Proteasomal inhibition leads to formation of ubiquitin/α-synuclein-immunoreactive inclusions in PC12 cells

Authors


Address correspondence and reprint requests to Leonidas Stefanis, Department of Neurology, Columbia University, Black Building, Room 326, New York, NY 10032, USA. E-mail: Is76@columbia.edu.

Abstract

Proteasomal dysfunction has been recently implicated in the pathogenesis of several neurodegenerative diseases, including Parkinson's disease and diffuse Lewy body disease. We have developed an in vitro model of proteasomal dysfunction by applying pharmacological inhibitors of the proteasome, lactacystin or ZIE[O-tBu]-A-leucinal (PSI), to dopaminergic PC12 cells. Proteasomal inhibition caused a dose-dependent increase in death of both naive and neuronally differentiated PC12 cells, which could be prevented by caspase inhibition or CPT-cAMP. A percentage of the surviving cells contained discrete cytoplasmic ubiquitinated inclusions, some of which also contained synuclein-1, the rat homologue of human α-synuclein. However the total level of synuclein-1 was not altered by proteasomal inhibition. The ubiquitinated inclusions were present only within surviving cells, and their number was increased if cell death was prevented. We have thus replicated, in this model system, the two cardinal pathological features of Lewy body diseases, neuronal death and the formation of cytoplasmic ubiquitinated inclusions. Our findings suggest that inclusion body formation and cell death may be dissociated from one another.

Abbreviations used
ALLN

acetyl-leucyl-leucyl-norleucinal

BAF

Boc-Asp-FMK

cdk

cyclin-dependent kinase

CFTR

cystic fibrosis transmembrane conductance regulator

DLBD

diffuse Lewy body disease

DMSO

dimethyl sulfoxide

NGF

nerve growth factor

PAGE

polyacrylamide gel electrophoresis

PD

Parkinson's disease

PSI

ZIE[O-tBu]-A-leucinal

SDS

sodium dodecyl sulfate

UCH-L1

ubiquitin C-terminal hydrolase L1.

Ubiquitin-dependent proteolysis is a major avenue for degradation of intracellular proteins. Proteins are targeted for degradation by attachment of polyubiquitin chains through the action of a family of ubiquitin conjugating enzymes and ligases, and are then degraded by the multicatalytic proteasome complex (Ciechanover 1998). Growing evidence suggests that ubiquitin-dependent protein degradation may be impaired in a number of neurodegenerative diseases, including Parkinson's disease (PD) and diffuse Lewy body disease (DLBD). A key pathological feature in PD and DLBD is the formation of ubiquitinated cytoplasmic inclusions (Gibb and Lees 1988; Iwatsubo et al. 1996). In PD, cytoplasmic inclusions known as Lewy bodies are formed within the dopaminergic neurons of the substantia nigra pars compacta. These inclusions contain not only ubiquitin, but also α-synuclein, subunits of the proteasome, and other proteins (Ii et al. 1997; Spillantini et al. 1997; Sulzer and Zecca 2000). Biochemical analyses of isolated cortical Lewy bodies from postmortem tissue from the closely related condition DLBD have shown that the ubiquitin present in Lewy bodies is in the form of polyubiquitin chains rather than ubiquitin monomer (Iwatsubo et al. 1996). This suggests that polyubiquitinated proteins could accumulate in inclusions because of a dysfunction at the level of degradation by the proteasome. This idea is consistent with a report of decreased proteasomal activity in PD nigral tissue (McNaught and Jenner 2001).

Several genetic findings in familial cases of PD suggest an involvement of the ubiquitin-dependent degradation system. Two point mutations have been identified in the gene encoding α-synuclein (Polymeropoulos et al. 1997; Kruger et al. 1998) that have been suggested to affect its degradation by the proteasome (Bennett et al. 1999; but see also Ancolio et al. 2000). Mutations have also been reported in the gene encoding parkin (Kitada et al. 1998), a protein that has E3 (ubiquitin ligase)-like activity. The mutations abolish or diminish parkin's E3-like activity (Shimura et al. 2000). Finally, a point mutation in the gene encoding ubiquitin C-terminal hydrolase L1 (UCH-L1), a neuronal-specific deubiquitinating enzyme, can lead to reduced catalytic activity of UCH-L1 (Leroy et al. 1998), at least in vitro. Such a mutation may thus be expected to indirectly affect proteasomal function.

