protein surveillance machinery in brains with spinocerebellar ataxia type 3: redistribution and...

Protein Surveillance Machinery in Brainswith Spinocerebellar Ataxia Type 3:

Redistribution and Differential Recruitmentof 26S Proteasome Subunits and Chaperones

to Neuronal Intranuclear InclusionsThorsten Schmidt, MS,1 Katrin S. Lindenberg, MS,2,3 Antje Krebs, BS,3 Ludger Schols, MD,4

Franco Laccone, MD,5 Jochen Herms, MD,6 Martin Rechsteiner, PhD,7 Olaf Riess, MD,1

and G. Bernhard Landwehrmeyer, MD2

Intracellular aggregates commonly forming neuronal intranuclear inclusions are neuropathological hallmarks of spino-cerebellar ataxia type 3 and of other disorders characterized by expanded polyglutamine-(poly-Q) tracts. To characterizecellular responses to these aggregates, we performed an immunohistochemical analysis of neuronal intranuclear inclu-sions in pontine neurons of patients affected by spinocerebellar ataxia type 3, using a panel of antibodies directed againstchaperones and proteasome subunits. A subset of the neuronal intranuclear inclusions stained positively for the chaper-ones Hsp90� and HDJ-2, a member of the Hsp40 family. Most neuronal intranuclear inclusions were ubiquitin positive,suggesting degradation by ubiquitin-dependent proteasome pathways. Surprisingly, only a fraction of neuronal intranu-clear inclusions were immunopositive for antibodies directed against subunits of the 20S proteolytic core, whereas mostinclusions were stained by antibodies directed against subunits of the 11S and 19S regulatory particles. These resultssuggest that the proteosomal proteolytic machinery that actively degrades neuronal intranuclear inclusions is assembledin only a fraction of pontine neurons in end stage spinocerebellar ataxia type 3. The dissociation between regulatorysubunits and the proteolytic core and the changes in subcellular subunit distribution suggest perturbations of the pro-teosomal machinery in spinocerebellar ataxia type 3 brains.

Ann Neurol 2002;51:302–310DOI 10.1002/ana.10101

An expansion of CAG triplet repeats forms the geneticbasis of a growing number of inherited neurodegenera-tive disorders and was shown to underlie at least six au-tosomal dominant spinocerebellar ataxias (SCAs), in-cluding SCA3 or Machado-Joseph disease.1–4 Since theCAG repeats are within the coding region, the mutantgene products are characterized by an expanded stretchof glutamines (poly-Q) in otherwise unrelated proteins.The current understanding of the mechanisms givingrise to a neurodegenerative phenotype of delayed onset isstill limited and incomplete. Current evidence favors theview that the expanded poly-Q stretch confers a newproperty on the respective proteins: poly-Q–expanded proteins display an increased propensity tomisfold and aggregate.5 One of the pathological hall-marks of these clinically and pathologically distinct

neurodegenerative disorders is therefore the presence ofnuclear aggregates of mutant proteins forming neuro-nal intranuclear inclusions (NIIs).6 The role of NIIs inthe pathogenesis of poly-Q disorders is unclear.7,8 Pre-vious work has demonstrated that the cellular responseto nuclear aggregates includes (1) the recruitment ofchaperones, or proteins involved in the folding of nas-cent translational products and in the resolubilizationof aggregated polypeptides; and (2) the ubiquitinationof aggregates, suggesting cellular attempts to degradedeposits of these mutant proteins via the ubiquitin-proteasomal pathway.9–11 The 26S proteasome repre-sents the most important site for protein degradationin eukaryotic cells12 and consists of a several compo-nents: 20S (containing the catalytic sites shielded insidea barrel-shaped multienzyme complex), 11S protea-

From the 1Department of Medical Genetics, University of Tubing-en; 2Department of Neurology, University of Ulm; 3Department ofNeurology, University of Freiburg; 4Department of Neurology,Ruhr-University Bochum; 5Institute of Human Genetics and 6De-partment of Pathology, University of Gottingen, Germany; and7Department of Biochemistry, University of Utah School of Medi-cine, Salt Lake City, UT.

Received May 23, 2001, and in revised form Nov 5. Accepted forpublication Nov 5, 2001.

Published online Feb 21, 2002.

Address correspondence to Dr Landwehrmeyer, Department ofNeurology, University of Ulm, Steinhoevelstrasse 9, D-89075 Ulm,Germany. E-mail: [email protected]

302 © 2002 Wiley-Liss, Inc.

some activator, and 19S regulatory particle (a multi-meric complex thought to mediate the recognition ofpolyubiquitin residues and the unfolding of proteins,thereby permitting access into the interior cavity of the20S component13). To characterize the cellular re-sponse to deposition of poly-Q–expanded ataxin-3, weperformed an immunohistochemical analysis of nuclearinclusions in the postmortem brains of patients af-fected by SCA3, using a panel of 33 antibodies di-rected against chaperones and proteasome subunits.

Patients and MethodsTissuesParaffin-embedded, frozen tissue from five patients with mo-lecular confirmation of a SCA3 mutation, an appropriatefamily history, and clinical findings consistent with SCA3, aswell as human brain tissue from five normal donors (withclinical, macroscopic, and histological findings within normalrange), were employed in this study as described previous-ly.14 Paraffin-embedded sections of human umbilical veinand carcinoma of the mamma were used as positive controlsfor the anti-Hsp antibodies.

