SEARCH

SEARCH BY CITATION

Keywords:

  • Mfn1;
  • mitochondria;
  • mitophagy;
  • parkin;
  • Parkinson’s disease;
  • ubiquitination

Abstract

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

J. Neurochem. (2011) 118, 636–645.

Abstract

Mutations in the parkin gene cause early-onset, autosomal recessive Parkinson’s disease. Parkin functions as an E3 ubiquitin ligase to mediate the covalent attachment of ubiquitin monomers or linked chains to protein substrates. Substrate ubiquitination can target proteins for proteasomal degradation or can mediate a number of non-degradative functions. Parkin has been shown to preserve mitochondrial integrity in a number of experimental systems through the regulation of mitochondrial fission. Upon mitochondrial damage, parkin translocates to mitochondria to mediate their selective elimination by autophagic degradation. The mechanism underlying this process remains unclear. Here, we demonstrate that parkin interacts with and selectively mediates the atypical poly-ubiquitination of the mitochondrial fusion factor, mitofusin 1, leading to its enhanced turnover by proteasomal degradation. Our data supports a model whereby the translocation of parkin to damaged mitochondria induces the degradation of mitofusins leading to impaired mitochondrial fusion. This process may serve to selectively isolate damaged mitochondria for their removal by autophagy.

Abbreviations used
CCCP

carbonyl cyanide 3-chlorophenylhydrazone

CHX

cycloheximide

GFP

green fluorescent protein

HA

hemagglutinin

IBR

in-between-RING finger

IP

immunoprecipitation

Mfn1

mitofusin 1

PBS

phosphate-buffered saline

PD

Parkinson’s disease

PINK1

PTEN-induced kinase 1

RING

really interesting new gene

shRNA

short hairpin RNA

Ub

ubiquitin

VDAC1

voltage-dependent anion channel 1

WT

wild-type

Mutations in the parkin gene (PARK2) cause early-onset, autosomal recessive forms of Parkinson’s disease (PD) (Gasser 2009). Parkin functions as an E3 ubiquitin ligase together with E1 and E2 enzymes to catalyze the covalent attachment of ubiquitin monomers to protein substrates (Shimura et al. 2000; Zhang et al. 2000; Moore 2006). A number of substrates have been identified for parkin in various experimental systems (Moore 2006), yet it has proved difficult to distinguish which if any of these substrates are important for mediating neurodegeneration because of loss-of-function mutations in parkin. Parkin is a relatively diverse ubiquitin ligase that can mediate the attachment of poly-ubiquitin chains to its substrates linked through various ubiquitin lysine residues, in addition to substrate mono-ubiquitination (Moore 2006). These alternative modes of ubiquitination potentially enable parkin to promote the proteasomal, lysosomal or autophagic degradation of its substrates, or impact non-degradative signaling pathways involving ubiquitin. However, the role of substrate ubiquitination in parkin-linked disease remains unclear.

Recent evidence supports a role for parkin in mitochondrial maintenance and dynamics. In Drosophila, the deletion of parkin produces flight muscle degeneration and mitochondrial morphological alterations which can be genetically rescued by promoting fission via increasing the gene dosage of Drp1 and by inhibiting the fusion-promoting factors Opa1 and mitofusin (Marf) (Greene et al. 2003; Pesah et al. 2004; Clark et al. 2006; Park et al. 2006; Deng et al. 2008; Poole et al. 2008). Deletion of PTEN-induced kinase 1 (PINK1), a mitochondrial kinase mutated in some recessive forms of PD, produces similar phenotypes in Drosophila to that of parkin mutants and functions upstream of parkin in a common pathway to preserve mitochondrial integrity (Clark et al. 2006; Park et al. 2006; Yang et al. 2006). These observations suggest that the PINK1/parkin pathway normally functions to regulate mitochondrial fission, and that loss of PINK1/parkin function may cause excess mitochondrial fusion (Deng et al. 2008; Poole et al. 2008; Yang et al. 2008; Park et al. 2009). As such, there is evidence of mitochondrial elongation in parkin mutant fibroblasts from human PD patients (Mortiboys et al. 2008). However, studies in parkin- or PINK1-deficient mammalian cells have reported variable effects on mitochondrial morphology (Exner et al. 2007; Dagda et al. 2009; Lutz et al. 2009; Morais et al. 2009; Sandebring et al. 2009). It is unclear at present how the regulation of mitochondrial dynamics by parkin or PINK1 preserves mitochondrial integrity.

