Miro phosphorylation sites regulate Parkin recruitmentand mitochondrial motilityEvgeny Shlevkova,b, Tal Kramera,b, Jason Schapanskyc, Matthew J. LaVoiec, and Thomas L. Schwarza,b,1

aF. M. Kirby Neurobiology Center, Boston Children’s Hospital, Boston, MA 02115; bDepartment of Neurobiology, Harvard Medical School, HarvardUniversity, Cambridge, MA 02138; and cAnn Romney Center for Neurological Diseases, Brigham and Women’s Hospital and Harvard Medical School,Boston, MA 02115

Edited by Don W. Cleveland, University of California, San Diego, La Jolla, CA, and approved August 23, 2016 (received for review July 30, 2016)

The PTEN-induced putative kinase 1 (PINK1)/Parkin pathway can tagdamaged mitochondria and trigger their degradation by mitophagy.Before the onset of mitophagy, the pathway blocks mitochondrialmotility by causing Miro degradation. PINK1 activates Parkin byphosphorylating both Parkin and ubiquitin. PINK1, however, has othermitochondrial substrates, including Miro (also called RhoT1 and -2),although the significance of those substrates is less clear. We showthat mimicking PINK1 phosphorylation of Miro on S156 promoted theinteraction of Parkin with Miro, stimulated Miro ubiquitination anddegradation, recruited Parkin to the mitochondria, and via Parkinarrested axonal transport of mitochondria. Although Miro S156Epromoted Parkin recruitment it was insufficient to trigger mitophagyin the absence of broader PINK1 action. In contrast, mimickingphosphorylation of Miro on T298/T299 inhibited PINK1-induced Miroubiquitination, Parkin recruitment, and Parkin-dependent mitochon-drial arrest. The effects of the T298E/T299E phosphomimetic weredominant over S156E substitution. We propose that the status of Mirophosphorylation influences the decision to undergo Parkin-dependentmitochondrial arrest, which, in the context of PINK1 action on othersubstrates, can restrict mitochondrial dynamics before mitophagy.

PINK1 | Parkin | Miro | mitochondrial transport | mitophagy

Parkinson’s disease (PD) is the second most common neurode-generative disorder, and is closely linked to mitochondrial dys-

function (1, 2). Two hereditary forms of recessive PD are caused bymutations in PINK1 (PTEN-induced putative kinase 1), a Ser/Thrmitochondrial kinase, and Parkin, a cytosolic E3 ubiquitin ligase (3,4). The realization that these proteins are in a single pathway, withPINK1 acting upstream of Parkin to influence mitochondrial prop-erties, was a critical step in uncovering the underlying pathologicalmechanisms of PD (5–7). This pathway can trigger the selectiveautophagy of damaged mitochondria, termed mitophagy (8, 9), butadditional cellular functions have also been indicated for PINK1 andParkin (10–15). Much, however, remains unclear about how thePINK1/Parkin pathway is regulated.In current models of PINK1/Parkin mitophagy (reviewed in ref. 1),

healthy mitochondria import a PINK1 precursor constitutively tothe inner membrane, where it is cleaved (16–18). The cleaved formthen returns to the cytoplasm and is degraded by the N-end rulepathway (19). Mitochondrial depolarization, or protein misfolding inthe matrix of energized mitochondria (20), prevent the import anddegradation of PINK1, resulting in the accumulation of PINK1 onthe outer mitochondrial membrane (OMM) (9, 21, 22). Once on theOMM, PINK1 kinase activity recruits Parkin from the cytosol (8, 9).Although Parkin adopts a self-inhibited conformation in solution(23–25), it becomes fully activated in a PINK1-dependent manner onthe mitochondria (9, 21). Parkin ubiquitinates numerous proteins ofthe OMM (26, 27), and thereby recruits autophagy-related proteinsto the damagedmitochondrion for autophagosome assembly (28–30).How PINK1 recruits and activates Parkin on mitochondria re-

mains incompletely understood, but two key components have beenidentified: PINK1 phosphorylation of both Parkin and ubiquitin (31–34). Together, these actions form a positive-feedback loop in whichPINK1 activates Parkin by phosphorylating Serine 65. Activated

Parkin adds ubiquitin to outer membrane-localized proteins at themitochondrial surface, providing more substrates for PINK1 and theresulting phosphoubiquitin is an allosteric activator of Parkin, fur-ther increasing its activity. Recruitment and activation of Parkin,although distinct processes, are thus tightly linked (35, 36) by thepositive feed-forward effect of ubiquitination by Parkin causing ad-ditional binding sites for Parkin, and additional binding causing fur-ther ubiquitination and allosteric activation of Parkin. Parkinsubstrates can also be de-ubiquitinated by Usp30 (37) andUsp15 (38).PINK1, however, can phosphorylate other proteins, including the

motor-adaptor protein Miro (also called RhoT1/2), Mitofusin 1/2,and Hsp75 (also known as Trap1) (15, 39–42), and the functionalrole of those modifications in regulating Parkin is unclear. Given thatsome of these proteins are also substrates of Parkin (15, 39), it ispossible that modification of these proteins can act as a layer ofregulation to modulate the overall levels of Parkin on mitochondriaor to modulate specific aspects of mitochondrial dynamics, such asmotility and fusion. Miro, an OMM GTPase involved in the regu-lation of mitochondrial traffic, is a well-established substrate ofParkin (15, 27, 43, 44). Data from our laboratory and others indi-cated that PINK1 can interact with Miro and cause Miro phos-phorylation of S156 in vitro (15, 45). PINK1 and Parkin-mediatedmodifications of Miro result in the proteasomal degradation of Miroand, because Miro is required to tether kinesin and dynein to themitochondrial surface, its degradation arrests mitochondrial move-ment. Genetic analysis indicated that, as in the mitophagic pathway,PINK1 acts upstream of Parkin in regulating motility (15). The arrestof mitochondrial movement may be a precursor to mitophagy or,independent of mitophagy, provide a means of regulating mito-chondrial dynamics. In Drosophila axons, for example, knockdown of

Significance

In mitophagy, damaged mitochondria stabilize PTEN-inducedputative kinase 1 (PINK1) and recruit Parkin, an E3-ligase thatubiquitinates proteins on the outer membrane and targetsmitochondria for degradation. The crucial roles of PINK1 phos-phorylation of Parkin and ubiquitin in mitophagy are well-established. Other substrates of PINK1, however, have also beenreported but the significance of those phosphorylations is lessclear. We now show that Miro phosphorylations can regulateParkin recruitment to Miro and trigger Miro degradation. Theconsequence of this branch of the PINK1/Parkin pathway is thedisruption of mitochondrial motility, an event that may spatiallyrestrict the deleterious effects of mitochondrial damage prior tothe mitophagic removal of the organelle.