AntibodiesDetails concerning the antibodies employed in the presentstudy are summarized in the Table. Twenty-five antibodiesdirected against 26S proteasomal subunits used in this studywere purchased from Affiniti Research Products (Exeter,United Kingdom). A total of 23 of these antibodies and an-tisera were subunit specific. An additional antibody directedto subunit 5a of the 19-S component was developed by oneof us (M.R.). The antibodies directed against heat shock pro-teins were delivered by NeoMarkers (Hsp 27, HDJ-2, Hsp60, Hsp 90�; Lab Vision Corp., Fremont, CA) and Stress-Gen (Victoria, Canada) (Hsp70/Hsc70, Hsc 70). Monoclo-nal and polyclonal antibodies directed against ubiquitin werepurchased from Biotrend (Cologne, Germany) and DAKO(Glostrup, Denmark), respectively. The polyclonal anti–ataxin-3 antibody used has been described previously,14 andthe monoclonal anti–ataxin-3 antibody (1H9) was kindlyprovided by Dr Y. Trottier.

ImmunohistochemistryImmunohistochemical experiments were conducted as previ-ously described.14 Briefly, after deparaffination, tissue sections(7�m) were rehydrated in a graded series of ethanols, treatedfor antigen retrieval,15 quenched (using 1% hydrogen peroxidein 40% methanol for 10 minutes), and blocked in phosphate-buffered saline (PBS) with 0.3% Triton X-100 containing 5%normal goat serum (polyclonal primary antiserum), 5% nor-mal horse serum (monoclonal primary mouse antibody), or5% normal rabbit serum (monoclonal primary rat antibody).Sections were incubated overnight at 4°C with the primaryantibody diluted in PBS containing 3% of the appropriatenormal serum (see table for dilutions) and rinsed, followed bydetection of the primary antibody using the Vectastain ABCelite system (Vector Laboratories, Burlingame, CA) accordingto the manufacturer’s directions and 3,3�-diaminobenzidine aschromogen (Sigma-Aldrich, Deisenhofen, Germany). Some

sections were lightly counterstained with hematoxylin, dehy-drated in a graded series of ethanols and xylene, coverslippedusing Eukitt (Kindler, Freiburg, Germany), and examined us-ing a light microscope equipped with Nomarski optics (Leica,Bensheim, Germany). Control sections incubated exclusivelywith the secondary antibody developed under the same con-ditions did not result in immunoreactivities (IR). For figures,pictures were digitally processed using Adobe Photoshop(Adobe Systems Inc., San Jose, CA).

Quantification of Nuclear Aggregates and SubcellularRedistribution of Core SubunitsFor quantification of nuclear aggregates, adjacent pontinesections sectioned perpendicularly to the brainstem axis were

Table. Dilutions of the Respective Antibodies andNomenclature of the Proteasome Subunits

Antibody/Subunit Type Dilution

Antibody against ataxin-3Ataxin-3 (1H9) mAb 1:1,000

Antibody against ubiquitinUbiquitin pAb 1:5,000

Antibodies against heat shock proteins/chaperonesHsp 27 mAb 1:500Hsp 40 (HDJ-2) mAb 1:500Hsp 60 mAb 1:500Hsp 70/Hsc 70 mAb 1:200Hsc 70 mrAb 1:50Hsp 90� (Hsp 86) pAb 1:1,000

Antibodies against 26S proteasome20S core pAb 1:1,000

20S �-type subunitsSubunits �1, 2, 3, 5, 6, and 7 mAb 1:500Subunit �2 (HC3) mAb 1:100Subunit �3 (HC9) mAb 1:200Subunit �4 (XAPC-7) mAb 1:100Subunit �5 (zeta) mAb 1:1,000Subunit �6 (HC2) mAb 1:2,000Subunit �7 (HC8) mAb 1:1,000

20S �-type subunitsSubunit �1 (Y) mAb 1:2,500Subunit �1i (Lmp2) pAb 1:1,000Subunit �2 (Z) mAb 1:2,000Subunit �2i (MECL-1) pAb 1:2,500Subunit �3 (HC10) mAb 1:1,000Subunit �5i (Lmp7) pAb 1:2,000Subunit �7 (HN3) mAb 1:1,000

11S (PA 28) regulator subunitsSubunit � (PA28 �) pAb 1:1,000Subunit � (PA28 �) pAb 1:1,000

19S (PA 700) regulator subunitsSubunit 4 (mts2) pAb 1:500Subunit 5a pAb 1:1,000Subunit 6a (Tbp1) pAb 1:1,000Subunit 6b (Tbp7) pAb 1:2,000Subunit 7 (Mss1) pAb 1:2,000Subunit 8 (p45) pAb 1:1,000Subunit 10a (p44) pAb 1:2,000Subunit 10b (p42) pAb 1:500

Nomenclature of the proteasome subunits according to Baumeisteret al.12

mAb � monoclonal mouse antibody; mrAb � monoclonal rat an-tibody; pAb � polyclonal rabbit antiserum.

