Research Article

Ubiquitination of a New Form of α-Synuclein by Parkin from Human Brain: Implications for Parkinson's Disease

See allHide authors and affiliations

Science  13 Jul 2001:
Vol. 293, Issue 5528, pp. 263-269
DOI: 10.1126/science.1060627

Abstract

Parkinson's disease (PD) is a common neurodegenerative disorder characterized by the progressive accumulation in selected neurons of protein inclusions containing α-synuclein and ubiquitin. Rare inherited forms of PD are caused by autosomal dominant mutations in α-synuclein or by autosomal recessive mutations in parkin, an E3 ubiquitin ligase. We hypothesized that these two gene products interact functionally, namely, that parkin ubiquitinates α-synuclein normally and that this process is altered in autosomal recessive PD. We have now identified a protein complex in normal human brain that includes parkin as the E3 ubiquitin ligase, UbcH7 as its associated E2 ubiquitin conjugating enzyme, and a new 22-kilodalton glycosylated form of α-synuclein (αSp22) as its substrate. In contrast to normal parkin, mutant parkin associated with autosomal recessive PD failed to bind αSp22. In an in vitro ubiquitination assay, αSp22 was modified by normal but not mutant parkin into polyubiquitinated, high molecular weight species. Accordingly, αSp22 accumulated in a non-ubiquitinated form in parkin-deficient PD brains. We conclude that αSp22 is a substrate for parkin's ubiquitin ligase activity in normal human brain and that loss of parkin function causes pathological αSp22 accumulation. These findings demonstrate a critical biochemical reaction between the two PD-linked gene products and suggest that this reaction underlies the accumulation of ubiquitinated α-synuclein in conventional PD.

Parkinson's disease is a highly prevalent neurodegenerative disorder that causes progressive motor dysfunction, variable cognitive impairment, and shortened life expectancy (1). The molecular pathogenesis of PD remains unclear, but genetic factors play a role in some cases. The genes encoding parkin (2), α-synuclein (αS) (3, 4), and ubiquitin carboxyl-terminal hydrolase (UCH)-L1 (5) have each been linked to familial forms of PD. Missense mutations in αS or UCH-L1 cause rare, autosomal dominant forms of PD. In contrast, mutations of parkin are a relatively common cause of autosomal recessive PD (ARPD), which often has early onset (6, 7). As is the case in conventional (“idiopathic”) PD, the neuropathologic changes of parkin-linked ARPD are largely confined to the brainstem and include loss of selected neurons and local gliosis. However, the αS- and ubiquitin-positive neuronal inclusions (Lewy bodies) that are a hallmark of conventional PD (8) are generally absent in parkin-linked ARPD (9-12).

Parkin consists of an NH2-terminal ubiquitin-like (Ubl) domain and a COOH-terminal RING box separated by a linker region (2, 13). The RING box encompasses three domains, termed RING1, IBR (in-between-RING), and RING2. Parkin was recently found in cell culture to act as a ubiquitin (Ub) ligase whose RING box recruits the E2 Ub conjugating enzymes, UbcH7 and UbcH8 (13-15). Ubiquitination is a vital cellular process by which a large variety of cellular proteins (targets or substrates) are conjugated with multimers of Ub, marking them for degradation by the proteasome (16). Conjugation requires a cascade of reactions that includes an E1 Ub activating enzyme, an E2 Ub conjugating enzyme, and an E3 Ub ligase. The E3 specifies both E2 recruitment and the recognition and binding of the substrate (16). As an E3 ligase, parkin conjugates Ub onto its unknown substrate(s) (designated “target X”) for subsequent degradation by the proteasome (13). It has also been shown that parkin can associate with CDC-rel1 (14) and actin filaments (17) in cultured cells.

The loss of functional parkin molecules in ARPD should lead to the gradual accumulation of nonubiquitinated substrates that would otherwise be polyubiquitinated by this E3 ligase and efficiently degraded (13). In this regard, the absence of Lewy bodies in ARPD brains suggested to us that both functional parkin and its unknown target(s) may be required for the formation of Lewy bodies.

