Novel Monoclonal Antibodies Demonstrate Biochemical Variation of Brain Parkin with Age*

Autosomal recessive juvenile parkinsonism is a movement disorder associated with the degeneration of dopaminergic neurons in substantia nigra pars compacta. The loss of functional parkin caused by parkin gene mutations is the most common single cause of juvenile parkinsonism. Parkin has been shown to aid in protecting cells from endoplasmic reticulum and oxidative stressors presumably due to ubiquitin ligase activity of parkin that targets proteins for proteasomal degradation. However, studies on parkin have been impeded because of limited reagents specific for this protein. Here we report the generation and characterization of a panel of parkin-specific monoclonal antibodies. Biochemical analyses indicate that parkin is present only in the high salt-extractable fraction of mouse brain, whereas it is present in both the high salt-extractable and RIPA-resistant, SDS-extractable fraction in young human brain. Parkin is present at decreased levels in the high salt-extractable fraction and at increased levels in the SDS-extractable fraction from aged human brain. This shift in the extractability of parkin upon aging is seen in humans but not in mice, demonstrating species-specific differences in the biochemical characteristics of murine versus human parkin. Finally, by using these highly specific anti-parkin monoclonal antibodies, it was not possible to detect parkin in α-synuclein-containing lesions in α-synucleinopathies, thereby challenging prior inferences about the role of parkin in movement disorders other than autosomal recessive juvenile parkinsonism.

Autosomal recessive juvenile parkinsonism (AR-JP) 1 is an especially insidious form of parkinsonism that can strike as early as the 1st decade of life. A major locus for this disease was mapped to chromosome 6q, and the gene was subsequently identified and termed parkin (1). Human parkin is a 465-amino acid protein with a predicted molecular mass of 52 kDa that contains an N-terminal ubiquitin-like domain, a linker region, and a C-terminal TRIAD domain consisting of two RING fingers on either side of an in-between RING (IBR) finger region (2). Deletions and insertions of one or more exons resulting in premature translation termination are some of the most common mutations in parkin, but numerous missense point mutations in parkin have also been shown to be causal of AR-JP (3). Since the identification of parkin, many studies focused on elucidating the function of this protein, and it has been shown it can function as a ubiquitin-protein isopeptide ligase and that its overexpression in vivo enhances the ubiquitinylation of synphilin-1, Pael-R, and CDCRel-1. Immunoprecipitated parkin has been reported to catalyze the ubiquination of these substrates in vitro as well as O-glycosylated ␣-synuclein p22 and cyclin E (4 -8). Furthermore, mutations linked to AR-JP have been reported to reduce the ubiquination of these substrates. Moreover, emerging data suggest that parkin may protect cells from premature death by targeting misfolded or damaged proteins for degradation via the ubiquitin proteasome pathway (4,18,19,38).
Numerous studies (5, 8 -22) have reported that parkin is extractable from cultured cells as well as from human and rodent brain by using low salt buffers with or without the addition of mild detergents. Parkin has also been found in Lewy body-enriched preparations from human brain (20). However, there is a discrepancy among these reports on the number and size of human versus rodent parkin isoforms as well as discrepancies in the subcellular distribution of parkin (12,13,15,21,23). To clarify some of these issues, sensitive and specific antibodies to parkin were generated. A panel of antiparkin monoclonal antibodies (mAb) that recognize different domains of parkin was produced, and their specificity in murine and human brain was demonstrated. Detailed analyses of the biochemical properties of parkin in mouse and human brain tissue by using serial fractionation with buffers of increasing protein extraction strength showed unusual properties of parkin in the human brain. Finally, by using these highly specific parkin mAbs in conjunction with biochemical and immunostaining methods, we were unable to detect parkin in ␣-synuclein containing lesions in patients with ␣-synucleinopa-thies thereby calling into question prior studies about the role of parkin in movement disorders other than AR-JP.

Antibody Generation, Epitope Mapping, and Isotyping
The rabbit polyclonal antibodies CS2132 and AB5112 were purchased from Cell Signaling Technology (Beverly, MA) and Chemicon International, Inc. (Temecula, CA), respectively.
