Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Cryo-EM reveals structural breaks in a patient-derived amyloid fibril from systemic AL amyloidosis

Cryo-EM reveals structural breaks in a patient-derived amyloid fibril from systemic AL amyloidosis ARTICLE https://doi.org/10.1038/s41467-021-21126-2 OPEN Cryo-EM reveals structural breaks in a patient- derived amyloid fibril from systemic AL amyloidosis 1 1 2 1 3 3 Lynn Radamaker , Julian Baur , Stefanie Huhn , Christian Haupt , Ute Hegenbart , Stefan Schönland , 1 1 1 Akanksha Bansal , Matthias Schmidt & Marcus Fändrich Systemic AL amyloidosis is a debilitating and potentially fatal disease that arises from the misfolding and fibrillation of immunoglobulin light chains (LCs). The disease is patient- specific with essentially each patient possessing a unique LC sequence. In this study, we present two ex vivo fibril structures of a λ3 LC. The fibrils were extracted from the explanted heart of a patient (FOR005) and consist of 115-residue fibril proteins, mainly from the LC variable domain. The fibril structures imply that a 180° rotation around the disulfide bond and a major unfolding step are necessary for fibrils to form. The two fibril structures show highly similar fibril protein folds, differing in only a 12-residue segment. Remarkably, the two structures do not represent separate fibril morphologies, as they can co-exist at different z-axial positions within the same fibril. Our data imply the presence of structural breaks at the interface of the two structural forms. 1 2 Institute of Protein Biochemistry, Ulm University, Ulm, Germany. Medical Department V, Section of Multiple Myeloma, Heidelberg University Hospital, 3 ✉ Heidelberg, Germany. Medical Department V, Amyloidosis Center, Heidelberg University Hospital, Heidelberg, Germany. email: marcus.faendrich@uni-ulm.de NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 1 1234567890():,; ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ystemic AL amyloidosis is a protein misfolding disease that IGLV3-19*01, IGLJ2*01, and IGLC2, which agrees with previous is newly diagnosed in 4–15 persons per one million per year data showing that the IGLV3-19 GL segment is linked to heart 1 6,7 Sin the United States of America and other parts of the involvement . The amino acid sequence of the fibril protein 2,3 world . The amyloid deposits underlying this disease frequently differs from the protein sequence of the translated GL segments occur in heart and kidneys, with cardiac involvement being the in several positions, probably as a result of the B-cell clone-spe- most important prognostic factor for the patient survival . The cific somatic hypermutation. The fibril protein sequence contains LC amino acid sequence is highly variable, as a consequence of five mutations with respect to the GL protein sequence within the the recombination of different variable (V), joining (J), and IGLV3-19 segment (Tyr31Ser, Tyr48Phe, Gly49Arg, Asn51Ser, constant (C) germ line (GL) segments, as well as somatic and Gly94Ala), one within IGLJ2*01 (Val97Gln), and one within hypermutation . Out of the resulting LC variants, the subtypes λ1, IGLC2 (Val135Gly). C mutations are rarely reported for patients λ2, λ3, λ6, and κ1 are in particular associated with AL with AL amyloidosis, possibly because the cDNA-based gene 6,7 amyloidosis . sequencing of the fibril protein precursor is often confined to the It is well established that amyloid fibrils and other LC aggre- V domain. 4,5 gates play a defining role in the pathogenicity of this disease . However, except for the chemotherapeutic removal of the pathogenic plasma cell clone, no pharmacological treatment Observation of two fibril structures in the fibril extracts. The options exist which directly prevent fibril formation or reverse extracted fibrils were subjected to cryo-EM and imaged at 300 kV fibril-induced organ damage . One reason for this paucity of (Supplementary Fig. 1a). Visual inspection of the recorded images treatment options is the lack of knowledge about the mechanism revealed, despite evidence for polymorphism , one apparently of LC misfolding and the structure of pathogenic amyloid fibrils dominant fibril morphology that corresponded to >95% of fibrils in vivo. Another reason is the patient-specific nature of systemic visible in the fibril extracts. Picking this fibril morphology for 3D AL amyloidosis, with each patient presenting an essentially reconstruction and performing two-dimensional (2D) and 3D unique LC precursor and fibril protein . classification resulted in two 3D classes showing two different To provide insight into the fibril structure and LC misfolding fibril structures, termed here A and B (Fig. 1 and Supplementary mechanism in vivo, we recently set up a research strategy in Fig. 1b). The corresponding reconstructions were refined to spatial which AL amyloid fibrils were extracted from diseased tissue and resolutions of 3.2 Å for fibril structure A and 3.4 Å for B (Sup- 10,11 subjected to biochemical analysis . The fibril proteins are plementary Table 1), based on the 0.143 Fourier shell correlation mainly derived from the LC variable (V ) domain of the fibril (FSC) criterion (Supplementary Fig. 2a). Their local resolution 10,11 12 protein precursor , consistent with earlier observations . varied in the fibril cross sections, with higher resolution occurring They contain the intramolecular disulfide bond that is also pre- at the fibril center and lower resolution toward the edges (Sup- 10,13 sent within the natively folded V domain . The fibrils are plementary Fig. 2b). Additional rounds of 3D classification did not polymorphic , but consistent amyloid fibril morphologies are further subdivide the data sets in a meaningful fashion. found in different organs/deposition sites within the same After the initial 3D classification, the data set contained patient . Different AL patients present different fibril 64,652 segments classified as fibril structure A and 36,667 as fibril 10,11 morphologies , suggesting that the variability of the LC structure B. The final reconstructions contained 11,003 segments sequence leads to different, or even patient-specific fibril for fibril structure A and 12,122 for fibril structure B (Supplemen- structures. tary Table 1). We interpreted the two reconstructions with To obtain insight into their molecular conformations, we and molecular models (Fig. 1c, d) and obtained model resolutions of others recently started to employ cryo-electron microscopy (cryo- 3.1 Å for reconstruction A and 3.2 Å for B (Supplementary Table 1 EM) combined with three-dimensional (3D) reconstruction. So and Supplementary Fig. 2c). 