Parkin colocalizes with α-synuclein in brainstem Lewy bodies.To test this hypothesis, we raised and purified high-affinity polyclonal antibodies to several regions of human parkin (18) and used these to probe Lewy bodies both immunohistochemically and biochemically. Staining with several antibodies to parkin (anti-parkin antibodies), e.g., HP2A [to amino acids (aa) 342–353], revealed that parkin epitopes specifically co-localized with αS in subsets of classical brainstem Lewy bodies found in both PD (Fig. 1A) and the related neurodegenerative disorder, dementia with Lewy bodies (DLB) (not shown). Anti-parkin positive Lewy bodies were also detected in sections of cingulate gyrus from DLB cortex and of sympathetic gangliocytes in the peripheral autonomic nervous system from PD (18). Moreover, affinity isolated Lewy bodies were found to contain intact 53-kD neural parkin. These and related biochemical findings on the subcellular distribution of a pool of parkin, αS, and UbcH7 to the same highly purified presynaptic fractions (18) led us to pursue the hypothesis that αS is a key substrate for the E3 ligase activity of parkin in human brain and that disease-associated mutations in parkin prevent the ubiquitination of αS.

Figure 1

Association of parkin and α-synuclein in human brain. (A) Immunohistochemistry on sections of PD substantia nigra. Intracellular Lewy bodies (arrows) are immunoreactive with anti-parkin (HP2A) and anti-αS (LB509) (18, 20). Competition with peptide antigen (HP2A absorbed) abolishes Lewy body staining but not the appearance of surrounding melanin granules. Bar, 10 μm. (B) Brain homogenates of frontal cortex of four control cases were immunoprecipitated (IP) with parkin antibodies HP1A or HP2A. HP1A* and HP2A* signify preabsorption of each antibody with its corresponding parkin peptide. The precipitates were analyzed by Western blots (WB) with HP6A, syn-1, or anti-UbcH7, as indicated. “Crude” signifies an aliquot of the starting homogenate for each IP. Double asterisks indicate partially denatured immunoglobulin G (IgG). (C) Homogenates of either frozen or fresh frontal cortex of control #2 processed with or without 0.2% NP-40 were immunoprecipitated with HP2A (or parkin-preabsorbed HP2A*), and the precipitates were blotted with HP6A, LB509, or anti-UbcH7, as indicated. Of note, αSp22 migration varies according to Tris-glycine gel conditions, ranging from 21 kD (10 to 20%) to 24 kD (4 to 20%).

Wild-type parkin binds UbcH7 and a new form of α-synuclein in human brain. To determine whether parkin interacts with αS in human brain, we carried out co-immunoprecipitiation experiments on homogenates of frozen frontal cortex from four normal brains (19). We used parkin antibodies HP1A (to aa 84–98) and HP2A for immunoprecipitation (19), followed by immunoblotting with either parkin antibody HP6A (to aa 6–15) (18), αS antibodies syn-1 and LB509, or an antibody to UbcH7 (20). In all cases, the 53-kD parkin protein was specifically precipitated by HP1A and HP2A, whereas preabsorbed HP1A and HP2A failed to do so (Fig. 1B, top panels). The reported E2 binding partner of parkin, UbcH7 (13, 15), co-precipitated with parkin (Fig. 1B, bottom panels), as expected. HP1A and HP2A also co-precipitated a new 22-kD isoform of αS (designated αSp22) but not the abundant 16-kD αS monomer (αSp16), as shown by two well-characterized αS antibodies, syn-1 (Fig. 1B, middle panels) and LB509 (not shown). αSp22 was never co-precipitated by preabsorbed parkin antibodies, and it was not detected when anti-parkin precipitates were blotted with syn-1 or LB509 that had been preabsorbed with recombinant human αS (data not shown) (20). These immunochemical results were confirmed by mass spectrometry of trypsin digests of the excised αSp22 band, which yielded multiple tryptic fragments of human αS (21).

