Analytical Biochemistry 385 (2009) 374–376

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/ locate /yabio

Notes & Tips

Amyloid fibril dynamics revealed by combined hydrogen/deuterium exchangeand nuclear magnetic resonance

Anders Olofsson *, A. Elisabeth Sauer-Eriksson, Anders Öhman *

Umeå Center for Molecular Pathogenesis, Umeå University, SE-901 87 Umeå, Sweden

a r t i c l e i n f o a b s t r a c t

Article history:Received 22 August 2008Available online 5 November 2008

0003-2697/$ - see front matter � 2008 Elsevier Inc. Adoi:10.1016/j.ab.2008.10.034

* Corresponding authors. Fax: +46 90 786 5944 (A.E-mail addresses: anders.olofsson@ucmp.umu.se (

ucmp.umu.se (A. Öhman).1 Abbreviations used: NMR, nuclear magnetic resonan

Ab, amyloid–b peptide; AFM, atomic force microscopyisopropanol; HSQC, heteronuclear single quantum corr

A general method to explore the dynamic nature of amyloid fibrils is described, combining hydrogen/deuterium exchange and nuclear magnetic resonance spectroscopy to determine the exchange rates ofindividual amide protons within an amyloid fibril. Our method was applied to fibrils formed by theamyloid-b(1–40) peptide, the major protein component of amyloid plaques in Alzheimer’s disease. Thefastest exchange rates were detected among the first 14 residues of the peptide, a stretch known to bepoorly structured within the fibril. Considerably slower exchange rates were observed in the remainderof the peptide within the b-strand-turn-b-strand motif that constitutes the fibrillar core.

� 2008 Elsevier Inc. All rights reserved.

Protein deposits in the form of amyloid fibrils are characteristicof a group of severe disorders in which Alzheimer’s and Creutz-feldt–Jakob’s diseases are among the most well-known examples[1]. Due to their size, insolubility, and noncrystalline nature, thestructural and dynamic characteristics of amyloid fibrils have beendifficult to elucidate using traditional methods such as crystal dif-fraction and solution nuclear magnetic resonance (NMR)1 spectros-copy. Solid-state NMR has revealed important structural details offibrils from several amyloidogenic proteins and has shown that theyform a general structure in which b-strands assemble perpendicularto the fibril axis, forming b-sheets along the fibril [2]. More recently,the combined use of quenched hydrogen/deuterium (H/D) exchangeand solution NMR spectroscopy has proven to be extremely valuablein investigating the structural characteristics of amyloid fibrils at aresidue-specific level [3–5] and can even be used for determiningthe three-dimensional structure if combined with pairwise muta-genesis [6]. Several other methods have also been used to elucidatethe static fibrillar structural organization [7–10]. However, there arefew experimental procedures suitable for investigating the dynamicbehavior of amyloid fibrils [3,10,11].

We previously described [4] and applied [12–14] a quenched H/D exchange NMR method that identifies labile amide protonstrapped within the core region of a fibril by secondary structureor solvent exclusion. The observed solvent protection patterntypically appears as a characteristic bell shape for each b-strand

ll rights reserved.

Öhman).A. Olofsson), anders.ohman@

ce; H/D, hydrogen/deuterium;; HFIP, 1,1,1,3,3,3-hexafluoro-elation.

involved in the fibril core and is acquired at a specific incubationtime in D2O, thereby providing crucial information about the staticstructural organization of the fibril. By varying the incubationtimes, here we demonstrate that the exchange rates of individualamide protons within an amyloid fibril are accessible. The currentreport describes a general method for experimentally determiningthe residue-specific exchange rates, providing unique informationabout the dynamic properties of the fibril.

The approach begins with the formation of fibrils in aqueoussolution under the conditions of choice. The fibrils are recoveredand placed in a D2O solution for a designated incubation time,TH/D. The H/D exchange reaction is terminated by freezing therecovered fibrils in liquid nitrogen, followed by a rapid conversionof the fibrils into a monomeric, NMR-detectable state under condi-tions of low back exchange. Unlike other techniques [3,5], ourmethod is not restricted to one dissolution agent; any solventcapable of dissolving the fibrils may be used. Labile amide protonsthat are protected from exchange by the secondary structure orexcluded from solvent interactions within the core of the fibrilare now trapped in the monomeric state. In an aprotic solvent,the degree of protection may be directly determined from the ratiobetween the NMR cross-peak intensities of each amide proton inthe H/D-exchanged sample and a protonated reference sample. Inpractice, however, the solvents used for fibril dissolution usuallycontain an excess of deuterons, resulting in a continued H/Dexchange within the monomeric state that appears as a post-trapdecay in signal intensity over time. To obtain the residue-specificprotection in the fibril state, the post-trap exchange must befollowed by a series of NMR spectra and the signal intensities mustbe fitted to a single exponential decay to determine the intensity attime zero (I0), that is, the level of protection in the fibril immedi-ately prior to dissolution. The calculated initial intensities I0 are

