OUP user menu

Oxidative stress regulates the ubiquitin–proteasome system and immunoproteasome functioning in a mouse model of X-adrenoleukodystrophy

Nathalie Launay, Montserrat Ruiz, Stéphane Fourcade, Agatha Schlüter, Cristina Guilera, Isidre Ferrer, Erwin Knecht, Aurora Pujol
DOI: http://dx.doi.org/10.1093/brain/aws370 891-904 First published online: 25 February 2013

Summary

Oxidative damage is a pivotal aetiopathogenic factor in X-linked adrenoleukodystrophy. This is a neurometabolic disease characterized by the accumulation of very-long-chain fatty acids owing to the loss of function of the peroxisomal transporter Abcd1. Here, we used the X-linked adrenoleukodystrophy mouse model and patient’s fibroblasts to detect malfunctioning of the ubiquitin–proteasome system resulting from the accumulation of oxidatively modified proteins, some involved in bioenergetic metabolism. Furthermore, the immunoproteasome machinery appears upregulated in response to oxidative stress, in the absence of overt inflammation. i-Proteasomes are recruited to mitochondria when fibroblasts are exposed to an excess of very-long-chain fatty acids in response to oxidative stress. Antioxidant treatment regulates proteasome expression, prevents i-proteasome induction and translocation of i-proteasomes to mitochondria. Our findings support a key role of i-proteasomes in quality control in mitochondria during oxidative damage in X-linked adrenoleukodystrophy, and perhaps in other neurodegenerative conditions with similar pathogeneses.

  • i-proteasome
  • oxidative stress
  • X-linked adrenoleukodystrophy
  • very-long-chain fatty acids
  • mitochondria

Introduction

The ubiquitin–proteasome system is pivotal in the rapid clearance of damaged, misfolded or aggregated proteins in both healthy and disease states (Moser, 2001; Goldberg, 2003). In the nervous system, proteasome impairment has been extensively studied in relation to conformational diseases such as polyglutamine diseases, tauopathies and synucleinopathies, which are characterized by an accumulation of insoluble deposits in the affected cells (Ciechanover and Brundin, 2003; Miller and Wilson, 2003). Notably, these conformational diseases involving proteasome malfunction are associated with oxidative stress, and in most cases, oxidative damage arises very early in the course of the pathology (Martinez et al., 2010).

Proteasomes are made up of a catalytic core particle (20S proteasome or 20S catalytic particle), either alone or, more frequently, bound to one or two terminal regulatory particles (Leggett et al., 2002). The 20S catalytic particle is organized into a structure resembling a hollow cylinder composed of two external α and two internal β-heptameric rings; the latter rings form the catalytic chamber, where the subunits β1 (PSMB6), β2 (PSMB7) and β5 (PSMB5) provide caspase-like, trypsin-like and chymotrypsin-like peptidase activities, respectively (for a review, see Ciechanover, 1994). During infection and inflammatory processes, the cytokines LPS, TNF-α and/or IFN-γ induce the expression of three alternate catalytic subunits: β5i/LMP7 (PSMB8), β2i/MECL1 (PSMB10) and β1i/LMP2 (PSMB9). These inducible catalytic subunits may replace their constitutive counterparts and form another catalytic particle known as the immunoproteasome (i-proteasome) (Kloetzel, 2001). i-Proteasomes play an important function in the generation of antigenic peptides for presentation via the major histocompatibility class I pathway, and its formation is a transient and rapid response to the proinflammatory cytokines produced by the innate immune system on infection (Heink et al., 2005; Ferrington and Gregerson, 2012).

In this study, we chose to specifically address the role of the ubiquitin–proteasome system and i-proteasomes in X-linked adrenoleukodystrophy, a neurodegenerative disease in which oxidative stress plays a major role, but is devoid of the accumulation of misfolded insoluble proteins (McKusick No. 300100) (Vargas et al., 2004; Powers, 2005; Fourcade et al., 2008; Galea et al., 2012). X-linked adrenoleukodystrophy is a severe and often lethal neurometabolic disorder characterized by progressive inflammatory demyelination in the brain and/or slowly progressing axonopathy in the spinal cord and peripheral nerves, adrenal insufficiency and the accumulation of very-long-chain fatty acids owing to inactivation of the Abcd1 peroxisomal transporter (Mosser et al., 1993; Moser, 2001; Ferrer et al., 2010). The mouse model for X-linked adrenoleukodystrophy is a classical knockout of the Abcd1 gene (Abcd1 null), which exhibits a late-onset axonopathy in the spinal cord without overt inflammatory features or demyelination, thus resembling the adult onset adrenomyeloneuropathy in humans (Lu et al., 1997; Pujol et al., 2002, 2004). Abcd1 null mice present overt motor disabilities and neuropathological phenotype at 20–22 months of age, although oxidative damage appears very early in life, ∼3 months of age (Fourcade et al., 2008). We recently showed that an excess of very-long-chain fatty acids generates radical oxygen species and oxidative damage to proteins in mice and humans (Fourcade et al., 2008; Galino et al., 2011) and, in a preclinical test, that a combination of antioxidants was able to halt disease onset and arrest progression (Lopez-Erauskin et al., 2011).

Here, we used functional genomic and redox proteomic approaches to investigate the status of the proteasome in an axonal degeneration mouse model of X-adrenoleukodystrophy and in fibroblasts derived from patients with X-linked adrenoleukodystrophy. We found that the ubiquitin–proteasome system and i-proteasome expression and function follow an adaptive response pattern that is modulated by oxidative stress in a cell environment that is devoid of overt inflammatory features. Our study underscores the rationale for using antioxidants as a therapeutic strategy for this and related conditions involving concomitant proteasome malfunction and oxidative stress. Moreover, our data suggest a role for i-proteasomes in mitochondrial protein quality control.

