Kidney International, Vol. 52 (1997), pp. 1543—1549

Ultrastructural organization of hemodialysis-associated132-microglobulin amyloid fibrilsSADAYUKL INOUE, MIE KUROIWA, KENIcHI OHASHI, MITSURU HA1, and ROBERT KISILEYSKY

Department of Anatomy and Cell Biology, McGill University, Montrea4 Quebec, Canada; Department of Pathology, Tokyo Medical andDental University School of Medicine, and Department of Pathology, Toranomon Hospital, Tokyo, Japan; and Department of Pathology,Queen's University and the Syl and Molly Apps Research Centre, Kingston General Hospital, Kingston, Ontario, Canada

Ultrastructural organization of hemodialysis-associated fi2-micro-globulin amyloid fibrils. Fibrils of hemodialysis-associated 32-microglobu-un amyloid were examined by high resolution electron microscopy andimmunohistochemical labeling. The amyloid containing tissues obtainedthrough autopsy were prepared for thin section observations. In contrastto other forms of amyloid, the most conspicuous feature of these fibrilswere their curved conformations. The fibril core showed ultrastructuraland immunohistochemical features in common with the core of connectivetissue microfibrils and of previously observed fibrils of experimentalmurine AA amyloidosis and familial amyloid polyneuropathy (FAP). Thecore was wrapped in a layer of 3 nm wide ribbon-like "double tracked"structures identified as chondroitin sulfate proteoglycan (CSPG) withimmunogold labeling as well as from the results of previous in Vitroexperiments. Finally, the outer surface of the fibril was associated with aloose assembly of I nm wide filaments immunohistochemically identifiedas 2-microglobulin. This is similar to the manner in which AA protein andtransthyretin filaments are associated with their respective fibrils. Theresults of this study provide an additional example for the concept thatamyloid fibrils in general are microfibril-like structures externally associ-ated with amyloid protein filaments. An unusual feature of the fibrils ofhemodialysis-associated amyloid, however, is the presence of a peripherallayer composed of CSPG rather than of heparan sulfate proteoglycan(HSPG) as in the case of the other two amyloids above. These chondroitinsulfate chains in the Outer CSPG layer may be less effective in providingrigidity to the fibril core, thus allowing for the curved conformations ofJ32-microglobulin amyloid fibrils.

Amyloidosis following on chronic hemodialysis is a relativelynew disease, having been first described about 12 to 15 years ago[1—41. The failure to see this disorder prior to these dates rests onthe necessity for long-term survival in the presence of renalfailure, a condition that was not met until hemodialysis becamecommon and allowed such patients to live for periods in excess of10 to 15 years. Over this period of time the majority of renalpatients develop a distinctive amyloid arthropathy that may affectmost major joints such as the spine, hips, knees, shoulders andperiarticular tissues [5]. Involvement of the gastrointestinal tractalso occurs [61, indicating that this process is systemic and not

Key words: amyloid fibrils, hemodialysis, f32-microglobulin, fibrils, familialaniyloid polyneuropathy, AA amyloidosis.

Received for publication April 21, 1997and in revised form July 18, 1997Accepted for publication July 21, 1997

© 1997 by the International Society of Nephrology

restricted to the bones and joints. The specific protein associatedwith these amyloid deposits is 32-microglobulin [1—41.

f32-microglobulin is a component of the major histocompatihil-ity loci of cells and a normal serum protein of a small size (Mr11,731) [7). This protein is known to accumulate in the blood ofpatients undergoing long-term hemodialysis to levels S to 10 timesnormal because of the inability of this procedure to remove theprotein from the blood plasma [8, 91.

This amyloid is ultrastructurally composed of deposits of fibrilsas in other types of amyloid. However, the fibrils of this particulartype of amyloid are unique in that they show an unusually atypical"curvilinear" conformation in contrast to straight fibrils in mostother amyloids [2, 10, 111. Fibrils of 132-microglobulin can formspontaneously in vitro in the supernatant of cultures of mononu-clear cells from dialysis patients apparently from intact precursorprotein molecules rather than from the product of proteolytictransformation [12, 131. These fibrils formed in vitro were approx-imately 8 to 10 nm in thickness again with curved appearances.