Author contributions: E.S. and T.L.S. designed research; E.S. and T.K. performed research;J.S. and M.J.L. contributed new reagents/analytic tools; E.S. and T.L.S. analyzed data; andE.S. and T.L.S. wrote the paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.1To whom correspondence should be addressed. Email: Thomas.Schwarz@childrens.harvard.edu.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1612283113/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1612283113 PNAS | Published online September 27, 2016 | E6097–E6106

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

either PINK1 or Parkin increased mitochondrial transport in theabsence of acute damage to the organelles (15). Moreover, Mirophosphorylations are needed for the survival of dopaminergic neu-rons and the proper development of the neuromuscular junction inDrosophila (46). Miro is thus a good candidate in which to explorethe functional roles of PINK1 phosphorylations of Parkin substrates.Previous experiments indicated that PINK1 can phosphorylate

Miro in vitro not only on S156 but also on T298/299 (15). Wegenerated phosphomimetic and nonphosphorylatable Miro mu-tants in all three sites. Mimicking phosphorylation on S156 stim-ulated the interaction of Miro with Parkin, promoted the recruit-ment of Parkin to the mitochondria, and induced mitochondrialarrest in axons, but was not sufficient alone to trigger mitophagy.Mimicking T298/T299 phosphorylation suppressed the effects ofthe phosphomimetic S156E. We conclude that Miro phosphoryla-tions on S156 and T298/T299 can regulate the onset of Parkinsignaling by modulating the levels of Parkin on mitochondria andcan regulate mitochondrial motility.

ResultsPhosphomimetic Mutant MiroS156E Enhances the Recruitment ofParkin to Mitochondria in a PINK1-Independent Manner. Mass spec-trometry of recombinant Drosophila Miro phosphorylated in vitroby human PINK1 immunoprecipitated from HEK293T cells pre-viously revealed two phospho-peptides, corresponding to aminoacids 177–190 and 319–330 (15). The homologous peptides inhuman Miro1 (hMiro1) contained the conserved sites S156, T298,and T299. To establish the likely consequences of the S156 mod-ifications at the cellular level, we asked whether expressing thephosphomimetic MiroS156E influenced Parkin recruitment tomitochondria and its consequent arrest of mitochondrial motility.The phosphomimetic substitution per se did not detectably disruptMiro function. Myc-MiroS156E coprecipitated with Xpress-taggedMilton equivalently to Myc-Miro when expressed in HEK293T(Fig. S1A). In addition, when expressed in neurons, the MiroS156Econstruct correctly localized to axonal mitochondria (Fig. S1B).Because overexpression of Miro constructs probably does not dis-place the endogenous Miro but increases Miro levels on mito-chondria, we used overexpressed wild-type Miro as a control forphosphorylation-independent effects of Miro. If Miro phosphory-lation can recruit and activate Parkin, we anticipated that thephosphomimetic Miro would have a dominant effect over the en-dogenous Miro. To determine if the S156E mutation alteredParkin distribution, YFP-Parkin was cotransfected into primary ratembryonic fibroblasts with either mock DNA, Myc-Miro, or Myc-MiroS156E and the abundance of YFP-Parkin in cytoplasmicand mitochondrial compartments was evaluated automatically ina blinded fashion using Mitolyzer1.0, a script we developed forImageJ (Materials and Methods) (Fig. 1 A and B). In the absence ofMiro overexpression, YFP-Parkin was not noticeably concentratedon mitochondria. Miro and MiroS156E isoforms were expressed atequivalent levels judged by α-Myc immunofluorescence levels (Fig.S1C) and highly localized to mitochondria (Fig. 1A). Althoughexpression of Myc-Miro increased slightly the levels of YFP-Parkinon mitochondria, MiroS156E caused a significantly greater mi-tochondrial accumulation of Parkin (Fig. 1 A and B), and con-sequently a higher ratio of Parkin to Miro on mitochondria (Fig.S1C). This effect cannot be attributed to cross-talk betweenfluorophores because the accumulation of YFP-Parkin is alsoapparent in the absence of the second fluorophore (Fig. S1D). Thus,mimicking phosphorylation of Miro on S156 can recruit Parkin tothe mitochondria. To observe with higher detail the recruitment ofParkin to the mitochondria by MiroS156E, we collected stacks ofconfocal images of a MiroS156E, YFP-Parkin–expressing fibroblast,and performed a 3D reconstruction. We also performed a line scanof the same region from a confocal image (Fig. S1E). Parkin-YFPsignal colocalizes with α-Tom20 signal in the reconstruction and inthe line-scan. Because Parkin recruitment to mitochondria is thought

to depend on its activation by PINK1 (35), we also examined theeffects of MiroS156E expression in primary embryonic fibroblastsfrom PINK1−/− rats (47). MiroS156E caused equivalent mitochon-drial localization of YFP-Parkin in these cells (Fig. 1B and Fig. S2).Thus, the phosphomimetic form of Miro was capable of recruitingParkin to mitochondria even in the absence of PINK1.

Phosphomimetic Mutant MiroS156E Enhances MitochondrialFragmentation. The mitochondrial network also appeared frag-mented in cells where MiroS156E was coexpressed with YFP-Parkin(Fig. 1A). To quantify this phenotype, we incorporated a particleanalysis feature to Mitolyzer1.0 and determined the numbers of small(0.2–5 μm in diameter) and rounded (circularity ∼0.5–1) mitochondriain the fibroblasts. The average number of small rounded mitochondriaper cell was ∼50 in control cells, but almost doubled with MiroS156Eexpression and this effect was equally apparent in PINK1−/− fibroblasts(Fig. 1C). Fragmentation also increased significantly when Miro wasexpressed, although to a lesser extent (∼30% of control). This frag-mentation was dependent on Parkin activity because it did not occurwhen we substituted YFP-ParkinC431F, a catalytically dead form ofParkin that is not recruited to mitochondria (36). YFP-ParkinC431Fwas not significantly recruited to mitochondria by MiroS156E and themitochondrial network was not fragmented by their coexpression (Fig.1 A, D, and E). Thus, Myc-MiroS156E, and to a lesser extent Myc-Miro overexpression, promoted Parkin recruitment to the mitochon-dria and induced mitochondrial fragmentation as a consequence ofthe catalytic activity of Parkin, and these effects of MiroS156E did notrequire additional activation of Parkin by PINK1.

Phosphomimetic MiroS156E Can Arrest Axonal Transport of Mitochondria.The PINK1/Parkin pathway, by triggering the degradation of Miro,halts the movement of axonal mitochondria (15). To determine ifS156E similarly altered mitochondrial behavior in neurons, wecoexpressed Mito-DsRed to label mitochondria, synaptophysin-GFP (Syp-GFP) to label axons, and either Myc-MiroWT or Myc-MiroS156E in either wild-type or Parkin KO mouse hippocampalneurons (48). We then imaged live neurons by time-lapse micros-copy and calculated the average time mitochondria spent in motionin each condition. Although expression of Miro did not signifi-cantly affect mitochondrial motility, expression of MiroS156E di-minished the percentage of mitochondria that were moving ineither direction (Fig. 1 F and G and Table S1). Mitochondrialarrest induced by MiroS156E was specific to mitochondria, as themovement of Syp-GFP vesicles was not altered (Fig. 1G and TableS1). Moreover, in Parkin−/− neurons, expression of MiroS156Efailed to induce mitochondrial arrest (Fig. 1 F and G and TableS1). Thus, mimicking PINK1 phosphorylation on S156 inducedmitochondrial arrest in a Parkin-dependent manner, similar to theeffect of PINK1 activation or overexpression and not through anonspecific disruption of the transport apparatus.