Parkin translocates from the cytoplasm to damaged or dysfunctional mitochondria to promote their degradation by autophagy (termed mitophagy) (Narendra et al. 2008). Parkin translocation and the subsequent induction of mitophagy require the ubiquitin ligase activity of parkin and are further dependent on PINK1 activity (Narendra et al. 2008, 2010; Geisler et al. 2010; Matsuda et al. 2010; Vives-Bauza et al. 2010). It is not clear how parkin activity induces the selective elimination of damaged mitochondria by mitophagy and what role mitochondrial dynamics plays in this process. Recent studies have shown that upon translocation to damaged mitochondria, parkin ubiquitinates the mitochondrial substrates voltage-dependent anion channel 1 (VDAC1) and mitofusin (Geisler et al. 2010; Poole et al. 2010; Ziviani et al. 2010). Parkin mediates the atypical poly-ubiquitination of VDAC1 which is required for mitophagy (Geisler et al. 2010). In Drosophila cells, the PINK1/parkin pathway mediates the parkin-dependent ubiquitination of Marf (mitofusin) which can influence its abundance (Poole et al. 2010; Ziviani et al. 2010). In the present study, we demonstrate in mammalian cells that parkin mediates the atypical poly-ubiquitination of mitofusin 1 (Mfn1) which leads to its enhanced turnover by the proteasome. During the preparation of our manuscript, two reports were published describing similar observations (Gegg et al. 2010; Tanaka et al. 2010). These complementary studies suggest that parkin may normally regulate mitochondrial dynamics by inhibiting mitochondrial fusion through the ubiquitination and degradation of mitofusins.

Materials and methods

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Expression plasmids and antibodies

Mammalian expression plasmids for full-length FLAG-tagged human parkin, hemagglutinin (HA)-tagged human parkin domains, Myc-tagged human parkin [wild-type (WT), T240R, R256C, G328E and P437L)] and HA-tagged ubiquitin variants (WT, K48, K48R, K63 and K0) have been described (Moore et al. 2008). A 10xMyc-tagged mouse Mfn1 plasmid was obtained from Addgene (plasmid #23212; Chen et al. 2003), and Myc-tagged human Drp1, Mfn2 and OPA1 plasmids were kindly provided by Dr. Manuel Rojo (Université Victor Segalen, France; Rojo et al. 2002; Guillery et al. 2008). A C-terminal V5-tagged human PINK1 plasmid was obtained from Addgene (plasmid #13320; Beilina et al. 2005). green fluorescent protein (GFP2)-tagged human ubiquitin was kindly provided by Dr. Michel Bouvier (Université de Montreal, Canada; Perroy et al. 2004). Short hairpin RNA sequences in plasmid pLKO.1 targeting human parkin (PRK1, TRCN0000000285; PRK2, TRCN0000000284; PRK3, TRCN0000000283) were obtained from Thermo Fisher Scientific (Open Biosystems, Huntsville, AL, USA). A non-targeting control short hairpin RNA (shRNA) sequence in plasmid pLKO.1 was obtained from Addgene (plasmid #1864; Sarbassov et al. 2005). The following antibodies were employed: mouse monoclonal anti-c-myc (clone 9E10), anti-c-myc-peroxidase, anti-HA (clone 12CA5) and anti-HA-peroxidase (Roche Applied Science, Basel, Switzerland); mouse monoclonal anti-FLAG-(M2), anti-FLAG-(M2)-peroxidase and anti-β-tubulin (clone TUB 2.1) (Sigma-Aldrich, Buchs, Switzerland); mouse monoclonal anti-parkin (clone PRK8, Cell Signaling Technology, Inc., Danvers, MA, USA); rabbit polyclonal anti-Mfn1 (kindly provided by Dr. Manuel Rojo, Université Victor Segalen, France; Guillery et al. 2008), anti-Mfn2 (N-terminal 38–55; Sigma-Aldrich) and anti-β-actin (Sigma-Aldrich); peroxidase-coupled anti-mouse and anti-rabbit IgG, light chain-specific secondary antibodies (Jackson ImmunoResearch, Inc., West Grove, PA, USA).

Cell culture and transient transfection

Human SH-SY5Y neuroblastoma cells were maintained in Dulbecco’s modified Eagle’s media supplemented with 10% foetal bovine serum and 1× penicillin/streptomycin at 37°C in a 5% CO2 atmosphere. For transient transfection, cells were transfected with plasmid DNA using FuGENE HD reagent (Roche Applied Science) according to manufacturer’s recommendations. Cells were routinely harvested at 48–72 h post-transfection.