In spite of these electron microscopic observations, the detailedultrastructural organization of the fibrils of hemodialysis-associ-ated amyloid is not clear. Our recent detailed high resolutionobservations have shown that the fibrils of both experimentalmurine AA amyloid [14) and FAP [Inoue et al, manuscript inpreparation) resembled microfibril-like structures, and the pro-teins specific to these amyloids, AA protein and transthyretin,respectively, were immunohistochemically identified as 1 to 2 nmwide flexible filaments associated with the surface of the fibril.

Microfibrils are unbranching rod-like structures approximately10 nm in width. They are one of the major components of theconnective tissue either in association with elastic fibers [51 orfree in the tissue space such as, for example, those composing asuspensory ligament (zonular fibers) in the eye [161. Amyloidfibrils in situ are also unbranching rods and resemble connectivetissue microfibrils in their size and appearance [17). Recent resultsfrcm our laboratories have shown that the 9 nm wide fibril core isidentical in microfibrils, and amyloid fibrils so far examined, andit is composed of peripheral CSPG with a center of amyloid Pcomponent (AP) [14, 18; lnoue et al, manuscript in preparation].The difference between these two distinct types of fibrils relates totheir surfaces, which in the case of the microfibril is associatedwith periodically distributed fibrillin and accumulations of HSPG[191, while in the case of amyloid fibril the core is associated witha layer of HSPG accompanied on its surface by I nm wide

1543

1544 lnoue et at: Stnwture of 2-microg1obu1in amyloid fibrils

filaments composed of an amyloid protein [14; Inoue et al,manuscript in preparation].

In this study the ultrastructure of the fibrils of hemodialysis-associated amyloid was examined with high resolution electronmicroscopy in combination with immunohistochemical labeling todetermine whether they are also microfibril-like in nature resem-bling the previous two described forms of amyloid.

METHODS

Ultrastructure

Hemodialysis-associated /32-microglobulin amyloid tissue wasobtained through autopsy from a 73-year-old male patient withchronic glomerulonephritis. Pieces of the tissue were fixed with2.5% glutaraldehyde in 0.1 M phosphate buffer, pH 7.2, for twohours at room temperature. They were washed with the samebuffer and further cut into blocks approximately 1 mm3. Theblocks were postfixed with 1% osmium tetroxide in the samebuffer for one hour at 4°C and briefly rinsed with distilled water,followed by dehydration and embedding in Epon. Thin sectionswere stained with uranyl acetate followed by lead citrate beforeexamination in a Philips EM400 or JEM-2000FX electron micro-scope. Magnification was calibrated with a grating replica and acatalase crystal standard. Size was measured on prints with a handmagnifier equipped with a 0.1 mm scale.

ImmunolabelingThin sections of glutaraldehyde- and osmium tetroxide-fixed,

Epon-embedded tissue of 32-microg1obuIin amyloid prepared asabove were pre-treated with sodium metaperiodate (NaIO4)according to the method of Bendayan and Zollinger [20] and thenlabeled for f32-microglobulin or chondroitin sulfate with theimmunogold method similar to that used previously [14]. Thinsections on nickel grids were exposed to a saturated aqueoussolution of Na104 for one hour. They were washed with distilledwater and then floated on the surface of a drop of 1% bovineserum albumin (BSA) in phosphate buffered saline (PBS) fol-lowed by a drop of (i) rabbit anti-human 132-microglobulin anti-serum (dilution 1:100; DAKO Corporation, Carpinteria, CA,USA) or (ii) monoclonal anti-mouse chondroitin sulfate anti-serum (dilution 1:100; Sigma, St. Louis, MO, USA), for one hourat room temperature. In the case of (ii), sections were pretreatedwith 10% hydrogen peroxide immediately after the Na104 treat-ment. Control sections were exposed to non-immune rabbit serumor non-immune mouse serum. The grids were washed by passingthem sequentially across three wells filled with PBS and then a setof drops of PBS containing 1% BSA. They were then floated onthe surface of drops of 0.05 M Tris-HCI buffer, pH 7.6, andexposed to goat anti-rabbit or anti-mouse immunoglobulins cou-pled to 5 nm gold particles (dilution 1:15; Janssen Pharmaceutica,Antwerp, Belgium) for 30 minutes at room temperature. Thegrids were washed by first placing them on drops of Tris-HCIbuffer, then in a gentle stream of the same solution followed bydistilled water. Finally, the sections were counterstained withuranyl acetate and lead citrate for four minutes and 15 seconds,respectively, before examination in the electron microscope. Forassessment of the extent of the labeling the number of goldparticles per II m2 square of the sections that were associatedwith 1 nm wide filaments (after labeling for 132-microglobulin), orwith 3 nm wide ribbon-like "double tracks" (after labeling for