To determine whether the observed association of parkin and αSp22 was influenced by freezing of the human tissue or by the use of detergents, we examined fresh human brain specimens obtained less than 18 hours after death (Fig. 1C). When these samples were processed in the presence or absence of 0.2% NP-40 and in parallel with a frozen specimen of the same brain, we observed indistinguishable co-precipitation of αSp22 and UbcH7 (Fig. 1C, middle and bottom panels) with parkin (Fig. 1C, top panels) in all three conditions. We conclude that in normal human brain, parkin, an E3 Ub ligase, is part of a stable protein complex that includes a known E2 protein, UbcH7, and a distinct αS isoform, αSp22.

We next examined whether mutant parkin proteins from four parkin-linked ARPD brains can bind αSp22 and UbcH7. ARPD cases 1 and 2 each carry a homozygous deletion of exon 4, which should lead to premature termination of parkin at aa 143 (2,22–24). ARPD cases 3 and 4 each carry a homozygous deletion of exon 3, which should cause premature termination of parkin at aa 96 (2, 23,24). Therefore, because these truncated mutant parkin proteins lack the epitope (aa 342–353) recognized by HP2A, this antibody could neither immunoprecipitate nor blot mutant parkin from any of the four ARPD brains (Fig. 2A, top panels), as expected. However, when we probed aliquots of the starting ARPD homogenates (crude) with HP6A (to aa 6–15) or HP7A (to aa 51–62), our two most NH2-terminal parkin antibodies (18), we unexpectedly failed to detect any mutant parkin proteins (Fig. 2A, top panels; data not shown). Thus, the truncated parkin proteins may be unstable and rapidly degraded in the brains of these four ARPD patients. Consistent with the lack of detectable parkin immunoreactivity, HP2A immunoprecipitates of the ARPD homogenates did not contain αSp22 or UbcH7 (Fig. 2A, lower left panels). However, UbcH7 and the unmodified αSp16 monomer were readily detected by immunoblotting of the crude ARPD brain homogenates (Fig. 2A, lower right panels), as expected.

Figure 2

Wild-type parkin specifically interacts with αSp22 and UbcH7 in human brain. (A) Lack of detection of parkin in ARPD brain. We analyzed HP2A immunoprecipitates from frontal cortex homogenates of ARPD cases #1 through 4 and control #2. The immunoprecipitates and crude extracts (starting material for the IPs) were evaluated by WB with HP2A, HP6A, LB509, or anti-UbcH7, as indicated. (B) Homogenates of normal human brain (control #2) were immunoprecipitated with antibodies to E6-AP or parkin (HP2A or preabsorbed HP2A*). The precipitates were immunoblotted with anti-E6-AP, HP6A, LB509, or anti-UbcH7 as indicated. (C) HP2A immunoprecipitates of normal brain were analyzed by Western blot with anti-UbcH7 that was either unabsorbed (unrelated peptide), or preabsorbed with recombinant UbcH7 or UbcH8.

Next, we examined the specificity of parkin versus another E3 Ub ligase of human brain for binding αSp22. A previously characterized antibody to E6-AP (25), an E3 Ub ligase that also recruits the E2 protein UbcH7 and is associated with Angelman syndrome (26,27), was used in parallel immunoprecipitation experiments with HP2A, followed by blotting with antibody to αS (anti-αS) LB509 or antibody to UbcH7 (anti-UbcH7). As expected, UbcH7 was co-precipitated from normal brain homogenates by the E6-AP antibody (Fig. 2B, bottom panel). In contrast, neither αSp22 nor the unmodified αSp16 were co-precipitated by the E6-AP antibody from brain homogenates (Fig. 2B, middle panel), demonstrating the specificity of the interaction between parkin and αSp22 in normal human brain.

To verify the identity of the endogenous E2 protein that associates with parkin in human brain (that is, to distinguish between the highly homologous proteins UbcH7 and UbcH8), we preabsorbed aliquots of the monoclonal UbcH7 antibody with either purified, recombinant UbcH7 or UbcH8 and used them to blot HP2A immunoprecipitates (28). Only when anti-UbcH7 was preincubated with UbcH7 peptide, not with UbcH8 peptide or an irrelevant peptide, was the co-immunoprecipitated band at 17-kD abolished (Fig. 2C). These data demonstrate the specificity of anti-UbcH7 and indicate that parkin associates specifically with UbcH7 as an E2 enzyme in human brain.