Schmidt et al: Proteasome and Chaperones in SCA3 303

viewed using a Leica microscope at �40 magnificationequipped with a video camera (Dage, Michigan City, MI)and an image processing system (MCID, Imaging Research,St. Catharines, Ontario, Canada). In some stainings nuclearinclusions were of rather small apparent size; in these in-stances counting was repeated at �100 magnification. In anappropriate number of random fields of view, the total num-ber of neurons and the number of neurons containing un-ambiguous nuclear inclusion bodies were counted by threeindependent observers until at least 100 neuronal profileswere encountered. The fields of view were randomly chosenbut systematically distributed over the entire section. Thepercentage of neurons containing nuclear inclusions for eachexperiment was estimated by calculating the mean of all re-spective counts. To determine the subcellular redistributionof 20S core antigens, sections containing pontine neuronswere inspected using �20 magnification until at least 100neuronal profiles were encountered.

ResultsFrequency of Neuronal Intranuclear Inclusions inPontine Neurons in SCA3To assess the frequency of NIIs, we stained pontinesections from SCA3 patients with a monoclonal anti-body directed to the carboxyterminus of ataxin-3,which is known to label NIIs with high intensity.14,16

In brains unaffected by SCA3, incubation with 1H9(Fig 1A) resulted in a neuronal staining, preferentiallyin the cytoplasmatic compartment. In pontine neuronsof SCA3 patients, however, an intense IR of NIIs wasobserved. NIIs as visualized by 1H9 (see Fig 1A) ap-peared round, with a mean cross-sectional area of5.2�m2 (standard deviation [SD], 2.1; range,1.6–8.7). In numerous neuronal nuclei, two NIIsforming a figure 8 could be seen. In all regions of the

Fig 1. Immunohistochemical staining of pontine neurons in spinocerebellar ataxia type 3 (SCA3) brains demonstrated recruitmentof chaperones and 19S proteasomal subunit antigens to neuronal intranuclear inclusions (NIIs). (A) Anti-ataxin-3 immunoreactivity(IR) using the monoclonal antibody 1H9. Note the double inclusion distinct from the nucleolus. (B) Anti-ubiquitin-IR in a pontineneuron counterstained with hematoxylin. IR is confined to NIIs. (C) Anti-Hsp90-IR. (D) Anti-subunit-5a-IR. Subunit 5a isthought to be a 19S subunit in the lid portion of the 19S particle mediating binding to polyubiquitin chains. (E) Anti-subunit-7-(Mss1)-IR. Subunit 7 is part of the base of the 19S particle and has adenosine triphosphatase activity (F) Anti-subunit-10a-IR.Subunit 10a is a component of the 19S particle without adenosine triphosphatase activity. Bar � 10�m. Nomarski optics.

304 Annals of Neurology Vol 51 No 3 March 2002

ventral pons, 1H9-IR-NIIs were found in pontine neu-rons with no obvious regional clustering. NIIs were ob-served in most pontine neurons: in 100 neuronal pro-files sampled at random, three independent observerscounted 72 (� 11) neurons bearing NIIs (Fig 2).

Staining with polyubiquitin antibodies (Fig 1B) inadjacent sections disclosed a somewhat lower frequencyof ubiquitin-IR NIIs (51 � 10 per 100 neuronal pro-files; see Fig 2). Ubiquitin-IR-NIIs tended to besmaller than 1H9-IR-NIIs (mean cross-sectional area,2.1�m2 SD, 1.4; range, 0.8–5.1). Thus, it appearedthat immunohistochemical staining with the 1H9 an-tibody disclosed the highest number of NIIs, settingthe gold standard for the estimation of NII frequency.Therefore, in the following paragraphs, NIIs as labeledby other antibodies will be expressed as percentages of1H9-IR-NIIs.

Antibodies against Heat Shock Protein StainNeuronal Intranuclear Inclusions in SCA3Immunohistochemical staining of pontine sectionswith antibodies to heat shock proteins demonstrated anintense staining of a subset of NIIs by a polyclonal an-tibody to Hsp90� (Fig 1C). In 100 pontine neuronssampled in 5 random sets of 20 neurons, we observedan average of 15 neurons containing Hsp90�-IR-NIIs;this number corresponded to a staining of 20% ofNIIs. However, a monoclonal antibody to HDJ-2,which is a member of the Hsp40 family, displayed IRonly in a small subset of NIIs. No staining of NIIs wasobserved with monoclonal antibodies for Hsp27,Hsp60, Hsp70, or Hsc70, despite clear immunoreac-tive signals in control tissues processed in a similarmanner. Compared with control brains, there was noobvious redistribution of Hsp-IR aside from an intense

labeling of some NIIs: as in controls, Hsp90�, Hsc/Hsp70, and Hsp40 IR were preferentially observed incytoplasm. There was no evidence of a marked increasein intensity of cytoplasmatic Hsp-IR in SCA3 brainscompared with controls; antibodies resulting in faint orno labeling in control brain tissue (eg, Hsp70) did notgive rise to more intense labeling in SCA3 brains.