Notes & Tips / Anal. Biochem. 385 (2009) 374–376 375

then fitted to a single exponential decay curve as a function of D2Oexposure time (TH/D) to obtain the exchange rates for individualamide protons within the fibril.

The applicability of our method is demonstrated in a study ofthe amyloid-b peptide (Ab), the main protein component of theplaques found in patients with the neurodegenerative disorderAlzheimer’s disease. Ab peptides are peptide fragments 39 to43 residues in length derived from the proteolytic cleavage ofthe significantly larger amyloid precursor protein. Of these,Ab(1–40) and Ab(1–42) are the most abundant fragments inamyloid plaques. We focused on Ab(1–40), which has been thor-oughly investigated in terms of its structural properties. It formsa fibril through the perpendicular stacking of peptides along thefibril axis via a parallel in-register assembly of a hairpin-like b-strand-turn-b-strand motif, creating a solvent-protected core oftwo intermolecular b-sheets [15]. We previously characterizedthe solvent protection of amide protons within Ab(1–40) fibrils[14], finding two protected bell-shaped regions with a particu-larly high protection (�90%) for residues Leu17 to Gly25 andAla30 to Val36, consistent with the known structural arrange-ment where two b-strands are connected with a turn involvingresidues Ser26 to Asn27. In the current study, this static pictureof the fibrillar solvent protection is extended to include thedynamic properties of the Ab(1–40) fibril.

Ab(1–40) fibrils were formed by incubating 15N-labeled Ab(1–40) (AlexoTech, Umeå, Sweden) in 10 mM acetate buffer (pH 5.0)and 150 mM NaCl at 37 �C for 10 to 14 days with agitation at130 rpm. The resulting gel-like solution contained the desired fi-brils as verified by atomic force microscopy (AFM) (Fig. 1A). Thissample was dispensed into seven tubes and subjected to H/D ex-change (TH/D) for 0 (reference), 2, 5, 10, 20, 120, or 720 min in10 mM acetate buffer (pD 4.6) and 150 mM NaCl at 37 �C. Duringthe final stage of the incubation time, the fibrils were pelletedvia centrifugation and frozen in liquid nitrogen, effectively termi-nating the H/D exchange. Dissolution of the fibrils into monomerswas carried out under previously optimized conditions [13,14](1,1,1,3,3,3-hexafluoroisopropanol [HFIP]/D2O, 80:20, pD 2.6,150 mM NaCl) in which the remaining protons are protected byan induced a-helical structure [16]. The pD of 2.6 and temperatureof 15 �C minimize the exchange of amide protons. For each TH/D, aseries of well-resolved two-dimensional 15N heteronuclear singlequantum correlation (HSQC) spectra were recorded (see examplesin Fig. 1B, D and F). The post-trap decays for each residue were

Fig. 1. H/D exchange analysis of Ab(1–40) fibrils. (A) Tapping mode AFM image showing tfrom a selected region of 15N HSQC NMR spectra of Ab(1–40) monomers recorded 7 min aand 720 min for panels B, D, and F, respectively. Assignments are indicated in panel B. (Cand G, respectively, relative to the reference sample (TH/D = 0 min), visualized onto a m[14,15], using MOLMOL [16]. The color code varies between the following extremes: blurate available are depicted in gray. (H) Initial signal intensities (I0) as a function of TH/D fotheir fibrillar H/D exchange rates (continuous and broken lines, respectively). The insetexperimental uncertainty of the measurements.