Phosphomimetic MiroS156E Does Not Induce Mitophagy. When mi-tochondrial damage causes PINK1 stabilization on the OMM,Parkin-recruitment to mitochondria and mitochondrial frag-mentation are a prelude to mitophagy (8). To assess if mimickingphosphorylation on S156 induced mitophagy, we expressed Myc-Miro, either wild-type or S156E, along with Parkin in fibro-blasts and examined three mitophagy markers: mitochondrialcontent, mitochondrial α- ubiquitin staining, and colocalizationof LC-3 with mitochondria (Fig. 2). We also quantified the levelsof Parkin recruitment to the mitochondria. We used cells onlyexpressing Miro as a negative control, and cells that overexpressedParkin and were treated with carbonyl cyanide 3-chlorophenyl-hydrazone (CCCP) as a positive control. As expected, in CCCP-treated cells, mitochondrial content, as judged by α-Tom20 immu-noreactivity, was drastically reduced (Fig. 2C), and α-ubiquitinimmunoreactivity, LC-3, and Parkin colocalization with mi-tochondria increased (Fig. 2 B and D–F) (49). In contrast,

E6098 | www.pnas.org/cgi/doi/10.1073/pnas.1612283113 Shlevkov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

Fig. 1. Cellular effects upon mimicking Miro phosphorylation on S156. (A) Rat embryonic fibroblasts were transfected with the indicated constructs andstained with α-Tom20 to label mitochondria and α-Myc to label Myc-Miro. YFP fluorescence was used to estimate Parkin levels and localization. (B) Theaverage median gray value of YFP-Parkin in PINK1+/+ (gray bars) and PINK1−/− (white bars) cells in the cytoplasmic and the mitochondrial compartmentsplotted for each condition. Overexpression of Myc-MiroS156E, and to a lesser extent Myc-Miro, increased YFP intensity exclusively in the mitochondrialcompartment. (C) Average number of small (0.2–5 μm in diameter) and rounded (circularity ∼0.5–1) mitochondria in cells expressing either GFP or YFP-Parkinin PINK1+/+ and PINK1−/− genetic backgrounds. (D and E) YFP intensity in cytoplasmic and mitochondrial compartments (D) and average number of mito-chondrial fragments (E) in cells coexpressing MiroS156E with either YFP-ParkinWT or YFP-ParkinC431F. n = 40–50 cells for each condition from three inde-pendent transfections per genotype. (F) Mitochondrial movement in representative axons transfected with mito-DsRed along with indicated constructs anddissected from Parkin+/+ (Left) or Parkin−/− (Right) animals. The first frame of each time-lapse imaging series is shown above a kymograph generated from themovie. The x axis represents mitochondrial position, and the y axis is time (moving from top to bottom). Vertical lines are stationary mitochondria, whereasdiagonal lines represent movement. (G) Average time spent in motion by each mitochondrion (mito-DsRed) or synaptic vesicle (syp-GFP) in axons of indicatedgenotypes. n = 100–130 mitochondria and n = 81–236 synaptic vesicles from nine axons/genotype and three independent biological replicates per genotype.(Scale bars, 10 μm.) *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. See also Table S1.

Shlevkov et al. PNAS | Published online September 27, 2016 | E6099

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

although Myc-MiroS156E coexpression with Parkin increasedParkin and α-ubiquitin colocalization with mitochondria relativeto Myc-MiroWT coexpression with Parkin, the S156E mutationdid not cause LC-3 recruitment. Miro coexpression with Parkincaused a modest decrease in mitochondrial content, but therewas no significant further effect of the S156E mutation. Theextent of mitochondrial ubiquitination induced by the S156Emutation presumably reflects Parkin activity, but at levels sig-nificantly lower than those induced by CCCP (Fig. 2 A and D).Thus, Parkin recruitment and activation by MiroS156E, althoughsufficient to induce mitochondrial fragmentation, was inadequateto trigger the full pathway for mitophagy; PINK1 activity on othersubstrates, such as ubiquitin and Parkin itself, is probably neededfor mitophagy to proceed.

Miro Phosphomimetic S156E Promotes the Interaction of Parkin withMiro. To explore the mechanism underlying MiroS156E-inducedParkin recruitment, we turned to established cell lines. Whenwild-type Miro (Myc-Miro) was coexpressed with Parkin andHA-ubiquitin in HeLa cells and the PINK1/Parkin pathway wasactivated by CCCP, Miro immunoreactivity appeared in highermolecular-mass bands than that of 75 kDa unmodified Myc-Miro(Fig. S3A). When we immunoprecipitated Myc-Miro, the highmolecular-mass bands were positive for both α-Myc and α-HA–

ubiquitin. Thus, Parkin-induced ubiquitination of Myc-Miro canbe detected as α-Myc+ bands migrating more slowly than 75 kDa.The apparent synergism of CCCP, an activator of PINK1, andParkin-expression in promoting Miro ubiquitination could bebecause of PINK1 phosphorylation of either Parkin or Miro orboth. Therefore, to examine the contribution of phosphorylationof MiroS156 we compared phosphoresistant Myc-MiroS156A(where Serine 156 was changed to Alanine) to Myc-Miro. Severalgroups have failed to detect differences in Myc-MiroS156Aubiquitination levels compared with Myc-Miro (44, 50). In agree-ment with these groups, we also see that, under some circum-stances, Myc-MiroS156A high molecular-mass bands can beinduced by Parkin (Fig. S3B). Our results indicate that in condi-tions where Parkin is strongly activated, the status of S156 is oflesser importance. However, if levels of Parkin activity are mod-erate to low, as in the following experiments, the phosphorylationstatus of S156 can determine the extent of Miro ubiquitination anddegradation. After coexpression of either Myc-Miro or S156A witheither PINK1 or Parkin in HEK cells, the Myc-Miro band hadsignificantly less immunoreactivity compared with Myc-MiroS156A(Fig. 3 A and B), suggesting that phosphorylation on S156 indeedfacilitates Miro degradation. The efficacy of the phosphoresistantmutation in also preserving Miro levels in the absence of PINK1overexpression or activation with CCCP suggests that this site may

Fig. 2. MiroS156E does not induce mitophagy. (A) Ratfibroblasts were transfected with the indicated con-structs and stained with α-Tom20 (red) and α-ubiquitin(blue). YFP fluorescence was used to estimate Parkinlevels and localization. (B) Fibroblasts were transfectedas in A, but m-Cherry Parkin was used instead of YFP-Parkin (false-colored in green for consistency) andautophagic vesicles were labeled by transfection ofLC-3–GFP, (false-colored in blue). (C–F) Quantificationof mitochondrial content (expressed as average area ofTom-20 immunoreactivity), α-ubiquitin immunoreac-tivity on mitochondria (expressed as average meangray value from 16-bit images) from cells transfectedas in A, the number of LC-3–GFP vesicles coextensivewith mitochondria from cells transfected as in B, andthe intensity of YFP-Parkin on mitochondria from 8-bitimages of cells transfected as in A. CCCP was added ata final concentration of 20 μM for 3 h before fixation.n = 60–70 cells per genotype from three independenttransfections. (Scale bars, 10 μm.) Statistical compari-sons are to Miro-only cells, except where other-wise indicated. *P < 0.05, **P < 0.01, ***P < 0.001,**** P < 0.0001.

