The extracellular chaperone clusterin influencesamyloid formation and toxicity by interacting withprefibrillar structures
Justin J Yerbury,* Stephen Poon,†,1 Sarah Meehan,‡,2 Brianna Thompson,*,3
Janet R. Kumita,† Christopher M. Dobson,† and Mark R. Wilson*,4
*School of Biological Sciences, University of Wollongong, Wollongong, Australia; †Department ofChemistry, University of Cambridge, Cambridge, UK; and ‡School of Chemistry and Physics,University of Adelaide, Adelaide, Australia
ABSTRACT Clusterin is an extracellular chaperonepresent in all disease-associated extracellular amyloiddeposits, but its roles in amyloid formation and proteindeposition in vivo are poorly understood. The currentstudy initially aimed to characterize the effects ofclusterin on amyloid formation in vitro by a panel ofeight protein substrates. Two of the substrates (Alzhei-mer’s beta peptide and a PI3-SH3 domain) were thenused in further experiments to examine the effects ofclusterin on amyloid cytotoxicity and to probe themechanism of clusterin action. We show that clusterinexerts potent effects on amyloid formation, the natureand extent of which vary greatly with the clusterin:substrate ratio, and provide evidence that these effectsare exerted via interactions with prefibrillar speciesthat share common structural features. Proamyloido-genic effects of clusterin appear to be restricted toconditions in which the substrate protein is present at avery large molar excess; under these same conditions,clusterin coincorporates with substrate protein intoinsoluble aggregates. However, when clusterin ispresent at much higher but still substoichiometric levels(e.g., a molar ratio of clusterin:substrate�1:10), it po-tently inhibits amyloid formation and provides substan-tial cytoprotection. These findings suggest that clus-terin is an important element in the control ofextracellular protein misfolding.—Yerbury, J. J., Poon,S., Meehan, S., Thompson, B., Kumita, J. R., Dobson,C. M., Wilson, M. R. The extracellular chaperoneclusterin influences amyloid formation and toxicity byinteracting with prefibrillar structures. FASEB J. 21,2312–2322 (2007)
Numerous age-related, systemic, and neurologicaldisorders are associated with the deposition of highlystructured protein aggregates, usually known as amy-loid or amyloid-like fibrils, including Alzheimer’s dis-ease, Parkinson’s disease, amyotrophic lateral sclerosis,and Creutzfeldt-Jakob disease (1). In many cases the
deposits formed in these and other protein conforma-tion disorders are located extracellularly, where theyexert pathogenic effects by organ disruption or bycytotoxicity. Although the processes that control thefolding of proteins inside cells are relatively well under-stood, little is known about the corresponding extracel-lular processes. Clusterin is the best-characterizedabundant extracellular chaperone and has recentlybeen proposed to form part of an extracellular proteinquality control system (2).
Amyloid fibrils found in vivo exhibit common struc-tural features independent of the identity of the parentprotein. Intracellular amyloid aggregates are foundcolocalized with components of the intracellular pro-tein quality control system, including chaperones andubiquitin (3). In a remarkable parallel observation, alldisease-associated insoluble extracellular protein de-posits tested, including those characterized as amyloid,colocalize with clusterin (Table 1). The roles ofclusterin in amyloid formation and protein deposi-tion are, however, poorly understood. Clusterin is awell-conserved secreted glycoprotein found in mostextracellular fluids; it has a potent ATP-independentchaperone action similar to that of the small heatshock proteins. It inhibits stress-induced amorphousprotein aggregation by binding to exposed regions ofhydrophobicity on non-native protein conformationsto form high-molecular-weight (HMW) but still solu-ble complexes (4).
Limited data from previous studies suggest that clus-terin can affect the amyloid-forming process both invitro and in vivo. Clusterin has been reported to inhibitin vitro amyloid formation by apolipoprotein C-II
1 Current address: School of Biological Sciences, Universityof Wollongong, Wollongong NSW 2522, Australia.
2 Current address: Department of Chemistry, University ofCambridge, Cambridge CB21EW, UK.
3 Current address: Intelligent Polymer Research Institute,University of Wollongong, Wollongong NSW 2522, Australia.
(apoC-II) (5), the A� peptide (6–8), and a fragment ofthe prion protein, PrP (106–126) (9). Depending onthe conditions, however, it has been reported to eitherpromote or suppress the cytotoxicity of A� (6–8, 10).Similarly, work with PDAPP mice (a transgenic mousemodel for Alzheimer’s disease) in which clusterin ex-pression is ablated has provided results that sometimesappear contradictory. For example, when comparedwith matched littermates, ablation of clusterin expres-sion decreased the levels of thioflavin S staining ofmaterial in brain sections and the number of visiblydamaged neurons in PDAPP mice. These findings wereinterpreted to indicate that clusterin expression pro-moted A� amyloid formation and toxicity (11). How-ever, in a background of apolipoprotein E-negative(apoE�/�) PDAPP mice, the ablation of clusterinexpression had the opposite effect, promoting the earlyonset of A� deposition and material staining withthioflavin S (12). Such evidence shows that the in vivoeffects of clusterin on amyloid formation are likely toinvolve multiple interactions and processes, making itcritical to better understand the nature and mecha-nism(s) of interactions between clusterin and amyloid-forming proteins through in vitro studies. This objectivewas the global aim of the current study in which theeffects of clusterin on amyloid formation by a broadrange of unrelated proteins were examined using avariety of complementary approaches. In addition, weselected two protein substrates—one associated withdisease (A�1–42, Alzheimer’s disease) and the other not(a PI3-SH3 domain)—to extend these investigations toexamine the effects of clusterin on amyloid-relatedtoxicity and to better characterize the mechanism(s) bywhich clusterin affects amyloid formation.
MATERIALS AND METHODS
Materials
Clusterin was purified from human serum obtained fromWollongong Hospital (Wollongong, NSW, Australia), as de-scribed previously (13). Hexafluoroisopropanol (HFIP), ly-sozyme (from hen egg white), bovine serum albumin (BSA),
and �-casein (from bovine milk) were purchased from Sigma(St. Louis, MO, USA). A plasmid encoding �-synuclein was agift from Dr. Robert Cappai (Department of Pathology,University of Melbourne, Melbourne, Australia). �-Synucleinwas expressed in Escherichia coli and purified by acid precipi-tation as described (14). Glutathione-S-transferase (GST)from Schistosoma japonicum was prepared by thrombin cleav-age of recombinant Jun leucine zipper-GST fusion proteinand purified by GSH-agarose affinity chromatography (15).The short coiled-coil � (cc�) peptide, originally designed denovo as a model that transforms from a helical conformationat 20°C into amyloid fibrils at 37°C (16), was modified byadding a tryptophan residue at its N terminus to producecc�W (a kind gift from Dr. Cait MacPhee, Department ofPhysics, University of Edinburgh, Edinburgh, UK). The cc�fibrils described herein were indistinguishable from thosedescribed previously. Calcitonin was purchased from bothAuspep (Melbourne, Australia) and Southampton Polypep-tides Limited (Southhampton, UK). �2-Microglobulin was akind gift from Prof. Sheena Radford (University of Leeds,UK). A plasmid encoding the amyloidogenic PI3-SH3 domainof bovine phosphatidyl-inositol-3�-kinase (hereafter referredto simply as SH3) as a GST fusion protein was a kind gift fromDr. Jesus Zurdo (University of Cambridge, UK). SH3 wasexpressed in E. coli, purified, and GST cleaved using a 5 mlGSTTrapFF cartridge (GE Healthcare, Sydney, Australia)following the manufacturer’s instructions. A�1–42 was pur-chased from Biopeptide (San Diego, CA, USA), resuspendedin HFIP, and divided into aliquots in which the solvent wasleft to evaporate (the peptide “film” was frozen at �80°C).Monoclonal anti-A� antibody WO2 supernatant was a kindgift from Dr. Kevin Barnham (Department of Pathology,University of Melbourne, Australia).
