Journal of Parkinson’s Disease 7 (2017) 13–29
DOI 10.3233/JPD-160989
IOS Press
13
Review
PINK1, Parkin, and Mitochondrial Quality
Control: What can we Learn about
Parkinson’s Disease Pathobiology?
Dominika Truban
a
,XuHou
a
, Thomas R. Caulfield
a,b
, Fabienne C. Fiesel
a,b
and Wolfdieter Springer
a,b,∗
a
Department of Neuroscience, Mayo Clinic, Jacksonville, FL, USA
b
Mayo Clinic Graduate School of Biomedical Sciences, Jacksonville, FL, USA
Accepted 10 November 2016
Abstract. The first clinical description of Parkinson’s disease (PD) will embrace its two century anniversary in 2017.
For the past 30 years, mitochondrial dysfunction has been hypothesized to play a central role in the pathobiology of this
devastating neurodegenerative disease. The identifications of mutations in genes encoding PINK1 (PTEN-induced kinase 1)
and Parkin (E3 ubiquitin ligase) in familial PD and their functional association with mitochondrial quality control provided
further support to this hypothesis. Recent research focused mainly on their key involvement in the clearance of damaged
mitochondria, a process known as mitophagy. It has become evident that there are many other aspects of this complex regulated,
multifaceted pathway that provides neuroprotection. As such, numerous additional factors that impact PINK1/Parkin have
already been identified including genes involved in other forms of PD. A great pathogenic overlap amongst different forms of
familial, environmental and even sporadic disease is emerging that potentially converges at the level of mitochondrial quality
control. Tremendous efforts now seek to further detail the roles and exploit PINK1 and Parkin, their upstream regulators and
downstream signaling pathways for future translation. This review summarizes the latest findings on PINK1/Parkin-directed
mitochondrial quality control, its integration and cross-talk with other disease factors and pathways as well as the implications
for idiopathic PD. In addition, we highlight novel avenues for the development of biomarkers and disease-modifying therapies
that are based on a detailed understanding of the PINK1/Parkin pathway.
Keywords: Parkinson’s disease, PINK1, Parkin, ubiquitin, mitochondria, mitophagy
THE ROLE OF MITOCHONDRIAL
DYSFUNCTION IN PD PATHOGENESIS
Mitochondria are energy producing organelles that
are enclosed by a double membrane and form a
complex and highly dynamic network regulated by
constant fission and fusion [1]. An electrochemi-
cal gradient (m) across the inner mitochondrial
membrane (IMM) is important for ATP production
∗
Correspondence to: Wolfdieter Springer, PhD, Department of
Neuroscience, Mayo Clinic, 4500 San Pablo Road, Jacksonville,
FL 32224, USA. Tel.: +1 904 953 6129; Fax: +1 904 953 7117;
E-mail: Springer.Wolfdieter@mayo.edu.
[2, 3]. Other important functions of mitochondria
include Ca
2+
buffering and the regulation of cel-
lular apoptosis [4, 5]. While mitochondria harbor
their own genome, it only encodes few subunits of
the mitochondrial respiratory chain complexes [6].
The majority of mitochondrial proteins are encoded
by the nuclear genome, thus requiring a coordinated
transcription, translation, import into, and protein
complex assembly inside mitochondria [7]. The
importance of intact, functional mitochondria is evi-
dent by a variety of symptoms that are associated with
classical mitochondrial diseases [1]. Especially vul-
nerable are tissues with a high energy demand such as
ISSN 1877-7171/17/$35.00 © 2017 – IOS Press and the authors. All rights reserved
This article is published online with Open Access and distributed under the terms of the Creative Commons Attribution Non-Commercial License (CC BY-NC 4.0).
14 D. Truban et al. / PINK1, Parkin, and Mitochondrial Quality Control
brain, muscle, heart, kidney, and liver. Neurological
symptoms include developmental delays, declining
mental abilities, seizures, blindness, and loss of motor
skills.
