Raft lipids as common components of human extracellular amyloid fibrils

1 downloads 0 Views 614KB Size Report
May 3, 2005 - from various human diseases (AA, ATTR, A 2M, AL , and AL amyloidosis) shows that these are associated with a common lipid component that ...
Raft lipids as common components of human extracellular amyloid fibrils Gerald P. Gellermann*, Thomas R. Appel*, Astrid Tannert*, Anja Radestock*, Peter Hortschansky†, Volker Schroeckh†, Christian Leisner*, Tim Lu¨tkepohl*, Shmuel Shtrasburg‡, Christoph Ro¨cken§, Mordechai Pras‡, Reinhold P. Linke¶, Stephan Diekmann*, and Marcus Fa¨ndrich*储 *Institut fu¨r Molekulare Biotechnologie, Beutenbergstrasse 11, 07745 Jena, Germany; †Leibniz-Inititut fu¨r Naturstoff-Forschung und Infektionsbiologie, Hans-Kno¨ll-Institut, Beutenbergstrasse 11, 07745 Jena, Germany; ‡Heller Institute of Medical Research, Tel Hashomer 52621, Israel; §Otto von Guericke University, Universita¨tsplatz 2, 39106 Magdeburg, Germany; and ¶Max Planck Institute of Biochemistry, Am Klopferspitz 18a, 82152 Martinsried, Germany Edited by Kai Simons, Max Planck Institute of Molecular Cell Biology and Genetics, Dresden, Germany, and approved January 20, 2005 (received for review September 23, 2004)

protein folding 兩 prion 兩 lipid rafts

T

he formation of amyloid fibrils represents a fundamental biochemical process that is also effective in the course of human aging, leading to specific diseases termed ‘‘amyloidoses’’ (1–3). Amyloid fibrils differ from ordinary protein fibrils by a conserved structural motif that encompasses ␤-sheets within a cross-␤ arrangement (4, 5). At present, ⬇30 nonhomologous polypeptide sequences are known to form such fibrils inside the body (3), but the basic ability to adopt this type of structure is common to many other, if not all, polypeptide sequences (1, 4, 6–7). Such observations promoted the view that amyloid fibrils arise primarily from an intrinsic property of the chiral polypeptide main chain that is often suppressed in nature by unfavorable physicochemical conditions, side-chain arrangements, or evolutionary adaptations (1, 6–7). Inside the human body, amyloid structures can occur as extracellular or intracellular deposits (3). In the case of the systemic amyloidoses that are characterized by massive deposits of fibrils, amyloidotic pathogenicity results in severe distortions of the affected tissues and compromises normal organ functions (2). Purified amyloid fibrils of the AA type may demonstrate the activity of an amyloid enhancing factor (8), resembling the case of prions (9). Moreover, specific amyloid structures, including protofibrils, are cytotoxic or detrimental to ordered cellular functions (10–11). The channel hypothesis relates this bioactivity to a perforation of biological lipid membranes through pore structures consisting of amyloid species (10). Indeed, amyloid structures can have an annular organization (10) or the ability to interact with lipids (12–13) or to disrupt lipid bilayers in vitro (14). Analysis of a few tissue-extracted amyloid fibril samples provided evidence that these polypeptide aggregates can be associated with lipids (15–17). Based on these single observations, we analyzed whether this evidence is of general nature and www.pnas.org兾cgi兾doi兾10.1073兾pnas.0407035102

determined the chemical composition of the lipid components from disease-associated amyloid deposits. Materials and Methods Tissue Staining and Fibril Extraction. Cryosections of murine AA amyloid-laden spleen (6 ␮m) were fixed in acetone, stained with hematoxylin, and incubated for 10 min in a filtered solution of 3% NaCl, 80% ethanol, and 0.01% NaOH. The same solution containing 0.5% Congo red (CR) was applied afterward for 10 min and washed with 80% and 100% ethanol. For Sudan III staining, sections were fixed in 4% paraformaldehyde in PBS, rinsed with water, and stained for 1 h with a filtered solution of 0.2% Sudan III in 70% ethanol. To extract amyloid fibrils from human tissues, we used the water method according to Pras et al. (18). To check for the lipid interactions of fibrils formed in vitro, we used the following purification microprotocol (19). Tissue aliquots were homogenized in 10 mM Tris䡠HCl兾138 mM NaCl兾2 mM CaCl2兾10 mM EDTA, pH 7.4. The homogenate was spun down for 30 min at 15,000 ⫻ g at 4°C. After discarding the supernatant, the pellet was resuspended and homogenized in water. Debris was spun down, and the fibril containing supernatant was retained. The pellet was subjected to 10 more extraction cycles. All supernatants were pooled and adjusted to 0.2 M NaCl and 10 mM EDTA. After vigorous mixing, fibrils were precipitated by centrifugation at 20,000 ⫻ g at 4°C for 2 h. The pellet was resuspended in water and lyophilized. Source of Polypeptides and in Vitro Fibril Formation. The myoglobin

