Cryo-EM structure of an amyloid fibril from systemic amyloidosis

Systemic AA amyloidosis is a worldwide occurring disease of humans and animals that arises from the misfolding of serum amyloid A protein. To provide insights into the molecular basis of this disease we used electron cryo-microscopy and determined the structure of an ex vivo amyloid fibril purified from AA amyloidotic mice at 3.0 Å resolution. The fibril consists of C-terminally truncated serum amyloid A protein arranged into a compactly folded all-β conformation. The structure identifies the protein N-terminus as central for the assembly of this fibril and provides a mechanism for its prion-like replication. Our data further explain how amino acid substitutions within the tightly packed fibril core can lead to amyloid resistance in vivo.

truncated serum amyloid A protein arranged into a compactly folded all-β conformation. The structure identifies the protein N-terminus as central for the assembly of this fibril and provides a mechanism for its prion-like replication. Our data further explain how amino acid substitutions within the tightly packed fibril core can lead to amyloid resistance in vivo. 5

Main Text:
Systemic AA amyloidosis is a major form of systemic amyloidosis that arises from the formation of amyloid fibrils from serum amyloid A (SAA) protein (1)(2)(3). The massive deposition of these fibrils in organs like spleen, liver and kidneys physically distorts and impairs the affected tissues (2,4). In addition, there can be toxic effects of oligomeric fibrillation intermediates, similar 10 to other amyloid diseases (5). Current treatments, which aim to reduce the level of SAA in the blood, are unable to control the disease in all cases, but therapies that directly target SAA misfolding are not available (3). Systemic AA amyloidosis has been a key model for understanding pathological protein misfolding, and its existence in animals has given rise to uniquely relevant mouse and other animal models (6). Murine AA amyloidosis reproduces crucial features of the 15 disease process in humans (2,7,8) and the fibril precursor proteins (mSAA1.1 in mouse and hSAA1.1 and hSAA1.3 in humans) are largely conserved (9).
The disease show prion-like characteristics in mice and other animals (8,10). Injection of AA amyloid fibrils or oligomers into mice with high serum mSAA1.1 levels leads to a very fast formation of amyloid deposits in the recipients (8,11). Transmission of the disease to mice is 20 possible via oral uptake and with amyloid fibrils purified from different species, including humans (6). However, there is no evidence that human AA amyloidosis is transmissible or contagious.
While AA amyloid fibrils are crucial for the prion-like features (8) and disease pathology (2), their 3 detailed structure remains unknown. In this study, we have used electron cryo-microscopy (cryo-EM) to determine the structure of an AA amyloid fibril from diseased murine spleen.
The protocol to extract the fibrils from amyloidotic tissue avoids harsh physical or chemical conditions and maintains the intact linear architecture of the amyloid fibrils (12). Based on transmission electron microscopy (TEM) more than 90 % of the extracted fibrils belong to the 5 same morphology that is defined by a width of ~12 nm and a cross-over distance of ~75 nm (fig. S1, A and B). The extracted AA amyloid fibrils contrast morphologically with 'amyloid-like fibrils' prepared from full-length mSAA1.1 or the fragment mSAA1.1    20 not develop amyloid deposits (amyloid grade 0). We conclude that the prion-like activity is associated with a protease resistant conformation adopted by residues 1-76. 4 Using cryo-EM, we determined the three-dimensional (3D) structure of the major amyloid fibril morphology (no proteinase K treatment) at a resolution of 3.0 Å ( Fig. 1; fig. S3A and table S2). Our 3D map resolves most side-chains and allowed us to unambiguously trace and assign residues 1-69 of mSAA1.1 ( Fig. 1D and table S3). The density corresponding to residues 70-82 is diffuse ( fig. S3B), likely due to structural disorder. Two-dimensional (2D) projections of the fitted Each PF represents a stack of fibril proteins and encompasses nine parallel cross-β sheets (β1-β9) ( Fig. 2A), consistent with the general cross-β architecture of amyloid fibrils (17, 18). The fibril conformation differs substantially from all known native conformations of SAA family members that uniformly belong to the all-α class of protein folds (Fig. 2B). The conformational differences between the fibril protein and its native precursor resemble the differences between 15 cellular and pathogenic prion protein (19) and imply the unfolding of native structure as a prerequisite of fibril formation in vivo. The PFs and the polypeptide chains appear slightly tilted relative to the main fibril axis (Fig. 1B) S2B). The fibril protein conformation superficially resembles the Greek key topology, but similar to α-synuclein-derived amyloid-like fibrils (16,20), there are no intramolecular backbone hydrogen bonds between the β-5 strands of the same polypeptide chain as required for a Greek key. Instead, the intramolecular strand-strand interactions are formed by the amino acid side-chains, and we hereafter refer to this motif as an 'amyloid key'.