E6100 | www.pnas.org/cgi/doi/10.1073/pnas.1612283113 Shlevkov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

A

B

F

G

K

L

H I

J

C

D

E

Fig. 3. MiroS156 promotes the interaction of Parkin with Miro. (A and B) Transfection of HEK293T cells with either 1 μg of DNA for PINK1-FLAG (A) or 1 μg ofDNA for YFP-Parkin (B) caused preferential degradation of Myc-Miro compared with Myc-MiroS156A (expressed equivalently from 2 μg of DNA). Dashed linesin B indicate the place where the order of the lanes on the gel was altered for clarity of presentation. (C and D) HEK293T cells expressing 0.1 μg of either Myc-Miro or Myc-MiroS156A for 24 h were pretreated with 5 μM MG132 for 12 h and with 10 μM CCCP for 2 h before lysis. CCCP caused preferential proteasomaldegradation of Myc-Miro compared with Myc-MiroS156A. A representative immunoblot (C) and quantification from three independent replicates (D) areshown. (E) α-Myc–Miro immunoprecipitates from cells expressing YFP-Parkin and HA-ubiquitin show greater α-HA–ubiquitin immunoreactivity on Myc-Mirothan Myc-MiroS156A. (F and G) α-Myc immunoprecipitates from lysates of HEK293T cells transfected with 1 μg of Myc-Miro and 2 μg of YFP-Parkin wereincubated with CIP (30 U) for 30 min. α-GFP staining of the immunoprecipitates showed less intensity of YFP-Parkin in the CIP-treated sample. A representativeexample (F) and quantification (G) from three independent replicates are shown. (H–J) Cells transfected with 1 μg of Myc-Miro or MycMiroS156A and 2 μg ofYFP-Parkin as indicated. YFP-Parkin expression caused a ladder of high molecular mass species immunoreactive for α-Myc. This ladder was more abundant incells expressing Myc-Miro than Myc-MiroS156A, as quantified in I, after normalization to actin. Upon immunoprecipitation with α-Myc, YFP-Parkin was moreabundant in Myc-Miro than in Myc-MiroS156A-expressing cells, as quantified in J after normalization to the 75 kDa myc-immunoreactive band. n = 3. (K) Theindicated purified recombinant proteins were combined and incubated for 30 min at 37 °C before immunoprecipitation for 2 h with α-MBP antibody.(L) Quantification of immunoprecipitated dMiro (anti-His immunoreactivity) from four experiments as in K. Background-subtraction and normalization toMBP levels as in J. *P < 0.05, **P < 0.01, **** P < 0.0001.

Shlevkov et al. PNAS | Published online September 27, 2016 | E6101

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

undergo phosphorylation even in their absence. If CCCP was usedto activate the endogenous PINK1/Parkin pathway, the influenceof the S156 site was also apparent. When we transfected low levelsof Myc-Miro or Myc-MiroS156A in HEK293T and treated thecells with CCCP for 1 h before lysis, Myc-Miro underwent moreproteasome-dependent degradation than Myc-MiroS156A (Fig. 3C and D). To test the significance of S156 for ubiquitination ofMiro, either wild-type or MiroS156A was expressed togetherwith Parkin and HA-ubiquitin in HEK293T cells and Miro wasimmunoprecipitated with α-Myc. Myc-MiroS156A had less anti–HA-ubiquitin immunoreactivity than Myc-Miro (Fig. 3E). We thenasked if phosphorylation stimulated the interaction of Miro withParkin. We coexpressed Miro and Parkin, immunoprecipitatedMiro with α-Myc, and then treated the immunoprecipitate with calfalkaline phosphatase (CIP). The interaction of Miro with Parkinwas significantly weaker upon the treatment with the phosphatase(Fig. 3 F and G), Thus, Miro–Parkin interaction is promoted byphosphorylations. To test if the interaction depended on Mirophosphorylated on S156, we coexpressed either Myc-Miro or Myc-MiroS156A with YFP-Parkin in HEK293T cells and examined theamount of Parkin coprecipitated with Miro. Less YFP-Parkincoprecipitated with Myc-MiroS156 than with Myc-Miro (Fig. 3 Gand H), and Myc-MiroS156A also had significantly less of theslowly migrating and presumably ubiquitinated bands (Fig. 3 H andI). Thus, the interactions of Parkin with Miro in HEK293T cellsdepend, at least in part, on phosphorylation of Miro on S156.The effects of the S156A mutation suggested that phosphoryla-

tion of this site promotes Parkin–Miro interactions. To test thishypothesis directly, we affinity-purified recombinant human Parkintagged with maltose binding protein (MBP) (Fig. S3C), which iscatalytically active (Fig. S3D), and incubated it with wild-type andphosphomimetic forms of bacterially expressed Miro. Becauserecombinant Drosophila Miro can be successfully expressed and pu-rified, we used Drosophila Miro with glutamate or aspartate substi-tutions at S182 (dMiroS182E or -D), the site equivalent tomammalian S156. After incubation of the proteins for 30 min at37 °C, we immunoprecipitated MBP-Parkin and evaluated the levelsof dMiro present in the precipitate. The interaction was significantlygreater for dMiroS182E and dMiro182D than for dMiroWT (Fig. 3K and L and Fig. S3E). Collectively, these lines of evidence suggestedthat Miro phosphorylation on S156 promotes the ubiquitination anddegradation of Miro by stimulating its interaction with Parkin.

Miro T298E and T299E Phosphomimetics Render Miro Resistant toParkin-Induced Ubiquitination and Degradation. To determine theconsequences of phosphorylation of T298 and T299, we replacedboth residues with a phosphomimetic glutamate (Miro298/9EE)mutant. We also generated a triple Miro phosphomimetic mu-tant (MiroEEE = MiroS156E,T298E,T299E). The mutations didnot affect the ability of Miro to localize to mitochondria or bindto Milton (Fig. S1). Although we anticipated that coexpression ofthe mutants with YFP-Parkin would promote the slower-migratingand presumably ubiquitinated bands of Miro similar to S156E,Miro298/9EE, and MiroEEE, mutants were resistant to the modi-fication (Fig. 4 A and B), even at high ratios of Parkin:Mirotransfection (Fig. 4 C and D). Moreover, Parkin expression condi-tions that caused wild-type Miro levels to be significantly reduceddid not reduce levels of MiroEEE (Fig. 4 E and F). We confirmedthat the slower-migrating bands of MiroWT and S156E were im-munoreactive for HA-ubiquitin but no HA-ubiquitin was observedin the immunoprecipitate when MiroEEE was coexpressed withParkin (Fig. S4). Thus, mimicking phosphorylation on T298 andT299 likely suppresses the ubiquitination and degradation of Miroby Parkin and does so even when S156 also has been replaced bya glutamate. However, PINK1 activation of Parkin upon depolar-ization of mitochondria with CCCP could override the effect ofT298/9EE; Miro levels were reduced and the slower-migrating Mirobands appeared, consistent with ubiqutination (Fig. 4G). The effect

of Miro phosphomimetic mutations thus depends on the level ofParkin activation.