A Small Subset of Neuronal Intranuclear Inclusionsin SCA3 Displays Proteasome 20S-Subunit-IRUsing a polyclonal 20S core antibody raised against aproteasomal preparation isolated from red blood cells,we observed IR in a small subset of NIIs (approximately3%). A monoclonal antibody recognizing a peptide se-quence shared by six of the seven subunits forming the�-ring of the 20S proteosome resulted in a similarly lowfrequency of immunoreactive NIIs (approximately 5%).Immunohistochemical staining using monoclonal anti-bodies selective for �-subunits 2–7 under conditions re-sulting in distinct signals in control brains disclosed IR-NIIs for antibodies selective for �2, �5, �6, and �7 atrelatively low abundance, ranging from 4 to 15% ofNIIs. Numerous neuronal profiles in pontine sections ofSCA3 brains displayed no NII staining when antibodiesselective for �-subunits were used. No NIIs were en-countered in sections stained with antibodies against �3and �4. Antibodies to subunits of the �-rings of the 20Sparticle resulted in a similarly low number of labeledNIIs. In comparison, the highest percentage of NIIs wasseen in sections stained with antibodies selective for �2and �3 (less than 20% of NIIs; Fig 3). Antibodies di-rected to three alternative �-subunits (�1i, �2i, and �5i)inducible by �-interferon did not result in staining ofNIIs. Taken together, antibodies directed to 20S sub-units labeled an average of 5 � 3 NIIs per 100 neuronalprofiles (see Fig 2).

Most Neuronal Intranuclear Inclusions in SCA3Exhibit 19S- and 11S-Proteasome-IRIn marked contrast to the results obtained with thepanel of antibodies directed to 20S proteasomal sub-units, subunit-specific antibodies directed to the 19Sparticle consistently resulted in labeling of numerousNIIs (Fig 1D, E, and F). The eight subunit-specific an-tibodies used stained an average of 37 � 10 NIIs per100 neuronal profiles (see Fig 2). For example, the an-tibody recognizing the structural subunit 5a, which isthought to mediate binding of the polyubiquitin chain17

(see Fig 1D), labeled a mean number of 35 NIIs per 100neuronal profiles. Antibodies directed to the subunit S7stained even higher numbers of NIIs (57 � 7) (see Fig1E). This subunit has been shown to be associated withadenosine triphosphatase activity and is presumably in-volved in unfolding proteins prior to degradation by the20S core particle. In yeast, the homologue of subunit S7

Fig 2. Bar graph indicating the relative frequency of neuronalintranuclear inclusions (NIIs) per 100 neuronal profiles inpontine neurons of spinocerebellar ataxia type 3 (SCA3)brains, with abundant inclusions labeled with antibodies toataxin 3 and ubiquitin and panels of antibodies to compo-nents of the 26S proteasome (see text). Error bars � standarddeviation.


is part of the base portion of the 19S particle. Similarnumbers were observed with an antibody to the struc-tural subunit S10a (see Fig 1F) of the lid portion of the19S particle (56 � 15 NIIs; see Fig 3).

Antibodies to the two subunits constituting the 11Sparticle, PA28� and �, resulted in preferentially cyto-plasmatic staining in pontine neurons of normal brain(Fig 4A and C) and labeled a substantial number ofNIIs (Fig 4B and D; see Fig 2), approaching the num-ber observed with antibodies to ubiquitin (see Fig 2). Asummary of the labeling of NII by proteasomal sub-units is provided in Fig 3.

Pontine Neurons in SCA3 DemonstrateRedistribution of Proteasome 26S-Subunit-IRIn both control and SCA3 brains, the overall cellularstaining pattern obtained with antibodies to proteasomalantigens varied substantially from one neuronal profileto the next. In some neurons a cytoplasmatic stainingpattern predominated, while neighboring neurons dis-played prominent nuclear staining, presumably reflectingdifferent functional states of the respective cells. Overallnuclear IR obtained with 26S antibodies in pontine neu-rons was less prominent in SCA3 patients (13%; SD �16) than in controls (67%; SD � 20), suggesting a rel-ative depletion of proteasomal subunits in the nuclearcompartment in at least a subset of pontine neurons. Inneurons containing NIIs, nuclear IR was concentrated toNIIs at the expense of overall nuclear staining in 54% of

neuronal profiles (Fig 5A–D). In contrast, cytoplasmaticIR in pontine neurons of SCA3 patients was over-all more prominent than in control brains, consistentwith redistribution or upregulation of components ofthe 26S proteasome (see Fig 5A–D). Counting morethan 2,000 pontine neurons stained with 20S antibodiesdemonstrated a higher percentage of neurons with pre-dominantly cytoplasmatic staining in SCA3 (34%;SD � 25) compared with controls (3%; SD � 7). Inaddition, using antibodies selective for inducible�-subunits of the 20S core component, we observed acytoplasmatic IR in many pontine neurons of SCA3brains, well in excess of the signals obtained in controlbrains (Fig 5E and F): 64% of pontine neurons in SCA3brains but only 6% of pontine neurons in control brainsdisplayed a predominantly cytoplasmatic IR using an an-tibody to �2i.

DiscussionIn this study we provide what we believe is thus far themost comprehensive immunohistochemical analysis ofchaperones and proteasomal subunits in pontine neu-rons of SCA3 brains, using a total of 33 antibodies.