Wild-type but not mutant parkin proteins ubiquitinate αSp22.In order to establish a functional role for parkin in human brain, we obtained immunoprecipitated (IP) parkin from frontal cortex homogenates and tested its E3 Ub ligase activity in a previously described in vitro assay (Fig. 3) (13). Normal brain-derived IP parkin was incubated at 37°C for 30 min with recombinant E1, excess UbcH7 as E2, and His-tagged Ub in the presence of adenosine triphosphate (ATP) (29). The reaction products were then probed with antibody to His (anti-His) for Ub and with LB509 for αS. High molecular weight (Mr) His-ubiquitinated proteins were observed in the presence of IP parkin, whereas no signal was noted in the absence of either IP parkin or E1 or UbcH7 (Fig. 3, left panels). These data confirm that endogenous parkin is an E3 Ub ligase for human brain–derived substrates, consistent with similar findings obtained in human neuroblastoma cells transfected withmyc-parkin cDNA (13). More important, αSp22, which co-immunoprecipitated with parkin as expected, was consumed as the high Mr ubiquitinated smear appeared (detected by anti-αS LB509) in the presence of E1, E2, IP parkin, and ATP (Fig. 3, right panels). In contrast, αSp22 was not further modified in the absence of either E1 or UbcH7. The loss of αSp22 and simultaneous gain of the high Mr αS-reactive smear of proteins during the incubation and the positive immunoreaction of the latter with both anti-His and anti-αS antibodies strongly suggest that these species represent, in part, polyubiquitinated forms of αSp22 [αSp22-Ubn]. Furthermore, the individual reactions shown in Fig. 3 demonstrate that brain-derived IP-parkin, in conjunction with its E2 binding partner, UbcH7, is necessary for this Ub ligase activity.

Figure 3

Neural parkin exhibits E3 ubiquitin ligase activity. Detection of ubquitination activity of IP-parkin from frontal cortex homogenate of control brain #1. IP-parkin was obtained as inFig. 1 and incubated with ATP, His-Ub, and the indicated combinations of E1 and E2. High Mr His-Ubn immunoreactivity was detected by anti-His (left panels), αS reactivity by LB509 (right panels). IP with HP1A (top panels) or HP2A (bottom panels) are shown. The last lanes in each panel show abolition of the ubiquitination reaction when parkin antibodies were preabsorbed with their cognate peptide. Asterisk signifies a His-reactive 30-kD band that is not recognized by LB509 and may be a distinct substrate for parkin's E3 Ub ligase activity. We obtained identical results in homogenates of control brain #2.

On the basis of these findings, we hypothesized that mutant parkin proteins resulting from parkin missense mutations in ARPD might fail to polyubiquitinate αSp22 for one of two reasons: (i) failure to bind αSp22 as a substrate at parkin's NH2-terminal Ubl domain or (ii) failure to recruit parkin's E2 binding partner, UbcH7, at its COOH-terminal RING box. Therefore, we generated recombinant myc-parkin Arg42 → Pro42 (R42P) with the Arg to Pro mutation in the Ubl domain and recombinant myc-parkin Thr240 → Arg240 (T240R) with the Thr to Arg mutation in the RING1 domain (30), in addition to wild-type myc-parkin (Fig. 4A). These parkin proteins were each expressed in transiently transfected HEK293 cells, immunoprecipitated with antibody to myc (Fig. 4B), and added to homogenates of frontal cortex from normal human brain. Exogenously expressed wild-type parkin was able to bind both αSp22 (Fig. 4C) and UbcH7 (Fig. 4D). In contrast, myc-parkin R42P was able to recruit UbcH7 but not αSp22, and myc-parkin T240R recruited αSp22 but not UbcH7 (Fig. 4, C and D). These findings are consistent with the reported inability of myc-parkin T240R to mediate recruitment of UbcH7 and the inability of myc-parkin R42P to interact with “target X” in cultured neuroblastoma cells (13). Furthermore, the mutant parkin proteins also failed to generate αSp22-Ubnconjugates. The IP-myc-parkin proteins were mixed with homogenates of normal frontal cortex and incubated at 37°C for 30 min with recombinant E1, UbcH7 as E2, and His-Ub in the presence of ATP (29). In this reaction, only wild-type IP-myc-parkin conferred E3 Ub ligase activity, while the mutant IP-myc-parkin fusion proteins showed no detectable Ub ligase activity (Fig. 4, E and F). We conclude that exogenously generated wild-type but not ARPD-associated mutant myc-parkin, when added to human brain, can confer in vitro assembly of the myc-parkin/UbcH7/αSp22 complex as well as E3 Ub ligase activity that posttranslationally modifies αSp22.

