Research ArticleCELL BIOLOGY

AMPK/ULK1-mediated phosphorylation of Parkin ACT domain mediates an early step in mitophagy

See allHide authors and affiliations

Science Advances  07 Apr 2021:
Vol. 7, no. 15, eabg4544
DOI: 10.1126/sciadv.abg4544


The serine/threonine kinase ULK1 mediates autophagy initiation in response to various cellular stresses, and genetic deletion of ULK1 leads to accumulation of damaged mitochondria. Here we identify Parkin, the core ubiquitin ligase in mitophagy, and PARK2 gene product mutated in familial Parkinson’s disease, as a ULK1 substrate. Recent studies uncovered a nine residue (“ACT”) domain important for Parkin activation, and we demonstrate that AMPK-dependent ULK1 rapidly phosphorylates conserved serine108 in the ACT domain in response to mitochondrial stress. Phosphorylation of Parkin Ser108 occurs maximally within five minutes of mitochondrial damage, unlike activation of PINK1 and TBK1, which is observed thirty to sixty minutes later. Mutation of the ULK1 phosphorylation sites in Parkin, genetic AMPK or ULK1 depletion, or pharmacologic ULK1 inhibition, all lead to delays in Parkin activation and defects in assays of Parkin function and downstream mitophagy events. These findings reveal an unexpected first step in the mitophagy cascade.


The timely sequestration and turnover of damaged mitochondria following mitochondrial poisons is a coordinated biochemical response, which results in mitochondrial fragmentation and ultimately degradation of the mitochondria via the cellular process of mitophagy. One of the core components of a highly conserved mitophagy pathway is Parkin, the gene product of the E3 ubiquitin ligase gene PARK2, which is the most frequent cause of early onset Parkinson’s disease (13). Studies of Parkin function first revealed its translocation from the cytosol to mitochondria following mitochondrial uncouplers like carbonyl cyanide m-chlorophenyl hydrazone (CCCP). Upon arrival at the mitochondria, Parkin becomes phosphorylated on Ser65 in its ubiquitin-like domain (UBL) by the protein kinase PINK1, whose levels are stabilized on the outer mitochondrial membrane (OMM) following mitochondrial damage (13). PINK1 also phosphorylates ubiquitin on a homologous Ser65 residue, which is required for timely mitophagy induction and for recruitment of LIR-domain containing mitophagy receptors (46). Activation of Parkin is well established to involve direct phosphorylation by PINK1 and the direct binding of phosphorylated ubiquitin to Parkin, resulting in additional conformation changes that release it from a proposed autoinhibited conformation (1). Recent hydrogen-deuterium exchange mass spectrometry studies in vitro recapitulating PINK1- and ubiquitin-dependent activation of Parkin unexpectedly revealed a previously uncharacterized conserved domain downstream of the UBL domain that was termed the activation (ACT) element and speculated to play a role in an unknown aspect of Parkin regulation (7).

Besides PINK1-Parkin, another cellular stress-sensing pathway that plays broad roles in mitochondrial homeostasis is the adenosine 5′-monophosphate (AMP)–activated protein kinase (AMPK) pathway (8). AMPK is rapidly activated following any stress that triggers a lowering of the intracellular adenosine 5′-triphosphate (ATP)/adenosine 5′-diphosphate (ADP) ratio, including all mitochondrial electron transport chain (ETC) inhibitors. Within 30 min of treatment of cells with ETC inhibitors, mitochondria rapidly fragment via AMPK induction of MFF-Drp1–mediated fission (9) and general macroautophagy is initiated via AMPK phosphorylation of ULK1 and other components of the upstream autophagy cascade (9). Genetic ULK1 deficiency or AMPK deficiency has been shown to lead to accumulation of defective mitochondria in mouse embryonic fibroblasts (MEFs) and erythrocytes (1012). However, little is known to directly connect ULK1 to established regulators of mitochondrial turnover.


To identify previously undiscovered substrates of ULK1 kinase activity, we generated an optimal ULK1 phosphorylation consensus motif, which uncovered multiple ULK1-dependent phosphorylation sites in several core autophagy pathway components, including Atg101, Beclin1, and Vps34, as described previously (13). Given the prominent defects in mitochondria associated with ULK1 deficiency, we also examined the phosphorylation of core components of the Parkin-PINK1 mitophagy complex in response to exogenous or endogenous ULK1 activation. We observed a profound electrophoretic mobility shift of yellow fluorescent protein (YFP)–tagged human Parkin on standard SDS–polyacrylamide gel electrophoresis (PAGE) gels when coexpressed with Myc-tagged human ULK1 [wild-type (WT) ULK1] (Fig. 1A) while V5-tagged PINK1 was only modestly affected (fig. S1A). This mobility shift was dependent on ULK1 kinase activity, as it was not observed following coexpression with Myc-ULK1–rendered kinase-inactive (KI ULK1) by a single point mutation, K46I. Furthermore, the Parkin band shift induced by catalytically active WT ULK1 was diminished by treatment of cell lysates with lambda-phosphatase, as was the mobility of ULK1 itself (Fig. 1B). Comparing phosphorylation sites on Parkin coexpressed with WT versus KI ULK1 using mass spectrometry revealed multiple phosphorylation sites specific to conditions when ULK1 was active (fig. S1B). However, many of these phosphorylation sites could be secondary events that are induced downstream of other kinases or phosphatases regulated by ULK1 and thus are not necessarily direct substrates of ULK1 itself in vivo. Of the phosphorylation sites specific to active ULK1 conditions, only Parkin Ser108 and Ser110 conform to the ULK1 substrate motif identified by degenerate peptide library analysis and validated in a number of previously identified ULK1 substrates (Fig. 1C). The Parkin Ser108-110 residues are well conserved in higher eukaryotes, with Ser108 specifically being conserved back to Caenorhabditis elegans (Fig. 1D). Phosphorylation of Ser108 in endogenous mouse Parkin was previously reported in a mouse tissue-specific phosphoproteomic analysis, with phosphorylation of this region identified uniquely in brown fat tissue (14). Although phosphorylation of Ser108 was identified in this phosphoproteomic analysis, it was not remarked on nor was the kinase responsible for mediating this phosphorylation identified. Regardless, this study reveals that physiological phosphorylation of this region in mouse Parkin occurs within a metabolic tissue characterized by a high density of uncoupled mitochondria. Because our mass spectrometry results in human Parkin indicated that any of the serine residues 108, 109, and 110 may be phosphorylated, we generated constructs mutating these sites and all the other phosphorylation sites identified by mass spectrometry, to nonphosphorylatable alanines, regardless of their lack of conformation to the ULK1 motif. Both S108A and S109/S110AA mutants reduced the ULK1-dependent band shift of Parkin, but the complete loss of band shift was only observed in the S108/109/110AAA (“SA3”) mutant (Fig. 1E and fig. S1C). To probe the regulation of Parkin S108-110 phosphorylation, we developed a phospho-specific antibody designed to detect phosphorylated Parkin Ser108 (the most evolutionarily conserved site) and thus subsequently referred to as “pParkin S108.” The pParkin S108 antibody detected robust signal by Western blotting in the presence of WT ULK1, which was not induced by KI ULK1, and mutation of the 108 to 110 serines to alanines caused the WT ULK1–induced signal to be entirely lost (Fig. 1F). Neither serine-to-alanine mutation of the Ser65 PINK1 phosphorylation site in Parkin nor mutagenesis of the commonly used representative Parkinson’s disease mutation Lys161Asn affected the WT ULK1–induced phosphorylation of Parkin Ser108 (Fig. 1G), indicating no dependency of these molecular events on the ability of ULK1 to phosphorylate Parkin. To validate that ULK1 itself can directly mediate phosphorylation of Parkin, we performed in vitro kinase assays combining immunoprecipitated Myc-tagged Parkin with recombinant ULK1 protein and ATP. In these in vitro assays, the Parkin S108 phospho-antibody detected phosphorylation of Parkin, similar to phosphorylation of the validated ULK1 substrate VPS34 (fig. S1D).

