Covalent ISG15 conjugation positively regulates the ubiquitin E3 ligase activity of parkin

Parkinson's disease (PD) is characterized by selective loss of dopaminergic neurons in the pars compacta of the substantia nigra and accumulation of ubiquitinated proteins in aggregates called Lewy bodies. Several mutated genes have been found in familial PD patients, including SNCA (α-synuclein), PARK2 (parkin), PINK1, PARK7 (DJ-1), LRRK2 and ATP13A2. Many pathogenic mutations of PARK2, which encodes the ubiquitin E3 ligase parkin, result in loss of function, leading to accumulation of parkin substrates and consequently contributing to dopaminergic cell death. ISG15 is a member of the ubiquitin-like modifier family and is induced by stimulation with type I interferons. Similar to ubiquitin and ubiquitination, covalent conjugation of ISG15 to target proteins (ISGylation) regulates their biochemical properties. In this study, we identified parkin as a novel target of ISGylation specifically mediated by the ISG15-E3 ligase HERC5. In addition, we identified two ISGylation sites, Lys-349 and Lys-369, in the in-between-ring domain of parkin. ISGylation of these sites promotes parkin's ubiquitin E3 ligase activity by suppressing the intramolecular interaction that maintains its autoinhibited conformation and increases its cytoprotective effect. In conclusion, covalent ISG15 conjugation is a novel mode of modulating parkin activity, and alteration in this pathway may be associated with PD pathogenesis.

Parkin is an ubiquitin E3 ligase, which promotes the degradation of target proteins through polyubiquitination or modulates their biochemical properties via mono-ubiquitination [3,4]. Parkin consists of an N-terminal ubiquitin-like (UBL) domain, three RING domains and an in-between-ring (IBR) domain [5]. Most pathogenic mutations of parkin are observed within these domains and are associated with a loss-of-function phenotype [6]. Several parkin substrates have been reported, including Pael receptor, the p38 subunit of aminoacyl-tRNA synthetase complex, ataxin-3, synaptotagmin XI, cyclin E, b-catenin and programmed cell death protein 2 isoform-1 [7]. These substrates accumulate when parkin activity is diminished, consequently contributing to dopaminergic cell death [7]. Therefore, tight regulation of parkin-mediated protein ubiquitination is important for cellular proteolysis and neuronal cell viability.
Recent evidence indicates that parkin is essential for elimination of damaged mitochondria via autophagy [8][9][10]. Under conditions of mitochondrial depolarization, PINK1 selectively accumulates on the outer membrane of damaged mitochondria, which results in recruitment and phosphorylation of parkin and ubiquitin. PINK1-activated parkin and phosphorylated ubiquitin then promote ubiquitination and subsequent degradation of many mitochondrial proteins, including mitofusion 1 and 2, TOM20, TOM70 and VDACs [8][9][10]. Parkin-mediated ubiquitination of mitochondrial proteins enhances recruitment of autophagy receptors, such as p62, optineurin and NBR1, resulting in targeting of the damaged mitochondria to LC3-positive autophagosomes for clearance [8][9][10].
A number of studies have shown that parkin's ubiquitin E3 ligase activity is regulated by various post-translational modifications (PTMs), including phosphorylation, S-nitrosylation, sulfhydration, oxidation and dopamine modification [11]. In addition, we have reported that parkin is the conjugation target of multiple UBL modifiers, such as ubiquitin, SUMO and NEDD8. Conjugation to these modifiers affects parkin's activity and/or neuroprotective effects [11][12][13].
ISGylation is the covalent conjugation of the protein interferon (IFN)-stimulated gene 15 (ISG15) to protein targets. ISG15 was first identified as being induced by type I IFNs [14]. It is composed of two N-terminal UBL domains and a C-terminal Gly-Gly motif, which is critical for target-protein conjugation [15,16]. Like ubiquitination, ISGylation also requires E1, E2 and E3 enzymes, each of which is induced by type I IFNs. UBE1 L is the first and only ISG15-activating enzyme [17]. UbcH8 acts as an ISG15-conjugating enzyme when cells are stimulated with IFN, lipopolysaccharide (LPS) or viral infection [18]. Three ISG15 E3 ligases have been reported: EFP, HHARI and HERC5 [19]. Covalently conjugated ISG15 can be removed from target proteins by UBL-specific isopeptidases, which function by mechanisms similar to those of deubiquitinating enzymes. UBP43, an ISG15 de-conjugating enzyme, cleaves isopeptide bonds between ISG15 and the substrate [20]. ISGylation is associated with various cellular functions, including regulation of JNK and NF-kB signal transduction, ubiquitination and antiviral responses [21][22][23]. ISG15 also acts as a cytokine, tumour suppressor and oncogenic factor [24][25][26]. Although more than 300 nascent proteins have been identified as ISGylation targets based on high-throughput proteomics screening [27,28], only a dozen are known to be functionally regulated.
According to the previous reports, low levels of parkin have been detected in blood and immune tissues, including bone marrow, lymph nodes, tonsils and spleen (http:// www.proteinatlas.org/ENSG00000185345-PARK2/tissue). In the central nervous system (CNS), parkin is expressed mainly in neurons, but it has also been detected in a small number of glial cells [29], suggesting that parkin may be a novel target of ISGylation. Here, we demonstrate that parkin is modified by ISG15 conjugation when ISGylation components are overexpressed or when cells are treated with type I IFN, LPS or other selected drugs. In addition, ISGylation enhances parkin's ubiquitin E3 ligase activity, consequently increasing its protective effects against cytotoxicity.
Site-directed mutagenesis was carried out using the QuikChange w XL Site-directed Mutagenesis kit (Stratagene, La Jolla, CA, USA), according to the manufacturer's protocol. The pRK5-HA-parkin and FLAG-tagged HERC5 constructs were amplified by PCR. The plasmids encoding HA-, Myc-or V5-tagged parkin-2KR or Myc-tagged parkin 81 -465 -2KR were generated by two rounds of mutagenesis and PCR amplification with combinations of the K349R and K369R primers. Six HA-tagged parkin mutants each having a single PD-associated amino acid substitution (R33Q, R42P, G328E, R334C, T415N or C418R) were produced by mutagenesis of HA-parkin-WT followed by PCR amplification. Primers used in mutagenesis are listed in the electronic supplementary material table S2.