fragment MYO(101–118) and serum amyloid A (SAA)1.1(2–21) were obtained from Jerini (Berlin). Transthyretin was a gift from M. Krebs and C. Robinson, (University of Cambridge, Cambridge, U.K.) The coding region of full-length murine SAA1.1 was cloned in fusion to the gene of the maltose-binding protein in the pMAL-c2X vector (NEB, Beverly, MA) and separated by a cleavage site for tobacco etch virus protease. The fusion protein was purified with nickel chelate and reverse-phase chromatography. The following conditions were used to form fibrils in vitro: A␤(1–40): 10 mg兾ml in 50 mM sodium borate, pH 9.0, at 20°C; MYO(101–118): 5 mg兾ml in 50 mM sodium acetate, pH 5.0, at 65°C; full-length murine SAA1.1: 5 mg兾ml in 50 mM sodium phosphate, pH 3.0, at 37°C; human SAA1.1(2–21): 10 mg兾ml in 50 mM sodium phosphate, pH 1.0, at 20°C; and transthyretin: 10 mg兾ml in 50 mM sodium acetate, pH 4.4, at This paper was submitted directly (Track II) to the PNAS office. Abbreviations: CR, Congo red; hpTLC, high-performance TLC; SAA, serum amyloid A; CH, cholesterol; CE, CH ester; PC, phosphatidylcholine; SM, sphingomyelin; MCD, methyl-␤cyclodextrin. 储To

whom correspondence should be addressed. E-mail: [email protected].

© 2005 by The National Academy of Sciences of the USA

PNAS 兩 May 3, 2005 兩 vol. 102 兩 no. 18 兩 6297– 6302

BIOCHEMISTRY

Amyloid fibrils are fibrillar polypeptide aggregates from several degenerative human conditions, including Alzheimer’s and Creutzfeldt–Jakob diseases. Analysis of amyloid fibrils derived from various human diseases (AA, ATTR, A␤2M, AL␭, and AL␬ amyloidosis) shows that these are associated with a common lipid component that has a conserved chemical composition and that is specifically rich in cholesterol and sphingolipids, the major components of cellular lipid rafts. This pattern is not notably affected by the purification procedure, and no tight lipid interactions can be detected when preformed fibrils are mixed with lipids. By contrast, the early and prefibrillar aggregates formed in an AA amyloidproducing cell system interact with the raft marker ganglioside-1, and amyloid formation is impaired by addition of cholesterolreducing agents. These data suggest the existence of common cellular mechanisms in the generation of different types of clinical amyloid deposits.

37°C. The presence of amyloid structure was confirmed with CR and negative-stain electron microscopy (20).

quantitative standard. Samples were applied in 1,2-dichlorpropan兾1-propanol (2:1 vol兾vol).

Lipid Extraction and High-Performance TLC (hpTLC). Lipids were extracted quantitatively from amyloid fibrils by chloroform兾 methanol at 2:1 (vol兾vol) extraction. The anhydrous mass of tissue-extracted amyloid samples was determined gravimetrically. Lipid extracts from normal porcine or bovine tissues from freshly slaughtered animals were obtained by homogenization in chloroform兾methanol at 1:1 (vol兾vol). Twenty microliters of the homogenate was mixed with 9 ml of 2 M KCl and spun down at 3,000 ⫻ g for 10 min. The lower phase was transferred into a tarred glass vial. The upper phase was reextracted twice. All lower phases were combined, lyophilized, and used immediately or stored at ⫺80°C under argon. HpTLC was performed by using silica gel 60 plates on glass (Macherey & Nagel, catalog no. 811043) soaked in 25:25:25:10:9 (vol兾vol) methylacetate兾1-propanol兾chloroform兾methanol兾 0.25% KCl) and activated by heating at 110°C (21). The first development was carried out in the same solvent and stopped as soon as the solvent front was 4.5 cm out of the preconcentration zone. Plates were dried thoroughly, developed on full length in n-hexane兾diethyl ether兾acetic acid 75:23:2 (vol兾vol), and dried. The third development was carried out in n-hexane. Dry plates were stained by dipping into 10% copper sulfate兾8% phosphoric acid. Excessive solvent was dried off. Quantifications were carried out within 1 h after charring (170°C) by using a Camag II TLC scanner operated at 428 nm. The total lipid content was obtained by adding up all lipids from a sample. Lipid standards were prepared by dissolving commercially available lipids (Sigma). Seventy-five to 500 ng per lipid class was used as a