chondroitin sulfate) in both experimental and control sectionswere enumerated. The numbers averaged from five randomlyselected squares in each specimen was used for quantitativeevaluation of the labeling.

RESULTS

Ultrastructure of j32-microglobulin amyloid fibrilsAmyloid fibrils were mainly localized to the space between

collagen fibrils as aggregates of various sizes (Fig. 1A). The mostconspicuous feature of these fibrils was their intense curving witha twisted appearance, contrasting with the fibrils of AA [14] andFAP amyloid [Inoue et a!, manuscript in preparation], which wereeither straight or only gently curved. Therefore, in thin sectionedtissues mainly short and curved fibril segments of approximately12 nm in width were observed (Fig. 1A). These intensely curvedfibrils, unable to form bundles, accumulated as loose randomassemblies. At a higher magnification only segments of limitedlengths of variously curved fibrils were seen, and their contourswere irregular and indistinct (Fig. 1B, bold arrow).

The fibrils in which the plane of section passed at various levelsfrom the surface to the interior were examined at high magnifi-cation. Each successive level was identified by examining sectionscut slightly oblique to the axis of the fibril so that the transitionfrom one level to the next could be seen. The space immediatelyabove and close to the surface of curved short segments ofindividual fibrils was examined first (Fig. 2A). The surface wasassociated with flexible filaments approximately I nm in thickness.When fibrils were sectioned at a level of or slightly below thesurface (Fig. 2B) this peripheral part of the fibril was found to becomposed of a random assembly of ribbon-like "double tracked"structures. They were composed of two parallel dark lines sepa-rated by a lucent space. They were also associated on both sideswith variable amounts of darkly stained material. The width ofthese characteristic ribbons was 3.0 0.2 nm (N 50; Fig. 3A).At a level of deeper sectioning of the fibril, the contour of the coreof the fibril, 8 to 9 nm in width, begins to emerge (Fig. 2C). Acharacteristic feature of the surface of the core is the presence ofcross striations. The pattern of cross striations was resolved into aseries of double tracked structures (Fig. 2C, paired arrows). Thewidth of these double tracked structures is 3.0 0.2 nm (N = 20)(Fig. 3B). They were indistinguishable in detailed appearancefrom those composing the peripheral layer of the fibril. In Figure2C the short segment of the fibril is curved within the planeperpendicular to the surface of the section. Because of thecurvature the lower three double tracks are tilted slightly awayfrom the face-on view, showing their side view with darkly stainedmaterial associated at their sides (single medium arrows).

The core was also sectioned both transversely (Fig. 4B) andlongitudinally through its center (Fig. 4 A, C, D). In longitudinalsections the core, approximately 9 nm in diameter, is composed ofthree parallel lines in which the middle one is often seen as beingcomposed of a series of dark dots (Fig. 4A). The core is often bentsharply (Fig. 4C) or more gently (Fig. 4D). In cross section one ofthe dots composing the middle line is present at the center of apentagonal frame approximately 9 nm in width (Fig. 4B).