Cell culture and DNA transfection
Human embryonic kidney 293 (HEK293) cells, mouse embryo fibroblast NIH3T3 cells, HeLa cells, mouse microglial BV2 cells and COS-7 cells were maintained in DMEM containing 10% FBS and 100 units ml 21 penicillin-streptomycin. Cells were grown at 378C in a humidified atmosphere of 5% CO 2 . All DNA transfections were performed using Lipofectamine and PLUS reagents according to the manufacturer's protocol.

RNA interference
Small interfering RNA (siRNA) targeting HERC5 and scrambled control siRNA were designed and synthesized by Bioneer (Seoul, Korea). The sequences of siRNA duplexes are shown in the electronic supplementary material, table S3. HEK293 cells were seeded into six-well plates and transfected with siRNA using Lipofectamine RNAiMAX.

Preparation of cytosolic and nuclear fractions
HEK293 cells were scraped from plates in ice-cold PBS and resuspended in hypotonic buffer (10 mM HEPES ( pH 7.9), 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT and protease inhibitor cocktail). Cells were incubated for 10 min on ice and lysed in 1% NP-40, followed by vortexing for 5 s. Cell lysates were centrifuged at 15 700g for 5 min at 48C. The supernatants were considered the cytosolic fractions. Nuclear pellets were washed in hypotonic buffer and resuspended in hypertonic buffer (27 mM HEPES ( pH 7.9), 2 mM MgCl 2 , 560 mM NaCl, 270 mM EDTA, 33% glycerol, 0.5 M DTT and protease inhibitor cocktail). Samples were then lysed in 1% Nonidet P-40, followed by vortex-mixing four times for 10 s each. Samples were incubated for 20 min on ice, followed by centrifugation at 15 700g for 20 min at 48C. The supernatants were considered the nuclear fraction.