Fibril formation in vitro
SH3 solutions (250 �M, unless otherwise indicated) or mix-tures of SH3 and clusterin in phosphate-buffered saline (PBS;137 mM NaCl, 2.7 mM KCl, 1.5 mM KH2PO4, and 8 mMNa2HPO4, pH 7.5) were heated to 60°C while shaking at 500rpm for 72 h in a Thermo Finemixer SH2000-DX (FinemouldPrecision Ind. Co., Seoul, Korea). Directly before use, A�1–42was resuspended in buffer (two parts 20 mM NaOH diluted inseven parts Milli Q water and 1 part 10 � PBS) and centri-fuged to remove any aggregated material. A�1–42 (10 �M,unless otherwise indicated) in the presence or absence ofclusterin was incubated at 37°C in oxidizing buffer thatconsisted of PBS containing 0.9 mM CaCl2, 0.5 mM MgCl2,100 �M CuCl2, 600 �M glycine, pH 7.5, while shaking for 8 hin a FLUOstar OPTIMA fluorescence plate reader (BMG
TABLE 1. Protein deposition disorders in which clusterin has been found colocalized with extracellular protein deposits
Disease Main constituent Referencea
Alzheimer’s disease A� (26)Creutzfeldt-Jakob disease PrP (27)Gerstmann-Straussler-Scheinker disease PrP (28)Gelatinous drop-like corneal dystrophy Keratoepithelin (29)Lattice type I corneal dystrophy M1S1 (29)Age-related macular degeneration Drusen (30)Pseudoexfoliation (PEX) syndrome PEX material (31)Down’s syndrome A� (32)HCHWA-Dutch type A� (33)Familial British dementia ABri (34)Atherosclerosis LDL/ApoB100 (35)
aReferences refer to studies that have found clusterin colocalized with deposits related to the respective disease.
2313CLUSTERIN INTERACTS WITH PREFIBRILLAR STRUCTURES
Labtech, Mornington, Victoria, Australia). In similar experi-ments, the relative ability of clusterin to inhibit fibril forma-tion in reactions that were seeded with preformed aggregatesof the same protein (sampled at different times during aprevious aggregation reaction) was examined. SH3 (500 �M)and A�1–42 (10 �M) were incubated as described above;samples were then taken from each aggregation reaction atvarious time points and stored frozen at �20°C. Thesesamples were used to seed fibril formation reactions at a finalmolar ratio of preaggregated substrate molecules to mono-mers of 1:9 in the presence or absence of clusterin (at molarratios of clusterin:SH3�1:100, clusterin:A��1:50). The fol-lowing proteins were all treated as described below, with andwithout added clusterin. Samples of �-synuclein (�-syn, 70�M), lysozyme (lys, 70 �M), and �-casein (�-cas, 52 �M) in 50mM Na2HPO4 buffer, pH 7.4, were shaken at 500 rpm and57°C for 192 h, 172 h, and 72 h, respectively, in an IKA VibraxVXR orbital shaker (IKA® Works, Inc., Wilmington, NC,USA). �2-Microglobulin (�2m, 85 �M) in 25 mM sodiumacetate, pH 2.1, was shaken for 336 h at 500 rpm and 37°C ina Thermo finemixer SH2000-DX. cc�W (60 �M) in 0.1 MNa2HPO4, pH 7.8, was shaken for 40 min at 37°C in aFLUOstar OPTIMA fluorescence plate reader. Calcitonin(calc, 150 �M) was incubated in 50 mM Na2HPO4 buffer, pH7.4, at 37°C for 20 h. In other experiments, to confirm thatclusterin did not form thioflavin T reactive aggregates underthe conditions used, clusterin (at 1.0–12.5 �M) was incubatedalone in 50 mM Na2HPO4 buffer, pH 7.4, for 24 h at 37°C andfor up to 192 h at 60°C or in 25 mM sodium acetate, pH 2.1,for 336 h at 37°C.
Thioflavin T fluorescence assays
Thioflavin T (50 �M) was added to aliquots of samples takenat specific time points after the initiation of fibril formation(lys, �-cas, �-syn, �2m, SH3) or to the reaction mixture at thebeginning of the time course (calc, A� and cc�W). Fluores-cence was measured on a FLUOstar OPTIMA fluorescenceplate reader using excitation and emission windows of 450 10 and 490 10 nm, respectively.
Transmission electron microscopy (TEM)
Formvar and carbon-coated nickel electron microscopy gridswere prepared by the addition of 2 �l of protein sample at aconcentration of 1 mg/ml. After several minutes, the gridswere washed with 3 � 10 �l H2O and negatively stained with10 �l of uranyl acetate [2% (w/v), Agar Scientific Ltd.,Stansted, Essex, UK]. The grids were dried with filter paperbetween each step. Samples were viewed under 20–125 Kmagnifications at 120 kV excitation voltages using a PhilipsCM100 transmission electron microscope and images wereanalyzed using the SIS Megaview II Image Capture system(Olympus, Munich, Germany).