Immunolocalization of 2-microglobulin and chondroitinsulfate

After immunolabeling for J32-microglobulin 5 nm gold particleswere associated with 1 nm wide filaments at the surface of the

Inoue Ct al. Structure of /32-microglobulin amyloid fibrils 1545

Fig. 1. Deposit of hemodialysis-associated j2-microgIobulin amyloid. (A) A large accumulation of amyloid fibrils (AF) is present among collagen fibrils(Col). Examples of short segments of intensely curved fibrils are indicated by arrows. (B) Higher power view of a part of A. The same short segmentof a curved fibril indicated by an arrow at the upper center in A is shown at a bold arrow (A X86,700; B x295,800).

fibrils (Fig. 5A). At high magnification it was confirmed that theparticles were preferentially associated with I nm wide filaments,and none of the other structures in the fibrils were labeled (Fig.SB). The number, averaged from five different areas, of goldparticles associated with 1 nm wide filaments was 425.4/jim2 whilein control sections average number of particles associated with thefilaments was 4.4/jim2. After labeling for chondroitin sulfate goldparticles were preferentially associated with 3 nm wide doubletracked structures, which were composed either the surface layerof the fibril (Fig. 6 A, B, C) or the peripheral component of thefibril core (Fig. 6D). The number of gold particles associated with3 nm wide "double tracks" averaged from five different areas, was34.6/jim2 and 2.6/jim2 in experimental and control sections,respectively.

DISCUSSION

Until 25 to 30 years ago, because of its unique tinctorial,ultrastructural, and conformational features, amyloid was thoughtto be a single entity regardless of its tissue localization or clinicalcontext. The advent of techniques to isolate amyloid fibrils [21],quickly led to the realization that there were many forms ofamyloid. Each was composed of a unique protein that wascharacteristic of the antecedent clinical condition. At least 17different amyloid proteins have now been identified, and these

now serve as the basis for classifying amyloids [22], rather than theclinical setting that was used previously.

In many cases the pure amyloid protein can, under promptingfrom non-physiologic in vitro conditions, be urged to aggregateinto fibrils. This has led to the concept that fibrillogenesis requiresonly the amyloid protein. Nevertheless, additional common amy-bid components, such as AP and HSPG, have been identified insitu irrespective of the type of amyloid. In some cases thesecomponents have been clearly implicated in the process offibrillogenesis [23—28]. How these additional components areinvolved in amyloid structure is only now beginning to emerge [14,

25].The fact that there are significant differences between the

reported ultrastructure of isolated amyloid fibrils [29—32] andthose of fibrils in situ [33—36] also indicates that the ultrastructuralorganization and biochemical composition of amyloid fibrils arenot as simple as previously believed.

In a recent high resolution ultrastructural study [14] the fibrilsof experimental murine AA amyloid were shown to be microfibril-like structures. The cores of both AA amyboid fibrils and connec-tive tissue microfibrils [18] are composed of helically wound 3 nmwide ribbon-like CSPG "double tracks" that enclose AP at theircenter. The cores of both AA amyloid fibrils and microfibrils [19]are covered by a layer of HSPG. In the ease of AA amyloid a loose

r

:'.- 'r-' -

• 'c-•- --

--

.

• V.

-S

••

:? . y ;;::.. -- - --- .- - - •

1546 Inoue et al: Structure of /32-microglobulin amyloid fibrils

Fig. 2. Ultrastructural organization of the fibril of 2-microglobulin amyloid as seen in single fibrils. Approximate length of the segments of curvedfibrils which span the thickness of the section is indicated between opposed bold arrows. (A) Flexible filaments 1 nm in width (small arrows) areassociated with the surface of the fibril. (B) Peripheral layer of the fibril is composed of random assemblies of 3 nm wide double tracks (paired arrows).(C) The surface of the fibril core. Cross striations made up of 3 nm wide double tracks (paired arrows) are present. Some of them are slightly tiltedbecause of the curvature of the segment and show their side view (single arrows) (A X557,600; B X634,100; C X676,600).

network of fine filaments, immunohistochemically identified asAA protein, was localized on the surface of the HSPG [14]. Asimilar organization was shown for the fibrils of familial amyloi-dotic polyneuropathy where the outermost surface of the micro-fibril-like structure was associated with a loose network of finefilaments immunohistochemically identified as transthyretin [In-oue et al, manuscript in preparation]. Unpublished work [Inoueand Kisilevsky] has revealed a similar organization for in situfibrils of AL and A amyloids as well. Therefore, it is possible thatamyloid fibrils in general may be microfibril-like in structure andprotein specific to each type of amyloid present in the form of finefilaments associated at the surface of the fibrils. In this study fibrilsof an additional type of amyloid, hemodialysis-associated /32-microglobulin amyloid, were examined at high resolution.