Distinct Chaperones Colocalize with a Subset ofNeuronal Intranuclear Inclusions in SCA3 BrainsWe observed labeling of NIIs with antibodies to heatshock protein (Hsp) 90� and HDJ-2, a member of theHsp 40 family of chaperones. Both chaperones stained

Fig 3. Schematic model of the association of neuronal intranuclear inclusions (NIIs) composed of carboxyterminal, polyglutamine-expanded fragments of ataxin-3, ubiquitin, chaperones (Hsp90 and HDJ-2), and proteasomal subunits. Proteasomal subunits arearranged according to proposals by Hendil and colleagues48 for 20S and Gorbea and colleagues with Glickman and colleagues49,50

for 19S. Shading indicates the frequency with which the antigens were detected in association with NIIs.


only a subset of NIIs. The presence of Hsp90 in NIIsof SCA3 brains has not been reported previously; nolabeling of NIIs in pontine SCA3 neurons was ob-served in an earlier study that used a different immu-noreagent for Hsp90.18 Hsp90 can prevent protein ag-gregation and participates in the folding of a specificsubset of proteins, eg, signal-transducing molecules,19

presumably by binding unstable folding intermediatesand maintaining them in a foldable state.20 In addi-tion, Hsp90 has been implicated in the degradation ofcertain mutant proteins21,22 and copurifies with theproteasome.23 Hsp90 is thought to act in associationwith several cochaperones, including Hsp40 and 70,giving rise to discrete subcomplexes (see Caplan for areview24). In SCA3 pontine neurons, a small subset ofNIIs was labeled by an antibody to HDJ-2, in agree-ment with independent observations.9,10,18 In ourmaterial we failed to detect NIIs labeled with the Hsp/Hsc70 antibodies used. Chai and colleagues demon-strated rare pontine neurons in SCA3 containingHsp70 positive NIIs; in addition, these cells displayedelevated cytoplasmatic IR for Hsp70, suggesting a heatshock response.18 These observations and similar find-ings in SCA110 indicate the relative paucity of NIIsdecorated by the Hsp40 and Hsp70 family of chaper-ones in human poly-Q brain (in contrast to a trans-genic animal model for SCA7, in which NIIs immu-noreactive to Hsp40 and Hsp70 represented most ifnot all NIIs15). A number of recent studies indicatethat overexpression of Hsp40 and Hsp70 reduces

poly-Q toxicity while altering but not abolishingNIIs.18,25–29 The paucity of NIIs decorated by Hsp40and Hsp70 and the lack of evidence of chaperone up-regulation suggest that this potential mechanism of re-ducing poly-Q toxicity is employed by only a minorityof neurons in end stage SCA3 brains.

Accompaniment of Neuronal Intranuclear Inclusionsin SCA3 by Proteasomal SubunitsThe ubiquitination of most NIIs in SCA3 brains andthe association of components of the 26S proteasomecomplex with NIIs indicate that the ubiquitin-proteasome pathway is involved in degrading mutantataxin-3.11 The present study extends previous reports,however, by disclosing that (1) the immunohistochem-ical composition of NIIs in pontine neurons of SCA3is not uniform, and (2) only a small fraction of NIIs inpontine neurons of end stage SCA3 assemble the pro-teosomal machinery thought to be required for proteindegradation.

To study the association of the catalytic 20S com-ponent consisting of four stacked, heptameric ringswith NIIs, we used antibodies to all seven subunits oftwo outer rings of �-subunits, six of which were sub-unit specific,30 and seven mono- and polyclonal anti-bodies to probe the two inner rings of �-subunits.30

An average of 8% of NIIs were immunoreactive tothe 15 antibodies directed to 20S (see Fig 2 and 3).Since these antibodies were directed to differentepitopes of the 20S core component, epitope masking

Fig 4. Immunohistochemical staining of pontine neurons in normal brain (A and C) and spinocerebellar ataxia type 3 (SCA3)brain (B and D) using antibodies directed to subunits of the 11S component of the proteasome. (A) Anti-PA28�-IR in normalbrain. (B) Anti-PA28�-IR in SCA3 brain. (C) Anti-PA28�-IR in normal brain. (D) Anti-PA28�-IR in SCA3 brain. Bar �10�m. Nomarski optics.


is unlikely to account for this disproportionally lowlabeling compared with 11S- and 19S-associatedepitopes. A similar low labeling using 20S antibodieswas observed in brains with Huntington’s disease(Krebs et al, unpublished results) and a cellular modelof Kennedy’s disease,9 suggesting that the proteasomalmachinery that actively degrades NIIs is assembled inonly a fraction of neurons. A failure to recruit theessential catalytic 20S particle to most NIIs may con-tribute to the impaired clearance and gradual accu-mulation of mutant ataxin-3 in SCA3 brains. In ad-dition, the shredderlike function of the 20Scomponent may be impeded when encountering poly-Q–expanded polypeptides, since they contain no pre-dicted cleavage sites.31

Prior to proteolytic cleavage, polypeptides have to beunfolded and then threaded through the narrow centralopenings of the outer �-rings of the 20S core parti-cle.13,32 The 19S component of the 26S proteasome is

thought to mediate this process in an adenosinetriphosphate–consuming fashion.13 Polyclonal anti-bodies selective for all six adenosine triphosphatases re-sulted consistently in the staining of numerous NIIs,suggesting that aggregates of mutant ataxin-3 resist un-folding or at least require the prolonged or persistentrecruitment of 19S subunits. Aside from its function inunfolding polypeptides, the 19S complex is critical inrecognition of ubiquitinated substrates and containssites for tight binding of polyubiquitin chains. An an-tibody to this putative “polyubiquitin receptor,” S5a, aswell as an antibody to a structural subunit located inthe lid portion of the 19S particle, S10a, resulted instaining of many NIIs, suggesting the recruitment ofthe entire 19S particle (and not only fragments thereof,eg, the base portion) to NIIs.

