- Email: [email protected]
Polymorphic Fibrillar Assembly of Human Amylin
JOURNAL OF STRUCTURAL BIOLOGY ARTICLE NO. SB973858
119, 17–27 (1997)
Polymorphic Fibrillar Assembly of Human Amylin Claire S. Goldsbury School of Biological Sciences, University of Auckland, New Zealand
Garth J. S. Cooper School of Biological Sciences and Department of Medicine, School of Medicine, University of Auckland, New Zealand
Kenneth N. Goldie School of Biological Sciences, University of Auckland, New Zealand
Shirley A. Mu¨ller M. E. Mu¨ller-Institute for Microscopy, Biozentrum, University of Basel, Switzerland
Etuate L. Saafi, W. T. M. Gruijters, and Martin P. Misur School of Biological Sciences, University of Auckland, New Zealand
Andreas Engel and Ueli Aebi M. E. Mu¨ller-Institute for Microscopy, Biozentrum, University of Basel, Switzerland
and Joerg Kistler School of Biological Sciences, University of Auckland, New Zealand Received November 6, 1996, and in revised form February 17, 1997
and occasionally large single-layered sheets. The mass-per-length (MPL) of the 5-nm protofibril is 10 kDa/nm. This has been established in two ways: first, the 8-nm fibril, which is formed by coiling two 5-nm protofibrils around each other, has an MPL of 20 kDa/nm. Second, higher order fibrils differ by increments of 10 kDa/nm. Hence, about 2.6 human amylin molecules (3904 Da) are packed in 1 nm of protofibril length. Similarities exist between amylin fibrils and those formed from other amyloid proteins, suggesting that the in vitro assembly of synthetic protein may serve as a useful model system in advancing our understanding of amyloid formation in disease. r 1997 Academic Press
Human amylin forms fibrillar amyloid between pancreatic islet cells in patients with non-insulindependent (type 2) diabetes mellitus. Fibrillar assemblies also form in vitro in aqueous solutions of synthetic human amylin. We now report on the structural polymorphism of these fibrils. The thinnest fibril, referred to as the protofibril, has an apparent width of 5 nm but is only rarely observed by itself. These protofibrils spontaneously assemble into higher order fibrillar structures with distinct morphologies. Prominent among these is an 8-nm fibril with a distinct 25-nm axial crossover repeat which is formed by left-handed coiling of two 5-nm protofibrils. Coiling of more than two 5-nm protofibrils results in cable-like structures of variable width depending on the number of protofibrils involved. Lateral (side-by-side) assembly of 5-nm protofibrils is also observed and produces ribbons which may contain two, three, four, or more protofibrils
INTRODUCTION
Human amylin (Cooper et al., 1989), also known as islet amyloid polypeptide (Johnson et al., 1989), is a 3904-Da peptide hormone. It is produced in the islet 17
1047-8477/97 $25.00 Copyright r 1997 by Academic Press All rights of reproduction in any form reserved.
18
GOLDSBURY ET AL.
b-cells of the pancreas, where it is synthesized and usually cosecreted with insulin (reviewed by Cooper, 1994). Amylin’s proposed normal function is as an endocrine hormone in the regulation of glucose metabolism (Cooper, 1994; Young et al., 1995). However, pathologically, it may contribute to insulin resistance (Leighton and Cooper, 1988) and is likely to play a significant role in the initiation and/or progression of non-insulin-dependent (type 2) diabetes mellitus (NIDDM) (Cooper, 1994; Verchere et al., 1996). Amylin was originally isolated from amyloid deposits in the islets of Langerhans of humans with NIDDM (Cooper et al., 1987). It belongs to a growing list of peptides which have been found to be abnormally deposited as fibrous protein aggregates in body tissues of individuals with one of a variety of diseases (reviewed by Sipe, 1992). Other well known examples are Alzheimer disease (b-amyloid protein (bA)), Creutzfeldt–Jakob disease (PrP), senile systemic amyloidosis (transthyretin), and medullary carcinoma of the thyroid (procalcitonin). It appears that the propensity of these proteins to form fibrils parallels their cytotoxicity. For example, fibrous preparations of human amylin have been shown to cause death of islet b-cells, whereas rat amylin, which does not form fibrils, has been shown to be nontoxic (Lorenzo et al., 1994). Similar cytotoxicity has been reported for fibrous bA (Lorenzo and Yankner, 1994; Schubert et al., 1995) and for fibrous PrP (DeArmond et al., 1985; Forloni et al., 1993). Information on the structure of amyloid fibrils is of great interest as it may contribute to an enhanced understanding of how these fibrils form and how the protein adopts cytotoxic properties. The different types of amyloid fibrils appear to have a common secondary structure which is dominated by beta sheets (reviewed by Bock and Goode, 1996). The best resolution has been achieved for transthyretin, the structure of which has been modeled on the basis of X-ray diffraction data to 2 Å resolution (Blake and Serpell, 1996). In this structure, beta sheets extended in regular twists along the length of the fibrils, implying that the polypeptide chains in the fibrils are hydrogen bonded together along the entire length, giving them great stability. Additionally, fragments of human amylin and of bA have been modeled as consisting of compact anti-parallel betasheet structures based on solid-state NMR spectroscopy and isotope-edited infrared spectroscopy (Griffiths et al., 1995; Lansbury et al., 1995). While fibrous assemblies have been documented by electron microscopy for all types of amyloid, only the assemblies of human calcitonin (Bauer et al., 1995) and the fibrillar structures contained in the neurofibrillary tangles of Alzheimer brains (Crowther, 1991) have been analyzed in detail and demonstrated to
exhibit polymorphism. From this work it is clear that protofibrils rarely remain single and instead have a great tendency to aggregate in various ways into higher order structures. The formation of fibrillar assemblies from synthetic human amylin has previously been reported but no details have been shown of fine structure (Charge´ et al., 1995; Westermark et al., 1996). Our report complements these previous ones and documents a distinct polymorphism of the higher order human amylin fibrils. Conventional transmission electron microscopy and mass-per-length measurements based on quantitative scanning transmission electron microscopy have been used to characterize these assemblies and to show how they are related to each other. Remarkable similarities have been found between human amylin fibrillar assemblies and those of other amyloid proteins, indicating that polymorphism in fibrillogenesis may be a widespread feature. MATERIALS AND METHODS In Vitro Assembly of Fibrillar Structures Assembly of fibrillar structures required only that lyophilized preparations of synthetic human amylin (Bachem, Torrence, CA) be solubilized in aqueous solution. We observed fibrils under a range of different buffer conditions and in pure water. As a standard condition we used human amylin solubilized at a concentration of 25 µM in 100 mM Tris–HCl, pH 7.2. In a separate experiment, to monitor the assembly of amylin fibrils, 125I-Tyr37 human amylin (Peninsula Laboratories, Belmont, CA) was added to the amylin solution to 35 000 cpm/ml, and incorporation of radioactivity into the forming fibrils measured following centrifugation (15 min, 20 000g) at various time points. Incorporation was 46% of total radioactivity after 24 hr and reached almost 100% after 96 hr. Assembly for 24 hr was used as a standard. Conventional Transmission Electron Microscopy (TEM) Fibril preparations were adsorbed to glow-discharged carboncoated collodion film on 400-mesh copper grids. For negative staining, the grids were washed with deionized water and stained with 2% uranyl acetate. For metal shadowing, the grids were washed and then drained such that only a thin layer of water remained. The grids were frozen in liquid nitrogen, freeze-dried, and unidirectionally metal shadowed (C/Pt/Ir) at an angle of 28°. Specimens were viewed and images recorded with a Hitachi 8000 or a Philips CM12 electron microscope operated at 100 or 80 kV, respectively. Measurements of fibril dimensions were taken from enlarged prints using a magnifying glass with a built-in calibrated ruler. Scanning Transmission Electron Microscopy (STEM) The mass-per-length (MPL) and full-width-half-maximum (FWHM) of unstained freeze-dried fibrils were determined by quantitative scanning transmission electron microscopy using a Vacuum Generators STEM HB5, operated at 80 kV as previously described (Mu¨ller et al., 1992). Amylin fibrils were adsorbed to glow-discharged ultrathin carbon films mounted on fenestrated carbon films adhered to gold-plated 200-mesh copper grids and washed five times in double quartz-distilled water. The grids were drained to leave only a thin water layer and subsequently