Co-immunoprecipitation and Western blotting

For co-immunoprecipitation (IP) assays, SH-SY5Y cells were transiently transfected with 3 μg of each plasmid in 10 cm dishes. After 48 h, cells were harvested in IP buffer [1× phosphate-buffered saline (PBS) pH 7.4, 1% Triton X-100, Complete Mini protease inhibitor cocktail (Roche Applied Sciences)]. Cell lysates were rotated at 4°C for 1 h and soluble supernatant fractions were obtained by centrifugation at 17 500 g for 15 min at 4°C. Soluble fractions were combined with 50 μL Protein G-Dynabeads (Invitrogen AG, Basel, Switzerland) pre-incubated with mouse anti-myc (5 μg) or anti-FLAG (5 μg) antibodies followed by overnight incubation by rotation at 4°C. Dynabead complexes were sequentially washed 1× with IP buffer supplemented with 500 mM NaCl, 2× with IP buffer and 3× with PBS. Immunoprecipitates were eluted by heating at 95°C for 5 min in 2× Laemmli sample buffer (Bio-Rad AG, Reinach, Switzerland) with 5% 2-mercaptoethanol. Immunoprecipitates and inputs (1% total lysate) were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, transferred to Protran nitrocellulose (0.2 μm; Perkin Elmer, Schwerzenbach, Switzerland), and subjected to Western blot analysis. Proteins were visualized by enhanced chemiluminescence (GE Healthcare, Glattbrugg, Switzerland) on a FujiFilm LAS-4000 Luminescent Image Analysis system.

Where indicated, SH-SY5Y cells were treated with MG132 (Enzo Life Sciences AG, Lausen, Switzerland) or dimethylsulfoxide as a control for 24 h prior to harvesting. For cycloheximide (CHX) chase assays, CHX (200 μg/mL; Sigma) was added to transfected cells at 48 h post-transfection, and cells were chased and harvested at 0, 1, 3, 6 and 24 h. For experiments with carbonyl cyanide 3-chlorophenylhydrazone (CCCP; Sigma-Aldrich), cells were treated with 10 μM CCCP for 1 h prior to harvesting.

Cell fractionation

Detergent-soluble and -insoluble protein fractions were prepared from SH-SY5Y cells in IP buffer [1× PBS pH 7.4, 1% Triton X-100, Complete Mini protease inhibitor cocktail (Roche Applied Sciences)]. Lysates were centrifuged at 17 500 g for 15 min at 4°C, and the resulting pellet (P1) and supernatant (Triton-soluble) fractions were collected. The P1 fraction was solubilized by sonication and heating at 95°C for 10 min in radioimmunoprecipitation assay (RIPA) buffer [10 mM Tris–HCl pH 7.4, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 2.5% sodium deoxycholate, 1% sodium dodecyl sulfate, Complete protease inhibitors (Roche Applied Sciences)] to produce the Triton-insoluble fraction (RIPA-soluble).

Ubiquitination assay

SH-SY5Y cells transiently transfected with HA- or GFP-tagged ubiquitin, and myc-tagged Mfn1 or Mfn2 with or without FLAG-tagged parkin, were harvested at 48 h post-transfection in IP buffer, and IP was conducted with anti-myc antibody (5 μg). IPs were washed stringently 5× in IP buffer supplemented with 500 mM NaCl and once with PBS, heated at 95°C for 5 min and eluted proteins were subjected to Western blot analysis with anti-HA, anti-GFP or anti-myc antibodies to detect Mfn-ubiquitin conjugates.

Results

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Parkin interacts with mitofusin 1 and 2

To begin to understand the effects of the PINK1/parkin pathway on mitochondrial dynamics and mitophagy in mammalian cells, we elected to explore the potential interaction of parkin with mitochondrial fission and fusion factors. Human SH-SY5Y neural cells transiently co-expressing FLAG-tagged parkin and myc-tagged Drp1, Mfn1 or Mfn2 were immunoprecipitated with anti-myc or anti-FLAG antibodies and analyzed by Western blotting (Fig. 1). The mitochondrial fission factor, Drp1, fails to interact with parkin (Fig. 1a), whereas the mitochondrial fusion factors, Mfn1 and Mfn2, both interact robustly with parkin (Fig. 1b and c). Our data demonstrates that parkin differentially interacts with components of the mitochondrial fission/fusion machinery in mammalian cells.

image

Figure 1.  Parkin interacts with Mfn1 and Mfn2. (a) Parkin fails to interact with Drp1. SH-SY5Y cells co-expressing FLAG-parkin and Myc-Drp1 were subjected to immunoprecipitation (IP) with anti-myc antibody and IPs and input lysates were probed with anti-FLAG or anti-myc antibodies. (b and c) Co-immunoprecipitation of parkin with Mfn1 and Mfn2. Cells co-expressing FLAG-parkin and Myc-Mfn1 (b) or Myc-Mfn2 (c) were subjected to IP with anti-FLAG antibody and IPs and inputs were probed with anti-myc or anti-FLAG antibodies. Blots are representative of at least three independent experiments.