Murine anti-human parkin mAbs were generated by immunization of mice with recombinant human parkin as described (30). Fusion was conducted by using spleen lymphocytes from immunized Balb/c mice and SP2 cells to produce hybridomas. Resulting hybridoma supernatants were screened by enzyme-linked immunosorbent assay (ELISA) using plates coated with parkin. The epitopes of the anti-parkin mAbs were mapped by Western blotting using proteins expressed by Gene-Porter 2 (Gene Therapy Systems, San Diego, CA)-mediated transfection of QBI293 cells with pcDNA3.1-myc/His (Invitrogen) harboring subcloned domains of parkin. To generate these constructs, PCR fragments spanning the coding region for each domain (ubiquitin-like domain, 1-77; linker region, 78 -220; first RING finger, 221-305; IBR region, 306 -398; second RING finger, 399 -465) were digested and ligated into the host vector. The predicted sequence of each construct was confirmed by DNA sequencing. Antibodies were isotyped by antigen-captured ELISA by using goat anti-mouse antibodies to each immunoglobulin subtype (Sigma).

Gel Electrophoresis and Western Blotting
Proteins were resolved on 10 and 15% SDS-polyacrylamide gels for parkin and ␣-synuclein, respectively, and electrophoretically transferred to nitrocellulose membranes (Schleicher & Schuell) as described previously (29,30). Western blotting was conducted by following published protocols by using either goat anti-mouse (Jackson Immuno-Research Laboratories, West Grove, PA) or goat anti-rabbit (Santa Cruz Biotechnology, Santa Cruz, CA) conjugated to horseradish peroxidase as secondary antibodies (29,30). Immuno-positive protein bands were detected with Renaissance Enhanced Luminol Reagents (PerkinElmer Life Sciences) and X-Omat Blue XB-1 film (Eastman Kodak Co.).

Tissue and Sequential Biochemical Fractionation
Parkin-null mice were generated by genomic deletion of exon 2 of parkin, removing most of the coding region for the ubiquitin-like do-main. 2 The parkin-null mice were maintained on a C57Bl6/J ϫ Sv129S hybrid background. Young wild type and parkin null mice used in these studies were 4 -12 weeks old, and aged wild type mice were Ͼ22 months old. Whole mouse brains were dissected for biochemical fractionation studies. The brain tissue samples from human subjects that were used for biochemical analyses are summarized in Table I. Two extraction methods (I and II) were used for these tissues as summarized below.
Method I-Murine or human tissue samples were homogenized in 2 ml/g of high salt (HS) buffer (50 mM Tris, pH 7.5, 2 mM EDTA, 750 mM NaCl) and sedimented at 125,000 ϫ g for 30 min at 4°C. Supernatants from the initial fractionation were saved as the HS fraction, and the pellets were washed and re-extracted sequentially with buffers of increasing protein extraction strengths. For each buffer, two cycles of extraction and washes were conducted using 2 ml/g of HS ϩ 1% Triton X-100 (HST) followed by 2 ml/g of RIPA (150 mM NaCl, 0.1% SDS, 0.5% sodium deoxycholate, 1% Nonidet P-40, 50 mM Tris, pH 8.0). Supernatants were saved as the HST and RIPA fractions, respectively. Floatation and removal of contaminating myelin using HS ϩ 20% sucrose was performed prior to the RIPA extraction. A final extraction using 2 ml/g of 2% SDS in 50 mM Tris, pH 7.6, was conducted on the pellet and sedimented at 125,000 ϫ g for 30 min at 22°C. The supernatant was saved as the SDS fraction.
Method II-Brain tissues from patients with ␣-synucleinopathies and control patients were homogenized in 10 ml/g low salt (LS) buffer (10 mM Tris, pH 7.5, 5 mM EDTA, 1 mM dithiothreitol, and 10% sucrose) and sedimented at 25,000 ϫ g for 30 min at 4°C. Supernatants were saved as the LS fraction, and pellets were washed by re-extraction in LS buffer. Resulting pellets were subjected to two sequential extractions in 10 ml/g Triton X (TX) buffer (LS ϩ 1% Triton X-100 ϩ 0.5 M NaCl) and sedimented at 180,000 ϫ g for 30 min at 4°C. Supernatants from the first of these extractions were saved as the TX fraction. Pellets were then homogenized in 10 ml/g Sarkosyl buffer (LS ϩ 1% N-lauroylsarcosine ϩ 0.5 M NaCl) and incubated at 22°C on a shaker for 1 h prior to sedimentation at 180,000 ϫ g for 30 min at 22°C. Supernatants were saved as the Sarkosyl extraction buffer fraction. Remaining pellets were extracted in 2.5 ml/g SDS buffer (2% SDS, 50 mM Tris, pH 7.6) prior to centrifugation at 25,000 ϫ g for 30 min at 22°C. Supernatants were saved as the SDS fraction.