Figure 4

Myc-parkin fusion proteins interact with αSp22 and UbcH7 from normal brain. (A) Diagrams of wild-type (WT) and R42P and T240R mutant parkin proteins. (B, C, and D) Myc-tagged wild-type and mutant human parkin proteins were expressed in HEK293 cells. Extracts (10 mg) were immunoprecipitated with anti-myc (13). The immunoprecipitates were incubated with (+) or without (–) frontal cortex homogenates (control case #1), washed and analyzed by WB with anti-myc (9E10) (B), anti-αS LB509 (C), or anti-UbcH7 (D). (E and F) Exogenous myc-parkin exhibits Ub ligase activity and conjugates His-Ub onto αS as αSp22 is consumed (in vitro ubiquitination assay performed as in Fig. 3). Asterisk signifies non-specific bands. We obtained identical results in homogenates of control brain #2.

αSp22 accumulates in parkin-deficient ARPD brain. Our findings raised the possibility of abnormal substrate (αSp22) accumulation in parkin-linked ARPD brains. To investigate this question, we first affinity-enriched for endogenous αS isoforms in normal and four parkin-deficient ARPD brains (Fig. 5). Immunoprecipitates obtained with the polyclonal anti-αS antibody, KC7071 (20), were analyzed by blotting with anti-αS LB509. Under these experimental conditions, we detected αSp22 solely in ARPD brains, while the regular αSp16 monomer (KC7071's actual antigen) was seen in large and comparable amounts in both ARPD and normal tissue (Fig. 5A, upper left panel). Neither αS isoform was precipitated from ARPD brain (case #2) by preimmune serum (Fig. 5A, upper right panel). These data suggested that αSp22 accumulates in ARPD brains relative to control brains, whereas the amount of unmodified αSp16 remains similar. To confirm this observation, we performed parallel pull-down assays (as described in Fig. 4C) by adding aliquots of wild-type IP myc-parkin to ARPD and control homogenates. After immunoblotting with LB509, we detected increased amounts of αSp22 in all four ARPD brains when compared with the co-migrating αSp22 species pulled down from the four control brains and the PD and DLB brains (Fig. 5A, lower panel). Therefore, the genetically and biochemically documented loss of parkin protein in our ARPD brains leads to the specific accumulation of its target substrate, αSp22. We believe that the low amounts of αSp22 in non-ARPD brains and its interaction with endogenous parkin prevented our conventional αS antibody from precipitating it in sufficient amounts to be visible on Western blots.

Figure 5

αSp22, a glycoprotein, accumulates in ARPD brains. (A) Upper panel, anti-αS KC7071 immunoprecipitates obtained from 500 μg of frontal cortex homogenates of control cases #1 through 4, ARPD brains #1 through 4, one patient with PD, and one with dementia with Lewy bodies (DLB) were blotted with anti-αS LB509. Right panel (ARPD case #2) shows specificity. Lower panel, anti-myc pull-down assay (performed as in Fig. 4C) of myc-parkin added to 500 μg of brain homogenates from the same cases as in the upper panel and blotted with LB509. (B) αSp22 is not mono-ubiquitinated. Frontal cortex homogenate (crude), HP2A immunoprecipitate, and recombinant Ub were each blotted with LB509 (left panel) or anti-Ub (right panel). (C) αSp22 isO-glycosylated. HP2A immunoprecipitates (each from 2 mg normal brain homogenate) were incubated with either buffer,N-glycosidase, sialidase A, O-glycosidase, sialidase A + O-glycosidase, or protein phosphatase-1 (31), as indicated, and blotted with syn-1.

