motifs, including the peptidyl-prolyl
cis-trans isomerase FKBP8 and FUN14 domain-containing protein 1
(FUNDC1), have been implicated in direct interactions with ATG8
proteins to promote mitophagy127,128. In the absence of parkin,
overexpression of FKBP8 and the ATG8 protein LC3A can promote
mitophagy, although a physiological setting for this form of
mitophagy is unknown. Similarly, overexpression of FUNDC1 can
promote mitophagy in response to hypoxia128 in a manner that is
regulated by the E3 ubiquitin-protein ligase MARCH5
(REF. 129). Understanding the physiological circumstances used
to promote direct mitophagy by ATG8 recruitment remains a goal for
future research.
CALCOCO2, calcium-binding and coiled-coil domain-containing protein
2; mtUPR, mitochondrial unfolded protein response; OPTN,
optineurin; ROS, reactive oxygen species. Nature Reviews |
Molecular Cell Biology
Mitophagy
Autolysosome
NIX
96 | FEBRUARY 2018 | VOLUME 19 www.nature.com/nrm
R E V I E W S
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
Nanobody A type of single-chain antibody frequently used to
stabilize weak interactions for structural biology.
further blocking E2 access (FIG. 2b). Third, the catalytic
Cys431 residue in RING2 is shielded by the unique parkin domain
(UPD; also known as RING0), located between the UBL domain and
RING1, thereby blocking transfer of ubiquitin from the charged E2
to the cata- lytic cysteine residue of parkin (FIG. 2b). Thus,
multiple conformational changes are necessary to remove these
auto-inhibitory constraints.
Since the discovery that PINK1 functions upstream of parkin14,27,
substantial effort has been focused on understanding how PINK1
directly regulates parkin activity. An early study found that PINK1
phosphoryl- ates parkin in human SH-SY5Y cells and that PINK1-
dependent phosphorylation promoted the ability of parkin to make
Lys63-linked ubiquitin chains44. However, the phosphorylation sites
in parkin and the activation mechanism were not reported. Key
insights into parkin activation came from the finding that PINK1
phosphorylates Ser65 in the UBL domain of parkin45,46 and that
phosphorylation of parkin at Ser65 (pSer65-parkin) promotes
ubiquitin chain assembly by parkin in vitro45. Subsequent
studies using parkin stoichio metrically phosphorylated on Ser65 in
con- junction with quantitative measurements of ubiquitin chain
formation by mass spectrometry revealed that pSer65-parkin
increases the chain assembly activity in vitro by
~2,400-fold47. Thus, phosphorylation of parkin on its UBL domain
dramatically activates its intrinsic ubiquitin ligase activity. UBL
domain phos- phorylation at Ser65 disrupts the UBL–RING1 inter-
action and is thought to allow movement of the UBL domain through
its tether to the UPD43,48,49 (FIG. 2b–d). Interestingly,
release of the UBL domain is mimicked by a mutation in parkin
(Trp403Ala) that is thought to partially release the REP element
from RING1 to facilitate binding of the E2 (REFS 39,43)
(FIG. 2b–d). This model is consistent with molecular dynamics
simu- lations, which suggested that UBL phosphorylation releases
the UBL domain from its close interaction with RING1 and leads to
small changes in the con- formation of the REP element, providing
access for the charged E2 to RING1 (REF. 50). Release of the
UBL domain from the parkin core provides access to the hydrophobic
Ile44 residue in the UBL domain known to be important for
phosphoryl ation by PINK1 (REF. 48), which explains in part
the need for UBL domain release before phosphorylation. Subsequent
structural alter- ations are propagated to the RING2 domain, as
indi- cated by the increased reactivity of the catalytic Cys431
residue (BOX 2).
Ubiquitin phosphorylation by PINK1 promotes parkin activation and
mitochondrial retention. A second mode of PINK1 action on damaged
mitochondria was identified when it was discovered that ubiquitin
and ubiquitin chains are PINK1 substrates47,51–54 and that
pSer65-Ub chains are linked to parkin activation and retention on
mitochondria47,55. The UBL domain of parkin is ~30% identical (~50%
similar) in amino acid sequence to ubiquitin, including substantial
con- servation of the surface containing Ile44, which is near
Ser65 (FIG. 2e,f). Importantly, Ser65 in the parkin UBL domain
was found to be conserved in ubi quitin, which led to the
demonstration that PINK1 can directly phos- phorylate ubiquitin on
Ser65 (REFS 51,52). Quantitative phosphoproteomics of
mitochondria during mitophagy induction47,53, unbiased
identification of candidate PINK1 substrates47 and biochemical
activation studies of parkin by PINK1 in the presence of
ubiquitin54 also independently identified ubiquitin as a PINK1
target. Initial biochemical studies indicated that monomeric
pSer65-Ub not only physically associates with parkin but can also
partially activate its ubi quitin chain assem- bly activity
independently of Ser65 phosphorylation on parkin47,51–53. Although
it seems that in vivo parkin associates with pSer65-Ub in the
form of ubiquitin chains on mitochondria (discussed below),
monomeric pSer65-Ub has served as a useful tool for understand- ing
the biochemical and structural basis of parkin activ- ation and
retention on the MOM. pSer65-Ub can bind to unphosphorylated parkin
and pSer65-parkin in vitro, but binding to pSer65- parkin
is ~20-fold stronger47,48,56. Moreover, while stoichiometric
binding of pSer65-Ub to unphosphoryl ated parkin activates chain
synthe- sis by ~1,000-fold, a complex of pSer65-parkin and
pSer65-Ub displays ~4,400-fold higher chain synthesis activity than
unphosphorylated parkin47,57.
Recently, insights into how PINK1 specifically recog- nizes
ubiquitin were obtained through structural analy sis of a
PINK1–ubiquitin complex stabilized via a nanobody. PINK1 is unique
among protein kinases in that its N-terminal lobe contains three
stretches of amino acid sequences, referred to as insertions, that
are absent from other protein kinases58. The structure reveals that
these insertions are stabilized by auto- phosphorylation of Ser202
and Ser204 in PINK1, creating a unique conformation of insertion 3
that enables the recogni- tion of ubiquitin as a substrate.
In addition, ubiquitin is in a unique
C-terminally retracted conformation ( ubiquitin-CR) when
bound to PINK1, which places its Ser65 residue in an extended
loop near the catalytic centre of PINK1. The ubiquitin- CR
conformation was thought to be unique to pSer65-Ub54, but it was
recently reported that it is found at low abundance in unmodi- fied
ubiquitin and is in rapid exchange with the con- ventional
conformation of ubiquitin59. The structure of PINK1 lacking
insertion 3 was also recently described60. These studies explain
key features of PINK1, includ- ing why PINK1 is highly selective
for ubiquitin and the parkin UBL domain as substrates and how many
of the PINK1 mutations identified in patients with Parkinson
disease disrupt either catalytic activity or
substrate recognition58.
Structural and functional studies have shown that, by enabling
important contacts within parkin, pSer65-Ub promotes the formation
of a central α-helix (H3) linking RING1 and the IBR domain
(FIG. 2a,d), and thus release of the UBL domain from the
parkin core48,49,56,61,62 (FIG. 2d). The linchpin in the
parkin– pSer65-Ub co-complex is a cluster of positively charged
residues in parkin (Lys151 and His302) that bind to the phosphate
in pSer65-Ub, the mutation of which
R E V I E W S
NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 19 | FEBRUARY 2018 |
97
M i to c h o n d r i a
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
4651
Ser65
PINK1-mediated UBL domain and ubiquitin phosphorylation
? (pUBL position unknown)
Phe146
Trp403
Asn273
e
Ser65 phosphorylation
0º 180º
98 | FEBRUARY 2018 | VOLUME 19 www.nature.com/nrm
R E V I E W S
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
diGLY capture proteomics In this approach, di-Gly-Gly ubiquitin
‘remnants’ that remain on substrate lysine residues after
trypsinization are captured using a specific antibody and
identified using mass spectrometry.