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN freeze-dried overnight in a pretreatment chamber directly connected to the STEM column. STEM dark-field images consisting of 512 3 512 pixels were recorded at nominal magnifications of 200 000. Using only the elastically scattered electrons, the image intensity signal was directly proportional to the mass distribution of the irradiated region, and following background subtraction, the MPL and FWHM of a sample could be determined quantitatively (Mu¨ller et al., 1992). MPL and FWHM measurements were determined for 100-nm-long fibril segments, and the results presented in the form of histograms fitted with Gaussian curves. The processing of STEM data was carried out using the IMPSYS computer package as previously described (Mu¨ller et al., 1992). Air dried Tobacco mosaic virus (TMV; kindly supplied by Dr. J. Witz, Institut de Biologie Moleculaire et Cellulaire, Strasbourg, France) prepared on ultrathin carbon film was imaged in the same way as described for amylin fibrils and used as a mass calibration standard and control. The known MPL for TMV of 131 6 3 kDa/nm was obtained in all control experiments. Mass loss correction was not required at the typical recording doses of 300 electrons/nm2. RESULTS
Classification of Polymorphic Amylin Fibrils Fibrillar structures formed in vitro from synthetic human amylin were polymorphic. They reached lengths of several micrometers and had variable widths, with a minimum of about 5 nm. Two fibril populations could readily be distinguished in negatively stained specimens (Fig. 1). One population contained fibrils which appeared smooth and had no noticeable axial repeat. Their width was variable, and it was evident that 5-nm fibrils could assemble laterally (side by side) to form ribbon-like structures or larger sheet-like arrays. The other population comprised helical fibrils of variable width. Most of these appeared relatively untextured by negative staining, but two exhibited distinct axial periodicities; one had an axial crossover repeat of 25 nm and an average width of 8 nm, the other had an average width of 13 nm and an axial crossover repeat of 50 nm. It is likely that these two types of fibrils represent stable end-points of coiling because more loosely coiled fibrils with larger and more variable crossover spacings were frequently seen. The 5-nm fibril was only rarely depicted by itself but could be readily identified as a distinct building block of some of the wider fibrils. It will therefore be referred to as the 5-nm protofibril (Fig. 2a). Two 5-nm protofibrils were often seen aligned side by side to form a 10-nm-wide ribbon (Fig. 2b). Frequently, these ribbons exhibited a slow twist along their length and yielded an apparent width of 5 nm at the crossovers. Ribbons formed by three, four, or more parallel protofibrils were less frequently but consistently observed (Figs. 2c and 2d). Ribbons were occasionally found to cross over at moderately regular intervals in a left-handed fashion. The thickness of the ribbons at these crossover sites was around 5 nm, indicating that these ribbons were single-
19
layered. Occasionally, sheet-like single-layered arrays of 5-nm protofibrils were also observed (Fig. 2e). It was more difficult to determine unambiguously the number of 5-nm protofibrils involved in the assembly of the coiled fibrils. The observation that the 8-nm fibril with a distinct 25-nm axial crossover repeat had alternating widths of 11 and 5 nm indicated that it was made of two protofibrils helically twisted around each other (Fig. 3a). By unidirectional metal shadowing of freeze-dried 8-nm fibrils, the two protofibrils were found to wind around each other with a left-handed pitch (Fig. 3b). Similarly, the 13-nm fibril with a 50-nm axial crossover repeat (Fig. 3c) was also left-handed as visualized by unidirectional metal shadowing (Fig. 3d). The number of protofibrils constituting this 13-nm fibril is likely to be three or four. Occasionally, two 8-nm fibrils were seen to join into an 11-nm-wide fibril without a distinct axial crossover repeat (Fig. 3e). Other higher order coiled fibrils with greater widths were consistently present in the preparations, but the number of protofibrils could not be identified. Mass-per-Length Analysis For a large portion of the higher order coiled fibrils, the number of protofibrils could not be determined unambiguously. We therefore measured their MPL and correlated this value with the fibril width as determined by their FWHM. For this purpose, unstained freeze-dried fibril preparations were analyzed by quantitative STEM (Fig. 4). Accordingly, the human amylin fibril preparations yielded MPL histograms which could be fitted by three Gaussian curves peaking at 19.9, 29, and 40.3 kDa/nm (Fig. 5a; Table I). It was evident that fibril mass increased in increments of approximately 10 kDa/nm, suggesting strongly that this represents the MPL of the 5-nm protofibril. The fibril width distribution also exhibited distinct peaks. They were at 7.9, 9.7, and 12.1 nm (Fig. 5b; Table I). To correlate the MPL with fibril width, fibrils falling into each of the three mass classes were analyzed separately for their width. The 20 kDa/nm MPL group correlated with fibrils which had a narrow width distribution averaging at 7.3 nm (Figs. 4a, 5c, and 5d; Table I). This group clearly contained the 8-nm coiled fibril with the distinct 25-nm axial crossover repeat (Fig. 4a). Ribbons made by the lateral association of two protofibrils probably also contributed to this group. The 30 kDa/nm MPL group of fibrils correlated with two width peaks, one at 8.3 nm and the other at 10.7 nm (Figs. 4b, 4c, 5e, and 5f; Table I). Given that these widths are less than 15 nm (i.e., the width measured for ribbons made from three side-by-side assembled 5-nm protofibrils, Fig. 2c), it appears more likely that the fibrils
20
GOLDSBURY ET AL.
FIG. 1. Polymorphism of human amylin fibrillar structures revealed by negative stain electron microscopy. Ribbons of variable width and coiled fibrils with distinct axial crossover repeats of 25 or 50 nm are evident. Bar, 100 nm.
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN
21
FIG. 2. Lateral (side-by-side) association of protofibrils. (a) A single 5-nm protofibril. (b–d) Ribbons assembled by lateral association of two (b), three (c), or five 5-nm protofibrils (d). The latter often crossed over in a left-handed sense at moderately regular intervals. (e) Lateral assembly of 5-nm protofibrils into single-layered, sheet-like arrays. Bar, 100 nm.
22
GOLDSBURY ET AL.
FIG. 3. Coiled association of protofibrils. (a, b) 8-nm fibrils with a left-handed 25-nm axial crossover repeat. (c, d) 13-nm fibrils with a left-handed 50-nm axial crossover repeat. (e) Two 8-nm fibrils join to form a fibril with an 11-nm diameter. Bar, 100 nm.