Materials and methods

X-linked adrenoleukodystrophy mice

The generation and genotyping of Abcd1 null mice have been previously described (Lu et al., 1997; Pujol et al., 2002). Mice used for experiments were on a pure C57BL/6J background. Abcd1 null and wild-type mice were separated into control and treated groups. Treated groups were fed a mixture of antioxidants that included 1000 IU/kg vitamin E and 0.5% α-lipoic acid in the diet and 1% N-acetylcysteine in water. Daily feedings were initiated at 8 months of age and were continued for 4 months until sacrifice. Animals were sacrificed and tissues were recovered and conserved at −80°C.

All methods used in this study were in accordance with the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH publication No. 85-23, revised 1996) and the ethical committees of Institut d'Investigació Biomédica de Bellvitge and the Generalitat de Catalunya.

Cell culture and treatment

Primary human fibroblasts were cultivated in Dulbecoo’s modified Eagle medium (containing 10% foetal bovine serum, 100 U/ml penicillin and 100 µg streptomycin) at 37°C in humidified 5% CO2/95% air. Unless otherwise stated, experiments were carried out with cells at 80% confluence. Skin biopsies to prepare fibroblasts were collected according to the institutional guidelines for sampling, including informed consent from the subjects involved or their representatives.

Cells were treated with either single doses of C26:0 (50 µM) or C18:0 (50 µM) in the presence or absence of N-acetylcysteine (1 mM) for 24 h. Proteasome activity was blocked with MG132 (20 µM) or epoxomicin (1 µM) for 4 h before harvesting the cells. Fatty acids were dissolved in ethanol (vehicle) and added to the medium for 24 h. Cell lines were used on passages 12–18.

Microarrays

To explore the transcriptome profiles of adrenoleukodystrophy mutant mice, a 22 K microarray produced at the microarray facility of the Institut de Génétique et de Biologie Moléculaire et Cellulaire, Illkirch, Strasbourg, France, was used (see Supplementary material for details). The microarray experiments were deposited in the Array Express Database under the accession number E-MTAB-79.

Reverse transcription polymerase chain reaction analysis

Total RNA was isolated from homogenized spinal cord using the RNeasy® mini kit (Qiagen), according to the manufacturer’s instructions. The expression of the candidate proteasome genes was analysed by real-time quantitative PCR using TaqMan® Gene Expression Assays (Applied Biosystems) (Supplementary material).

Proteasome activity assay

Tissues were homogenized and centrifuged at 12 000 g for 10 min. Chymotrypsin-like activity was determined using Suc-LLVY-7-amino-4-methylcoumarin as a substrate. Equal amounts of extracts were incubated with the substrate in proteasome activity assay buffer for 30 min at 37°C. The free AMC fluorescence was quantified with a fluorescence multiplate FLUOstar OPTIMA FL reader (BMG) with excitation and emission wavelengths at 380 and 460 nm, respectively (Supplementary material).

Proteasome characterization

To characterize of the proteasome from tissue lysates, the procedure of Elsasser et al. (2005) was followed. For substrate overlays, native gels were incubated in 50 mM Tris–HCl, pH 7.4, 5 mM MgCl2, 2 mM ATP, 50 µM Suc-LLVY-7-amino-4-methylcoumarin and read out by UV excitation (Supplementary material).

Two-dimensional electrophoresis

Two-dimensional analyses were performed as previously described (Galino et al., 2011). Sample preparation is described in the Supplementary material.

Four 2D gels were electrophoresed in parallel. The first gel was Coomassie G250-stained to detect whole proteins, and the three others gels were transferred to nitrocellulose membranes to detect oxidized, poly-ubiquitinated and K48-ubiquitinatediquitinated proteins.

Mitochondrial isolation

Fresh spinal cord was homogenized in sucrose buffer (sucrose 0.32 M, HEPES 4 mM pH 7.4 and protease inhibitor cocktail) using a Dounce with a tight pestle, and then centrifuged for 10 min (1000g). The supernatant (S1a) was kept in ice and the pellet resuspended in ice-cold sucrose buffer, and then centrifuged for 10 min (1000g). The supernatant (S1b) was kept in ice and the pellet resuspended in ice-cold sucrose buffer and centrifuged for 10 min (1000g). The supernatant (S1c) was kept in ice and the pellet discarded. The supernatants S1a, S1b and S1c were centrifuged (16 000 g) for 10 min. The pellet contains the mitochondrial-enriched fractions. Purity of mitochondrial fractions was checked by western blotting with aldolase A and the mitochondrial marker Cox IV antibodies.

Mass spectrometry

Proteins were identified in the Proteomic Unit of the Institut de Recerca Vall d’Hebron (Barcelona) (Supplementary Table 2) as previously described (Galino et al., 2011).

Statistical analysis

Statistical significance (P < 0.05) was assessed using the Student’s t-test whenever two groups were compared. When analysing multiple groups, we used ANOVA and Tukey’s post hoc test to determine statistical significance. Data are presented as mean + standard deviation (SD). A P-value < 0.05 was considered significant (*P < 0.05; **P < 0.01; ***P < 0.001).

Results

Oxidative stress induces increased ubiquitin conjugates levels and regulates proteasome activity during X-linked adrenoleukodystrophy progression

We first measured proteasome activity in 3- and 12-month-old Abcd1 null mouse spinal cords. At 3 months, we found that the chymotrypsin-like activity significantly diminished under ATP-dependent (26S proteasomes) and ATP-independent (20S proteasomes) conditions. The stronger inhibition of activity obtained for the 26S proteasomes than for the 20S proteasomes is consistent with studies showing that 20S proteasome activity is more resistant to oxidative stress than ATP- and ubiquitin-dependent 26S proteasome activity, at least in vitro (Fig. 1A) (Reinheckel et al., 1998). Twelve-month-old wild-type mice exhibited a decrease in proteasome activity; this finding is in line with previous studies showing declines in proteasome activity during ageing (Farout and Friguet, 2006). In contrast, in 12-month-old Abcd1 null mice, the chymotrypsin-like activity from 26S proteasomes increased in comparison with 3-month-old Abcd1 null mice. No alterations in proteasome activity at 3 or 12 months were found in the cortex or liver of Abcd1 null mice, which argues for a selective alteration of proteasome activity in the target organ of the disease (Supplementary Fig. 1A and B).