Preparation of cytosolic and mitochondrial fractions
HEK293 cells were scraped from plates in ice-cold PBS and resuspended in ice-cold digitonin buffer (10 mM PIPES ( pH 6.8), 0.015% digitonin, 300 mM sucrose, 100 mM NaCl, 3 mM MgCl 2 , 5 mM EDTA and 1 mM PMSF). Samples were mixed vigorously on a rotating shaker for 10 min at 48C and centrifuged at 480g for 10 min at 48C. The supernatants were considered the cytosolic fraction. Pellets were washed twice with digitonin buffer and resuspended in ice-cold Triton X-100 buffer (10 mM PIPES ( pH 7.4), 0.5% Triton X-100, 300 mM sucrose, 100 mM NaCl, 3 mM MgCl 2 , 5 mM EDTA and 1 mM PMSF). Samples were mixed vigorously on a rotating shaker for 30 min at 48C and centrifuged at 5000g for 10 min at 48C. The supernatants were considered the mitochondrial fraction.

Cell viability assays
After transfection with DNA for 24 h, HeLa cells were left untreated or stimulated with IFN-a for an additional 48 h. Medium was removed and Cell Counting Kit-8 (CCK-8; Dojindo Laboratories, Kumamoto, Japan) solution diluted 1 : 20 in media was added to each well. Plates were incubated for 1 h at 378C, and absorbance at 450 nm was measured using a microplate reader.

Statistical analyses
The statistical significance of cell viability data was determined using unpaired Student's t-tests and SIGMA PLOT rsob.royalsocietypublishing.org Open Biol. 6: 160193 v. 9.0. Values are expressed as the mean + standard error of the mean (s.e.m.).

HERC5 mediates covalent ISG15 conjugation to parkin in mammalian cells
First, we examined whether parkin is modified by ISG15 in mammalian cells. As shown in figure 1a, two slowly migrating forms of parkin were observed in HEK293 cells co-transfected with expression vectors encoding parkin and wild-type ISG15 (ISG15-WT). These two bands suggest that the upper band is due to conjugation of two ISG15 chains, whereas the lower band corresponds to attachment of a single ISG15 chain. However, these bands were not observed in cells co-expressing parkin and ISG15-AA, the conjugationdefective ISG15 mutant that has the C-terminal Gly -Gly residues substituted with Ala-Ala residues (figure 1a). To confirm whether these two bands represent ISG15 -parkin conjugates, we performed co-immunoprecipitation assays of cell lysate with anti-HA IgG, followed by immunoblotting with anti-parkin antibody. As shown in figure 1b, two identically shifted bands were also observed in lysates of cells overexpressing parkin and ISG15-WT by anti-parkin antibody. To rule out the possibility that parkin is not covalently conjugated to ISG15 but instead binds to intracellular ISGylated proteins, we performed co-immunoprecipitation after cell lysates were boiled at 1008C. Under these conditions, the shifted bands were still observed in the cell lysates of cells expressing HA-parkin and FLAG-ISG15-WT ( figure 1c). In addition, these two shifted bands selectively disappeared in the presence of co-expressed UBP43 (figure 1d). Based on the previous report that the principal targets of ISG15 conjugation are nascent polypeptide chains [31], we next analysed parkin ISGylation after cycloheximide treatment of cells. As shown in figure 1e, ISGylated parkin levels decreased after cycloheximide treatment in a time-dependent manner. These results suggest that newly synthesized parkin is a target of covalent ISG15 conjugation.
To further examine whether the action of a specific ISG15-E3 ligase is required for parkin ISGylation, we tested the rsob.royalsocietypublishing.org Open Biol. 6: 160193 interaction between parkin and HERC5. As shown in figure 2a-d, parkin specifically binds to wild-type HERC5 (HERC5-WT) and its catalytically inactive mutant (HERC5-C994A) when the proteins were co-expressed with the three components of the ISG15 conjugation system (i.e. ISG15, UBE1 L and UbcH8). These results indicate that the interaction between HERC5 and parkin might be mediated by another, yet unknown, component. Next, we tested whether HERC5 acts as an ISG15 E3 ligase to promote parkin ISGylation. As shown in figure 2e, wild-type HERC5 markedly enhanced the ISGylation of parkin, whereas HERC5-C994A had no effect. To confirm the specific effect of HERC5 on parkin ISGylation, we evaluated the effect of siRNA-mediated HERC5 knockdown on parkin ISGylation (figure 2f ). When cells were treated with siRNA targeting HERC5, the two parkin-ISG15 conjugate bands were abolished (figure 2g). However, parkin ISGylation was unaffected by treatment with control siRNA ( figure 2g). Lastly, we tested whether another ISG15 E3 ligase, EFP [19], promotes parkin ISGylation. Unlike HERC5, EFP has no effect on parkin ISGylation (figure 2h). These results demonstrate that HERC5 is a specific ISG15 E3 ligase of parkin.