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN
23
FIG. 4. MPL analysis of human amylin fibrils by quantitative scanning transmission electron microscopy. (a) The 8-nm fibril with a distinct axial crossover repeat of 25-nm belongs to the 20 kDa/nm MPL group. (b) Fibrils belonging to the 20 or 30 kDa/nm MPL groups. (c) Fibrils which can be assigned to the 20 or 30 kDa/nm MPL groups and a rare example which cannot be unambiguously assigned to any of the three discrete MPL groups. (d) Fibrils belonging to the 20, 40, or 50 kDa/nm MPL groups. Bar, 50 nm.
24
GOLDSBURY ET AL.
FIG. 5. Correlation of MPL and FWHM (width) data. (a) MPL and (b) width values of the pooled data fitted with Gaussian curves. To correlate MPL and width, the fibril widths within a distinct MPL peak were plotted separately. The 20 kDa/nm peak: (c) MPL; (d) width. The 30 kDa/nm peak: (e) MPL; (f) width. The 40 kDa/nm peak: (g) MPL; (h) width.
in this group have three protofibrils coiled around each other. Fibrils in this width range were certainly abundantly seen in negatively stained preparations, but they did not clearly exhibit a three-stranded structure. However, this feature would be difficult to visualize at 20 Å resolution. Finally, the 40 kDa/nm MPL group correlated with three fibril widths: 9.3, 12, and 13.8 nm (Figs. 4c, 4d, 5g, and 5h; Table I). Fibrils with either the 12- or 13.8-nm width are likely to represent the 13-nm fibril with the distinct 50-nm axial crossover repeat, indicating that it contains four protofibrils (Figs. 3c and 3d). The fact that 40 kDa/nm fibrils exist with a width of 9.3 nm indicates that four protofibrils can also coil around each other in tighter conformations.
DISCUSSION
It is evident that synthetic human amylin peptide molecules spontaneously assemble into 5-nm protofibrils which further associate into higher order fibrils either by lateral (side-by-side) association or by coiling around each other. Images derived from conventional TEM of negatively stained or metalshadowed preparations and the mass-per-length data obtained with the STEM are consistent with each other in that the higher order fibrils are polymorphic. Our interpretation of the major fibrillar assemblies is summarized in Fig. 6. The lateral aggregation of 5-nm protofibrils produces ribbon-like structures which in some cases grow laterally into
25
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN
TABLE I Correlation of MPL and FWHM of Human Amylin Fibrils
All types 20 kDa/nm 30 kDa/nm 40 kDa/nm Average
MPL 1
MPL 2
MPL 3
Width 1
Width 2
Width 3
19.9 6 .3 20.4 6 .3
29.0 6 .3
40.3 6 .3
7.9 6 .5 7.3 6 .6 8.3 6 .7
9.7 6 .7
12.1 6 .7
10.7 6 .6 9.3 6 .7 9.9 6 .6
12.0 6 .4 12.1
29.3 6 .3 20.2
29.2
40.7 6 .3 40.5
7.8 6 .4
Width 4
13.8 6 .4 13.8
Note. MPL peaks (kDa/nm) and FWHM peaks (width in nm) of all fibril types are listed in the top row. Below, individual MPL groups of fibrils are correlated separately with the width distribution in each respective fibril group.
sheet-like arrays. Coiling of 5-nm protofibrils produces cable-like structures of variable width, depending on the number of protofibrils involved. Among these cable-like structures, two displayed distinct axial crossover repeats of 25 or 50 nm, while others appeared untextured at the relatively low resolution employed in this study. The 25-nm axial crossover repeat observed in both
the stained (Fig. 3a) and unstained (Fig. 4a) preparations documents the correlation between the stained 8-nm fibril and the unstained 20 kDa/nm MPL fibril. This 8-nm fibril is made from two protofibrils; therefore, the 5-nm protofibril has a MPL of 10 kDa/nm. This is further strongly supported by the discrete MPL distribution, which exhibits peaks spaced by 10 kDa/nm. In the case of other fibrillar assemblies,
FIG. 6. Scheme depicting the polymorphism of human amylin fibrillar assembly. Accordingly, the basic building block, i.e., the 5-nm protofibril, assembles either into ribbon-like or sheet-like arrays (top row) or into coiled fibrils (bottom row). Note that the four stranded fibrils can be of two types: one is formed by the coiling of four 5-nm protofibrils, and the other is generated by joining two 8-nm (with 25-nm axial crossover repeat) fibrils which each contain two 5-nm protofibrils.
26
GOLDSBURY ET AL.
some uncertainties exist concerning the FWHM values because negative staining and freeze-drying protocols are expected to result in different degrees of fibril flattening, and fibril width may also vary somewhat depending on the stain thickness. However, these uncertainties do not invalidate the scheme in Fig. 6; it does not depend in a major way on the precise correlation between stained and unstained images. Instead, the two data sets contribute to the scheme in a complementary fashion, one by distinguishing discrete fibrillar structures and recognizing the 5-nm protofibril as the common assembly unit, the other by contributing MPL values which differ by units of 10 kDa/nm and thus entirely agree with the notion of a basic assembly unit. Our data complement previous reports by others and confirm that human amylin readily forms fibrillar assemblies in vitro (Charge´ et al., 1995; Westermark et al., 1996) which appear similar to those found in pathological situations in vivo (De Koning et al., 1994). The present analysis goes further in that it describes distinct polymorphic structures and defines the assembly unit as a 5-nm protofibril with a mass-per-unit length of 10 kDa/nm. Based on this MPL, the 5-nm protofibril contains 2.6 of the 3904-Da human amylin molecules/nm of protofibril length. Using the equation MPL 5 prr2 with r 5 820 Da/nm3 (protein packing density), a cylindrical protofibril would have a diameter of 4 nm. Hence, it is likely that the fine structure of the protofibril is slightly more open than strictly cylindrical. Many different amyloid polypeptides have in common that they assume a beta-pleated sheet conformation and form fibrillar structures (Bock and Goode, 1996). This conformation allows hydrogen bonding over long distances along the fibril axis and thereby explains the great stability of amyloid fibrils (Blake and Serpell, 1996). It is therefore not unexpected that amylin fibrils share some lower resolution structural features with fibrils of other amyloid proteins. Among these, the human transthyretin fibril (Serpell et al., 1995), the human calcitonin fibril (Arvinte et al., 1993; Bauer et al., 1995), and the paired helical filament isolated from Alzheimer brains (Crowther, 1991) have been most extensively characterized by electron microscopy. The predominant transthyretin fibril has a width of 13 nm and consists of four protofibrils of 5–6 nm diameter. Its amylin counterpart would be the 40 kDa/nm fibril, also consisting of four protofibrils and assuming widths ranging from 9.3 to 13.8 nm. The human calcitonin protofibrils, which have an MPL of 9 kDa/nm and an apparent diameter of 4 nm, form an 8-nm fibril consisting of predominantly three 4-nm protofibrils coiled with a left-handed helical pitch. Higher order calcitonin assemblies appear to use this 8-nm fibril as their
