Parkin coregulates glutathione metabolism in adult mammalian brain

Abstract

We recently discovered that the expression of PRKN, a young-onset Parkinson disease-linked gene, confers redox homeostasis. To further examine the protective effects of parkin in an oxidative stress model, we first combined the loss of prkn with Sod2 haploinsufficiency in mice. Although adult prkn^-/-//Sod2^± animals did not develop dopamine cell loss in the S. nigra, they had more reactive oxidative species and a higher concentration of carbonylated proteins in the brain; bi-genic mice also showed a trend for more nitrotyrosinated proteins. Because these redox changes were seen in the cytosol rather than mitochondria, we next explored the thiol network in the context of PRKN expression. We detected a parkin deficiency-associated increase in the ratio of reduced glutathione (GSH) to oxidized glutathione (GSSG) in murine brain, PRKN-linked human cortex and several cell models. This shift resulted from enhanced recycling of GSSG back to GSH via upregulated glutathione reductase activity; it also correlated with altered activities of redox-sensitive enzymes in mitochondria isolated from mouse brain (e.g., aconitase-2; creatine kinase). Intriguingly, human parkin itself showed glutathione-recycling activity in vitro and in cells: For each GSSG dipeptide encountered, parkin regenerated one GSH molecule and was S-glutathionylated by the other (GSSG + P-SH [Formula: see text] GSH + P-S-SG), including at cysteines 59, 95 and 377. Moreover, parkin's S-glutathionylation was reversible by glutaredoxin activity. In summary, we found that PRKN gene expression contributes to the network of available thiols in the cell, including by parkin's participation in glutathione recycling, which involves a reversible, posttranslational modification at select cysteines. Further, parkin's impact on redox homeostasis in the cytosol can affect enzyme activities elsewhere, such as in mitochondria. We posit that antioxidant functions of parkin may explain many of its previously described, protective effects in vertebrates and invertebrates that are unrelated to E3 ligase activity.

Keywords: Early-onset Parkinson disease; Glutathione metabolism; Mass spectrometry; Parkin; Posttranslational modification; Prkn; Redox stress; Sod2.

Fig. 1

Parkin deficiency increases cytosolic hydrogen peroxide in the brain when MnSOD activity is reduced. a Schema of mouse chromosome 17, where prkn and Sod2 loci are separated by 1 centimorgan (cM) and b the breeding strategy used to generate bi-genic prkn^−/−//Sod2^± mice and littermate controls. c Representative Western blot of parkin, MnSOD and actin levels from ~ 3 month-old mouse brains (representative of n = 3 mice/genotype). d Relative MnSOD activity in isolated mitochondria from whole brain lysates of wild-type (WT) and bi-genic littermates, as shown in (c). e–f Ratio of endogenous levels of H₂O₂ (μM) to total protein concentration (μg/μL) in the cortex (e) and midbrain (f) homogenates from 6 month-old mice. g Representative Western blots of constituents from mitochondrial and cytosolic fractions of WT and prkn^−/− mouse brains, with parkin, Dj-1, MnSOD, aconitase-2 and actin as markers (*denotes a non-specific band). h–i Ratio of endogenous levels of H₂O₂ (μM) to total protein concentration (μg/μL) in mitochondria-enriched (h) and cytosolic fractions of the brain (i) from 6 to 8 month-old WT and prkn^−/− animals (left panel), and from 2 to 4 month-old prkn^−/−, Sod2^± as well as bi-genic mice (right panel). Data represent the mean normalized to WT using n = 3/genotype d–f or n = 4–7/genotype (h–i) ± SEM. Significance was determined using unpaired Student T-test d, h and 1-way ANOVA with Tukey’s post-hoc (e, f, i), where *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001

Fig. 2

Parkin lowers chronic oxidative stress-induced damage in the cytosol of mammalian brain. a, b Total protein nitrotyrosination in midbrain homogenates from 6 month-old mice of indicated genotypes, where each lane represents a separate mouse. Ponceau S was used as a loading control to quantify relative nitrotryrosine signals (b). c Total protein nitrotyrosination in cytoplasmic (C) versus mitochondrial (M) fractions from midbrains of 6 month-old mice. d Relative, mean protein carbonyl content in brain homogenates from 6 month-old mice, and e homogenates of human frontal cortices from control subjects and age- and ethnicity-matched, parkin-deficient autosomal recessive Parkinson disease (ARPD) patients as well as from age-matched, non-PRKN-linked parkinsonism (PSM) cases. f Relative, mean protein carbonyl content of brain mitochondria and in g of cytosolic fractions from 6-month-old WT, prkn^± and prkn^−/− mice. Data in b–g represent n = 3–4/genotype ± SEM. Significance was determined using a 1-way ANOVA with Tukey’s post-hoc analysis (b, d–g), where *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001

