Caspase-mediated parkin cleavage in apoptotic cell death.

The parkin protein is important for the survival of the neurons that degenerate in Parkinson's disease as demonstrated by disease-causing lesions in the parkin gene. The Chinese hamster ovary and the SH-SY5Y cell line stably expressing recombinant human parkin combined with epitope-specific parkin antibodies were used to investigate the proteolytic processing of human parkin during apoptosis by immunoblotting. Parkin is cleaved during apoptosis induced by okadaic acid, staurosporine, and camptothecin, thereby generating a 38-kDa C-terminal fragment and a 12-kDa N-terminal fragment. The cleavage was not significantly affected by the disease-causing mutations K161N, G328E, T415N, and G430D and the polymorphism R366W. Parkin and its 38-kDa proteolytic fragment is preferentially associated with vesicles, thereby indicating that cleavage is a membrane-associated event. The proteolysis is sensitive to inhibitors of caspases. The cleavage site was mapped by site-directed mutagenesis of potential aspartic residues and revealed that mutation of Asp-126 alone abrogated the parkin cleavage. The tetrapeptide aldehyde LHTD-CHO, representing the amino acid sequence N-terminal to the putative cleavage site was an efficient inhibitor of parkin cleavage. This suggests that parkin function is compromised in neuropathological states associated with an increased caspase activation, thereby further adding to the cellular stress.

Parkinson's disease is the second most common neurodegenerative disorder, and its symptoms arise primarily from a rather selective loss of dopaminergic neurons in the substantia nigra of the brain stem (1). Lesions in the parkin gene on chromosome 6 are responsible for a large number of patients affected by early onset Parkinson's disease (2,3). The parkin gene encodes the intracellular parkin protein that consists of 465 amino acids with an N-terminal ubiquitin-like domain and a C-terminal domain with two Ring finger motifs (2). Ring finger domains are present in one class of E3 1 ubiquitin ligases (4), and parkin exhibits ubiquitin ligase activity although it is unclear whether it is a bona fide E3 ubiquitin ligase or exists as part of a multisubunit ubiquitin ligase complex (5)(6)(7). The C-terminal Ring finger domain mediates the binding to E2 ubiquitin conjugases (2,7).
The ubiquitin-proteasome system is responsible for the majority of intracellular protein catabolism and relies on the concerted action of two ATP-dependent enzyme systems, the ubiquitin system and the proteasome (8). Ubiquitination of substrate proteins is carried out by sequential reactions catalyzed by the ubiquitin-activating enzyme (E1), ubiquitin conjugating enzymes (E2s), and ubiquitin ligases (E3s). The E3 ubiquitin ligases function either as templates that bring the ubiquitin-bearing E2 molecules in the proximity of the substrate proteins or by directly transferring the ubiquitin moiety to the substrate protein. The target proteins, ranging from misfolded proteins to highly controlled cellular regulators of e.g. the cell cycle, become polyubiquitinated, thereby making them recognizable by the 26 S proteasomal multicatalytic protease complex. The expression of the parkin gene is enhanced during unfolded protein stress as part of the unfolded protein response, and parkin confers cytoprotection toward unfolded protein stress (6,9). Several disease-causing lesions in the parkin gene result in a loss of ubiquitin ligase function (5,7,9). The loss of function elicits a preferential degeneration of the dopaminergic neurons in the substantia nigra despite parkin being widely expressed in the central nervous system (10 -13). This indicates that parkin performs a vital function for the survival of this particular population of nerve cells. Accordingly, pathological signaling pathways that impinge on parkin function may represent potential contributors to the neurodegeneration in sporadic Parkinson's disease.
Caspases comprise a family of cysteine-proteases that regulate the process of apoptosis at several levels and exhibit an ultimate requirement for aspartic residues in the P 1 position of their substrates (14 -16). Caspase-mediated cell death is generally considered a swift process, where caspase activation is rapidly followed by cellular fragmentation and phagocytic removal (17). However, transient caspase activation is compatible with cellular survival (18,19), and nerve cells display a relative resistance toward caspase activity that may allow lowlevel caspase activation to play a role in long term neurodegenerative processes (20). This is corroborated by the demonstration of activated caspases or their specific proteolytic products in degenerating nerve cells of brain tissue affected by Parkinson's disease (21) and Alzheimer's disease (22,23). Proteolytic cascades catalyzed by caspases may thus represent signaling pathways involved in the progression of common neurodegenerative diseases.