The central structural element of the amyloid key is an N-terminal β-arch (residues 1-21), the most hydrophobic and amyloidogenic segment of the protein (21-23). This β-arch is Å along the fibril axis and interacts with six other protein molecules, four within the same PF and two of the opposite PF (Fig. 2D). This non-planarity originates from the tilt of the PF with respect 10 to the fibril axis (Fig. 1B) and a GPGG motif (residues 47-50) that induces a ~5.5 Å height change of the polypeptide chain relative to the fibril axis.
The fibril surface is predominantly hydrophilic, whereas residues Phe3, Phe5, Phe10, Met16, Trp17 and Ala19 form a tightly packed hydrophobic core within the N-terminal β-arch ( Fig. 2E). The more C-terminal segments of the protein pack onto the outer face of the N-terminal 15 β-arch, mediated by a complex pattern of ionic, polar and hydrophobic interactions ( fig. S5). Some of these interactions occur across different molecular layers of the fibril, interdigitating the PF structure. Salt bridges structure the interface between the two PFs which is remarkably small and formed by only two residues (Asp59 and Arg61) from the C-terminal β-arch (Fig. 3). These residues make bidentate, reciprocal contacts with the respective residues in chains i+1 and i-1 in 20 the opposing PF, which is consistent with the observed pseudo-21 symmetry.
Our structure leads to a simple mechanism for the protein-like replication of the fibril structure that depends on the β1/β7 zipper region and the exposure of unpaired half zippers at 6 either end of the fibril ( fig. S6). Pairing of the half zipper with a complementary strand from an incoming mSAA1.1 molecule helps to orient the newly added chain at the fibril tip and nucleates the formation of a new half zipper in the added molecule such that the fibril structure can be further replicated. A related mechanism has recently been proposed for amyloid-like fibrils formed from Aβ(1-42) peptide (15). 5 The buried position and conformation of residues 1-21 explains the previously described role of the protein N-terminus for fibril formation (22) and for the resistance of Mus musculus czech and CE/J mice to development of systemic AA amyloidosis. This resistance originates from the expression of mSAA1.5 and mSAA2.2 proteins in these animals (24, 25). Both proteins differ from mSAA1.1 by a synonymous Ile6Val substitution, which is consistent with our amyloid fibril 10 morphology and a Gly7His substitution (Fig. 4A) that is structurally disruptive and places a bulky, charged residue into the tightly packed, hydrophobic core of the N-terminal β-arch (Fig. 4B).
Importantly, mSAA2.2 is able to readily form amyloid-like fibrils in vitro (23), demonstrating that the amyloid resistance of CE/J mice does not arise from an intrinsic inability to form cross-β structures but rather from its specific incompatibility with the pathogenic fibril architecture 15 described here. The importance of Gly7 is further corroborated by its conservation in the human pathogenic proteins hSAA1.1 and hSAA1.3 ( Fig. 4A) but not in mSAA2.1, mSAA3 and mSAA4, which carry a bulky, positively charged residue at this position and are unable to form amyloid in vivo (9).
Murine AA amyloidosis can be cross-seeded by amyloid fibrils purified from other species, 20 including humans (6,26). Although detailed AA amyloid fibril structures from other species are not known, the human fibril precursor proteins hSAA1. conformational differences in these more C-terminal structural elements. 5 This first detailed structure of an amyloid fibril from systemic amyloidosis underscores the importance of working with ex vivo fibrils when drawing conclusions on the disease process. Our structure provides mechanistic clues about fibril formation and reveals a previously unknown relevance of electrostatic interactions for structuring AA amyloid fibrils. It explains the importance of the protein N-terminus for fibril formation and the resistance of two mouse variants to the 10 development of amyloidosis. The observed protein fold and its compactness are remarkable given that mSAA1.1 was optimized by biological evolution to adopt a fundamentally different, but nevertheless compact conformation. The disease-associated fibril differs in morphology from all amyloid-like fibrils examined in this study and it is also more proteinase resistant (fig. S1). As proteinase resistance is a common feature of many ex vivo amyloid fibrils and prions (14,19, 29), the ability to form a compact and/or protease resistant cross-β structure may explain the accumulation of only certain forms of aggregates inside the body. While clarifying these issues will require detailed structural information on other amyloid and amyloid-like fibrils, our current study demonstrates the general feasibility to obtain such information in systemic amyloidosis.

