Amyloidogenic nature of spider silk John M. Kenney 1 , David Knight 2 , Michael J. Wise 3 and Fritz Vollrath 2,4 1 Institute for Storage Ring Facilities, University of Aarhus, Denmark; 2 Department of Zoology, University of Oxford, UK; 3 Department of Genetics, University of Cambridge, UK; 4 Department of Zoology, University of Aarhus, Denmark In spiders soluble proteins are converted to form insoluble silk fibres, stronger than steel. The final fibre product has long been the subject of study; however, little is known about the conversion process in the silk-producing gland of the spider. Here we describe a study of the conversion of the soluble form of the major spider-silk protein, spidroin, directly extracted from the silk gland, to a b-sheet enriched state using circular dichroism (CD) spectroscopy. Combined with electron microscopy (EM) data showing fibril forma- tion in the b-sheet rich region of the gland and amino-acid sequence analyses linking spidroin and amyloids, these results lead us to suggest that the refolding conversion is amyloid like. We also propose that spider silk could be a valuable model system for testing hypotheses concerning b-sheet formation in other fibrilogenic systems, including amyloids. Keywords: spider silk; spidroin, amyloids; CD spectroscopy; low complexity peptides. Spiders have evolved sophisticated systems not only to produce and store proteins at high concentrations (‡ 30%) in solution, but also to control the conversion of these soluble proteins to form insoluble silk fibres [1]. The conversion process appears to rely on refolding of one or two members of the spidroin family of major silk proteins [2,3], but is poorly understood. Although the structure of the insect silkworm (Bombyx mori) silk has been well characterized [4] and a model [5] has been proposed for insect silk formation based on regenerated fibroin (the major protein in silkworm silk), the structure of spider silk is less well understood and appears to be different from insect silk in several respects [6]. Though progress has also been made in spinning fibres from recombinant spider silk produced in mammalian cells [7], the formation process in the spider has remained largely unexplored. Here, we describe the conversion of spidroin in the spider (Nephila edulis) using circular dichroism spectroscopy, electron microscopy and amino-acid sequence analyses. The results reveal a striking molecular-level similarity between spider- silk processing and amyloid-fibril formation. We conclude that the spider-silk production process, particularly the mechanisms that the spider employs to secrete, store and manipulate silk protein, could prove to be a valuable model system for exploring fibrilogenesis of amyloid, prion and other related proteins. EXPERIMENTAL PROCEDURES Spider and sample preparation Prior to dissection, Nephila edulis female spiders were kept in plastic boxes (10 · 40 · 40 cm) and exclusively fed flies for several weeks. Under these conditions the spiders do not make webs thus ensuring a large accumulation of spidroin in their silk glands. The major-ampullate dragline-produ- cing silk gland of a freshly decapitated spider was removed and the epithelium of the ampulla was carefully stripped off under spider Ringer solution (pH 7.4) [8]. The highly viscous remainder containing concentrated spidroin was separated into fractions derived from the two morphologi- cally distinct regions of the gland known as the A and B zones [9]. Freshly prepared samples from individual glands of undiluted solute and solute diluted (1 : 4) in spider Ringer solution were used for CD analysis. Circular dichroism spectroscopy and secondary structure prediction The samples were carefully loaded into a 0.01 mm light path quartz cell (Hellma 124-QS) to avoid shearing. Prior to CD analysis they were examined by polarized light microscopy to confirm that loaded samples did not exhibit birefringence and therefore had not suffered from shear-induced poly- merization. Spectra were recorded for each sample, two raw (undiluted) and three diluted from each zone, over a series of temperatures from 20 °Cupto75°CusingaJasco J720 CD UV spectrometer equipped with a water-bath temperature controller. Each spectrum is the average of 16 scans (each 3 min in duration) from 260 nm to 190 nm with 0.5 nm steps. The samples were allowed to equilibrate for at least 10 min at each temperature before the CD spectrum was recorded. Several times the spectrum was repeatedly recorded at the same temperature to ensure that it was stable in time. The spectra of five (three diluted and two undiluted) different samples of A zone were essentially identical, though the spectra of the undiluted samples exhibited a lower signal-to-noise ratio, probably due to light losses resulting from absorption by the high protein concentration and scattering by inhomogeneous distribution of the sample material in the cell. However, the similarity of the spectra of diluted and undiluted samples confirmed that the diluted samples spectra are valid measures of the in vivo molecular Correspondence to J. M. Kenney, Institute for Storage Ring Facilities, University of Aarhus, 8000 Aarhus C, Denmark. Fax: + 45 8612 0740, Tel.: + 45 8942 3721, E-mail: kenney@ifa.au.dk (Received 27 March 2002, revised 12 June 2002, accepted 12 July 2002) Eur. J. Biochem. 