Figure 1

Oxidative stress induces specific poly-ubiquitinated conjugate accumulation and altered proteasome activity in Abcd1 null mouse spinal cord. (A) The 20S and 26S chymotrypsin-like (CTL) proteasome activity of wild-type (WT) and Abcd1 null (Abcd1) mouse spinal cord were assessed at 3 and 12 months (mean ± SD; n = 7). (B) Treatment with antioxidants (Aox) for 4 months normalized the 26S chymotrypsin-like proteasome activity of 12-month-old Abcd1 mice (mean ± SD; n = 4). (C) Increased poly-ubiquitinated conjugates were detected in Abcd1 mouse spinal cords as early as 3 months. Poly-ubiquitinated conjugates were preferentially K48-linked in 12-month-old Abcd1 mice. No significant difference in K63-ubiquitinated conjugated proteins was observed at 3 and 12 months. γ-Tubulin (γTub), loading control; bottom panels, ratios of poly-ubiquitinated, K48-ubiquitinated and K63-ubiquitinated conjugated proteins relative to wild-type (mean ± SD; n = 6). (D) Antioxidants prevent the accumulation of poly-ubiquitinated conjugated proteins in Abcd1 mouse spinal cords. γ-tubulin (γTub), loading control; right panel, ratios of poly-ubiquitinated proteins relative to wild-type littermates (mean ± SD; n = 4). *P < 0.05; **P < 0.01; ***P < 0.001.

We previously showed that a combination of vitamin E, α-lipoic acid and N-acetylcysteine efficiently reduced reactive oxygen species production in vitro, reversed oxidative damage to proteins and DNA in spinal cords, along and arrested axonal degeneration and disability in X-linked adrenoleukodystrophy mice (Lopez-Erauskin et al., 2011). Interestingly, treatment with this antioxidant cocktail over 4 months normalized the altered 26S proteasome activity (Fig. 1B).

We found a concomitant accumulation of poly-ubiquitinated conjugates in Abcd1 null mouse spinal cord as early as 3 months (Fig. 1C), suggesting that the proteolysis machinery can be altered very early in adulthood in the X-linked adrenoleukodystrophy mouse model. Antioxidant treatment of Abcd1 null mice prevented the increase in poly-ubiquitinated conjugates (Fig. 1D), suggesting that oxidative stress regulates ubiquitin–proteasome system functioning during X-linked adrenoleukodystrophy disease progression. To rule out a direct effect of antioxidants on proteasome activities, we compared chymotrypsin-like activity and poly-ubiquitinated levels between wild-type mice and wild-type mice treated with antioxidants (wild-type + antioxidant). No difference was observed (Supplementary Fig. 1C and D).

The functional role of ubiquitination and, thus, the fate of the modified proteins largely depend on the type of linkage between the ubiquitinated moieties within a poly-ubiquitinated chain (Mukhopadhyay and Riezman, 2007). It is traditionally accepted that proteins modified with K48-linked ubiquitinated conjugates are mainly degraded by the ubiquitin–proteasome system (Thrower et al., 2000; Ikeda and Dikic, 2008), whereas K63-linked ubiquitinated conjugates are degraded by autophagy (Olzmann and Chin, 2008; Tan et al., 2008). To investigate the types of poly-ubiquitinated conjugates, we performed an immunoblot study using specific K48-ubiquitinated and K63-ubiquitinated antibodies. We found that the accumulated poly-ubiquitinated conjugates were preferentially K48-linked in 12-month-old Abcd1 null mouse spinal cords. This correlates with an accumulation of oxidized proteins with time, as we previously reported (Fourcade et al., 2008). Also, accumulation of K48-labelled proteins may indicate that the above described increase of chymotrypsin-like enzymatic activity falls short to degrade the increasing load of damaged proteins. No significant changes in K63 poly-ubiquitinated chains were observed (Fig. 1C), suggesting that autophagic processes were not engaged.

Oxidative stress modulates the expression of the proteasome machinery during X-linked adrenoleukodystrophy disease progression

According to our transcriptomic analysis (Schluter et al., 2012), among 26 proteasome-related genes of the murine genome present in the microarray, the gene expression of 18 subunits was increased at 3 months, whereas the expression of 10 subunits was reduced at 12 months in Abcd1 null mice when compared with wild-type littermates (Fig. 2A and Supplementary Table 1). Notably, 9 of 14 subunits of the 20S proteasome exhibited increased gene expression at 3 months, whereas the expression of four subunits was repressed at 12 months of age. A similar pattern was observed for the 19S regulatory particle; nine subunits were upregulated at 3 months of age, whereas six subunits were downregulated at 12 months.

Figure 2

Oxidative stress alters proteasome subunit expression in Abcd1 null mouse spinal cords. (A) A schematic summary of microarray and western blot results showing proteasome alterations at 3 and 12 months in Abcd1 null (Abcd1) mouse spinal cords. Upregulated and downregulated genes are represented in red and green, respectively. Over-expressed and under-expressed proteins from western blotting experiments are indicated in dark red and dark green, respectively. See also Supplementary Table 1. (B–E) Real-time reverse transcription PCR (B and D) and immunoblot (C and E) analyses of proteasome subunits, using 20S proteasome-, i-proteasome- and 19S regulatory particle-specific primers/antibodies show upregulation at 3 months and downregulation at 12 months of constitutive proteasome subunits and a strong induction of i-proteasome subunits at 12 months in Abcd1 mouse spinal cords. Antioxidant (Aox) treatment reverses the RNA/protein level alterations previously observed in Abcd1 mice (D and E). γ-tubulin (γTub), loading control (mean ± SD; n = 6 for real-time PCR, n = 8 for western blot). *P < 0.05; **P < 0.01; ***P < 0.001.