MiroT298E,T299E Inhibits Parkin Effects on Mitochondrial Dynamics.The partial resistance of MiroEEE mutants to degradation incell lines suggested that 298/9EE might also suppress the effectsof Parkin on mitochondrial dynamics. In primary rat fibroblasts,Miro298/9EE prevented S156E from recruiting Parkin; the triplephosphomimetic induced neither mitochondrial accumulation ofParkin nor fragmentation (Fig. 5 A–C). Consistent with thesefindings, the 298/9EE mutant also suppressed the effect of S156Eon mitochondrial movement in axons; neither Myc-Miro298/9EEnor EEE arrested their transport (Fig. 5 D and E and Table S2).To test the effects of 298/9EE on mitochondrial dynamics when

endogenous PINK1 was activated, we transfected rat hippocampalneurons with either with Myc-Miro or Myc-Miro298/9EE and ap-plied 10 μM antimycin A to activate the PINK1/Parkin pathway(51). Antimycin A caused mitochondrial arrest in both populations(Fig. 5G), although a low amount of axonal mitochondrial move-ment was visible in Miro298/9EE-expressing neurons at late time-points (Fig. 5 F and G and Table S3) that was significantly higherthan in control neurons expressing Myc-Miro. This experimentindicated that even upon depolarization, the presence of the 298/9EE mutation can modestly delay the onset of Parkin activity withregard to Miro. Therefore, we concluded that the presence of aphosphate-mimicking modification on T298 and T299 can, via thesuppression of Miro ubiquitination and degradation, negativelyregulate the ability of Parkin to regulate mitochondrial dynamics.

DiscussionThe physiological role of PINK1 modification of Parkin sub-strates has been unclear. We report here that: (i) mimicking phos-phorylation of the Parkin substrateMiro at S156 can recruit Parkin tomitochondria; (ii) Parkin can be activated by phosphomimetic Miroin a PINK1-null background; (iii) substrate-based activation ofParkin can cause mitochondrial fragmentation and mitochondrialarrest, but will not proceed to mitophagy without PINK1 activationof Parkin; and (iv) the effect of the modification of S156 is reversibleby phosphomimetic substitutions on two other Miro residues.Solving how PINK1 recruits Parkin to damaged mitochondria

will be instrumental for the rational design of Parkin-modulatorytherapies. Besides modifying ubiquitin and Parkin, PINK1 phos-phorylates several known Parkin substrates on the mitochondrialouter membrane. Phosphorylation-dependent recruitment of E3ligases to their substrates is a common regulatory theme (52), andwe propose that Parkin is no exception to this pattern. To cir-cumvent PINK1 phosphorylation of Parkin and isolate the conse-quences of the substrate phosphorylation, we used phosphomimeticsubstitutions either in wild-type fibroblasts where PINK1 was notactivated pharmacologically or in rat embryonic fibroblasts lackingPINK1, where even background activity of the kinase would belacking. In either case, mimicking PINK1 phosphorylation of Miroon S156 promoted Parkin recruitment to mitochondria in fibro-blasts (Fig. 1 A and B and Fig. S2). In vitro, mimicking this modi-fication enhanced the interaction of Miro with Parkin (Fig. 3 K andL and Fig. S3E). Thus, the phosphorylation state of a Parkin sub-strate can stimulate Parkin interaction with that substrate and wassufficient to recruit Parkin to the mitochondria. In neurons, the celltype most affected by pathological mutations in PINK1 and Parkin,mimicking S156 phosphorylation was sufficient to cause Parkin-dependent arrest of mitochondria (Fig. 1 F and G), indicating thatthe S156E modification can stimulate endogenous Parkin. Substratephosphorylation has also been reported to cause Mitofusin1/2degradation upon Parkin recruitment (41), and thus may influ-ence a variety of Parkin targets. Because Parkin-mediated degra-dation of Miro and Mitofusin1/2 happens before bulk clearance ofmitochondrial proteins by mitophagy (15), selective phosphorylation-promoted recruitment of Parkin to particular high-priority targets

E6102 | www.pnas.org/cgi/doi/10.1073/pnas.1612283113 Shlevkov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

may play a role in the early steps of Parkin recruitment, whereasubiquitin phosphorylation is instrumental in driving the subsequent

feed-forward loop that broadly ubiquitinates many proteins at themitochondrial surface and induces mitophagy.

A

C

B

D

E G

F

Fig. 4. MiroT298/9EE is resistant to Parkin-mediated ubiquitination and degradation. (A and B) HEK293T cells cotransfected with Myc-Miro mutant constructs andYFP-Parkin DNA as indicated. Intensities of the ∼75-kDa band and bands higher than ∼75 kDa in size were normalized to α-ATPase5β intensity. Bar graphs representthe mean normalized intensity from four independent transfections. (C and D) Increasing YFP-Parkin DNA concentrations were cotransfected with either Myc-Miroor Myc-Miro298/9EE and quantified as in B. (E and F) Increasing amounts of YFP-Parkin was coexpressed with the indicated constructs in HEK293T cells. Bar graphsrepresent α-Myc intensities in four independent transfections. (G) HeLa cells transfected with the indicated constructs and treated with 40 μM CCCP for 3 h beforelysis. Dashes indicate places where the ordering of the lanes on the gel was altered for clarity of presentation. *P < 0.05, **P < 0.01.

Shlevkov et al. PNAS | Published online September 27, 2016 | E6103

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

Fig. 5. MiroT298/9EE suppresses S156E-induced effects on mitochondrial dynamics. (A) Rat embryonic fibroblasts transfected with the indicated constructs and stainedwith α-Tom20 (red) and α-Myc (blue). (B) Average median gray value of YFP signal and (C) average number of small, α-Tom20+ mitochondrial particles per cell in cellstransfected with the indicated constructs. n = 40–50 cells from three biological replicates. (D) Kymographs from distal axons of mouse hippocampal neurons transfectedwith mito-DsRed and the indicated constructs. (E) Average time each mitochondrion (mito-DsRed) spent in motion in axons transfected with the indicated constructs andaverage time each synaptic vesicle (syp-GFP) spent in motion in the same axons. n = 125–149 mitochondria and 171–240 synaptic vesicles/axon from nine axons and threebiological replicates (see also Table S2). (F) Representative kymographs of mitochondrial motility in axons transfected with the indicated constructs and imaged after35 min incubation with 10 μM Antimycin A. (G) Average time spent in motion by mitochondria at different times after application of 10 μM Antimycin A. n > 100mitochondria and time points from nine axons and four biological replicates per time point. (Scale bars, 10 μm.) *P < 0.05, **P < 0.01, ****P < 0.0001. See also Table S3.