269, 4159–4163 (2002) Ó FEBS 2002 doi:10.1046/j.1432-1033.2002.03112.x structure. The same is true of the spectra of B zone samples. Background (spider Ringer solution) reference spectra were recorded before and after each sample spectrum was recorded. Background-corrected spectra of the samples were analyzed using the SUPER3 algorithm [10], courtesy of B. Wallace (Birkbeck College, London, UK). This is the only published concentration independent method for analysing CD spectra and is the only one that could be used because the protein concentration of each sample is not precisely known. For other CD analysis algorithms to provide reliable results the protein concentration must be known to approximately 1%. The normalized standard deviation (NRMSD) [11] measure of the accuracy of the undiluted samples was worse (NRMSD undiluted >0.2)than for the diluted ones (0.1 < NRMSD diluted < 0.2); how- ever, the estimate of secondary structure content was similar to that of the diluted sample spectral analysis, again confirming the validity of the analysis of the diluted samples spectra. Their analyses showed a systematic variation (attributed to a systemic error in background correction) in the predicted secondary structure content (by 5–15%) from one sample to the next, however, the trends were identical in all cases; e.g. increasing b-sheet content with temperature for the A zone samples. The reproducibility of the spectra and the moderately low NMRSD suggest that, though indicative rather than exact, the secondary structure predictions are valid and support the proposition that the spidroin at the downstream end of the B zone is rich in b-sheet and that heating converts the A zone secreted spidroin to a b-sheet enriched state. Electron microscopy Major ampullate silk glands and ducts were dissected from Nephila edulis spiders, fixed in a modified Karnovsky fixative, embedded in Spurr’s resin and sectioned on a diamond knife [9]. The sections were stained with uranyl acetate and lead citrate. Images were recorded at 40 000 times on a Philips 400 transmission electron microscope. RESULTS AND DISCUSSION Circular dichroism shows a b sheet transition We used CD spectroscopy to characterize the molecular structure of spidroin [12] in the major ampullate silk gland of the spider Nephila Edulis and its conversion along the silk production pathway. This gland has two morphologically distinct zones. Upstream in the silk production pathway the A zone secretes material that makes the silk fibre core and downstream the B zone secretes a thin coating [9]. Analyses of the CD spectra show that there is a remarkable difference in the initial as well as temperature-induced refolding of the secondary structure of the protein extracted from two different zones in the silk gland (Fig. 1). At ambient temperature the solute extracted from the A zone has a stable structure poor in a helix and b sheet, while from the downstream end of the B zone the extracted sample, composed of a mixture of A- and B zone secretions, is primarily b sheet. As the temperature is increased the B zone protein sample secondary structure remains relat- ively stable. With heating, however, the A zone protein sample undergoes a dramatic refolding conversion at approximately 35 °Ctoanenrichedb-sheet state that persists to at least 70 °C. This b-sheet enriched state remains stable upon return to ambient temperature and is also seen to be irreversible by modulated differential scanning calori- metry. This shows that energy is required to convert the b-sheet poor state to the b-sheet rich state of spidroin. It is unlikely that the spider uses heat to facilitate the conversion, though it is interesting that the spiders were observed not to spin silk above 35 °C. The spiders take advantage of these two states to control the production, transportation and conversion of spidroin in the gland so that the enrichment for b sheet can take place just prior to spinning the silk fibre. These