All extraction buffers contained a mixture of protease inhibitors (1 g/ml each of leupeptin, pepstatin A, N-p-tosyl-L-phenylalanine chloromethyl ketone, N ␣ -p-tosyl-L-lysine chloromethyl ketone, soybean trypsin inhibitor, and 0.5 mM PMSF). For experiments using quantitative amounts of protein, total protein concentration was determined with the BCA protein assay kit (Pierce) using bovine serum albumin as 2 F. A. Perez and R. D. Palmiter, manuscript in preparation.

Immunopurification of Parkin from Human Brain
PRK8-coupled dextran beads were generated using the Carbo-Link column kit (Pierce) following the manufacturer's protocol. Proteins were extracted from human frontal cortex tissue by two extractions with HST, and myelin was floated and removed as described above. The resulting pellet was homogenized in 8 M urea in 50 mM Tris, pH 7.6, with protease inhibitors. Following sedimentation at 125,000 ϫ g, the supernatant was diluted to 2 M urea and incubated by gentle rocking with PRK8-coupled beads for 4 h. The incubation and all subsequent steps were performed at 4°C. The beads were washed 3 times with 2 M urea, 1% Triton X-100 in 50 mM Tris, pH 7.6, and bound parkin was then eluted with Pierce ImmunoElution Buffer by gentle rocking for 1 h. SDS sample buffer was added to an aliquot of the eluant. Samples were not boiled to limit carbamoylation of protein in the presence of urea.

Immunofluorescence and Immunohistochemistry
The harvesting, fixation, and further processing of the tissue specimens used in this study were conducted as described previously (24,25). Briefly, tissue blocks of cingulate cortex from DLB or cerebellum from MSA brains were fixed with 70% ethanol, 150 mM NaCl or neutral buffered formalin and infiltrated with paraffin. The diagnostic assessment of all DLB and MSA cases (both of which are ␣-synucleinopathies characterized by ␣-synuclein inclusions) was performed in accordance with published guidelines (26,37). Whole mouse brains (wild type or parkin null) were fixed with 70% ethanol, 150 mM NaCl and paraffin-embedded.
In order to try to stain ␣-synuclein pathological inclusions in human diseased cases with parkin antibodies, a variety of fixation and retrieval methods were applied. These include the following: 1) formalin-or ethanol-fixed, paraffin-infiltrated tissue followed by antigen retrieval with urea, microwave irradiation, or formic acid; 2) formalin-or ethanol-fixed floating sections; 3) frozen sections post-fixed with ethanol or formalin; and 4) fresh frozen un-fixed sections.
Double labeling indirect immunofluorescence analyses for brain sections were performed as described previously (29). The rabbit anti-␣syn antibody SNL4 (30) and several of the parkin mAbs were used to examine possible co-localization of ␣-synuclein and parkin in ␣-synuclein inclusions by immunofluorescence. For tissue culture experiments, cells were grown on glass coverslips, transfected as described above, and fixed for 10 min with 4% paraformaldehyde in PBS followed by the permeabilization with 0.2% Triton X-100 in PBS for 10 min. Alexa Fluor 488-(green) and Alexa Fluor 594 (red)-conjugated secondary antibodies (Molecular Probes, Eugene, OR) were used to detect immunostaining, and slides were coverslipped with Vectashield 4Ј,6-diamidino-2-phenylindole (DAPI) mounting medium (Vector Laboratories, Burlingame, CA).

Immunostaining of Paraffin-embedded Sections of CHO Cells Stably Expressing Parkin
CHO cells were maintained in ␣-minimal essential medium, 10% fetal bovine serum (Invitrogen), and 100 units/ml penicillin, 100 units/ml streptomycin. Cells were transfected with the plasmid parkin/pcDNA 3.1 expressing full-length un-tagged parkin using Lipo-fectAMINE reagents (Invitrogen) according to the manufacturer's instructions. Cells stably transfected with parkin/pcDNA 3.1 were selected and maintained with 200 g/ml geneticin (Invitrogen).
Following harvesting with trypsin, cells were washed with PBS, fixed with 70% ethanol, 150 mM NaCl, compacted into a pellet by centrifugation at 13,000 ϫ g, and paraffin-embedded. Following sectioning, immunocytochemistry or immunofluorescence staining was performed as described above.