Fig. 1 Parkin Ser108 is regulated by AMPK/ULK1 signaling.

(A) Immunoblots of Parkin cotransfected with myc-tagged wild-type (WT) or kinase-inactive (KI) ULK1 in human embryonic kidney (HEK) 293T cells. (B) Immunoblots as in (A), but here, cell lysates were treated with lambda-phosphatase (PPase) as indicated. (C) Parkin Ser108 and Ser110 aligned with known ULK1 phosphorylation sites. Residues conforming to optimal ULK1 substrate motif highlighted. (D) ClustalW alignment of evolutionary conservation surrounding Parkin Ser108. (E) Immunoblots of indicated Parkin and ULK1 cDNAs cotransfected into HEK293T cells. Phosphorylation monitored by Parkin mobility shift. (F and G) As in (E) but monitored with Parkin phospho-ser108 antibody. (H) Immunoblots of cells transfected with Parkin and treated for 1 hour with vehicle (D), 20 μM CCCP (C), 10 μM 991, 10 μM 6965 ULK1 inhibitor (65), or 6965 combined with CCCP or 991 (65/C: 65/991). (I) Immunoblots of liver lysates from control or inducible knockout of AMPKa1/a2 (AMPK DKO) mice treated with vehicle or AMPK activator MK8722 for 2 hours. See Materials and Methods for details. (J) Immunoblots as in (I) but control or inducible KO of Ulk1/Ulk2. (K) Immunoblots of lysates from primary mouse hepatocytes from WT or AMPK liver DKO mice treated for 1 hour with vehicle or metformin.

Having established that ULK1 can phosphorylate Parkin in vitro and in cells with overexpressing ULK1, we next sought to determine whether phosphorylation of Parkin Ser108 occurs under conditions in which Parkin is known to be activated and whether this involves the induction of endogenous ULK1 kinase activity. The mitochondrial uncoupler CCCP is widely used to induce Parkin activity, and CCCP induced phosphorylation of Parkin Ser108 in human embryonic kidney (HEK) 293T cells, within 1 hour of treatment (Fig. 1H). Phosphorylation of ULK1 Ser555 by AMPK, which is known to stimulate ULK1 (10, 15), is also increased by CCCP treatment. Treatment of cells with compound 991 (hereafter, 991), a small molecule that directly binds and activates AMPK independent of effects on mitochondria or cellular energy levels (16), was also sufficient to fully induce Parkin Ser108 phosphorylation (Fig. 1H). The phosphorylation event of Parkin Ser108 induced by CCCP and 911 can be attenuated by a small-molecule ULK1/2 inhibitor, SBI-0206965 (“65”), further reinforcing the observation that these events are mediated by ULK1 or ULK2 kinase activity (13). It is worth noting that no substrates unique to ULK1 or ULK2 are known currently, and they behave genetically redundant in mice and most tissues examined to date, thus to inhibit “ULK” activity in cells in which both are expressed one needs to inhibit both ULK1 and ULK2, which SBI-0206965 does (13).

We also sought to examine regulation of Parkin Ser108 relative to other known stimuli that activate Parkin and/or AMPK-ULK1. Unlike PINK1, which requires mitochondrial depolarization for its activation, AMPK and ULK1 can be activated fully by mitochondrial ETC inhibitors, which have no impact on mitochondrial membrane potential (17). We observed here that the complex I inhibitors phenformin and rotenone potently activate the AMPK/ULK1 pathway and trigger phosphorylation of Parkin Ser108 to the same extent as CCCP (fig. S2A). INK128, a catalytic kinase inhibitor of mechanistic target of rapamycin (mTOR) and a potent autophagy inducer, was not sufficient to trigger phosphorylation of Parkin Ser108, suggesting a selective response of pParkin S108 to mitochondrial insults (fig. S2A). Consistent with previous observations that depolarization is required for Parkin catalytic activation, the ubiquitylation of the Parkin substrate CISD1/MitoNEET was observed following treatment with the mitochondrial depolarizing agents CCCP and valinomycin (4, 18) but not with the complex I inhibitors (rotenone and phenformin) or direct AMPK activator 991, despite their ability to comparably activate AMPK and ULK1 and all induce Ser108 phosphorylation equivalently (fig. S2A). Transient introduction of YFP-Parkin followed by Ulk1 or control small interfering RNA and treatment with these stimuli revealed that ULK1 was required for phosphorylation of Parkin Ser108 after these stimuli (fig. S2B). In immunofluorescence experiments with cell stably expressing YFP-Parkin, even prolonged treatment (6 hours) with phenformin, rotenone, or 991 was insufficient to induce Parkin translocation to the mitochondria, in contrast to treatment with CCCP or valinomycin, which did induce translocation of Parkin to the mitochondria (fig. S2C). Similarly, only CCCP, valinomycin, or the combination of antimycin and oligomycin, but not phenformin, rotenone, 991, or INK128, was capable of inducing Parkin ubiquitin ligase activity in thioester conjugation assay using C431S mutation in Parkin (fig. S2D) (18).

We next examined phosphorylation of endogenous Parkin S108 in liver lysates following treatment of mice with MK-8722 (Fig 1I), an orally available compound 991 analog that robustly induces endogenous AMPK activation (19). Endogenous Parkin pS108 was readily detected in mouse liver after MK-8722 treatment in WT mice but not AMPKα1/α2 (Prkaa1l/l; Prkaa2l/l) liver-specific knockout (KO) mice (Fig. 1I), which have no ULK1 Ser555 phosphorylation in the absence of AMPK expression. Similarly, endogenous Parkin pS108 was attenuated in genetically disrupted Ulk1/Ulk2 double KO livers (Fig. 1J). The phosphorylation of endogenous Parkin on the S108 site can be also observed in WT but not AMPK KO primary hepatocytes after metformin stimulation, which is widely prescribed type 2 diabetes drug and is well known to activate AMPK in vivo (Fig. 1K). We also confirmed that metformin induces endogenous P-Ser108 of Parkin in the mouse liver under the more pathologically relevant conditions of C57/Bl6 mice on a high-fat diet (fig. S2E). To summarize, we found Parkin as a novel ULK1 substrate in vitro and in vivo, and these phosphorylation events on endogenous Parkin occur in response to mitochondrial stress or synthetic AMPK activation.