common building block (Bauer et al., 1995). This is in contrast to human amylin, where higher order fibrillar structures increase in increments of the 5-nm, 10 kDa/nm protofibril unit. Fibrillar polymorphism has further been observed among the paired helical filaments isolated from Alzheimer brains (Crowther, 1991). Both the predominant paired helical filament and the less frequent straight filament have been reported to be assembled from a common protofibril (Crowther, 1991). In each case, two protofibrils coil around each other in a slightly different fashion producing two distinct fibrillar isoforms. It is evident that the polymorphism of fibrillar assemblies is widespread among human amyloid polypeptides. As more structural data on these assemblies is available, it becomes clear that while higher order fibril structures may differ somewhat between different amyloid proteins, they have in common that they form by lateral and/or helical association of protofibrils. Synthetic human amylin has an overwhelmingly strong tendency to form fibrillar assemblies in aqueous solutions. We have observed the same polymorphism of these structures under a wide variety of solution conditions of differing pH and salt concentrations. Consequently, we believe that our in vitro system is suitable for the characterization of the molecular interactions which drive fibrillogenesis and that this process of fibrillogenesis is likely to imitate that occurring in the in vivo disease state. Ultimately, the knowledge gained with in vitro systems, such as that with synthetic human amylin presented in this report, may contribute to the development of novel therapeutic means to prevent amyloid formation and fibril-associated cytotoxicity. We thank Cynthia Tse and Christina Buchanan for technical support and helpful discussions and Robi Wyss for drawing Fig. 6. This work was supported with grants from the Endocore Research Trust, the Lottery Grants Board of New Zealand, the Health Research Council of New Zealand, and the Auckland Medical Research Foundation. Andreas Engel and Ueli Aebi are supported by the M. Mu¨ller Foundation of Switzerland, the Swiss National Science Foundation, and the Kanton Basel-Stadt. REFERENCES Arvinte, T., Cudd, A., and Drake, A. F. (1993) The structure and mechanism of formation of human calcitonin fibrils, J. Biol. Chem. 268, 6415–6422. Bauer, H. H., Aebi, U., Ha¨ner, M., Hermann, R., Mu¨ller, M., Arvinte, T., and Merkle, H. P. (1995) Architecture and polymorphism of fibrillar supramolecular assemblies produced by in vitro aggregation of human calcitonin, J. Struct. Biol. 115, 1–15. Blake, C., and Serpell, L. (1996) Synchrotron X-ray studies suggest that the core of the transthyretin amyloid fibril is a continuous beta-sheet helix, Structure 4, 989–998. Bock, G. R., and Goode, J. A. (eds.) (1996) The Nature and Origin of Amyloid Fibrils, Ciba Foundation Symposium 199, Wiley, Chichester, UK.
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN Charge´, S. B. P., de Koning, E. J. P., and Clark, A. (1995) Effect of pH and insulin on fibrillogenesis of islet amyloid polypeptide in vitro, Biochemistry 34, 14588–14593. Cooper, G. J. S., Willis, A. C., Clark, A., Turner, R. C., Sim, R. B., and Reid, K. B. M. (1987) Purification and characterisation of a peptide from amyloid-rich pancreases of type 2 diabetic patients, Proc. Natl. Acad. Sci. USA 84, 8628–8632. Cooper, G. J. S., Willis, A. C., and Leighton, B. (1989) Amylin hormone, Nature 340, 272. Cooper, G. J. S. (1994) Amylin compared with calcitonin generelated peptide: Structure, biology, and relevance to metabolic disease, Endocr. Rev. 15, 163–201. Crowther, R. A. (1991) Straight and paired helical filaments in Alzheimer disease have a common structural unit, Proc. Natl. Acad. Sci. USA 88, 2288–2292. DeArmond, S. J., McKinley, M. P., Barry, R. A., Braunfeld, M. B., McColloch, J. R., and Prusiner, S. B. (1985) Identification of prion amyloid filaments in scrapie-infected brain, Cell 41, 221–235. De Koning, E. J. P., Morris, E. R., Hofhuis, F. M. A., Posthuma, G., Hoeppener, J. W. M., Morris, J. F., Capel, P. J. A., Clark, A., and Verbeek, S. (1994) Intra- and extracellular amyloid fibrils are formed in cultured pancreatic cells of transgenic mice expressing human islet amyloid polypeptide, Proc. Natl. Acad. Sci. USA 91, 8467–8471. Forloni, G., Angeretti, N., Chiesa, R., Monzani, E., Salmona, M., Bugiani, O., and Tagliavini, F. (1993) Neurotoxicity of a prion protein fragment, Nature 362, 543–546. Griffiths, J. M., Ashburn, T. T., Auger, M., Costa, P. R., Griffin, R. G., and Lansbury, P. T., Jr. (1995) Rotational resonance solid-state NMR elucidates a structural model of pancreatic amyloid, J. Am. Chem. Soc. 117, 3539–3546. Johnson, K. H., O’Brien, T. D., Betsholtz, C., and Westermark, P. (1989) Islet amyloid, islet-amyloid polypeptide, and diabetes mellitus, N. Engl. J. Med. 321, 513–518. Lansbury, P. T., Jr., Costa, P. R., Griffiths, J. M., Simon, E. J., Auger, M., Halverson, K. J., Kocisko, D. A., Hendsch, Z. S., Ashburn, T. T., Spencer, R. G. S., Tidor, B., and Griffin, R. G.
27
(1995) Structural model for the b-amyloid fibril based on interstrand alignment of an antiparallel-sheet comprising a C-terminal peptide, Nature Struct. Biol. 2, 990–998. Leighton, B., and Cooper, G. J. S. (1988) Pancreatic amylin and calcitonin gene-peptide cause resistance to insulin in skeletal muscle in vitro, Nature 335, 632–635. Lorenzo, A., Razzaboni, B., Weir, G. C., and Yankner, B. A. (1994) Pancreatic islet cell toxicity of amylin associated with type-2 diabetes mellitus, Nature 368, 756–760. Lorenzo, A., and Yankner, B. A. (1994) b-amyloid neurotoxicity requires fibril formation and is inhibited by Congo red, Proc. Natl. Acad. Sci. USA 91, 12243–12247. Mu¨ller, S. A., Goldie, K. N., Bu¨rki, R., Ha¨ring, R., and Engel, A. (1992) Factors influencing the precision of quantitative scanning transmission electron microscopy, Ultramicroscopy 46, 317–334. Schubert, D., Behl, C., Lesley, R., Brack, A., Dargusch, R., Sagara, Y., and Kimura, H. (1995) Amyloid peptides are toxic by a common oxidative mechanism, Proc. Natl. Acad. Sci. USA 92, 1989–1993. Serpell, L. C., Sunde, M., Fraser, P. E., Luther, P. K., Morris, E. P., Sangren, O., Lundgren, E., and Blake, C. C. F. (1995) Examination of the structure of the transthyretin amyloid fibril by image reconstruction from electron micrographs. J. Mol. Biol. 254, 113–118. Sipe, J. D. (1992) Amyloidosis, Annu. Rev. Biochem. 61, 947–975. Verchere, C. B., D’Alessio, D. A., Palmiter, R. D., Weir, G. C., Bonner-Weir, S., Baskin, D. G., and Kahn, S. E. (1996) Islet amyloid formation associated with hyperglycemia in transgenic mice with pancreatic beta cell expression of human islet amyloid polypeptide, Proc. Natl. Acad. Sci. USA 93, 3492–3496. Westermark, P., Li, Z. C., Westermark, G., Leckstrom, A., Steiner, D. F. (1996) Effect of beta cell granule components on human islet amyloid polypeptide fibril formation, FEBS Lett. 379, 203–206. Young, A., Pittner, R., Gedulin, B., Vine, W., and Rink, T. (1995) Amylin regulation of carbohydrate metabolism, Biochem. Soc. Trans. 23, 325–331.