fitted to a single exponential decay to obtain the initial signalintensity at time zero (I0) for each TH/D. To obtain the residue-spe-cific exchange rates of the fibrillar species, the initial signal inten-sities were plotted as a function of TH/D and fitted to a singleexponential decay, as demonstrated in the examples of Fig. 1H. Thisdecay is further illustrated by mapping I0 for three TH/D on theAb(1–40) fibril model in Fig. 1C, E, and G. Exchange rates were deter-mined for 34 of 39 observable residues in the Ab(1–40) fibrils, andthey display a large variation from 0.0054 to 0.21 min�1. The resultsare depicted in Fig. 2A and are mapped on a model of the Ab(1–40)fibril in Fig. 2B. Most of the N-terminal residues (Asp1–His14) are fastexchanging, as is expected because they have previously beenshown to have exposed solvent-accessible positions within thefibril. Residues Gln15 to Gly25 display significantly slower andmore uniform exchange rates (0.0054–0.0095 min�1) and corre-spond to the b-strand stretch in the most well-protected b-sheetcore of the fibril. The b-turn region, residues Ser26 to Lys28, hasfaster exchange rates (0.011–0.019 min�1), in agreement with theirposition in the less solvent-protected edge of the b-sheet. Forresidues in the C-terminal b-strand core region, Gly29 to Val40,slow but rather dispersed exchange rates (0.0056–0.014 min�1)are observed. Clearly, our results agree well with what is expectedfrom the protected b-strand-turn-b-strand conformation foundwithin Ab(1–40) fibrils [15].

The presented method also has the ability to discriminate be-tween a simple and a more complex exchange mechanism of thefibril. The Ab(1–40) fibrils studied here clearly have a complex ex-change behavior given that most residues (also among the N-ter-minal residues Asp1–His14) have a signal intensity decay thatreaches a positive plateau value at infinite time (Iinf) (see Fig. 1H)and hence a considerable degree of residual intensity (Iinf/I0), thatis, a high protection ratio (see Fig. 2C). A similar exchange behaviorwas observed previously [11] and can be attributed to an exchangemechanism such as the molecular recycling model [10] or a struc-tural heterogeneity of the amyloid fibril [11]. Although the protec-tion pattern (Fig. 2C) is very similar to our previous studyperformed at pH 7.0 [14], the observed ratios at pH 5.0 are reduced.Because acidic pH is known to increase the aggregation rate ofAb(1–40) and modify the resulting fibril assembly [17], a likelyexplanation for this result is that peptides form a more heteroge-neous assembly at pH 5.0.

In summary, we have described a general quenched H/Dexchange NMR method that determines residue-specific exchange

he morphology of Ab(1–40) fibrils. The scale bar is 0.5 lm. (B, D and F) Contour plotsfter fibril dissolution. The fibrillar H/D exchange incubation times (TH/D) were 0, 10,

, E and G) Degree of exchanged protons at TH/D of 0, 10, and 720 min for panels C, E,onomer of the cross-b unit taken from a cross section of an Ab(1–40) fibril modele for no exchange and red for completed H/D exchange. Residues with no exchanger Gln15 (open circles) and Ser26 (filled triangles) and fitted functions to determineimage shows an enlarged view of the first five data points. Error bars indicate the

Fig. 2. Exchange rates within Ab(1–40) fibrils. (A) Determined residue-specificexchange rates within Ab(1–40) fibrils plotted against the residue name andnumber of Ab(1–40). Crosses (�) represent residues with exchange rates that aretoo fast to be detected. Error bars indicate the experimental uncertainty of themeasurements. (B) Obtained exchange rates in Ab(1–40) fibrils are mapped onto amonomer cross-b unit taken from a cross section of an Ab(1–40) fibril model [14,15]using MOLMOL [16]. The color code varies between the following extremes: blue forslowest and red for fastest exchange rates. Residues with no exchange rate availableare depicted in gray. (C) Determined residue-specific protection ratios at infinitetime (Iinf/I0) within Ab(1–40) fibrils plotted against the residue name and number ofAb(1–40).

376 Notes & Tips / Anal. Biochem. 385 (2009) 374–376

rates within an amyloid fibril and that thereby provides detailedinformation about the fibril’s dynamic properties. The methodwas successfully applied to Ab(1–40) fibrils, and details of the ex-change rates of these fibrils were presented. With its applicabilityto fibrils, and potentially to other protein aggregates and com-plexes, this method is expected to be particularly useful withinthe field of amyloid research.