We investigated the presence of apoptosis-associated parkin modifications and demonstrate that parkin is cleaved during apoptosis by caspase-mediated proteolysis. Asp-126 represents the major cellular parkin cleavage site in parkin-expressing Chinese hamster ovary (CHO) and SH-SY5Y dopaminergic neuroblastoma cell lines as determined by site-directed mu-tagenesis. Proteolysis of parkin after Asp-126 will mimic the disease-causing in-frame truncation of exons 2-3 comprising amino acid residues 3-137 (3) and is thus incompatible with a functional parkin molecule.
Cell Lines-CHO cells were maintained in UltraCHO medium (Bio-Whittaker) supplemented with 1% fetal calf serum (BioWhittaker). Human dopaminergic neuroblastoma SH-SY5Y cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells were transfected with 1 g of DNA using FuGENE 6 transfecting reagent (Roche). Transfected cell lines were selected in 300 g/ml Zeocin (CHO cells) or 25 g/ml Zeocin (SH-SY5Y cells) and tested for the expression of parkin by immunoblotting. The P8-CHO and CP3-SH-SY5Y cell lines stably expressing human recombinant parkin were used in the present study. Apoptosis was induced by incubating the cells with 300 nM okadaic acid (OA), 10 M staurosporine, or 14 M camptothecin. Control cells were incubated with the solvent of the toxins, dimethyl sulfoxide. The peptide aldehyde caspase inhibitors DEVD-CHO and YVAD-CHO and the custom-synthesized LHTD-CHO (Bachem) were also solubilized in dimethyl sulfoxide.
Cellular Extraction and Subcellular Fractionation-For the direct analysis of parkin cleavage, cells were scraped of the culture surface, rinsed in phosphate-buffered saline, pelleted, and dissolved in 8 M urea and 2% sodium dodecyl sulfate. Insoluble proteins were pelleted prior to determination of the protein concentration by the bicinchoninic method (Sigma).
Parkins association to vesicles were assessed by the fractionation of a postnuclear supernatant into a cytosol and vesicle fraction by centrifugation at 350,000 ϫ g for 2 h at 4°C. The postnuclear supernatant was prepared by Dounce homogenization (10 strokes at 800 rpm) in 9% sucrose, 0.1 mM EGTA, 1 mM MgCl 2 , 10 mM Hepes, pH 7.4, on ice followed by pelleting of nuclei and debris by centrifugation at 625 ϫ g for 5 min. Samples from the postnuclear supernatant, the cytosol, and the vesicle fractions were sonicated in 2% sodium dodecyl sulfate, 20 mM dithioerythritol, 20% glycerol, 25 mM Tris, pH 6.4, and heated to 95°C for 3 min prior to SDS-PAGE.
Antibodies and Immunoblotting-The PAR-N1 antibody was produced by immunizing rabbits with the synthetic peptide, EVDS-DTSIFQLKEVVAKC, corresponding to amino acid residues 16 -32 of human parkin with a cysteine residue in the C-terminal position. PAR-C1 was produced by immunization with CEWNRVSMGDHWFDV corresponding to parkin amino acid residues 451-465 with cysteine 457 changed to a serine. Both peptides were coupled to keyhole limpet hemocyanin prior to immunization. The PAR-N1 serum was affinitypurified on a column with the synthetic peptide coupled via its Cterminal cysteine to thiopropyl-Sepharose (Amersham Biosciences). The T160 antibody was produced by immunizing rabbits with partially purified recombinant hexahistidine-tagged human parkin. The IgG from the PAR-C1 and T160 sera was isolated by protein-A-Sepharose chromatography (Amersham Biosciences). The mouse monoclonal antihuman poly(A)DP-ribose polymerase antibody 66401A was from PharMingen. The mouse monoclonal anti-human Bip/GRP78 antibody was from BD Transduction Laboratories. The immunoblotting was performed essentially as described (24) using the antibodies PAR-N1, PAR-C1, T160, 66401A, and Bip/GRP78 at concentrations of 25, 10, 6, 1, and 0.25 g/ml, respectively. Quantification of immunoblots was performed by scanning of the bands on a GS 300 transmittance scanning densitometer (Hoefer Scientific Instruments) and analyzing the data using the GS 365W electrophoresis data reduction system (Hoefer Scientific Instruments).
