Report

Amyloid Fibrils of the HET-s(218–289) Prion Form a β Solenoid with a Triangular Hydrophobic Core

See allHide authors and affiliations

Science  14 Mar 2008:
Vol. 319, Issue 5869, pp. 1523-1526
DOI: 10.1126/science.1151839

Abstract

Prion and nonprion forms of proteins are believed to differ solely in their three-dimensional structure, which is therefore of paramount importance for the prion function. However, no atomic-resolution structure of the fibrillar state that is likely infectious has been reported to date. We present a structural model based on solid-state nuclear magnetic resonance restraints for amyloid fibrils from the prion-forming domain (residues 218 to 289) of the HET-s protein from the filamentous fungus Podospora anserina. On the basis of 134 intra- and intermolecular experimental distance restraints, we find that HET-s(218–289) forms a left-handed β solenoid, with each molecule forming two helical windings, a compact hydrophobic core, at least 23 hydrogen bonds, three salt bridges, and two asparagine ladders. The structure is likely to have broad implications for understanding the infectious amyloid state.

Prions are infectious proteins capable of self-replicating their conformation and are best known as the agent of diseases such as scrapie in sheep (1), bovine spongiform encephalopathy in cattle (2), and a new variant of Creutzfeldt-Jakob disease in humans (3). Prions have also been described in yeast and filamentous fungi (4). The infectious form of prions has been characterized as a β sheet–rich molecular aggregate termed an amyloid fibril (5, 6).

HET-s is a protein of the filamentous fungus Podospora anserina. In its prion form, HET-s plays a role in heterokaryon incompatibility, a fungal self/nonself recognition phenomenon that prevents different forms of parasitism. The proteinase K–resistant core of the prion fibrils formed by the C-terminal residues 218 to 289 is unstructured in solution and forms infectious fibrils in vitro (7). Earlier work showed that HET-s(218–289) in its fibrillar state consists of four β strands forming two windings of a β solenoid (8). However, it gave no information about the intermolecular β-sheet propagation (parallel or antiparallel), and the model was not based on distance restraints.

We used solid-state nuclear magnetic resonance (NMR) to determine the structure of the rigid core of HET-s(218–289) fibrils. Previous work on amyloid fibrils by NMR has shown that structural information can be obtained from these noncrystalline entities (914). Resonance assignment of HET-s(218–289) solid-state NMR spectra have been described previously (15, 16). We determine a large number of distance restraints from uniformly labeled samples and specifically identify purely intra- and intermolecular restraints by using differently labeled samples. The 1H-1H and 13C-13C internuclear distance restraints are derived from 13C-detected proton-spin diffusion [carbon-proton-proton-carbon (CHHC) and nitrogen-proton-proton-carbon (NHHC) experiments (17, 18)] and from proton-driven 13C spin diffusion (PDSD) (19), respectively (for a list of experiments, see table S1). Part of a CHHC spectrum from a uniformly isotopically labeled (U-[13C, 15N]) sample is shown in Fig. 1A as a representative example. From this spectrum, 41 structurally meaningful restraints were obtained (table S2b). These are translated to upper distance restraints of 3.5 and 4.5 Å, for strong and weak cross peaks, respectively. Sixty-two distance restraints (table S2b) were identified from a PDSD spectrum of uniformly 2-13C–labeled fibrils and translated to upper distance limits of 7.5 Å [2-13C glycerol was used as a carbon source (20, 21) in this sample to reduce spectral overlap, line width, and relayed spin-diffusion effects].

Fig. 1.

Solid-state NMR spectra of HET-s(218–289) fibrils. Blue labels identify signals that correspond to short-range restraints (|ij| ≤ 1); red labels, signals that correspond to register-restraints (35 ≤ | ij | ≤ 37); and green labels, restraints that define the hydrophobic core. (A) CHHC spectrum of U-[13C, 15N]–labeled HET-s(218–289) fibrils recorded at 19-kHz magic-angle spinning (MAS) with 150-μs 1H–1H mixing with register and side-chain contacts useful for structure calculation. (B) PDSD spectrum of fibrils containing 25% 2-13C–labeled monomers diluted in unlabeled HET-s(218–289), recorded at 10-kHz MAS with 250-ms 13C–13C mixing giving intramolecular contacts. (C) PDSD spectrum of fibrils containing 25% 1,3-13C–labeled molecules diluted in unlabeled HET-s(218–289), recorded at 19-kHz MAS with 500-ms 13C–13C mixing giving further intramolecular contacts. (D) NHHC spectrum of a U-[13C, 15N]–labeled sample recorded at 19-kHz MAS with 150-μs 1H–1H mixing. The expected peak positions calculated for proton pairs separated by less than 3.5 Å in the lowest-energy structure are indicated by colored crosses. For a β sheet, three types of HNi-Hαj contacts are mainly expected: intraresidue (j = i), sequential (j = i – 1), and interstrand register (|ji| = 36, intra- and intermolecular) peaks are indicated with blue, orange, and red symbols, respectively. The green crosses indicate the positions of other expected peaks. (E) NHHC spectrum of fibrils containing a 1:1 mixture of U-13C– and U-15N–labeled monomers, recorded at 9.5-kHz MAS with 150-μs 1H–1H mixing. Only intermolecular proton contacts give rise to peaks in this spectrum.