Validation of these results by quantitative real-time PCR and immunoblot showed that levels of several subunits of the 20S proteasome and the 19S regulatory particle were markedly increased at 3 months and repressed at 12 months in Abcd1 null mice, whereas the levels of the inducible catalytic subunits β1i/LMP2 and β5i/LMP7 were greatly elevated in the spinal cords of 12-month-old Abcd1- mice but not in 3-month-old mice (Fig. 2B and C). The stable levels of the S4 protein of the 19S regulatory particle, as a function of time and genotype (Supplementary Fig. 2A), make this protein a suitable loading control for proteasome contents and argues for a selective modulation of given subunits.

In contrast, no significant changes in any of these protein levels were found in mouse cortex or liver at the same ages (Supplementary Fig. 2B and C), indicating a time-dependent and target organ-specific dysregulation of proteasome expression.

To determine whether proteasome expression is regulated by oxidative stress in Abcd1 null mice, we performed quantitative real-time PCR and western blot experiments with tissue lysates of 12-month-old antioxidant-treated Abcd1 null mice after 4 months of oral antioxidant treatment. The treatment exerted opposing effects on proteasome and i-proteasome subunits; it stimulated the transcription of the constitutive 20S proteasome and 19S regulatory particle subunits but prevented the induction of i-proteasome β1i/LMP2 and β5i/LMP7 subunits (Fig. 2D and E). Moreover, when comparing wild-type and wild-type + antioxidant samples, we could not detect any difference in messenger RNA levels of β2, β1, β5i (LMP7), β1i (LMP2), Rpt5 and Rpn10, nor did we observe differences on protein levels of Rpn2, β5, β5i (LMP7) and β1i (LMP2) protein levels between wild-type and wild-type + antioxidant mice (Supplementary Fig. 2D and E). This suggests that expression of proteasome subunits is not directly regulated by antioxidants under non-pathological conditions. These results are compatible with a replacement of the constitutive proteasome catalytic subunits by their inducible counterparts, as previously shown in amyotrophic lateral sclerosis and Huntington’s disease mouse models (Diaz-Hernandez et al., 2004; Cheroni et al., 2009). Additionally, these results provide insight into the fine-tuning mechanisms of proteasome activity under chronic oxidative stress in vivo, and they may provide an explanation for the increased proteasome activity observed at 12 months of age in Abcd1 null spinal cords.

Very-long-chain fatty acid-induced oxidative stress leads to enhanced levels of K48-linked ubiquitinated conjugates and increased proteasome activity in X-linked adrenoleukodystrophy fibroblasts

Previously, we demonstrated that excess very-long-chain fatty acids generate reactive oxygen species and oxidative lesions to proteins in human X-linked adrenoleukodystrophy fibroblasts (Fourcade et al., 2008). Thus, in the present study, we investigated a possible direct link between an excess of very-long-chain fatty acids and the proteasome dysfunction. Monitoring of the total amount of poly-ubiquitinated conjugates under baseline conditions revealed a significant enrichment in the fibroblasts derived from patients with X-linked adrenoleukodystrophy compared with control fibroblasts (Fig. 3A). After treatment with a pathophysiologically relevant dose of very-long-chain fatty acids (C26:0; 50 µM for 24 h), the levels of poly-ubiquitinated proteins increased in X-linked adrenoleukodystrophy fibroblasts and, to a lesser extent, in control fibroblasts as well. The enhanced level of poly-ubiquitinated conjugates was dependent on the type of fatty acid used; excess oleic acid (C18:1) did not induce any accumulation of poly-ubiquitinated proteins (Fig. 3A). Notably, excess C18:1 did not generate reactive oxygen species at the doses used (Fourcade et al., 2008).

Figure 3

Very-long-chain fatty acid-induced oxidative stress leads to enhanced levels of K48-ubiquitinated conjugates and increased proteasome activity in X-linked adrenoleukodystrophy (X-ALD) fibroblasts. (A) Human control (Ctrl) and X-linked adrenoleukodystrophy fibroblasts were stimulated with very-long-chain fatty acids (C26:0; 50 µM, 24 h) or long-chain fatty acids (C18:1; 50 µM, 24 h). The accumulation of poly-ubiquitinated conjugates was visualized by immunoblotting. γ-Tubulin (γTub), loading control; right panel, the ratios of poly-ubiquitinated proteins relative to the control (mean ± SD; n = 4). (B) 26S chymotrypsin-like (CTL) proteasome activity was analysed in human control (Ctrl) and X-linked adrenoleukodystrophy fibroblasts exposed to very-long-chain fatty acids (C26:0) or long-chain fatty acids (C18:1) (mean ± SD, n = 4). N-acetylcysteine (NAC) normalizes 26S chymotrypsin-like (CTL) proteasome activity in X-linked adrenoleukodystrophy fibroblasts exposed to very-long-chain fatty acids (C26:0) (mean ± SD; n = 4). (C) N-acetylcysteine treatment strongly reduced K48-ubiquitinated conjugate levels in X-linked adrenoleukodystrophy fibroblasts. Human fibroblasts were exposed to very-long-chain fatty acids for 24 h in presence or absence of N-acetylcysteine and were either treated with proteasome inhibitor epoxomicin (Epo) for the last 4 h of the incubation period or left untreated. Proteasome inhibition resulted in a strong accumulation of K48-ubiquitinated conjugates in X-linked adrenoleukodystrophy fibroblasts on exposure to very-long-chain fatty acids. γ-Tubulin (γTub), loading control; bottom panel, the ratios of K48 poly-ubiquitinated conjugated proteins relative to the control (mean ± SD; n = 4). See also Supplementary Fig. 2A. (D) Immunoblots of human control and X-linked adrenoleukodystrophy fibroblasts treated either with or without very-long-chain fatty acids (C26:0) and N-acetylcysteine labelled with 19S subunit Rpn2, 20S subunit β5 and i-proteasome subunits β5i/LMP7 and β1i/LMP2. γ-Tubulin (γTub), loading control; bottom panel, the ratios of distinct subunits relative to the control littermates (mean ± SD; n = 4). See also Supplementary Fig. 2B. *P < 0.05; **P < 0.01; ***P < 0.001.