E6104 | www.pnas.org/cgi/doi/10.1073/pnas.1612283113 Shlevkov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

What role, if any, does PINK1 modification of Parkin substratesplay in regulating the catalytic activity of Parkin? Parkin is a self-inactivated enzyme whose activity is stimulated by two PINK1phosphorylations: (i) phosphorylation of S65 in the ubiquitin-likedomain of Parkin and (ii) phosphorylation of an equivalent serine inubiquitin, which can allosterically activate Parkin. Our data indicatethat PINK1 phosphorylation of Miro also can activate Parkin, evenin the absence of the other two events. Overexpression of wild-typeMiro itself could cause Parkin to translocate to mitochondria, butthe effect was enhanced by the phosphomimetic mutation (Fig. 1 Aand B). The action of the phosphorylation, therefore, is less like anon/off switch than like an increase in the affinity for Parkin, andhence an increase in its ability to activate Parkin. Because Parkincatalytic activity is needed for Parkin to remain on mitochondriaafter mitochondrial depolarization (36, 53), Parkin recruitment tomitochondria can be used as a proxy for Parkin activity. Consistentwith activation of Parkin by the MiroS156E phosphomimetic, wild-type Parkin, but not catalytically inactive Parkin, accumulated onmitochondria. Further evidence for catalytic activation of Parkin byMiro S156E was the fragmentation of the mitochondrial networkupon its coexpression with Parkin in fibroblasts, a phenotype thatalso did not occur when the catalytically dead Parkin was expressed(Fig. 1 A and B). Importantly, PINK1 was not needed in theseexperiments, as we observed the same phenotypes in PINK1-nullcells (Fig. S2). Thus, PINK1 phosphorylations of ubiquitin or Parkinare not absolutely required for Parkin activation. It will be in-teresting to know if substrate-driven activation is specific to Miro orcan extend to other Parkin substrates as well.Although PINK1 phosphorylation of S156 is supported both by

previous data from our laboratory and the functional experimentspresented here, other groups have failed to observe a role of PINK1-induced S156 phosphorylation in the ubiqutination of Miro (44, 50).One reason for the disparity may lie in the high levels of Parkin usedin these experiments. In agreement with those studies, we also findthat nonphosphorylatable Miro gets ubiquitinated by Parkin in thoseconditions (Fig. S3B). However, in those experiments where onlyendogenous Parkin was present, we observe a clear effect of theS156A mutation (Fig. 3 A, and C and D). Therefore, althoughPINK1-mediated S156 phosphorylation is not required for Mirodegradation, the functional effect of mutating this site can berevealed when Parkin activity is low to moderate. The in vivo sig-nificance of the phosphorylation site was observed inDrosophila (46)and is also supported by our previous observations on mitochondrialmotility in hippocampal neurons (15).Once Parkin has been recruited to Miro, it could have access to

other mitochondrial substrates. The fragmentation of mitochondriawe observed in MiroS156E could, for example, reflect the sec-ondary loss of proteins that support fusion (39, 41, 54), although itmight also be because of the loss of kinesin from mitochondriawhen Miro is degraded (55). However, MiroS156E recruitment ofParkin did not induce detectable mitophagy, at least in the timeframe that we analyzed (Fig. 2). Therefore, PINK1 modification ofthis single substrate, although able to recruit Parkin to Miro andeven to activate Parkin for Miro ubiquitination, was not sufficientto induce the widespread ubiquitination of proteins on the mito-chondrial surface, and thereby cause the recruitment of adaptorproteins and mitophagy (29, 30, 36, 56, 57). When PINK1 is acti-vated in the context of mitochondrial damage, it will act on mul-tiple targets, including Parkin and ubiquitin, and thereby achievefull enzymatic activity of Parkin and trigger mitophagy.The possibility of limited activation of the PINK1/Parkin

pathway offers a potential mechanism to separate mitochondrialarrest from mitophagy: some PINK1 and Parkin substrates, in-cluding Miro and Mitofusin1/2 (26, 39), may undergo proteaso-mal degradation rather than autophagic degradation because oftheir special relationship with PINK1 and Parkin. We do not knowwhether the degradation of Miro in mammalian cells under normalphysiological conditions is always an early “pro-clearance” step in a

series that will ultimately produce mitophagy, or if there are levels ofPINK1 activation that will stop the process with Miro degradationand a decrease in mitochondrial motility. In Drosophila, however,alterations in mitochondrial motility can be seen with manipulationsof PINK1 and Parkin that do not appear to involve widespreadmitophagy (15, 46, 58). If Miro degradation is a step toward clear-ance of either a segment of the OMM or mitophagy of the entireorganelle (8, 59), it is likely to be important, along with Mitofusin1/2degradation, as a means to quarantine damaged mitochondria rap-idly before the slower steps of autophagosome-dependent mitophagyor mitochondria-derived vesicle-based quality control.To our surprise, phosphomimetic modifications on T298 and

T299 acted in an opposite manner to S156. The physiologicalsignificance of these sites is currently less clear. There is un-certainty at present concerning the kinase that acts at these sitesin vivo; it may not be PINK1. However, the unexpected findingthat phophomimetics at these residues can prevent the effects ofS156 phosphorylation raises the possibility that they are part of anegative regulatory pathway requiring further exploration.Better understanding of the regulation of Parkin by PINK1 may

lead to rational strategies for treating Parkinson’s disease. Althoughmany studies have focused on Parkin and ubiquitin phosphorylationby PINK1, our findings highlight the need to consider also the rolethat PINK1 substrates like Miro play in the onset of Parkin activity.The PINK1/Parkin mechanism now appears more complex and caninvolve proteasomal degradation of specific substrates as well as ly-sosomal clearance of either all or part of a mitochondrion. BecauseParkin can be activated and recruited or inhibited by substratephosphorylations, the state of particular substrates may influence theend results of the pathway.

Materials and MethodsReagents. Myc-Miro mutants were generated by mutagenizing Myc-Miroplasmid (60) with a QuikChange Site Directed Mutagenesis Kit (Agilent).YFP-ParkinC431F mutant was generated as above using YFP-Parkin (8) asbackbone. All mutants were sequence-verified. Sequences used for muta-genesis are in SI Materials and Methods. For antibody and chemical reagentsinformation, see SI Materials and Methods.

Cell Culture and Transfection. DNA transfections in HEK293T cells were per-formedwith the calciumphosphatemethod and Lipofectamine2000 (ThermoFisherScientific) or GenJet (SignaGen Laboratories) were used for transfections in HeLacells. Immunocytochemistry was done on fibroblasts of Long Evans wild-type orLong Evans or PINK1−/− rats (47). Rat and mouse hippocampal neurons were iso-lated according to standard procedures and cultured as in ref. 61. Rat and mouseprocedures were approved by the Institutional Animal Care Committee at theBoston Children’s Hospital and by the Harvard HMA Standing Committee onAnimals, respectively. See SI Materials and Methods for details.

Image Acquisition and Image Analysis. Live-cell imaging of axonal mitochondria wasperformed as described in ref. 61 (see also SI Materials and Methods). For fixedsamples, cells were fixed in 2.5% (vol/vol) PFA 24 h after transfection, permeabilizedin 0.5% (wt/vol) saponin, blocked in 1% (wt/vol) BSA, 0.5% (wt/vol) saponin, andincubated overnight at 4 °C with primary antibodies. Confocal images were takenon a Zeiss LSM700 confocal microscope using 63× Zeiss plan-Apochromat oil,1.4 NA or 100× Zeiss plan-Apochromat oil, 1.3 NA objectives. To analyze YFP-Parkinand ubiquitin intensity as well as mitochondrial morphology, images were pro-cessed in batch using a custom ImageJ script (SI Materials and Methods).

In Vitro Experiments, Immunoprecipitations, and Western Blotting. dMiro constructswere mutagenized using standard procedures and then expressed in bacteria,purified, and stored as described in ref. 62. MBP-Parkin was purified according toref. 63. MBP-Parkin was used within 2 wk after purification for in vitro reactions.Validation of MBP-Parkin activity was carried out following the protocols de-scribed in ref. 63. See SI Materials and Methods for details.