Download figure to PowerPoint

Parkin enhances the ubiquitination of mitofusin 1

We next sought to determine whether parkin could mediate the ubiquitination of Mfn1 and Mfn2. Cells co-expressing HA-tagged ubiquitin, FLAG-tagged parkin and myc-tagged Mfn1 or Mfn2 were immunoprecipitated with anti-myc antibody and analyzed by Western blotting to evaluate ubiquitination (Fig. 2). Parkin over-expression markedly promotes the ubiquitination of Mfn1 (Fig. 2a). Mfn1 ubiquitination by parkin is also clearly detected using a GFP-tagged ubiquitin construct (Fig. 2b). Proteasome inhibition with MG132 leads to an increase in ubiquitinated Mfn1 that is further enhanced by parkin over-expression (Fig. 2a). These observations suggest that the ubiquitination of Mfn1 by parkin may target Mfn1 for proteasomal degradation. In contrast, we find no evidence for parkin-mediated ubiquitination of Mfn2 (Fig. 2c), Drp1 or OPA1 (Fig. 3) in the presence or absence of proteasome inhibition. Mfn2 is normally ubiquitinated in SH-SY5Y cells at a low level, as revealed by the appearance of Mfn2-ubiquitin conjugates of increased molecular mass (Fig. 2c), but this ubiquitination is not altered by the over-expression of parkin. We note, however, that parkin over-expression leads to a small reduction in the steady-state levels of Mfn2 that is partly restored by proteasome inhibition (Figs 1b and 2c), consistent with a role for parkin in mediating its degradation. PINK1 is required for the recruitment of parkin to damaged mitochondria to mediate their removal by autophagy (Geisler et al. 2010; Matsuda et al. 2010; Narendra et al. 2010; Vives-Bauza et al. 2010). To determine whether PINK1 influences Mfn1 ubiquitination by parkin, ubiquitination assays were conducted in the presence or absence of exogenous human PINK1. PINK1 over-expression markedly enhances the parkin-mediated ubiquitination of Mfn1 (Fig. 2d). Collectively, our data demonstrate that parkin mediates the ubiquitination of Mfn1 and that PINK1 can further enhance this modification.

image

Figure 2.  Parkin mediates the ubiquitination of Mfn1 but not Mfn2. (a) In vivo ubiquitination assay revealing Mfn1 ubiquitination by parkin. SH-SY5Y cells co-expressing combinations of Myc-Mfn1, HA-ubiquitin and FLAG-parkin were treated with or without the proteasome inhibitor MG132 (5 μM) for 24 h and then subjected to IP with anti-myc antibody and IPs and inputs were probed with anti-HA, anti-myc and anti-FLAG antibodies. Mfn1 is ubiquitinated in the presence of parkin, and this is further enhanced following proteasome inhibition. (b) Confirmation of Mfn1 ubiquitination. Ubiquitination assay in SH-SY5Y cells co-expressing GFP-ubiquitin, Myc-Mfn1 and FLAG-parkin. Lysates were subjected to IP with anti-myc antibody and probed with anti-GFP, anti-myc and anti-FLAG antibodies. (c) Mfn2 is not ubiquitinated by parkin. Similar ubiquitination assay to (a) on cells co-expressing Myc-Mfn2, HA-ubiquitin and FLAG-parkin treated with or without MG132 (5 μM) for 24 h. Mfn2 is ubiquitinated at a low level in the presence or absence of parkin as revealed by the appearance of Mfn2-ubiquitin conjugates, which are not further enhanced by proteasome inhibition. Notice the reduction in Mfn2 levels in the presence of parkin. (d) PINK1 enhances the ubiquitination of Mfn1 by parkin. Cells co-expressing combinations of Myc-Mfn1, HA-ubiquitin, FLAG-parkin and V5-PINK1 were subjected to IP with anti-myc antibody and probed with anti-HA, anti-myc, anti-FLAG and anti-V5 antibodies. Blots are representative of at least three independent experiments.

Download figure to PowerPoint

image

Figure 3.  Parkin does not ubiquitinate Drp1 or OPA1. (a) Drp1 is not ubiquitinated by parkin in the presence or absence of proteasome inhibition. SH-SY5Y cells co-expressing Myc-Drp1, HA-ubiquitin and FLAG-parkin were treated with or without MG132 (5 μM) for 24 h and subjected to IP with anti-myc antibody and probing with anti-HA, anti-myc and anti-FLAG antibodies. Drp1-ubiquitin conjugates are not detected following parkin over-expression with or without proteasome inhibition. Also, Drp1 levels are not altered following parkin over-expression. (b and c) Similar ubiquitination assay on cells co-expressing Myc-OPA1, HA-ubiquitin and FLAG-parkin following treatment with (c) or without (b) MG132 (5 μM) for 24 h. OPA1-ubiquitin conjugates are not detected following parkin over-expression and proteasome inhibition. Also, OPA1 levels are not reduced following parkin over-expression. Blots are representative of at least two independent experiments.