Acknowledgments

This work was supported by the Magn. Bergvalls Foundation,Carl Trygger Foundation, Alzheimerfonden, Socialstyrelsen, Insam-lingsstiftelsen at Umeå University, Bernhard och Signe BäckströmsStiftelse, medical faculty at Umeå University, Hjärnfonden, Åke Wi-bergs Foundation, Göran Gustafssons Foundation, Swedish Re-search Science Council, and FAMY/AMYL patients association.

Appendix A. Supplementary data

Supplementary data associated with this article can be found, inthe online version, at doi:10.1016/j.ab.2008.10.034.

References

[1] F. Chiti, C.M. Dobson, Protein misfolding, functional amyloid, and humandisease, Annu. Rev. Biochem. 75 (2006) 333–366.

[2] R. Tycko, Molecular structure of amyloid fibrils: Insights from solid-state NMR,Q. Rev. Biophys. 39 (2006) 1–55.

[3] S.T. Alexandrescu, An NMR-based quenched hydrogen exchange investigationof model amyloid fibrils formed by cold shock protein A, Pac. Symp.Biocomput. 6 (2001) 67–78.

[4] J.H. Ippel, A. Olofsson, J. Schleucher, E. Lundgren, S.S. Wijmenga, Probingsolvent accessibility of amyloid fibrils by solution NMR spectroscopy, Proc.Natl. Acad. Sci. USA 99 (2002) 8648–8653.

[5] M. Hoshino, H. Katou, Y. Hagihara, K. Hasegawa, H. Naiki, Y. Goto, Mapping thecore of the b2-microglobulin amyloid fibril by H/D exchange, Nat. Struct. Biol. 9(2002) 332–336.

[6] T. Lührs, C. Ritter, M. Adrian, D. Riek-Loher, B. Bohrmann, H. Döbeli, D.Schubert, R. Riek, 3D structure of Alzheimer’s amyloid-b(1–42) fibrils, Proc.Natl. Acad. Sci. USA 102 (2005) 17342–17347.

[7] M. Sunde, C. Blake, The structure of amyloid fibrils by electron microscopy andX-ray diffraction, Adv. Protein Chem. 50 (1997) 123–159.

[8] O.S. Makin, L.C. Serpell, Structures for amyloid fibrils, FEBS J. 272 (2005) 5950–5961.

[9] I. Kheterpal, R. Wetzel, Hydrogen/Deuterium exchange massspectrometry: A window into amyloid structure, Acc. Chem. Res. 39(2006) 584–593.

[10] N. Carulla, G.L. Caddy, D.R. Hall, J. Zurdo, M. Gairí, M. Feliz, E. Giralt, C.V.Robinson, C.M. Dobson, Molecular recycling within amyloid fibrils, Nature 436(2005) 554–558.

[11] K. Yamaguchi, H. Katou, M. Hoshino, K. Hasegawa, H. Naiki, Y. Goto, Core andheterogeneity of b2-microglobulin amyloid fibrils as revealed by H/Dexchange, J. Mol. Biol. 338 (2004) 559–571.

[12] A. Olofsson, J.H. Ippel, S.S. Wijmenga, E. Lundgren, A. Öhman, Probing solventaccessibility of transthyretin amyloid by solution NMR spectroscopy, J. Biol.Chem. 279 (2004) 5699–5707.

[13] A. Olofsson, A.E. Sauer-Eriksson, A. Öhman, The solvent protection ofAlzheimer amyloid-b(1–42) fibrils as determined by solution NMRspectroscopy, J. Biol. Chem. 281 (2006) 477–483.

[14] A. Olofsson, M. Lindhagen-Persson, A.E. Sauer-Eriksson, A. Öhman, Amidesolvent protection analysis demonstrates that amyloid-b(1–40) and amyloid-b(1–42) form different fibrillar structures under identical conditions, Biochem.J. 404 (2007) 63–70.

[15] A.T. Petkova, W.M. Yau, R. Tycko, Experimental constraints on quaternarystructure in Alzheimer’s b-amyloid fibrils, Biochemistry 45 (2006) 498–512.

[16] R. Koradi, M. Billeter, K. Wüthrich, MOLMOL: a program for display andanalysis of macromolecular structures, J. Mol. Graph. 14 (1996) 51–55.

[17] S.J. Wood, B. Maleeff, T. Hart, R. Wetzel, Physical, morphological, andfunctional differences between pH 5.8 and 7.4 aggregates of the Alzheimer’samyloid peptide Ab, J. Mol. Biol. 256 (1996) 870–877.