ε-Amino group Refers to the NH3
+ group in a lysine side chain, which is often used as a recipient
for ubiquitin transfer in proteins.
abolished binding to pSer65-Ub and in vitro parkin activation
by pSer65-Ub43,48,63. Crystallographic analysis of pSer65-Ub bound
to a form of parkin that is missing 59 residues of the UBL domain
linker (FIG. 2a) led to an alternative model for activation
involving a parkin dimer, wherein pSer65-Ub binding opens a surface
on the IBR domain that then interacts with the donor ubiquitin of a
charged ubiquitin~E2 thioester (where ~ indicates the
thioester bond), with the E2 itself being bound to RING1 of the
neighbouring parkin mol- ecule64. This model is supported by the
finding that mutations in IBR residues that are predicted to
directly interact with the donor ubiquitin reduce chain synthe- sis
by parkin in vitro, although these mutants were not tested
in vivo. Biophysical analysis of active, full-length
pSer65-parkin in complex with pSer65-Ub revealed a monomeric 1:1
complex47,56. While it is conceivable that during the catalytic
process, a pSer65-parkin dimer assembles transiently, the UBL
domain in pSer65- parkin seems to be unable to promote dimer
formation at least in the absence of charged UBCH7 at concentra-
tions used for biophysical measurements. Biophysical and molecular
dynamics measurements have also sug- gested that binding of parkin
lacking its UBL domain to the ubiquitin~E2 triggers large-scale
diffusional motion of the RING2 domain towards RING1, thereby
facilitating ubiquitin transfer to Cys431 located in RING2
(REF. 65). Analysis of full-length pSer65-parkin–
pSer65-Ub–ubiquitin~E2–substrate complexes in both the pre-transfer
state and a state containing ubiquitin- charged RING2 is required
to confirm which of these models is correct. It is also noteworthy
that activ- ation mechanisms involving allosteric UBL domains have
been described for other RBR E3s64,66, suggest- ing conserved
elements of activation mechanisms for RBR E3s.
Mitochondrial surface ubiquitylation Within minutes of
mitochondrial damage, parkin is recruited and activated on the MOM
and initiates ubi- quityl ation of local substrates10,46,67–70.
Much effort has been focused on the identification of parkin
substrates in response to mitochondrial damage. Early studies
identified several targets of diverse function, including mitofusin
1 (MFN1), MFN2, voltage-dependent anion- selective channel (VDAC)
proteins, mitochondrial fission 1 protein (FIS1), mitochondrial
import receptor subunit TOM20 homologue (TOMM20) and CDGSH
iron-sulfur domain-containing protein 1 (CISD1), which are all
located on the MOM70–72. To identify parkin targets and primary
ubiquitylation sites, quanti- tative diGLY capture proteomics
following mitochondrial depolarization was performed73–75.
Ubiquitylation sites were found on the cytoplasmic domain of
numerous MOM proteins and on several cytoplasmic proteins that are
recruited to mitochondria in response to parkin activation (see
below).
Ubiquitylation on mitochondria in HeLa cells over- expressing
parkin occurs in two phases: an initial phase (in the first two
hours) during which the primary substrates are cytosolic domains of
MOM proteins, followed by a second phase during which a cohort of
proteins that were localized inside mitochondria become targeted
for ubiquitylation75. Mitochondrial depolarization in the presence
of overexpressed parkin can lead to rupture of the MOM, thereby
potentially exposing inner membrane proteins to the action of
parkin or other E3s70,76. However, it is unclear whether MOM
rupture occurs at endogenous parkin levels in neurons and whether
it has a specific role in mitophagy.
The diversity of parkin substrates on the MOM and the absence of an
obvious substrate recognition ele- ment within parkin suggest that
parkin lacks inherent substrate specificity73. Thus, it is possible
that the iden- tity of substrates on the MOM is less important than
the density of ubiquitin chains that are assembled on these
substrates for specifying mitophagy. Indeed, as described below,
mitophagy receptors have the abil- ity to bind to particular types
of ubiquitin chains on mitochondria. Ubiquitin chains can be
assembled through the ε-amino group in each of the seven lysine
residues (Lys6, Lys11, Lys27, Lys29, Lys33, Lys48 and Lys63) on
ubiquitin as well as through the α-amino group of its N-terminal
methionine residue, and the chains can be either linear, branched
or mixed. The different types of chain linkages can be
distinguished by specific ubiquitin binding domains (UBDs), such as
those found in mitophagy receptors77. We now know that in cells
overexpressing parkin in the context of endogenous ubiquitin,
mitochondrial depolarization leads to the formation of Lys6, Lys11,
Lys48 and Lys63 chain linkages on the MOM, and parkin can catalyse
the formation of these same linkages in vitro, but it is not
known how such chains are distributed across dif- ferent
mitochondrial substrates, how long chains built on each type of
substrate are or the extent to which there might be chains with
mixed or branched chain topologies47,78. While poly-ubiquitylation,
as opposed to
NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 19 | FEBRUARY 2018 |
99
M i to c h o n d r i a
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
mono-ubiquitylation, is presumed to be crucial for the recruitment
of mitophagy receptors, it is also possible that
mono-ubiquitylation plays an important role by serving as a target
for phosphorylation by PINK1.
A model for parkin activation on mitochondria The mechanisms of
ubiquitin and parkin phosphoryl- ation described above have led to
a model whereby mitochondrial damage promotes rapid ubiquitin chain
polymerization as a result of two mechanisms acting in parallel
that, together, generate a positive feedback
loop43,47,51–53,55,57,63 (FIG. 3).
On one hand, the accumulation of pSer65-Ub on mitochondria as a
result of PINK1 phosphorylating pre-existing ubiquitin molecules,
or chains that are built by parkin, promotes the recruitment of
cytosolic unphosphorylated parkin through direct inter action with
pSer65-Ub48,55,63 (FIG. 3Ba). As many as 20% of ubi- qui tin
molecules on damaged mitochondria in HeLa cells are phosphorylated
in a PINK1-dependent manner upon mitochondrial depolarization in
the presence of catalytically active parkin47,57. The
parkin–pSer65-Ub interaction has two major consequences:
it partially activates the ubiquitin ligase activity of parkin
by ~1,000-fold47,51–53, thereby contributing to ubiquitin chain
assembly on the MOM57; and it greatly increases the rate at which
PINK1 phosphorylates the parkin UBL domain, as was shown
in vitro48. Similar results were obtained using a fluorescent
ubiquitin probe, which also revealed that addition of
mitochondrial
Rho GTPase (MIRO) as a parkin substrate can further increase rates
of ubiquitin transfer and chain poly- meriza tion79. The
binding of pSer65-Ub to pSer65- parkin is ~20-fold stronger than to
unphosphoryl ated par kin, thus favouring the retention of fully
active pSer65- parkin on damaged MOMs45,47,48,55,63. Moreover,
because pSer65-parkin bound to pSer65-Ub is opti- mally active
(~4,400-fold activation)57, its retention on the MOM promotes
further ubiquitin chain assem- bly and provides additional
ubiquitin molecules for phosphorylation by PINK1, creating the
feedforward mechanism (FIG. 3).
On the other hand, parkin can be directly phos- phorylated and
activated by PINK1 on the MOM (inde- pendently of its initial
encounter with pSer65-Ub) to locally generate ubiquitin chains
(FIG. 3Bb) that become substrates for PINK1 to recruit
more pSer65-parkin to the MOM, thereby serving as an initial
amplification step47. The importance of ubiquitin chain phosphoryl-
ation in the feedforward process is highlighted by the observation
that cells expressing a mutant form of ubiquitin
(ubiquitin-Ser65Ala) that cannot be phos- phorylated display
decreased ubiquitin chain synthesis on MOM proteins, dramatically
reduced recruitment of parkin to the MOM and reduced rates of
mitophagy57. Ubiquitin conjugation is further reduced when parkin
is mutated in its PINK1 phosphorylation site (parkin- Ser65Ala),
which is consistent with the finding that the most active form of
parkin is phosphorylated on its UBL domain and bound to
pSer65-Ub57. The impor- tance of ubiquitin phosphorylation for
parkin recruit- ment was also shown by the ability of overexpressed
Ser65 phosphomimetic linear tetrameric ubiquitin chains to promote
parkin recruitment in the absence of
PINK1 (REF. 55).