Chromatin Staining-Cells were grown on glass slides, washed in phosphate-buffered saline, and fixed in 4% formaldehyde at room temperature for 20 min. The chromatin was stained with bis-benzimide (Roche Molecular Biochemicals) at room temperature for 20 min, after which the cells were visualized by immunofluorescence microscopy at 200ϫ magnification. Four random images from each culture were captured, and the nuclear morphology was examined for the appearance of intensely fluorescent condensed chromatin by counting 300 consecutive nuclei.

RESULTS
Antibodies and Cell Lines-The proteolytic processing of parkin is studied in the CHO and the human dopaminergic neuroblastoma SH-SY5Y cell line stably transfected with a human parkin expression vector. All investigations are performed on both cell lines, and similar results were observed in all cases. Representative data from one of the cell lines are presented.
All the antibodies employed bind to an ϳ52-kDa band in extracts of parkin expressing P8-CHO cells (Fig. 1, lanes 1, 3,5) in agreement with the calculated 51.6-kDa molecular mass of human parkin. The specificity of the antibody binding to the 52-kDa band is demonstrated by its inhibition by prior absorption of the antibody with its antigenic peptide ( Fig. 1, lanes 2 and 4) and partially purified glutathione-S-transferase-parkin fusion protein (Fig. 1, lane 6). Moreover, none of the antibodies bind to a 52-kDa band in the non-transfected parental cell lines as demonstrated for the T160 binding to the CHO cell extract (Fig. 1, lane 7). The ϳ100-kDa T160-immunoreactive band in the transfected CHO cells (Fig. 1, lane 5) and SH-SY5Y cells (not shown) represents a cross-reacting band as demonstrated by its absence in the preabsorption control (Fig. 1, lane 6). All antibodies bind to recombinant human glutathione-S-transferase-parkin and hexahistidine-parkin fusion proteins (data not shown).
Apoptosis-associated Parkin Cleavage-Apoptosis is induced in the parkin-expressing cell lines by treatment with the established inducers of apoptosis OA (a protein phosphatase 2Ainhibitor), staurosporine (a broad-spectrum kinase inhibitor), and camptothecin (a topoisomerase inhibitor). No significant difference in either the morphology or the kinetics of cell death was observed in the cells expressing and not expressing parkin during the process of apoptosis induced by the above mentioned inducers (data not shown).
Treating the P8-CHO cell line with 300 nM okadaic acid causes a time-dependent accumulation of a 38-kDa T160-immunoreactive band that becomes detectable after 6 h and reaches a maximal plateau after ϳ12 h, where it remains stable for more than 24 h ( Fig. 2A). The T160 binding to the 38-kDa band was specific as it was inhibited by preabsorption with partially purified recombinant glutathione-S-transferaseparkin (data not shown). The formation of the 38-kDa band is not a late apoptotic phenomena as it precedes apoptosis-associated nuclear condensation and fragmentation that comprise 2, 10, and 82% of the nuclei after 6, 12, and 24 h, respectively, as measured by bis-benzimide immunofluorescence analysis (Fig. 2C).