ISG15 is conjugated to the Lys349 and Lys369 residues of parkin
To determine the exact ISGylation site(s) within parkin, several truncated parkin mutants (parkin 81 -465 , parkin 226 -465 , parkin 291 -465 and parkin 381 -465 ) were examined to determine whether they were targets of ISGylation (figure 3a). Co-immunoprecipitation assays revealed that parkin 81 -465 and parkin 226 -465 fragments were ISGylated to the same degree as wild-type parkin (figure 3b,c). In addition, parkin 291 -465 was  rsob.royalsocietypublishing.org Open Biol. 6: 160193 the target of greatly reduced double-ISG15 conjugation, but of a still significant level of mono-ISG15 conjugation. However, substantial ISGylation was not observed with the parkin 381 -465 mutant (figure 3b,c). These results indicate that the region of amino acids 291-380 of parkin is critical for ISGylation. Similar to ubiquitin, the C-terminal Gly-Gly residue of ISG15 is conjugated to the e-amine group of lysine (K) residues on the substrate. Because three lysine residues (K299, K349 and K369) are present in the region of amino acid residues 291-380 of parkin, we next determined which lysine residue is the targeting site of ISGylation. To map the exact ISGylation site(s), we generated three parkin mutants, each with substitution of arginine for one of the lysines (K299R, K349R and K369R) and then tested the effect of the substitutions on parkin ISGylation. The K299R mutant was ISGylated to an extent similar to that of wild-type parkin (figure 3d). By contrast, a significant level of ISGylation of the parkin-K349R and -K369R mutants was not observed. Based on our previous finding that two ISG15 moieties can attach to parkin, these data suggest that K349 and K369 are the specific parkin ISGylation sites. To confirm this hypothesis, we created parkin mutant 2KR in which both K349 and K369 were substituted with arginine. Co-immunoprecipitation assays revealed that ISGylation bands were completely abolished in the 2KR mutant compared with wild-type parkin (figure 3e). To assess the specificity of these lysine residues for HERC5 binding, we performed rsob.royalsocietypublishing.org Open Biol. 6: 160193 co-immunoprecipitation assays of lysates of cells co-expressing wild-type parkin or the 2KR mutant and HERC5. As shown in figure 3f, the parkin-2KR mutant interacts with HERC5, indicating that residues K349 and K369 are important for ISG15 conjugation, but do not affect the binding of HERC5. Taken together, our data suggest that the residues K349 and K369 within the parkin IBR domain are the sites for covalent ISG15 conjugation.