results illuminate three important aspects of the conversion of the soluble spidroin as it flows through the silk gland on its way to becoming an insoluble fibre: (a) there is a refolding of the structure to enrich for b-sheet structure; (b) that the refolding can be induced by adding energy (e.g. heat); and (c) that the enriched b-sheet state is stable and irreversible. Fibrils seen by electron microscopy and structural similarities to amyloids Moderated conversion to a stable b-sheet enriched state is also a signature of amyloid fibril formation [13] and other related disease-associated proteins, e.g. prions [14]. In the spider the conversion of silk solute to solid fibre occurs during the spinning process. Using electron microscopy (EM) we examined the silk gland and duct to determine the site of fibril formation. No fibrils could be seen in the EM of the luminal contents of the A zone part of the gland where the CD spectroscopy analysis shows that the protein exhibits a b-sheet poor structure. Fibrils, however, are found downstream of the B zone (Fig. 2), where the CD spectroscopy analysis shows that the extracted samples are rich in b sheet. This is close to the previously proposed site of b-sheet formation [15]. Here, the fibrils nucleate within the A zone secreted material. The fibrils are approximately 10 nm wide and are similar in aspect to amyloid, prion and other amyloidogenic fibrils, such as a-synucleins [16]. The appearance of fibrils at the end of the silk production pathway implies that a b-sheet enriching structural conver- sion of spidroin plays a role in their formation. Both amyloids and spider silk contain b sheets with antiparallel strands [6,13]. In the final dragline silk fibre spun by the spider the strands are found to be aligned with the fibre axis making a parallel-b structure [6], this should not be confused with a parallel-b sheet which refers to the relative (parallel) orientation of adjacent strands. On the other hand, in amyloid fibrils the strands are perpendicular to the fibril axis forming a cross-b structure [17]. However, when cross-b silk fibres are stretched they assume the parallel-b structure characteristic of Group 3 silk proteins (including spider silk) [6]. Therefore, it is possible that the fibrils we see appearing in the b sheet enriched region of the silk gland could have a cross-b structure (like amyloid protein fibrils) which is converted to the parallel-b structure observed in the final silk fibre product due to the extension of the fibre during the spinning process. This is supported by the observation that the orientation function (i.e. alignment of the strands with respect to the axis) increases with the speed that the silk is drawn from the spider [18]. Addition- ally, there are other structural similarities between amyloido- 4160 J. M. Kenney et al.(Eur. J. Biochem. 269) Ó FEBS 2002 genic and spider silk proteins. Specifically, there is a link between the prion octapeptide repeats and the spidroin Pro- Gly repeats which is evident in the similar structures that have been calculated for them in the context of prion fibrils and dragline silk, respectively. The octapeptide repeat has been found to have a polyproline Form II structure [19] while the Pro-Gly repeat is a 3 1 helix [20]. Both of these are left-handed helices with exactly three residues per turn. Sequence low complexity: spidroin and amyloids The similarity between spidroin and fibril-forming amyloido- genic proteins suggested by our CD and EM results was further investigated by amino-acid sequence analyses. We were first struck by the obvious low complexity (highly nonrandom with substantial regularities) [21] of the spi- droin 2 sequence, SPD2-NEPCL; repetitions of the peptide PGGYGPGQQG (a low complexity sequence itself) are interleaved with polyalanine stutters. When the SWISSPROT database (Version 40, as at March 15, 2002) was probed with SPD2-NEPCL using the SCANPS implementation of the Smith Waterman sequence alignment algorithm [22] many other low complexity sequences were returned. High on the list are the sequences of elastomers [23] such as collagen (e.g. CA1-BOVIN, CA24 °CANFA), elastin (ELS-RAT), silk-moth fibroin (FBOH-BOMMO), glutenin (GLT5-WHEAT) and mucin (MUC1-HUMAN). Some- what down the list is the trinucleotide disease protein DRPL-HUMAN (dentatorubral-pallidoluysian atrophy protein). The first of the prion sequences, PRIO-MUSVI, Fig. 1. Typical CD spectra and predicted molecular secondary structure of silk protein are shown. (A) CD spectra of A zone sample at 20 °C (black), 30 °C(blue),35°C(green),40 °C (orange), 45 °C(red)and55 °C (pink); (B) fraction of predicted secondary structure of A zone sample for a helix (solid), b sheet (dashed) and other (dotted); (C) CD spectra of downstream B zone sample at 20 °C(black),30°C (blue), 40 °C (orange) and 50 °C (red); (D) fraction of secondary structure of B zone sample for a helix (solid), b sheet (dashed) and other (dotted). Ó FEBS 2002 Amyloidogenic nature of spider silk (Eur. J. Biochem. 