To confirm that excess very-long-chain fatty acids alter proteasome activity, we performed proteasome assays with cellular lysates of hexacosanoic acid-treated controls and fibroblasts from patients with X-linked adrenoleukodystrophy (Fig. 3B). Under baseline conditions, there was increased 26S chymotrypsin-like activity in X-linked adrenoleukodystrophy fibroblasts. Moreover, we found that excess C26:0 induced this activity in X-linked adrenoleukodystrophy fibroblasts only. In contrast, no differences in proteasome activity were detected in X-linked adrenoleukodystrophy fibroblasts exposed to oleic acid (C18:1) (Fig. 3B). Importantly, the treatment of human fibroblasts with N-acetylcysteine normalized 26S proteasome activity (Fig. 3B).

We next used a complementary assay for proteasomal function, consisting of monitoring degradation of K48 ubiquitinated proteins on excess of very-long-chain fatty acids, using epoxomicin, a specific inhibitor of proteasomes. K48-labelled proteins accumulated both in control and X-linked adrenoleukodystrophy fibroblasts on epoxomicin treatment, as expected. Interestingly, the accumulation of K48-labelled proteins was greater in X-linked adrenoleukodystrophy cells, which may indicate that the ubiquitin–proteasome system is prone to malfunction (Fig. 3C). The same results were obtained with proteasome inhibitor MG132 (Supplementary Fig. 3A). These data suggest that excess hexacosanoic acid stimulates the formation of poly-ubiquitinated chains with K48 linkages that target these ubiquitinated conjugates for proteasome degradation. Moreover, the treatment of X-linked adrenoleukodystrophy fibroblasts with the antioxidant N-acetylcysteine prevented very-long-chain fatty acid-induced accumulation of K48 poly-ubiquitinated conjugate under baseline conditions, and prevented the increase in K48 poly-ubiquitinated conjugates that occurs after inhibition with epoxomicin (Fig. 3C).

These findings support our data obtained from the spinal cords of Abcd1 null mice on treatment with a combination of the antioxidants N-acetylcysteine and lipoic acid (Fig. 1D).

Oxidative stress generated by an excess of hexacosanoic acid modulates the expression of immunoproteasome subunits in X-linked adrenoleukodystrophy fibroblasts

We next addressed the question of the putative effect of excess hexacosanoic acid on proteasome subunit gene expression, which may underlie the observed increase in ubiquitin–proteasome system function. Immunoblot results showed that the levels of the constitutive β5 subunit of the core 20S proteasome were diminished in X-linked adrenoleukodystrophy fibroblasts compared with control fibroblasts. In contrast, the levels of the inducible catalytic subunits β1i/LMP2 and β5i/LMP7 were significantly elevated in X-linked adrenoleukodystrophy fibroblasts in comparison with their controls under basal conditions (Fig. 3D).

Excess hexacosanoic acid induced these subunits to even higher levels in X-linked adrenoleukodystrophy fibroblasts (Fig. 3D), whereas treatment with N-acetylcysteine prevented β1i/LMP2 and β5i/LMP7 induction on exposure to very-long-chain fatty acids, suggesting that i-proteasomes may play an important role in very-long-chain fatty acid-induced oxidative stress response (Fig. 3D and Supplementary Fig. 3B).

Taken together, these data indicate that acute excess hexacosanoic acid produces enhanced levels of poly-ubiquitinated proteins and proteins with K48-linked ubiquitinated conjugates, the upregulation of i-proteasomes subunit expression and concomitant increases in 26S proteasome activity in X-linked adrenoleukodystrophy fibroblasts.

Thus, these results suggest that proteasome activity and subunit expression are regulated by oxidative stress in X-linked adrenoleukodystrophy and underscore the use of anti-oxidants to prevent proteasome dysfunction, particularly i-proteasome induction, during disease progression.

Oxidative stress induces changes in proteasome composition in X-linked adrenoleukodystrophy

The induction of i-proteasome subunits and the concomitant increase in proteolytic activity can be explained by a change in proteasome composition, that is, an increased inclusion of i-proteasome subunits into the complexes substituting the core subunits. Thus, we next determined whether immunoproteasome assembly is induced in X-linked adrenoleukodystrophy using native gel and subsequent in-gel assay and immunoblotting with β5i/LMP7 antibody (Fig. 4 and Supplementary Fig. 4A).

Figure 4

Oxidative stress induces changes in proteasome composition in Abcd1 null spinal cords and human X-linked adrenoleukodystrophy (X-ALD) fibroblasts. (A and B) Proteasome complexes of spinal cord lysates from 3- and 12-month-old wild-type (WT) and Abcd1 null (Abcd1) mice were separated on native PAGE. The proteasome populations were analysed by (A) in-gel substrate overlay assay and (B) immunoblot with β5, Rpn2 and β5i/LMP7 antibodies. (C and D) N-acetylcysteine prevents the proteasome composition changes induced by very-long-chain fatty acids (C26:0) in human X-linked adrenoleukodystrophy fibroblasts. Proteasome complexes of total lysates from human control (Ctrl) and X-linked adrenoleukodystrophy fibroblasts treated with very-long-chain fatty acids (C26:0) with/without N-acetylcysteine for 24 h were separated on native PAGE, and proteasome populations were analysed by (C) in-gel chymotryptic activity and (D) immunoblot with β5i/LMP7 antibody. In native PAGE, arrows denote bands corresponding to the 20S or 26S proteasome. All data are representative of at least three experiments.