ISGylation does not affect the localization of parkin
We next investigated whether ISGylation affects the biochemical and functional properties of parkin. Although parkin is primarily localized in the cytoplasm in the resting state [32], it moves into the nucleus or mitochondria upon non-covalent protein interactions or PTMs [12,33]. Based on these findings, we first evaluated whether ISGylation changes the subcellular localization of parkin. Consistent with previous findings, overexpression of wild-type parkin resulted in predominantly cytoplasmic localization ( figure 4a,b). Moreover, co-expression of wild-type parkin and ISG15 does not trigger translocation of cytosolic parkin into the nuclear or mitochondrial fraction ( figure 4a,b), suggesting that parkin localization is unaffected by ISG15 conjugation. Overall, although we could not clarify the intracellular location of ISG15-conjugated parkin directly, these data suggest that ISGylation does not affect parkin subcellular localization.

ISGylation increases the ubiquitin E3 ligase activity of parkin
We next examined whether ISGylation affects the ubiquitin E3 ligase activity of parkin. Parkin activity was measured using autoubiquitination assays. As shown in figure 5a, parkin autoubiquitination was increased in the presence of wild-type ISG15. This effect was not observed in the presence of the conjugation-defective ISG15-AA mutant. Next, we determined whether parkin ISGylation on the K349 and K369 residues enhanced its E3 ubiquitin ligase activity. Compared with wild-type parkin, autoubiquitination of the parkin-2KR mutant was significantly reduced in the ISG15-overexpressing cells (figure 5b). Although parkin is self-modified by multiple mono-ubiquitination events, the specific ubiquitination sites have not been identified [4,34]. Therefore, the K349 and K369 residues are likely to be dual targets of ubiquitination and ISGylation; thus, mutation of these sites might affect autoubiquitination of parkin. However, because the pattern of parkin mutant 2KR autoubiquitination was identical to that of wildtype parkin (figures 5b and 8b), these two sites are likely to be specifically modified by ISGylation, not ubiquitination. Previous reports revealed that several PTMs regulate the ubiquitin E3 ligase activity of parkin, thus modulating steady state levels of target proteins, such as p38, TRAF2 and Mfn1/2, through ubiquitination [7]. To verify the positive regulatory effect of ISGylation on parkin activity, we investigated whether ISGylation affects ubiquitination of the parkin target-protein p38. As ISGylation can also alter ubiquitination [26], we tested whether p38 itself could be modified by ISG15 conjugation, as a control. Co-immunoprecipitation assays showed no apparent ISGylated p38 bands in the presence of ISG15 (figure 5c). In addition, when we tested the effect of parkin on p38 ubiquitination, it was greatly increased by wild-type parkin but not by its conjugation-resistant mutant. We then examined the effect of parkin on changes in p38 ubiquitination. Ubiquitination of p38 was enhanced in cells co-expressing wild-type parkin and ISG15 compared with cells expressing wild-type parkin alone (figure 5d). However, p38 ubiquitination was similar in cells co-expressing parkin mutant 2KR and ISG15 and those expressing the 2KR mutant alone ( figure 5d).
Recently, several studies demonstrated that activation and recruitment of parkin to damaged mitochondria involves PINK1-mediated phosphorylation of both parkin and ubiquitin during mitophagy [35][36][37]. Based on these findings, we next investigated whether ISGylation affects PINK1-mediated parkin phosphorylation. Analyses of cell lysates by electrophoresis through Phos-tag gels revealed that PINK1 directly phosphorylates parkin as in cells treated with carbonyl cyanide m-chlorophenyl hydrazone (CCCP), a potent mitochondrial oxidative phosphorylation uncoupler. In addition, parkin phosphorylation by wild-type PINK1 was greatly increased in the presence of wild-type ISG15, but not with the conjugation-defective ISG15-AA mutant (figure 5e). Moreover, this effect was not seen with the PINK1 kinase-deficient mutant, verifying that phosphorylation of parkin is specifically mediated by the action of PINK1 (figure 5e).
Taken together, our data suggest that ISGylation positively regulates the ubiquitin E3 ligase activity of parkin.  . Samples were then immunoprecipitated with anti-HA antibody, followed by immunoblotting with anti-FLAG or anti-HA antibody. Tubulin served as a marker for the cytosolic fraction, histone H3 as a marker for the nuclear fraction and VDAC as a marker for the mitochondrial fraction. rsob.royalsocietypublishing.org Open Biol. 6: 160193