119, 17–27 (1997)
Polymorphic Fibrillar Assembly of Human Amylin Claire S. Goldsbury School of Biological Sciences, University of Auckland, New Zealand
Garth J. S. Cooper School of Biological Sciences and Department of Medicine, School of Medicine, University of Auckland, New Zealand
Kenneth N. Goldie School of Biological Sciences, University of Auckland, New Zealand
Shirley A. Mu¨ller M. E. Mu¨ller-Institute for Microscopy, Biozentrum, University of Basel, Switzerland
Etuate L. Saafi, W. T. M. Gruijters, and Martin P. Misur School of Biological Sciences, University of Auckland, New Zealand
Andreas Engel and Ueli Aebi M. E. Mu¨ller-Institute for Microscopy, Biozentrum, University of Basel, Switzerland
and Joerg Kistler School of Biological Sciences, University of Auckland, New Zealand Received November 6, 1996, and in revised form February 17, 1997
and occasionally large single-layered sheets. The mass-per-length (MPL) of the 5-nm protofibril is 10 kDa/nm. This has been established in two ways: first, the 8-nm fibril, which is formed by coiling two 5-nm protofibrils around each other, has an MPL of 20 kDa/nm. Second, higher order fibrils differ by increments of 10 kDa/nm. Hence, about 2.6 human amylin molecules (3904 Da) are packed in 1 nm of protofibril length. Similarities exist between amylin fibrils and those formed from other amyloid proteins, suggesting that the in vitro assembly of synthetic protein may serve as a useful model system in advancing our understanding of amyloid formation in disease. r 1997 Academic Press
Human amylin forms fibrillar amyloid between pancreatic islet cells in patients with non-insulindependent (type 2) diabetes mellitus. Fibrillar assemblies also form in vitro in aqueous solutions of synthetic human amylin. We now report on the structural polymorphism of these fibrils. The thinnest fibril, referred to as the protofibril, has an apparent width of 5 nm but is only rarely observed by itself. These protofibrils spontaneously assemble into higher order fibrillar structures with distinct morphologies. Prominent among these is an 8-nm fibril with a distinct 25-nm axial crossover repeat which is formed by left-handed coiling of two 5-nm protofibrils. Coiling of more than two 5-nm protofibrils results in cable-like structures of variable width depending on the number of protofibrils involved. Lateral (side-by-side) assembly of 5-nm protofibrils is also observed and produces ribbons which may contain two, three, four, or more protofibrils
INTRODUCTION
Human amylin (Cooper et al., 1989), also known as islet amyloid polypeptide (Johnson et al., 1989), is a 3904-Da peptide hormone. It is produced in the islet 17
1047-8477/97 $25.00 Copyright r 1997 by Academic Press All rights of reproduction in any form reserved.
18
GOLDSBURY ET AL.
b-cells of the pancreas, where it is synthesized and usually cosecreted with insulin (reviewed by Cooper, 1994). Amylin’s proposed normal function is as an endocrine hormone in the regulation of glucose metabolism (Cooper, 1994; Young et al., 1995). However, pathologically, it may contribute to insulin resistance (Leighton and Cooper, 1988) and is likely to play a significant role in the initiation and/or progression of non-insulin-dependent (type 2) diabetes mellitus (NIDDM) (Cooper, 1994; Verchere et al., 1996). Amylin was originally isolated from amyloid deposits in the islets of Langerhans of humans with NIDDM (Cooper et al., 1987). It belongs to a growing list of peptides which have been found to be abnormally deposited as fibrous protein aggregates in body tissues of individuals with one of a variety of diseases (reviewed by Sipe, 1992). Other well known examples are Alzheimer disease (b-amyloid protein (bA)), Creutzfeldt–Jakob disease (PrP), senile systemic amyloidosis (transthyretin), and medullary carcinoma of the thyroid (procalcitonin). It appears that the propensity of these proteins to form fibrils parallels their cytotoxicity. For example, fibrous preparations of human amylin have been shown to cause death of islet b-cells, whereas rat amylin, which does not form fibrils, has been shown to be nontoxic (Lorenzo et al., 1994). Similar cytotoxicity has been reported for fibrous bA (Lorenzo and Yankner, 1994; Schubert et al., 1995) and for fibrous PrP (DeArmond et al., 1985; Forloni et al., 1993). Information on the structure of amyloid fibrils is of great interest as it may contribute to an enhanced understanding of how these fibrils form and how the protein adopts cytotoxic properties. The different types of amyloid fibrils appear to have a common secondary structure which is dominated by beta sheets (reviewed by Bock and Goode, 1996). The best resolution has been achieved for transthyretin, the structure of which has been modeled on the basis of X-ray diffraction data to 2 Å resolution (Blake and Serpell, 1996). In this structure, beta sheets extended in regular twists along the length of the fibrils, implying that the polypeptide chains in the fibrils are hydrogen bonded together along the entire length, giving them great stability. Additionally, fragments of human amylin and of bA have been modeled as consisting of compact anti-parallel betasheet structures based on solid-state NMR spectroscopy and isotope-edited infrared spectroscopy (Griffiths et al., 1995; Lansbury et al., 1995). While fibrous assemblies have been documented by electron microscopy for all types of amyloid, only the assemblies of human calcitonin (Bauer et al., 1995) and the fibrillar structures contained in the neurofibrillary tangles of Alzheimer brains (Crowther, 1991) have been analyzed in detail and demonstrated to
exhibit polymorphism. From this work it is clear that protofibrils rarely remain single and instead have a great tendency to aggregate in various ways into higher order structures. The formation of fibrillar assemblies from synthetic human amylin has previously been reported but no details have been shown of fine structure (Charge´ et al., 1995; Westermark et al., 1996). Our report complements these previous ones and documents a distinct polymorphism of the higher order human amylin fibrils. Conventional transmission electron microscopy and mass-per-length measurements based on quantitative scanning transmission electron microscopy have been used to characterize these assemblies and to show how they are related to each other. Remarkable similarities have been found between human amylin fibrillar assemblies and those of other amyloid proteins, indicating that polymorphism in fibrillogenesis may be a widespread feature. MATERIALS AND METHODS In Vitro Assembly of Fibrillar Structures Assembly of fibrillar structures required only that lyophilized preparations of synthetic human amylin (Bachem, Torrence, CA) be solubilized in aqueous solution. We observed fibrils under a range of different buffer conditions and in pure water. As a standard condition we used human amylin solubilized at a concentration of 25 µM in 100 mM Tris–HCl, pH 7.2. In a separate experiment, to monitor the assembly of amylin fibrils, 125I-Tyr37 human amylin (Peninsula Laboratories, Belmont, CA) was added to the amylin solution to 35 000 cpm/ml, and incorporation of radioactivity into the forming fibrils measured following centrifugation (15 min, 20 000g) at various time points. Incorporation was 46% of total radioactivity after 24 hr and reached almost 100% after 96 hr. Assembly for 24 hr was used as a standard. Conventional Transmission Electron Microscopy (TEM) Fibril preparations were adsorbed to glow-discharged carboncoated collodion film on 400-mesh copper grids. For negative staining, the grids were washed with deionized water and stained with 2% uranyl acetate. For metal shadowing, the grids were washed and then drained such that only a thin layer of water remained. The grids were frozen in liquid nitrogen, freeze-dried, and unidirectionally metal shadowed (C/Pt/Ir) at an angle of 28°. Specimens were viewed and images recorded with a Hitachi 8000 or a Philips CM12 electron microscope operated at 100 or 80 kV, respectively. Measurements of fibril dimensions were taken from enlarged prints using a magnifying glass with a built-in calibrated ruler. Scanning Transmission Electron Microscopy (STEM) The mass-per-length (MPL) and full-width-half-maximum (FWHM) of unstained freeze-dried fibrils were determined by quantitative scanning transmission electron microscopy using a Vacuum Generators STEM HB5, operated at 80 kV as previously described (Mu¨ller et al., 1992). Amylin fibrils were adsorbed to glow-discharged ultrathin carbon films mounted on fenestrated carbon films adhered to gold-plated 200-mesh copper grids and washed five times in double quartz-distilled water. The grids were drained to leave only a thin water layer and subsequently