αSp22 is an O-linked glycosylated isoform of α-synuclein. To test whether αSp22 was a mono-ubiquitinated form of αS arising as a parkin-independent intermediate in polyubiquitination, we immunoblotted HP2A precipitates of normal human brain, which contained αSp22, with LB509 and anti-Ub (Fig. 5B). αSp22 was detected by LB509, as expected, but not by the Ub antibody, providing no evidence that αSp22 is mono-ubiquitinated. Treatment of the HP2A immunoprecipitates with protein phosphatase-1 did not alter the electrophoretic mobility of αSp22, and N-glycosidase treatment likewise had no effect (Fig. 5C) (31). However, co-incubation with O-glycosidase and sialidase A shifted the 22-kD species to a 16-kD position (31), where it now co-migrated with the unmodified αS monomer, αSp16, from crude brain extracts (Fig. 5C). We conclude from our data obtained by mass spectrometry analysis (21) and these enzymatic digestions that αSp22 is a posttranslationally modified form of human αS containing O-linked sugars.

Discussion. This study has uncovered a direct functional relation in human brain between the two principal gene products associated to date with inherited, monogenic forms of PD, parkin and αS. The data identify a multimeric ubiquitination complex that contains a novel, O-glycosylated αS isoform, αSp22, as an Ub-accepting substrate for parkin. We confirm that neural parkin is an E3 Ub ligase and show that wild-type parkin, but not mutant parkin from four previously genotyped ARPD brains, binds αSp22 and the necessary E2 protein, UbcH7, in vivo and generates high Mr species of αSp22-Ubn conjugates in vitro. Such polyubiquitination is known to serve as a universal signal for degradation by the proteasome (16) (Fig. 6).

Figure 6

Working model for the interaction of parkin with αSp22 in human brain. Step 1, αSp16 undergoes a posttranslational modification by the addition of O-linked sugars to generate αSp22. Step 2a, human parkin recruits UbcH7 at its RING box domain, and thus E1 protein and Ub, and binds αSp22 at its Ubl domain. Step 3, parkin conjugates Ub onto αSp22 to generate αSp22-Ubn. Step 4, αSp22-Ubn undergoes proteasomal degradation and/or sequestration in Lewy bodies (stained with anti-parkin HP2A). Interference with step 2a will lead to accumulation of αSp22n in ARPD (step 2b) and lack of αSp22-Ubn formation.

Our findings also provide a dynamic in vitro assay for the molecular effects of naturally occurring parkin missense mutations. Exogenously expressed myc-parkin, when added to human brain homogenates in vitro, confers not only the assembly of the myc-parkin/UbcH7/αSp22 complex but also E3 Ub ligase activity. Mutant myc-parkin isoforms carrying ARPD missense mutations in the Ubl domain (R42P) or the RING1 domain (T240R) fail to do so, because of an inability to bind either the substrate (αSp22) or UbcH7, respectively (Figs. 4 and 6).

Such specificity of Ub ligases for their substrate(s) and E2 protein(s) is principally determined by at least two domains of the E3 protein (32). Parkin's specificity for its substrate in human brain is demonstrated by the failure of E6-AP, a non-PD related neural E3 ligase, to bind αSp22, despite the shared ability of both ligases to recruit UbcH7. Thus, the Ubl domain appears to be the principal motif that confers selectivity for parkin's target substrate(s) in vivo. Nevertheless, other motifs (for example, within the linker region) may contain alternate substrate recognition sites. It is likely that parkin has additional physiological targets in normal brain besides αSp22 (18), of which some may also contribute to the pathogenesis of PD. Of note, we did not detect co-immunoprecipitation of neural parkin and β-synuclein (33), a close homolog of αS (not shown), or human parkin and αSp16, in our assays. Such remarkable specificity of an E3 ligase in recognizing a modified isoform (but not unmodified or homologous proteins) as substrate has been observed with other E3 ligase/substrate interactions [e.g. (34, 35)]. As concerns the RING box-mediated recruitment of parkin's E2 protein, our data identify UbcH7 as a binding partner for parkin in the αSp22-containing complex (Fig. 2). However, this finding does not preclude the possibility of parkin's interaction with other E2 proteins, an assumption that is strengthened by the localization of a sizeable pool of neural parkin in postsynaptic terminals of adult brain, where no UbcH7 (or αS) can be detected (18).