We observed two major active bands corresponding to double-capped 26S and 20S proteasomes. Although the 26S and 20S relative bands from Abcd1 null mice exhibited lower signal intensity than the corresponding bands from wild-type mice at 3 months, in agreement with proteasome activity measurements in whole-tissue extracts (Fig. 4A), their relative proportions remained unaltered (Fig. 4B and Supplementary Fig. 4B and C). This result suggests that the impairment of proteasome activity was mainly owing to the inhibition of peptidase activity and not to a change in proteasome assembly. In accordance with this result, we found increased levels of carbonylated proteasome-associated proteins in 3-month-old Abcd1 null mouse spinal cords (Supplementary Fig. 4B and C), suggesting that overloading of the proteasome machinery with an excess of oxidized proteins and/or oxidative damage of proteins functionally related to the proteasome may explain the inhibition of proteasome activity observed at this age. Moreover, it has been shown that proteasome activity impairment induces the transcriptional activation of proteasome genes and de novo formation of matured proteasomes (Meiners et al., 2003). Thus, the early upregulation of constitutive proteasome expression in 3-month-old Abcd1 null mouse spinal cord can be accounted for by a feedback mechanism that seeks to compensate for the reduced proteasome activity observed and may therefore be considered an early adaptive proteasome response.

At 12 months of age, the 20S relative bands were not different, and the 26S band exhibited greater intensity in Abcd1 null mouse spinal cords compared with wild-type spinal cords (Fig. 4A). These findings indicate that the reassembly of the main proteasome complexes may occur before disease onset, at which point, the expression of i-proteasome subunits increases. The reorganization of proteasomes was confirmed by immunoblotting of native polyacrylamide gel electrophoresis (PAGE), which shows the formation of 26S-i-proteasomes (i26S) through the incorporation of the β5i/LMP7 subunit (Fig. 4B).

Next, we studied the composition of proteasome complexes in control and X-linked adrenoleukodystrophy fibroblasts on excess very-long-chain fatty acids using native PAGE (Fig. 4C). Consistent with the enhanced proteasomal activity after 24 h of very-long-chain fatty acids (C26:0) exposure, i26S proteasome levels were strongly enhanced in X-linked adrenoleukodystrophy fibroblasts. The reassembly of proteasomes on very-long-chain fatty acids exposure was determined with immunoblotting experiments that showed greater levels of β5i/LMP7 subunit in i26S proteasomes (Fig. 4D). This increase in the levels of β5i/LMP7 was abolished when cells were co-cultured with N-acetylcysteine, supporting the hypothesis that i26S proteasome formation is a response to very-long-chain fatty acid-induced oxidative stress (Fig. 4C and D).

Mitochondrial proteins involved in energy metabolism are oxidized and K48-linked ubiquitinated conjugated in the spinal cords of X-linked adrenoleukodystrophy mice

Several reports have tied ubiquitin-dependent protein degradation to oxidative stress. Levels of oxidized ubiquitinated proteins have been shown to increase as a consequence of either oxidative modifications of already ubiquitinated proteins or the ubiquitination of oxidized proteins pending degradation (Shang and Taylor, 2011).

To identify potential targets of the ubiquitin–proteasome system, we performed 2D gel analysis to determine whether the oxidant-damaged proteins are targeted by the proteasome for degradation. Interestingly, we identified 17 oxidized poly-ubiquitinated proteins, including 12 proteins of energy metabolism and four cytoskeletal/transport proteins (Fig. 5A and B and Supplementary Table 2). These results are consistent with a previous study, in which we reported the oxidation of the following five key enzymes of glycolysis and the tricarboxylic acid cycle that contribute to bioenergetic failure in Abcd1 null mouse spinal cords: aldolase A, phosphoglycerate kinase, pyruvate kinase, dihydrolipoamide dehydrogenase and mitochondrial aconitase (Galino et al., 2011).

Figure 5

The concomitant presence of K48 poly-ubiquitinated conjugates and increased i-proteasome levels in the mitochondria from Abcd1 null mouse spinal cord. (A) The determination of oxidized proteins (Ox.proteins), poly-ubiquitinated (Poly-ub), K48-linked (K48-ub) ubiquinated conjugates in Abcd1 null (Abcd1) mouse spinal cord lysates by 2D western blot analysis. Coomassie G250-stained 2D gels (left) were used as a loading control. The protein marker is shown on the left. (B) Correlations between oxidized (OX) proteins, poly-ubiquitinated (Ub) and K48-ubiquitinated conjugates (K48) in the different spots are shown. See also Supplementary Table 2. (C) Levels of K48-ubiquitinated conjugates (long and short exposure), i-proteasome β5i/LMP7 and β1i/LMP2 subunits increased in cytosolic and mitochondria fractions from 12-month-old Abcd1 mouse spinal cords in comparison with wild-type littermates. Aldolase A (AldoA), cytosolic purity control; COX IV, mitochondrial loading control; γ-tubulin (γ-tub), loading control; right panel, the ratios of K48-ubiquitinated and both i-proteasome subunits relative to the control littermates (mean ± SD; n = 6). (D) Antioxidant treatment of Abcd1 mice prevents increased K48 poly-ubiquitinated conjugates and i-proteasome levels in mitochondria-enriched fractions. COX IV, mitochondrial loading control; right panel, the ratios of K48-ubiquitinated and β5i/LMP7 and β1i/LMP2 subunits relative to the control littermates (mean ± SD; n = 4).