In fibrils with uniform isotope labeling, distance restraints measured by NMR are difficult to assign to either intra- or intermolecular contacts. A comparison of spectra from uniformly labeled samples with those from “diluted samples,” in which isotopically labeled monomers are mixed with unlabeled material before fibrilization (ratio 1:2.5), can in some cases resolve the ambiguities. Following this strategy, we identified a total of 30 intramolecular restraints from a CHHC spectrum on a diluted U-[13C, 15N]–labeled HET-s(218–289) sample and PDSD spectra on diluted extensively labeled HET-s(218–289) (Fig. 1, B and C) (20, 21). Intermolecular restraints were obtained from a sample fibrilized from a mixture of U-[13C]–labeled and U-[15N]–labeled HET-s(218–289) molecules (22). Polarization transfer between 15N- and 13C-bound protons, respectively (17) (Fig. 1E), indicates short HN–Hα contacts (∼3.0 Å) between β sheets of different monomers and thus selectively characterizes the intermolecular interface. Comparing these spectra to similar spectra on uniformly labeled compounds allowed us to identify seven H bonds as intramolecular and seven H bonds as intermolecular, and nine H bonds remained ambiguous (intra- or intermolecular). Details are given in the supporting online material (23). These H bonds define a parallel in-register arrangement of the intra- and intermolecular β-sheet interfaces, which is supported by 69 of the experimental restraints that correlate residues i and i + (36 ± 1) (Fig. 1 and fig. S1). Fourteen of these can be identified unequivocally as intramolecular.

All restraints used for the structure calculation are summarized in Table 1 (a comprehensive list of cross peaks is given in table S2). In total, 90 13C–13C and 44 1H–1H distance restraints (i.e., 2.8 per assigned residue, all of them containing nontrivial structural information) were identified, together with the 23 β-sheet H bonds (fig. S2). In addition, 74 angle restraints obtained by TALOS (24) were used (fig. S2). From NMR (8) and mass-per-length measurements (25), it is known that the thinnest HET-s(218–289) fibril consists of a stack of single molecules all having the same structure. To implement the resulting quasi–one-dimensional (1D) symmetry, we used 206 additional intermolecular distance restraints (fig. S2). The NMR-structure calculation was performed with CYANA (26) using the restraints of Table 1 on a set of seven molecules, and yielded the structure depicted in Fig. 2. The backbone heavy-atom average root mean square deviation to the mean structure of the 20 lowest-energy conformers (200 calculated structures) is 0.4 Å for the backbone and 1.0 Å for all heavy atoms, considering only the rigid core of one HET-s(218–289) molecule (residues N226 to G242, N262 to G278) (Fig. 2 and fig. S3).

Fig. 2.

Structure of the HET-s(218–289) fibrils. The fibril axis is indicated by an arrow. (A) Side view of the five central molecules of the lowest-energy structure of the HET-s(218–289) heptamer calculated from the NMR restraints. (B) Top view of the central molecule from (A). β3 and β4 lie on top of β1 and β2, respectively. A view orthogonal to the fibril axis is given in fig. S7. (C) NMR bundle: superposition on residues N226 to G242, N262 to G278 of the 20 lowest-energy structures of a total of 200 calculated HET-s(218–289) structures. Only the central molecule of the heptamer is shown. (D) Representation of the well-defined central core of the fibril (N226 to G242, N262 to G278). Hydrophobic residues are colored white, acidic residues red, basic residues blue, and others green (lowest-energy structure). (E and F) Schematic representations of the two windings in (D): the first winding [N226 to G242, displayed in (E)] of the β solenoid is located beneath the second one [N262 to G278, displayed in (F)]. Abbreviations for the amino acid residues are as follows: A, Ala; C, Cys; D, Asp; E, Glu; F, Phe; G, Gly; H, His; I, Ile; K, Lys; L, Leu; M, Met; N, Asn; P, Pro; Q, Gln; R, Arg; S, Ser; T, Thr; V, Val; W, Trp; and Y, Tyr.

Table 1.

Number of structural restraints used for the structure calculation (per monomer).

View this table:

The overall organization of a HET-s(218–289) fibril is a left-handed β solenoid with two windings per molecule (Fig. 2A). The core of the fibril is defined by three β strands per winding (six β strands per molecule) that form continuous in-register parallel β sheets. An additional β sheet outside the core is formed by β2b and β4b (Fig. 2B). This organization is consistent with the fold proposed earlier (8), but the details show that each previously proposed β strand is split into two shorter segments (e.g., β1 into β1a and β1b). The segments β1a and β1b (β3a and β3b) are connected by a two-residue β arc, changing the inside-outside pattern of side chains and leading to an approximately rectangular kink in the strand at K229 and E265, for β1 and β3, respectively (Fig. 2B). The connection between β1b and β2a (and similarly, between β3b and β4a) is provided by a three-residue β arc, allowing for the orientation change of the polypeptide backbone by ∼150°. A disruption of the β-sheet pattern is also observed between β2a and β2b (β4a and β4b), leading to ∼90° β arcs. β1-β2 and β3-β4 are pseudo-repeats and form both parallel intramolecular and intermolecular H bonds, as follows: β1a-β3a, β1b-β3b, β2a-β4a, β2b-β4b (Fig. 2, A and D).