In addition, an increasing body of evidence sug-
gests that mitochondrial dysfunction is associated
with several age-related diseases and the aging pro-
cess itself. Initial studies showed that dysfunctional
mitochondria accumulate during aging [8], which
is a major risk factor for many neurodegenerative
diseases including PD [9]. The characteristic motor
and non-motor symptoms of PD result from striatal
depletion of the neurotransmitter dopamine (DA),
secondary to the relatively selective degeneration of
DA neurons in the substantia nigra pars compacta
(SNpc). DA neurons are characterized by distinctive
physiological features that could contribute to their
vulnerability and sensitivity toward mitochondrial
dysfunction [10–12]. One of the theories explaining
the susceptibility of DA neurons is based on their high
metabolic activities resulting in the increased produc-
tion of harmful reactive oxygen species (ROS), which
subsequently induce mitochondrial damage [13]. The
deleterious effects could then spread within the entire
mitochondrial network, eventually resulting in neu-
ronal cell death [14].
The first direct evidence linking mitochondrial
dysfunction to PD appeared with the observation that
accidental exposure to 1-methyl-4-phenyl-1,2,3,6-
tetrahydropyridine (MPTP) caused parkinsonism and
DA neuron degeneration in patients [15]. The product
of MPTP oxidation, MPP+, was shown to selectively
inhibit complex I, a component of the mitochon-
drial respiratory chain, which subsequently induces
DA neuronal death [16, 17]. Later, reduced activity
of complex I was also observed in different tissues
of sporadic PD patients, such as the SNpc, skele-
tal muscle and platelets [18–20]. The exposure to
other toxins such as the pesticide rotenone and the
herbicide paraquat that inhibit complex I or induce
ROS, respectively, also result in loss of nigrostriatal
DA neurons and have been linked to parkinsonian
phenotypes in humans and in animal models [21].
THE RECESSIVE PD GENES PINK1
AND PARKIN
Among others, the identification of the two genes
associated with PD, PINK1 (PARK6) [22] and Parkin
(PARK2) [23], became milestones in PD research.
Homozygous or compound heterozygous mutations
in both genes are the most common causes of reces-
sive early-onset PD (usually below age of 40) [24].
Though both genes are ubiquitously expressed, com-
plete loss of either function results in selective loss of
DA neurons and manifestation of tremor, bradykine-
sia, and rigidity. Besides earlier disease onset, PINK1
and Parkin associated PD develops generally more
benign than idiopathic disease, with a slower disease
progression as well as a good and sustained response
to levodopa treatment [25, 26]. Aside from the cardi-
nal symptoms of PD, PINK1 patients more frequently
show psychiatric symptoms (depression, anxiety, and
psychosis) [27]. Neuropathological examination has
casted some doubts as to whether Parkin-related PD
has the same disease mechanism as sporadic disease,
since several studies described the absence of Lewy
bodies, the pathognomonic sign of PD. However, to
date, at least five studies reported the presence of
Lewy bodies in Parkin cases [28–32]. In addition,
two out of three PINK1 cases that were studied so
far reported Lewy bodies [33–35]. The reason for
the neuropathological heterogeneity is unclear but
might relate to some specific mutations within PINK1
or Parkin and/or the age of the patients when they
decease.
While a mitochondrial function had been sug-
gested for PINK1 from the beginning simply based
on its mitochondrial targeting sequence (MTS), a
mitochondrial role for Parkin remained elusive for
several years. The first studies supporting the involve-
ment of both in regulating mitochondrial function
were performed in Drosophila [36–38]. PINK1
−/−
and Parkin
−/−
mutant flies exhibited similar mito-
chondrial morphological abnormalities, locomotor
deficits, muscle degeneration, male sterility as well
as neuronal loss [37, 38]. The PINK1
−/−
pheno-
type was rescued by Parkin overexpression, but not
vice versa, suggesting that PINK1 acts upstream of
Parkin in a common, linear pathway [37–39]. Mito-
chondrial abnormalities and rescue of PINK1 loss
by Parkin, but not PD-associated mutations were
confirmed in human cell lines and primary fibrob-
lasts [40]. However, PINK1 or Parkin knockout mice
showed only subtle phenotypes with some mito-
chondrial dysfunction, yet without overt pathological
changes in ultrastructure [41, 42].
In a breakthrough study in 2008, massive Parkin
translocation from the cytosol to damaged mitochon-
dria was observed after treatment with the uncou-
pler carbonyl cyanide m-chlorophenylhydrazone
(CCCP), a chemical that dissipates m [43].