RESULTS
To produce antibodies specific to parkin, a battery of murine mAbs was raised against recombinant human parkin as described under "Experimental Procedures." We identified 60 hybridomas with strong immunoreactivity for the recombinant protein by ELISA. To confirm the specificity of these parkin mAbs, we conducted Western blot analyses using brain extracts from wild type and parkin-null mice as well as extracts from the human brain. Surprisingly, out of our panel of 60 mAbs, only four anti-parkin antibodies (PRK8, PRK28, PRK106, and PRK109) were able to detect protein bands with apparent molecular masses corresponding to that of parkin while at the same time showing no cross-reactivity with proteins in extracts from parkin null mice ( Fig. 1 and Table II). These four parkin mAbs detected ϳ50and ϳ44-kDa bands in mouse brain extracts but not in extracts from parkin-null mice; representative Western blots are shown in Fig. 1B. A similar doublet of ϳ50and ϳ46-kDa immunobands also was detected in human brain extracts with PRK8, PRK28, PRK35, PRK106, PRK109, and a commercially available antibody, CS2132 (Figs. 1B and 2). A very minor cross-reacting band at ϳ142 kDa was detected in both wild type and parkin-null mouse brain extracts by some of the parkin mAbs, but this ϳ142-kDa band is the major species recognized by CS2132 (Fig. 1B). The other commercially purchased anti-parkin antibody AB5112 did not recognize authentic parkin since this antibody detected a number of protein bands in mouse brain extracts that were also present in brain extracts from parkin-null mice (Fig. 1B).
To identify specific domains within parkin that are recognized by our mAbs, we conducted immunoblot analysis using lysates generated from QBI293 cells overexpressing each of the Myc/His-tagged parkin domains (Fig. 1, C and D, and Table II). The majority of the mAbs appeared to bind to the RING2 domain at the C terminus, whereas two of the mAbs detected the ubiquitin-like domain with a single mAb detecting the IBR domain and the region between amino acid residues 76 and 238 of parkin. Despite the generation of mAbs specific for different FIG. 2. Biochemical distribution of parkin in mouse and human brain. Western blot analysis of sequential extractions (method I; see "Experimental Procedures") from wild type (A) and parkin-null mouse brain (B) using the antibody PRK8. Immunoblot analysis with antibody PRK8 of HS-or HST-(C) and RIPA-or SDS (D)-soluble extracts generated by method I from five cortical regions of cases A-1 (upper panels) and A-2 (lower panels). 40 g of total protein were loaded per lane for HS and HST fractions, and 7.5 g of total protein were loaded per lane for RIPA and SDS fractions. The mobility of molecular mass markers (kDa) is shown on the left of each panel. rec, 2 ng of recombinant tagged human parkin. cing, frnt, tmp, occ, and par correspond to cingulate, frontal, temporal, occipital and parietal cortices, respectively. E, immunoprecipitation (IP) followed by Western blotting (WB) analysis to demonstrate that the RIPA-resistant, SDS-soluble immunoband is parkin. RIPA-insoluble human brain lysate was immunoprecipitated with PRK8, and Western blot analysis was performed using the indicated anti-parkin mAbs. One-quarter adjusted volume of the starting material (S) was loaded relative to the amount of immunoprecipitated material (IP) to account for the incomplete pull-down of parkin.  b Ability of antibody to detect an immunoreactive protein band(s) that may correspond to human parkin in the 2% SDS fraction of human brain extract: ϩ, weak; ϩϩ, moderate; ϩϩϩ, strong.
c Ability of antibody to detect a band corresponding to parkin in the high salt fraction of wild-type mouse brain which is absent in the parkin-null mouse brain: Ϫ, unable to detect specific band; ϩ, weak specific band; ϩϩ, moderate specific band; ϩϩϩ, strong specific band.
domains of parkin overexpressed in QBI293 cells, only PRK8, PRK28, PRK106, and PRK109 reacted strongly with fulllength parkin in brain extracts without evidence of other contaminating protein bands. Thus, only mAbs to the RING2 domain of parkin were suitable for direct in vivo studies. Although it is possible that some antibodies to other domains of parkin (e.g. PRK3, PRK35, PRK36, and PRK101) may be specific for human parkin and they may not cross-react with murine parkin (Table II), it was not possible to unequivocally confirm their specificity in extracts from human brains.