Interestingly, the following 7 of the 17 oxidized and poly-ubiquitinated proteins were found to be K48 poly-ubiquitinated: (i) one protein of the mitochondrial tricarboxylic acid cycle proteins (aconitase hydratase, ACON); (ii) two mitochondrial enzymes of glutamine metabolism (glutamate dehydrogenase 1 and glutamic-oxaloacetic transaminase 2); (iii) three cytoskeletal/transport proteins (neurofilament polypeptide, alpha-internexin and glial fibrillary acidic protein); and (iv) the heat-shock protein hsc71 (Fig. 5A and B). Notably, this is the first time that the mitochondrial enzymes cited above have been identified as ubiquitination targets. For example, although aconitase has been previously reported as a preferential substrate of the Lon protease (Ngo and Davies, 2007), our results suggest that oxidized aconitase may be degraded in an ubiquitin–proteasome system-dependent pathway. Given the rapid degradation of K48-linked ubiquitinated proteins and the limitations of 2D-redox proteomics experiments, we do not rule out the possibility that larger number of K48-linked ubiquitinated proteins might be identified in Abcd1 null mouse spinal cords.

The data suggest that proteins of energy metabolism, including mitochondria proteins, are oxidized and K48-linked ubiquitinated conjugated in X-linked adrenoleukodystrophy mouse spinal cord. These results are in agreement with recent evidence supporting the ubiquitination and ubiquitin–proteasome system-dependent degradation of proteins localized in the mitochondrial intermembrane space and the inner membrane (Margineantu et al., 2007; Radke et al., 2008; Azzu and Brand, 2010).

The recruitment of immunoproteasomes to mitochondria in Abcd1 null mouse model spinal cords

We previously showed that excess hexacosanoic acid depolarizes the mitochondria of human fibroblasts by decreasing their membrane potentials (Fourcade et al., 2008). Because proteasomes have been shown to localize to dysfunctional, depolarized mitochondria (Chan et al., 2011), we wondered whether i-proteasomes could also be recruited to the mitochondria in X-linked adrenoleukodystrophy, and thus be implicated in K48-linked mitochondrial protein degradation. Therefore, we performed immunoblot analyses with β5i/LMP7, β1i/LMP2 and K48-ubiquitinated antibodies in mitochondrial-enriched fractions from 12-month-old Abcd1 null mouse spinal cords. Our results indicated a substantial increase in K48-linked poly-ubiquitinatediquitination and i-proteasome in mitochondria-enriched fractions of Abcd1 null spinal cords (Fig. 5C). This increase was prevented by antioxidant treatment, suggesting that oxidative stress induced i-proteasome recruitment to the mitochondria in mouse spinal cords (Fig. 5D).

Oxidative stress induced by excess of very-long-chain fatty acids triggers colocalization of immunoproteasome with mitochondria in human X-linked adrenoleukodystrophy fibroblasts

We used X-linked adrenoleukodystrophy fibroblasts to investigate whether oxidative stress mediated by an excess of very-long-chain fatty acids could induce colocalization of i-proteasomes to mitochondria. We performed immunostaining with immunoproteasome β5i/LMP7 antibody and the mitochondrial probe MitoTracker under baseline conditions, exposure to C26:0 or C26:0 plus N-acetylcysteine. Confocal laser scanning microscopy analysis showed a higher degree of colocalization of i-proteasomes with mitochondria in X-linked adrenoleukodystrophy fibroblasts under baseline conditions; this effect was exacerbated by incubation with C26:0. Treatment with N-acetylcysteine abolished this effect, suggesting that i-proteasome localization to mitochondria is triggered by very-long-chain fatty acid-induced oxidative stress (Fig. 6A and B and Supplementary Fig. 5).

Figure 6

i-Proteasomes are recruited to mitochondria in human X-linked adrenoleukodystrophy fibroblasts on oxidative stress. (A and B) The colocalization of i-proteasomes with mitochondria was visualized by confocal immunofluorescence in X-linked adrenoleukodystrophy fibroblasts using the β5i/LMP7 antibody (green) and the mitochondrial marker MitoTracker (red). Scale bar = 40 µM. Nuclei were stained with 4',6-diamidino-2-phenylindole (DAPI) (blue). Images were merged with the DAPI image. Insets: Higher magnification of representative cells. Scale bar = 10 µM. n = 3 per group. (B) The quantification of colocalization was determined using ImageJ software (mean ± SD; n = 4). See also Supplementary Fig. 4. (C) Significant increases in K48-ubiquitinated conjugates and i-proteasomes detected in mitochondrial fractions from human fibroblasts exposed to C26:0 in the presence of epoxomicin (Epo). COX IV, mitochondrial loading control; right panels, the ratios of K48-ubiquitinated conjugated proteins and β5i/LMP7 relative to the control (mean ± SD; n = 4). *P < 0.05; **P < 0.01; ***P < 0.001.

Finally, using the selective inhibitor of proteasome activity epoxomicin, we isolated mitochondrial-enriched fractions from control and X-linked adrenoleukodystrophy fibroblasts that were exposed to excess very-long-chain fatty acids and N-acetylcysteine. We found that mitochondrial levels of K48 poly-ubiquitinated conjugates and i-proteasomes were significantly increased in X-linked adrenoleukodystrophy fibroblasts and, to a lesser extent, in control fibroblasts after exposure to excess hexacosanoic acid (Fig. 6C).