The availability of specific pharmacological inhibitors of the proteasome has provided a means to examine the cellular consequences of proteasomal dysfunction. Proteasomal inhibitors induce death in both neuronal and non-neuronal cells (Drexler 1997; Qiu et al. 2000), and increase levels of polyubiquitinated proteins (Figueiredo-Pereira et al. 1994; Ohtani-Kaneko et al. 1998). Proteasomal inhibitors also potentiate the formation of cytoplasmic and nuclear inclusions induced by over-expression of mutant polyglutamine repeat or other misfolded proteins (Johnston et al. 1998; Cummings et al. 1999). These inhibitors have also been found to induce formation of perinuclear ubiquitinated aggregations (Wojcik et al. 1996) that resemble the protein aggregates termed aggresomes formed following over-expression of wild-type or mutant cystic fibrosis transmembrane conductance regulator (CFTR; Johnston et al. 1998). However, whether proteasomal inhibition alone is sufficient to induce formation of ubiquitinated inclusions in dopaminergic neuronal cells, whether such inclusions contain α-synuclein, and the potential link between inclusion formation and cell death in this context has not been addressed. We used pharmacological inhibitors of the proteasome to create a cellular model of proteasomal dysfunction in PC12 cells, a dopaminergic cell line that assumes a neuronal phenotype following exposure to the neurotrophin nerve growth factor (NGF) (Greene and Tischler 1976), and has been extensively studied as a model for neuronal degeneration (Rukenstein et al. 1991; Stefanis et al. 1996).

Materials and methods

Cell culture

PC12 cells were grown as previously described (Stefanis et al. 1996) in Roswell Park Memorial Institute (RPMI) medium containing 5% fetal bovine serum and 10% heat-inactivated horse serum, and penicillin/streptomycin. Neuronally differentiated PC12 cells were established by treating PC12 cells with NGF (100 ng/mL; Upstate Biotechnology, Lake Placid, NY, USA) in RPMI/1% horse serum for 10–12 days.

Application of agents

Lactacystin (Kamiya, Seattle, WA, USA) and PSI (ZIE[O-tBu]-A-leucinal; Calbiochem, San Diego, CA, USA) were prepared as 1 mm (in H2O) and 10 mm (in dimethyl sulfoxide; DMSO) stock solutions, respectively, and diluted in serum-free medium prior to addition to the cells. The broad spectrum caspase inhibitors, Boc-Asp-FMK (BAF) or zVAD-fmk (Enzyme Systems Products, Dublin, CA, USA) were prepared as 100 mm stock solutions in DMSO. The cAMP analogue, CPT-cAMP was obtained from Roche (Indianapolis, IN, USA) and prepared as a 10-mm stock solution in serum-free RPMI.

Cell survival

For estimates of PC12 cell survival, cells were plated in 24-well plates and, at the indicated times, lysed as described (Rukenstein et al. 1991). Intact nuclei from triplicate wells were counted on a hemocytometer. Parallel cultures were plated on glass coverslips, fixed in 3.7% formaldehyde for 30 min at 4°C and stained with the nucleic acid specific dye YOYO-1 (1 : 1500; Molecular Probes, Eugene, OR, USA). To quantify apoptotic cell death, the percentage of cells with condensed and fragmented nuclei in three fields of 100 cells was determined. Each experiment was conducted at least twice.

Immunofluorescence

For immunofluorescence staining, fixed cells were blocked with 10% normal goat serum in phosphate buffer containing 0.4% Triton-X-100, and incubated with antibodies raised against ubiquitin (rabbit polyclonal; Dako, Carpinteria, CA, USA; 1 : 100), the rat homologue of α-synuclein, synuclein-1 (mouse monoclonal; Transduction Labs, Lexington, KY, USA; 1 : 40), or the cyclin dependent kinase inhibitor (cdk) p27 (mouse monoclonal, Santa Cruz Biotechnology, Santa Cruz, CA, USA; 1 : 100) for 1 h at room temperature followed by incubation with Alexa 488 (Molecular Probes; 1 : 100), Cy3 (Jackson Immuno-Research, West grove, PA, USA; 1 : 250) conjugated secondary antibodies. In some experiments, the cells were counterstained with the nuclear dyes YOYO-1 or Hoechst 33342 (Sigma, St Louis, MO, USA; 1 µg/mL). Images were acquired on a Zeiss LSM410 laser confocal microscope using oil immersion 40× or 100× objectives, or using standard epifluorescence and phase contrast microscopy with 40× objectives, as indicated. To quantify ubiquitinated cytoplasmic inclusions, slides were visualized under standard epifluorescence using a 100× oil-immersion objective. Cells were scored as positive if the ubiquitin immunoreactivity was localized to a discrete area, distinct from the nucleus, not encompassing the entire cytoplasmic volume.