ISGylation inhibits parkin's intramolecular interaction
We next investigated how ISGylation enhances the ubiquitin E3 ligase activity of parkin. According to previous report, native parkin exists in a closed conformation via an intramolecular autoinhibitory interaction between the UBL domain and the PUB motif [38]. Therefore, we examined the effect of parkin ISGylation on this intramolecular interaction. Owing to the technical difficulty inherent in efficiently detecting an To confirm that the intramolecular interaction of parkin is inhibited by ISGylation, we examined whether the parkin 81 -465 fragment was a target of ISG15 conjugation under the same assay conditions as in figure 6a. Co-immunoprecipitation analyses revealed that the parkin 81 -465 fragment was highly modified by ISG15 conjugation (figure 6b). Next, we determined whether parkin ISGylation on the K349 and K369 residues inhibit its intramolecular interaction. Compared with lysates of cells expressing the wild-type parkin 81 -465 fragment, the intramolecular interaction of parkin 81 -465 -2KR was significantly increased (figure 6c). These data demonstrate that ISGylation suppresses the intramolecular interaction that maintains parkin's autoinhibited conformation, which may underlie the ability of ISG15 conjugation to stimulate parkin activity.

Type I interferon enhances parkin E3 ubiquitin ligase activity
Type I IFNs (IFN-a and -b) are essential components of the innate antiviral response [39]. The JAK family of tyrosine kinases is activated when IFNs bind to cell surface receptors, resulting in receptor dimerization [40]. Activated JAKs phosphorylate STAT proteins, and STAT proteins bind IFN response elements at the promoters of IFN-stimulated genes, including ISG15 [41]. Therefore, treatment with type I IFNs induces the accumulation of ISG15 and its conjugating enzyme system, which subsequently increases ISGylation of target proteins. We next investigated whether parkin is a target of ISGylation in response to type I IFN treatment.
Based on a report that treatment with type I IFN induces protein ISGylation in NIH3T3 cells [19], co-immunoprecipitation analyses of NIH3T3 cells showed that endogenous parkin was ISGylated after treatment with IFN-b (figure 7a) as well as with IFN-a (data not shown). As ISG15 is strongly induced by various types of external stimuli, such as virus infection, LPS, genotoxic stresses and retinoic acid [42][43][44][45], we then investigated whether ISG15 conjugation to parkin occurs in response to other selected signals, such as LPS, camptothecin (CPT) and etoposide (Eto). First, we examined whether expression of ISG15 is induced by LPS, CPT and Eto in various cell types. As shown in figure 7b, these stimuli, but not IFN-b, failed to induce ISG15 in NIH3T3 cells. However, BV2 cells treated with LPS and IFN-b showed specific induction of ISG15, whereas COS-7 cells showed induction of ISG15 in response to CPT, Eto and IFN-b (figure 7c,d). We then determined whether parkin is modified by some of these ISG15 inducers. As shown in figure 7e-g, endogenous parkin was ISGylated in response to LPS in BV2 cells, and to CPT and Eto in COS-7 cells. These results suggest the parkin ISGylation occurs in response to selected stimuli in specific cell types.
We next investigated whether type I IFN treatment regulates parkin ubiquitin E3 ligase activity through ISGylation. As shown in figure 8a, parkin autoubiquitination was ). Next, we evaluated the effect of mutating the parkin -ISG15 conjugation sites K349 and K369 on parkin's ubiquitin E3 ligase activity after type I IFN treatment. For these assays, we utilized HeLa cells, which express no endogenous parkin [46], rather than NIH3T3 cells. As shown in figure 8b, parkin autoubiquitination was significantly increased upon IFN-a treatment. However, this increase was not observed in the presence of the parkin-2KR mutant (figure 8b). Finally, we examined the effect of the parkin-2KR mutant on p38 ubiquitination after IFN-a stimulation. Assays of p38 ubiquitination demonstrated that IFN-a treatment, and subsequent increase in parkin ISGylation, also promotes p38 ubiquitination (figure 8c). Accordingly, this effect was greatly diminished in the presence of the parkin-2KR mutant (figure 8c). These results suggest that type I IFN treatment enhances parkin E3 ubiquitin ligase activity through ISGylation.