Statistical Analysis. All statistical analyses were performed using GraphPad Prismv6.0. All samples were first subjected to a D’Angostino–Pearson omnibus normalitytest. If values came from a Gaussian distribution, t-test analysis was used for pairedcomparisons and one-way ANOVA along with Bonferroni correction for multiplecomparisons. If values came from a non-Gaussian distribution, a Mann–Whitney U

Shlevkov et al. PNAS | Published online September 27, 2016 | E6105

CELL

BIOLO

GY

PNASPL

US

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020

test was used for paired comparisons and Kruskall–Wallis nonparametric ANOVAtest was used for multiple comparisons. Graph bars represent the mean ± SEMP values: *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

ACKNOWLEDGMENTS. We gratefully acknowledge the gift of plasmid con-structs from Drs. P. Aspenstrom for myc-hMiro1/2; G. Hajnoczky for Mito-DsRed;R. Youle for Parkin constructs; D. Selkoe for PINK1-FLAG; Sheila Thomas for GFP–LC-3; Noriyuki Matsuda forMBP-Parkin; G.W. Hart for Xpress-hMilton1; H. T. Clinefor syp-GFP; and S. Rice for dMiro proteins. We thank S. Vasquez for assistancewith primary hippocampal neuron cultures; K. Apaydin for assistance with neuron

cultures and protein purification; Dr. K. Kapur for help with statistical analyses;A. Jackson for the help coding Mitolyzer1.0; Drs. G. Pekkurnaz and M. Cronin forhelp with the initial experiments and for fruitful discussions; the Intellectual andDevelopmental Disabilities Research Center (IDDRC) Molecular Genetics and Im-aging Cores and L. Mkhitarian for general help; and Drs. C. Gutierrez, E. Martinez-Salas, and P. Gutierrez-Martinez for critically reading the manuscript. This workwas funded by NIH Grants P30 HD018655 (to the IDDRC Imaging, Proteomics,and Molecular Genetics Cores) and NS065013 (to M.J.L); PDF-FBS-1320 from theParkinson’s Disease Foundation (to E.S.); and the Mathers Foundation and NIHGrant R01GM069808 (to T.L.S.).

1. Pickrell AM, Youle RJ (2015) The roles of PINK1, parkin, and mitochondrial fidelity inParkinson’s disease. Neuron 85(2):257–273.

2. Exner N, Lutz AK, Haass C, Winklhofer KF (2012) Mitochondrial dysfunction inParkinson’s disease: Molecular mechanisms and pathophysiological consequences.EMBO J 31(14):3038–3062.

3. Kitada T, et al. (1998) Mutations in the Parkin gene cause autosomal recessive juvenileparkinsonism. Nature 392(6676):605–608.

4. Valente EM, et al. (2004) Hereditary early-onset Parkinson’s disease caused by mu-tations in PINK1. Science 304(5674):1158–1160.

5. Poole AC, et al. (2008) The PINK1/Parkin pathway regulates mitochondrial morphol-ogy. Proc Natl Acad Sci USA 105(5):1638–1643.

6. Clark IE, et al. (2006) Drosophila pink1 is required for mitochondrial function andinteracts genetically with Parkin. Nature 441(7097):1162–1166.

7. Park J, et al. (2006) Mitochondrial dysfunction in Drosophila PINK1 mutants iscomplemented by Parkin. Nature 441(7097):1157–1161.

8. Narendra D, Tanaka A, Suen DF, Youle RJ (2008) Parkin is recruited selectively toimpaired mitochondria and promotes their autophagy. J Cell Biol 183(5):795–803.

9. Narendra DP, et al. (2010) PINK1 is selectively stabilized on impaired mitochondria toactivate Parkin. PLoS Biol 8(1):e1000298.

10. Morais VA, et al. (2014) PINK1 loss-of-function mutations affect mitochondrial com-plex I activity via NdufA10 ubiquinone uncoupling. Science 344(6180):203–207.

11. Morais VA, et al. (2009) Parkinson’s disease mutations in PINK1 result in decreasedcomplex I activity and deficient synaptic function. EMBO Mol Med 1(2):99–111.

12. Berger AK, et al. (2009) Parkin selectively alters the intrinsic threshold for mito-chondrial cytochrome c release. Hum Mol Genet 18(22):4317–4328.

13. Henn IH, et al. (2007) Parkin mediates neuroprotection through activation of IkappaBkinase/nuclear factor-kappaB signaling. J Neurosci 27(8):1868–1878.

14. Müller-Rischart AK, et al. (2013) The E3 ligase Parkin maintains mitochondrial in-tegrity by increasing linear ubiquitination of NEMO. Mol Cell 49(5):908–921.

15. Wang X, et al. (2011) PINK1 and Parkin target Miro for phosphorylation and degra-dation to arrest mitochondrial motility. Cell 147(4):893–906.

16. Jin SM, et al. (2010) Mitochondrial membrane potential regulates PINK1 import andproteolytic destabilization by PARL. J Cell Biol 191(5):933–942.

17. Meissner C, Lorenz H, Weihofen A, Selkoe DJ, Lemberg MK (2011) The mitochondrialintramembrane protease PARL cleaves human Pink1 to regulate Pink1 trafficking.J Neurochem 117(5):856–867.

18. Greene AW, et al. (2012) Mitochondrial processing peptidase regulates PINK1processing, import and Parkin recruitment. EMBO Rep 13(4):378–385.

19. Yamano K, Youle RJ (2013) PINK1 is degraded through the N-end rule pathway.Autophagy 9(11):1758–1769.

20. Jin SM, Youle RJ (2013) The accumulation of misfolded proteins in the mitochondrialmatrix is sensed by PINK1 to induce PARK2/Parkin-mediated mitophagy of polarizedmitochondria. Autophagy 9(11):1750–1757.

21. Matsuda N, et al. (2010) PINK1 stabilized by mitochondrial depolarization recruitsParkin to damaged mitochondria and activates latent Parkin for mitophagy. J Cell Biol189(2):211–221.

22. Vives-Bauza C, et al. (2010) PINK1-dependent recruitment of Parkin to mitochondriain mitophagy. Proc Natl Acad Sci USA 107(1):378–383.

23. Riley BE, et al. (2013) Structure and function of Parkin E3 ubiquitin ligase revealsaspects of RING and HECT ligases. Nat Commun 4:1982.

24. Trempe JF, et al. (2013) Structure of Parkin reveals mechanisms for ubiquitin ligaseactivation. Science 340(6139):1451–1455.

25. Wauer T, Komander D (2013) Structure of the human Parkin ligase domain in anautoinhibited state. EMBO J 32(15):2099–2112.

26. Chan NC, et al. (2011) Broad activation of the ubiquitin-proteasome system by Parkinis critical for mitophagy. Hum Mol Genet 20(9):1726–1737.

27. Sarraf SA, et al. (2013) Landscape of the PARKIN-dependent ubiquitylome in responseto mitochondrial depolarization. Nature 496(7445):372–376.

28. Itakura E, Kishi-Itakura C, Koyama-Honda I, Mizushima N (2012) Structures containingAtg9A and the ULK1 complex independently target depolarized mitochondria atinitial stages of Parkin-mediated mitophagy. J Cell Sci 125(Pt 6):1488–1499.