The relative importance of the two mechanisms that contribute to
the feedforward loop is unclear. Parkin mutants that cannot bind
pSer65-Ub failed to be recruited to mitochondria and promote
ubiquitin chain assembly despite retaining catalytic activity when
activ ated by UBL domain phosphorylation43,48,63. This finding
suggests that recruitment and full parkin activ- ation requires
phosphorylation of pre- existing ubi- quitin on mitochondria by
PINK1 (FIG. 3). However, the catalytically defective
parkin-Cys431Ser mutant is not detectably recruited to depolarized
mito chon- dria39,47,80,81, suggesting that ubiquitin chain synthe-
sis by parkin is necessary for sufficient pSer65-Ub
to accumulate on mitochondria and recruit parkin to detectable
levels, at least in the HeLa cell model sys- tem that was used.
While it is clear that binding of parkin to pSer65-Ub accelerates
PINK1-dependent phosphorylation of its UBL domain48,63, a parkin-
His302Ala mutant that binds ~270-fold more weakly to pSer65-Ub can
still be phosphorylated on its UBL domain in cells upon
mitochondrial depolarization63,82. Moreover, several parkin mutants
that were isolated from patients with Parkinson disease and are not
stably recruited to depolarized mitochondria can neverthe- less be
phosphorylated by PINK1 to the same extent as wild-type parkin,
indicating that PINK1-dependent
Box 2 | Parkin Cys431 reactivity as a tool for monitoring
activation status
The E2 conjugating enzyme UBCH7 functions together with the RING1
domain of the E3 ubiquitin-protein ligase parkin to discharge
ubiquitin onto the catalytic Cys431 residue. As such, the ability
of Cys431 to form a thioester bond with ubiquitin, as well as
formation of the more stable oxyester in the context of the
parkin-Cys431Ser mutant, has been a useful tool for monitoring
conformational changes that alleviate parkin autoinhibition.
Indeed, phosphorylation of parkin on Ser65 is sufficient to allow
discharge of ubiquitin from UBCH7~ubiquitin (where ~ indicates a
thioester bond) to the catalytic residue and greatly
increases modification of Cys431 in parkin in vitro by
ubiquitin-vinyl sulfone, a reactive catalytic site probe47,62,81.
Thus, parkin phosphorylation seems to render RING1 more accessible
to charged E2s and the catalytic cysteine in RING2 more accessible
to activated forms of ubiquitin. Interestingly, binding of
pSer65-Ub to unphosphorylated parkin does not promote reactivity of
Cys431 towards ubiquitin-vinyl sulfone in vitro, suggesting
that this interaction alone is not sufficient to fully release
autoinhibition of the RING2 domain and Cys431 (REFS 47,82).
Similarly, parkin phosphorylation and binding to pSer65-Ub is
required to facilitate optimal reactivity towards an activity-based
ubiquitin~E2 transthiolation probe, and pSer65-Ub alone is only
weakly supportive of reactivity with the ubiquitin~E2 probe82,
consistent with the idea that pSer65-Ub binding is only partially
able to activate unphosphorylated parkin57. Parkin-Cys431Ser has
also been used to scan for defects in activation in cells in the
context of a large set of parkin mutations in patients with
Parkinson disease, revealing defects in activation for most patient
mutations upon mitochondrial damage86. These defects largely
correlate with the efficiency of recruitment of the parkin mutant
to damaged mitochondria86. Importantly, phospho-mimetic
mutants such as parkin-Ser65Glu and ubiquitin-Ser65Glu are poor
mimics of activation54,57. While genetic studies in Drosophila
melanogaster have suggested that parkin-Ser65Glu can rescue mutant
mitochondrial phenotypes of PTEN-induced putative kinase 1
(PINK1)83, the underlying mechanisms are unclear at present.
Moreover, ubiquitin-Ser65Glu is a poor mimic of pSer65-Ub in terms
of parkin activation in vitro54,57, despite the fact that
these mutants are extensively used in overexpression experiments to
examine the role of phospho-ubiquitin.
R E V I E W S
100 | FEBRUARY 2018 | VOLUME 19 www.nature.com/nrm
R E V I E W S
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
Piecemeal mitophagy A process through which subdomains of
mitochondria harbouring misfolded matrix proteins are separated
from the areas of mitochondria that are healthy before engulfment
by autophagy.
phosphorylation can occur without stable association with
mitochondria46,47,63. Considering all these data, the simplest
model is that very small amounts of pSer65-Ub — generated either
from ubiquitin already present on mitochondria before damage or
from ubi- quitin chains synthesized locally by pSer65-parkin —
are necessary to initiate the feedforward mech anism but are not
sufficient for full parkin activation. Given the mechanistic basis
of the feedforward process, parkin or PINK1 overexpression could
contribute to artificial activation of the pathway. This
appears to be the case in experiments using parkin-Ser65Ala,
as independent studies reported different levels of mitochondrial
recruitment, possibly reflecting dif- ferences in expression
levels43,50,52,53,57,83. Since parkin- Ser65Ala can bind to and be
activated by pSer65-Ub in vitro47,51, high levels of
parkin-Ser65Ala expression together with low levels of pSer65-Ub
could promote a feedforward reaction artificially. Given the funda-
mental role of mitochondrial ubiquitin chains in pro- moting
PINK1-dependent parkin activation, it is not
surprising that ubiquitin chain disassembly via DUBs could reduce
available mitochondrial ubiquitin for initiation of the feedforward
process36 (BOX 3). Recent studies have identified a role for
the E3 ubiquitin- protein ligase HUWE1 in making Lys6 ubiquitin
chains on mitochondria, and these types of chains on TOMM20 are
removed by the mitochondrial DUB USP30 (REF. 84) (BOX 3).
Thus, HUWE1 could control the level of basal ubiquitin on
mitochondria that may participate in parkin
activation (FIG. 3Ba).
A key question going forward concerns how mito- phagy is controlled
spatially. Insights into this ques- tion have come from a recent
study85 that revealed ‘sub domains’ of mitochondria harbouring
misfolded matrix proteins initiate piecemeal mitophagy via PINK1
activ ation and parkin recruitment to the damaged sub- domain.
Interestingly, blocking mitochondrial fission in this context
increased mitophagy while decreasing selectivity for damaged
‘domains’, suggesting a new model whereby fission protects healthy
mitochondrial regions from unchecked PINK1–parkin activity85.
Figure 3 | Feedforward mechanism of parkin activation in response
to mitochondrial depolarization. In healthy mitochondria,
PTEN-induced putative kinase 1 (PINK1) is imported and rapidly
degraded in a presenilins-associated rhomboid-like protein,
mitochondrial (PARL)-dependent manner. When mitochondria are
depolarized, PINK1 is stabilized on the translocase of the outer
membrane (TOM) complex (step A), where it can access its substrates
— ubiquitin chains near the TOM complex (step Ba) or the E3
ubiquitin-protein ligase parkin, which may come in contact with
PINK1 through a diffusion-limited mechanism (step Bb).
Phosphorylation of ubiquitin by PINK1 creates a binding site for
parkin (step C), which can then increase parkin chain assembly
activity by ~1,000-fold and facilitate phosphorylation of Ser65 on
the ubiquitin-like domain of parkin by PINK1 (step Da). Direct
phosphorylation of parkin by PINK1 leads to local activation of its
E3 activity (~2,400-fold), providing additional ubiquitin for
phosphorylation by PINK1 (steps Db–F). Parkin that is
phosphorylated on Ser65 and is associated with Ser65-phosphorylated
ubiquitin (pSer65-Ub) is maximally active in ubiquitin chain
assembly (~4,400-fold) (step E). The combination of parkin
activation, recruitment to pSer65-Ub chains, further ubiquitin
chain synthesis and further PINK1-dependent phosphorylation of
ubiquitin constitutes the feedforward mechanism (step F). DUB,
deubiquitinating enzyme; MIM, mitochondrial inner membrane; MOM,
mitochondrial outer membrane; USP30, ubiquitin carboxyl-terminal
hydrolase 30.