To confirm the ability of anti-parkin mAbs to detect endogenous parkin in cell lines, we performed Western blots by using lysates of CHO cells stably transfected with human parkin (CHO-PAR4) as well as untransfected CHO, HeLa, Neuro2A, SH-SY5Y, and HEK-293 cells. Both PRK8 and PRK28 detected an ϳ50-kDa band corresponding to endogenous parkin in CHO, SH-SY5Y, and HEK-293 cells, although the levels of parkin in these cell lines were much lower than in the CHO-PAR4 line (Fig. 1E).
To characterize the biochemical properties of parkin, we performed serial extractions on brain tissue from mouse and human using buffers of increasing protein extraction strengths (Fig. 2). In mouse, parkin was extractable in HS buffer alone as shown with antibody PRK8 (Fig. 2A). No parkin was detected in the HST, RIPA, or SDS fractions, indicating that little or no RIPA-resistant parkin was present in the mouse brain. The ϳ44and 50-kDa protein bands detected with PRK8 in HS fraction were confirmed to correspond to parkin, because these immunobands are not present in extract from null-parkin mice (Fig. 2B). Similar results were obtained with antibodies PRK28 and PRK109 (data not shown). By contrast, the biochemical properties of parkin in aged human brain differed greatly from that of mouse brain. Little or no human parkin was detected in the HS fraction, and the majority of parkin was found in the RIPA-resistant, SDS-extractable fraction (Fig. 2, C and D). The data shown in Fig. 2, C and D, with antibody PRK8 was confirmed with PRK28, PRK35, and PRK109 (data not shown). Indeed, SDS-extractable parkin was persistently present in the gray matter of cingulate, frontal, temporal, occipital, and parietal cortex from six aged human cases examined here (Fig. 2, C and D, and Table III). Furthermore, in addition to 2% SDS, we found that RIPA-resistant parkin in human brain also can be extracted by buffer containing 8 M urea (data not shown). To verify that the immunoreactive species extracted by either 2% SDS or 8 M urea from human brain was indeed parkin, we immunoprecipitated parkin using PRK8-coupled beads after diluting the 8 M urea fraction to 2 M urea. Immunoprecipitated parkin was detected by parkin mAbs that recognize either the N (PRK35) or C termini (PRK8, PRK28, and PRK109) of parkin (Fig. 2E). The ability of this panel of mAbs to detect PRK8immunoprecipitated material demonstrated that parkin was indeed present in a highly insoluble form in human brain. Parkin also was variably detectable in the HS and HST fractions in some of the cases examined here. The presence or absence of HS/HST-extractable parkin varied from case to case, and when present, it was typically most highly extractable from temporal cortex (Table III). This analysis was extended to include the caudate, putamen, globus pallidus, and cerebellum from three of the aged human cases. Although the majority of the parkin was recovered in the RIPA-resistant, SDS-extractable fraction from these regions, some HS/HST-extractable parkin was present in cases that exhibit an HS/HST-extractable cortical pool of parkin (data not shown).
It is possible that the differences in the biochemical properties of murine and human parkin are related to age or different postmortem intervals (PMIs). To address these possibilities, we compared serially extracted parkin from brain tissue of young human (14 -22 years) and young (4 -12 weeks) and aged (Ͼ22 months) mice, as well as in tissue from mice with variable PMI. There was no observable difference in the extractability of parkin extracted from young or old mice or as a function of increasing PMI up to 6 h (data not shown). However, the extractability of parkin in the frontal cortex of the three young human cases differed from parkin in the frontal cortex of the elderly human cases in that significant amounts of parkin were recovered from both the HS-and RIPA-resistant, SDS-extract- SDS ϩϩϩϩ ϩϩϩϩ ϩϩϩϩ ϩϩϩϩ ϩϩϩϩ able fraction of young cases (Fig. 3A). Interestingly, human parkin was not detected in the RIPA fractions. Despite the presence of higher levels of HS-extractable parkin in brain tissue from young human cases, significant amounts of SDSextractable material also were present in these cases (Fig. 3, B and C). These results were confirmed with two other parkin antibodies, PRK28 and PRK109 (data not shown). Thus, there is an effect of aging on parkin in human but not in mouse brains.