29. Yamano K, Fogel AI, Wang C, van der Bliek AM, Youle RJ (2014) Mitochondrial RabGAPs govern autophagosome biogenesis during mitophagy. eLife 3:e01612.

30. Wong YC, Holzbaur EL (2014) Optineurin is an autophagy receptor for damagedmitochondria in Parkin-mediated mitophagy that is disrupted by an ALS-linked mu-tation. Proc Natl Acad Sci USA 111(42):E4439–E4448.

31. Kane LA, et al. (2014) PINK1 phosphorylates ubiquitin to activate Parkin E3 ubiquitinligase activity. J Cell Biol 205(2):143–153.

32. Kazlauskaite A, et al. (2014) Parkin is activated by PINK1-dependent phosphorylationof ubiquitin at Ser65. Biochem J 460(1):127–139.

33. Koyano F, et al. (2014) Ubiquitin is phosphorylated by PINK1 to activate parkin.Nature 510(7503):162–166.

34. Ordureau A, et al. (2014) Quantitative proteomics reveal a feedforward mechanism formitochondrial PARKIN translocation and ubiquitin chain synthesis. Mol Cell 56(3):360–375.

35. Lazarou M, Jin SM, Kane LA, Youle RJ (2012) Role of PINK1 binding to the TOMcomplex and alternate intracellular membranes in recruitment and activation of theE3 ligase Parkin. Dev Cell 22(2):320–333.

36. Lazarou M, et al. (2013) PINK1 drives Parkin self-association and HECT-like E3 activityupstream of mitochondrial binding. J Cell Biol 200(2):163–172.

37. Bingol B, et al. (2014) Themitochondrial deubiquitinase USP30 opposes parkin-mediatedmitophagy. Nature 510(7505):370–375.

38. Cornelissen T, et al. (2014) The deubiquitinase USP15 antagonizes Parkin-mediatedmitochondrial ubiquitination and mitophagy. Hum Mol Genet 23(19):5227–5242.

39. Tanaka A, et al. (2010) Proteasome and p97 mediate mitophagy and degradation ofmitofusins induced by Parkin. J Cell Biol 191(7):1367–1380.

40. Pridgeon JW, Olzmann JA, Chin LS, Li L (2007) PINK1 protects against oxidative stressby phosphorylating mitochondrial chaperone TRAP1. PLoS Biol 5(7):e172.

41. Chen Y, Dorn GW, 2nd (2013) PINK1-phosphorylated mitofusin 2 is a Parkin receptorfor culling damaged mitochondria. Science 340(6131):471–475.

42. Lai YC, et al. (2015) Phosphoproteomic screening identifies Rab GTPases as noveldownstream targets of PINK1. EMBO J 34(22):2840–2861.

43. Kazlauskaite A, et al. (2014) Phosphorylation of Parkin at Serine65 is essential foractivation: elaboration of a Miro1 substrate-based assay of Parkin E3 ligase activity.Open Biol 4(3):130213.

44. Liu S, et al. (2012) Parkinson’s disease-associated kinase PINK1 regulates Miro proteinlevel and axonal transport of mitochondria. PLoS Genet 8(3):e1002537.

45. Weihofen A, Thomas KJ, Ostaszewski BL, Cookson MR, Selkoe DJ (2009) Pink1 forms amultiprotein complex with Miro and Milton, linking Pink1 function to mitochondrialtrafficking. Biochemistry 48(9):2045–2052.

46. Tsai P-I, et al. (2014) PINK1-mediated phosphorylation of Miro inhibits synapticgrowth and protects dopaminergic neurons in Drosophila. Sci Rep 4:6962.

47. Baptista MA, et al. (2013) A strategy for the generation, characterization and distri-bution of animal models by The Michael J. Fox Foundation for Parkinson’s Research.Dis Model Mech 6(6):1316–1324.

48. Goldberg MS, et al. (2003) Parkin-deficient mice exhibit nigrostriatal deficits but notloss of dopaminergic neurons. J Biol Chem 278(44):43628–43635.

49. Narendra D, Kane LA, Hauser DN, Fearnley IM, Youle RJ (2010) p62/SQSTM1 is re-quired for Parkin-induced mitochondrial clustering but not mitophagy; VDAC1 isdispensable for both. Autophagy 6(8):1090–1106.

50. Birsa N, et al. (2014) Lysine 27 ubiquitination of the mitochondrial transport protein Miro isdependent on serine 65 of the Parkin ubiquitin ligase. J Biol Chem 289(21):14569–14582.

51. Ashrafi G, Schlehe JS, LaVoie MJ, Schwarz TL (2014) Mitophagy of damaged mito-chondria occurs locally in distal neuronal axons and requires PINK1 and Parkin. J CellBiol 206(5):655–670.

52. Hunter T (2007) The age of crosstalk: Phosphorylation, ubiquitination, and beyond.Mol Cell 28(5):730–738.

53. Zheng X, Hunter T (2013) Parkin mitochondrial translocation is achieved through anovel catalytic activity coupled mechanism. Cell Res 23(7):886–897.

54. Misko A, Jiang S, Wegorzewska I, Milbrandt J, Baloh RH (2010) Mitofusin 2 is nec-essary for transport of axonal mitochondria and interacts with the Miro/Miltoncomplex. J Neurosci 30(12):4232–4240.

55. Wang C, et al. (2015) Dynamic tubulation of mitochondria drives mitochondrialnetwork formation. Cell Res 25(10):1108–1120.

56. Heo JM, Ordureau A, Paulo JA, Rinehart J, Harper JW (2015) The PINK1-PARKIN mi-tochondrial ubiquitylation pathway drives a program of OPTN/NDP52 recruitmentand TBK1 activation to promote mitophagy. Mol Cell 60(1):7–20.

57. Lazarou M, et al. (2015) The ubiquitin kinase PINK1 recruits autophagy receptors toinduce mitophagy. Nature 524(7565):309–314.

58. Devireddy S, Liu A, Lampe T, Hollenbeck PJ (2015) The organization of mitochondrialquality control and life cycle in the nervous system in vivo in the absence of PINK1.J Neurosci 35(25):9391–9401.

59. McLelland GL, Soubannier V, Chen CX, McBride HM, Fon EA (2014) Parkin and PINK1function in a vesicular trafficking pathway regulating mitochondrial quality control.EMBO J 33(4):282–295.

60. Fransson A, Ruusala A, Aspenström P (2003) Atypical Rho GTPases have roles in mi-tochondrial homeostasis and apoptosis. J Biol Chem 278(8):6495–6502.

61. Pekkurnaz G, Trinidad JC, Wang X, Kong D, Schwarz TL (2014) Glucose regulates mito-chondrial motility via Milton modification by O-GlcNAc transferase. Cell 158(1):54–68.

62. Klosowiak JL, et al. (2013) Structural coupling of the EF hand and C-terminal GTPasedomains in the mitochondrial protein Miro. EMBO Rep 14(11):968–974.

63. Matsuda N, et al. (2006) Diverse effects of pathogenic mutations of Parkin that cat-alyze multiple monoubiquitylation in vitro. J Biol Chem 281(6):3204–3209.

E6106 | www.pnas.org/cgi/doi/10.1073/pnas.1612283113 Shlevkov et al.

Dow

nloa

ded

by g

uest

on

Mar

ch 2

3, 2

020