OA, staurosporine, and camptothecin generate a similar 38-kDa T160-immunoreactive band, whereas their kinetics is distinct under each apoptosis-stimulating condition. Staurosporine and OA result in a fast cleavage comparable with the more protracted proteolysis induced by camptothecin as exemplified by the 24-and 48-h time points in Fig. 2D, lanes 2-4. A distinct ϳ12-kDa T160-immunoreactive band is observed in both cell lines concomitant with the appearance of the 38-kDa band but only in cultures with a high degree of parkin cleavage and/or when loading larger samples (Fig. 2D, lanes 2 and 3). The nature of the apoptosis-associated T160-immunoreactive bands is investigated with the epitope-specific PAR-N1 and PAR-C1 antibodies. PAR-N1 binds to the 52-kDa band only (Fig. 2D, lane 5), whereas PAR-C1 binds to both the 52-kDa and the 38-kDa bands (Fig. 2D, lane 7). The PAR-C1 binding to the 38-kDa band is inhibited by preabsorption of the antibody with its antigen prior to immunoblotting (data not shown). Accordingly, the 38-kDa band represents a C-terminal fragment of parkin based on immunoreactive criteria. The 12-kDa band likely represents the N-terminal fragment as demonstrated below.
The apoptotic nature of the cell death is, in addition to bis-benzimide staining (Fig. 2C), verified morphologically by phase-contrast microscopy showing that the cells become round and lose their attachment to the substratum without lysis (data not shown). A biochemical signature of apoptosis is demonstrated by the caspase-mediated degradation of the 116-kDa poly(A)DP-ribose polymerase to the cleaved 86-kDa fragment (Fig. 2B, lower part). The extent of poly(A)DP-ribose polymerase cleavage does not mirror the parkin cleavage exactly indicating that they might be caused by different proteolytic activities (Fig. 2B, lower versus top part).
Some disease-causing missense mutations in the parkin gene exist along with polymorphisms at single amino acid positions. We investigated whether a range of these may make parkin a better substrate for apoptosis-associated proteolysis in CHO cell lines stably expressing the mutant or polymorphic parkin proteins (Fig. 2E). No significant changes in the cleavage pattern were observed for the mutations K161N, G328E, T415N, and G430D (3,25) and the polymorphism R366W (26) after induction of apoptosis by 300 nM OA for 9 h (data not shown) and 24 h (Fig. 2E).
Caspase-mediated Parkin Cleavage-Apoptosis is the result of imbalances between pro-and antiapoptotic cellular factors and is orchestrated by proteolysis performed by the caspase class of cysteine proteases. Analyses using the caspase 1 and 3 inhibitors YVAD-aldehyde and DEVD-aldehyde demonstrate that both inhibitors abrogate the parkin cleavage when used in extracellular concentrations of 100 M (Fig. 3A, lanes 3 and 9  versus 2 and 8). This demonstrates that caspases directly or via activation of down-stream proteinases are responsible for the parkin cleavage as the solvent for the inhibitors dimethyl sulfoxide neither inhibit nor induce parkin cleavage (data not shown).
The cleavage into a C-terminal 38-kDa fragment and a putative 12-kDa N-terminal fragment indicates a cleavage site close to amino acid residue 100. Caspases cleave their substrates after an aspartic residue, and parkin contains several aspartic residues around residue 100 (Fig. 3B, upper part). We chose initially to mutagenize the six aspartic residues shown in Fig. 3B, upper part to glutamic acid residues. Glutamic acid was preferred to alanine because it results in a more conserved change taking into account that parkin is very sensitive to disease-causing mutations in its peptide back bone. Moreover, caspases display a high degree of selectivity for aspartic acid as compared with glutamic acid in the P 1 position as shown by aspartic/glutamic selectivity ratios ranging from 400 to 20,000 for different caspases (16).