Because several reports have suggested that parkin and ␣-synuclein are linked functionally or pathologically (5,17,20), we examined the ability of our anti-parkin mAbs to detect a biochemical association between these two proteins as well as evidence of parkin in ␣-synuclein inclusions of patients with DLB and MSA. First, we showed that LS-extractable ␣-synuclein, as detected by LB509 (33), is present in similar amounts from cingulate cortical gray matter of DLB and control cases and from cerebellar white matter from MSA and control cases (Fig. 3D). However, parkin levels were consistently low in the LS fraction of both disease and control cases except in one case (A-10) wherein high levels of LS-extractable parkin were most likely due to the relatively young age (43 years) of this individual. As shown previously (24, 33-36), pathological ␣-synuclein was present in the RIPA-resistant, SDS-extractable fraction from only diseased cases (Fig. 3E), and as expected, parkin was present in the SDS-extractable fraction from both disease and normal cases. The variation in parkin levels in the SDS-extractable fraction could be due to variable cell loss in disease cases. Thus, RIPA-insoluble parkin levels do not strongly correlate with ␣-synuclein levels in pathological brain tissue.
To explore further the association of parkin with ␣-synuclein, we asked whether or not parkin could be detected in Lewy bodies and Lewy neurites from patients with DLB or PD, as well as glial cytoplasmic inclusions (GCIs) from patients with MSA. Immunohistochemical studies using parkin mAbs PRK8, PRK28, or PRK109 and anti-␣-synuclein antibodies on adjacent brain sections from DLB and MSA cases failed to detect parkin in any ␣-synuclein inclusions (Fig. 4, A-D; data not shown). Double labeling indirect immunofluorescence studies also revealed no detectable parkin staining in Lewy bodies, Lewy neurites, or GCIs (Fig. 4, E-H). In addition to the methods shown here, a variety of antigen preservation or retrieval methods (see "Experimental Procedures") was used to try to detect parkin in ␣-synuclein pathological inclusions. To confirm that our parkin antibodies can detect parkin in fixed cells, immunofluorescence analysis of transfected cells overexpressing parkin was performed. Several parkin mAbs, including FIG. 3. Biochemical distribution of brain parkin in young and aged human as well as in patients with DLB and MSA. All the fractions generated by method I from case Y-1 (A) and the specific fractions HS (B) and SDS (C) from frontal cortex of three young (Y-1, Y-2, and Y-3) and three aged (A-7, A-8, and A-9) cases were loaded on SDS-polyacrylamide gels, and following electrophoretic transfer onto membranes the blots were probed with PRK8. LS (D) and SDS (E) fractions generated by sequential extraction method II from gray matter of cingulate cortex from DLB cases and cerebellar white matter from MSA cases, and from control cases were analyzed by Western blot analysis and probed with PRK8 (upper panels) and the anti-␣-synuclein antibody LB509 (lower panels). The mobility of molecular mass markers (kDa) is shown at the left of each panel. rec, 2 ng of tagged recombinant human parkin.
Moreover, we also wanted to demonstrate that our novel anti-parkin antibodies could specifically detect parkin in fixed, paraffin-embedded transfected cells and in mouse brain. CHO cells, transfected to stably express human parkin (see "Experimental Procedures"), were generated, and these cells were processed and paraffin-embedded using the standard method used for human tissue. Following sectioning and incubation with several anti-parkin antibodies (i.e. PRK8 and PRK28 (Fig.  5) or PRK109 (data not shown)), cells expressing parkin could readily be detected (Fig. 5). Furthermore, PRK28 was able to detect endogenous parkin in mouse brain by immunohistochemistry. In mouse cerebellum, PRK28 detected parkin in white matter, the molecular layer, and Purkinje cells (Fig. 5M). This staining was not seen in parkin null mice (Fig. 5N), indicating the specificity of this antibody for mouse parkin. PRK28 revealed a similar diffuse staining pattern throughout the brains of wild type mice, whereas parkin null mice showed no staining (data not shown). DISCUSSION We have generated a library of murine anti-parkin mAbs that detect recombinant parkin by ELISA and immunoblotting. Initial analyses using extracts from human and mouse tissues, however, were variable and difficult to interpret. Attempts at detecting parkin by using published protocols and our mAbs as well as commercially available and published parkin antibodies from other groups (data not shown) resulted in the detection of protein bands that, although close to the predicted size of 52 kDa for parkin, differed subtly in size depending on the antibody used. The difficulty in identifying antibodies specific to parkin is likely borne out by recent analysis of human genomic data, which identified RING fingers as the fourth most abundant protein domain in the human genome, as it is present in 379 predicted human proteins (31). Moreover, the most recent analysis of the human proteome performed by the EMBL-EBI (www.ebi.ac.uk/index.html) indicates that the average protein length in the human proteome is 459 amino acid residues, very close to that of parkin. It is therefore likely that there are many RING finger proteins with a molecular mass close to that of parkin detectable as cross-reacting species by many anti-parkin antibodies generated here as well as by commercial sources and other investigators. Nevertheless, four of our anti-parkin mAbs detected parkin specifically because these mAbs recognized protein bands from human and mouse brains but not in  the brains of parkin-null mice. Five additional mAbs appeared to detect human, but not mouse, parkin specifically, although it was not possible to unequivocally confirm their specificity in human brain extracts (see Table II).