Download figure to PowerPoint

PD-associated mutations impair parkin-mediated ubiquitination of mitofusin 1

To further define the interaction between parkin and Mfn1, we conducted co-immunoprecipitation experiments to map the protein domain of parkin that mediates this interaction. Parkin contains an N-terminal ubiquitin-like domain (Ubl) and an enzymatic C-terminal really interesting new gene (RING) box domain consisting of two RING finger motifs (RING1 and RING2) separated by an in-between-RING finger (IBR) motif (Fig. 4a). SH-SY5Y cells co-expressing myc-tagged Mfn1 and HA-tagged parkin deletion mutants were immunoprecipitated with anti-myc antibody and analyzed by Western blotting (Fig. 4a). Mfn1 interacts robustly with parkin deletion mutants containing the RING1 (R1) and IBR domains (ΔUbl, R1-IBR, IBR-R2 and R1 domains) but only modestly with the RING2 (R2) domain, relative to the input levels of each parkin deletion mutant (Fig. 4a). These data suggest that the RING box domain of parkin is sufficient for the interaction with Mfn1. The effects of PD-associated mutations on the parkin-mediated ubiquitination of Mfn1 were also examined. Cells co-expressing HA-tagged ubiquitin and myc-tagged Mfn1 together with myc-tagged parkin harboring various disease-associated missense mutations were immunoprecipitated with anti-myc antibody and analyzed by Western blotting to evaluate Mfn1 ubiquitination (Fig. 4b). PD-associated mutations in the RING1 (T240R and R256C), IBR (G328E) and RING2 (P437L) domains of parkin markedly impair the ubiquitination of Mfn1 as compared to WT parkin. Collectively, these data demonstrate that parkin interacts with Mfn1 via its RING box domain and that PD-associated mutations located throughout this domain compromise the parkin-mediated ubiquitination of Mfn1.

image

Figure 4.  PD-associated mutations impair the parkin-mediated ubiquitination of Mfn1. (a) Mfn1 interacts with the RING box domain of parkin. SH-SY5Y cells co-expressing combinations of Myc-Mfn1 and HA-parkin deletion mutants were subjected to IP with anti-myc antibody and probed with anti-HA and anti-myc antibodies. Mfn1 interacts robustly with parkin deletion mutants containing the RING1 and IBR domains (ΔUbl, R1-IBR, IBR-R2 and R1), but less well with the RING2 (R2) domain. (*) indicates non-specific protein bands. The domain structure and amino acid positions of parkin deletion mutants are shown, together with the location of PD-associated mutations. (b) PD-associated mutations compromise the parkin-mediated ubiquitination of Mfn1. Cells co-expressing Myc-Mfn1, HA-ubiquitin and Myc-parkin variants (WT, T240R, R256C, G328E or P437L) were subjected to IP with anti-myc antibody and probed with anti-HA and anti-myc antibodies. T240R, R256C, G328E and P437L parkin mutations impair the ubiquitination of Mfn1 compared to WT parkin. Blots are representative of at least three independent experiments.

Download figure to PowerPoint

Parkin mediates the attachment of atypical poly-ubiquitin chains to mitofusin 1

To further understand the implications of Mfn1 ubiquitination induced by parkin over-expression, we evaluated the mode of ubiquitin attachment to Mfn1. Ubiquitin monomers are first covalently conjugated to a protein via linkage of glycine-76 of ubiquitin with a lysine (K) side chain within the substrate (Glickman and Ciechanover 2002). Poly-ubiquitin chains are formed through the sequential linkage of a new ubiquitin monomer to an internal lysine side chain of the previously attached ubiquitin moiety. Ubiquitin contains seven lysine residues that are all capable of supporting poly-ubiquitin chain formation (Xu et al. 2009). Poly-ubiquitin chains linked through K48 target proteins for degradation by the 26S proteasome, whereas K63-linked poly-ubiquitin chains promote an array of non-degradative functions (Komander 2009). The function of K6, K11, K27, K29 and K33-linked chains are not as well characterized whereas proteins can also be modified by single or multiple mono-ubiquitination (Komander 2009). To distinguish between these possibilities, we explored the parkin-mediated ubiquitination of Mfn1 using various lysine mutants of ubiquitin (Ub) (Fig. 5). Compared to WT-Ub, the parkin-mediated ubiquitination of Mfn1 is equivalently supported by K48R-Ub but is prevented by K0-Ub (Fig. 5a). K0-Ub contains all seven lysine residues mutated to arginine and is only capable of mono-ubiquitin attachment. Our data therefore suggests that Mfn1 undergoes poly-ubiquitination via non-K48-linked chains rather than mono-ubiquitination. Next, we performed similar experiments using K48-Ub or K63-Ub that can only support K48- or K63-linked poly-ubiquitin chain formation, respectively, or mono-ubiquitination. Both K48-Ub and K63-Ub effectively block Mfn1 ubiquitination by parkin suggesting that poly-ubiquitin chains are linked through alternative lysine residues (Fig. 5b). Collectively, our data suggests that parkin mediates the robust poly-ubiquitination of Mfn1.