Western blotting

Naive and neuronally differentiated PC12 cells were exposed to lactacystin (10 µm) for 8 or 24 h. Cells were collected and washed in ice-cold PBS and resuspended in lysis buffer (25 mm HEPES–NaOH pH 7.4, 5 mm MgCl2, 1 mm EDTA, 1 mm EGTA) containing 0.5% Triton X-100 and incubated on ice for 20 min. Detergent insoluble material was pelleted by centrifugation at 100 000 g for 20 min at 4°C, resuspended in sodium dodecyl sulfate (SDS) sample buffer and boiled for 10 min. Protein concentration of both detergent-soluble and -insoluble fractions was determined by the Bradford assay (Bio-Rad Laboratories, Hercules, CA, USA). Equal amounts of protein from both fractions were separated by SDS–PAGE (8% or 12%) and transferred to nitrocellulose membranes, blocked in 5% non-fat milk for 1 h at room temperature and incubated overnight at 4°C with the same monoclonal synuclein-1 antibody used to detect cytoplasmic inclusions (1 : 1000; Transduction Labs). Changes in ubiquitination were visualized by probing the membranes with a rabbit anti-ubiquitin antibody (Dako; 1 : 1000). Immunoreactive bands were visualized with horseradish peroxidase conjugated secondary antibodies (Pierce, Rockford, IL, USA) and the DuraWest chemiluminescence substrate (Pierce). To demonstrate equal protein loading, the membranes were stripped and re-probed with an anti-ERK2 antibody (rabbit polyclonal, 1 : 5000; Santa Cruz). Autoradiographic films were scanned and prepared using adobe photoshop software.

Results

Proteasomal inhibition leads to cell death of PC12 cells

As reported previously for PC12 cells and mouse neuroblastoma cells, proteasomal inhibition by either lactacystin or PSI elicited neurite outgrowth. Also, consistent with reports in both neuronal (Pasquini et al. 2000) and non-neuronal cells (Drexler 1997), inhibition of the proteasome with either lactacystin (Fig. 1) or PSI (not shown) induced cell death in naive PC12 cells. We found that lactacystin induced a dose-dependent increase in the percentage of naive PC12 cells with apoptotic nuclei at 24 h following exposure (Fig. 1a).

Figure 1.

Figure 1.

 Proteasomal inhibition causes cell death of naive PC12 cells. (a) Counts of apoptotic nuclei of naive PC12 cells exposed to vehicle (open bar) or increasing concentrations of lactacystin (filled bars) for 24 h, fixed and stained with the DNA binding dye YOYO-1. Means reflect counts of three fields of 100 cells from two or more experiments. (b) Appearance of apoptotic naive PC12 cells stained with YOYO-1 following exposure to lactacystin (10 µm) for 24 h (arrows); Scale bar = 10 µm. (c) Counts of intact nuclei of naive PC12 cells at baseline, or 40 h following exposure to vehicle, lactacystin (10 µm) or lactacystin simultaneously with BAF (100 µm), CPT-cAMP (100 µm), or NGF (100 ng/mL). The increase in control numbers relative to baseline reflects ongoing PC12 cell proliferation. Means reflect counts of three wells from two or more independent experiments. (d) Counts of apoptotic nuclei of naive PC12 cells exposed to vehicle, lactacystin (10 µm) or lactacystin together with BAF (100 µm), CPT-cAMP (100 µm), or NGF (100 ng/mL). Means reflect counts of three fields of 100 cells from two or more experiments. *p < 0.05 compared with baseline; **p < 0.05 compared with lactacystin alone, determined by one-way anova, with Neuman–Keul's post hoc tests.

Other cultures were treated with lactacystin simultaneously with agents known to prevent apoptosis induced by trophic factor deprivation in PC12 cells (Rukenstein et al. 1991; Stefanis et al. 1996), including the general caspase inhibitors BAF (100 µm) or zVAD-fmk (100 µm, not shown), the cAMP analogue CPT-cAMP (100 µm), and NGF (100 ng/mL). Each of these agents significantly reduced lactacystin-induced death, assessed by counting intact nuclei or apoptotic nuclei (Figs 1c and d).

PC12 cells cultured in the presence of NGF extend neurites and differentiate into a sympathetic neuron-like phenotype (Greene and Tischler 1976). Lactacystin reduced the survival of such neuronal PC12 cells dose-dependently, as determined by counts of intact nuclei (Fig. 2). Additionally, as in naive PC12 cells, lactacystin-induced death was prevented by the general caspase inhibitor BAF or the cAMP analogue CPT-cAMP (Fig. 2) in neuronally differentiated PC12 cells.