A recent study reported detailed analysis of in vitro conformation changes in Parkin during its activation by PINK1 and the binding of phospho-ubiquitin, revealing a here-to-fore undiscovered additional domain that lay in between the UBL domain and the UPB domain, termed the ACT element (Fig. 2A) (7). Previously undetected in earlier crystal structures, this region was determined to be juxtaposed against the UBL domain and was subject to repositioning during Parkin activation in vitro. Within this larger 30–amino acid domain, a deletion of only 9 amino acids from amino acids 101 to 109 prevented the ability of Parkin to be activated properly in vitro (7). Hence, we sought to examine phosphorylation of Ser108 within this short ACT region relative to other known Parkin modification during Parkin activation following mitochondrial damage. We performed an acute time course after antimycin A and oligomycin A (AO) in HEK293T cells and observed robust Parkin Ser108 phosphorylation between 2 and 5 min after AO that remained high with diminishment starting ~30 min after AO, coincident with an increase of PINK1-induced phosphorylation of Parkin S65 (Fig. 2B). Thirty minutes after AO, TBK1 activation at Ser172 also became apparent and evidence of Parkin substrate (CISD1) ubiquitylation and LC3 maturation was also observed (Fig. 2B). At 60 min after AO, PINK1-phosphorylation of ubiquitin Ser65 was increasingly apparent and continued to increase at later time points. In contrast to PINK1 and TBK1 activation at 30 min and after, but similar to P-Parkin Ser108, AMPK appeared fully activated at 2 min after AO, as visualized by endogenous phosphorylation of four downstream substrates (ULK1 Ser555, Raptor Ser792, MFF Ser146, and ACC Ser79). However, unlike Parkin Ser108, the phosphorylation of AMPK and its three downstream substrates remained comparably induced between 2 min and 60 min after AO, all only partly attenuating at 120 min. Endogenous phosphorylation of two other ULK1 substrates, Beclin Ser30 (13) and ATG16L1 Ser278 (20), were observed to parallel P-Parkin Ser108: induced within 2 min and diminishing after 30 min. Analogous kinetics of the different phosphorylation events described above were observed when CCCP was used to initiate mitochondria damages (Fig. 2C).

Fig. 2 Parkin Ser108phosphorylation is rapidly induced following AO and CCCP and precedes activation of PINK1 and TBK1.

(A) Schematic of overall Parkin domain structure with inset of ClustalW alignment of residues 59 to 115 of human Parkin. The evolutionary conservation of the PINK1 site Ser65 and ULK1 site Ser108 in Parkin across vertebrates is shown in red. Ser108 is one of only two fully conserved serines in the recently described ACT element (7) involved in Parkin activation (residues 101 to 110). At right, structural model of Ser108 in ACT element shown in red, Arg104 shown in green is mutated in a familial Parkinson’s patient (R104W). (B) Immunoblots of HEK293T cells stably expressing YFP-Parkin was subjected to AO (2.5 μM antimycin A and 5 μM oligomycin) treatment for the times indicated. (C) Immunoblots of HEK293T cells stably expressing YFP-Parkin were subjected to 10 μM CCCP treatment for the times indicated.

These data demonstrate unambiguously that phosphorylation of Parkin Ser108 is extremely rapid and evidently precedes all prior noted biochemical steps in the mitophagy cascade. To examine what may be triggering AMPK and ULK1 activation at 2 min, we examined the ADP/ATP ratio, a conventional measure of ATP loss that triggers AMPK activation, as well as mitochondrial reactive oxygen species (mtROS) levels, which is also hypothesized to activate AMPK in some contexts (21), though recent reports suggest that the effects of mtROS may be indirect and primarily via effects on ATP loss (22, 23). Escalating doses of CCCP induced notable changes in the ADP/ATP ratio and changes in mtROS, both detectable starting at 2 min (fig. S3, A and B).

PINK1, as a mitochondrial membrane-bound protein, exclusively phosphorylates Parkin that is recruited to the OMM (1). A question provoked by the rapid ULK1-induced phosphorylation that we observed was whether the pool of Parkin being phosphorylated by ULK1 may also be localized to the OMM or occurs exclusively in the cytosol. To examine this, we used a time course of cytoplasmic—mitochondrial fractionation on cells bearing WT or SA3 Parkin following CCCP treatment. Again, in the WT Parkin–expressing cells, phosphorylation of Parkin Ser108 was observed at 2 min after CCCP, most robustly in the cytoplasmic fraction, while we also noticed a small fraction of pParkin-S108 signals appeared on mitochondria roughly after 10 min, which was proportional to the fraction of total Parkin at the mitochondria (Fig. 3A). Furthermore, Parkin Ser65 phosphorylation is observed in the mitochondrial fraction in the WT Parkin cells, but only at later (30 to 60 min) time points, consistent with the signaling observed in total cell lysates (Fig. 2C). P-Parkin S65 was significantly diminished in the mitochondrial fraction in the SA3 mutant cells, indicating a reduction of active Parkin on mitochondria compared to the control cells.

Fig. 3 Parkin is phosphorylated at Ser108 in the cytosol and its timely translocation to mitochondria requires Ser108-Ser110.

(A) Immunoblots of cytoplasmic and mitochondrial extracts isolated from YFP-Parkin-WT and YFP-Parkin-SA3 expressing HEK293T cells by cell membrane disruption and differential centrifugation, after treatment with 20 μM CCCP for the indicated time. Cytoplasmic proteins (30 μg) and mitochondria proteins (20 μg) per lane were analyzed. (B) Immunoblots of HEK293T cells transiently transfected with YFP-Parkin or YFP-Parkin-SA3 or YFP-Parkin-R104W then treated with 20 μM CCCP for the times indicated. MG132 (10 μM) was added to prevented proteasome-mediated degradation. (C) Quantification of Parkin mitochondrial translocation phenotype by immunofluorescence. Individual cells from the images in (D) were analyzed for Person correlation coefficient (PCC) analysis by ImageJ software. Data are shown in a violin plot. n = 19 for YFP-Parkin-WT; n = 24 for YFP-Parkin-SA3, **P < 0.01, ****P < 0.0001 by two-way analysis of variance (ANOVA). (D) Time-lapse live-cell images of U2OS cells stably expressing WT or SA3 YFP-Parkin treated with dimethyl sulfoxide (DMSO) or CCCP (20 μM) for 150 min. YFP-Parkin is shown as green, mitochondria were stained with MitoTracker (red). Clustered mitochondria with translocated YFP-Parkin are indicated with white arrowheads. Scale bars, 20 μm.

A familial Parkinson’s disease mutation R104W lies within this same amino acid stretch (the aforementioned ACT region), and the adjacent V105 is a hydrophobic amino acid critical for recognition of the ULK1 phosphorylation motif (13), residing three residues upstream of the candidate target site (Fig. 2A, left-hand inset). Treatment of cells expressing R104W mutant Parkin with CCCP revealed a complete inability of this patient mutant to be phosphorylated at Ser108 (Fig. 3B), and Parkin Ser65 phosphorylation by PINK1 was partially affected by R104W, similar to SA3 Parkin. To rule out the possibility that the R104W mutation simply disrupts the antibody recognition, we deliberately compared its ability to be phosphorylated at the Ser108 site with ULK1 overexpression or in vitro or when dependent on endogenous ULK1 activation by upstream stimuli. The results showed that the R104W Parkin can still be phosphorylated and detected by the phospho-specific antibody when ULK1 is overly abundant (fig. S4A). This is also true when we tested this by ULK1 in vitro kinase assay (fig. S4B). However, we observed that R104W is not phosphorylated at S108 under endogenous ULK1 activation conditions following 991 (fig. S4A) or CCCP treatment (Fig. 3B). Moreover, both SA3 and R104W mutants displayed parallel defects in PINK1 phosphorylation of Parkin S65 and Parkin ubiquitylation of its substrate (CISD1) (Fig. 3B and fig. S4C), strongly suggesting ULK1-mediated phosphorylation within the ACT region could play a crucial role in normal Parkin function and that the R104W patient mutation is compromised for Ser108 phosphorylation in cells following mitochondrial damage.

Having observed that the abundance of SA3 Parkin on mitochondria is reduced compared to WT Parkin by biochemical methods, we next examined a series of time points following CCCP using time-lapse video microscopy to visualize the extent of Parkin association with the mitochondria, quantified in Fig. 3C. We observed the association of WT YFP-Parkin with mitochondria within 60 min, which by 90 min had developed into foci of Parkin completely colocalized with condensed mitochondria (Fig. 3D). These foci continued to develop robustly, as shown at 120 and 150 min. In contrast, SA3-YFP-Parkin displayed delayed association with mitochondria, with translocation largely absent until 90 min, and incomplete development of foci even at 150 min. It is worth noting that the WT Parkin/mitochondria foci tended to be clustered together at one side of the cells (polarized) in contrast to much dispersed pattern with the SA3 Parkin, which may have implications in functional discrepancy to be examined in future studies.