The cores of microfibrils [18] as well as the fibrils of experimen-tal murine AA amyloid [14] and FAP [Inoue et al, manuscript inpreparation] were shown to be composed of helically arranged 3nm wide CSPG double tracks enclosing AP at the center. Thedetailed ultrastructural feature of the /32-microglobulin fibril coreof dialysis associated amyloid, (illustrated in Fig. 2C and Fig.4A-D), is identical to that seen in microfibrils and the fibrils ofabove two types of amyloid. Also, 3 nm wide double trackedstructures at the periphery of the core were immunolaheled forchondroitin sulfate, Together with the fact that they closelyresembled, in size and detailed appearance, the double trackedstructures reconstituted in vitro after incubation of a CSPGpreparation at body temperature [18], it is likely that these 3 nm

A30

25

20

15

>10

ci)

05

U-

0

15

10

5

0

Width, nm

Fig. 3. Width of ribbon.like double tracked structures. (A) Double trackscomposing the pcdpheral layer of the fibril. (B) Double tracks present atthe surface of the core of the flhril as cross bands.

1 2 3 4 5

Pick 4f: tiAt.

'• tf'1'At

-

WtA itt

Inoue et at: Structure of /32-microglobulin amyloid fibrils 1547

Fig. 4. The core of amyloid fibrils sectioned transversely or longitudi-nally. (A) The core longitudinally sectioned through the center is com-posed of three parallel lines (sets of three arrows) in which the middle oneis made up of a series of successive dots. The core is often curved eithersharply (C) or more gently (D). Transverse section of the core shows apentagonal frame with a lumen containing a central dot (B, circled) (A, Cand D X574,100; B x559,100).

wide double tracked structures are composed of CSPG. Insummary, it is therefore likely that the f32-microglobulin fibril coreis also made up of helical CSPG double tracks with enclosed AP,as schematically shown in Figure 7. In addition, the outermostsurface of the dialysis-associated amyloid fibril was found to beassociated with f32-microglobulin filaments. Thus, this study pro-vides an additional example of amyloid fibrils that are microfibril-like structures with a protein specific to the type of amyloidpresent on the surface as a loose network of fine filaments.

There are, however, two unusual features in the fibril of/32-microglobulin amyloid when compared with the fibrils ofexperimental AA amyloid and of FAP. First, morphologically,fibrils of f32-microglobulin amyloid are localized in the vicinity ofcollagen fibrils that may be related to the known strong affinity ofthis amyloid for collagen in Vitro [37, 38], and they are intenselycurved instead of being straight in accord with observations madein previous reports [2, 10, 111. Second, the layer that encloses theAP/CSPG fibril core is made up of random assemblies of 3 nmwide double tracks immunolabeled for chondroitin sulfate andultrastructurally indistinguishable from double tracked structurespreviously reconstituted in vitro from CSPG, and thus are likely tobe CSPG as discussed above. That is, the layer is made up of asecond CSPG layer instead of HSPG, which is the proteoglycanlayer seen in the other two types of amyloid. This latter finding isin keeping with light microscopic analyses of 32-microg1obulinamyloid where HSPG was shown to be present but as a minorcomponent relative to CSPG [39]. These two features may herelated to one another. Glycosaminoglycans (GAGs) containing

Fig. 5. linmunogold labeling of hemodialysis.associated amyloid for j2microglobulin. (A) Five nm gold particles are present over and betweenamyloid fibrils in association with 1 nm wide flexible filaments (arrows).(B) High power view of the association of gold particles with the filaments(thick arrows) (A ><361,800; B X581,600).

predominantly iduronic acid, such as heparan sulfate, are knownto have tighter binding properties as compared to GAGs, whichcontain more glucuronic acid such as CSPG [40]. This differencein GAG composition may be responsible for a more flexible, lessrigid, arrangement of fibril structure, and the more curved entitiesseen in dialysis type amyloid than in AA and FAP.