ISGylation promotes the cytoprotective effect of parkin
Based on the finding that parkin has cytoprotective activity against cell death induced by various toxic agents, including 6-OHDA, H 2 O 2 and rotenone [47], we examined whether ISG15 conjugation affects the cell-protective actions of parkin after treatment with IFN-a at concentrations high enough to trigger cytotoxic effects. First, after HeLa cells were treated with various concentrations of IFN-a, we measured cell viability. As shown in figure 9a, HeLa cell viability decreased dose-dependently in the presence of IFN-a, reaching its maximum effect at 4000 U ml 21 . We further examined whether ISGylation affects the cytoprotective activity of parkin. As expected, HeLa cells transfected with wild-type parkin showed much higher viability in the presence of 4000 U ml 21 IFN-a (figure 9b). In addition, the cytoprotective effect against IFN-a was much lower in cells expressing the parkin-2KR mutant (figure 9b). These results demonstrate that ISG15 conjugation enhances the protective function of parkin against IFN-a-induced cell death through K349 and K369 modifications.  (figure 10a,b). Among the mutations, ISGylation was decreased by more than 70% when mutations occurred in the PUB motif (T415N and C418R). In addition, ISGylation was decreased by more than 40% when the mutation occurred in the UBL (R33Q) or IBR domain (R334C) (figure 10b). However, the extent of parkin ISGylation was increased by more than 50% with the R42P mutant, whereas the G328E mutant showed an almost identical extent of ISG15 conjugation as wild-type parkin (figure 10b). These results suggest that the effect of cytotoxicity by some parkin loss-of-function mutations might be triggered through dysregulation of parkin ISGylation, which may also play a role during PD progression.