Figure 2.

 Proteasomal inhibition induces death of neuronally differentiated PC12 cells. PC12 cells were differentiated in the presence of NGF for 10–12 days. The percentage of intact nuclei expressed as a percentage of control following exposure to 1, 5, or 10 µm lactacystin for 36 h, or lactacystin (10 µm) together with BAF (100 µm) or CPT-cAMP (100 µm). Means reflect counts of three wells from two or more independent experiments. *p < 0.05 compared with control; #p < 0.001 compared with control; **p < 0.05 compared with lactacystin alone, determined by one-way anova, with Neuman–Keul's post hoc tests.

Proteasomal inhibition leads to the formation of ubiquitin/synuclein-positive cytoplasmic inclusions

As inhibition of the proteasome has also been shown to increase cellular levels of ubiquitinated proteins (Figueiredo-Pereira et al. 1994), and induce the appearance of ubiquitin and proteasome-positive perinuclear structures in HeLa cells (Wojcik et al. 1996), we asked if proteasomal dysfunction could also lead to the formation of discrete cytoplasmic ubiquitinated inclusions in dopaminergic PC12 cells. PC12 cells that were treated with either lactacystin or PSI were fixed and stained with antibodies against ubiquitin. Control PC12 cells displayed low-level cytoplasmic ubiquitin staining (Fig. 3a1). In lactacystin-exposed cells, there was an overall increase in ubiquitin levels shown by immunofluorescence (see Fig. 3b1). Apoptotic cells often showed a very intense pattern of diffuse ubiquitin immunoreactivity filling the entire cytoplasmic volume (Figs 3b1 and 3b2). In some other cells, the proteasome inhibitors induced a focal cytoplasmic accumulation of ubiquitin immunoreactivity (Figs 3c1, 3e1 and 3f1). Strikingly, these cytoplasmic ubiquitinated inclusions were seen exclusively in viable cells, based on nuclear morphology (see Fig. 3f2). The focal inclusions were distinctly different from the intense diffuse ubiquitin staining observed in apoptotic cells (compare Figs 3b1 to 3c1 and e1).

Figure 3.

Figure 3.

 Proteasomal inhibition leads to the formation of cytoplasmic inclusions that contain ubiquitin and synuclein-1. (a–c) Naive PC12 cells exposed to lactacystin (b1–2, 10 µm) for 24 h were fixed and stained for ubiquitin and the DNA binding dye YOYO-1. Note the intense diffuse ubiquitin staining in the apoptotic cell (b1, arrow) compared with the focal accumulation of ubiquitin (c1, arrow). Scale bar = 10 µm. (d–f) Naive PC12 cells were exposed to lactacystin (e1,2, 10 µm) or PSI (f1,2, 10 µm) for 24 h, fixed and double-stained for ubiquitin and synuclein-1. Note the increase in cell size following proteasomal inhibition. The nucleus of the cell in panel E is indicated by an asterisk. The cytoplasmic inclusions seen in e and f are positive for both ubiquitin and synuclein-1 (arrows). Scale bar = 10 µm.

We examined the cytoplasmic ubiquitinated inclusions to determine whether they also contained synuclein-1, the rat homologue of human α-synuclein, a major component of Lewy bodies (Spillantini et al. 1998). In control naive PC12 cells (Fig. 3d1), synuclein-1 exhibits a low-level cytoplasmic pattern of immunoreactivity, as we reported previously (Stefanis et al. 2001). Following proteasomal inhibition with lactacystin (Fig. 3e1) or PSI (Fig. 3f1), some of the ubiquitin-positive cytoplasmic inclusions also contained synuclein-1 immunoreactivity. As expected from the distribution of ubiquitinated inclusions, the synuclein-1 immunopositive inclusions were also observed only in viable cells based on nuclear staining with Hoechst (not shown). To examine the possibility that this colocalization was not specific, and merely reflected a generalized accumulation of polyubiquitinated proteins targeted for proteasomal degradation, we double immunostained PC12 cells for ubiquitin and a protein known to be regulated by the proteasome, the cdk inhibitor p27 (Pagano et al. 1995). p27 immunoreactivity was observed in control PC12 cells, localized to the nucleus. In lactacystin treated PC12 cells, p27 was never observed to colocalize with ubiquitin within the cytoplasmic inclusions formed following proteasomal inhibition (not shown). This suggests that α-synuclein may be preferentially susceptible to sequestration within ubiquitinated inclusions as a result of proteasomal dysfunction.