Common tinctorial and structural characteristics of amyloidsinclude positive Congo red staining, which show red-green hire-fringence under polarized light, the presence of cross structureof amyloid proteins as detected by X-ray diffraction, and identi-fication of approximately 10 nm wide fibrils at the electronmicroscopic level. In the present study with /32-microglobulinamyloid only the last feature was demonstrated. What elements ofour proposed structure is responsible for the other charactersticsis not clear at present. Since microfibrils themselves do not showthese staining and polarization charateristics, 132-microglobulin inthe form of the fine filaments at the surface of the fibril may be thecomponent that is related to these specific characteristics.

Preliminary observations in our laboratories (unpublished re-sults) showed that the 1 nm wide filaments of AA protein inexperimental murine AA amyloid fibrils observed either in situ at

CSPGCSPG

1548 Inoue et al: Structure of f32-microglobulin amyloid fibrils

Fig. 6. Immunogold labeling of hemodialysis-associated amyloid for chondroitin sulfate. (A-c) Gold particles are preferentially associated with 3 nmwide "double tracks" (paired arrows) composing a surface layer of the fibril. (D) Three nm wide double tracks (paired arrows) composing a shortsegment of the core of the fibril (bctween thick arrows) are attached by gold particles (A-D x649,400).

Fig. 7. Schematic drawing of ultrastructuralorganization of f32-microglobulin amyloid fibril.AP, subunits of amyloid P component; CSPG,chondroitin sulfate proteoglycan in the form of3 nm wide ribbon-like "double tracks"; /32-microglobulin, I nm wide flexible filamentscontaining 32-microglobulin. See text for adetailed explanation.

the surface of the fibril or detached from it, had a tendency to coilthemselves to form 3 nm wide tight helices. In favorable areas ofAA amyloid tissues these helical rods were oriented parallel to thedirection of the fibril, and also to one another, with a regularcenter-to-center distance of approximately 5 nm at the surface ofthe fibril. If the arrangement of /32-microglobulin filaments issimilar to that of AA protein filaments, grooves 1.5 to 2 nm inwidth between parallel helical rods at the surface of the fibril maybe a favorable site for attachment and orientation of molecules ofCongo red. Such an arrangement would be adequate for theproduction of optical polarization. However, the cross-/3 confor-mation of the amyloid protein is difficult to explain from theresults of this study.

ACKNOWLEDGMENTS

This work was supported by grants from the Medical Research Councilof Canada, MT-3153 (to RK), MA-12279 (to SI), Neurochem Inc., andGrant-in-Aid from the Japanese Ministry of Health and Welfare (to MH).We thank Mrs. M. Cheung, Mrs. J. Mui, and Mrs. R. Tan for theirtechnical assistance. KO was a visiting research scientist (RK's lab)supported by the Japanese Ministry of Education, Science and Culture.MK is a visiting research scientist (SI's lab) from 2nd Department of OralAnatomy, School of Dentistry, Showa University, and Department ofClinical Pathology, Tokyo Metropolitan Institute of Gerontology, Tokyo,Japan.

Reprint requests to Dr. S. Inoue, Department of Anatomy and CellBiology,McGill University, 3640 University Street, Montreal, Quebec, Canada H3A2B2.

REFERENCES

1. GEJYO F, YAMADA T, ODANI S, NAKAGAwA Y, ARAKAWA M, KuNI-TOMO T, KATAOKA H, SUZUKI M, HIRASAWA Y, SHIRAHAMA T, COHENAS, SCHMID K: A new form of amyloid protein associated with chronichemodialysis was identified as /32-microglobulin. Biochem Biophys ResCommun 129:701—706, 1985

2. GOREVIC PD, CASEY 'VT, STONE WJ, DIRAIMONDO CR, PRELLI FC,FRANGIONE B: Beta-2-microglobulin is an amyloidogenic protein inman. J Clin Invest 76:2425—2429, 1985

3. SHIRAHAMA T, SKINNER M, COhEN AS, GEJYO F, ARAKAWA M,SuzuKi M, HIRASAWA Y: Histochemical and immunohistochemicalcharacterization of amyloid associated with chronic hemodialysis asbeta-2-microglobulin. Lab Invest 53:705—709, 1985