MOM MIM
Local ubiquitylation of substrates and phosphorylation of ubiquitin
by PINK1
Feedforward amplification of ubiquitin chain synthesis
Parkin makes Lys6, Lys11, Lys48 and Lys63 chains, and stochiometry
of phosphorylation of ubiquitin on mitochondria can reach
~20%
Diffusion-limited encounter
Some parkin phosphorylation before significant ubiquitin chain
assembly (Bb)
Further parkin recruitment and phosphorylation of both parkin and
ubiquitin by PINK1
A
C
Phosphorylation of local ubiquitin by PINK1 recruits inactive
parkin (Ba)
Opposing DUB (USP30) activity to reverse ubiquitin tagging of
mitochondria
PINK1
Activation of parkin by ~1,000-fold (Da) or ~2,400-fold (Db)
Parkin phosphorylated and bound to pSer65-Ub: further activation
(now ~4,400-fold) and retention on the MOM
Parkin binds to single phosphorylated ubiquitin molecules (Kd ~17
nM) and to phosphorylated ubiquitin chains
Phosphorylation Ubiquitylation
NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 19 | FEBRUARY 2018 |
101
M i to c h o n d r i a
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
Mutations in patients with Parkinson disease Numerous mutations
have been identified in PRKN genes through sequencing of patient
genomes, spanning all domains in the protein6. Functional analysis
of such parkin mutant proteins provides important insights into the
defective cellular mechanisms that underlie this form of Parkinson
disease, and results are consistent with the hypothesis that
multiple structural elements contribute to parkin
activity37–39,48,86.
Perhaps the most distinguishing characteristic of par- kin mutants
is that most are defective in their recruit- ment to mitochondria.
Given that parkin-dependent ubiquitin chain assembly is necessary
for stable parkin recruitment to mitochondria, the inability of
individ- ual parkin mutants to be recruited could be caused by the
absence of catalytic activity or the inability to be retained on
the MOM, for example, through defects in binding to pSer65-Ub. For
example, parkin-Lys161Asn and parkin-Lys211Asn mutants are strongly
defective in intrinsic activation by phosphorylation on Ser65,
pSer65-Ub-dependent activation and recruitment to damaged
mitochondria despite being capable of bind- ing to Lys63-linked
pSer65-Ub chains in vitro47. Thus, these and other parkin
mutants identified in patients may be primarily defective in steps
that support the feedforward mechanism.
A major question in this medical research field is whether it is
possible to design small molecules that target a mutant parkin
protein and reinstate its catalytic activity by locking it in an
active conformation. The finding that parkin undergoes multiple
conformational
changes during activation (FIG. 2) provides several oppor-
tunities for identifying small molecules that bind and stabil ize
one or more active forms. It is also possible that small molecules
that stabilize active forms of wild-type parkin downstream of PINK1
could facilitate the removal of damaged mitochondria in forms of
Parkinson disease that are unlinked genetically to parkin
and PINK1.
Decoding ubiquitin chains for mitophagy The assembly of ubiquitin
chains by parkin on damaged mitochondria initiates the process of
decoding by ubi- quitin-chain-binding autophagy receptors
(FIG. 4). These receptors, which include SQSTM1, next to BRCA1
gene 1 protein (NBR1), OPTN, calcium-binding and coiled-coil
domain-containing protein 2 (CALCOCO2; also known as NDP52) and
Tax1-binding protein 1 (TAX1BP1), contain a C-terminal UBD and a
short hydrophobic sequence (known as an LC3-interacting region
(LIR)) (FIG. 4a) that can bind to ATG8 proteins to potentially
pro- mote the recruitment of autophagosomal membranes via a
canonical autophagy mechanism87 (FIGS 1,4b).
SQSTM1 is recruited to depolarized mitochondria in a
parkin-dependent manner72, but it is not required for mitophagy in
most cell lines examined thus far33,34,71. SQSTM1 is instead
required for mitochondrial cluster- ing71. HeLa cells lacking OPTN,
CALCOCO2 and TAX1BP1, but still expressing SQSTM1 and NBR1, are
defective in mitophagy, with the most prominent defects shown in
cells lacking OPTN32–35. These data suggest some level of
functional redundancy between receptors, with the relative
contributions of individual
Box 3 | Factors acting in opposition of ubiquitin chain
assembly
Chain synthesis by ubiquitin ligases is often negatively regulated
by deubiquitinating enzymes (DUBs) (see also FIG. 3). Several
studies have examined the role of DUBs in parkin-dependent
mitochondrial ubiquitylation and mitophagy36,47,54,74,78,130–134.
Ubiquitin carboxyl-terminal hydrolase 30 (USP30), which is tethered
to the mitochondrial outer membrane (MOM) via a single
transmembrane domain, is perhaps the best understood DUB
antagonizing parkin activity, although USP8 and USP15 have also
been implicated36. Alteration in USP30 levels has a profound effect
on PTEN-induced putative kinase 1 (PINK1)– parkin-dependent
mitophagy in multiple systems. First, elevated USP30 expression
blocks ubiquitin chain assembly by parkin on the MOM47,74, thereby
reducing parkin recruitment74,132, while depletion of USP30 from
postmitotic neurons increased rates of mitophagy74, consistent with
USP30 actively antagonizing parkin function. Second, depletion of
USP30 in neuronal cultures reduced basal mitochondrial oxidative
stress74. Third, reduced USP30 expression substantially
improved behavioural phenotypes in pink and park mutant Drosophila
melanogaster, which is thought to be related to mitochondrial
function74. Finally, depletion of USP30 specifically in
dopaminergic neurons in D. melanogaster partially rescued
paraquat-induced behavioural phenotypes thought to model
mitochondrial damage74. Mechanistically, USP30 can remove ubiquitin
chains from numerous parkin substrates on the MOM but,
interestingly, is also a target of parkin-dependent
mono-ubiquitylation, leading to proteasomal degradation of USP30
(REFS 74,133). Thus, degradation of USP30 may help enforce the
feedforward mechanism for parkin function by eliminating the
negative regulator in the system (FIG. 3). Indeed, the
antagonistic relationship between parkin and USP30 suggests that
the balance between the activities of these two enzymes
(ubiquitin chain synthesis and ubiquitin chain disassembly) sets a
threshold for mitophagic
flux that is dictated by the activity of PINK1 on the mitochondria,
thereby controlling which mitochondria are detected as damaged.
Although the precise activities differ, multiple studies78,133,134
indicate a preference of USP30 in vitro for Lys6 linkages over
other chain types, although other chain types can also be
hydrolysed. This selectivity can be explained by unique
interactions revealed in the structure of USP30 bound to Lys6-
di-ubiquitin133,134. Interestingly, ubiquitin phosphorylation
within chains, but particularly when phosphorylated on the distal
ubiquitin within a tetrameric chain, also reduces activity133,134.
This result, along with the finding that PINK1 displays a kinetic
preference for phosphorylation of distal ubiquitin moieties in Lys6
tetramers but less so in other chain types, suggests that PINK1
could ‘end protect’ Lys6 ubiquitin chains from disassembly by
USP30. Consistent with this, depletion of USP30 specifically
increased the abundance of Lys6 chains on mitochondrial import
receptor subunit TOM20 homologue (TOMM20) but did not increase Lys6
chains on several other parkin targets133. Distinct models have
emerged for other DUBs36. USP8 has been proposed to remove Lys6
ubiquitin chains from parkin itself to alter its ability to be
recruited to mitochondria in cancer cell lines131. By contrast,
USP15 overexpression does not block parkin recruitment to the MOM
but does block MOM ubiquitylation, while its depletion in
fibroblasts from patients with Parkinson disease induces
mitophagy130. However, USP15 is thought to participate in mRNA
splicing135, and it is currently unclear if there is a contribution
of altered gene expression to the phenotypes observed, rather than
a direct effect on mitochondrial ubiquitylation. Further studies
are necessary to understand how these DUBs as well as a second
mitochondrial DUB, USP35 (REF. 132), are linked with parkin
function. Importantly, USP30 is a candidate target for small
molecules that will increase mitophagy in patients with decreased
parkin or PINK1 function36.