We then sought to determine whether such inclusions could also be seen in neuronally differentiated PC12 cells. Differentiated PC12 cells, grown under control conditions, fixed and stained with anti-ubiquitin antibodies appeared phase-bright with low-level ubiquitin staining using phase-contrast optics combined with epifluorescence (Figs 4a1 and 4a2). In contrast, a proportion of the PC12 cells treated with lactacystin for 24–36 h and identically processed displayed discrete phase-dark cytoplasmic structures (Fig. 4b1, arrow) which colocalized with discrete cytoplasmic ubiquitin immunostaining (Fig. 4b2, arrow). Higher-power images obtained using confocal microscopy (Fig. 4d) show that these discrete ubiquitinated inclusions occupy only a proportion of the cytoplasmic volume. Similar phase-dark cytoplasmic structures were observed in naive PC12 cells treated with proteasomal inhibitors (not shown).

Figure 4.

Figure 4.

 Proteasomal inhibition leads to formation of ubiquitin and synuclein-1 positive cytoplasmic inclusions in neuronally differentiated PC12 cells. PC12 cells were differentiated in the presence of NGF for 10–12 days. (a and b) Differentiated PC12 cells grown on glass coverslips were maintained in normal medium or treated with lactacystin (10 µm, 36 h), fixed and then stained for ubiquitin. Phase contrast and epifluorescent images were acquired with a 40 × objective. The nucleus in b1 is indicated by the arrowhead. Note the phase-dark cytoplasmic inclusion (arrow) in b1 that stains positive for ubiquitin (arrow) in b2. (c and d) Differentiated PC12 cells were treated as in a and b, fixed and stained for ubiquitin. Images were acquired by confocal microscopy using a 100X oil-immersion objective. Note that the ubiquitin-positive inclusion (arrow) does not occupy the entire cytoplasm. Scale bar = 10 µm. (e and f) Differentiated PC12 cells were treated with vehicle or lactacystin (10 µm, 30 h), fixed and double-stained for ubiquitin (red) and synuclein-1 (green). Note the diffuse synuclein-1 staining in control cells throughout the cytoplasm and nucleus. (f1–f2) Show a well defined cytoplasmic inclusion that is positive for both ubiquitin and synuclein-1 (arrow), and a less clearly defined inclusion (arrowhead) staining predominantly for synuclein-1. The cell shown in (f3–f4) is the same cell as in (f1–f2), however, in a lower focal plane, separated by approximately 3 µm. Note that the inclusion shown in (f1–f2) (arrow) is within the border of the cell and clearly distinct from the nucleus (indicated by an asterisk). Scale bar = 10 µm.

We then examined if synuclein-1 also colocalized with the ubiquitin-positive inclusions in neuronally differentiated PC12 cells. As in the naive PC12 cells treated with lactacystin, a proportion of the ubiquitin-positive cytoplasmic inclusions also contained intense synuclein-1 immunoreactivity (arrow in Figs 4f1 and f2). In Fig. 4, the same cell displayed in f1 and f2 is shown in f3 and f4 at a different focal plane, 3 µm apart, to demonstrate that the nucleus (asterisk in Fig. f4) is not contiguous with the inclusion.

The survival agents ZVAD-FMK and CPT-cAMP increase the percentage of ubiquitin-positive cytoplasmic inclusions

To examine the effects of the survival agents CPT-cAMP and zVAD-FMK on either inclusion formation, or the appearance of intense diffuse ubiquitin immunoreactivity characteristic of apoptotic cells (as shown in Fig. 3b), naive PC12 were exposed to these agents and immunostained for ubiquitin. Cells that had been treated with zVAD-FMK or CPT-cAMP in the absence of proteasome inhibitors showed no inclusion formation by these agents (not shown). Following proteasomal inhibition with lactacystin (36 h), the percentage of cells with inclusions (Fig. 5a), and the percentage of cells with intense diffuse ubiquitin immunoreactivity (Fig. 5b) both showed a significant increase. Both zVAD-FMK and CPT-cAMP increased the percentage of cells bearing inclusions compared with lactacystin alone (Fig. 5a), but prevented the increase in intense diffuse ubiquitin staining, characteristic of apoptotic cells (Fig. 5b).

Figure 5.

 Inhibition of cell death modulates ubiquitin staining in PC12 cells following proteasomal inhibition. Naive PC12 cells were exposed to lactacystin (10 µm) for 24 h, fixed and stained for ubiquitin. (a) Discrete ubiquitin positive inclusions present in a focal region within the cytoplasm were counted in three representative coverslips. Three fields of 100 cells were counted. (b) The percentage of PC12 cells with intense diffuse ubiquitin staining throughout the entire cytoplasmic volume. *p < 0.05 compared with control cells; **p < 0.05 compared with lactacystin alone, determined by one-way anova, with Neuman–Keul's post hoc tests.