Cytotoxicity
SH-SY5Y cells (a kind gift from Dr. Kevin Barnham, Depart-ment of Pathology, University of Melbourne, Melbourne,Australia) were cultured at 37°C and 5% (v/v) CO2 in fullmedium, which consisted of DMEM:F12 medium containing2.5% (v/v) fetal bovine serum (FBS) (both from TraceBiosciences, Melbourne, Australia). Cells suspended in fullmedium were added to a 96-well plate (100 �l/well contain-ing 5000 cells) and left to attach overnight before washingwith DMEM:F12. The cells were then cultured as above for48 h in FBS-free AIM-V medium (Invitrogen, Melbourne,Australia), with or without additives. Fibril formation by SH3
(250 �M) or A� (10 �M) was initiated as described above; insome reactions, clusterin or a control protein (BSA) wasincluded to give molar ratios of clusterin/BSA:substrate of1:500 (SH3) or 1:10 (A�). To analyze cytotoxicity, aliquots ofSH3 and A� reactions (taken at 12 h and 2 h, respectively)were added to cells alone to give final concentrations of 10�M and 1 �M, respectively, or in some cases (from reactionslacking clusterin/BSA) were supplemented with clusterin orBSA to give clusterin/BSA:substrate � 1:10. In other experi-ments, clusterin or BSA alone was added to cells to give a finalconcentration of 1.0 �. Calcein-AM was used to measurecell viability (17). Calcein-AM (1 �M) was added to cells andleft to incubate for 30 min before analyzing fluorescenceusing a FLUOstar OPTIMA plate reader and excitation andemission windows of 485 10 nm and 520 10 nm,respectively. The significance of differences in fluorescencewas assessed using the Student’s t test. Calcein-AM is mem-brane permeable and nonfluorescent; it is cleaved by ester-ases in the cytoplasm of viable cells to release fluorescent,membrane-impermeable calcein that remains trapped insideviable cells. Thus, the resulting level of cell-associated calceinfluorescence is proportional to the number of viable cells.
Effects of clusterin on the sedimentation properties ofsubstrate protein aggregates
At the conclusion of in vitro fibril formation time courses,samples of SH3 and A�, with or without clusterin, werecentrifuged for 30 min at 10,000 g. The supernatant wasremoved and the pellet was resuspended, then washed repeat-edly in PBS. SH3 samples were analyzed by 15% SDS-PAGEand stained with Coomassie blue. A� samples were analyzedby 15% SDS-PAGE and subsequent immunoblotting usinganti-A� monoclonal antibody. The presence of clusterin insupernatant and pelleted fractions was tested by applyingthese fractions to nitrocellulose membranes that were subse-quently blocked with HDC [1% (w/v) heat-denatured casein,0.04% (w/v) thimerosal, in PBS]. The presence of clusterinwas detected using a mixture of G7, 41D, and 78E monoclo-nal anticlusterin antibodies (18). Bound anti-A� and anticlus-terin antibodies were detected with HRP-conjugated sheepanti-mouse Ig antibody (Silenus, Melbourne, Australia), fol-lowed by enhanced chemiluminescence (ECL) with Supersig-nal West Pico Chemiluminescent Substrate (Pierce, Rock-ford, IL, USA).
Detection of stable clusterin-substrate complexes
Samples of SH3 and A�, with or without clusterin, taken fromthe beginning and end of in vitro fibril formation time courseswere centrifuged for 30 min at 10,000 g. The supernatantswere shaken end-over-end for 1 h at room temperature withanticlusterin antibody coupled to Sepharose beads (G7-Sepharose; 100 �l packed volume) (4). The G7-Sepharosewas washed by centrifugation three times with PBS, thenincubated in 2M guanidine HCl in PBS at pH 7.5 for 15 minto elute bound protein. The mixture was then centrifugedusing a 0.45 �m Ultrafree-MC centrifugal filter device (Milli-pore, Billerica, MA, USA) to separate the beads from theeluted proteins, which were subsequently analyzed by SDS-PAGE under nonreducing conditions. To detect A�, immu-noblotting was performed as described above.
Immuno-dot blots
Samples of amyloidogenic proteins were taken at various timepoints during fibril formation and frozen at �20°C untilrequired. Samples (1 �g) were spotted onto nitrocellulose
2314 Vol. 21 August 2007 YERBURY ET AL.The FASEB Journal
membranes (Pall, Pensacola, FL, USA) and allowed to dry;the membranes were then blocked with HDC. The mem-branes were incubated for 2 h at 37°C in PBS containing 10�g/ml clusterin or control proteins GST and ovalbuminbefore being washed with PBS. Bound clusterin was detectedusing a mixture of G7, 78E, and 41D antibodies. Rabbitanti-GST (Silenus, Melbourne, Australia) and anti-ovalbumin(a gift from S. Easterbrook-Smith, University of Sydney) wereused to detect any bound control protein. Bound primaryantibodies were detected with sheep anti-mouse Ig-HRP orsheep anti-rabbit Ig-HRP (Silenus, Melbourne, Australia)using ECL as described above.
RESULTS
Clusterin affects fibril formation in vitro
Under the various conditions used, the aggregatesformed from all proteins showed 1) increased thioflavinT fluorescence (Fig. 1A), 2) green birefringence whenstained with Congo Red (data not shown), and 3)fibrillar structures detected by TEM (Fig. 1B) charac-teristic of amyloid species. The addition of clusterin tosolutions of all proteins tested showed a dose-depen-dent decrease in thioflavin T fluorescence (Fig. 1A).However, for calcitonin, �-synuclein, and A�, molarclusterin:substrate ratios of 1:50, 1:100, and 1:500,respectively, produced an increase in thioflavin T fluo-rescence. The minimum molar ratio of clusterin:sub-strate required to affect the levels of thioflavin Tfluorescence differed greatly among the systems tested,which could reflect differences between the substratesand/or the conditions used. A ratio of clusterin:CC�wof only 1:100 was sufficient to reduce thioflavin Tfluorescence to almost background levels. Molar ratiosof clusterin:substrate needed to inhibit 70–80% of thethioflavin T fluorescence measured in the absence ofclusterin were 1:100 for �-casein and 1:50 for �2-microglobulin, SH3, and A�. Although under the con-ditions tested much higher levels of clusterin wereneeded to similarly inhibit fibril formation of calcito-nin, lysozyme, and �-synuclein, even in these cases theeffect was markedly substoichiometric (clusterin:sub-strate �1:10). The presence of a control protein, BSA,had no measurable effect on thioflavin T fluorescencefor any of the proteins tested (data not shown). Whenincubated alone, clusterin did not develop significantthioflavin T reactivity under any of the conditionstested (data not shown).
TEM was used to examine the morphology of proteinaggregates (at the highest molar ratio of clusterin:substrate shown for each protein in Fig. 1A). In theabsence of clusterin, fibrillar aggregates of dimensionsexpected for amyloid fibrils in these well-characterizedsystems were observed for all proteins (Fig. 1B). Underthe conditions tested, clusterin inhibited the formationof fibrillar structures in all cases (Fig. 1B). A variety ofstructures was observed in reactions containing clus-terin, including spherical particles of differing diame-ter (Fig. 1B; �-cas, A�, and CC�w) and amorphousaggregates (Fig. 1B; Lys, calc, �2M, and SH3). TEM
Figure 1. Effect of clusterin on fibril formation. A) Thiofla-vin T fluorescence (in arbitrary units, AU) of proteinspecies present at the end of aggregation reactions in thepresence (gray bars) or absence (white bars) of variousamounts of clusterin (indicated as molar ratios of clusterin:substrate). In each case the data shown are means oftriplicates and the error bars are ses of the mean. B) TEMimages of final samples taken from aggregation reactionscontaining either no clusterin (�clust) or the highestmolar ratio of clusterin:substrate indicated in panel A(�clust). In all cases, results shown are representative oftwo or more individual experiments. The scale bar shownin the upper right panel applies to all TEM images shownand represents 200 nm.