Another regulator of proteasome function is the 11Sregulatory particle, a heterodimeric complex composedof three pairs of alternating subunits called PA28� and

Fig 5. Immunohistochemical staining of pontine neurons in normal brain (A, C, E) and spinocerebellar ataxia type 3 (SCA3)brain (B, D, F) using subunit-specific antibodies directed to subunits of the 20S core component of the proteasome demonstratedreduced nuclear staining and more prominent cytoplasmatic staining in SCA3 tissues. (A) Anti-�5-IR in normal brain. (B) Anti-�5-IR in SCA3 brain. (C) Anti-�2-IR in normal brain. (D) Anti-�2-IR in SCA3 brain. (E) Anti-�2i-IR in normal brain. (F)Anti-�2i-IR in SCA3 brain. Bar � 20�m.


�,33 which can form a cap at either end of the 20Score (see Fig 3).12,34,35 We used antibodies selective foreach subunit and observed a surprisingly high numberof NIIs labeled with either antibody, suggesting the ac-companiment of most NIIs by the 11S particle. Theassociation of 19S, 20S, and 11S with NIIs may indi-cate the recruitment of heterocomplexes of protea-somes, in which a 19S complex caps one opening ofthe 20S core and an 11S particle the other open-ing.36,37 Alternatively, the PA28 complex in associationwith NIIs may have a function independent from 20S:Minami and colleagues recently demonstrated thatPA28 has a critical role in Hsp90-dependent proteinrefolding.20

Proteasomal Redistribution in Pontine Neuronsof SCA3There is evidence for a redistribution of proteasomalantigens in SCA3 brains. In neurons containing NIIs,nuclear IR obtained with subunit-specific proteasomalantibodies was often concentrated in NIIs at the ex-pense of overall nuclear staining, suggesting a depletionof certain proteasomal subunits. The apparent deple-tion was particularly obvious for antibodies to 19S sub-units and some 20S subunits, resulting in intense nu-clear staining under physiological conditions. Clearly,the assumption that the apparent sequestration of someproteasomal subunits into NIIs results in a functionaldepletion of some proteasomal subunits in the nuclearcompartment requires biochemical confirmation.

Aside from these alterations in the nuclear stainingpattern, overall cytoplasmatic staining in pontine neu-rons of SCA3 patients appeared more intense than incontrol brains (see Fig 4). In particular, cytoplasmaticstaining with antibodies to inducible �-subunits of 20Swas more prominent than in control brains, compati-ble with redistribution or upregulation of some com-ponents of the 26S proteasome. A quantitative confir-mation using Western blot analyses and mRNA studiesis required; we could not carry out these studies due toa lack of suitable material. The functional conse-quences of this apparent redistribution are currentlyunknown. However, the appearance of inducible cata-lytic 20S �-subunits in the cytoplasmatic compartmentsuggests that protein catabolism in the cytoplasmaticcompartment may be enhanced, possibly contributingto atrophy of affected neurons similar to muscle wast-ing in cachexia.38

Perturbations of Proteasomal Degradation PathwaysTaken together, these observations suggest a perturba-tion of proteasomal protein degradation in SCA3brains. Recent studies of cellular models of poly-Q dis-orders indeed support impaired proteasomal functionin the presence of poly-Q–expanded proteins.39,40 Cel-lular consequences of impaired proteasomal function

may include alterations in gene expression (as describedin poly-Q disorders)41–45 caused by interference withthe degradation of transcriptional regulators or withmitochondrial functions, potentially leading to neuro-nal death.39,40 Since protein deposits are a pathologicalhallmark of many neurodegenerative diseases, includingAlzheimer’s disease, Pick’s disease, Parkinson’s disease,and prion disorders, these disorders may be viewed asproteinopathies in which disease-specific proteins mis-fold46 and eventually form aggregates tagged by ubiq-uitin and associated with proteasomal antigens.47 Per-turbations of the proteasomal machinery asdemonstrated in this study for SCA3 may therefore bepart of the pathophysiology of other neurodegenerativediseases as well.

This study was supported by grants from the Deutsche Forschungs-gemeinschaft (Ri682/7-1, awarded to OR, and SFB 505, TPC2,awarded to GBL); the German Hereditary Ataxia Foundation(awarded to OR); and the Graduierten-Kolleg fur Molekulare Medi-zin (awarded to KSL)

We thank the families of the patients, whose generosity made thisresearch possible. We also thank Dr Y. Trottier for providing anti-bodies directed to ataxin-3.