The loss of normal parkin expression and thus function in ARPD leads to the accumulation of αSp22, as documented by two distinct approaches in all four of the parkin-genotyped ARPD brains analyzed by us (Fig. 5A). In the first and more direct approach, we detected αSp22 only in ARPD brains owing to the relatively low affinity of anti-αS KC7071 for αSp22 (but not for its cognate antigen, αSp16) under nondenaturing conditions and the overall low amount of αSp22 in normal brain (Fig. 5A, upper panels). Second, using myc-parkin proteins added to the homogenates in indirect pull-down assays, we demonstrated an increase of αSp22 in ARPD brains when compared with that pulled down from normal brains. The development of specific antibodies with high affinity for this modified O-glycosylated form will determine whether αSp22 also accumulates in subtle amounts in conventional PD and DLB cases (Fig. 5A, lower panel). In this regard, we have also observed a 22- to 24-kD αS-immunoreactive band in extracts of affinity-isolated Lewy bodies from DLB cortex (18), raising the possibility that both αSp22 as substrate and parkin as its processing enzyme are present in ubiquitinated αS inclusions (Fig. 6).

Several observations are now consistent with the conclusion that functional parkin molecules contribute to the formation of Lewy bodies in human brain: (i) the general absence of Lewy bodies in parkin-deficient ARPD brains (9-12), (ii) the presence of polyubiquitinated αS in Lewy bodies (36), (iii) the presence of parkin protein in Lewy bodies (Fig. 1A) (18), and (iv) parkin's ubiquitin ligase function and the resultant generation of αSp22-Ubn conjugates (see above).

Taken together, our findings in normal and ARPD brains raise the possibility that the inherited and conventional (idiopathic) forms of PD involve etiologically distinct but biochemically related alterations of a shared metabolic pathway. In this hypothetical model, the loss of E3 ligase function (that is, parkin) in ARPD brain leads to the accumulation of nonubiquitinated αSp22, accelerated neuronal loss and a generally younger age of disease onset (2, 6,7). In contrast, in conventional Lewy body-positive PD, wild-type parkin mediates the formation of polyubiquitinated substrates, including αSp22-Ubn. Some of these αSp22-Ubn–conjugates are not efficiently processed by proteasomal degradation, perhaps due to other genetically or environmentally determined causes, and, thus, gradually accumulate (with other proteins) as inclusions in Lewy bodies (37,38). In this regard, it is notable that the only other nuclear gene defect linked to the classical PD phenotype to date occurs in uch-l1 (5).

Such a scenario shows striking parallels to emerging information about the pathogenesis of polyglutamine expansion mutations in the spinocerebellar ataxin-1 gene, sca-1 . Transgenic mice expressing the SCA-1[Q82 ] mutant develop nuclear inclusions of insoluble, polyubiquitinated SCA-1[Q82] in cerebellar Purkinje cells (39). However, the phenotype of these mice is altered when they are crossed with mice lacking the gene for E6-AP Ub ligase (39). Although the degree of SCA-1[Q82]–containing nuclear inclusions is markedly decreased when the cognate E3 ligase (E6-AP) is absent, cerebellar neurons actually show augmented neuronal injury, and the mice undergo earlier neurological impairment (39). This murine model of accelerated neurodegeneration associated with defective ubiquitination and altered proteasomal degradation offers intriguing parallels to the genetic, neuropathological, and biochemical features of idiopathic and inherited PD. Future work will determine whether soluble,O-glycosylated αSp22 mediates specific physiological functions [as found for other O-glycosylated cytoplasmic proteins (40)] and/or has neurotoxic effects in PD. Of equal interest is to examine whether inherited missense mutations in αS (3, 4) as well as the unknown precipitants of idiopathic PD dysregulate the metabolic pathway described here.

  • * These authors contributed equally to this work.

  • To whom correspondence should be addressed. E-mail: schloss{at}cnd.bwh.harvard.edu

  • These authors contributed equally to this work.

REFERENCES AND NOTES

View Abstract

Navigate This Article