2315CLUSTERIN INTERACTS WITH PREFIBRILLAR STRUCTURES
analysis of samples identified as having increased thio-flavin T fluorescence resulting from a low molar ratioof clusterin:substrate showed them to contain fibrillaraggregates (data not shown).
Effects of clusterin on aggregate toxicity
Compared to SH3 alone, at a clusterin:SH3 ratio of1:10, the toxicity of aggregates generated after 72 h wassignificantly reduced (Fig. 2A; P 0.05), but at a clus-terin:SH3 ratio of 1:500 the toxicity of aggregatesgenerated after 48 and 72 h was significantly increased(Fig. 2A; P 0.05). At a clusterin:A� 1:10 ratio, clusterinsignificantly suppressed the toxicity of aggregates gen-erated after 2 h and 8 h (P 0.05), but at a clusterin:A�1:500 ratio the toxicity of aggregates at 4 h was signifi-cantly enhanced (P 0.05). Moreover, when clusterinwas added to aggregates of SH3 and A� preformed inthe absence of clusterin, it significantly decreased theircytotoxicity (Fig. 2B; P 0.001 and P 0.05, respec-tively). Under the conditions tested, the addition ofeither clusterin or BSA alone to cultures had nosignificant effect on cell viability (data not shown).
Clusterin binds amyloid-forming proteins and affectstheir sedimentation properties
Clusterin, at various ratios relative to the substrate, wasadded to fibril formation reactions of SH3 and A�; atthe end of the time course the samples were centri-fuged to separate supernatant and pellet fractions. SH3fractions were analyzed by SDS-PAGE; in the absence ofclusterin most of the SH3 protein was found in thepellet (P) fraction (Fig. 3A). However, at clusterin:SH3ratios of 1:100–1:10, almost all the SH3 protein wasfound in the supernatant (S) fraction. Even at a ratio ofclusterin:SH3 of 1:500, most of the SH3 remained inthe S fraction (Fig. 3A). Similarly, immunoblots of A�fibril formation reactions showed that the proportionof A� in the S fraction increased as the ratio ofclusterin:A� increased until (at clusterin:A��1:10) ef-fectively all of the A� was present in the S fraction (Fig.3A). In the absence of clusterin, A� was detected as itsmonomeric form (at �4.5 kDa) or as SDS-resistant highmolecular mass aggregates of � � 180 kDa (19).However, at clusterin:A� � 1:10, much of the nonsedi-menting A� was detected in a broad HMW band, whichmay represent an SDS-resistant clusterin-A� complex(6). Indeed, immunoblotting demonstrated the pres-ence of both A� and clusterin in this HMW band (datanot shown). For both SH3 and A�, at a clusterin:substrate ratio of 1:10, clusterin was detected only inthe S fractions, but at a lower ratio (1:500) was foundpredominantly in the P fractions (Fig. 3B). To deter-mine whether clusterin was forming stable complexeswith the substrate proteins, we immunoadsorbed clus-terin (and hence anything to which it was bound) fromthe initial and final samples of fibril formation reac-tions of SH3 and A�, and analyzed this material bySDS-PAGE and immunoblotting. In fractions prepared
from samples of clusterin and substrate protein, beforeincubation no SH3 or A� was detected; similarly, whenincubated in the absence of clusterin for the durationof the respective time courses, neither SH3 nor A� wasbound by immobilized anticlusterin Ig (Fig. 3C). How-ever, immunoadsorbed fractions prepared from finalsamples containing clusterin and substrate protein
Figure 2. Effects of clusterin on cytotoxicity of SH3 and A�. A)Reduction in SH-SY5Y cell viability (measured as a decrease incalcein fluorescence; see Materials and Methods), expressedas a percentage of corresponding values for untreated controlcultures, after incubation for 48 h with protein aggregatestaken from SH3 and A� fibril formation reactions at the timesindicated on the x axis in the absence or presence of clusterin(see key). *Significant differences between the effects of aclusterin-containing sample and the corresponding SH3 orA�-only sample (P 0.05); in each case, the significantlydifferent data point is the one closest to the asterisk. B)Changes in the viability of SH-SY5Y cells were measured asabove 48 h after adding 1) aggregates generated in SH3 or A�fibril formation reactions lacking clusterin (�clust) or 2) thesame aggregates added together with clusterin (to give clus-terin:SH3�1:10 and clusterin:A��1:10) (�clust). A, B) Datapoints represent means of triplicate determinations; errorbars � se. All results shown are representative of at least twoindependent experiments.
2316 Vol. 21 August 2007 YERBURY ET AL.The FASEB Journal
showed bands corresponding to SH3 and A� (Fig. 3C).This result demonstrates that clusterin formed stablecomplexes with the substrate proteins at some pointduring the course of the aggregation reactions.
Clusterin binds to intermediate structures on thefibril-forming pathway
Samples taken at different times from fibril formationreactions performed (in the absence of clusterin) forthe same panel of eight proteins previously examinedwere spotted on nitrocellulose membranes; the thiofla-vin T reactivity of species present in these reactions was
also measured as a function of time (Fig. 4). Whenassayed by immuno-dot blot, clusterin did not bind tothe native proteins present in samples before incuba-tion, and bound only weakly or not at all to finalsamples. Clusterin showed no detectable binding tofibrils after they had been centrifuged and washed(data not shown). In most cases, the strongest bindingof clusterin detected was to transient “intermediate”species; in the case of A�, �-synuclein, and cc�w,maximum binding was detected in species presentduring the transition between the lag and growthphases (Fig. 4). For lysozyme and �-casein, which underthese conditions have no detectable lag phase, thestrongest binding was to species present early in thegrowth phase (Fig. 4). For SH3, the strongest bindingwas detected at 4–6 h, the initial stage at whichthioflavin T fluorescence reached a maximum value.There was no binding of clusterin detected in speciespresent at any time in �2-microglobulin and calcitoninfibril formation reactions (data not shown). In addi-tion, there was no detectable binding of the controlproteins GST and ovalbumin to any of the samples ofall eight proteins tested (data not shown).