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN freeze-dried overnight in a pretreatment chamber directly connected to the STEM column. STEM dark-field images consisting of 512 3 512 pixels were recorded at nominal magnifications of 200 000. Using only the elastically scattered electrons, the image intensity signal was directly proportional to the mass distribution of the irradiated region, and following background subtraction, the MPL and FWHM of a sample could be determined quantitatively (Mu¨ller et al., 1992). MPL and FWHM measurements were determined for 100-nm-long fibril segments, and the results presented in the form of histograms fitted with Gaussian curves. The processing of STEM data was carried out using the IMPSYS computer package as previously described (Mu¨ller et al., 1992). Air dried Tobacco mosaic virus (TMV; kindly supplied by Dr. J. Witz, Institut de Biologie Moleculaire et Cellulaire, Strasbourg, France) prepared on ultrathin carbon film was imaged in the same way as described for amylin fibrils and used as a mass calibration standard and control. The known MPL for TMV of 131 6 3 kDa/nm was obtained in all control experiments. Mass loss correction was not required at the typical recording doses of 300 electrons/nm2. RESULTS
Classification of Polymorphic Amylin Fibrils Fibrillar structures formed in vitro from synthetic human amylin were polymorphic. They reached lengths of several micrometers and had variable widths, with a minimum of about 5 nm. Two fibril populations could readily be distinguished in negatively stained specimens (Fig. 1). One population contained fibrils which appeared smooth and had no noticeable axial repeat. Their width was variable, and it was evident that 5-nm fibrils could assemble laterally (side by side) to form ribbon-like structures or larger sheet-like arrays. The other population comprised helical fibrils of variable width. Most of these appeared relatively untextured by negative staining, but two exhibited distinct axial periodicities; one had an axial crossover repeat of 25 nm and an average width of 8 nm, the other had an average width of 13 nm and an axial crossover repeat of 50 nm. It is likely that these two types of fibrils represent stable end-points of coiling because more loosely coiled fibrils with larger and more variable crossover spacings were frequently seen. The 5-nm fibril was only rarely depicted by itself but could be readily identified as a distinct building block of some of the wider fibrils. It will therefore be referred to as the 5-nm protofibril (Fig. 2a). Two 5-nm protofibrils were often seen aligned side by side to form a 10-nm-wide ribbon (Fig. 2b). Frequently, these ribbons exhibited a slow twist along their length and yielded an apparent width of 5 nm at the crossovers. Ribbons formed by three, four, or more parallel protofibrils were less frequently but consistently observed (Figs. 2c and 2d). Ribbons were occasionally found to cross over at moderately regular intervals in a left-handed fashion. The thickness of the ribbons at these crossover sites was around 5 nm, indicating that these ribbons were single-
19
layered. Occasionally, sheet-like single-layered arrays of 5-nm protofibrils were also observed (Fig. 2e). It was more difficult to determine unambiguously the number of 5-nm protofibrils involved in the assembly of the coiled fibrils. The observation that the 8-nm fibril with a distinct 25-nm axial crossover repeat had alternating widths of 11 and 5 nm indicated that it was made of two protofibrils helically twisted around each other (Fig. 3a). By unidirectional metal shadowing of freeze-dried 8-nm fibrils, the two protofibrils were found to wind around each other with a left-handed pitch (Fig. 3b). Similarly, the 13-nm fibril with a 50-nm axial crossover repeat (Fig. 3c) was also left-handed as visualized by unidirectional metal shadowing (Fig. 3d). The number of protofibrils constituting this 13-nm fibril is likely to be three or four. Occasionally, two 8-nm fibrils were seen to join into an 11-nm-wide fibril without a distinct axial crossover repeat (Fig. 3e). Other higher order coiled fibrils with greater widths were consistently present in the preparations, but the number of protofibrils could not be identified. Mass-per-Length Analysis For a large portion of the higher order coiled fibrils, the number of protofibrils could not be determined unambiguously. We therefore measured their MPL and correlated this value with the fibril width as determined by their FWHM. For this purpose, unstained freeze-dried fibril preparations were analyzed by quantitative STEM (Fig. 4). Accordingly, the human amylin fibril preparations yielded MPL histograms which could be fitted by three Gaussian curves peaking at 19.9, 29, and 40.3 kDa/nm (Fig. 5a; Table I). It was evident that fibril mass increased in increments of approximately 10 kDa/nm, suggesting strongly that this represents the MPL of the 5-nm protofibril. The fibril width distribution also exhibited distinct peaks. They were at 7.9, 9.7, and 12.1 nm (Fig. 5b; Table I). To correlate the MPL with fibril width, fibrils falling into each of the three mass classes were analyzed separately for their width. The 20 kDa/nm MPL group correlated with fibrils which had a narrow width distribution averaging at 7.3 nm (Figs. 4a, 5c, and 5d; Table I). This group clearly contained the 8-nm coiled fibril with the distinct 25-nm axial crossover repeat (Fig. 4a). Ribbons made by the lateral association of two protofibrils probably also contributed to this group. The 30 kDa/nm MPL group of fibrils correlated with two width peaks, one at 8.3 nm and the other at 10.7 nm (Figs. 4b, 4c, 5e, and 5f; Table I). Given that these widths are less than 15 nm (i.e., the width measured for ribbons made from three side-by-side assembled 5-nm protofibrils, Fig. 2c), it appears more likely that the fibrils
20
GOLDSBURY ET AL.
FIG. 1. Polymorphism of human amylin fibrillar structures revealed by negative stain electron microscopy. Ribbons of variable width and coiled fibrils with distinct axial crossover repeats of 25 or 50 nm are evident. Bar, 100 nm.
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN
21
FIG. 2. Lateral (side-by-side) association of protofibrils. (a) A single 5-nm protofibril. (b–d) Ribbons assembled by lateral association of two (b), three (c), or five 5-nm protofibrils (d). The latter often crossed over in a left-handed sense at moderately regular intervals. (e) Lateral assembly of 5-nm protofibrils into single-layered, sheet-like arrays. Bar, 100 nm.
22
GOLDSBURY ET AL.
FIG. 3. Coiled association of protofibrils. (a, b) 8-nm fibrils with a left-handed 25-nm axial crossover repeat. (c, d) 13-nm fibrils with a left-handed 50-nm axial crossover repeat. (e) Two 8-nm fibrils join to form a fibril with an 11-nm diameter. Bar, 100 nm.