4. CASEY 11, STONE WJ, DIRAIMONDO CR, BRANTLEY BD, DIRAI-MONDO CHV, GoREvic PD, PAGE DL: Tumoral amyloidosis of boneof beta 2-microglobulin origin in association with long-term hemodi-alysis: A new type of amyloid disease. Hum Pathol 7:731—738, 1986

5. GEJYO F, HOMMA N, ARAKAWA M: Long-term complications ofdiallysis-pathogenic factors with special reference to amyloidosis.Kidney mt 43(Suppl 41):S78—S82, 1993

6. GAL R, KORZETS A, SCHWARTZ A, RATH-WOLFSON L, GAFFER U:Systemic distribution of /32-microglobulin-derived amyloidosis in pa-tients who undergo long-term hemodialysis. Arch Pathol Lab Med118:718—721, 1994

-j;'; -. - ? ,-_A4t-?I. —

I..• .

.i 'I

Inoue et at: Structure of /32-microglobulin amyloid fibrils 1549

7. BERGGARD I, BEARN AG: Isolation and properties of a low molecularweight f32-globulin occurring in human biological fluids. J Biol Chem243:4095—4103, 1968

8. VINCENT C, REVILLARD JP, GALLAND M, TRAEGER J: Serum 13,-microglobulin in hemodialyzed patients. Nephron 21:260—268, 1978

9. WIBELL L, EVRIN PE, BERGGARD I: Serum f32-microglobulin in renaldiseases. Nephron 10:320—331, 1973

10. DIRAIMONDO CR, CASEY TT, DIRAIMONDO CV, STONE WJ: Patho-logic features associated with idiopathic amyloidosis of bone inchronic hemodialysis patients. Nephron 43:22—27, 1986

11. MORITA T, SUZUKI M, KAMIMURA A, HIRASAWA Y: Amyloidosis ofpossible new type in patients on long-term hemodialysis. Arch PatholLab Med 109:1029—1032, 1985

12. CONNORS LH, SHIRAHAMA T, SKINNER M, FENVES A, COHEN AS: Invitro formation of amyloid fibrils from intact /32-microglobulin. Bio-chem Biophys Res Common 131:1063—1068, 1985

13. CAMPISTOL JM, SOLE M, BOMBI JA, RODRIGUEZ R, MIRAPEIX E,MUOZ-GOMEZ J, REVENT OW: In vitro spontaneous synthesis ofj32-microglobulin amyloid fibrils in peripheral blood mononuclear cellculture. Am J Pathol 141:241—247, 1992

14. IN0UE 5, KISILEVSKY R: A high resolution ultrastructural study ofexperimental murine AA amyloid. Lab Invest 74:670—683, 1996

15. CLEARY EC, GIBSON MA: Elastin-associated microfibrils and micro-fibrillar proteins. mt Rev Connect TLsue Res 10:97—209, 1983

16. RAVIOLA G: The fine structure of the ciliary zonule and ciliaryepithelium. With special regard to the organization and insertion ofthe zonular fibrils invest. Ophthalmol 10:851—869, 1971

17. COHEN AS, CALKINS E: Electron microscopic observations on afibrous component in amyloid of diverse origins. Nature 183:1202—1203, 1959

18. INOUE S: Ultrastructural organization of connective tissue microfibrilsin the posterior chamber of the eye in vivo and in vitro. Cell Tissue Res279:291—302, 1995a

19. INouc 5: Ultrastructural and immunohistochemical studies of micro-fibril-associated components in the posterior chamber of the eye. CellTissue Res 279:303—313, 1995b

20. BENDAYAN M, ZOLLINGER M: Ultrastructural localization of antigenicsites on osmium-fixed tissues applying the protein A-gold technique.J Histochem Cytochem 31:101—109, 1983

21. PRAS M, SCHUBERT M, ZUCKER-FRANKLIN D, RIMON A, FRANKLINEC: The characterization of soluble amyloid prepared in water. J ClinInvest 47:924—933, 1968

22. HUSBY G: Classification of amyloidosis, in Clinical Rheumatology (vol8, No 3), edited by HUSBY G, London, Baillière Tindall, 1994, pp503—511

23. SNow AD, WILLMER J, KISILEVSKY R: Sulfated glycosaminoglycans: Acommon constituent of all amyloid? Lab Invest 56:120—123, 1987