Clusterin affects nucleation and is less effective atsuppressing fibril elongation
Increasing the concentration of amyloidogenic pro-teins will generally favor the self-association of mono-mers into oligomers able to nucleate the aggregationprocess. Therefore, to test whether clusterin affectsnucleation, one clusterin:substrate ratio was selected ineach case and the concentration of the substrate pro-tein was varied. As expected, as the concentrations ofSH3 and A� were increased, the level of thioflavin Treactive species increased and the preaggregation lagtime shortened (Fig. 5A). Clusterin inhibited the devel-opment of thioflavin T reactive species in all cases; asthe concentrations of SH3 and A� were increased,however, this effect became progressively smaller. Forsamples containing 0.25 mM SH3, a 1:50 M ratio ofclusterin:SH3 suppressed the thioflavin T fluorescencemeasured on the final sample by �90%. However,when SH3 was at a concentration of 1 mM, the sameratio of clusterin:SH3 decreased thioflavin T fluores-cence by only �35% (Fig. 5B). Similarly, a 1:50 M ratioof clusterin:A� suppressed the final thioflavin T fluo-rescence by �90% for samples containing 15 �M A�,but only by �45% when A� was at 60 �M (Fig. 5B). Asanother means of examining the point at which clus-terin exerts its effects on amyloid formation, sampleswere taken at different times during fibril formationand used to seed amyloid formation in subsequentexperiments in the presence or absence of clusterin.Under the conditions used in this study, in the absenceof preformed amyloid species, SH3 and A� exhibitedlag phases of �6 h and �2 h, respectively (Fig. 6A).Note that the kinetics of SH3 aggregation in thisreaction were faster than in most of the subsequentSH3 reactions reported here (Fig. 6B), as the latter
Figure 3. Clusterin reduces the formation of sedimentableaggregates of SH3 and A� by forming stable complexes withfibrillogenic intermediates. A) Images of Coomassie blue-stained SDS-PAGE gel (SH3) and immunoblot (A�) showingsupernatant (S) and pellet (P) fractions prepared by centrif-ugation of final samples taken from fibril formation reactionscontaining various ratios of clusterin:substrate (indicatedabove the corresponding lanes). On the A� panel, HMWindicates SDS-resistant aggregates migrating at an apparentmolecular mass of �180 kDa; 6 kDa indicates the position ofa 6 kDa marker. B) Immuno-dot blot detection of clusterinassociated with supernatant and pellet fractions preparedfrom final samples taken from SH3 and A� fibril formationreactions containing various ratios of clusterin:substrate (in-dicated). C) Composite image showing proteins that copre-cipitate with immunoadsorbed clusterin from initial and finalsamples of SH3 and A� fibril formation reactions (clusterin:SH3�1:10, clusterin: A��1:10). Immunoprecipitations werealso performed for final samples of control reactions lackingclusterin. SH3 was detected by Coomassie staining SDS-PAGEgels; A� was detected by immunoblotting. In all cases theresults shown are representative of two or more individualexperiments.
2317CLUSTERIN INTERACTS WITH PREFIBRILLAR STRUCTURES
were performed using a lower substrate concentration(see Fig. 6 legend). In both reactions used to generatepreformed amyloid species, the elongation phase wasrapid, after which the level of thioflavin T reactivematerial remained relatively constant. When samplestaken early in these time courses or from the transition
region between lag and growth phases were used toseed subsequent aggregation reactions, clusterin wasable to suppress the generation of thioflavin T reactivespecies by up to �75–90% (Fig. 6B, C). However, whenthe samples used to seed the reaction were taken laterin the initial aggregation time courses, the lag phases of
Figure 4. Binding of clusterin to intermediatestructures on the fibril-forming pathway. Mainpanels show the thioflavin T fluorescence ofspecies present in fibril formation reactions as afunction of time. Panel insets show the resultsof immuno-dot blot assays measuring the bind-ing of clusterin to protein species present atdifferent times during fibril formation. In theseexperiments, the aggregation reactions con-tained 1 mM SH3 or 10 �M A�. The resultsshown are representative of two or more indi-vidual experiments.
Figure 5. Effect of substrate protein concentra-tion on the relative ability of clusterin to inhibitfibril formation. A) Time-dependent changes inthe thioflavin T fluorescence of species present infibril formation reactions containing differentconcentrations of substrate protein and (wherepresent) a constant clusterin:substrate ratio. Solu-tions containing SH3 without (empty symbols) orwith (filled symbols) clusterin (at clusterin:SH3�1:100) were incubated as described in Ma-terials and Methods at various concentrations ofSH3 (1.0 mM, circles; 0.75 mM, triangles; 0.50mM, squares; 0.25 mM, diamonds). Similarly,solutions of A� without (empty symbols) or with(filled symbols) clusterin (at clusterin:A��1:50)were incubated at various concentrations of A�(15 �M, triangles; 30 �M, diamonds; 45 �M,squares; 60 �M, circles). The data shown here areaverages of triplicate measurements and are rep-resentative of two individual experiments. B)Thioflavin T fluorescence of final samples fromfibril formation reactions containing clusterin,expressed as a percentage of the respective valuesfor reactions lacking clusterin. In each case, thedata shown represent means of triplicate determi-nations and the error bars are se.
2318 Vol. 21 August 2007 YERBURY ET AL.The FASEB Journal
the subsequent aggregation reactions were shortened(Fig. 6B) and the inhibitory effects of clusterin weresignificantly less (25–50% inhibition, Fig. 6C).
DISCUSSION
Previous work has indicated that clusterin can inhibitthe in vitro formation of amyloid aggregates by theprion protein, A�, and apolipoprotein C-II (5, 7, 9). Wehave shown here that clusterin can potently inhibit invitro amyloid formation by a broad range of unrelatedproteins, suggesting that this activity is not stronglydependent on the identity of the specific polypeptidesubstrate or is associated with disease (for a summary ofresults, see Table 2 and Table 3). In all cases tested,significantly substoichiometric ratios of clusterin:sub-strate (from 1:10 to 1:3.5) completely inhibited theformation of fibrils detectable by fluorescence andTEM (Fig. 1B). For three of the eight substrate proteinstested (calcitonin, �-synuclein, and A�), when clusterinwas present at very low levels relative to the substrate itsignificantly increased the level of thioflavin T reactivematerial formed (Fig. 1A). One possible explanation isthat at least for calcitonin, �-synuclein, and A�, whenpresent at very low levels clusterin may facilitate amy-loid formation by stabilizing the otherwise unstableprotein aggregates required to initiate fibril formation.
Given the demonstration that, under differing con-ditions, clusterin exerts variable effects on amyloidfibril formation, we explored whether this is also truefor its effects on amyloid toxicity. Both SH3 and A� areknown to generate cytotoxic intermediates during amy-loidogenesis (20, 21). For both these proteins, at leastfor some of the time points tested, clusterin enhancedthe cytotoxicity of aggregates formed when presentduring fibril formation at a clusterin:substrate ratio of1:500, but had the opposite effect when present at themuch higher but still substoichiometric ratio of 1:10(Fig. 2A). These results clearly indicate that the effectsof clusterin on the cytotoxicity of aggregates are com-plex and depend on both the stage of amyloid forma-tion at which the aggregates are formed and on theclusterin:substrate ratio. Regions of exposed hydropho-bicity have been strongly implicated in the cytotoxicity
Figure 6. Effects of clusterin on seeded fibril growth. A)Panels showing time-dependent changes in the thioflavin Tfluorescence of protein species present in fibril formationreactions in which the concentrations of SH3 and A� were500 �M and 10 �M, respectively. Aliquots taken at differenttimes from these reactions were used to seed similar reactions(see panel B). B) Thioflavin T fluorescence as a function oftime of protein species in fibril formation reactions that, at
zero time, were seeded with preformed aggregates taken atthe times indicated from the aggregation mixtures repre-sented in panel A. These reactions contained 250 �M SH3 or10 �M A� and were conducted in the absence (emptysymbols) or presence of clusterin (filled symbols; clusterin:substrate�1:100 and 1:50, respectively). Unseeded aggrega-tion reactions had time courses similar to those of 1 h seededreactions (data not shown). C) Thioflavin T fluorescence offinal samples taken from fibril formation reactions containingclusterin, expressed as a percentage of the respective valuesfor reactions lacking clusterin (calculated from results such asthose shown in panel B). In each case, the data shownrepresent means of duplicate determinations and the errorbars are ranges. The data shown are representative of at leasttwo independent experiments.