It has been well documented that Parkin initiates a series of biochemical events upon its translocation to mitochondria, including ubiquitylation of OMM proteins and further recruitment of TBK1 and autophagy adaptor proteins (13). Here, we used a recently developed technique, called Mito-IP (24), to rapidly isolate mitochondria to further dissect the downstream events and the impact of ULK1-mediated signaling. The HA-tagged OMM protein OMP25 was stably introduced into our previously used HEK293T cells lines bearing Parkin alleles and mitochondria were rapidly purified by magnetic bead isolation. Consistent with the fractionation experiments, Mito-IP on SA3 Parkin cells showed reduced recruitment of Parkin to mitochondria following CCCP and attenuated PINK1-mediated phosphorylation of both Parkin and ubiquitin was also observed in the isolated mitochondria of SA3 Parkin cells compared to WT Parkin cells (Fig. 4A). In addition, ubiquitylation of Parkin substrates (CISD1 and Mfn2), mitochondrial recruitment of autophagy adaptor proteins (NDP52 and TAX1BP1), and lipidated LC3B (LC3B-II) were all compromised in the SA3 cells compared to the WT Parkin cells. To further validate the deficiency in recruitment of autophagy receptors to the mitochondria, we examined the appearance of NDP52 puncta in response to CCCP, which has been previously defined in HeLa cells (25). We first confirmed that HeLa cell lines stably expressing WT versus SA3 Parkin exhibited the same defects in biochemical mitophagy induction as observed in the HEK293T cells (less mitochondrial pParkin S65 and pUb S65) (fig. S5A). In agreement with the mitochondria purification, the HeLa cells stably expressing SA3 Parkin also showed significant reduction in the puncta size of endogenous NDP52 by immunofluorescence staining after CCCP when compared to the WT Parkin–expressing cells (Fig. 4B).

Fig. 4 Recruitment of Parkin and downstream mitophagy effectors to mitochondria is controlled by AMPK/ULK1 signaling following CCCP.

(A) Immunoblots of Mito-tag purified mitochondria from HEK293T cells stably expressing WT or SA3 YFP-Parkin treated with 20 μM CCCP for indicated times. All cells also treated with 10 μM MG132 for 4 hours. (B) Representative immunocytochemistry of NDP52 (white) and Tom20 (red) in HeLa cells stably expressing WT or SA3 YFP-Parkin. Cells treated with DMSO or CCCP for 4 hours. Nuclei stained with 4′,6-diamidino-2-phenylindole (blue). NDP52 puncta number and puncta size were analyzed by ImageJ with 10 individual images for each condition. Data are shown as the means ± SEM. **P < 0.001; ****P < 0.0001 by two-way ANOVA. Scale bar, 20 μm. n.s., not significant. (C) Immunoblots of Mito-tag purified mitochondria from WT YFP-Parkin expressing cells treated with DMSO or 10 μM 6965 for 15 min before CCCP treatment for indicated times. Cells treated with MG132 as in (A). (D) Immunoblots of cell lysates from WT YFP-Parkin expressing cells treated with CCCP ± 15 min pretreated with 50 μM 991 or 10 μM 6965 as indicated. (E) Immunoblots of purified mitochondria from cells stably expressing WT or SA3 YFP-Parkin treated with CCCP ± 50 μM 991 for 1 hour. All cells also treated with 10 μM MG132 for 1 hour. (F) Immunoblots of supernatant (top) and purified mitochondria (bottom) from control or CRISPR-mediated AMPK KO cells treated with 20 μM CCCP for indicated times. Cells treated with MG132 as in (A).

Next, we examined the effect of acute ULK1/2 kinase inhibition by pretreating cells with the selective ULK1/2 kinase inhibitor SBI-0206965 (“6965”) before CCCP treatment and then performing Mito-IP–based mitochondrial isolation. 6965 (10 μM) pretreatment rapidly inhibited ULK1 activity in cells (as visualized by pBeclin Ser30) (fig. S5B) and also greatly reduced mitochondria-associated PINK phosphorylation of Parkin and ubiquitin, evidence of an early block in mitophagy initiation (Fig. 4C). Subsequent mitophagy events in the 65- and CCCP-treated WT Parkin cells—including CISD1 ubiquitylation and recruitment of TAXBP1, NDP52, and LC3 to mitochondria—mirrored the phenotypes observed in cells stably expressing the SA3 mutant, arguing against a structural defect in the SA3 mutant that might yield an artifactual inflation of the role of phosphorylation of these sites in early mitophagy initiation (Fig 4C).

Having observed that synthetic AMPK agonist 991 (and its bioavailable analog MK-8722) could induce pParkin Ser108 (Fig. 1, H to J), we examined whether 991 would have an effect on ser108 beyond what was observed with CCCP alone. We observed that 991 cotreatment with CCCP can provoke even higher levels of Parkin Ser108 phosphorylation (which 6965 prevents) (Fig 4D). In Mito-IP experiments, the combination treatment (991 + CCCP) also enhanced PINK1-mediated P-Parkin S65 and P-Ub S65 at the mitochondria, as well as downstream events including TAXBP1, NDP52, and LC3 recruitment to the mitochondria (Fig. 4E). These effects of 991 to enhance CCCP-induced mitophagy events were severely attenuated in the SA3-Parkin cells (Fig. 4E), indicating that Parkin Ser108-110 is a primary target of the ability of 991 to enhance biochemical induction of mitophagy.

To complement these gain-of-function AMPK experiments, we next created HEK293T cells bearing CRISPR KO of AMPKα1/α2 and performed Mito-IP experiments following CCCP. As compared to control single-guide RNA (sgRNA) HEK293T cells, AMPK KO cells displayed loss of pParkin S108 after CCCP and defects in Parkin activation (PINK1 phosphorylation of Parkin and Ub; CISD1 and MFN2 ubiquitylation) and the recruitment of downstream effectors (TAXBP1, NDP52, and LC3B), phenocopying the SA3 Parkin cells and the 6965-treated WT Parkin cells (Fig 4F). Similar effects were observed in HeLa cells stably expressing WT Parkin and stably expressing CRISPR sgRNA for AMPK but not control (fig. S5C).

We next wanted to examine end point readouts of Parkin function, focused around its ability to result in mitochondrial clearance. Examinations stable-tagged Parkin localization at fixed time points following CCCP in another cell type—here MEFs—revealed that SA3 Parkin displayed a modest delay in mitochondria translocation compared to S65A and a classic Parkinson’s patient mutation K161N (fig. S6A). In these stable Parkin expressing cell lines, we assessed an end-stage readout of Parkin function: the clearance of mitochondria in the final step of Parkin-mediated mitophagy, when the damaged mitochondria are degraded in the lysosome and thus not detected by immunofluorescence staining, leading to diffuse cytosolic Parkin in cells where the mitochondria appear absent (“gone”). We observed that cells expressing SA3 Parkin displayed significantly less mitochondrial clearance than cells expressing WT Parkin after 16 hours of CCCP treatment (fig. S6B). SA3 cells also displayed a greater percentage of cells in which no Parkin colocalization or clearance of mitochondria was observed at all, similar to S65A mutant cells, which were also less severely affected than K161N cells (fig. S6B). To complement these mitochondrial clearance assays, we examined the ubiquitin ligase activity of Parkin as aforementioned (fig. S2D), assessing its ability undergo a C431S-thioester activation in response to CCCP, valinomycin, or antimycin/oligomycin. Despite equivalent stress pathway activation induced by mitochondrial ATP loss in WT and SA3 cells (as observed by P-ACC and P-Raptor), the SA3 mutant was greatly attenuated in its ability to promote the active thioester conjugated form when compared to WT (Fig. 5A). In addition, we examined the ability of ULK1 RNA interference (RNAi) or the small-molecule ULK1 kinase inhibitor SBI-0206965 to phenocopy the effects on Parkin activity observed with SA3 mutation. ULK1 RNAi reduced Parkin Ser108 phosphorylation (fig. S2B), mitochondrial clearance (fig. S6C), and CISD1 ubiquitylation (fig. S6D), and pretreatment with SBI-0206965 also reduced C431S thioester conjugation (Fig. 5B), comparable to the magnitude of effect of the SA3 mutation, confirming that loss of ULK1 kinase activity recapitulates the phenotype of SA3 Parkin.