We conclude therefore, that there is a dissociation between the effects of these agents on survival and on inclusion formation.

Synuclein-1 levels do not appreciably increase following proteasomal inhibition in PC12 cells

The issue of whether α-synuclein is ubiquitinated and degraded by the proteasome is controversial, with some groups reporting that synuclein levels are regulated by the proteasome (Bennett et al. 1999; Imai et al. 2000; but see Ancolio et al. 2000). Additionally, it has been shown that α-synuclein binds to a subunit of the proteasomal complex (Ghee et al. 2000). Given the localization of synuclein-1 within a proportion of the ubiquitin-positive cytoplasmic inclusions, we examined whether there were alterations in either overall levels of synuclein-1, or its migration pattern, following treatment with lactacystin, as would be expected if it is normally degraded through the proteasome.

We examined the migration pattern of endogenous synuclein-1 by SDS–PAGE in PC12 cells following proteasomal inhibition with lactacystin (8–24 h). As a control for the detection of accumulation of polyubiquitinated proteins, we used immunoblotting with an anti-ubiquitin antibody. Naive and differentiated PC12 cells were treated with lactacystin and then separated into detergent-soluble and detergent-insoluble fractions, to increase the sensitivity of detection of proteins with low solubility.

The upper panel of Fig. 6a1 shows the time-dependent accumulation of high molecular weight ubiquitinated proteins in both the detergent-soluble and -insoluble fractions from naive PC12 cells treated with lactacystin. A large proportion of the protein remained within the stacking gel (indicated by an arrow in a1 and a2) in the detergent-insoluble fraction. The middle panel of Fig. 6a2 shows the migration pattern of synuclein-1 in the same lysates. Levels of the 18 kDa species within the detergent-soluble fraction did not increase over the time points examined following proteasomal inhibition (arrowhead). We also observed a 45–50 kDa band, present only within the detergent-soluble fraction (indicated by as asterisk). Importantly, although high molecular weight bands were detected on these blots, we did not observe an increase in these bands, as would be expected if synuclein-1 were accumulating in a polyubiquitinated form. To demonstrate equal protein loading, the blots were stripped and re-probed with anti-ERK2 antibodies (Fig. 6a3).

Figure 6.

Figure 6.

 Synuclein-1 levels do not change following proteasomal inhibition. (a) Naive PC12 cells were treated with lactacystin (10 µm) for 8 or 24 h, lysed in low-detergent buffer (0.5% Triton X-100) and separated by centrifugation in to detergent soluble and insoluble fractions. a1 shows the accumulation of high molecular weight (HMW) ubiquitin-protein conjugates with lactacystin treatment over time. Note the increased accumulation particularly within the detergent insoluble fraction, and that considerable protein remains within the stacking gel (delineated by an arrow). a2 shows the pattern of synuclein-1 migration following proteasomal inhibition. Note the 45–50 kDa synuclein-1 reactive band (asterisk) and the very faint levels of the 18 kDa species of synuclein-1 (arrowhead). We detected no increase in synuclein-1 levels, nor did we detect an increase in HMW synuclein-1 reactive bands. The membrane in a2 was stripped and re-probed with ERK-2 (a3) to show equal protein levels. (b) PC12 cells were differentiated in the presence of NGF for 10–12 days, treated with lactacystin (10 µm) for 8 or 24 h and separated into detergent soluble and insoluble fractions as in naive PC12 cells. b1 shows the accumulation of HMW ubiquitin-protein conjugates following lactacystin treatment particularly within the insoluble fraction (stacking gel indicated by arrow). b2 shows the 18 kDa species of synuclein-1 (arrowhead) predominantly detected within the soluble fraction. Prolonged exposure of the same membrane (b3) revealed no increase in HMW synuclein-1 bands. The membrane in b3 was stripped and re-probed with ERK-2 (b4) to show equal protein levels.

The 45–50 kDa synuclein-1 immunoreactive band was seen previously in naive PC12 cells (Stefanis et al. 2001), whereas the expected 18 kDa band was not observed in these cells. The presence of the 18 kDa band in the present experiments may reflect clonal variations in the PC12 cell lines used.