269) 4161 occurs at position 638 in the list (Pr ¼ 0.0032), while the avian prion protein, PRIO-CHICK, is at position 974 (Pr ¼ 0.068). However, it is inevitable that this search will also encompass proteins such as the fish antifreeze protein ANP-NOTCO, that match the polyalanine stutters, so an alternate search was undertaken using just the octapeptide repeats from PRIO-HUMAN. As expected, mammalian prion proteins fill the top of the list, but PRIO-CHICK is found at position 199 with a probability of 2.71. SPD2- NEPCL is at position 543. The extremely low probability scores occur because the query sequence is very short and, in general, the statistical model used in scoring alignments is better suited to globular proteins (which are predominantly high complexity sequences) rather than these structural proteins. The common theme among all of the proteins discussed above is that they contain low complexity peptide sequences. Interestingly, if a set of 32 known amyloidogenic sequences is examined using 0j.py [24], a tool specifically designed for demarcating low complexity protein sequences, we find that 20 (i.e. > 60%) have scores greater than or equal to 3 vs. 41 607 out of 104 939 for SwissProt as a whole, which gives rise to a one-sided binomial-distribution probability of 0.0075. In other words, the amyloidogenic sequences have significantly lower complexity than proteins generally. The set of sequences includes proteins such as SAA5-MESAU, amyloid protein A5 and SYUA-MOUSE, alpha-synuclein protein, both of which also appear in the list returned by the similarity search using SPD2-NECLP but not the prion proteins discussed above. We hypothesize therefore that low sequence complexity may be a property of many, if not most, fibril-forming proteins, including spidroin. Experimental evidence of such a link is already present in the literature. Proteins created from alternating polar and nonpolar residues were found to form fibril-like structures while at the same time being rare in nature [25]. In another study, water soluble peptides of the form DPKGDPKG-(VT) n -GKGDPKPD-NH 2 , n ¼ 3–8, were found to self assemble into fibril-like structures [26]. Both of these are examples of low complexity peptides. CONCLUSION Elucidating the details of the spidroin conversion process in the spider is vital to understanding silk fibre formation. Using CD spectroscopy, electron microscopy and sequence analysis we have revealed similarities in the nature of spider silk and amyloidogenic fibril formation. This suggests that spider-silk processing, not the least for the large quantities of material produced in vivo in the spider, may also promise to be beneficial for studying amyloid-like fibril formation in general. Significant applications could include the develop- ment of genetically modified spidroin to test hypotheses on the formation and inhibition of pathogenic fibrils. ACKNOWLEDGEMENTS We thank Karen Wise (Zoology, Oxford, UK) for assistance with EM; Cait MacPhee, Klaus Doering and Michael Gross (OCMS, Oxford, UK) for help with CD experimental design and interpretation; and Bonnie Wallace (Birkbeck College, London, UK) for guidance in quantitative analysis and the use of the SUPER3 algorithm. The work was funded in part by a grant from the Danish Natural Science Research Council and supported by the European Science Foundation. Michael Wise is supported by a grant from Bristol-Myers Squibb, which is gratefully acknowledged. REFERENCES 1. Vollrath, F. & Knight, D.P. (2001) Liquid crystalline spinning of spider silk. Nature 410, 541–548. 2. Simmons, A., Michal, C. & Jelinski, L. (1996) Molecular orientation and two-component nature of the crystalline fraction of spider dragline silk. Science 271, 84–87. 3. Gosline, J.M., Guerette, P.A., Ortlepp, C.S. & Savage, K.N. (1999) The mechanical design of spider silks: from fibroin sequence to mechanical function. J. Exp. Biol. 202, 3295–3303. 4. Takahashi, Y., Gehoh, M. & Yuzuriha, K. (1999) Structure refinement and diffuse streak scattering of silk (Bombyx mori). Int. J. Biol. Macromol. 24, 127–138. 5. Li, A., Zhou, P., Shao, Z., Xie, X., Chen, X., Wang, H., Chunyu, L. & Yu, T. (2001) The natural silk spinning process. Eur. J. Biochem. 268, 6600–6606. 6. Craig, C.L. (1997) Evolution of Arthropod silks. Annu. Rev. Entomol 42, 231–267. 