Strikingly, Parkin recruitment to mitochondria was
D. Truban et al. / PINK1, Parkin, and Mitochondrial Quality Control 15
demonstrated shortly after to depend on the presence
of catalytically active PINK1 by several laborato-
ries [44–47]. Since PINK1 and Parkin were shown
to coordinate the selective clearance of damaged
mitochondria via the autophagy/lysosome pathway
(mitophagy), this also provided a link between the
two major cellular dysfunctions implicated in PD
pathogenesis, namely alterations in mitochondria
and in cellular degradation [48]. Importantly, PD-
associated mutations in either gene abrogated this
sequential process at different steps and through dis-
tinct pathomechanisms, underscoring the etiological
importance of this pathway [44, 49, 50].
THE MITOCHONDRIAL KINASE PINK1
PINK1 was initially identified in an expression
analysis of genes that were upregulated upon intro-
duction of the tumor growth suppressor phosphatase
and tensin homolog (PTEN) and was named accord-
ingly (PTEN-induced putative kinase 1) [51]. PINK1,
a 581 amino acid protein, contains an N-terminal
MTS, a transmembrane domain (TM), a highly
conserved serine/threonine kinase domain, and a
C-terminal auto-regulatory domain (Fig. 1A). The
overall frequency of PINK1 mutations in early-onset
PD ranges between 1–9% depending on the ethnicity
of the subjects [52]. Although crystal structures of
PINK1 are currently unavailable, in silico analysis
and structural modeling indicated some similari-
ties between its kinase domain and members of the
calmodulin-dependent kinase family [53–56].
Two critical regulatory regions within PINK1 are
the cleavage sites of the mitochondrial process-
ing peptidase (MPP) and the presenilin-associated
rhomboid-like protease (PARL) [57–59]. PINK1
activity is determined by autophosphorylation on
three residues (Ser228, Thr257, and Ser402) in the
activation loop [60–62]. While a complete loss of
kinase activity is unequivocally linked to early-onset
PD, as seen in, for instance, homozygous p.Q456X
carriers [63], a single PINK1 mutation that only
causes partial reduction in enzymatic activity could
also result in a milder phenotype or contribute to
disease vulnerability later in life. While this is a
matter of debate, mild PD symptoms were observed
in heterozygous individuals carrying the p.W437X
or p.Q456X mutation [64–66]. In addition, data
from our laboratory suggested that the heterozygous
PINK1 p.G411S mutation increases risk for PD by
a dominant-negative mechanism [67]. Though not
all PINK1 mutations appear to be alike, these stud-
ies highlight the disease effects of particular variants
and encourage a more detailed genetic and functional
analysis of heterozygous mutations in recessive PD
genes.
THE CYTOSOLIC E3 UBIQUITIN LIGASE
PARKIN
Parkin, a 465 amino acid protein, is a RING-
in-between-RING (RBR)-type E3 ubiquitin (Ub)
ligase [68] that catalyzes (multi-) mono- and
poly-ubiquitylation of numerous substrates that are
structurally and functionally divers, including itself
[68, 69]. Together with specific co-enzymes, Parkin
adds the small modifier protein Ub (76 amino acids)
to lysine residues of substrate proteins including Ub
that itself contains seven internal lysines. Consecutive
rounds of conjugation result in growth of poly-Ub
chains that, depending on the linkage type, present
with distinct topologies and thus biological functions.
Structurally, Parkin consists of an N-terminal Ub-like
(UBL) domain followed by a flexible linker and four
cysteine-rich regions that each coordinate two Zn
2+
atoms (Fig. 1B). Of those, three are really interest-
ing new gene (RING) domains (RING0, RING1 and
RING2), the last two of which are separated by an
in-between-RING (IBR) domain. Parkin mutations
are the primary cause of familial early-onset PD that
are found in nearly 50% of all young PD patients
(≤40 years old) [70, 71]. To date, over 150 PD-related
Parkin mutations have been identified across various
ethnic groups [24]. In addition to genetic mutations in
familial PD, Parkin inactivation by post-translational
modifications in sporadic disease has been suggested
[72].