As seen in Fig. 2, D to F, the β-sheet arrangement is stabilized by favorable side-chain contacts. The first three β strands of each pseudo-repeat enclose a triangular hydrophobic core that is tightly packed and contains almost exclusively hydrophobic residues (Ala, Leu, Ile, and Val) with numerous experimental restraints (indicated by green labels in the spectra of Fig. 1) between hydrophobic side chains. The packing is dense and defines a dry interface. The only polar residues in the core are T233 and S273, which can form side-chain H bonds, thereby stabilizing the formation of the turn between β1 and β2. In contrast, all charged residues face outside and are mostly located in β-arc regions, where the solvent accessibility is high. Three of them are arranged on top of each other such that charges are compensated (Fig. 2, D to F) and the formation of salt-bridges becomes possible. Several experimental restraints support the existence of the salt-bridges K229-E265, E234-K270, and R236-E272 (fig. S4). Because the stacking is parallel, we expect the charge compensation to have both intra- and intermolecular character. This may explain why fibrils have high stability against denaturation by nonionic urea at neutral pH, but are destabilized by urea at acidic or basic pH (27). The two asparagines next to the hydrophobic core are stacked and can form a ladder (N226-N262), further contributing to the fibril stability through side-chain H bonds. Another asparagine ladder can be formed outside the hydrophobic core (N243-N279) (28).

For the structure calculation, only a limited number of well-resolved NMR peaks was used. To assess whether the structure is consistent with all the peaks observed in the spectra, we calculated the expected cross peaks from the internuclear distances in the structure. As an example, all expected resonances in the NHHC spectrum are shown as symbols in Fig. 1D. Similar good agreement was found for the other spectra (figs. S5 and S6). In our structure, the cross-section of the fibril is approximately circular (Fig. 2C), a feature which might explain why no helical twist of HET-s(218–289) fibrils has been detected in electron micrographs (25). The structure given in Fig. 2 explains all the details of the chemical-shift data, H/D exchange, water accessibility, and mutant studies by Ritter et al. (8), even the ones that could not be explained by the earlier straight-stranded fold (8). The non–β-sheet chemical shift of K229 and E265 residues, together with their fast H/D exchange rates, can now be explained by the β arc at this position. Residues D230 and T266, which were found to be highly solvent-accessible as probed by chemical cross-linking of cysteine mutants (8), are indeed solvent-exposed in our structure (Fig. 2, D to F), whereas I231, T233, V239, L241, V267, and V275 are buried in the hydrophobic core of the fibrils, explaining their very low water accessibility. G242, N279 and E280, for which intermediate values were found, are located in a region of β2b and β4b where the NMR structure is less well defined and where both sides of the β strand are somewhat protected (Fig. 2C). The present model also explains the requirement for a minimal length of the loop connecting β2b and β3a to maintain infectivity (8).

The structure of HET-s(218–289) shows an overall β-helical fold that is of higher structural complexity than that of peptide fibrils (29, 30)—a complexity that is reminiscent of soluble protein folds. Part of the complexity, e.g., the favorable alternation of positive and negative charges in ladders along the fibril axis, can be realized only because of the pseudo-repeat (β1/β3 and β2/β4), leading to a structure in which one molecule forms two turns of the solenoid. (31). This feature distinguishes HET-s from other amyloids and prions that have also been modeled by β solenoids (32, 33). It leads to the formation of three salt-bridges that stabilize the structure, in contrast to the finding in yeast prion protein Sup35 where the presence of pseudo-repeats is probably related to structural variability and the existence of prion strains (31). The three-stranded triangular hydrophobic core indeed bears some resemblance to β-solenoid structures of soluble proteins like filamentous hemagglutinin (34) and the P22 tailspike protein (35). In contrast to HET-s(218–289), these structures are not periodic, but the geometry of the triangular core is quite similar. Furthermore, a β-solenoid fold has also been proposed for the prion state of the human prion protein PrP on the basis of modeling and electron microscopy (36) and for the yeast prion Sup35 (37, 38).

The well-organized structure of the HET-s prion fibrils can explain the extraordinarily high order in these fibrils, as seen by NMR, as well as the absence of polymorphism caused by different underlying molecular structures at physiological pH conditions, because the specific nature of the interactions in the fibril excludes polymorphic molecular conformations with comparable stability. The fibril structure of HET-s(218–289) exemplifies the well-defined structure of a functional amyloid and illustrates the interactions that can stabilize their fold (39).

Supporting Online Material

www.sciencemag.org/cgi/content/full/319/5869/1523/DC1

Materials and Methods

Figs. S1 to S7

Tables S1 to S3

References

References and Notes

View Abstract

Navigate This Article