Accordingly, we generated the following stably transfected CHO and SH-SY5Y cell lines: the double mutant parkin (D86E/ D87E) and the point mutations parkin ((D106E), (D115E), (D126E), and (D130E)). The cell lines were subsequently challenged with the previously mentioned apoptogenic factors as demonstrated for OA in Fig. 3B, lower part. Only the D126E mutation abrogates the generation of the 38-kDa parkin fragment thus indicating that the Asp-126/Ser-127 peptide bond represents the cleavage site (Fig. 3B, lower part). Subsequently, a CHO cell line stably expressing a D126A parkin mutant was generated to verify that the inhibition of cleavage not was an artifact of the Asp to Glu mutation. This cell line displayed the same inhibitory effect on the apoptosis-associated cleavage as the D126E mutation (data not shown). The absent parkin cleavage in the D126E cells is not due to inhibition of caspase activity as demonstrated by OA-induced poly-(A)DP-ribose polymerase cleavage in the cells (data not shown). No effect is observed for the D86E/D87E double mutation despite that this site has been reported to represent a caspase cleavage site (40) (Fig. 3B). The D126E mutagenized, parkinexpressing CHO and SHSY5Y cell lines did not exhibit more resistance to the toxic effects of OA, staurosporine, and camptothecin than did the non-transfected parental cells or those expressing wild type parkin as determined by their morphology and bis-benzimide immunofluorescence staining (data not shown).
Besides the Asp-126 cleavage, some quantitatively minor but significant degradation does occur in apoptotic parkin-expressing cells. This is demonstrated in highly exposed immunoblots as a smear between the 52-kDa parkin band and the 38-kDa cleavage product in cells expressing parkin and as a smear between parkin and the 38-kDa marker in the D126E cells (Fig.  3D). The non-Asp-126 proteolysis is also mediated by caspases as verified by its sensitivity to the DEVD caspase inhibitors (Fig. 3D). This demonstrates the presence of additional caspase cleavage sites in parkin.
A peptide corresponding to the amino acid sequence N-terminal to a proteolytic cleavage site is expected to represent a pseudosubstrate for the responsible protease and thus act as an inhibitor of the reaction. The tetrapeptide aldehyde LHTD-CHO, corresponding to the parkin residues 123-126, is as an efficient inhibitor of parkin cleavage in our cellular assay as is the established caspase inhibitor DEVD (Fig. 3C). Accordingly, the Asp-126/Ser-127 peptide bond represents the major apoptosis-associated caspase cleavage site in SH-SY5Y and CHO cells.
Parkin and Its 38-kDa C-terminal Cleavage Fragment Is Associated to Cellular Vesicles-The subcellular distribution of parkin was determined by fractionating the P8-CHO postnuclear supernatant into the cytosolic and vesicular fractions. The majority of parkin resides in the vesicular fraction (Fig.  4A, lower panel, lanes 2 versus 3) as demonstrated by the copelleting with the vesicular marker BIP (Fig. 4A, top panel). The 38-kDa parkin fragment cofractionates with the noncleaved parkin after induction of apoptosis (Fig. 4A, lower  panel, lanes 5 versus 6). Quantification of the bands using laser scanning densitometry showed that the vesicle-associated fraction accounted for 70% of the 52-kDa parkin in the control and OA-treated cells and ϳ85% of the 38-kDa band in the OAtreated cells. The localization of the 12-kDa N-terminal cleavage fragment was difficult to ascertain due to the low and variable concentration (Fig. 2D, lanes 2 and 3), but our data indicated a more cytosolic localization (data not shown). We made the P1-126-CHO cell line, stably expressing the N-terminal 126-amino acid residues of parkin (Fig. 2D, lane 8) to investigate the localization of this fragment of parkin. Fig. 4B, lanes 3 versus 4 demonstrate that the recombinant 12-kDa parkin-(1-126) peptide is almost exclusively cytosolic. This indicates that the vesicle-associated parkin protein is cleaved in this compartment, after which the C-terminal fragments remain associated with the vesicles, whereas the N-terminal fragment is free to dissociate into the cytosol. DISCUSSION Apoptosis-associated proteolysis cleaves parkin and generates a 38-kDa C-terminal fragment as demonstrated by (i) its specific binding of polyclonal antibodies raised against recombinant human parkin and a synthetic peptide corresponding to the 15 C-terminal amino acid residues of human parkin and (ii)  1 