Fig. 5 ULK1-dependent phosphorylation of Parkin is required for maximal Parkin activity.

(A) Immunoblots of HEK293T cells transfected with YFP-Parkin-C431S or YFP-Parkin-C431S/SA3. Twenty-four hours after transfection, cells were treated with CCCP (20 μM), valinomycin (5 μM), AO (2.5 μM antimycin A + 5 μM oligomycin) or vehicle (DMSO) for 2 hours. Red arrow indicates higher molecular weight Parkin-ubiquitin thioester species. (B) YFP-Parkin-C431S–expressing HEK293T cells treated as in (A) ± SBI-0206965 (10 μM) for 2 hours. (C) Functional analysis of Parkin-mediated mitophagy activity in HEK293T cells stably bearing mito-Keima reporter. The graph indicates the mitophagy-positive cell population quantified by flow cytometry. Data are shown as the means ± SEM of two independent experiments. *P < 0.05 when compared to the control cells at 2 or 4 hours, respectively, by two-way ANOVA. (D) Immunoblot analysis of endogenous Parkin in hESC-derived induced neurons treated with AO for indicated times or 60 min with 10 or 50 μM 6965 pretreatment. (E) Model for Parkin activation. CCCP treatment causes rapid increases in AMP and mtROS, which activate AMPK to phosphorylate and activate ULK1 within minutes. ULK1, in turn, phosphorylates Parkin maximally at Ser108 in the cytoplasm within 2 min of CCCP treatment. Meanwhile, PINK1 is slowly becoming stabilized on the mitochondrial outer membrane following CCCP, with detection of its phosphorylation of ubiquitin and Parkin Ser65 appearing around 30 min and becoming maximal at 60 min or later.

To more quantitatively compare the activity of different Parkin mutants, we introduced a dual-color fluorescence reporter, mito-Keima, into the cells to perform a fluorescence-activated cell sorting–based mitophagy assay (fig. S7A) (26). With transient expression of YFP-Parkin mutants, SA3 Parkin displayed a 50% reduction in the mitophagy-positive population at 2 hours compared to WT Parkin (Fig. 5C). In line with previous results, R104W imposed a slightly greater defect than SA3. We also generated a serine-to-threonine mutant of the ULK1 sites on Parkin (ST3), which is unphosphorylatable by ULK1 (fig. S4B) and would be predicted to have less perturbation of normal Parkin structure than serine to alanine substitutions, and ST3-Parkin phenocopied SA3-Parkin in the mito-Keima assay. A recent study reported AMPK can also directly phosphorylate Parkin at the Ser9 position specifically under conditions of necroptosis induction (27). It is noted that the surrounding amino acids of Parkin Ser9 do not conform to the well-established optimal AMPK consensus substrate motif, and we were unable to detect evidence of AMPK phosphorylation of the Parkin Ser9 when using the AMPK substrate motif antibody as used in that study (27). Nonetheless, we sought to examine the impact of Ser9 mutation on Parkin activity here. As reported in the previous study, we did not observe any defects of S9A Parkin in the mito-Keima assay (Fig. 5C) nor on Parkin function on downstream biochemical events at the mitochondria (fig. S7B). Together, these data suggest that if Ser9 is an AMPK substrate under any conditions, it is not playing any role after mitochondrial damage.

Last, to further examine our findings in a neuronal context, which is the cell type of importance in Parkinson’s disease, we examined whether endogenously expressed Parkin in human embryonic stem cell (ESC) line–derived neurons can be regulated by AMPK/ULK1 signaling. AO can induce endogenous Parkin phosphorylation at ser108 just like the ectopic Parkin in other cell types, and the pretreatment of ULK1 inhibitor attenuated this phosphorylation event (Fig. 5D). Moreover, the molecular markers of mitophagy induction imposed by AO are also partially inhibited by ULK1 inhibitor in a dose-dependent manner.


Together, these results in this study present an unexpected and rapid new step in the biochemical activation of Parkin following mitochondrial damage (Fig. 5E). Immediately following mitochondrial depolarization, AMPK is fully activated in the cytosol within 2 min and directly phosphorylates its substrates Raptor, ACC, and ULK1. Parkin is phosphorylated by ULK1 at Ser108 in its recently described nine amino acid ACT element at this early time point; an event that appears to occur in the cytoplasm, though it should be noted that Ser108-phosphorylated Parkin can be weakly detected in mitochondrial fractions at 2 min. ULK1 and AMPK, in turn, translocate to the mitochondria within 10 min of CCCP, where AMPK directly phosphorylates MFF to promote mitochondrial fission. Thirty to 60 min later, PINK1 phosphorylates Parkin and ubiquitin at the mitochondria, accompanied by TBK1 activation and ubiquitylation of the Parkin substrates CISD1 and MFN2. PINK1-dependent phosphorylation of Parkin and ubiquitin may be comparatively slower due to the temporal nature of PINK1 stabilization on the OMM. Recent studies detail a later wave of ULK1 activation recruited by NDP52 and TBK1 at time points after full PINK1 activation at the mitochondria (1 hour or later) (28).

The amino acid sequence flanking Parkin Ser108 makes it not only a consensus phosphorylation site for ULK1 but also a divergent consensus for AMPK itself (Hyd-R-x-x-x-pS-x-x-x-L), making it possible that either kinase may be capable of inducing this site under some circumstances. It is also worth noting that SBI-0206965 can inhibit AMPK directly at doses ~5× higher than those with inhibiting ULK1. As we used SBI-0206965 at 10 μM throughout the study, we additionally tested the impact of this dose on AMPK signaling, observing no effect on AMPK phosphorylation of its well-studied substrate P-ACC1 yet complete loss of P-Ser108 Parkin (Fig. 4D). Given the complex interplay between AMPK, ULK1, and other components of the autophagy pathway (9), future studies will be needed to further delineate the role and relative importance of AMPK and ULK1/2 in Parkin activation and Parkin function in dopaminergic neurons, innate immunity, and inflammation.

PINK1/Parkin-mediated mitophagy is traditionally considered in the context of mitochondrial depolarization by promoting the turnover of isolated damaged mitochondria, but ULK1-mediated Parkin phosphorylation downstream of AMPK directly connects Parkin to the highly conserved AMPK energy sensing response, which is activated by a wide range of more mild stresses and hormones on the organismal as well as cellular level (16). Unlike PINK1, AMPK and ULK1 are activated fully by even mild mitochondrial perturbations, including the frontline type 2 diabetes drug metformin, which we find triggers phosphorylation of Ser108 endogenous Parkin in the liver. Coupled with the original observation of Ser108 phosphorylation in brown fat, this expands the physiological contexts in which Parkin is traditionally thought of playing a homeostatic role in mitochondrial health.