Cell Cultivation, Staining, and Analysis. The monocytic fraction was isolated from human blood by centrifugation on a Histopaque centrifuge (PAA, Linz, Austria). Cells were incubated on glass slides in a 24-well plate in DMEM (BE12-604F, Cambrex, East Rutherford, NJ) containing 15% FCS (at 37°C in an atmosphere of 10% CO2). The culture medium was supplemented with full-length SAA1.1 protein (final concentration of 130 ␮g兾ml) or, if applicable, 1 or 4 ␮M lovastatin or 1.1 mM methyl-␤cyclodextrin (MCD; Sigma). Addition of the latter does not notably affect cell morphology or differentiation by light microscopy. Cell viability was examined with lactate dehydrogenase and 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide kits (Roche Diagnostics, Mannheim, Germany) according to the manufacturer’s instructions. For CR analysis, cells were fixed in ice-cold methanol for 10 min, stained for 45 min with CR prepared in alkaline 80% ethanol (22), and examined with a Leica DMRE polarizing microscope. For fluorescence staining, the culture medium was replaced with icecold Dulbecco’s PBS (Sigma) containing 1% BSA. The cells were incubated under gently rocking in the presence of cholera toxin subunit B (23) coupled with Alexa Fluor 647 (1 ␮g兾ml, Molecular Probes) for 30 min on ice, washed thoroughly with PBS, and fixed in 4% paraformaldehyde兾PBS. Transferrin receptor was detected with monoclonal murine antibody (20 ␮g兾ml, Sigma) and murine SAA1.1 with affinity-purified R48 antibody (1:50, polyclonal rabbit) applied in PBS and 1% BSA for 75 min at room temperature. Excessive primary antibodies were washed off thoroughly with PBS before the addition of secondary antibodies conjugated with FITC

Fig. 1. Tissue-deposited amyloid fibrils are associated with lipids. (A–C) Cryosections of AA-laden spleen tissue after CR (A, brightfield; B, darkfield) or Sudan III (C) staining. (Scale bar, 200 ␮m.) (D) HpTLC plate of lipid extracts from AA, AL, ATTR amyloid (lanes 2–9), normal tissue (NT; liver lane 10, brain lane 11), and a lipid standard (lane 1). Load: ⬇4 ␮g lipid per lane. (E) Relative lipid abundance in the eight AA cases, and the average. (F) Relative lipid abundance in different amyloid type patterns (AA, AL␭, AL␬, and ATTR), a single A␤2M sample, and NT average (liver, heart, and brain). To guide the eye, the main membrane lipids (Left) and the rare membrane lipids (Right) are shown in separation. Lipid abbreviations: CA, cardiolipin; FA, mono fatty acid (oleyl acid); GA, gangliosides; GC, galactosylceramides; LPC, lyso-PC; MG, monoacylglyceride; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; PS, phosphatidylserine; SQ, squalen; SU, sulfatides; TG, triacylglyceride. 6298 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0407035102

Gellermann et al.

Table 1. Tissue-extracted amyloid fibril samples used in this study Amyloid classification AA AA AA AA AA AA AA AA ATTR ATTR ATTR ATTR A␤2M AL␭ AL␭ AL␭ AL␭ AL␭ AL␭ AL␭ AL␭ AL␭ AL␭ AL␭ AL␬ AL␬ AL␬

Tissue of origin

Clinical manifestation

Lipid content of the anhydrous mass, %

Liver Liver Spleen Spleen Spleen Spleen Kidney Thyroid Heart Spleen Heart Heart Skin Spleen Liver Spleen Spleen Spleen Spleen Spleen Spleen Spleen Brain Muscle Breast Liver Spleen

Systemic, inflammatory Hypernephroma Systemic, inflammatory FMF FMF SJRA FMF FMF FAP FAP FAP Senile, systemic Subcutaneous node Systemic Systemic Factor X deficiency Systemic Systemic Systemic Systemic Systemic Systemic Local amyloidoma Systemic Systemic Systemic No clinical data available