7. Lazaris,A.,Arcidiacono,S.,Huang,Y.,Zhou,J F.,Duguay,F., Chretien, N., Welsh, E.A., Soares, J.W. & Karatzas, C.N. (2002) Spider silk fibers spun from soluble recombinant silk produced in mammalian cells. Science 295, 472–476. Fig. 2. EM of a transverse section of the contents of the third limb of the major ampullate duct of Nephila edulis just before the start of the internal draw down taper. The material in the B zone coat appears as a band of homogeneous material about 300 nm thick near the bottom of the figure and lying just inside the denser outer layer of the duct’s cuticle at the very bottom of the figure. The A zone (upper majority of the figure) shows an increasing density of tangled fibrils close to the junction with the B zone. The scale bar is 500 nm. 4162 J. M. Kenney et al.(Eur. J. Biochem. 269) Ó FEBS 2002 8. Schartau, W. & Leidersher, T. (1983) Composition of the hemo- lymph of the tarantula Eurypelma californicum. J. Comp. Physiol. B. Biochem. Sys. Environ. Physiol. 152, 73–77. 9. Vollrath, F. & Knight, D.P. (1999) Structure and function of the silk production pathway in the spider Nephila edulis. Int. J. Biol. Macromol. 24, 243–249. 10. Wallace, B.A. & Teeters, C.L. (1987) Differential absorption flattening optical effects are significant in the circular dichroism spectra of large membrane fragments. Biochemistry 26, 65–70. 11. Mao, D., Wachter, E. & Wallace, B.A. (1982) Folding of the H + -ATPase proteolipid in phospholipid vesicles. Biochemistry 21, 4960–4968. 12.Hijirida,D.H.,Do,K.G.,Michal,C.,Wong,S.,Zax,D.& Jelinski, L.W. (1996) 13 CNMRofNephila clavipes major ampull- ate silk gland. Biophys. J. 71, 3442–3447. 13. Dobson, C.M. (2001) The structural basis of protein folding and its links with human disease. Trans. R. Soc. Lond. B. Biol. Sci. 356, 133–145. 14. Jackson, G.S., Hosszu, L.L.P., Power, A., Hill, A.F., Kenney, J.M., Saibil, H., Craven, C.J., Waltho, J.P., Clarke, A.R. & Collinge, J. (1999) Reversible conversion of monomeric human prion protein between native and fibrilogenic conformations. Science 283, 1935–1937. 15. Knight, D.P., Knight, M.M. & Vollrath, F. (2000) Beta transition and stress-induced phase separation in the spinning of spider dragline silk. Int. J. Biol. Macromol. 27, 205–210. 16. Jensen,P.H.,Islam,K.,Kenney,J.M.,Nielsen,M.S.,Power,J.& Gai, W.P. (2000) Microtubule-associated Protein 1B is a compo- nent of cortical lewy bodies and binds a-synuclein filaments. J. Biol. Chem. 275, 21500–21507. 17. Sunde, M., Serpell, L.C., Bartlam, M., Fraser, P.E., Pepys, M.B. & Blake, C.C. (1997) Common core structure of amyloid fibrils by synchrotron X-ray diffraction. J. Mol. Biol. 273, 729– 739. 18. Riekel, C., Madsen, B., Knight, D. & Vollrath, F. (2000) X-ray diffraction on spider silk during controlled extrusion under a synchrotron radiation X-ray beam. Biomacromolecules 1, 622–626. 19. Smith, C.J., Drake, A.F., Banfield, B.A., Bloomberg, M.S., Clarke, A.R. & Collinge, J. (1997) Conformational properties of the prion octa-repeat and hydrophobic sequences. FEBS Lett. 405, 378–384. 20. van Beek, J., Ku ¨ mmerlen, D.K., Vollrath, F. & Meier, B.H. (1999) Solid state NMR on supercontracted spider dragline silk. Int. J. Biol. Macromol. 24, 173–178. 21. Wooton, J.C. (1994) Sequences with ÔUnusualÕ Amino Acid Compositions. Curr. Opin. Struct. Biol. 14, 413–421. 22. Barton, G.J. (1993) An efficient algorithm to locate all locally optimal alignments between two sequences allowing for gaps. Computer Appl. Biol. Sci. 9, 729–734. 23. Tatham, A.S. & Shewry, P.R. (2000) Elastomeric proteins: Biological roles, structures and mechanisms. Biochem. Sci. 25, 567–571. 24. Wise, M.J. (2001) 0jpy: a Software Tool for Low Complexity Proteins and Protein Domains. Ninth International Conference on Intelligent Systems in Molecular Biology (ISMB¢01), Supple- mentary Issue of Bioinformatics, 17. S288–295. 25. Broome, B.M. & Hecht, M.H. (2000) Nature disfavors sequences of alternating polar and non-polar amino acids: Implications for amyloidogenesis. J. Mol. Biol. 296, 961–968. 26. Janek, K., Behlke, J., Zipper, J., Fabian, H., Georgalis, Y., Beyermann, M., Bienert, M. & Krause, E. (1999) Water-soluble b sheet models which self-assemble into fibrillar structures. Biochem. 38, 8246–8252. Ó FEBS 2002 Amyloidogenic nature of spider silk (Eur. J. Biochem. 269) 4163 . fraction of secondary structure of B zone sample for a helix (solid), b sheet (dashed) and other (dotted). Ó FEBS 2002 Amyloidogenic nature of spider silk. of study; however, little is known about the conversion process in the silk- producing gland of the spider. Here we describe a study of the conversion of