Initially Parkin was identified as a classical RING-
type E3 Ub ligase [73] that bridges the interaction
between the E2 Ub-conjugating enzyme and a sub-
strate without physically receiving Ub. More recently
it was shown that Parkin has a catalytic cysteine
residue (Cys431) in its RING2 domain that forms
a transient thioester with Ub before further trans-
fer onto substrate protein, similar to E6-AP carboxyl
terminus (HECT)-type enzymes [74]. Indeed, the PD-
associated mutation p.C431F abolishes the E3 Ub
ligase activity of Parkin as it can no longer bind the
Ub moiety [75, 76]. This novel RING/HECT-hybrid
mechanism lead to its re-classification as a mem-
ber of the RBR-type family of E3 Ub ligases that
are auto-inhibited through several intra-molecular
16 D. Truban et al. / PINK1, Parkin, and Mitochondrial Quality Control
Fig. 1. PINK1 and Parkin domain structures and PD-related mutations. (A-B) Given are schematic, color-coded domain representations
of PINK1 and Parkin. PD-associated missense and nonsense mutations from the PD Mutation Database (http://www.molgen.vib-
ua.be/PDMutDB/) are displayed on top of each structure with their respective locations. Mutations in red have been experimentally verified
as loss-of-function mutations and are considered pathogenic, while functional defects for variants shown in black remain unclear. Underlined
mutations are common variants based on the ExAC database (http://exac.broadinstitute.org) with allele frequencies greater than 1:10 000.
(A) Domain structure of PINK1 (581 amino acids): mitochondrial targeting sequence (MTS, orange), transmembrane region (TM, red),
N-terminal regulatory region (NT, gray), N-lobe of the kinase domain (cyan), C-lobe of the kinase domain (purple) and the C-terminal
domain (CTD, blue). PD-associated mutations are listed on the top. Mitochondrial protease (MPP and PARL) cleavage sites and PINK1
auto-phosphorylation sites are displayed at the bottom. (B) Domain structure of Parkin (465 amino acids): ubiquitin-like domain (UBL,
red), linker (gray), really-interesting-new-gene (RING)/unique Parkin domain (R0/UPD, green), RING1 (R1, cyan), in-between-RING (IBR,
purple), repressor element of Parkin (REP, yellow), and RING2 (R2, pink). E2 co-enzyme and p-Ser65-Ub binding sites as well as Ser65
phosphorylation and Cys431 catalytic sites are displayed at the bottom. (C) Closed, inactive conformation of full-length human Parkin (left:
front view and right: back view). The structure is shown in colored ribbons that correspond to the respective domain colors. The solvent-
accessible surface area of each domain is shown in semi-transparent rendering in the same color. Ser65 is highlighted in Van der Waal
representation with standard atom coloring (hydrogen: white, oxygen: red, nitrogen: blue). The zinc-finger motifs of Parkin are rendered in
licorice stick with standard atom coloring and the corresponding zinc ions as spheres (cyan).
interactions [68]. While a role of the UBL domain
was already suggested earlier [77], elucidation of
Parkin’s inactive structure provided a complete pic-
ture of the auto-inhibitory mode [78–82]. In addition
to the N-terminus that keeps Parkin in a closed con-
formation, its activity is further repressed by the
REP (repressor of Parkin) region that blocks the
E2 binding site in its canonical RING1 (Fig. 1C).
D. Truban et al. / PINK1, Parkin, and Mitochondrial Quality Control 17
Moreover, RING0 intersects between RING1 and
RING2 and thereby buries Cys431 resulting in a
spatial separation between the active sites of E2
and Parkin, which disrupts Ub transfer. Targeted
mutations in the auto-inhibitory regions indeed acti-
vated Parkin and increased its auto-ubiquitylation
activity [79]. Based on the closed structure, it has
become apparent that Parkin must get activated
and has to undergo major structural rearrangements
in order to gain enzymatic E3 Ub ligase activity
[82, 83].