11 ⫾ 1.65 9 ⫾ 1.35 8 ⫾ 1.2 6 ⫾ 0.9 3 ⫾ 0.45 1 ⫾ 0.15 1 ⫾ 0.15 1 ⫾ 0.15 10 ⫾ 1.5 6 ⫾ 0.9 4 ⫾ 0.6 2 ⫾ 0.3 7 ⫾ 1.05 12 ⫾ 1.8 10 ⫾ 1.5 10 ⫾ 1.5 9 ⫾ 1.35 9 ⫾ 1.35 3 ⫾ 0.45 3 ⫾ 0.45 2 ⫾ 0.3 2 ⫾ 0.3 1 ⫾ 0.15 3 ⫾ 0.45 12 ⫾ 1.8 11 ⫾ 1.65 5 ⫾ 0.75

FMF, familial mediterranean fever; FAP, familial amyloid polyneuropathy; SJRA, systemic juvenile rheumatic arthritis.

ImmunoGold Labeling. Five microliters of a solution of AA-amyloid

was placed onto a Formvar- and carbon-coated grid and incubated for 10 min, followed by blocking with 1% BSA solution in PBS (20 min). Primary antibody (affinity purified R48) was applied in PBS and 1% BSA for 1 h. After being washed three times with pure water, the secondary antibody (anti-rabbit IgG conjugated with 10-nm gold particles, Sigma) was applied (1 h), and the grid was stained with 2% uranyl acetate. Results Tissue Deposits of Amyloid Fibrils Contain Lipids. Serial sections cut

through AA amyloidotic tissue and stained alternately with amyloid-specific (CR) and lipophilic (Sudan III) dye demonstrate lipids in native amyloid deposits (Fig. 1 A–C). The chemical composition of these lipids was established by means of a specifically adapted hpTLC method that enables separation of most natural lipid classes in a single experiment (Fig. 1D). The resulting spots, corresponding to individual lipid classes, can be assigned and quantified by comparison with a lipid standard (Fig. 1D, lane 1). By using this setup, we have analyzed 27 samples from 5 different types of human, extracellular amyloidosis (Table 1). One set of samples (AA) arises from a largely ␣-helical lipoprotein, the SAA protein. The other samples are derived Gellermann et al.

from polypeptides that fold natively into ␤-sheet structures, such as transthyretin (ATTR amyloidosis), ␤2-microglobulin (A␤2M), and the Ig light chains ␭ and ␬ (AL␭ and AL␬). The latter proteins are not known to interact strongly with lipids. HpTLC shows that all amyloid samples contain lipids corresponding to 1–12% lipids of the total anhydrous mass (Table 1). The Conserved Lipid Pattern of Disease-Associated Amyloid Fibrils.

Comparison of different types of human amyloidosis reveal a conserved lipid pattern that consists mainly of hydrophobic lipids, such as cholesterol (CH), and only small amounts of the more polar lipids, such as phosphatidylcholine (PC) and phosphatidylethanolamine (Fig. 1D; for lipid abbreviations, see the Fig. 1 legend). Different samples obtained from the same type of amyloidosis can differ with respect to the relative abundance of the individual lipid classes (Fig. 1E). However, this heterogeneity is reduced substantially when comparing the lipid ‘‘type patterns’’ in which all of the cases from the same type of amyloidosis have been averaged (Fig. 1E). The main membrane lipids present in the fibrils are CH (15–24%), sphingomyelin (SM; 7–15%) and glycosphingolipids, such as galactocerebrosides, sulfatides, and gangliosides. The latter migrate at the border of the preconcentration zone and do not enter the quantitative area of the hpTLC plate. Nevertheless, glycosphingolipids can also be evidenced by the ability of tissue-extracted amyloid fibrils to bind cholera toxin subunit B, a ligand of ganglioside GM-1 (data not shown). In addition to these main membrane lipids, amyloid fibrils contain also large quantities of those lipids that occur only rarely in cellular membranes, such as CH esters (CEs; 7–20%) and free fatty acids (14–30%). This pattern of lipids is effectively independent from the sequence of the underling polypeptide chain and type of amyloidosis (AA, ATTR, and AL␭ and AL␬) (Fig. 1F). Only the single sample obtained from a PNAS 兩 May 3, 2005 兩 vol. 102 兩 no. 18 兩 6299

BIOCHEMISTRY

(Dianova, Hamburg, Germany) or Alexa Fluor 546 (Molecular Probes). Cells were examined in a Zeiss LSM510 confocal laserscanning microscope equipped with a ⫻63 objective lens. Dyes were excited by using a 488-nm argon laser, a 543-nm HeNe laser, and a 633-nm HeNe laser. The emission band path was set to 505–530 nm when using the green channel, to 560–615 nm when using the bright red channel, and to and 650 nm when using the deep-red channel.