R E V I E W S
102 | FEBRUARY 2018 | VOLUME 19 www.nature.com/nrm
R E V I E W S
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
receptors probably reflecting their relative abundance in
individual cell types rather than intrinsically distinct
activities33. Indeed, expression of several receptors is
tissue-specific, which has implications for disease, as t hey may
function in and affect only specific cell lineages33.
For example, there is clear evidence of an essential role for
SQSTM1 in parkin-dependent mitophagy in mouse macrophages
in vivo88 and in mouse embryonic fibro- blasts undergoing
mitophagy as a result of high levels of
oxidative phosphorylation89.
Figure 4 | Principles of mitophagy receptor recruitment and
activation. a | Domain structures of the major ubiquitin
(Ub)-binding autophagy receptors in mammals. b | Scheme
depicting how ubiquitin-binding autophagy receptors may function to
recruit phagophores through ATG8-dependent (canonical) or
independent (non-canonical) mechanisms. ‘?’ represents a
hypothetical protein involved in non-canonical phagophore
recruitment. c,d | Models for phosphorylation- dependent
regulation of optineurin (OPTN)–ubiquitin binding.
e | Mechanism of feedforward phosphorylation upon binding
of OPTN to ubiquitin chains on damaged mitochondria. Cytosolic
OPTN–serine/threonine-protein kinase TBK1 complexes are recruited
to ubiquitin chains in response to mitochondrial activity of the E3
ubiquitin-protein ligase parkin and PTEN-induced putative kinase 1
(PINK1). Engagement of ubiquitin chains by the UBAN (ubiquitin
binding domain in ABINs and NEMO) domain of OPTN leads to TBK1
phosphorylation on Ser172 by an unknown process. TBK1 activation
promotes phosphorylation of OPTN on its LC3-interacting region
(LIR) and UBAN domains, which increases both association of OPTN
with ATG8 and binding of OPTN to ubiquitin chains. The feedforward
process promotes accumulation of OPTN–TBK1 on ubiquitin chains on
mitochondria. Note that the length of ubiquitin chains on
individual substrates is unknown. AR, autophagy receptor;
CALCOCO2, calcium-binding and coiled-coil domain-containing protein
2; CC, coiled-coil domain; PB, Phox and Bem1p domain; SKICH,
skeletal muscle and kidney-enriched inositol phosphatase carboxyl
homology domain; SQSTM1, sequestosome 1; TAX1BP1, Tax1-binding
protein 1; UBA, ubiquitin-associated domain; UBZ, ubiquitin binding
zinc-finger domain; UFD, ubiquitin fold domain; ZnF, zinc-finger
domain.
PINK1-mediated and parkin- mediated ubiquitylation on mitochondrial
outer membrane proteins
Kinase X
Phagophore membrane
Phagophore membrane
d Reduced OPTN binding to phosphorylated ubiquitin chains in
vitro
b Canonical autophagy Non-canonical autophagy
LIR
Substrate
a Autophagy receptors
–
NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 19 | FEBRUARY 2018 |
103
M i to c h o n d r i a
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
Xenophagy The process by which intracellular bacteria are targeted
for autophagy.
A major focus of research is on understanding which types of
ubiquitin chains are recognized by auto- phagy receptors.
Replacement of Lys6 or Lys63 in ubiquitin with arginine, which
cannot be conjugated to ubi quitin, reduced mitophagy rates
compared with Lys11 replacement, which led to only a minor
reduction57. Overexpression of Lys6Arg or Lys48Arg and Lys63Arg
ubiquitin mutants also inhibited mitophagy78. Moreover,
overexpression of ubiquitin-Lys11Arg reduced mitophagy to the
greatest extent among the mutants tested (~50% reduction)57. The
reason for this discrepancy is unknown but might reflect indirect
effects of ubiquitin- Lys11Arg on ubiquitin chain synthesis by
parkin, a hypothesis that has not been examined. In vitro, SQSTM1,
OPTN and CALCOCO2 bind more efficiently to Lys63 chains than to
Lys48 chains34,57,90, consistent with a role for Lys63 chains in
mitophagy, although the other chain types have not been tested
systematically.
The finding that ~20% of ubiquitin molecules on mitochondria are
phosphorylated upon depolarization in HeLa cells overexpressing
parkin47,57 raises the ques- tion of whether this modification
plays a direct role in the recruitment of mitophagy receptors.
Indeed, over- expression of PINK1 artificially targeted to
mitochon- dria promotes OPTN recruitment and mitophagy in the
absence of parkin, albeit with an extensive delay and lower
efficiency compared with when parkin activity is present33.
Although these observations have led to the conclusion that
pSer65-Ub is the receptor for OPTN and other autophagy receptors,
other data suggest that unphosphorylated forms of ubiquitin
conjugates on mitochondria function to recruit autophagy receptors
(FIG. 4c,d). In particular, in vitro experiments indicate
that OPTN, CALCOCO2 and SQSTM1 bind efficiently to unphosphorylated
Lys63 (but not Lys48) ubiquitin chains, and phosphorylation of
these chains on Ser65 (with a stoichio metry of ~0.7) largely
abolishes direct interactions between autophagy receptors and
ubiquitin chains34,47,57.
These findings are inconsistent with pSer65-Ub being directly
involved in the decoding of ubiquitin conjugates to promote
mitophagy. Moreover, quantitative proteomic experiments
demonstrated recruitment of endogenous OPTN, SQSTM1, TAX1BP1 and
CALCOCO2 to dam- aged mitochondria in the presence of wild-type
PINK1 and parkin, but recruitment was absent in PINK1−/− HeLa cells
or in cells expressing parkin-Ser65Ala (which failed to build
ubiquitin chains on mitochondria but expressed active PINK1)75.
Given that PINK1 should phosphorylate ubiquitin molecules that are
present on the mitochon- drial surface to promote OPTN recruitment,
these data suggest that endogenous levels of PINK1 are not suffi-
cient for receptor recruitment in the absence of ubiqui- tin chain
assembly by parkin. Moreover, imaging studies have shown that
pSer65-Ub signals uniformly cover damaged mitochondria, whereas
mitophagy receptors are recruited to highly focal puncta that cover
only a small part of the surface area of mitochondria34, indicat-
ing that pSer65-Ub conjugated to mitochondria is not sufficient to
directly recruit autophagy receptors and that additional signals
are necessary to direct receptors to these puncta.
Further studies are required to identify the signals that enable
mitophagy receptors to decode ubiquitin chains. Based on structural
data, a minimum of two ubiquitin molecules in a Lys63 chain are
necessary for binding to the ubiquitin binding domain in ABINs and
NEMO (UBAN) modules as found in OPTN91,92. How the ubiqui- tin
chain length is optimized to provide sufficient sites for parkin
recruitment to pSer65-Ub and sufficient con- jugates of two or more
ubiquitin molecules to support autophagy receptor recruitment is
unknown.
TBK1 promotes mitophagy A common feature of autophagy receptors is
their ability to interact with the kinase TBK1. Early studies
examining Salmonella enterica turnover by xenophagy demonstrated
that TBK1 can phosphorylate serine residues adjacent to the LIR
motif in OPTN and that this phosphoryl- ation promotes binding of
OPTN to ATG8 proteins to increase xenophagy93. It is now known that
TBK1 is required for multiple types of selective autophagy that
rely on ubiquitin- binding autophagy receptors34,94–97 and it can
phosphorylate OPTN, TAX1BP1, CALCOCO2 and SQSTM1 when
overexpressed35.
The best understood role for TBK1 is in regulating OPTN during
mitophagy and xenophagy. TBK1 phos- phoryl ates residues within and
adjacent to the UBAN motif in OPTN to increase its affinity for
unphos- phoryl ated Met1, Lys48 and Lys63 chains34,35,97
(FIG. 4c–e). Inter estingly, a positive feedback mechanism
controls TBK1-dependent phosphorylation of OPTN (FIG. 4e). The
ability of TBK1 to phosphorylate OPTN in response to mitochondrial
depolarization depends on OPTN binding to ubiquitin chains
(FIG. 4c,e). Inhibition of TBK1 activity with small molecule
inhibitors or dele- tion of TBK1 reduces OPTN recruitment to
damaged mitochondria, indicating that TBK1 activity is needed for
OPTN binding to ubi quitin chains34,35. Importantly, mutation of
serine residues in OPTN phosphorylated by TBK1 reduces its associ
ation with damaged mito- chondria and TBK1-dependent activation,
which delays mitophagy. Reciprocally, OPTN binding to ubiquitin
chains is required for TBK1 phos phorylation on Ser172, which
activates its kinase activity34,35. These findings have revealed
another crucial feedforward loop that is required for the efficient
clearance of defective mitochondria through mitophagy
(FIG. 4e).