The findings here that the phosphorylation of Ser108 in the ACT domain of Parkin is required for rapid and maximal Parkin phosphorylation by PINK1 on Ser65 after mitochondrial uncouplers reveals an expected critical and early role for the AMPK-ULK1 axis in mitophagy after mitochondrial damage. Although previous studies indicated a minimal role for AMPK in later stages of mitophagy, once PINK1 and Parkin are fully activated (28), though that does not rule out additional targets/functions of AMPK in the initial hour after mitochondrial damage where much of AMPK action takes place. It is also worth noting that ULK1 may connect to additional regulators of mitophagy and/or PINK1/Parkin function, as ULK1 was recently reported to directly phosphorylate VCP/p97 (29), which was previously shown to be recruited to sites of Parkin-dependent ubiquitylation (30).

In conclusion, our study has identified Ser108 as a novel site of Parkin regulation directly downstream of AMPK/ULK1 pathway activation and forces a revision of dogma regarding when and where Parkin function may be important. The ability of ULK1 to phosphorylate Ser108 in the Parkin ACT element following even mild mitochondrial stresses—including metformin—begets questions of whether this event serves as an “early alert signal” of mitochondrial damage and may play a surveillance/proteostatic role in some biological contexts by modulating Parkin interactions before full Parkin catalytic activation. The direct connection of AMPK and ULK1 to Parkin opens up several avenues for future investigation in important in vivo contexts, including Parkinson’s disease, cancer, and diabetes.


Antibodies and reagents

Cell Signaling Technology (CST) antibodies used were ACC (#3662), phospho-ACC S79 (#3661), AMPK (#2532), phospho-AMPK T172 (#2535), Atg13 (#13273), glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (#5174), green fluorescent protein (GFP) (#2956), Myc-tag (#2278), Raptor (#2280), phospho-Raptor (#2083), TBK1 (#3013), phospho-TBK1 S172 (#5483), ULK1 (#8054), phospho-|ULK1 S317 (#12753), phospho-ULK1 S555 (#5869), phospho-ubiquitin S65 (#37642), PINK1 (#6946), phospho-MFF S146 (#49281), VDAC (#4661), TAX1BP1 (#5105), NDP52 (#60732), and ATG16L1 (#8089). Phospho-Parkin S108 (#36728) was developed in collaboration with Thorsten Wiederhold at CST. Phospho-Vps34 S249 antibody was developed in collaboration with G. Kasof at CST as reported previously (13). Sigma antibodies used were ULK1 (#A7481), β-actin (#A5541), Mitofusin 2 (#M6319), and Flag tag polyclonal (#F7425). MitoNEET/CISD1 (#16006-1-AP) and total MFF (#17090-1-AP) antibodies were from ProteinTech. Phospho-ATG16L1-S278 was purchased from Abcam (ab195242). Earle’s Balanced Salt Solution (EBSS) (#14155-063) was from Gibco/Life Technologies. INK-128 (#A-1023) was from Active Biochem. CCCP (#C2759), AO (#75351), antimycin A (A8674), rotenone (#R8875), and phenformin (#P7045) were from Sigma-Aldrich. Valinomycin was from Thermo Fisher Scientific (#V-1644). Compound 991 was from purchased from Glixx Laboratories Inc. (#GLXC-09267). Anti-Parkin phospho-S65 antibody (Abcam, MJF17) was a gift from M. Muqit. MitoTracker Deep Red was from Invitrogen (M22426). NDP52 for immunostaining was purchased from GeneTex (#GTX115378).


YFP-Parkin plasmid was from Addgene (#23955). The cDNA encoding human Vps34 was obtained from Invitrogen. The Flag tag and attL1 sites (for BP Gateway reaction) were cloned by polymerase chain reaction (PCR) using standard methods. cDNAs were subcloned into pDONR221 with BP clonase (Invitrogen), and site-directed mutagenesis was performed using QuikChange II XL (Stratagene). Kinase-dead ULK1 was achieved by introduction of a K46I mutation. WT and mutant alleles in pDONR221 were sequenced in their entirety to verify no additional mutations were introduced during PCR or mutagenesis steps and then put into either pcDNA3 Myc or pcDNA3 Flag mammalian expression vectors, or pQCXIN retroviral destination vector from Addgene (#17399) by LR Gateway reaction (Invitrogen). pMXs-3XHA-EGFP-OMP25 was a gift from D. Sabatini (Addgene plasmid, # 83356;; RRID:Addgene_83356). pLenti-mt-mKeima was subcloned from pCHAC-mt-mKeima (a gift from R. Youle; Addgene plasmid, # 72342;; RRID:Addgene_72342).

Cell culture, transient transfections, and cell lysis

HEK293T and SV40 immortalized WT MEF cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Mediatech) containing 10% fetal bovine serum (FBS) (Hyclone, Thermo Fisher Scientific) and penicillin/streptomycin (Gibco) at 37°C in 10% CO2. For transient expression, cells were transfected using Lipofectamine 2000 (Invitrogen) following the manufacturer’s protocol. Negative control (−) for cDNA transfections was pEBG empty backbone glutathione S-transferase (GST) fusion vector. Cells were harvested 16 hours after transfection, and cell lysates were prepared and immunoblotted as previously described (10). RNAi knockdown was performed using Dharmacon ON-TARGETplus SMARTpool or ON-TARGETplus Non-targeting Control Pool (20 nM) using RNAiMAX (Invitrogen) according to the manufacturer’s protocol for 72 hours. Immunoprecipitations were performed 16 hours after transfection using GFP-Trap _A (GTA, Bulldog Bio) or using the indicated antibodies in combination with recombinant Protein A sepharose (Invitrogen, #101041) or Protein G sepharose (Invitrogen, 101242) beads as indicated. Phosphatase treatment was performed on immunoprecipitations using lambda protein phosphatase and the included Protein MetalloPhosphatases (PMP) buffer negative control from New England Biolabs (#P0753). To perform in vitro kinase assays, Parkin and VPS34 were isolated by transfection of WT Myc-Parkin or Flag-VPS34 into HEK293T cells and subsequent immunoprecipitation 16 hours after transfection using monoclonal antibodies directed toward the epitope tag (anti-Myc and anti-Flag, respectively) and sepharose beads conjugated to recombinant Protein G. In vitro kinase assays were performed using either Parkin or VPS34 as substrates for ULK1 Recombinant Human Protein (Thermo Fisher Scientific, #PV6430) and ATP for 30 min in kinase buffer [50 mM tris (pH 7.5), 10 mM MgCl2, and 2 mM dithiothreitol (DTT)] at 30°C; the reaction was stopped with 6× loading buffer. CRISPR guide RNA sequences (31) were cloned into pLenti-CRISPR v2 vector (32). Stable CRISPR KO cells were made after viral infection and puromycin selection.

Generation of stable cell lines and fluorescence microscopy

Stable retroviral expression of WT and mutant YFP-Parkin in MEF cells was performed as described previously (33). Briefly, the pQCXIN YFP-Parkin constructs were transfected along with the ampho packaging plasmid into HEK293T cells, the virus-containing supernatants were collected 48 hours after transfection and passed through 0.45-μm filters, and target MEFs were infected in the presence of polybrene (5 μg/ml; Sigma-Aldrich, #107689). Twenty-four hours after infection, cells were selected with neomycin (0.5 mg/ml; Invitrogen, #10131027). Immunofluorescence was performed on cultured cells either stably expressing YFP-Parkin from the pQCXIN retroviral destination vector or transiently transfected as described, which were plated on glass coverslips and subsequently fixed in 4% paraformaldehyde (Electron Microscopy Sciences, #15710). Immunofluorescence blocking buffer was 0.1% Triton X-100, 0.02% SDS, and bovine serum albumin (BSA) (10 mg/ml). Mitochondria were visualized using Tom20 primary antibody (1:200; Santa Cruz Biotechnology, #11415) and Alexa Fluor 555 secondary (1:1000; Invitrogen, #A-21429); nuclei were visualized with 4′,6-diamidino-2-phenylindole (Sigma-Aldrich, #D9542). Coverslips were mounted with Fluoromount-G (Thermo Fisher Scientific, #OB10001). Images were acquired on a Zeiss LSM 700 confocal microscope, and scoring was performed blind to cellular genotype/expression and treatment, using previously described methods (34, 35). Data are shown as the means ± SEM of three independent experiments with ≥100 cells counted for each condition for each replicate. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by two-way analysis of variance (ANOVA).