Discussion
Although ubiquitin is the best studied post-translational modifier, there is a growing family of UBL proteins that modify cellular targets in pathways that are parallel to, but distinct from, the ubiquitin pathway. These proteins include SUMO, NEDD8, ISG15, FAT10 and Hub1, and have novel functions and influence diverse biological processes. Regarding the link between UBL modification and PD pathogenesis, SUMO-1 expression is increased in the Lewy bodies of PD patient brains as well as in the unilateral rotenone-lesioned mouse model of PD [48,49]. NEDD8 accumulation has also been observed in neuronal and glial inclusions in neurodegenerative disorders [50,51]. Moreover, several familial gene products are modified by UBLs. For example, a-synuclein and DJ-1 are targets of SUMOylation [52][53][54], whereas parkin and PINK1 are regulated by NEDDylation [13,55]. Here, we demonstrated that parkin is a novel target of ISG15 conjugation. Our current work also proposes that ISG15 modification affects toxic protein accumulation by positively modulating parkin activity. These findings suggest that UBL conjugation is important for maintaining neuronal cell viability, and its alteration could promote neurodegenerative diseases, including PD.
According to previous reports, type I IFNs are key components of proinflammatory pathways via JAK/STAT signalling [56]. Inflammation has been implicated in PD pathophysiology, and inhibition of microglial activation attenuates dopaminergic neuron degeneration in animal models of PD. Studies have shown that inflammatory processes modulate PD risk because higher plasma concentrations of various proinflammatory cytokines correlate with an increased risk of developing PD. Regarding the link between parkin and the inflammatory response, loss of parkin function increases the vulnerability of nigral dopaminergic neurons to inflammation-related degeneration [57]. Moreover, parkin mRNA and protein are detectable in brain-resident microglia, astrocytes and peripheral macrophages, and chronic inflammatory conditions reduce parkin levels [58,59]. Specifically, parkin suppresses inflammation and cytokine-induced cell death by promoting proteasomal degradation of E3 ubiquitin-protein ligase TRAF6 [60]. These observations suggest that parkin loss-of-function mutations increase susceptibility to inflammation-mediated degeneration of the nigrostriatal pathway and development of PD. This study reveals a novel relationship between inflammation-related factors and parkin. We demonstrate that excessive inflammatory reactions increase the neuroprotective function of parkin through ISGylation, which contributes to the inhibition of nigral dopaminergic neuronal cell death and consequently prevents PD.
We mapped two specific ISGylation sites within parkin: the K349 and K369 residues. Because poly-ISGylation has not been reported, two ISG15 chains are likely to be individually conjugated to these sites. If two ISG15 moieties were rsob.royalsocietypublishing.org Open Biol. 6: 160193 conjugated at each site, we would have observed a 1X-ISGparkin band when cells overexpressed parkin-K349R or parkin-K369R. However, that band was not observed, and both 2X-and 1X-ISG-parkin bands were abolished. These results suggest that a single substitution at either site changes the tertiary structure of parkin, and therefore affects ISG15 conjugation at the other lysine residue. Native parkin is maintained in an autoinhibited state under resting conditions via intramolecular interactions between the UBL domain and the PUB motif [38]. Here, we showed that ISGylation enhances the ubiquitin E3 ligase activity of parkin. This finding implies that ISGylation might affect parkin autoinhibition and disrupt its inhibitory intramolecular interaction. The activated state of parkin would be triggered and maintained by HERC5. Structural analyses of parkin revealed that the IBR domain is associated with the RING1 domain via a small hydrophobic patch [61]. Moreover, the IBR domain interacts with the RING2 domain over a distance of 50 Å [62]. We demonstrated that the parkin-2KR mutant displays reduced autoubiquitination and thus acts like a dominant-negative mutant. These results suggest that these mutations within the fold of the IBR domain might affect and alter the normal structure of the RING2 domain. This hypothesis is further supported by the finding that intramolecular interaction between parkin 1 -80 , which includes the UBL domain, and the parkin 81 -465 fragment, which includes the PUB motif, is inhibited when cells overexpress these two fragments, as well as ISG15-WT, E1 and E2. By contrast, these interactions are enhanced when ISG15-AA is expressed. Therefore, the IBR domain of the parkin-2KR mutant might also interrupt the structural integrity of parkin and binding to the E2 enzyme.
Parkin autoubiquitination is greatly diminished by certain pathogenic missense mutations, including T415N, C418R, G430D, C431F, M434 K and C441R [4,30]. Among these, T415N and C418R are located in the PUB motif, whereas G430D, C431F, M434 K and C441R are located in the RING2 domain. According to X-ray analyses of parkin structure (PDB code-4K95), R33 and R334 residues are located close to the parkin ISGylation sites (K349 and K369), whereas R42 and G328 are located far from these sites [63]. Therefore, mutations R33Q and R334C might interrupt parkin ISGylation through the conformational change in targeting sites and/or domain. Chaugule et al. [38] previously showed that the R42P mutant disrupts the fold of the UBL domain, which could inhibit the autoinhibitory interaction of parkin with substrates and consequently maintain parkin in a constitutively active state. Constitutively, active parkin may also have easy access to ISGylation machinery for its modification. Based on these findings, we speculate that inhibition of parkin activity by some of these mutants is largely attributed to their stimulatory effects on autoinhibition. Compared with wild-type parkin, these mutants might maintain the intramolecular interactions and subsequent autoinhibited state more strongly. For conversion of the autoinhibited to the activated state, parkin might require specific effector(s) that relieve the autoinhibitory binding of the UBL domain to the PUB motif.
Taken together, our current findings and the presence of diverse PTMs for parkin regulation highlight the functional importance of parkin within the CNS for the maintenance of cellular homeostasis. In addition, our data support the hypothesis that ISG15 and HERC5 have novel roles that contribute to the pathogenesis of PD.