The Lipid Composition of Amyloid Fibrils Is Not Related to Purification.

The lipid composition of the amyloid fibril samples does not correlate obviously with the total lipid content, representing the stringency of the purification procedure (Fig. 2 A and B). Moreover, when fibrils are formed in vitro from pure polypeptide chains [ATTR, A␤(1–40), SAA(2–21), the smallest SAA1.1 fragment from human AA deposits (25), and MYO(101–118), which is only known to form such fibrils in vitro (21) (Fig. 2 D–G)], and mixed with homogenized tissues (Fig. 2C), repurified fibrils contain only 0.01–0.3% lipids (Fig. 2H). This range is much lower than what we obtained with tissue-extracted fibrils (Table 1) and is effectively indistinguishable from control samples where tissue material was subjected to the extraction procedures in the absence of fibrils. Furthermore, there is no significant enrichment of certain lipid classes. Less stringent purification conditions increase the lipid content without significant effects on the lipid composition of amyloid fibrils. We conclude that purification does not affect the lipid pattern substantially and that preformed fibrils do not bind lipids with strong affinity. Interaction Between Prefibrillar SAA Aggregates and GM-1 in Amyloidogenic Monocytes. SM, CH, and glycosphingolipids are main

Fig. 2. Purification does not lead to significant lipid association. PC兾CH ratio (A) and SM plus CH fraction (B) within all membrane lipids (ML) of tissueextracted amyloid fibrils and normal tissue. Values of normal tissues are indicated by arrows (light gray, brain; dark gray, liver; black, heart). (C) Strategy of purification (see main text for details). (D–G) Electron micrographs of fibrils formed in vitro from SAA(2–21) (D), transthyretin (E), MYO(101–118) (F), and A␤(1– 40) (G). (H) HpTLC plate showing lipid extracts from normal tissues (lanes 1–3) and from 3 mg of fibrils after repurification from 25 mg of normal tissue incubated for 1.5 h. We applied 25% of the total lipid extract in lanes 4 –10. Lanes 1–3 contain ⬇4 ␮g of lipid. For lipid abbreviations see Fig. 1.

subcutaneous node of a patient with A␤2M amyloidosis corresponds less well to this pattern and shows only small amounts of SM and CE and a very high triacylglyceride content. The lipid components revealed here are not incorporated into the internal structure of the fibrils, because electron microscopy and CR binding confirm that the fibrils are not disrupted by lipid extraction (data not shown). Furthermore, they differ significantly from the lipid patterns of normal tissues (Fig. 1F) that are characterized by large quantities of phospholipids, leading to high PC兾CH ratios and smaller relative amounts of SM and CH within the membrane lipid fraction (Fig. 2 A and B). 6300 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0407035102

components of specific membrane microdomains termed lipid rafts (24), thus raising questions as to whether amyloid fibrils (or their precursors) might interact with raft lipids in vivo. To address this question, we have established a modified version of the monocytic cell culture model of the AA amyloidosis (25–27) where primary human monocytes convert externally added full-length murine SAA1.1 into amyloid. Within 12 days of incubation, large quantities of amyloid plaques can be detected in the culture dish (Fig. 3 A and B). These plaques show typical CR green birefringence and can be stained by R48 antibody, which positively recognizes by ImmunoGold-labeling fibrils formed from murine SAA1.1 in vitro (Fig. 3C). HpTLC analysis of lipid extracts from cell culture-derived amyloid fibrils reveals the same lipid pattern that we described above for amyloid fibrils from diseased human tissues (Fig. 3D). These data show that the cell culture model used here faithfully reconstitutes the lipid interactions occurring during amyloidogenesis inside the human body. The monocytes were then examined by using confocal fluorescence microscopy. Interestingly, after 3 days of incubation, a punctuate SAA staining emerges that colocalizes significantly with the raft marker GM-1 (Fig. 3E). In contrast to this result, there is only little colocalization with a typical non-raft marker, such as transferrin receptor (Fig. 3E), and no such interactions are seen when preformed mature fibrils or maltose-binding protein are added to the cells (data not shown). Maltose-binding protein is a globular protein that does not aggregate readily under the conditions used in this study, and the resulting staining with anti-maltose-binding protein antibody shows a diffuse and unspecific staining. Amyloid Formation Is Impaired by CH-Inhibiting or -Depleting Agents.