References1. Kawaguchi Y, Okamoto T, Taniwaki M, et alCAG expansions

in a novel gene for Machado-Joseph disease at chromosome14q32.1. Nat Genet 1994;8:221–228.

2. Durr A, Stevanin G, Cancel G, et al. Spinocerebellar ataxia 3and Machado-Joseph disease: clinical, molecular, and neuro-pathological features. Ann Neurol 1996;39:490–499.

3. Schols L, Amoiridis G, Epplen JT, et al. Relations between ge-notype and phenotype in German patients with the Machado-Joseph disease mutation. J Neurol Neurosurg Psychiatry 1996;61:466–470.

4. Schols L, Amoiridis G, Langkafel M, et al. Machado-Josephdisease mutations as the genetic basis of most spinocerebellarataxias in Germany. J Neurol Neurosurg Psychiatry 1995;59:449–450.

5. Zoghbi HY, Orr HT. Glutamine repeats and neurodegenera-tion. Annu Rev Neurosci 2000;23:217–247.

6. Davies SW, Beardsall K, Turmaine M, et al. Are neuronal in-tranuclear inclusions the common neuropathology of triplet-repeat disorders with polyglutamine-repeat expansions? Lancet1998;351:131–133.

7. Davies SW, Turmaine M, Cozens BA, et al. Formation of neu-ronal intranuclear inclusions underlies the neurological dysfunc-tion in mice transgenic for the HD mutation. Cell 1997;90:537–548.

8. Klement IA, Skinner PJ, Kaytor MD, et al. Ataxin-1 nuclearlocalization and aggregation: role in polyglutamine-induced dis-ease in SCA1 transgenic mice. Cell 1998;95:41–53.

9. Stenoien DL, Cummings CJ, Adams HPet al. Polyglutamine-expanded androgen receptors form aggregates that sequesterheat shock proteins, proteasome components and SRC-1, andare suppressed by the HDJ-2 chaperone. Hum Mol Genet1999;8:731–741.

10. Cummings CJ, Mancini MA, Antalffy B, et al. Chaperone sup-pression of aggregation and altered subcellular proteasome lo-calization imply protein misfolding in SCA1. Nat Genet 1998;19:148–154.


11. Chai Y, Koppenhafer SL, Shoesmith SJ, et al. Evidence for pro-teasome involvement in polyglutamine disease: localization tonuclear inclusions in SCA3/MJD and suppression of polyglu-tamine aggregation in vitro. Hum Mol Genet 1999;8:673–682.

12. Baumeister W, Walz J, Zuhl F, et al. The proteasome: para-digm of a self-compartmentalizing protease. Cell 1998;92:367–380.

13. Voges D, Zwickl P, Baumeister W. The 26S proteasome: a mo-lecular machine designed for controlled proteolysis. Annu RevBiochem 1999;68:1015–1068.

14. Schmidt T, Landwehrmeyer GB, Schmitt I, et al. An isoform ofataxin-3 accumulates in the nucleus of neuronal cells in affectedbrain regions of SCA3 patients. Brain Pathol 1998;8:669–679.

15. Yvert G, Lindenberg KS, Picaud S, et al. Expanded polyglu-tamines induce neurodegeneration and trans-neuronal alter-ations in cerebellum and retina of SCA7 transgenic mice. HumMol Genet 2000;9:2491–2506.

16. Paulson HL, Perez MK, Trottier Y, et al. Intranuclear inclu-sions of expanded polyglutamine protein in spinocerebellarataxia type 3. Neuron 1997;19:333–344.

17. Deveraux Q, Ustrell V, Pickart C, et al. A 26S protease subunitthat binds ubiquitin conjugates. J Biol Chem 1994;269:7059–7061.

18. Chai Y, Koppenhafer SL, Bonini NM, Paulson HL. Analysis ofthe role of heat shock protein (Hsp) molecular chaperones inpolyglutamine disease. J Neurosci 1999;19:10338–10347.

19. Pratt WB. The role of the hsp90-based chaperone system insignal transduction by nuclear receptors and receptors signalingvia MAP kinase. Annu Rev Pharmacol Toxicol 1997;37:297–326.

20. Minami Y, Kawasaki H, Minami M, et al. A critical role for theproteasome activator PA28 in the Hsp90-dependent protein re-folding. J Biol Chem 2000;275:9055–9061.

21. Imamura T, Haruta T, Takata Y, et al. Involvement of heatshock protein 90 in the degradation of mutant insulin receptorsby the proteasome. J Biol Chem 1998;273:11183–11188.

22. Takata Y, Imamura T, Iwata M, et al. Functional importanceof heat shock protein 90 associated with insulin receptor oninsulin-stimulated mitogenesis. Biochem Biophys Res Commun1997;237:345–347.

23. Conconi M, Petropoulos I, Emod I, et al. Protection from ox-idative inactivation of the 20S proteasome by heat-shock pro-tein 90. Biochem J 1998;333:407–415.

24. Caplan AJ. Hsp90’s secrets unfold: new insights from structuraland functional studies. Trends Cell Biol 1999;9:262–268.

25. Wyttenbach A, Carmichael J, Swartz J, et al. Effects of heatshock, heat shock protein 40 (HDJ-2), and proteasome inhibi-tion on protein aggregation in cellular models of Huntington’sdisease. Proc Natl Acad Sci U S A 2000;97:2898–2903.