and 7) and presence of 300 nM OA for 24 h (lanes 2-6 and 8 -12) and in the absence (lanes 2 and 8) and presence of 100, 10, 1, or 0.1 M of the tetrapeptide aldehyde inhibitors YVAD-CHO (lanes 3-6) and DEVD-CHO (lanes 9 -12). Both inhibitors caused a dose-dependent decrease in the formation of the 38-kDa parkin fragment. B, site-directed mutagenesis of Asp-126 inhibits apoptosis-associated parkin cleavage. Upper part, the amino acid sequence of human parkin from residue 85 to 131. The aspartate residues are shown in bold with their positions marked above and the sequence underlined LHTD used for synthesizing an inhibitor of parkin cleavage. Lower part, the effect of site-directed mutagenesis of the above marked aspartate residues to glutamate residues on the generation of the apoptosis-associated 38-kDa parkin fragment in SH-SY5Y cells stably expressing the mutagenized parkin proteins. Apoptosis was induced by incubation with 300 nM OA for 24 h. Neither the double mutant D86E/D87E nor the single mutants D106E, D115E, or D130E inhibited the parkin cleavage. This was only accomplished by the D126E mutation. C, a tetrapeptide aldehyde corresponding to parkin amino acid sequence 123-126 inhibits apoptosis-associated parkin cleavage. The LHTD-CHO peptide aldehyde was compared with DEVD-CHO (both 100 M) with respect to their effect on parkin cleavage in CP3-SH-SY5Y cells induced by 300 nM OA for 24 h. Note that they both efficiently inhibit the formation of the 38-kDa fragment. D, cleavage after Asp-126 represents the major cellular caspase cleavage site in parkin. CP3-SH-SY5Y cells expressing wild type parkin (WT) and D126E-SH-SY5Y cells (D126E) were treated with 300 nM OA for 24 h in the absence and presence of 100 M DEVD-CHO (DEVD) and analyzed for parkin proteolysis by highly exposed T160 immunoblots. The D126E point mutation inhibits most of the apoptosis-induced cleavage but some caspase inhibitor-sensitive degradation does occur in the parkin- the abrogated formation of the 38-kDa fragment in the presence of peptide inhibitors of caspases. Parkin cleavage is induced by as different inducers of apoptosis as the protein phosphatase 2A inhibitor OA, the broad-spectrum kinase inhibitor staurosporine, and the topoisomerase inhibitor camptothecin. This indicates that parkin cleavage is associated to apoptosis in general and not merely reflects the activation of a specific cellular signal transduction pathway. A direct involvement of caspase activity in the cleavage reaction is supported by several lines of evidence. Firstly, two peptide-based caspase inhibitors inhibit the cleavage, thereby demonstrating a caspase as being upstream of or as responsible for the cleavage. Secondly, the inhibition of cleavage by site-directed mutagenesis of Asp-126 to Glu and Ala indicates that the protease requires an aspartate in the P 1 position. Thirdly, the requirement for an aspartate in the P 1 position is corroborated by the efficient inhibition of the cleavage by the tetrapeptide aldehyde LHTD-CHO based on the amino acid sequence N-terminal to the putative scissile peptide bond Asp-126 -Ser-127. Caspases may cleave other sites in parkin since some caspase inhibitor-sensitive proteolysis of parkin-(D126E) was observed as a smear below the uncleaved 52-kDa parkin. Asp-86 -Asp-87 has been reported as a cleavage site in parkin (40), but this site has no quantitative significance in our cellular models based on sitedirected mutagenesis. The identity of the parkin cleaving caspase cannot be inferred from the potency of the inhibitors employed as they have a low cell permeability. The LHTD sequence N-terminal to the cleavage site does not conform to any of the optimal sequences for caspase cleavage sites, but may be cleaved with some efficiency by caspases 1, 2, and 9 as judged on the combinatorial approach employed by Thornberry et al. (27). The identification of the responsible caspase will have to await the development of a cell-free assay, specific cell permeable caspase inhibitors, or the use of cell lines established from caspase gene knock-out mice.
Subcellular fractionation of parkin-expressing cells showed that parkin predominantly is a vesicle-associated protein as previously demonstrated (28). This localization is maintained in the apoptotic cells where the 38-kDa fragment remains vesicle-associated. This indicates that the caspase-mediated parkin cleavage occurs at or near the membranes where caspases previously have been localized (29,30).