Our data suggest that lipids such as CH might play a potential role in the formation of amyloid structures in vivo. To test this idea, we added lovastatin to the culture medium of amyloidogenic macrophages. Lovastatin is an inhibitor of the HMG-CoA reductase that reduces the plasma CH level of humans (28), as well as the CH content of cultured cells, by 10–20% (29, 30). hpTLC analysis shows that the addition of 1 or 4 ␮M lovastatin reduces, under the conditions of the present study, the cellular CH level by 15–20%. In addition, filipin staining shows also a reduction of the cellular CH (data not shown). After 8 days, lovastatin-free control reactions show plaques, whereas no plaques are seen on day 8 in the lovastatin-treated cultures. After 12 days, all incubations were terminated to determine the Gellermann et al.

Fig. 4. Amyloid formation is impaired by CH-reducing agents. (A) The number of plaques per slide was quantified from reactions where 1.5 ⫻ 106 cells were inoculated per slide and incubated with full-length SAA1.1 for 12 days in the absence (control) or presence of 1.1 mM MCD or lovastatin (concentrations are as indicated). The number of plaques was averaged form a minimum of four individual measurements. Significance was established with an one-side unpaired t test (*, ␣ ⫽ 0.05). (B) Lactate dehydrogenase assay of cells incubated for 48 h without (column A) or with 1 ␮M (column B) or 4 ␮M (column C) lovastatin. (C) A 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay of cells incubated for 12 days with 4 ␮M lovastatin (column B) or 1.1 mM MCD (column C) and control (column A). Error bars represent the SE of the mean.

Fig. 3. Colocalization of SAA and GM-1 in amyloidogenic monocytes. Amyloid plaques in the AA macrophage model after 12 days using brightfield (A) and darkfield (B) microscopy. (Scale bar, 20 ␮m.) (C) Electron micrograph of fibrils formed from murine SAA1.1 in vitro-stained with R48 and a secondary antibody coupled with 10-nm gold particles. (D) HpTLC plate of lipid standard (lane 1) and lipid extracts from monocytes after amyloid formation (lane 2), cell culture-extracted amyloid (lane 3), and disease-associated amyloid (lane 4). (E) Confocal sections showing two monocytes in the plane of the plasma membrane on day 3. Distribution of SAA, GM-1, and transferrin receptor (TFR) is color-coded and combined as shown in each image. (Scale bar, 5 ␮m.)

number of plaques per cover slide. The amyloid load was graded for each plaque separately, and similar to a method developed by Westermark and coworkers (8) for murine tissues, we assigned scores from ⫹1, representing a small plaque, to ⫹3, representing a very large plaque. This analysis shows that the presence of 1 or 4 ␮M lovastatin leads to a marked reduction in the total number of plaques and their size (Fig. 4A). The addition of low levels of mevalonate (0.3 ␮M) to lovastatin-treated cells has only a small, if any, effect on this result, although the addition of mevalonate may cause a slight increase in the number of Gellermann et al.

Discussion By using an improved hpTLC setup for the analysis of lipid extracts from biological samples, we were able to show that raft lipids represent a common component of disease-associated, extracellular amyloid fibrils. In addition, amyloidogenic monocytes interact with SAA aggregates through raft-like microdomains of their plasma membrane, and amyloid formation can be attenuated with CHdepleting agents. These data reconcile the single previous observations of lipids in tissue-extracted amyloids (15–17) with the presence of a common pattern of lipids. Altogether, raft lipids have been demonstrated in the following types of amyloidosis: AA, ATTR, AL␭, AL␬, and previously in the prion protein-dependent amyloidosis APr (16). In addition, amyloid deposits are enriched in lyso-PC, free fatty acids, and CEs, i.e., lipids that occur rarely in biological membranes and accumulate characteristically in the course of tissue degradation or necrosis (31, 32). Although such degrading processes have only small effects on the main membrane lipids, such as SM and CH (32), we believe, based on the CE and lyso-PC content, that they represent an additional factor affecting the clinical deposition of amyloid. The general lipid pattern of amyloid fibrils described here is not obviously related to factors, such as purification procedure, clinical syndrome, case history, organ of deposition, amino acid sequence, structure, function, or lipid interaction of the native protein. However, tissue- or casePNAS 兩 May 3, 2005 兩 vol. 102 兩 no. 18 兩 6301