We similarly examined the migration of ubiquitin and synuclein-1 in neuronally differentiated PC12 cells treated with lactacystin. As with naive PC12 cells, there was a dramatic accumulation of high-molecular weight ubiquitin immunoreactive bands in both the detergent-soluble and -insoluble fractions of differentiated PC12 cells exposed to lactacystin (Fig. 6b1). We have previously shown that synuclein-1 is selectively upregulated in PC12 cells following differentiation with NGF (Stefanis et al. 2001). Similarly, the second and third panel of Fig. 6(b2 and b3), 1 and 10 s film exposures, respectively, show high levels of the 18 kDa synuclein-1 species predominantly within the detergent-soluble fraction (arrowhead). However these bands do not significantly change with proteasomal inhibition. Additionally, we detected an 18-kDa band within the detergent-insoluble fraction of differentiated PC12 cells; levels of this species also did not change with proteasomal inhibition. As in naive cells, we did not detect an increase in higher-molecular weight synuclein-1 immunoreactive bands within either the detergent-soluble or the detergent-insoluble fractions. The blots were again stripped and re-probed with anti-ERK2 antibodies to demonstrate equal protein loading (Fig. 6b4).

We conclude that, although polyubiquitinated proteins accumulate in naive and neuronal PC12 cells treated with lactacystin, proteasomal inhibition does not produce a pattern consistent with synuclein-1 polyubiquitination or an increase in overall synuclein-1 levels.

Discussion

We show for the first time that following proteasomal inhibition, a population of both naive and neuronally differentiated PC12 cells form discrete cytoplasmic inclusions that appear as phase-dark cytoplasmic structures that are immunopositive for ubiquitin. Additionally, a proportion of the inclusions also show synuclein-1 immunoreactivity. These two proteins are the major components of Lewy bodies (Spillantini et al. 1997). We have thus reproduced in these cultures the two hallmarks of Lewy body disease: cell death and formation of cytoplasmic inclusions.

It has been shown previously that an over-expressed mutant form of the cystic fibrosis-related protein, CFTR, accumulates in perinuclear aggresomes that also stain with antibodies to ubiquitin and cytoskeletal associated proteins such as vimentin (Johnston et al. 1998). The formation of these aggresomes was accelerated by treatment with ALLN, an inhibitor of calpain and the proteasome. The inclusions observed in the present study share some similarities with the CFTR aggresomes, but are also different in some respects. Although a perinuclear localization was common (see Figs 3c1 and f1), in many cases the inclusions were not adjacent to the nucleus (see Figs 3e1 and 4f1). Aggresomes, on the other hand, typically assume a perinuclear distribution associated with the microtubule organizing center (see also Wojcik et al. (1996) who observed perinuclear inclusions in HeLa cells exposed to PSI). In addition, in most of the previous work reporting aggresomes, a misfolded protein has been overexpressed in the cells. In contrast, the present findings show that proteasomal dysfunction alone, without over-expression of an aberrant or misfolded protein, promotes formation of ubiquitin/synuclein-positive inclusions in naive and neuronally differentiated PC12 cells. Presumably, endogenous misfolded proteins are, at least in part, components of these inclusions.

The sequestration of endogenous ubiquitin and synuclein into cytoplasmic inclusions was also recently reported by Betarbet et al. (2000) in the substantia nigra of rats following chronic infusion of a complex I inhibitor. Transgenic mice (Masliah et al. 2000) or flies (Feany and Bender 2000) over-expressing wild-type or mutant human α-synuclein also show evidence of inclusion formation. The present work shows that physiological levels of synuclein can be sequestered within ubiquitinated inclusions following proteasomal inhibition. Such sequestration is evident even in naive PC12 cells, which express very low levels of synuclein-1 (Figs 3 and 6). It should be noted, however, that synuclein/ubiquitin-positive inclusions formed only a subpopulation of the total number of ubiquitin-positive inclusions, and that this percentage varied substantially amongst different clonal lines, ranging from nondetectable to approximately 20%.

Using SDS–PAGE analyses of detergent soluble and insoluble fractions of both naive and differentiated PC12 cells, we found a rapid accumulation of high molecular weight ubiquitin protein conjugates following proteasomal inhibition with lactacystin. The accumulation of high molecular weight ubiquitin bands was observed within both fractions, but particularly so in the detergent insoluble fraction, especially in the area of the stacking gel, where insoluble material is trapped. Using immunocytochemistry with anti-ubiquitin antibodies, we found discrete cytoplasmic inclusions within a relatively small percentage of cells treated with lactacystin, and only after 24–36 h of treatment. This suggests that there may be intermediate stages of protein aggregation within the cell following proteasomal inhibition that are detectable by western blotting, but do not appear in situ as focal ubiquitin-positive inclusions. The nature of the polyubiquitinated proteins that accumulate in the inclusions is unclear, but our results with the lack of p27 localization to the inclusions argue that not all polyubiquitinated substrates form part of these structures.