Proteolysis of parkin after Asp-126 will liberate the ubiquitin-like domain from the remaining large polypeptide containing the Ring box domain (Fig. 5). This is incompatible with a functional parkin enzyme based on several data. Clinical data demonstrate that a single missense mutation in the ubiquitinlike domain R42P or in-frame deletion of exons 2 and 3, resulting in the loss of amino acid residues 3-137, causes early onset autosomal-recessive juvenile parkinsonism based on parkin dysfunction (3,25). Moreover, biochemical and cellular data demonstrate that deletion of the ubiquitin-like domain abrogates in vivo and in vitro ubiquitin ligase activity and binding of some parkin substrates (5-7, 9, 31). Subcellular fractionation of the P1-126-CHO1 cells expressing the N-terminal 126-amino acid parkin peptide demonstrates its localization in the cytosol. This suggests that it may function independently in this environment e.g. as suggested by inhibiting the proteasome (40).
Caspase activation has for several years been associated to the swift apoptotic process observed in cell culture models and certain animal models. However, increasing evidence demonstrate that activated caspases and their specific proteolytic products are present in viable neurons in brain tissue from normal-aged control brains (32,33) and brains affected by Parkinson's disease (17,34) and other neurodegenerative diseases (18,19,31,32,35,36) where they are likely to persist for months. The caspase-positive neurons display degenerative characteristics but are viable, thus demonstrating that lowlevel proapoptotic caspase activity is compatible with continued survival of functionally compromised nerve cells (17)(18)(19)36) probably due to antiapoptotic mechanisms (for a review, see Ref. 37). This idea is corroborated by cell culture experiments where cultured neurons display strong resistance toward microinjection of active caspases as compared with cell lines that die rapidly (16). Shimura et al. (10) have previously reported the existence of an anti-parkin-immunoreactive band of ϳ41 kDa in brain extracts from patients with Parkinson's disease, which is absent in control substantia nigra. However, with our current antibodies we are unable to unambiguously demonstrate a parkin cleavage product compatible with a Asp-126 -  1-3) and presence of 300 nM OA (lanes 4 -6) for 18 h were homogenized prior to isolation of a postnuclear supernatant (lanes 1 and 4) that were further fractionated into a cytosolic (lanes 2 and 5) and vesicular fraction (lanes 3 and 6). Comparable volumes of the three fractions were analyzed as described in Fig. 1  Ser-127 proteolysis in extracts of substantia nigra (data not shown). The demonstration of such putative in vivo cleavage of parkin must await the generation of antibodies that specifically recognize Asp-126-cleaved parkin as previously demonstrated, e.g. amyloid precursor protein (35).
Parkin cleavage may thus contribute to a vicious circle in neurons where caspase activation, initiated by e.g. hitherto unknown causes in sporadic Parkinson's disease, compromises parkin function that subsequently lowers the cellular stress threshold and thus leads to further caspase activation. Whether the degenerating caspase-positive neurons represent a reversible state is unknown but a reversible dysfunctional degenerative state has been demonstrated in a transgenic mouse model of Huntington's disease, where inhibition of the expression of the transgene polyglutamine-containing protein was followed by improvement of the motor symptoms of the mice (38).
We did not observe any cytoprotective effect of the D126E mutation as compared with wild type parkin when challenging the cells with OA, staurosporine, and camptothecin. However, this does not exclude the existence of such an effect when using another apoptotic paradigm as the protective effect of parkin previously have been demonstrated to be rather selective. Accordingly, parkin cleavage may represent a novel drug target in neuroprotective strategies against Parkinson's disease aiming at slowing the functional decline of those dopaminergic neurons that remain at the time of diagnosis. Transplantation of dopamine-producing cells represents another treatment strategy used in Parkinson's disease. Such cells are subject to a considerable stress as evidenced by the large caspase-dependent cell loss upon transplantation (39). Transgenic expression of parkin mutagenized at Asp-126 in these cells might represent an attractive method for improving the survival provided the mutation does not affect the enzymatic activity of parkin.