Several questions remain to be addressed to fully understand this
pathway. First, by what mechanism is TBK1 activated in response to
the binding of OPTN to ubiquitin chains on mitochondria? Small
molecule inhib itors of TBK1 that prevent trans-autoactivation fail
to block its activation via phosphorylation on Ser172, suggesting
that one or more additional kinases may be involved34. Second, it
is unclear whether phosphoryl ation of the autophagy receptors
CALCOCO2 or TAX1BP1 by TBK1 (REF. 35) also increases their
affinity for ubi quitin chains. Phosphorylation on the UBDs of OPTN
and SQSTM1 increases their affinity for ubiquitin chains34,35,98,
suggesting that similar mechanisms regulate other recep- tors.
Third, it is not clear whether TBK1 has other func- tions in
addition to receptor phosphorylation. In this
R E V I E W S
104 | FEBRUARY 2018 | VOLUME 19 www.nature.com/nrm
R E V I E W S
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
regard, it has been shown that fusion of TBK1 lacking its
C-terminal autophagy receptor binding domain to UBDs from OPTN and
other proteins can rescue the clearing of Salmonella enterica in
TBK1-null cells95. However, the role of TBK1 kinase activity in the
context of Salmonella- infected cells remains unknown and could
involve phos- phorylation of proteins other than ubiquitin-binding
cargo receptors.
Finally, TBK1 and OPTN, but not parkin or PINK1, can be mutated in
patients with ALS and frontal tem- poral dementia17–19, and these
mutations affect the TBK1–OPTN interaction18,97. This suggests that
forms of TBK1–OPTN-driven selective autophagy, possibly involving
cargo other than damaged mitochondria, may be important for the
health of motor neurons and possibly other neuronal cell
types19.
Evolving roles of ATG8 proteins in mitophagy Canonical models for
mitophagy posit that recruitment of autophagy receptors to
ubiquitylated mitochondria leads to the recruitment of
autophagosomal membrane precursors to the surface of mitochondria
and subsequent engulfment of the damaged mitochondria by the auto-
phagosome6,19 (FIGS 1,4b). This recruitment is expected to
occur through association of the LIR elements in OPTN or CALCOCO2
with ATG8 proteins within the grow- ing autophagosome93,99.
However, the canonical model for ATG8 receptor recognition does not
explain recent studies reporting that human cells lacking all six
ATG8 proteins are still capable of building autophagosomes around
damaged mitochondria100. Thus, the canonical association of LIR
sequences in autophagy receptors with ATG8 proteins does not seem
to be required for this form of selective autophagy (FIG. 4b).
Moreover, cells lacking the ATG8 conjugation system still support
signifi cant (~30%) flux for starvation-induced bulk autophagy,
with a decreased frequency of autophagosomal closure and decreased
rates of autophagosomal inner membrane breakdown upon fusion with
lysosomes101,102. Thus, capture of some types of autophagic cargo
may occur independently of ATG8–LIR interactions. It is possible
that such ATG8-conjugation-independent forms of what is otherwise
considered to be selective autophagy involve interaction of
distinct sequences in either the autophagy receptor or associated
TBK1 with the autophagosomal machinery (FIG. 4b).
Nevertheless, ATG8 proteins and the ATG8 lipidation machinery are
required for mitophagic flux, and their absence correlates with
defects in fusion of lysosomes with autophagosomes100,103.
Recent studies have proposed a role for prohibitin 2,
a poorly understood mitochondrial inner membrane protein, as
an autophagy receptor for mitochondria that functions by directly
interacting with ATG8 pro- teins104. Given the proposed direct
binding to ATG8, it is unclear whether prohibitin 2 plays a
role in inducing mitophagy independently of the ubiquitin
conjugation pathway. In addition, it is not clear why cells lacking
OPTN, CALCOCO2 and TAX1BP1 fail to undergo mitophagy33–35 if
prohibitin 2 is a direct autophagy recep- tor as proposed104.
Further investigations are needed to understand when LIR–ATG8
interactions are required
and to understand any interplay between ubiquitin bind- ing
mitophagy receptors and prohibitin 2. In this regard, multiple
prohibitin 2 ubiquitylation sites are detected in a
parkin-dependent and PINK1-dependent manner and in a kinetically
delayed manner relative to most primary parkin targets, consistent
with ubiquitylation occurring after MOM rupture75.
Coupling parkin function to antigen presentation While much of the
efforts to understand the role of par- kin and PINK1 in disease
have focused on the removal of mitochondria via mitophagy, parkin
can also function to suppress the presentation of
mitochondria-derived anti- gen105, suggesting a novel autoimmune
mechanism for Parkinson disease in patients with mutations in
parkin and PINK1.
In macrophages and dendritic cells as well as fibroblasts, parkin
suppresses heat stress and lipopoly sac char ide (LPS)-dependent
production of mito chondrial antigens by ubiquitin-dependent
turnover of sorting nexin 9 (SNX9), a protein required for
parkin- independent prod uction of mitochondrial-derived vesicles
(MDVs)106. In the absence of parkin, MDVs transfer
mitochondrial contents into endosomes, where peptides are
ultimately presented on major histo compatibility complex
class I molecules105. This results in targeting of
T cells to antigen- presenting cells. These data suggest a
non-cell autono- mous mechanism that could contribute to Parkinson
disease, whereby cytotoxic T cell activity promotes the loss
of dopaminergic neurons. Intriguingly, LPS induces selective loss
of dopaminergic neurons in Prkn−/− mice107, raising the possibility
that age-dependent neuroinflam- mation could underlie neuronal loss
in humans105. Further explor ation of this system and the pathways
by which parkin selectively degrades SNX9 to promote its
proteasomal turnover to block production of MDVs might provide a
new paradigm in parkin function.
Conclusions and future questions Parkin-dependent mitophagy has
provided a paradigm for understanding the molecular mechanisms that
couple ubiquitin chain synthesis to recruitment of auto- phagy
receptors that are required to induce mitophagy and possibly other
types of organellar autophagy.
Parkin is unique in that it is the only ubiquitin ligase known to
be activated by binding to pSer65-Ub. It is intriguing to consider
that parkin may have evolved to function only in the presence of
pSer65-Ub. PINK1 is the only known Ser65-ubiquitin kinase, although
pSer65-Ub is also seen in budding yeast108, which lacks an obvious
PINK1 orthologue. Therefore, mitophagy is the only signalling
system that has been discovered for parkin activation. A plethora
of candidate parkin substrates were reported before the discovery
that parkin is activ- ated by PINK1 and pSer65-Ub109–115, but
virtually none of these studies examined what we now understand to
be the active form of parkin. It therefore remains unclear whether
there are alternative protein kinases that can phosphorylate parkin
and/or ubiquitin in the context of other signalling pathways that
parkin has been implicated in. Interestingly, pSer65-Ub can be
detected at very low
Starvation-induced bulk autophagy The process by which nutrient
deprivation leads to engulfment of cytosolic contents in
autophagosomes followed by delivery to lysosomes.
Dendritic cells Antigen-presenting immune cells that activate T
cells.
T cells Lymphocytes that function in cell-mediated immunity
and contain the T cell receptor on their cell surface.
R E V I E W S
NATURE REVIEWS | MOLECULAR CELL BIOLOGY VOLUME 19 | FEBRUARY 2018 |
105
M i to c h o n d r i a
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
levels in PINK1-null cells47, suggesting the existence of
additional Ser65-ubiquitin kinases. Moreover, parkin is required
for xenophagic removal of M. tuberculosis from mammalian
cells, but this function is apparently inde- pendent of PINK1,
suggesting that parkin activation for this form of xenophagy may
involve a distinct kinase sig- nalling pathway26. Exploration of
other signals that pro- mote pSer65-Ub may yield new insights into
signalling pathways regulated by parkin.