2319CLUSTERIN INTERACTS WITH PREFIBRILLAR STRUCTURES
of protein aggregates (1). The enhancement of toxicityseen at very low clusterin:substrate ratios may resultfrom there being sufficient clusterin present underthese conditions to physically stabilize in solution ag-gregates bearing multiple hydrophobic surfaces, butinsufficient clusterin to bind to and inhibit the cytotox-icity of all hydrophobic surfaces. If this is correct, thenit would be expected that at higher clusterin:substrateratios clusterin would be more effective at maskingexposed hydrophobicity on protein aggregates andreducing their cytotoxicity. This was found to be thecase when clusterin was present during amyloid forma-tion and when it was added to aggregates of SH3 andA� preformed in its absence (see clusterin:sub-strate�1:10 data in Fig. 2A, B). Earlier studies havereported that clusterin enhances (10) or suppresses (7)the cytotoxicity of A�; but the conditions used in eachof these studies differed greatly, including the type ofpeptide (A�1–40 vs. A�1–42), whether or not clusterinwas preincubated with A�, and the effective ratios ofclusterin:A� tested. The results of the present studyshow that very different effects can be observed underdifferent experimental conditions, suggesting that thefindings of both earlier studies may be broadly correct.
When increasing amounts of clusterin were added tofibril formation reactions containing either SH3 or A�,a progressive increase in the proportion of substrateprotein present in the nonsedimenting supernatant (S)fraction at the end of the time course was observed(Fig. 3A). At a “high” clusterin:substrate ratio (1:10),the substrate remained in the S fraction, but at a muchlower ratio (1:500) it was incorporated into the sedi-menting pellet (P) fraction (Fig. 3B). We found thatclusterin did not detectably bind to washed, preformedfibrils formed from any of the eight substrate proteinstested (data not shown). These results suggest that
under amyloidogenic conditions, as is the case underconditions when amorphous aggregates are formed(4), clusterin can enhance substrate protein solubilityeven when present at significantly substoichiometricratios. In the case of proteins forming amorphousaggregates, clusterin appears to exert this effect bypromoting the formation of soluble HMW complexeswith substrate molecules present in non-native confor-mations (4, 22). The presence of clusterin and sub-strate molecules in supernatant fractions preparedfrom amyloid-forming reactions in which clusterin hadinhibited fibril formation (Fig. 3B) suggests that similarclusterin-substrate complexes might be forming underthese conditions. This was confirmed for both SH3 andA� substrates by immunoprecipitation analyses (Fig. 3C).
Clusterin inhibits the amorphous aggregation of avariety of proteins in vitro when present at molarclusterin:substrate ratios of 1:5.5–1:0.65 (4); this actionappears to be a consequence of clusterin binding toregions of exposed hydrophobicity on substrate mole-cules (23). In contrast, data presented here and else-where (5, 8) show that clusterin can influence amyloidformation at much lower clusterin:substrate ratios (aslow as 1:500). The apparent difference in stoichiometrybetween the effects of clusterin on protein aggregation-producing amorphous vs. amyloid-type aggregateslikely relates to the low concentration of protein oli-gomers that nucleate fibril formation thought to bepresent in amyloid-forming reactions (24). The verylow clusterin:substrate ratios that affect the amyloid-forming pathway are consistent with clusterin interact-ing with species present in low abundance such as thosethat nucleate growth or their precursors. Immuno-dotblot analyses showed that for most proteins tested,clusterin bound most strongly to transient proteinspecies, which are presumably more abundant duringfibril nucleation and/or growth (Fig. 4). It is not clearwhy binding to corresponding species present in calci-tonin and �2m fibril formation reactions was not de-tected, although one possibility is that the interactingspecies in these cases are at very low levels. In the caseof SH3, maximum binding was traced to speciespresent at 4–6 h, corresponding to the initial stage atwhich maximum thioflavin T fluorescence was ob-tained. Electron microscopic examination of these sam-ples indicated that they consisted of a mixture ofspherical and short fibrillar aggregates (data notshown). Thus, in this case also it is feasible thatclusterin was binding to transient intermediate species.In SH3 and A�, clusterin suppressed amyloid formation
TABLE 2. Summary of the effects of clusterin on amyloid formation by eight different proteins
Effect lys calc �-cas �-syn �2m SH3 A� cc�w
Dose-dependent decrease in thioT fluorescence Yes Yes Yes Yes Yes Yes Yes YesIncreased thio T fluorescence at low clusterin:substratea No Yes No Yes No No Yes NoInhibition of fibril formation at high clusterin: substrateb Yes Yes Yes Yes Yes Yes Yes YesClusterin binding to transient intermediates detected Yes No Yes Yes No Yes Yes Yes
aLow clusterin:substrate ratio varies between substrates but is in the range of (1:50–1:500). bHigh clusterin:substrate ratios are � 1:10.
TABLE 3. Summary of additional results for A� and SH3relating to the effects of clusterin on the toxicity associated withamyloid formation and the mechanism of clusterin action
Effect SH3 A�
Cytoprotection (at high clusterin:substrate) Latea YesCytotoxicity (at low clusterin:substrate) Yes Mida
Clusterin retained in soluble fraction Yes YesInhibition of seeding by early aggregates Yes YesMore efficiently inhibits fibril formation at
low substrate concentrations Yes Yes
aRefers to the stage of the time course: late for SH3 is 72 h andmid for A� is 4 h.
2320 Vol. 21 August 2007 YERBURY ET AL.The FASEB Journal
on a molar basis more efficiently at lower substrateconcentrations (when self-association of substrate mol-ecules with non-native conformations into oligomers isless favored; Fig. 5). Similarly, clusterin appears moreefficient at inhibiting fibril formation reactions that areseeded with species taken from early time points ofpreceding reactions than with those seeded with speciestaken from later time points (Fig. 6). These resultssuggest that clusterin exerts its effects on amyloidformation primarily by interacting with transient pro-tein species that are most abundant prior to fibrilelongation; these interacting species are probably smalloligomers that are either structural precursors to, orfully functional, aggregation nuclei.