BIOCHEMISTRY

plaques. Finally, we added 1.1 mM MCD to the culture medium. MCD is known to deplete CH from the exofacial leaflet of cellular membranes. Similar to lovastatin, it leads also to a reduction in the number of plaques and their intensity (Fig. 4A). The concentrations of MCD and lovastatin used here do not reduce the viability of these cells significantly when using 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide and lactate dehydrogenase assays (Fig. 4 B and C). Furthermore, they do not inhibit aggregation of full-length SAA1.1 in vitro (data not shown).

specific factors will inevitably modify this pattern, thus giving rise to the variability seen in samples derived from different patients. These findings could have two major implications. First, they demonstrate a substance class to represent a conserved secondary component of extracellular amyloid fibrils. So far, these secondary components include mainly proteins, such as serum amyloid P component, collagen, or apolipoprotein E, and glycosaminoglycanes. It is interesting to note that some of these substances, namely glycosaminoglycanes and serum amyloid P component, have been suggested to also modulate the pathogenicity of the underlying amyloid structures, and are currently tested as possible target structures in clinical therapy (2, 33, 34). Second, the present data suggest the possibility that at least some extracellular amyloid fibril deposits form by a common cellular mechanism involving raft lipids. It remains to be established, however, whether these lipids might act actively on amyloid formation or whether they define the cellular context in which amyloid formation can readily occur. However, several lines of evidence argue, in these specific cases, in favor of the latter idea. For example, the efficiency of amyloid formation by the monocyte system is increased when SAA is associated with lipids (27). Raft lipids have been reported to be involved in the cellular conversion of the prion protein into its pathogenic isoform (35) or the proteolytic generation of Alzheimer’s A␤ peptide (30). The CH and ceramide metabolism is abnormally altered in aging brains and Alzheimer’s patients (36), and chronic administration of CH-reducing agents (statins)

diminishes the prevalence of Alzheimer’s disease (37). CH-rich or raft-like lipid environments have been shown in vitro to promote the aggregation and fibril formation from A␤ (12) as well as peptide fragments or the prion protein and human SAA, SAA(1–18) (13, **). Nevertheless, it is also clear that there is no general biophysical requirement of polypeptides to bind to lipids to aggregate (1, 6–7) and present data derived from several, and normally noncerebral, amyloidoses suggests that these lipids must play a much more general role in the biology of amyloidotic processes than was noted previously for the specific cases quoted above. Improving our understanding of these generalized processes has the potential, therefore, to provide a novel basis for the development of drug candidates that are effective on more than one type of amyloidosis.

1. Dobson, C. M. (2001) Philos. Trans. R. Soc. London B 356, 133–145. 2. Pepys, M. B. (2001) Philos. Trans. R. Soc. London B 356, 203–210. 3. Westermark, P., Benson, M. D., Buxbaum, J. N., Cohen, A. S., Frangione, B., Ikeda, S., Masters, C. L., Merlini, G., Saraiva, M. J. & Sipe, J. D. (2002) Amyloid 9, 197–200. 4. Glenner, G. G., Eanes, E. D., Bladen, H. A., Linke, R. P. & Termine, J. D. (1974) J. Histochem. Cytochem. 22, 1141–1158. 5. Sunde, M., Serpell, L. C., Bartlam, M., Fraser, P. E., Pepys, M. B. & Blake, C. C. (1997) J. Mol. Biol. 273, 729–739. 6. Chiti, F., Webster, P., Taddei, N., Clark, A., Stefani, M., Ramponi, G. & Dobson, C. M. (1999) Proc. Natl. Acad. Sci. USA 96, 3560–3594. 7. Fa¨ndrich, M., Fletcher, M. A. & Dobson, C. M. (2001) Nature 410, 165–166. 8. Lundmark, K., Westermark, G. T., Nystrom, S., Murphy, C. L., Solomon, A. & Westermark, P. (2002) Proc. Natl. Acad. Sci. USA 99, 6979–6984. 9. Prusiner, S. B. (1998) Proc. Natl. Acad. Sci. USA 95, 13363–13383. 10. Caughey, B. & Lansbury, P., T., Jr. (2004) Annu. Rev. Neurosci. 26, 267–298. 11. Reixach, N., Deechongkit, S., Jiang, X., Kelly, J. W. & Buxbaum, J. N. (2004) Proc. Natl. Acad. Sci. USA 101, 2817–2822. 12. Kakio, A., Nishimoto, S., Kozutsumi, Y. & Matsuzaki, K. (2003) Biochem. Biophys. Res. Commun. 303, 514–518. 13. Sanghera, N. & Pinheiro, T. J. (2002) J. Mol. Biol. 315, 1241–1256. 14. Arispe, N., Pollard, H. B. & Rojas, E. (1996) Proc. Natl. Acad. Sci. USA 93, 1710–1715. 15. Kim, I. C., Shirahama, T. & Cohen, A. S. (1967) Am. J. Pathol. 50, 869–886. 16. Klein, T. R., Kirsch, D., Kaufmann, R. & Riesner, D. (1998) Biol. Chem. 379, 655–666. 17. Damas, A., Sebastiao, M. P., Domingues, F. S., Costa, P. P. & Saraiva, M. J. (1995) Amyloid: Int. J. Exp. Clin. Invest. 2, 173–178. 18. Pras, M., Schubert, M., Zucker-Franklin, D., Rimon, A. & Franklin, E. C. (1968) J. Clin. Invest. 47, 924–933. 19. Tennent, G. A. (1999) Methods Enzymol. 309, 26–47.