Materials and Methods
Figs. S1 to S6 Tables S1 to S3 15

Recombinant mSAA proteins
Recombinant mSAA1.1, mSAA1.1(1-76) and mSAA1.1(1-82) were expressed and purified as described previously (23). 5 Animal experiments and extraction of AA amyloid fibrils from murine tissue Animal experiments were performed with female 6-to 8-week-old NMRI mice (Charles River Laboratories). Mice were divided into 4 experimental groups with two animals per group. Tris buffer pH 8. The sample was desalted using a ZipTip (Merck Millipore). Matrix-assisted laser desorption/ionization MS spectra were recorded as described previously (23). Based on our set up a maximum error of 2 Da was assumed. 15 Negative stain TEM Negative-stain TEM specimens were prepared by loading 5 µL of the sample (0.2 mg/mL) onto a formvar and carbon coated 200 mesh copper grid (Plano). After incubation of the sample for 1 min at room temperature, the excess solvent was removed with filter paper. The grid was washed three times with water and stained three times with 2 % uranyl acetate solution. Grids were 20 examined in a JEM-1400 transmission electron microscope (JEOL) that was operated at 120 kV. 23 Platinum shadowing Formvar and carbon coated 200 mesh copper grids (Plano) were glow-discharged using a PELCO easiGlow glow discharge cleaning system (TED PELLA). A 5 µl droplet of the AA amyloid fibril sample (0.2 mg/mL) were placed onto a grid and incubated for 30 s at room temperature. Excessive solution was removed with filter paper (Whatman). Grids were washed 5 three times with water and dried at room temperature for 30 min. Platinum was evaporated at an angle of 30° using a Balzers TKR 010 to form a 1 nm thick layer on the sample. Grids were examined in a JEM-1400 transmission electron microscope (JEOL), operated at 120 kV.
Cryo-EM 10 A 4 µL aliquot of a sample of AA amyloid fibrils (0.2 mg/mL) was applied to glowdischarged holey carbon coated grids (C-flat 2/1, 200 mesh), blotted with filter paper and plungefrozen in liquid ethane using a Vitrobot Mark 3 (Thermo Fisher Scientific) Images were acquired using a K2-Summit detector (Gatan) in super-resolution counting mode on a Titan Krios transmission electron microscope (Thermo Fisher Scientific) at 300 kV. 1063 images that were 15 composed of 40 frames each, were recorded in two sessions with exposure times of 0.6 s (session 1) and 0.5 s (session 2), exposure rates of 1.97 e -/Å 2 /s (session 1) and 2.52 e -/Å 2 /s (session 2) and a pixel size of 1.35 Å. Defocus values ranged from -1.3 to -5.5 µm.
Helical reconstruction 20 Super-resolution movie frames were corrected for gain reference using IMOD (32). Motion correction, dose-weighting and binning by a factor of 2 was done using MOTIONCOR2 (33). To obtain a total electron dose of 20 e -/Å 2 per aligned image, frames 3 -19 (session 1) or frames 3 -24 18 (session 2) of each recorded movie were used. The contrast transfer function was estimated from the motion-corrected images using Gctf (34). Helical reconstruction was performed using RELION 2.1 (35). Fibrils were selected manually from the aligned micrographs. Segments were extracted using a box size of 210 pixel and an inter-box distance of 21 pixel (10 % of box length).
Reference-free 2D classification with a regularization value of T = 2 was used to select class  Once assembled, the added polypeptide chain exposes a β1 (top) or β7 (bottom) half zipper itself.
The exposure of different half zippers at either end of the polar fibril implies different mechanisms and/or outgrowth kinetics to occur at the two fibril ends, which was observed with other fibril 10 systems (15,53 Molprobity score 1.07 Table S3. Structural statistics of model building and refinement.