image

Figure 5.  Parkin mediates the atypical poly-ubiquitination of Mfn1. (a) Mfn1 undergoes poly-ubiquitination and not mono-ubiquitination by parkin. SH-SY5Y cells co-expressing combinations of Myc-Mfn1, HA-ubiquitin variants (WT, K48R or K0) and FLAG-parkin, were subjected to IP with anti-myc antibody and probed with anti-HA, anti-myc and anti-FLAG antibodies. K48R-ubiquitin supports normal parkin-mediated Mfn1 ubiquitination compared to WT-ubiquitin whereas K0-ubiquitin (lysine-less mutant) impairs ubiquitination. (b) Mfn1 poly-ubiquitination is independent of K48 or K63 ubiquitin linkages. Similar ubiquitination assay on cells co-expressing Myc-Mfn1, HA-ubiquitin variants (WT, K48 or K63) and FLAG-parkin. K48- and K63-ubiquitin block the normal parkin-mediated ubiquitination of Mfn1. Blots are representative of at least two independent experiments.

Download figure to PowerPoint

Parkin promotes the degradation of mitofusin 1 by the proteasome

To address the consequences of Mfn1 ubiquitination by parkin, we explored the effects of parkin over-expression on Mfn1 turnover and proteasomal degradation (Fig. 6). To ascertain the effect of parkin on the steady-state levels of soluble Mfn1, cells were transfected with a constant amount of Mfn1 in the presence of increasing quantities of parkin. Mfn1 levels are progressively reduced with an increasing dosage of parkin (Fig. 6a). The effect of parkin on Mfn1 levels is partly reduced by proteasomal inhibition with MG132 (Fig. 6a). The reduction of soluble Mfn1 by increasing quantities of parkin is not because of reduced solubility since Mfn1 fails to correspondingly reappear in the detergent-insoluble fraction (Fig. 6a). These data suggest that parkin promotes the degradation of Mfn1, at least in part, by the proteasome. To assess the effect of parkin on Mfn1 turnover, new protein synthesis was inhibited by treatment with CHX and Mfn1 levels were monitored over a subsequent 24-h period. CHX assays demonstrate an increased turnover of Mfn1 because of parkin over-expression (Fig. 6b). Collectively, our data suggest that parkin promotes the turnover of Mfn1 in a manner dependent, at least in part, upon proteasomal degradation.

image

Figure 6.  Parkin enhances the turnover of Mfn1 by proteasomal degradation. (a) Parkin expression reduces the steady-state levels of Mfn1 via the proteasome. SH-SY5Y cells co-expressing Myc-Mfn1 and increasing quantities of FLAG-parkin (0.1–1 μg plasmid DNA) were treated with or without MG132 (5 μM) for 24 h, and equivalent Triton-X-100 (TX)-soluble or TX-insoluble (RIPA-soluble) extracts were probed with anti-myc and anti-FLAG antibodies, or with anti-β-tubulin antibody to demonstrate equal protein loading. Increasing parkin expression results in a progressive reduction of detergent-soluble Mfn1 levels that does not reappear in the detergent-insoluble fraction. Proteasome inhibition impairs the effects of increasing parkin dosage on the steady-state levels of Mfn1. (b) Parkin enhances the turnover of Mfn1. SH-SY5Y cells expressing Myc-Mfn1 with or without FLAG-parkin were treated with cycloheximide (CHX, 200 μg/mL) at 48 h post-transfection, and cells were harvested at 0, 1, 3, 6 and 24 h thereafter. Equivalent detergent-soluble extracts were probed with anti-myc and anti-FLAG antibodies to monitor Mfn1 turnover, or with anti-β-actin antibody to demonstrate equal protein loading. Blots are representative of at least two independent experiments.