Our understanding of pSer65-Ub in animals is limited. In brain
tissue of mice lacking parkin, pSer65-Ub is increased upon
mutations in POLG, which encodes the mitochondrial DNA polymerase
gamma, catalytic subunit, which leads to mitochondrial stress as a
result of defects in respiratory chain assembly116. Moreover,
pSer65-Ub is detectable in human brain using specific antibodies,
and the signal increases with age and disease in limited patient
samples but is absent when PINK1 is mutated117. Thus, much work
still needs to be done to understand pSer65-Ub pathways in normal
and patho- genic conditions. Signals that activate PINK1 to promote
parkin function in both embryos and haematopoietic stem cells
remain to be identified (BOX 1).
Our understanding of the activation mechanism for parkin and the
role for pSer65-Ub has advanced rapidly since the discovery of
pSer65-Ub in 2014, but several ques- tions remain unresolved. We do
not have a clear under- standing of how pSer65-Ub and parkin
phosphorylation modify parkin structure to fully activ ate its
chain assem- bly function. The current structures48,64 with
pSer65-Ub still have some elements of auto inhibition in place that
are presumably removed upon full activation. An addi- tional
question concerns how parkin initially encounters PINK1. Parkin
does not seem to form a stable complex with activated PINK1 on the
TOM complex isolated from depolarized cells31, suggesting that if
initial activ ation of parkin occurs in the context of a PINK1–TOM
complex, this interaction is transient. Structural analysis of
PINK1– parkin complexes may shed light on this question. More-
over, it is unclear whether speci fic protein phosphatases act on
either pSer65-parkin or pSer65-Ub, thereby potentially providing a
second threshold to overcome in addition to that imposed by DUBs
(BOX 3).
Because most studies in the pathway have employed cells with
overexpressed parkin or PINK1, which can affect the amplitude and
persistence of the feedforward system, we do not fully understand
the temporal order of
individual steps in the pathway and the relative impor- tance of
PINK1-dependent phosphorylation of ubiqui- tin present on
mitochondria before damage versus direct parkin activation, for
example, in postmitotic neurons. It is possible that the
levels of pre-existing ubiquitin on the MOM are cell-type dependent
or are regulated by distinct mitochondrial E3s linked with
mitochondrial dynamics, such as mitochondrial ubiquitin ligase
activ- ator of NFKB 1 (MUL1) or the E3 ubiquitin-protein ligase
MARCH5 (REFS 89,118), thereby facilitating the feed forward
initiation mechanism directly. Alternatively, USP30 or other DUBs
may act to control the abun- dance of pre- existing ubiquitin on
the MOM (BOX 3). Understanding the biochemical steps necessary
for par- kin activation may facilitate the identification of mol-
ecules that can promote parkin activation in the context of disease
alleles in parkin itself or PINK1.
Finally, only recently has the link between parkin and
mitochondrial antigen presentation been made. This link provides a
completely novel regulatory func- tion for parkin that may provide
key insights into the fate of neurons in patients with Parkinson
disease with mutations in PINK1 or parkin through a potential auto-
immune mechanism. According to current models105,106, when under
stress, antigen-presenting cells harbouring mutations in parkin
produce MDVs that allow presen- tation of mitochondrial-derived
antigens on the cell surface and subsequent T cell activation.
Neurons that present mitochondrial- derived antigens on their
surface could be subsequently recognized by mitochondrial
antigen-specific T cells, thereby triggering a cytotoxic
response ultimately leading to neuronal death through an
autoimmune-type mechanism. Major questions concern how signals
downstream of stressors linked with mito- chondrial antigen
presentation are coupled to parkin and PINK1 activation, the extent
to which parkin activation in antigen-presenting cells is uncoupled
from the canoni- cal mitophagy system, and how parkin selectively
recog- nizes SNX9 for ubiquitylation to suppress mitochondrial
antigen presentation. Furthermore, it is crucial to know the cell
types in vivo in which the pathway is active and thereby might
be targeted by cytotoxic T cells. The availability of mice
expressing reporters of mitophagic flux119,120 will greatly
facilitate a physiological understand- ing of spatial and temporal
control of mitophagy and the genetic requirements for the PINK1–
parkin system across a broad range of tissues.
1. Haynes, C. M., Petrova, K., Benedetti, C.,
Yang, Y. & Ron, D. ClpP mediates activation of a
mitochondrial unfolded protein response in C. elegans. Dev.
Cell 13, 467–480 (2007).
2. Munch, C. & Harper, J. W. Mitochondrial
unfolded protein response controls matrix pre-RNA processing
and translation. Nature 534, 710–713 (2016).
3. Lin, Y. F. & Haynes, C. M. Metabolism
and the UPRmt. Mol. Cell 61, 677–682 (2016).
4. Shpilka, T. & Haynes, C. M. The mitochondrial
UPR: mechanisms, physiological functions and implications in
ageing. Nat. Rev. Mol. Cell Biol. http://dx.doi.org/
10.1038/nrm.2017.110 (2017).
5. Lin, Y. F. et al. Maintenance and propagation of
a deleterious mitochondrial genome by the mitochondrial
unfolded protein response. Nature 533, 416–419 (2016).
6. Pickrell, A. M. & Youle, R. J. The roles
of PINK1, parkin, and mitochondrial fidelity in Parkinson’s
disease. Neuron 85, 257–273 (2015).
7. Yamano, K., Matsuda, N. & Tanaka, K. The
ubiquitin signal and autophagy: an orchestrated dance leading to
mitochondrial degradation. EMBO Rep. 17, 300–316 (2016).
8. De Duve, C. & Wattiaux, R. Functions of lysosomes.
Annu. Rev. Physiol. 28, 435–492 (1966).
9. Kitada, T. et al. Mutations in the parkin gene cause
autosomal recessive juvenile parkinsonism. Nature 392, 605–608
(1998).
10. Narendra, D. P. et al. PINK1 is selectively
stabilized on impaired mitochondria to activate Parkin.
PLoS Biol. 8, e1000298 (2010).
11. Valente, E. M. et al. Hereditary early-onset
Parkinson’s disease caused by mutations in PINK1. Science 304,
1158–1160 (2004).
12. Walden, H. & Muqit, M. M. Ubiquitin and
Parkinson’s disease through the looking glass of genetics. Biochem.
J. 474, 1439–1451 (2017).
13. Clark, I. E. et al. Drosophila pink1 is required
for mitochondrial function and interacts genetically
with parkin. Nature 441, 1162–1166 (2006).
14. Park, J. et al. Mitochondrial dysfunction in
Drosophila PINK1 mutants is complemented by parkin. Nature 441,
1157–1161 (2006). References 13 and 14 provide the initial genetic
evidence that PINK1 and parkin function in a common pathway to
control mitochondrial function.
15. Narendra, D., Tanaka, A., Suen, D. F. &
Youle, R. J. Parkin is recruited selectively to impaired
mitochondria and promotes their autophagy. J. Cell Biol. 183,
795–803 (2008). This study provides a link between parkin
recruitment to mitochondria and a role in mitophagy.
R E V I E W S
106 | FEBRUARY 2018 | VOLUME 19 www.nature.com/nrm
R E V I E W S
© 2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved. ©
2018
Macmillan
Publishers
Limited,
part
of
Springer
Nature.
All
rights
reserved.
http://dx.doi.org/10.1038/nrm.2017.110
http://dx.doi.org/10.1038/nrm.2017.110
16. Maruyama, H. et al. Mutations of optineurin in
amyotrophic lateral sclerosis. Nature 465, 223–226 (2010).
17. Cirulli, E. T. et al. Exome sequencing in
amyotrophic lateral sclerosis identifies risk genes and pathways.
Science 347, 1436–1441 (2015).
18. Freischmidt, A. et al. Haploinsufficiency of TBK1
causes familial ALS and fronto-temporal dementia. Nat. Neurosci.
18, 631–636 (2015). References 17 and 18 report the identification
of TBK1 mutations in patients with sporadic and familial
ALS.