Fig. 3

Parkin mediates the recycling of oxidized to reduced glutathione, resulting in its own S-glutathionylation. a Silver staining of recombinantly expressed maltose binding protein (MBP; tag only) and MBP-tagged, human parkin proteins separated on SDS/PAGE under reducing conditions. b–d Fluorescence-based quantification of eosin (E)-labelled GSH following incubation of full-length (FL) MBP-parkin (b), MBP-IBR-RING2-parkin (c), or untagged, recombinant (r-) human parkin (d) with 20 mM Di-E-GSSG, monitored over 10 min; (n = 2 runs (a, b, c) in triplicate wells). e Quantification of free GSH levels, as measured in the monochlorobimane assay, following incubation of indicated levels of untagged glutathione (mM) at various GSH:GSSG ratios in the presence of 1 mM of untagged, full-length, human r-parkin (n = 3 ± SEM). A 1-way ANOVA with Dunnett’s post-hoc test was used to compare all values to r-parkin incubated with 10 μM GSH, where *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001. f In vitro S-glutathionylation studies of recombinant parkin, where preparations of 10 mM MBP-IBR-RING2 parkin were treated with 20 mM Di-E-GSSG followed by SDS/PAGE (lanes 1 and 2; both panels). Deglutathionylation studies in the presence of 5 mM DTT alone or in the presence of either 1 mM glutaredoxin 1 (Grx1) or Grx2 (together with: 1 mM NADPH; 5 mM GSH; 0.1 mM glutathione reductase), as indicated (lanes 3–5). Left panel shows a transilluminated gel; the right panel a Coomassie-stained gel. Results are representative of three independent experiments. g (upper) Schema of streptavidin-based enrichment of cellular biotin-labelled S-glutathionylated myc-parkin following treatment of cells with biotin-tagged GSSG (BioGEE) where S-glutathionylated proteins elute from the streptavidin beads in the presence of DTT and are then resolved by SDS-PAGE. (lower) Western blot of S-glutathionylated myc-parkin isolated from CHO-parkin cells either untreated (−) or treated with (+) BioGEE (20 µM; 3 h). Both were exposed to 1 mM H₂O₂ for 10 min prior to lysis. A fraction of input lysate is shown for each condition; high (5 µg) and low (2 µg). h,i Examples of LC–MS/MS-generated spectra following trypsin digestion of MBP-parkin proteins incubated with Di-E-GSSG showing S-glutathionylation (as in f) are shown at two residues: (h) identification of residue Cys59 within human parkin peptide aa 52–75, and in (i) of Cys95 within human parkin peptide aa 90–104

Fig. 4

Prkn expression alters glutathione metabolism in murine brain, including the activity of glutathione reductase. a Tietze method-based quantification of reduced glutathione (GSH), oxidized glutathione (GSSG), the ratio of GSH:GSSG, and total glutathione (GSH + 2GSSG) in mouse brain homogenates of two genotypes, as indicated. b Schema of GSH synthesis pathway, where glutamate cysteine ligase (GCL) is the rate limiting enzyme; GS = glutathione synthase. c Relative GCLC, GCLM and Dj-1 mRNA levels in the brains of 6 month-old mice (–C and –M denote catalytic and modifying subunits respectively; n = 3 mice/genotype). d HPLC-based quantification of GSH, GSSG, the ratio of GSH:GSSG, and the total glutathione pool (GSH + 2GSSG) in brains of 6 month-old mice, as indicated. e, f Western blot analysis of glutathione reductase (GR) protein levels brains from 6 month-old mice, as separated by SDS/PAGE (reducing conditions) using Ponceau S as the loading control, and its quantification by densitometry (n = 6/genotype) in (f). g GR activity in freshly prepared brain homogenates of 6-month-old mice (n = 3/genotype), as indicated. Significance was determined using unpaired Student T-test (a, c, d, f, g) where * represents p ≤ 0.05 and **p ≤ 0.01

Fig. 5

Parkin alters glutathione metabolism in human brain and affects the activity of redox state-dependent enzymes in mice. a HPLC-based quantification of reduced glutathione (GSH), oxidized glutathione (GSSG), the ratio of GSH:GSSG, and total glutathione (GSH + 2GSSG) in cortex homogenates from age-matched human control and PRKN-deficient autosomal recessive PD (ARPD) cases; b, c Western blot analysis of glutathione reductase (GR) protein levels, as separated by SDS/PAGE (reducing conditions) c its quantification by densitometry (normalized to actin) and d GR activity measured in cortex homogenates from the same 8 cases. e Western blot results of aconitase-2 (Aco2), mitochondrial creatine kinase (mtCK), and parkin expression in membrane extracts of the four control and four ARPD cortices (as in b–c). f, g Quantification of relative expression levels of Aco2 and mtCK (shown in e). h Protein levels of murine Aco2, mtCK, Tom20 and VDAC in mitochondrial extracts from wild-type (WT) and prkn^−/− brains of 12 month-old mice, as shown by Western blotting. i Aco2 and j mtCK activities, as measured in freshly isolated mitochondria from WT and prkn^−/− brains with or without exogenous treatment of 4 μM H₂O₂. Data in a-b are plotted as mean values (nmol/µg total protein) ± SEM. Significance was determined using an unpaired Student T-test (a, c, d, f–h) and 2-way ANOVA with Tukey’s post-hoc analysis (i, j), where *p ≤ 0.05, **p ≤ 0.01, and ***p ≤ 0.001

Fig. 6

Working model for parkin-dependent effects on the cytosolic redox state in mammalian brain. Graphical depiction of redox changes identified herein, and as published. Highlighted are: GSH recycling; generation of ROS levels (i.e., H₂O₂; superoxide) leading to parkin’s variable states of oxidation [53]; protein carbonylation [36]; metabolism of nitric oxide (NO) leading to nitrosylation of parkin and nitrotyrosination of proteins [8, 61]; and function of glutathione reductase (GR), as monitored in normal, mammalian brain. P, parkin; -SH, reduced thiol group; P-S-SG, S-glutathionylation of parkin; MnSOD, Mn²⁺-dependent superoxide dismutase (SOD2); GPx, glutathione peroxidase; Grx, glutaredoxin