There is great disagreement in the literature regarding the number and relative electrophoretic mobility of parkin isoforms in rodent and human, which most likely reflects crossreactivity of parkin antibodies with other proteins. Although most studies report a band of ϳ52 kDa for parkin, several studies report the detection of additional putative isoforms from rodent tissues ranging from ϳ22 to ϳ65 kDa (8 -16) and from human tissue ranging from ϳ22 to ϳ65 kDa (5,12,13,(17)(18)(19)(20)(21)(22). The specific antibodies described here detect only doublets of ϳ50 and ϳ44 kDa in mouse and ϳ50 and ϳ46 kDa in human brains. It is possible that the ϳ50-kDa band is fulllength parkin, whereas the ϳ44/46-kDa species represents protein from an alternatively spliced mRNA or post-translational proteolytic cleavage of full-length parkin. Alternatively, the 50-kDa species could represent post-translational modification of the ϳ44/46-kDa form, although this is less likely because recombinant full-length parkin has an apparent molecular mass of ϳ50 kDa. Because only the parkin mAbs that recognized the C terminus of rodent parkin effectively detected parkin in rodent brain lysates, it is possible that they do not detect shorter isoforms of parkin lacking the C terminus in mice. However, we have mAbs that recognize both N-and C-terminal regions of human parkin, and they do not detect any additional human isoforms of parkin.
Interestingly, although mouse parkin was easily extracted from brain by HS buffer, most human parkin was only extracted from brain with harsher buffers, such as 2% SDS, especially in elderly humans. One explanation for this phenomenon is that human parkin becomes modified with age or interacts with other molecules that become modified with age, and these alterations influence extractability. These modifications may not occur in mice either because of differences in mouse proteins or because of the shorter life span of mice. It will be important to determine whether the biological functions of parkin are affected by changes in extractability. If the less readily extracted forms of parkin are compromised in their intrinsic activity or availability to substrates, that would support the observation that haplo-insufficiency might contribute to PD (32). Alternatively, parkin may be part of different protein complexes that have different extraction properties, and the relative abundance of these complexes may change with age. The recent finding (6) that parkin is present in Skp1, cullin, and F-box protein-like complexes supports such a possibility.
Whereas several reports (17,20) have detected parkin in ␣-synuclein pathologies with some anti-parkin antibodies, we were unable to detect parkin in Lewy bodies, Lewy neurites, or GCIs by immunohistochemical or immunofluorescence techniques. We were, however, able to detect diffuse cytoplasmic staining of parkin in cells overexpressing parkin. We were also able to detect parkin in ethanol-fixed, paraffin-embedded cells stably expressing parkin as well as at endogenous expression levels in mouse brain, demonstrating that our mAbs are capable of recognizing parkin by using immunohistochemical methods. Because disease-related proteins tend to be quite concentrated in characteristic aggregates and we failed to detect parkin in these ␣-synuclein containing inclusions, our data argue against an interaction between parkin and ␣-synuclein in ␣-synucleinopathies.
The generation of sensitive anti-parkin antibodies, with their specificity confirmed by testing on parkin-null mouse tissues, has provided us with powerful tools for examining the properties of parkin in mouse and human brains. Two isoforms of parkin have been identified in both mouse and human brains, although the precise nature of these two isoforms remains unclear. We have determined that the biochemical properties of parkin are inherently different between mouse and human brains and that there is a shift in parkin from the HS/HST-extractable to RIPA-resistant fractions with age in the human brain. It is possible that alterations in the biochemical properties of parkin with age are a contributing factor to the development of PD, but further work is necessary to examine the nature and roles of the multiple parkin isoforms and to determine the mechanism(s) whereby the extractability of parkin changes with age.