24. MCCUBBIN WD, KAY CM, NARINDRASORASAK 5, KISILEVSKY R:Circular dichroism and fluorenscence studies on two murine serumamyloid A proteins. Biochem J 256:775—783, 1988

25. SNOW AD, KISILEVSKY R: A close ultrastructural relationship between

sulfated proteoglycans and AA amyloid fibrils. Lab Invest 57:687—698,1988

26. LYON AW, NARINDRASORASAK 5, YOUNG ID, ANASTASSIADES T,COUCHMAN JR, MCCARTHY K, KISII.EVSKY R: Co-deposition ofbasement membrane components during the induction of murinesplenic AA amyloid. Lab Invest 64:785—790, 1991

27. KISILEVSKY R, FRASER P: Proteoglycans and amyloid fibrillogenesis, inThe Nature and Origin of Amyloid Fibrils, edited by BOCK GR, GOODEJA, Chichester, Wiley, 1996, pp 58—67

28. PEPYS MB, TENNENT GA, Boom DR, BELLO1-rI V, LOVAT LB, TANSY, PERSEY MR, HUTCHINSON WL, BOOTH SE, MADHOO 5, SOUTARAK, HAWKINS PN, VANZYLSMIT R, CAMPISTOL JM, FRASER PE,RADFORD SE, ROBINSON CV, SUNDE M, SERPELL LC, BLAKE CCF:Molecular mechanisms of fibrillogenesis and the protective role ofamyloid P component: Two possible avenues for therapy, in TheNature and Origin of Amyloid Fibrils, edited by BOCK GR, GOODE JA,Chichester, Wiley, 1996, pp 73—89

29. COHEN AS: The constitution and genesis of amyloid. Int Rev ExpPathol 4:159—243, 1965

30. COHEN AS, SHIRAHAMA T, SKINNER M: Electron microscopy ofamyloid, in Electron Microscopy of Proteins, edited by HARRIS JR,London, Academic Press, 1982, pp 165—205

31. SHIRAHAMA T, COHEN AS: High-resolution electron microscopicanalysis of the amyloid fibrils. J Cell Biol 33:679—708, 1967

32. SORENSEN GD, FINKE E: The ultrastructure of amyloid, in Amyloid-osis: Proceedings of the Symposium of Amyloidosis, University of Gro-ningen, The Netherlands, edited by MANDEMA E, RUINEN L, SCHOLTENJH, COHEN AS, Amsterdam, Excerpta Medica Foundation, 1968, pp184 —193

33. GUEI.-T B, GHIDONI JJ: The site of formation and ultrastructure ofamyloid. Am J Pathol 43:837—854, 1963

34. MIYAKAWA T, KATSURAGI S, KURAMOTO R: Ultrastructure of perivas-cular amyloid fibrils in Alzheimer's disease. Virchows Arch Cell Pathol56:21—24, 1988

35. MIYAKAWA T, KATSURAGI S, WATANABE K: Ultrastructure of amyloidfibrils in Alzheimer's disease and Down's syndrome, in Amyloid andAmyloidosis: Proceedings of 5th international Symposium on Amyloid-osis: Hakone, Japan, edited by IS0BE T, ARAKI 5, UCHINO F, K.A.To 5,TSIJBURA E, New York, Plenum Press, 1990, pp 573—578

36. TERRY RD, GONATAS NK, WEISS M: Ultrastructural studies inAlzheimer's presenile dementia. Am J Pathol 44:269—297, 1964

37. Sn's JD: Amyloidosis. Crit Rev Clin Lab Sci 31:325—354, 199438. GEJYO F, ARAKAWA M: /3,-microglobulin-associated amyloidoses.

J Int Med 232:531—532, 199239. OHASHI K, HARA M, YANAGISHITA M, KAWAI R, TACHIBANA S,

OGURA Y: Proteoglycans in hemodialysis-related amyloidosis. Vir-chows Arch 427:49—59, 1995

40. CASU B, PETITOU M, PROVASOLI M, SINAi P: Conformational flexi-bility: A new concept for explaining binding and biological propertiesof iduronic acid-containing glycosaminoglycans. Trends Biochem Sci13:221—225, 1988