26. Muchowski PJ, Schaffar G, Sittler A, et al. Hsp70 and hsp40chaperones can inhibit self-assembly of polyglutamine proteinsinto amyloid-like fibrils. Proc Natl Acad Sci U S A 2000;97:7841–7846.

27. Jana NR, Tanaka M, Wang G, et al. Polyglutamine length–de-pendent interaction of Hsp40 and Hsp70 family chaperoneswith truncated N-terminal huntingtin: their role in suppressionof aggregation and cellular toxicity. Hum Mol Genet 2000;9:2009–2018.

28. Warrick JM, Chan HY, Gray-Board GL, et al. Suppression ofpolyglutamine-mediated neurodegeneration in Drosophila bythe molecular chaperone HSP70. Nat Genet 1999;23:425–428.

29. Kazemi-Esfarjani P, Benzer S. Genetic suppression of polyglu-tamine toxicity in Drosophila. Science 2000;287:1837–1840.

30. Hendil KB, Kristensen P, Uerkvitz W. Human proteasomesanalysed with monoclonal antibodies. Biochem J 1995;305:245–252.

31. Kuttler C, Nussbaum AK, Dick TP. An algorithm for the pre-diction of proteasomal cleavages. J Mol Biol 2000;298:417–429. See also PAProC (Prediction Algorithm for Protea-somal Cleavages) online at URL: http://www.paproc.de

32. Braun BC, Glickman M, Kraft R, et al. The base of the pro-teasome regulatory particle exhibits chaperone-like activity. NatCell Biol 1999;1:221–6.

33. Song X, Mott JD, von Kampen J, et al. A model for the qua-ternary structure of the proteasome activator PA28. J BiolChem 1996;271:26410–26417.

34. Dubiel W, Pratt G, Ferrell K. Purification of an 11 S regulatorof the multicatalytic protease. J Biol Chem 1992;267:22369–22377.

35. Ma CP, Slaughter CA, DeMartino GN. Identification, purifi-cation, and characterization of a protein activator (PA28) of the20 S proteasome (macropain). J Biol Chem 1992;267:10515–10523.

36. Tanahashi N, Murakami Y, Minami Y, et al. Hybridproteasomes: induction by interferon-gamma and contributionto ATP-dependent proteolysis. J Biol Chem 2000;275:14336–14345.

37. Hendil KB, Khan S, Tanaka K. Simultaneous binding of PA28and PA700 activators to 20 S proteasomes. Biochem J 1998;332:749–754.

38. Lecker SH, Solomon V, Mitch WE, Goldberg AL. Muscle pro-tein breakdown and the critical role of the ubiquitin-proteasome pathway in normal and disease states. J Nutr 1999;129:227S–237S.

39. Bence N, Sampat R, Kopito R. Impairment of the ubiquitin-proteasome system by protein aggregation. Science 2001;292:1467–1468.

40. Jana NR, Zemskov EA, Wang G, Nukina N. Altered proteaso-mal function due to the expression of polyglutamine-expandedtruncated N-terminal huntingtin induces apoptosis by caspaseactivation through mitochondrial cytochrome c release. HumMol Genet 2001;10:1049–1059.

41. Cha JH. Transcriptional dysregulation in Huntington’s disease.Trends Neurosci 2000;23:387–392.

42. Lin X, Antalffy B, Kang D, et al. Polyglutamine expansiondown-regulates specific neuronal genes before pathologicchanges in SCA1. Nat Neurosci 2000;3:157–163.

43. Luthi-Carter R, Strand A, Peters NL, et al. Decreased expres-sion of striatal signaling genes in a mouse model of Hunting-ton’s disease. Hum Mol Genet 2000;9:1259–1271.

44. Wood JD, Nucifora FC, Duan K, et al. Atrophin-1, thedentato-rubral and pallido-luysian atrophy gene product, inter-acts with ETO/MTG8 in the nuclear matrix and represses tran-scription. J Cell Biol 2000;150:939–948.

45. McCampbell A, Taylor JP, Taye AA, et al. CREB-binding pro-tein sequestration by expanded polyglutamine. Hum MolGenet 2000;9:2197–2202.

46. Hardy J, Gwinn-Hardy K. Genetic classification of primaryneurodegenerative disease. Science 1998;282:1075–1079.

47. Ii K, Ito H, Tanaka K, Hirano A. Immunocytochemical co-localization of the proteasome in ubiquitinated structures inneurodegenerative diseases and the elderly. J Neuropathol ExpNeurol 1997;56:125–131.

48. Hendil KB, Welinder KG, Pedersen D, et al. Subunit stoichi-ometry of human proteasomes. Enzyme Protein 1993;47:232–240.

49. Gorbea C, Taillandier D, Rechsteiner M. Assembly of the reg-ulatory complex of the 26S proteasome. Mol Biol Rep 1999;26:15–19.

50. Glickman MH, Rubin DM, Fu H, et al. Functional analysis ofthe proteasome regulatory particle. Mol Biol Rep 1999;26:21–28.


protein surveillance machinery in brains with spinocerebellar ataxia type 3: redistribution and...

Documents