Discussion

The normal brain is an immunologically privileged organ with very low levels of major histocompatibility and i-proteasomes (Xiao and Link, 1998; Piccinini et al., 2003). The induction of i-proteasomes is therefore considered to be a marker of pathological states, is usually viewed as a consequence of the inflammation processes that favour the production of antigenic peptides presented by major histocompatibility class I molecules (Ferrington and Gregerson, 2012) and is also found in chronic neurodegenerative conditions such as Huntington’s disease (Diaz-Hernandez et al., 2003) and Alzheimer's disease (Aso et al., 2012). Emerging evidence suggests that i-proteasomes may be implicated in physiological processes in the nervous system and may have additional non-immune functions in the stress-response pathway (Ding et al., 2006; Kotamraju et al., 2006; Ferrington et al., 2008; Hussong et al., 2010). Recently, it was reported that i-proteasomes participate in the rapid degradation of oxidized ubiquitinated proteins during IFN-γ-induced oxidative stress to cope with the increased formation of aggresomes or aggresome-like induced structures (Seifert et al., 2010). Our findings show that i-proteasomes may play a pivotal role in cytosolic and mitochondrial-oxidized protein homeostasis before axonal degeneration occurs in the X-linked adrenoleukodystrophy model.

We propose the following scenario for the role of i-proteasome induction in X-linked adrenoleukodystrophy (Fig. 7). Initially, an excess of oxidized proteins results in an increase in poly-ubiquitinated proteins, the inhibition of proteasome activity and the subsequent upregulation of constitutive proteasome subunits. Later, constitutive proteasome catalytic subunits are replaced by their inducible counterparts in an attempt to increase the efficiency of the degradation of K48-ubiquitinated conjugated oxidized proteins. Indeed, it has been demonstrated that the incorporation of i-subunits increases proteasomal activity and improves accessibility of the active sites for protein substrates, facilitating efficient poly-ubiquitinated substrate turnover (Gaczynska et al., 1993; Strehl et al., 2008). Therefore, in the absence of overt inflammation features or infection, i26S induction may constitute a secondary adaptive response to chronic oxidative damage that helps maintain cellular homeostasis and viability in X-linked adrenoleukodystrophy.

Figure 7

Oxidative stress regulates i-proteasomes in X-linked adrenoleukodystrophy. Very-long-chain fatty acids induce oxidative stress and the accumulation of defective proteins in X-linked adrenoleukodystrophy. At the early stage of the disease (3 months), a first adaptive proteasome response occurs through the upregulation of constitutive subunits. However, oxidative stress leads to the accumulation of poly-ubiquitinated proteins and inhibition of the proteasome machinery. Later (12 months), i-proteasomes subunits are induced, which increases the degradation of cytosolic and mitochondrial K48-linked ubiquitinated-oxidized proteins and prevents the cytotoxic effects due to their accumulation. VLCFA = very-long-chain fatty acids.

Mitochondria are a major source of reactive oxygen species in cells. As in with most other neurodegenerative diseases, mitochondrial dysfunction and oxidative stress are intertwined in X-linked adrenoleukodystrophy (Fourcade et al., 2008; Galino et al., 2011; Lopez-Erauskin et al., 2012). The accumulation of oxidized and ubiquitined conjugates of mitochondrial proteins likely represents a key early cellular response during neuronal stress, since they have been reported to accumulate in the mitochondria of cortical and hippocampal neurons following cerebral ischemia (Hayashi et al., 1992).

Several recent reports have demonstrated the importance of the ubiquitin–proteasome system in the clearance of damaged mitochondria (Tanaka et al., 2010; Chan et al., 2011; Yoshii et al., 2011). In addition, proteasomes are required for mitochondrial homeostasis and functioning during quiescence; proteasome inactivation results in an accumulation of reactive oxygen species and diminished mitochondrial function (Takeda and Yanagida, 2010). This finding suggests that the ubiquitin–proteasome system-dependent elimination of oxidized mitochondrial proteins may represent a first line of defense against proteotoxicity in the mitochondria. In Abcd1 null mouse spinal cords, we found a clear correlation among mitochondrial dysfunction, enhanced level of mitochondrial-oxidized K48 poly-ubiquitinated proteins, and i26S proteasome induction as well as enrichment of i-proteasomes in the mitochondrial fractions. While this manuscript was in preparation, the mitochondrial localization of i-proteasomes was described in chronic steatohepatitis (French et al., 2011), which supports our findings. Although further studies will be required to confirm our data, we provide here in vivo evidence suggesting the involvement of i26S proteasomes in the mitochondrial protein quality control process.

In previous studies, the combination of antioxidants tested was able to halt axonal degeneration in Abcd1 null mice by reversing the oxidative damage to proteins, including mitochondrial proteins, and other key proteins involved in energy metabolism in whole spinal cords (Galino et al., 2011; Lopez-Erauskin et al., 2011). These effects are accompanied by the induction of constitutive proteasome subunits from the catalytic core and the 19S regulatory particle, normalization of proteasome activity and decreases in poly-ubiquitinatediquitination levels. Moreover, antioxidant treatment prevents not only i-proteasome induction in Abcd1 null mouse spinal cords and in X-linked adrenoleukodystrophy fibroblasts on very-long-chain fatty acids exposure but also, importantly, the translocation of i-proteasomes to presumably dysfunctional mitochondria. These results suggest a mechanism of action by which antioxidants exert their pleiotropic effect in vivo that deserves further investigation, given its implications for several neurodegenerative conditions in which oxidative stress plays a role (Barnham et al., 2004; Bossy-Wetzel et al., 2004).

Funding

European Commission (FP7-241622); the European Leukodystrophy Association (ELA2009-036C5; ELA2008-040C4); the Spanish Institute for Health Carlos III (FIS PI080991 and FIS PI11/01043 );and the Autonomous Government of Catalonia (2009SGR85 to A.P.). The study was developed under the COST action BM0604 (to A.P.). S.F. was a fellow of the European Leukodystrophy Association (ELA 2010-020F1) and the Spanish Institute for Health Carlos III (Miguel Servet program CP11/00080). The CIBER on Rare Diseases (CIBERER) and CIBER on Neurodegenerative Diseases (CIBERNED) are initiatives of the ISCIII. FIS ECA07/055 to A.S.

Supplementary material

Supplementary material is available at Brain online.

References

View Abstract