NDP52 recruitment assay

HeLa cells with stable expression of WT or SA3 YFP-Parkin were treated with dimethyl sulfoxide (DMSO) or CCCP for 4 hours. Cells were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS), permeabilized with 1× PBS/0.25% Triton X-100 (PBST) for 15 min, and then blocked with 1× PBS/0.25% Triton X-100/5% BSA for 30 min. Primary antibodies (1:200; NDP52 or Tom20) were added in antibody dilution buffer (1× PBS/0.25% Triton X-100/1% BSA) for overnight incubation. Anti-mouse Alexa Fluor 555 and anti-rabbit Alexa Flour 647 secondary antibodies were applied for 1 hour. At least 10 individual images were taken per genotype per condition. Three independent experiments were performed. NDP52 puncta number and their sizes were quantified by ImageJ for any positive foci larger than 1 pixel. The puncta size was calculated and converted into the actual metrics on the basis of the scale bar. Data are shown as the means ± SEM of three independent experiments with ≥50 cells counted for each condition for each replicate. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 by two-way ANOVA.

Mass spectrometry

YFP-Parkin was overexpressed in HEK293T cells as described and immunoprecipitated with GFP-Trap (Bulldog Bio), run out on SDS-PAGE gels, and stained with GelCode blue reagent (Thermo Fisher Scientific). Bands on the gels were cut out and subjected to reduction with DTT, alkylation with iodoacetamide, and in-gel digestion with trypsin or chymotrypsin overnight at pH 8.3, followed by reversed-phase microcapillary/tandem mass spectrometry [liquid chromatography (LC)/MS/MS]. LC/MS/MS was performed using an Easy-nLC nanoflow HPLC (Proxeon Biosciences) with a selfpacked 75-μm id × 15-cm C18 column coupled to a LTQ-Orbitrap XL mass spectrometer (Thermo Fisher Scientific) in the data-dependent acquisition and positive ion mode at 300 nl/min. Peptide ions from predicted phosphorylation sites were also targeted in MS/MS mode for quantitative analyses. MS/MS spectra collected via collision induced dissociation in the ion trap were searched against the concatenated target and decoy (reversed) single entry and full Swiss-Prot protein databases using Sequest (Proteomics Browser Software, Thermo Fisher Scientific) with differential modifications for serine/threonine/tyrosine phosphorylation (+79.97) and the sample processing artifacts Met oxidation (+15.99), deamidation of asparagine and glutamine (+0.984) and cysteine alkylation (+57.02). Phosphorylated and unphosphorylated peptide sequences were identified if they initially passed the following Sequest scoring thresholds against the target database: 1+ ions, Xcorr ≥ 2.0 Sf ≥ 0.4, P ≥ 5; 2+ ions, Xcorr ≥ 2.0, Sf ≥ 0.4, P ≥ 5; 3+ ions, Xcorr ≥ 2.60, Sf ≥ 0.4, P ≥ 5 against the target protein database. Passing MS/MS spectra were manually inspected to be sure that all b- and y-fragment ions aligned with the assigned sequence and modification sites. Determination of the exact sites of phosphorylation was aided using FuzzyIons and GraphMod and phosphorylation site maps were created using ProteinReport software (Proteomics Browser Software suite, Thermo Fisher Scientific). False discovery rates of peptide hits (phosphorylated and unphosphorylated) were estimated below 1.5% on the basis of reversed database hits.

Mouse studies

AMPK α1/α2 floxed allele (Prkaa1fl/fl; Parkaa2fl/fl) mice or ULK1/2 floxed allele (Ulk1fl/fl; Ulk2fl/fl) mice expressing Ubc-creERT2 or floxed allele mice without cre (Control) were treated with tamoxifen (3 mg/kg per day) for five consecutive days. Eight weeks after Tamoxifen (TAM) injection, mice were fasted overnight, refed for 1 hour, then treated for 2 hours with vehicle or MK-8722 (30 mg/kg) before sacrificing. Similarly, WT C57/Bl6 mice on high-fat diet for 8 weeks mice were fasted overnight, refed for 1 hour, then treated for 2 hours with saline or metformin (250 mg/kg). Livers were collected, lysates were prepared in CST lysis buffer supplemented with 5× EDTA-free protease inhibitor cocktail (Roche) and Calyculin A, and subsequently run on standard SDS-PAGE gels or Phos-tag acrylamide gel (36) as indicated, then immunoblotted with the indicated antibodies. In Fig. 1K, primary hepatocytes were made and treated with metformin as previously described (17). All animal procedures were approved by the Salk Institute’s Institutional Animal Care and Use Committee.

Detection of Parkin-ubiquitin oxyester

The YFP-Parkin-C431S mutant was generated using PCR-based site-directed mutagenesis (forward primer: 5′-GTGGAAAAAAATGGAGGCAGCATGCACATGAAGTGTC-3′; reverse primer: 5′-GACACTTCATGTGCATGCTGCCTCCATTTTTTTCCAC). HEK239T cells were transiently transfected with YFP-Parkin or YFP-Parkin-C431S or YFP-Parkin-SA3/C431S by using Lipofectamine 2000 (Invitrogen) according to the manufacture’s protocol. Twenty-four hours after transfection, the indicated drug treatments were given for 2 hours with a media change. Cells were lysed on ice for 30 min in 50 mM tris (pH 7.5), 150 mM NaCl, 0.5% NP-40, and 10% glycerol with complete protease inhibitors (Roche). Lysates were clarified for 10 min at 16,000g centrifugation at 4°C, and the supernatant protein was quantified by the bicinchoninic acid (BCA) reagent (Thermo Fisher Scientific). To remove oxyester-linked Parkin, a portion of each lysate was treated with 100 mM NaOH at 37°C for 1 hour. Lysates were mixed with SDS sample buffer and boiled for 10 min before SDS-PAGE analysis. For optimal cellular signaling detection, duplicated plates were lysed in the CST lysis (Cell Signaling Technology) buffer with phosphatase and protease inhibitors.

Mitochondria fractionation

HEK293T (2 × 107) cells with stable YFP-Parkin expression were resuspended in 1 ml of MS buffer [210 mM mannitol, 70 mM sucrose, 5 mM tris-HCl (pH 7.5), and 1 mM EDTA (pH 7.5)]. Cell homogenate was obtained by disrupting cell membrane through a 27-gauge needle for 10 strokes. The degree of homogenization was monitored with a phase-contrast microscope. The homogenate was first centrifuged at 1500g for 5 min to remove nuclei and cell debris. The supernatant was transferred to a new tube and a second centrifugation was carried out at 10,000g for 15 min to pellet down mitochondria. The upper supernatant was saved as cytoplasmic fraction, and the mitochondria pellet was washed with MS buffer and centrifuged again at 10,000g for 15 min. This crude mitochondria pellet was then lysed in the CST lysis buffer for Western blot analysis. The entire process of mitochondria isolation was performed on ice or at 4°C. Protease inhibitors and phosphatase inhibitors were included in MS buffer and CST lysis buffer.