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN
23
FIG. 4. MPL analysis of human amylin fibrils by quantitative scanning transmission electron microscopy. (a) The 8-nm fibril with a distinct axial crossover repeat of 25-nm belongs to the 20 kDa/nm MPL group. (b) Fibrils belonging to the 20 or 30 kDa/nm MPL groups. (c) Fibrils which can be assigned to the 20 or 30 kDa/nm MPL groups and a rare example which cannot be unambiguously assigned to any of the three discrete MPL groups. (d) Fibrils belonging to the 20, 40, or 50 kDa/nm MPL groups. Bar, 50 nm.
24
GOLDSBURY ET AL.
FIG. 5. Correlation of MPL and FWHM (width) data. (a) MPL and (b) width values of the pooled data fitted with Gaussian curves. To correlate MPL and width, the fibril widths within a distinct MPL peak were plotted separately. The 20 kDa/nm peak: (c) MPL; (d) width. The 30 kDa/nm peak: (e) MPL; (f) width. The 40 kDa/nm peak: (g) MPL; (h) width.
in this group have three protofibrils coiled around each other. Fibrils in this width range were certainly abundantly seen in negatively stained preparations, but they did not clearly exhibit a three-stranded structure. However, this feature would be difficult to visualize at 20 Å resolution. Finally, the 40 kDa/nm MPL group correlated with three fibril widths: 9.3, 12, and 13.8 nm (Figs. 4c, 4d, 5g, and 5h; Table I). Fibrils with either the 12- or 13.8-nm width are likely to represent the 13-nm fibril with the distinct 50-nm axial crossover repeat, indicating that it contains four protofibrils (Figs. 3c and 3d). The fact that 40 kDa/nm fibrils exist with a width of 9.3 nm indicates that four protofibrils can also coil around each other in tighter conformations.
DISCUSSION
It is evident that synthetic human amylin peptide molecules spontaneously assemble into 5-nm protofibrils which further associate into higher order fibrils either by lateral (side-by-side) association or by coiling around each other. Images derived from conventional TEM of negatively stained or metalshadowed preparations and the mass-per-length data obtained with the STEM are consistent with each other in that the higher order fibrils are polymorphic. Our interpretation of the major fibrillar assemblies is summarized in Fig. 6. The lateral aggregation of 5-nm protofibrils produces ribbon-like structures which in some cases grow laterally into
25
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN
TABLE I Correlation of MPL and FWHM of Human Amylin Fibrils
All types 20 kDa/nm 30 kDa/nm 40 kDa/nm Average
MPL 1
MPL 2
MPL 3
Width 1
Width 2
Width 3
19.9 6 .3 20.4 6 .3
29.0 6 .3
40.3 6 .3
7.9 6 .5 7.3 6 .6 8.3 6 .7
9.7 6 .7
12.1 6 .7
10.7 6 .6 9.3 6 .7 9.9 6 .6
12.0 6 .4 12.1
29.3 6 .3 20.2
29.2
40.7 6 .3 40.5
7.8 6 .4
Width 4
13.8 6 .4 13.8
Note. MPL peaks (kDa/nm) and FWHM peaks (width in nm) of all fibril types are listed in the top row. Below, individual MPL groups of fibrils are correlated separately with the width distribution in each respective fibril group.
sheet-like arrays. Coiling of 5-nm protofibrils produces cable-like structures of variable width, depending on the number of protofibrils involved. Among these cable-like structures, two displayed distinct axial crossover repeats of 25 or 50 nm, while others appeared untextured at the relatively low resolution employed in this study. The 25-nm axial crossover repeat observed in both
the stained (Fig. 3a) and unstained (Fig. 4a) preparations documents the correlation between the stained 8-nm fibril and the unstained 20 kDa/nm MPL fibril. This 8-nm fibril is made from two protofibrils; therefore, the 5-nm protofibril has a MPL of 10 kDa/nm. This is further strongly supported by the discrete MPL distribution, which exhibits peaks spaced by 10 kDa/nm. In the case of other fibrillar assemblies,
FIG. 6. Scheme depicting the polymorphism of human amylin fibrillar assembly. Accordingly, the basic building block, i.e., the 5-nm protofibril, assembles either into ribbon-like or sheet-like arrays (top row) or into coiled fibrils (bottom row). Note that the four stranded fibrils can be of two types: one is formed by the coiling of four 5-nm protofibrils, and the other is generated by joining two 8-nm (with 25-nm axial crossover repeat) fibrils which each contain two 5-nm protofibrils.
26
GOLDSBURY ET AL.
some uncertainties exist concerning the FWHM values because negative staining and freeze-drying protocols are expected to result in different degrees of fibril flattening, and fibril width may also vary somewhat depending on the stain thickness. However, these uncertainties do not invalidate the scheme in Fig. 6; it does not depend in a major way on the precise correlation between stained and unstained images. Instead, the two data sets contribute to the scheme in a complementary fashion, one by distinguishing discrete fibrillar structures and recognizing the 5-nm protofibril as the common assembly unit, the other by contributing MPL values which differ by units of 10 kDa/nm and thus entirely agree with the notion of a basic assembly unit. Our data complement previous reports by others and confirm that human amylin readily forms fibrillar assemblies in vitro (Charge´ et al., 1995; Westermark et al., 1996) which appear similar to those found in pathological situations in vivo (De Koning et al., 1994). The present analysis goes further in that it describes distinct polymorphic structures and defines the assembly unit as a 5-nm protofibril with a mass-per-unit length of 10 kDa/nm. Based on this MPL, the 5-nm protofibril contains 2.6 of the 3904-Da human amylin molecules/nm of protofibril length. Using the equation MPL 5 prr2 with r 5 820 Da/nm3 (protein packing density), a cylindrical protofibril would have a diameter of 4 nm. Hence, it is likely that the fine structure of the protofibril is slightly more open than strictly cylindrical. Many different amyloid polypeptides have in common that they assume a beta-pleated sheet conformation and form fibrillar structures (Bock and Goode, 1996). This conformation allows hydrogen bonding over long distances along the fibril axis and thereby explains the great stability of amyloid fibrils (Blake and Serpell, 1996). It is therefore not unexpected that amylin fibrils share some lower resolution structural features with fibrils of other amyloid proteins. Among these, the human transthyretin fibril (Serpell et al., 1995), the human calcitonin fibril (Arvinte et al., 1993; Bauer et al., 1995), and the paired helical filament isolated from Alzheimer brains (Crowther, 1991) have been most extensively characterized by electron microscopy. The predominant transthyretin fibril has a width of 13 nm and consists of four protofibrils of 5–6 nm diameter. Its amylin counterpart would be the 40 kDa/nm fibril, also consisting of four protofibrils and assuming widths ranging from 9.3 to 13.8 nm. The human calcitonin protofibrils, which have an MPL of 9 kDa/nm and an apparent diameter of 4 nm, form an 8-nm fibril consisting of predominantly three 4-nm protofibrils coiled with a left-handed helical pitch. Higher order calcitonin assemblies appear to use this 8-nm fibril as their
common building block (Bauer et al., 1995). This is in contrast to human amylin, where higher order fibrillar structures increase in increments of the 5-nm, 10 kDa/nm protofibril unit. Fibrillar polymorphism has further been observed among the paired helical filaments isolated from Alzheimer brains (Crowther, 1991). Both the predominant paired helical filament and the less frequent straight filament have been reported to be assembled from a common protofibril (Crowther, 1991). In each case, two protofibrils coil around each other in a slightly different fashion producing two distinct fibrillar isoforms. It is evident that the polymorphism of fibrillar assemblies is widespread among human amyloid polypeptides. As more structural data on these assemblies is available, it becomes clear that while higher order fibril structures may differ somewhat between different amyloid proteins, they have in common that they form by lateral and/or helical association of protofibrils. Synthetic human amylin has an overwhelmingly strong tendency to form fibrillar assemblies in aqueous solutions. We have observed the same polymorphism of these structures under a wide variety of solution conditions of differing pH and salt concentrations. Consequently, we believe that our in vitro system is suitable for the characterization of the molecular interactions which drive fibrillogenesis and that this process of fibrillogenesis is likely to imitate that occurring in the in vivo disease state. Ultimately, the knowledge gained with in vitro systems, such as that with synthetic human amylin presented in this report, may contribute to the development of novel therapeutic means to prevent amyloid formation and fibril-associated cytotoxicity. We thank Cynthia Tse and Christina Buchanan for technical support and helpful discussions and Robi Wyss for drawing Fig. 6. This work was supported with grants from the Endocore Research Trust, the Lottery Grants Board of New Zealand, the Health Research Council of New Zealand, and the Auckland Medical Research Foundation. Andreas Engel and Ueli Aebi are supported by the M. Mu¨ller Foundation of Switzerland, the Swiss National Science Foundation, and the Kanton Basel-Stadt. REFERENCES Arvinte, T., Cudd, A., and Drake, A. F. (1993) The structure and mechanism of formation of human calcitonin fibrils, J. Biol. Chem. 268, 6415–6422. Bauer, H. H., Aebi, U., Ha¨ner, M., Hermann, R., Mu¨ller, M., Arvinte, T., and Merkle, H. P. (1995) Architecture and polymorphism of fibrillar supramolecular assemblies produced by in vitro aggregation of human calcitonin, J. Struct. Biol. 115, 1–15. Blake, C., and Serpell, L. (1996) Synchrotron X-ray studies suggest that the core of the transthyretin amyloid fibril is a continuous beta-sheet helix, Structure 4, 989–998. Bock, G. R., and Goode, J. A. (eds.) (1996) The Nature and Origin of Amyloid Fibrils, Ciba Foundation Symposium 199, Wiley, Chichester, UK.