Download figure to PowerPoint

Parkin is required for the degradation of mitofusin 1

To examine the role of endogenous parkin on the degradation of Mfn1, we assessed the effects of reducing parkin expression on the steady-state levels of Mfn1. To silence the expression of endogenous parkin in SH-SY5Y cells, we first evaluated shRNA constructs targeting human parkin. Transient transfection of cells with three independent shRNA constructs (PRK1–3) leads to a marked reduction of parkin protein levels compared to a non-targeting control shRNA (Fig. 7a). To examine the effects of parkin silencing on Mfn1 levels, cells co-expressing myc-tagged Mfn1, HA-tagged ubiquitin and shRNAs were assessed by Western blotting. Under conditions known to support Mfn1 ubiquitination and degradation (refer to Fig. 2a), silencing of endogenous parkin expression results in a marked increase in the steady-state levels of exogenous Mfn1 (Fig. 7b). These data suggest that the co-over expression of Mfn1 and ubiquitin is sufficient to induce the degradation of Mfn1, and that endogenous parkin mediates this degradation. We also assessed the impact of parkin silencing on the steady-state levels of endogenous Mfn1 and Mfn2 in SH-SY5Y cells by Western blotting (Fig. 7c). Consistent with recent studies (Gegg et al. 2010; Tanaka et al. 2010), reducing parkin expression fails to alter the levels of endogenous Mfn1 and Mfn2 under basal conditions (Fig. 7c). Similar assays were therefore conducted following treatment of cells with 10 μM CCCP for 1 h to induce mitochondrial damage. CCCP treatment induces mitochondrial depolarization which has been shown to promote the translocation of parkin to mitochondria (Narendra et al. 2008). CCCP treatment induces the ubiquitination and degradation of endogenous Mfn1, whereas reducing parkin expression results in a modest increase in the steady-state levels of Mfn1 (Fig. 7c). Parkin silencing also leads to an increase in the steady-state levels of endogenous Mfn2 following CCCP treatment (Fig. 7c). These data suggest that endogenous parkin mediates the degradation of endogenous Mfn1 and Mfn2 following mitochondrial damage. Notably, whereas the over-expression of Mfn1 and ubiquitin is sufficient to induce the degradation of exogenous Mfn1 in a parkin-dependent manner (Fig. 7b), endogenous Mfn1 (and Mfn2) requires mitochondrial damage to induce its parkin-dependent degradation (Fig. 7c). Collectively, these data demonstrate that endogenous parkin is required for the degradation of Mfn1.

image

Figure 7.  The degradation of Mfn1 is dependent on parkin expression. (a) Validation of parkin-directed short hairpin RNAs. Soluble extracts from SH-SY5Y cells expressing non-targeting control shRNA (Con) or parkin-directed shRNAs (PRK1–3) were probed with anti-parkin and anti-β-tubulin antibodies. PRK1–3 shRNAs lead to a marked reduction of endogenous parkin levels. (b) Effect of parkin silencing on exogenous Mfn1 levels. Soluble extracts from cells co-expressing Myc-Mfn1, HA-ubiquitin and shRNAs (Con, PRK1, PRK3) were probed with anti-myc, anti-parkin and anti-β-tubulin antibodies. Reducing parkin expression with PRK1 and PRK3 shRNAs increases the steady-state levels of Mfn1. (c) Effect of parkin silencing on endogenous Mfn1 and Mfn2 levels. SH-SY5Y cells expressing control (Con) or parkin-directed (PRK1, PRK3) shRNAs were treated with or without 10 μM CCCP for 1 h to induce mitochondrial depolarization. Soluble cell extracts were probed with anti-Mfn1, anti-Mfn2, anti-parkin and anti-β-tubulin antibodies. Reducing parkin expression with PRK1 and PRK3 shRNAs increases the steady-state levels of endogenous Mfn1 and Mfn2 following treatment with CCCP but not under basal conditions. Notice that CCCP treatment leads to the formation of Mfn1-ubiquitin conjugates and a reduction of full-length Mfn1 levels. Blots are representative of at least two independent experiments.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

In this study, we demonstrate that parkin interacts with the mitochondrial fusion factors, Mfn1 and Mfn2, but not with Drp1 or OPA1 in human cells. Parkin mediates the robust ubiquitination of Mfn1 whereas the ubiquitination of Mfn2, Drp1 or OPA1 was not detected. The parkin-mediated ubiquitination of Mfn1 is further enhanced by PINK1. The RING box domain of parkin is sufficient for the interaction with Mfn1 and PD-associated mutations located within this domain impair the parkin-mediated ubiquitination of Mfn1. The ubiquitination of Mfn1 occurs through the covalent attachment of atypical poly-ubiquitin chains rather than through K48- or K63-linked ubiquitin chains or the single attachment of ubiquitin monomers. Parkin promotes the turnover of Mfn1 through enhancing its degradation by the proteasome. Collectively, our data demonstrates that Mfn1 is a substrate of parkin-mediated ubiquitination and provides support for the parkin-dependent regulation of mitochondrial dynamics through the inhibition of mitochondrial fusion. Parkin could induce the autophagic degradation of damaged mitochondria by impairing the fusion machinery and thus favoring fission, which may aid in the selective isolation of damaged mitochondria (Gegg et al. 2010; Poole et al. 2010; Tanaka et al. 2010; Ziviani et al. 2010).