In vitro kinase assays

HEK293T cells transiently transfected with YFP alone, YFP-Parkin, or YFP-tagged Parkin mutants were lysed and subjected to YFP immunoprecipitation. Immunoprecipitates were washed three times in lysis buffer, followed by three times in kinases assay buffer [50 mM tris (pH 7.5) and 10 mM MgCl2]. Subsequently, immunoprecipitates were subjected to a kinase reaction containing 0.1 mM [γ32P]-ATP (PerkinElmer) and 2 mM DTT with or without 0.5 μg of active recombinant GST-ULK1 (Life Technologies, #PV6430) in the presence of kinase assay buffer. The reaction was incubated at 30°C for 30 min. Reactions were terminated with lithium dodecyl sulfate (LDS) sample buffer and resolved by SDS-PAGE electrophoresis. Proteins were detected with Coomassie staining. Dried gels were exposed to UltraCruz autoradiography film overnight, in an autoradiography cassette and the films were later developed using an autodeveloper.

Immunopurification of mitochondria (Mito-IP)

Mito-IP HEK293T cells were generated by subsequent infection with viruses containing 3xHA-EGFP-OMP25 and YFP-Parkin. The protocol of Mito-IP was modified from Chen et al. (24). Briefly, 2 × 107 HEK293T cells were washed with PBS and then resuspended in 1 ml of cold KPBS (136 mM KCl, 10 mM KH2PO4, pH 7.25). Cells were homogenized in a plain plunger with 25 strokes. Nuclei were removed by a quick spin at 1000g for 2 min. Transfer the postnuclear supernatant to a new tube with prewashed anti-hemagglutinin (HA) magnetic beads. The immunoprecipitation was carried out on a rotator at 4°C for 15 min. The beads were washed with KPBS three times and the mitochondria lysates were obtained by a direct incubation of the beads with CST lysis buffer.

Cellular ADP/ATP ratio and mitochondrial ROS levels

For measuring cellular ADP/ATP ratio, HEK293T cells were plated at 1 × 104 cells per well in a 96-well cell culture plate. Next day, cells were incubated in fresh media containing 0, 5, or 10 μM CCCP for various time points and then were lysed immediately after CCCP depolarization by using a bioluminescent ADP/ATP Ratio Assay kit (Abcam, #ab65313) according to the manufacturer’s protocol. The luminescence intensity was measured in a microplate reader (Tecan Infinite M1000 PRO). The relative ADP/ATP ratio was normalized to the value of untreated cells (0 μM CCCP) at each time point, respectively. To monitor mtROS levels, 1 × 104 U2OS cells were first incubated with 5 μM MitoSOX dye (Invitrogen, #M36008) in Hanks’ balanced salt solution (HBSS) with 2% FBS at 37°C for 30 min. Excess dye was washed away with HBSS for two times. Cells were then incubated with HBSS (2% FBS) containing 0, 5, or 10 μM CCCP. The fluorescence intensity was measured in a microplate reader (Tecan Infinite M1000 PRO) with excitation/emission at 510/580 nm.

Mitophagy assay with mito-Keima reporter

HEK293T cells stably expressing mitochondrially targeted mt-Keima (mito-Keima). YFP-Parkin and referred Parkin mutants were transiently expressed in mt-Keima cells for 24 hours before the assay. Cells were treated with DMSO or 10 μM CCCP for 2 and 4 hours and then analyzed on an LSR-II flow cytometer (Flow Cytometry Core, Salk). For each sample, 50,000 events were collected and YFP-positive population was first gated and subjected to analysis of mt-Keima intensity. Measurement of mt-Keima was performed by a dual-excitation at 405 nm (Keima in cytosol, ~pH 7) and 561 nm (Keima in acidic environment, ~pH 4) lasers and detection through a 610/20 filter. The percentage of cell population with increased ratio of 561:405 emission was accessed as a readout of mitophagy activity.

In vitro neural precursor cell differentiation

Human in vitro differentiated neurons were provided by Salk Stem Cell Core. In brief, H9 ESCs were differentiated to neural precursor cells [hES–neural precursor cells (NPCs)] using previously published methods. Purified NPCs were matured in neurogenic conditions [DMEMF12-based media with 1× N2, 1× B27, glial cell line–derived neurotrophic factor (20 ng/ml), brain-derived neurotrophic factor (20 ng/ml), 1 mM cyclic AMP, and 200 nM ascorbic acid] for 4 weeks to generate electrophysiologically active neurons. Mature neurons were harvested for end point assays after 4 weeks. Neural precursor cells and mature neurons were derived from the H9 human ESC line (procured as a deidentified cell line from the WiCell cell repository). As these cells were procured as a deidentified commercial product from an established cell repository human subjects oversite and approval (Institutional Review Board) was not required. This work was completed by personnel in the Salk Institute Stem Cell Core facility with the oversite and approval of the Salk Institute Embryonic Stem Cell Research and Oversite (ESCRO) committee (protocol 08-0006sc). The purpose of the Salk Institute ESCRO committee is to provide oversight of human ESC research and other stem cell research covered by the California Institute for Regenerative Medicine and California Department of Public Health regulations to ensure Salk research meets the highest scientific and ethical standards. The ESCRO Committee reviews new protocols, modifications to currently approved research, and continuing research using or creating human pluripotent stem cells or stem cell lines, any proposed collection and use of germ or other cells designed to generate pluripotent stem cells, and any covered cells as required by State or Federal Law.


Supplementary material for this article is available at

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.


Acknowledgments: We thank T. Wiederholden at Cell Signaling Technology for partnership in developing the pSer108 Parkin antibody. We thank M. Muquit for providing the Phospho-Ser65 Parkin antibody in early stages of the project, D. Egan for assistance with fig. S1A and early stages of this project, and K. Lamia for assistance with Fig. 3A model. Funding: This study was supported by grants to R.J.S. from the National Institute of Health (NIH) (R35CA220538 and R01DK080425), the Leona M. and Harry B. Helmsley Charitable Trust (grant no. 2012-PG- MED002), and the AHA-Allen Foundation funding to Aging Research at the Salk Institute (P01CA120964) to R.J.S. and J.M.A. C.-M.H. was supported by a Salk Institute Pioneer Fund Fellowship and funding from the Bruce Ford and Anne Smith Bundy Foundation. P.S.L. was supported by the NIH Cell and Molecular Genetics Training Program (grant 5T32GM007240-35) and the Chapman Charitable Trust. S.N.B. was supported by NIH T32 postdoctoral training grant to the Salk Institute Cancer Center (5T32CA009370-33) and an NIH postdoctoral fellowship (F32CA206400). J.L.V.N. was supported by a Damon-Runyan Postdoctoral Fellowship. K.H. was supported by a Hewitt Foundation Fellowship. Author contributions: P.S.L., C.-M.H., and R.J.S. designed the experiments and wrote the manuscript with input from all authors. P.S.L. originated the project and performed experiments in Fig. 1 (A to H), and figs. S1, S2 (A to C), and S6. C.-M.H. performed experiments in Figs. 2 to 4 and figs. S2 (D and E) to S5 and S7. N.M. performed in vitro kinase assay in fig. S4B and assisted with Fig. 1K and fig. S5. J.L.V.N. performed mouse experiments for Fig. 1K. S.N.B., J.B., K.H., and D.G. assisted with experiments in Fig. 1 (I to K). K.D. differentiated hESC cells into neurons for Fig. 4F. J.M.A. performed mass spectrometry in fig. S1B. Competing interests: All authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article