THE LIFE CYCLE OF PINK1 – A SENSOR
FOR MITOCHONDRIAL DAMAGE
PINK1 is imported into healthy mitochondria
through the general import machinery, the translo-
case of the outer membrane (TOM) and translocase of
the inner membrane (TIM) [84]. Inside mitochondria,
PINK1 undergoes consecutive cleavages by the mito-
chondrial proteases MPP that removes the MTS in the
matrix and by PARL that cleaves the TM domain in
the IMM [57–59]. As a result of both events, cleaved
PINK1 is re-exported and subsequently degraded by
the N-end rule pathway through the Ub/proteasome
system resulting in low levels of endogenous PINK1
in healthy mitochondria [85] (Fig. 2A). However,
inside mitochondria, PINK1 has been implicated in
the phosphorylation of different proteins, although
for some of them it is unclear whether this occurs
directly. PINK1-mediated phosphorylation of inner
mitochondrial proteins is generally thought to be pro-
tective. Among the reported PINK1 substrates are the
PD-linked serine protease HTRA2/Omi (PARK13)
[86], the chaperone TRAP1 [87] in the intermem-
brane space and the complex I subunit NDUFA10
[88] located on the IMM. In flies, overexpression
of TRAP1, for which direct phosphorylation by
PINK1 has been shown, completely rescued pheno-
types of PINK1, but not Parkin, deficiency [89, 90].
The phospho-mimetic version of NDUFA10 reversed
PINK1 loss-of-function induced deficits in mouse
cells and fly models [88].
Import of proteins, including PINK1, into mito-
chondria requires an active proton gradient and
is inhibited upon mitochondrial damage induced
by mitochondrial depolarizing agents (e.g. CCCP)
[91]. Consequently, full-length PINK1 accumulates
on the outer mitochondrial membrane (OMM) [91]
(Fig. 2B), where it is stabilized by autophospho-
rylation [61] and the formation of a large 700kDa
multimeric protein complex with the TOM machin-
ery [84, 92]. Components of the TOM complex are
essential for the alternative insertion of PINK1 into
the OMM [93, 94]. The dynamic nature of PINK1
protein levels and its dual mitochondrial localization
explain how PINK1 serves as a sensor of mitochon-
drial impairment and initiates the degradation of
selectively damaged mitochondria.
PINK1-MEDIATED PARKIN ACTIVATION
When full-length PINK1 is stabilized on the OMM
of damaged mitochondria, its kinase domain is fac-
ing the cytosol. Several cytosolic and OMM proteins
that modulate mitochondrial dynamics or apoptotic
signaling have also been suggested as candidate sub-
strates for PINK1 [95–97]. PINK1 was demonstrated
to phosphorylate Parkin at serine-65 (Ser65) in the N-
terminal UBL domain [60, 98, 99]. PINK1-mediated
phosphorylation of Parkin was shown to activate its
enzymatic functions and to induce its translocation
to damaged mitochondria (Fig. 2B) [98, 99]. Using
molecular modeling and dynamics simulation, we
suggested that phosphorylation of Parkin releases its
auto-inhibited structure and initiates opening confor-
mations that could lead to its enzymatic activation
[82]. Second, PINK1 was recently identified as the
first Ub kinase, which phosphorylates not only the
E3 Ub ligase Parkin but also the Ub itself at this con-
served residue (p-Ser65-Ub) [100–103]. Binding of
p-Ser65-Ub to the RING1 domain of Parkin leads to
straightening of a helix that results in an initial release
of the auto-inhibitory UBL domain [104–106]. In
turn, this facilitates phosphorylation of Parkin, which
causes further major structural rearrangements, de-
repression of enzymatic activity, and maintenance of
an active conformation [107–109]. While p-Ser65-
Ub also serves as the receptor for Parkin on damaged
mitochondria [106], both PINK1-mediated phospho-
rylation events together fully activate its enzymatic
functions.
Recruitment of Parkin to damaged mitochondria
then initiates a feedforward loop (Fig. 2B), wherein
activated Parkin further conjugates Ub moieties onto
OMM proteins and thereby provides more Ub sub-
strates for PINK1-dependent phosphorylation, which
in turn amplifies Parkin activation, recruitment, and
E3 Ub ligase activity [101, 102, 110]. As a result of
the concerted action of PINK1 and Parkin, damaged
mitochondria are coated with p-Ser65-Ub chains.
While the detection of endogenous, activated PINK1