Collectively, our results indicate that at substoichio-metric levels, clusterin exerts substantial effects on invitro amyloid formation by a broad variety of substrateproteins, and suggest that these effects result frominteractions with prefibrillar species of very low abun-dance. At least in vitro, clusterin does not bind detect-ably to either native substrate proteins or matureamyloid fibrils. The relatively low level of substratespecificity characterizing this action suggests that clus-terin interacts with species sharing common structuralfeatures present on the amyloid-forming pathways ofmany different proteins. We have shown that clusterinforms stable complexes with fibril-forming proteins andreduces their propensity to sediment from solution.Our results and those from a recent study using humanlysozyme as a substrate protein (36) suggest that clus-terin does not bind to native or non-native monomers,but instead support the hypothesis that clusterin inter-acts with oligomeric species that may function as nucle-ation points for fibril formation. Our results clearlyshow that an important determinant of the nature andextent of the effects of clusterin on both amyloidformation and toxicity is the clusterin:substrate ratio.This conclusion has important potential implicationsfor the likely effects of clusterin on amyloid formationin vivo, as clusterin may exert differing effects on theextracellular folding landscape depending on its rela-tive abundance compared to amyloid-forming substrateproteins in certain biological contexts. For example,although clusterin is present in human plasma at �100�g/ml (25), in cerebrospinal fluid it is present atonly �2 �g/ml (26). It is therefore feasible that in thecentral nervous system, clusterin might under someconditions enhance amyloid formation and toxicity(11) and yet elsewhere in the body, where it is moreabundant (e.g., in plasma), exert the opposite effect.The evidence that clusterin does not bind to matureamyloid fibrils, but at very low clusterin:substrate ratiosis incorporated into insoluble amyloid material, alsoprovides a tenable explanation for the otherwise curi-ous observation that clusterin is associated with a varietyof disease-related amyloid deposits in vivo (Table 1). Itappears likely that the association of clusterin withthese deposits represents its failed attempts to maintainthe solubility of amyloid-forming species under disease-specific conditions of high molar substrate excess.
In conclusion, the abundant extracellular chaperoneclusterin exerts potent effects on the formation ofamyloid in vitro by a wide range of proteins, includingexamples known to be associated with disease andothers that are not. The nature and extent of theseeffects show a limited dependence on the identity ofthe substrate protein, but (in contrast) vary with theclusterin:substrate ratio. Clusterin appears to interactwith prefibrillar species that share common structuralfeatures and, depending on the prevailing conditions,can either promote or suppress amyloid formation andtoxicity. However, in all cases tested, at molar ratios ofclusterin:substrate of 1:10 or greater, clusterin potentlyinhibited amyloid formation and provided substantialcytoprotection. These findings suggest that clusterinmay be an important part of an armory of mechanismsthat defend against the consequences of extracellularprotein misfolding. They also raise the possibility that,at least in some circumstances, increasing the levels ofclusterin in vivo could be a therapeutic tool in the fightagainst extracellular protein deposition disorders.
The authors thank the University of Wollongong and theInstitute of Biomolecular Science for grant support. M.R.W. issupported by an Australian Research Council DiscoveryProject grant (DPO773555). J.J.Y is grateful for an AustralianPostgraduate Research Award. Research by C.M.D was sup-ported in part by programme grants from the Wellcome Trustand the Leverhulme Trust.
REFERENCES
1. Chiti, F., and Dobson, C. M. (2006) Protein misfolding, func-tional amyloid, and human disease. Annu. Rev. Biochem. 75,333–366
2. Yerbury, J. J., Stewart, E. M., Wyatt, A. R., and Wilson, M. R.(2005) Quality control of protein folding in extracellular space.EMBO Rep. 6, 1131–1136
3. Sherman, M. Y., and Goldberg, A. L. (2001) Cellular defencesagainst unfolded proteins: a cell biologist thinks about neuro-degenerative diseases. Neuron 29, 15–32
4. Humphreys, D. T., Carver, J. A., Easterbrook-Smith, S. B., andWilson, M. R. (1999) Clusterin has chaperone-like activitysimilar to that of small heat shock proteins. J. Biol. Chem. 274,6875–6881
5. Hatters, D. M., Wilson, M. R., Easterbrook-Smith, S. B., andHowlett, G. J. (2002) Suppression of apolipoprotein C-II amy-loid formation by the extracellular chaperone, clusterin. Eur.J. Biochem. 269, 2789–2794
6. Matsubara, E., Soto, C., Governale, S., Frangione, B., and Ghiso,J. (1996) Apolipoprotein J and Alzheimer’s amyloid beta solu-bility. Biochem. J. 316, 671–679
7. Oda, T., Wals, P., Osterburg, H. H., Johnson, S. A., Pasinetti,G. M., Morgan, T. E., Rozovsky, I., Stine, W. B., Snyder, S. W.,and Holzman, T. F. (1995) Clusterin (ApoJ) alters the aggrega-tion of amyloid beta-peptide (A�1–42) and forms slowly sedi-menting A� complexes that cause oxidative stress. Exp. Neurol.136, 22–31
8. Hughes, S. R., Khorkova, O., Goyal, S., Knaeblein, J., Heroux, J.,Riedel, N. G., and Sahasrabudhe, S. (1998) Alpha-2-macroglob-ulin associates with beta-amyloid peptide and prevents fibrilformation. Proc. Natl. Acad. Sci. U. S. A. 95, 3275–3280
9. McHattie, S., and Edington, N. (1999) Clusterin prevents aggre-gation of neuropeptide 106–126 in vitro. Biochem. Biophys. Res.Commun. 259, 336–340
10. Boggs, L. N., Fuson, K. S., Baez, M., Churgay, L., McClure, D.,Becker, G., and May, P. C. (1996) Clusterin (ApoJ) protects
2321CLUSTERIN INTERACTS WITH PREFIBRILLAR STRUCTURES
against in vitro amyloid-beta (1–40) neurotoxicity. J. Neurochem.67, 1324–1327
11. DeMattos, R. B., O’Dell, M., A., Parsadanian, M., Taylor, J. W.,Harmony, J. A., Bales, K. R., Paul, S. M., Aronow, B. J., andHoltzman, D. M. (2002) Clusterin promotes amyloid plaqueformation and is critical for neuritic toxicity in a mouse modelof Alzheimer’s disease. Proc. Natl. Acad. Sci. U. S. A. 99, 10843–10848