19. Weishaupt, J. H., Hyman, T. & Dikic, I.
Common molecular pathways in amyotrophic lateral sclerosis and
frontotemporal dementia. Trends Mol. Med. 22, 769–783 (2016).
20. Novak, I. et al. Nix is a selective autophagy
receptor for mitochondrial clearance. EMBO Rep. 11, 45–51
(2010).
21. Chourasia, A. H. et al. Mitophagy defects
arising from BNip3 loss promote mammary tumor progression
to metastasis. EMBO Rep. 16, 1145–1163 (2015).
22. Li, W. et al. MicroRNA-137 is a novel hypoxia-
responsive microRNA that inhibits mitophagy via regulation of two
mitophagy receptors FUNDC1 and NIX. J. Biol. Chem. 289,
10691–10701 (2014).
23. Ito, K. et al. Self-renewal of a purified Tie2+
hematopoietic stem cell population relies on mitochondrial
clearance. Science 354, 1156–1160 (2016).
24. Boland, M. L., Chourasia, A. H. &
Macleod, K. F. Mitochondrial dysfunction in cancer.
Front. Oncol. 3, 292 (2013).
25. Ney, P. A. Mitochondrial autophagy: origins,
significance, and role of BNIP3 and NIX. Biochim. Biophys.
Acta 1853, 2775–2783 (2015).
26. Manzanillo, P. S. et al. The ubiquitin ligase
parkin mediates resistance to intracellular pathogens. Nature 501,
512–516 (2013).
27. Okatsu, K., Kimura, M., Oka, T., Tanaka, K.
& Matsuda, N. Unconventional PINK1 localization
to the outer membrane of depolarized mitochondria drives
Parkin recruitment. J. Cell Sci. 128, 964–978 (2015).
28. Jin, S. M. et al. Mitochondrial membrane
potential regulates PINK1 import and proteolytic destabilization by
PARL. J. Cell Biol. 191, 933–942 (2010).
29. Greene, A. W. et al. Mitochondrial processing
peptidase regulates PINK1 processing, import and Parkin
recruitment. EMBO Rep. 13, 378–385 (2012).
30. Yamano, K. & Youle, R. J. PINK1 is degraded
through the N-end rule pathway. Autophagy 9, 1758–1769
(2013).
31. Lazarou, M., Jin, S. M., Kane, L. A.
& Youle, R. J. Role of PINK1 binding to the TOM
complex and alternate intracellular membranes in recruitment and
activation of the E3 ligase Parkin. Dev. Cell 22, 320–333
(2012).
32. Wong, Y. C. & Holzbaur, E. L.
Optineurin is an autophagy receptor for damaged mitochondria in
parkin-mediated mitophagy that is disrupted by an ALS-linked
mutation. Proc. Natl Acad. Sci. USA 111, E4439–E4448 (2014).
33. Lazarou, M. et al. The ubiquitin kinase PINK1
recruits autophagy receptors to induce mitophagy. Nature 524,
309–314 (2015).
34. Heo, J. M., Ordureau, A.,
Paulo, J. A., Rinehart, J.
& Harper, J. W. The PINK1-PARKIN mitochondrial
ubiquitylation pathway drives a program of OPTN/ NDP52 recruitment
and TBK1 activation to promote mitophagy. Mol. Cell 60, 7–20
(2015).
35. Richter, B. et al. Phosphorylation of OPTN by TBK1
enhances its binding to Ub chains and promotes selective autophagy
of damaged mitochondria. Proc. Natl Acad. Sci. USA 113,
4039–4044 (2016). References 32–35 define the roles of
ubiquitinbinding autophagy receptors in the capture of
ubiquitylated mitochondria to promote mitophagy.
36. Bingol, B. & Sheng, M. Mechanisms of mitophagy:
PINK1, Parkin, USP30 and beyond. Free Radic. Biol. Med. 100,
210–222 (2016).
37. Chaugule, V. K. et al. Autoregulation of Parkin
activity through its ubiquitin-like domain. EMBO J. 30,
2853–2867 (2011).
38. Wauer, T. & Komander, D. Structure of the human
Parkin ligase domain in an autoinhibited state. EMBO J. 32,
2099–2112 (2013).
39. Trempe, J. F. et al. Structure of parkin reveals
mechanisms for ubiquitin ligase activation. Science 340, 1451–1455
(2013).
40. Riley, B. E. et al. Structure and function of
Parkin E3 ubiquitin ligase reveals aspects of RING and HECT
ligases. Nat. Commun. 4, 1982 (2013). References 37–40 define the
structure of autoinhibited parkin and define structural elements
that impinge upon RING1 and RING2 functions.
41. Fajner, V., Maspero, E. & Polo, S. Targeting
HECT-type E3 ligases — insights from catalysis, regulation and
inhibitors. FEBS Lett. 591, 2636–2647 (2017).
42. Wenzel, D. M., Lissounov, A.,
Brzovic, P. S. & Klevit, R. E. UBCH7
reactivity profile reveals parkin and HHARI to be RING/HECT
hybrids. Nature 474, 105–108 (2011).
43. Tang, M. Y. et al. Structure-guided mutagenesis
reveals a hierarchical mechanism of Parkin activation. Nat. Commun.
8, 14697 (2017).
44. Sha, D., Chin, L. S. & Li, L.
Phosphorylation of parkin by Parkinson disease-linked kinase PINK1
activates parkin E3 ligase function and NF-κB signaling.
Hum. Mol. Genet. 19, 352–363 (2010).
45. Kondapalli, C. et al. PINK1 is activated by
mitochondrial membrane potential depolarization and stimulates
Parkin E3 ligase activity by phosphorylating Serine 65. Open
Biol. 2, 120080 (2012).
46. Shiba-Fukushima, K. et al. PINK1-mediated
phosphorylation of the Parkin ubiquitin-like domain primes
mitochondrial translocation of Parkin and regulates mitophagy. Sci.
Rep. 2, 1002 (2012). References 45 and 46 identify Ser65 in parkin
as a target for PINK1.
47. Ordureau, A. et al. Quantitative proteomics reveal a
feedforward mechanism for mitochondrial PARKIN translocation and
ubiquitin chain synthesis. Mol. Cell 56, 360–375 (2014). This study
describes a feedforward mechanism for parkin activation by PINK1
and pSer65Ub binding.
48. Wauer, T., Simicek, M., Schubert, A. &
Komander, D. Mechanism of phospho-ubiquitin-induced PARKIN
activation. Nature 524, 370–374 (2015). This study provides the
first structural analysis of pSer65Ub bound to parkin and defines
multiple mechanisms of parkin activation.
49. Aguirre, J. D., Dunkerley, K. M.,
Mercier, P. & Shaw, G. S. Structure of
phosphorylated UBL domain and insights into PINK1-orchestrated
parkin activation. Proc. Natl Acad. Sci. USA 114, 298–303
(2017).
50. Caulfield, T. R. et al. Phosphorylation by PINK1
releases the UBL domain and initializes the conformational opening
of the E3 ubiquitin ligase Parkin. PLoS Comput. Biol. 10, e1003935
(2014).
51. Kazlauskaite, A. et al. Parkin is activated by
PINK1-dependent phosphorylation of ubiquitin at Ser65.
Biochem. J. 460, 127–139 (2014).
52. Koyano, F. et al. Ubiquitin is phosphorylated by
PINK1 to activate parkin. Nature 510, 162–166 (2014).
53. Kane, L. A. et al. PINK1 phosphorylates
ubiquitin to activate Parkin E3 ubiquitin ligase activity.
J. Cell Biol. 205, 143–153 (2014).
54. Wauer, T. et al. Ubiquitin Ser65 phosphorylation
affects ubiquitin structure, chain assembly and hydrolysis. EMBO J.
34, 307–325 (2015). References 47 and 51–54 identify Ser65 of
ubiquitin as a substrate for PINK1 and demonstrate that pSer65Ub
activates parkin.
55. Okatsu, K. et al. Phosphorylated ubiquitin chain is
the genuine Parkin receptor. J. Cell Biol. 209, 111–128
(2015).
56. Sauve, V. et&nbs