POLYMORPHIC FIBRILLAR ASSEMBLY OF HUMAN AMYLIN Charge´, S. B. P., de Koning, E. J. P., and Clark, A. (1995) Effect of pH and insulin on fibrillogenesis of islet amyloid polypeptide in vitro, Biochemistry 34, 14588–14593. Cooper, G. J. S., Willis, A. C., Clark, A., Turner, R. C., Sim, R. B., and Reid, K. B. M. (1987) Purification and characterisation of a peptide from amyloid-rich pancreases of type 2 diabetic patients, Proc. Natl. Acad. Sci. USA 84, 8628–8632. Cooper, G. J. S., Willis, A. C., and Leighton, B. (1989) Amylin hormone, Nature 340, 272. Cooper, G. J. S. (1994) Amylin compared with calcitonin generelated peptide: Structure, biology, and relevance to metabolic disease, Endocr. Rev. 15, 163–201. Crowther, R. A. (1991) Straight and paired helical filaments in Alzheimer disease have a common structural unit, Proc. Natl. Acad. Sci. USA 88, 2288–2292. DeArmond, S. J., McKinley, M. P., Barry, R. A., Braunfeld, M. B., McColloch, J. R., and Prusiner, S. B. (1985) Identification of prion amyloid filaments in scrapie-infected brain, Cell 41, 221–235. De Koning, E. J. P., Morris, E. R., Hofhuis, F. M. A., Posthuma, G., Hoeppener, J. W. M., Morris, J. F., Capel, P. J. A., Clark, A., and Verbeek, S. (1994) Intra- and extracellular amyloid fibrils are formed in cultured pancreatic cells of transgenic mice expressing human islet amyloid polypeptide, Proc. Natl. Acad. Sci. USA 91, 8467–8471. Forloni, G., Angeretti, N., Chiesa, R., Monzani, E., Salmona, M., Bugiani, O., and Tagliavini, F. (1993) Neurotoxicity of a prion protein fragment, Nature 362, 543–546. Griffiths, J. M., Ashburn, T. T., Auger, M., Costa, P. R., Griffin, R. G., and Lansbury, P. T., Jr. (1995) Rotational resonance solid-state NMR elucidates a structural model of pancreatic amyloid, J. Am. Chem. Soc. 117, 3539–3546. Johnson, K. H., O’Brien, T. D., Betsholtz, C., and Westermark, P. (1989) Islet amyloid, islet-amyloid polypeptide, and diabetes mellitus, N. Engl. J. Med. 321, 513–518. Lansbury, P. T., Jr., Costa, P. R., Griffiths, J. M., Simon, E. J., Auger, M., Halverson, K. J., Kocisko, D. A., Hendsch, Z. S., Ashburn, T. T., Spencer, R. G. S., Tidor, B., and Griffin, R. G.
27
(1995) Structural model for the b-amyloid fibril based on interstrand alignment of an antiparallel-sheet comprising a C-terminal peptide, Nature Struct. Biol. 2, 990–998. Leighton, B., and Cooper, G. J. S. (1988) Pancreatic amylin and calcitonin gene-peptide cause resistance to insulin in skeletal muscle in vitro, Nature 335, 632–635. Lorenzo, A., Razzaboni, B., Weir, G. C., and Yankner, B. A. (1994) Pancreatic islet cell toxicity of amylin associated with type-2 diabetes mellitus, Nature 368, 756–760. Lorenzo, A., and Yankner, B. A. (1994) b-amyloid neurotoxicity requires fibril formation and is inhibited by Congo red, Proc. Natl. Acad. Sci. USA 91, 12243–12247. Mu¨ller, S. A., Goldie, K. N., Bu¨rki, R., Ha¨ring, R., and Engel, A. (1992) Factors influencing the precision of quantitative scanning transmission electron microscopy, Ultramicroscopy 46, 317–334. Schubert, D., Behl, C., Lesley, R., Brack, A., Dargusch, R., Sagara, Y., and Kimura, H. (1995) Amyloid peptides are toxic by a common oxidative mechanism, Proc. Natl. Acad. Sci. USA 92, 1989–1993. Serpell, L. C., Sunde, M., Fraser, P. E., Luther, P. K., Morris, E. P., Sangren, O., Lundgren, E., and Blake, C. C. F. (1995) Examination of the structure of the transthyretin amyloid fibril by image reconstruction from electron micrographs. J. Mol. Biol. 254, 113–118. Sipe, J. D. (1992) Amyloidosis, Annu. Rev. Biochem. 61, 947–975. Verchere, C. B., D’Alessio, D. A., Palmiter, R. D., Weir, G. C., Bonner-Weir, S., Baskin, D. G., and Kahn, S. E. (1996) Islet amyloid formation associated with hyperglycemia in transgenic mice with pancreatic beta cell expression of human islet amyloid polypeptide, Proc. Natl. Acad. Sci. USA 93, 3492–3496. Westermark, P., Li, Z. C., Westermark, G., Leckstrom, A., Steiner, D. F. (1996) Effect of beta cell granule components on human islet amyloid polypeptide fibril formation, FEBS Lett. 379, 203–206. Young, A., Pittner, R., Gedulin, B., Vine, W., and Rink, T. (1995) Amylin regulation of carbohydrate metabolism, Biochem. Soc. Trans. 23, 325–331.