12. DeMattos, R. B., Cirrito, J. R., Parsadanian, M., May, P. C.,O’Dell, M. A., Taylor, J. W., Harmony, J. A., Aronow, B. J., Bales,K. R., Paul, S. M., and Holtzman, D. M. (2004) ApoE andclusterin cooperatively suppress A� levels and deposition: evi-dence that ApoE regulates extracellular A� metabolism in vivo.Neuron 41, 193–202
13. Wilson, M. R., and Easterbrook-Smith, S. B. (1992) Clusterinbinds by a multivalent mechanism to the Fc and Fab regions ofIgG. Biochim. Biophys. Acta 1159, 319–326
14. Souza, J. M., Giasson, B. I., Lee, V. M., and Ischiropoulos, H.(2000) Chaperone-like activity of synucleins. FEBS Lett. 474,116–119
15. Heuer, K. H., Mackay, J. P., Podzebenko, P., Bains, N. P., Weiss,A. S., King, G. F., and Easterbrook-Smith, S. B. (1996) Develop-ment of a sensitive peptide-based immunoassay: application todetection of the Jun and Fos oncoproteins. Biochemistry 35,9069–9075
16. Kammerer, R. A., Kostrewa, D., Zurdo, J., Detken, A., Garcia-Echeverria, C., Green, J. D., Muller, S. A., Meier, B. H., Winkler,F. K., Dobson, C. M., and Steinmetz, M. O. (2004) Exploringamyloid formation by a de novo design. Proc. Natl. Acad. Sci.U. S. A. 101, 4435–4440
17. Lichtenfels, R., Biddison, W. E., Schulz, H., Vogt, A. B., andMartin, R. (1994) CARE-LASS (calcein-release-assay), an im-proved fluorescence-based test system to measure cytotoxic Tlymphocyte activity. J. Immunol. Methods 172, 227–239
18. Lakins, J. N., Poon, S., Easterbrook-Smith, S. B., Carver, J. A.,Tenniswood, M. P., and Wilson, M. R. (2002) Evidence thatclusterin has discrete chaperone and ligand binding sites.Biochemistry 41, 282–291
19. Klug, G. M., Losic, D., Subasinghe, S. S., Aguilar, M. I., Martin,L. L., and Small, D. H. (2003) Beta-amyloid protein oligomersinduced by metal ions and acid pH are distinct from thosegenerated by slow spontaneous ageing at neutral pH. Eur.J. Biochem. 270, 4282–4293
20. Bucciantini, M., Giannoni, E., Chiti, F., Baroni, F., Formigli, L.,Zurdo, J., Taddei, N., Ramponi, G., Dobson, C. M., and Stefani,M. (2002) Inherent toxicity of aggregates implies a commonmechanism for protein misfolding diseases. Nature 416, 507–511
21. Dahlgren, K. N., Manelli, A. M., Stine, W. B., Jr., Baker, L. K.,Krafft, G. A., and LaDu, M. J. (2002) Oligomeric and fibrillarspecies of amyloid-beta peptides differentially affect neuronalviability. J. Biol. Chem. 277, 32046–32053
22. Poon, S., Easterbrook-Smith, S. B., Rybchyn, M. S., Carver, J. A.,and Wilson, M. R. (2000) Clusterin is an ATP-independentchaperone with very broad substrate specificity that stabilizesstressed proteins in a folding-competent state. Biochemistry 39,15953–15960
23. Poon, S., Rybchyn, M. S., Easterbrook-Smith, S. B., Carver, J. A.,Pankhurst, G. J., and Wilson, M. R. (2002) Mildly acidic pHactivates the extracellular molecular chaperone clusterin. J. Biol.Chem. 277, 39532–39540
24. Harper, J. D., and Lansbury, P. T., Jr. (1997) Models of amyloidseeding in Alzheimer’s disease and scrapie: mechanistic truthsand physiological consequences of the time-dependent solubil-ity of amyloid proteins. Annu. Rev. Biochem. 66, 385–407
25. Morrissey, C., Lakins, J., Moquin, A., Hussain, M., and Tennis-wood, M. (2001) An antigen capture assay for the measurementof serum clusterin concentrations. J. Biochem. Biophys. Methods48, 13–21
26. Calero, M., Rostagno, A., Matsubara, E., Zlokovic, B., Frangione,B., and Ghiso, J. (2000) Apolipoprotein J (clusterin) andAlzheimer’s disease. Microsc. Res. Tech. 50, 305–315
27. Freixes, M., Puig, B., Rodriguez, A., Torrejon-Escribano, B.,Blanco, R., and Ferrer, I. (2004) Clusterin solubility and aggre-gation in Creutzfeldt-Jakob disease. Acta Neuropath. 108, 295–301
28. Chiesa, R., Angeretti, N., Lucca, E., Salmona, M., Tagliavini, F.,Bugiani, O., and Forloni, G. (1996) Clusterin (SGP-2) inductionin rat astroglial cells exposed to prion protein fragment 106–126. Eur. J. Neurosci. 8, 589–597
29. Nishida, K., Quantock, A. J., Dota, A., Choi-Miura, N. H., andKinoshita, S. (1999) Apolipoproteins J and E co-localise withamyloid in gelatinous drop-like and lattice type I cornealdystrophies. Br. J. Ophthalmol. 83, 1178–1182
30. Sakaguchi, H., Miyagi, M., Shadrach, K. G., Rayborn, M. E.,Crabb, J. W., and Hollyfield, J. G. (2002) Clusterin is present indrusen in age-related macular degeneration. Exp. Eye Res. 74,547–549
31. Zenkel, M., Kruse, F. E., Junemann, A. G., Naumann, G. O., andSchlotzer-Schrehardt, U. (2006) Clusterin deficiency in eyeswith pseudoexfoliation syndrome may be implicated in theaggregation and deposition of pseudoexfoliative material. In-vest. Ophthalmol. Vis. Sci. 47, 1982–1990
32. Kida, E., Choi-Miura, N. H., and Wisniewski, K. E. (1995)Deposition of apolipoproteins E and J in senile plaques istopographically determined in both Alzheimer’s disease andDown’s syndrome brain. Brain Res. 685, 211–216
33. Maat-Schieman, M. L., van Duinen, S. G., Bornebroek, M.,Haan, J., and Roos, R. A. (1996) Hereditary cerebral hemor-rhage with amyloidosis-Dutch type (HCHWA-D): II—A review ofhistopathological aspects. Brain Pathol. 6, 115–120
34. Ghiso, J., Plant, G. T., Revesz, T., Wisniewski, T., and Frangione,B. (1995) Familial cerebral amyloid angiopathy (British type)with nonneuritic amyloid plaque formation may be due to anovel amyloid protein. J. Neurol. Sci. 129, 74–75
35. Witte, D. P., Aronow, B. J., Stauderman, M. L., Stuart, W. D.,Clay, M. A., Gruppo, R. A., Jenkins, S. H., and Harmony, J. A.(1993) Platelet activation releases megakaryocyte-synthesizedapolipoprotein J, a highly abundant protein in atheromatouslesions. Am. J. Pathol. 143, 763–773
36. Kumita, J. R., Poon, S., Caddy, G. L., Hagan, C. L., Dumoulin,M., Yerbury, J. J., Stewart, E. M., Robinson, C. V., Wilson, M. R.,and Dobson, C. M. (2007) The extracellular chaperone clus-terin potently inhibits human lysozyme amyloid formation byinteracting with prefibrillar species. J. Mol. Biol., in press (ac-cepted 26 Feb 2007).
Received for publication December 19, 2006.Accepted for publication March 1, 2007.
2322 Vol. 21 August 2007 YERBURY ET AL.The FASEB Journal