Parkin mediates the poly-ubiquitination and degradation of Mfn1 but not Mfn2 in human cells. However, we observe a small decrease in Mfn2 steady-state levels following the over-expression of parkin suggestive of Mfn2 degradation. It is possible that Mfn2 ubiquitination is below the limit of detection in our assay or that Mfn2-ubiquitin conjugates are labile because of either rapid degradation or de-ubiquitination. Inhibition of the proteasome partly recovers full-length Mfn2 but fails to recover Mfn2-ubiquitin conjugates perhaps suggesting the possibility of modest Mfn2 degradation through an ubiquitin-independent pathway. Reducing parkin expression led to an increase in the steady-state levels of endogenous full-length Mfn2 without evidence for Mfn2 ubiquitination, further supporting the parkin-dependent degradation of Mfn2 through an alternative mechanism. Two recent studies have similarly reported the parkin-dependent ubiquitination of Mfn1 and Mfn2 in human cells (Gegg et al. 2010; Tanaka et al. 2010). These studies demonstrate that the ubiquitination and proteasomal degradation of endogenous Mfn1 and Mfn2 in human cells induced by mitochondrial depolarization is dependent on the expression of parkin and PINK1 (Gegg et al. 2010; Tanaka et al. 2010), a finding also replicated in the present study. Mfn1 ubiquitination was more robust than that of Mfn2 in this study and recent studies (Gegg et al. 2010; Tanaka et al. 2010) perhaps suggesting a preference of parkin for Mfn1 over Mfn2. Our study has largely employed the over-expression of parkin and Mfn1 to induce the parkin-dependent ubiquitination and degradation of Mfn1, which most likely mitigates the requirement for mitochondrial damage to induce a similar outcome. Accordingly, this study and recent studies demonstrate that the parkin-dependent ubiquitination and degradation of endogenous Mfn1 (and Mfn2) requires mitochondrial depolarization (Gegg et al. 2010; Tanaka et al. 2010). Our findings using combined over-expression of parkin and Mfn1 are complementary to and extend recent studies (Gegg et al. 2010; Tanaka et al. 2010) and have permitted the detailed investigation of the effects of PD-associated parkin mutations on Mfn1 ubiquitination, parkin domains interacting with Mfn1, the mode of Mfn1 ubiquitination, and the assessment of Mfn1 turnover and proteasomal degradation. We demonstrate that Mfn1 poly-ubiquitination occurs through atypical lysine-linked chains that are independent of K48 or K63. Although K48-linked poly-ubiquitin chains are well established as mediators of proteasomal degradation (Komander 2009), a recent study suggests that all non-K63-linked poly-ubiquitin chains may target proteins for degradation (Xu et al. 2009). Parkin can also mediate the atypical poly-ubiquitination of VDAC1 predominantly via K27-linked chains which does not appear to mediate the degradation of VDAC1 (Geisler et al. 2010). Parkin is also known to mediate K48- and K63-linked poly-ubiquitination in addition to mono-ubiquitination, supporting the notion that parkin is a multi-purpose E3 ubiquitin ligase (Moore 2006).

Our data suggest that parkin may regulate mitochondrial dynamics by promoting the degradation of mitofusins leading to inhibition of the mitochondrial fusion machinery. Recent evidence suggests that the parkin-mediated degradation of mitofusins may act to prevent the refusion of damaged mitochondria with healthy mitochondria following depolarization-induced mitochondrial fission (Tanaka et al. 2010). Thus, parkin may normally function to limit mitochondrial fusion to aid in the selective isolation and elimination of damaged mitochondria by mitophagy. PD-associated mutations that impair the parkin-mediated ubiquitination and degradation of Mfn1 may result in excessive mitochondrial fusion and impaired mitophagy leading to an accumulation of damaged or dysfunctional mitochondria. It will be important to clarify this potential disease mechanism in future studies.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

This work was supported by funding from the Swiss National Science Foundation (grant no. 310030_127478) and the Ecole Polytechnique Fédérale de Lausanne (EPFL). The authors are grateful to Dr. Manuel Rojo (Université Victor Segalen, France) for providing Mfn2, Drp1 and OPA1 expression plasmids and Mfn1 antibody.

References

  1. Top of page
  2. Abstract
  3. Materials and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References