The presence of synuclein-1 within a proportion of the ubiquitinated inclusions following proteasomal inhibition suggested the possibility that the sequestered protein may be ubiquitinated and relatively insoluble and therefore resistant to detergent extraction. We examined the levels of the native (18 kDa) form of synuclein-1, as well as the appearance of higher molecular weight bands which would be suggestive of ubiquitination. Under the conditions we employed, we did not detect any changes in either the 18 kDa species of synuclein-1, or an accumulation of higher molecular weight bands suggestive of synuclein ubiquitination, in either naive or neuronally differentiated PC12 cells. There could be a number of reasons for this. First, given that there were relatively few numbers of ubiquitinated inclusions, and that only a portion of these also stain positive for synuclein-1, it is possible that our method of detection was simply not sensitive enough to detect the form of synuclein-1 present within the inclusions. Second, it is also possible that aggregated ubiquitinated synuclein is not soluble even in SDS buffer. Alternatively, synuclein-1 may not be degraded by the proteasome. Previous studies reporting proteasomal regulation of α-synuclein (Bennett et al. 1999; Imai et al. 2000) showed an accumulation of overexpressed human α-synuclein following proteasomal inhibition, as opposed to changes in levels of endogenous rat synuclein-1, as we have examined.

In the present study we have shown that pharmacological inhibition of the proteasome induces death of both naive and neuronally differentiated PC12 cells. Previous work has shown that proliferating cells are vulnerable to proteasomal inhibition, and that they die via an apoptotic mechanism, leading to early speculation that these cells were uniquely sensitive to proteasomal dysfunction, whereas quiescent postmitotic cells, such as neurons, were not (Drexler 1997). However, the present work and recent studies by others (Lopes et al. 1997; Qiu et al. 2000) show that postmitotic neurons and neuronally differentiated PC12 cells are also sensitive to proteasomal inhibition. Additionally, the cell death induced by proteasomal inhibition in both naive and neuronally differentiated PC12 cells was reduced by agents previously shown to prevent death induced by trophic factor deprivation (Rukenstein et al. 1991; Stefanis et al. 1996).

The inclusions induced in PC12 cells by proteasomal inhibitors were seen only in viable cells. While we cannot rule out the possibility that the intense diffuse ubiquitin immunoreactivity observed in apoptotic cells obscured the presence of ubiquitinated inclusions, we consider this an unlikely possibility for two reasons: first, under phase microscopy, the cytoplasmic inclusions appeared as phase-dark structures which were not observed in degenerating cells. Second, we did not observe synuclein-positive inclusions in degenerating cells; as synuclein-1 levels are quite low in such degenerating cells, the presence of inclusions should have been clearly visible against the background.

In contrast to viable cells harboring inclusions, apoptotic cells often showed a very intense diffuse pattern of ubiquitin staining. Agents that prevented cell death reduced the percentage of cells with intense, diffuse ubiquitin staining, but increased the percentage of cells with discrete cytoplasmic inclusions (see Fig. 5) without affecting overall levels of ubiquitination. These data show a dissociation between inclusion formation and cell death, and suggest that these phenomena are mediated by distinct cellular processes. Several authors have previously shown a similar dissociation between cell death and inclusion formation (Saudou et al. 1998; Kim et al. 1999) in other contexts. We cannot however, rule out the possibility that inclusion formation lies within the death pathway upstream of the point of action of CPT-cAMP or BAF.

While it was not possible in the current study to monitor the fate of individual cells harboring inclusions, one possibility is that the inclusions that are formed in nonapoptotic cells disassemble as the cell begins to undergo cell death, giving rise to the diffuse ubiquitin immunostaining. When death is prevented, the inclusions remain in the cell longer, and the intense diffuse ubiquitin immunostaining pattern does not occur.

Alternatively, the formation of inclusions and cell death could occur in different subpopulations of cells. This may be particularly relevant in neuronal populations in which α-synuclein levels, for example, are not homogeneous across all cells, as is the case in neuronally differentiated PC12 cells (Stefanis et al. 2001). This would be consistent with the possibility that cells able to form inclusions are relatively resistant to death. Future studies will attempt to clarify the relationship between inclusion formation and death following proteasomal inhibition and the underlying molecular components that regulate each of these cellular processes.

Acknowledgements

Supported by: NPF (HJR, KEL), PDF (LS, DS), Burroughs Wellcome Fund, NIH grant RO1 NS38586, and Matheson Foundation (LS), and a Udall Parkinson's Center of Excellence Award (DS). We wish to thank Dr Lloyd Greene for helpful discussions.

Ancillary