20. Fa¨ndrich, M., Forge, V., Buder, K., Kittler, M., Dobson, C. M. & Diekmann S. (2003) Proc. Natl. Acad. Sci. USA 100, 15463–15468. 21. Yao, J. K. & Rastetter, G. M. (1985) Anal. Biochem. 150, 111–116. 22. Puchtler, H., Sweat, F. & Levine, M. (1962) J. Histochem. Cytochem. 10, 355–364. 23. Harder, T., Scheiffele, P., Verkade, P. & Simons, K. (1998) J. Cell Biol. 141, 929–942. 24. Simons. K. & Ehehalt, R. (2002) J. Clin. Invest. 110, 597–603. 25. Ro ¨cken, C. & Shaeksepeare, A. (2002) Virchows Arch. 440, 111–122. 26. Kluve-Beckerman, B., Liepnieks, J. J., Wang, L. & Benson, M. D. (1999) Am. J. Pathol. 155, 123–133. 27. Elimova, E., Kisilevsky, R., Szarek, W. A. & Ancsin, J. B. (2004) FASEB J. 14, 1749–1751. 28. Eghdamian, B. &. Ghose, K. (1998) Drugs Today 11, 943–956. 29. Cignarella, A., Brennhausen, B., von Eckardstein, A., Assmann, G. & Cullen, P. (1998) Arterioscler. Thromb. Vasc. Biol. 8, 1322–1329. 30. Simons M., Keller, P., De Strooper, B., Beyreuther, K., Dotti, C. G. & Simons, K. (1998) Proc. Natl. Acad. Sci. USA 95, 6460–6464. 31. Wender, M., Adamczewska-Goncerzewicz, Z., Dorszewska, J., Szczech, J., Godlewski, A., Pankrac, J. & Talkowska, D. (1991) Neuropatol. Polska. 29, 87–94. 32. Takano, T. (1992) Acta. Pathol. Jpn. 42, 625–631. 33. Gervais, F., Chalifour, R., Garceau, D., Kong, X., Laurin, J., McLaughlin, R., Morissette, C. & Paquette, J. (2001) Amyloid 8, 28–35. 34. Kisilevsky, R. (2000) J. Struct. Biol. 130, 99–108. 35. Baron, G. S., Wehrly, K., Dorward, D.W., Chesebro, B. & Caughey, B. (2002) EMBO J. 21, 1031–1040. 36. Cutler, R. G., Kelly, J., Storie, K., Pedersen, W. A., Tammara, A., Hatanpaa, K., Troncoso, J. C. & Mattson, M. P. (2004) Proc. Natl. Acad. Sci. USA 101, 2070–2075. 37. Zamrini, E., McGwin, G. & Roseman, J. M. (2004) Neuroepidemiology 23, 94–98.

6302 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0407035102

**Liang, J. S., Kirschner, D. A., Szumowski, K. E., Salmona, M., Cathcart, E. S. & Sipe, J. D. (1997) Protein Sci. 6, Suppl. 2, p. 63 (abstr.).

We thank S. Fricke, P. Hemmerich, U. C. Hipler, R. Joswig, J. Lindermayer, R. Oos, D. Schwertner, S. Weidtkamp-Peters, and the Elektronmikroskopisches Zentrum, Jena, for technical assistance; M. Krebs and C. Robinson for the gift of full-length transthyretin; and J. Sipe and K.-J. Halbhuber for helpful discussions. This work was supported in part by the Exist-HighTEP Programme and a BioFuture Grant (both from the Bundesministerium fu ¨r Bildung und Forschung). R.P.L. is supported by Deutsche Forschungsgemeinschaft Grant Li 247兾12-13.

Gellermann et al.