2D projections of the models far, two AL amyloid fibrils were analyzed with this combination correspond well to the 2D class averages of the original segments of methods, one derived from a λ1 (ref. ), termed hereafter (Supplementary Fig. 3). Both models depict polar fibrils with C1 FOR006, and one from a λ6LC . The two fibril proteins showed helical symmetry (Supplementary Table 1), consisting of only one markedly different folds, and their conformations differed fun- protofilament and a single stack of fibril proteins (Fig. 1b). All damentally from natively folded LCs. In the present study, we peptide bonds of the fibril proteins, including the two X-Pro bonds, analyze the structure of fibrils that were purified from the heart are modeled as trans isomers. muscle tissue of a patient (FOR005) with λ3 LC-derived amyloid Both 3D maps contain diffuse density decorating the ordered fibrils. Using cryo-EM, we obtained two different fibril structures, fibril core (Fig. 1c, d, blue star), reminiscent of the two previously 13,14 termed here A and B. The two structures coexist at different z- reported cryo-EM structures of ex vivo AL amyloid fibrils . axial positions within the same fibril, which implies the presence These diffuse density regions may represent disordered parts of of structural breaks in these patient-derived amyloid fibrils. the fibril protein or non-fibril components. In addition, there is a well-defined density feature (Fig. 1c, d, red star) that appears to stem from a peptide segment in β-sheet conformation, owing Results to the zigzag pattern and a 4.8 Å rise along the fibril axis Extraction of the fibril protein and sequence analysis. The (Supplementary Fig. 4a, b). Similar well-defined density islands analyzed AL amyloid fibrils were extracted from the explanted were previously observed with in vitro formed fibril 15,16 heart of a female patient (FOR005) with systemic AL amyloidosis. structures . One study suggested that the density islands were The patient suffered from severe cardiomyopathy and underwent formed from a segment of the fibril protein that was protruding heart transplantation at the age of 50 years. We previously from the main fibril core . In another study, the density island obtained the amino acid sequence of the fibril protein by protein originated from a peripherally attached fibril protein that adopted sequencing, and the nucleotide sequence of the precursor LC by a single, short cross-β-strand but that was otherwise conforma- 10 16 cDNA sequencing . The tissue-deposited fibril protein consists tionally disordered . As all segments outside the FOR005 fibril of residues Ser2–Ser116 of the λ3 precursor LC, which corre- core are too short to reach our density islands (Supplementary spond to the V domain and a few residues (Gly109–Ser116) of Fig. 4c), we conclude that non-covalently attached fibril proteins, the LC constant (C ) domain. Bioinformatic analysis of the LC or fibril protein fragments, are the most plausible explanation of cDNA sequence indicated that it originates from the GL segments the density islands in the FOR005 fibril structure. 2 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE Fig. 1 Two different fibril protein conformations underlie the FOR005 amyloid fibrils. a Side views of the 3D maps of fibril structures A and B (left, gray), and corresponding molecular models (right, green/magenta). b Side view of longer segments of the two molecular models. c, d Cross-sectional views of the fibril protein conformations A (c) (EMD-11031) and B (d) (EMD-11030). Blue asterisk: region with blurry density surrounding the fibril core. Red asterisk: extra density decorating the fibril core, indicating an ordered peptide conformation. e Overlay of the molecular models of fibril structures A (PDB: 6Z10) and B (PDB: 6Z1I). The N- and C-terminal residues of the model are highlighted. Comparison of the fibril protein conformations A and B.The bonds that extend between the strands of the cross-β-sheets. In two fibril structures arise from similar but slightly different protein addition, there are side chain–side chain interactions, such as conformations. The fibril proteins are essentially indistinguishable polar ladders of asparagine or glutamine residues, or stacked at residues Ala9–Arg49 and Leu72–Val107. Residues Ser2–Pro8 and hydrophobic or aromatic groups. These features are shown for Leu108–Ser116 could not be assigned to any well-defined density in residues Gln37 and Phe48 in the Supplementary Fig. 6a, b. The the 3D map, which implies that these segments are structurally fibril backbones show axial height changes of 7.5 Å (fibril A) and heterogeneous or disordered. The main difference between the two 7.0 Å (fibril B), which lead to polar fibril topologies and sterically structures lies in the segment Arg60–Ser71 (Fig. 1e). In fibril A, interlock the fibril layers. Each fibril protein layer interacts only residues Arg60–Ser71 are in a stable conformation encompassing with the layers above and below, e.g., Lys38 from layer i interacts an arch, while residues Lys50–Asp59 are not well defined in the 3D with Asp81 from layer i + 1 (Supplementary Fig. 6c). map (Fig. 1c). In fibril B, residues Asn68–Ser71 are in a relatively The surfaces of both fibrils are rich in charged and polar amino extended conformation and the region of structural disorder occurs acids (Fig. 2c ad Supplementary Fig. 5b). The fibril cores contain between residues Lys50–Gly67 (Fig. 1d).Importantly,massspec- small hydrophobic patches, such as the one formed by residues trometry previously demonstrated the fibril protein to be con- Val10, Val12, Leu14, Val98, Phe99, Leu105, and Val107 (Fig. 2c tinuous and to extend from Ser2 to Ser116 (ref. ). Thus, the fibril and Supplementary Fig. 5b), as well as patches of buried polar core as seen in our 3D map is not made up of two fibril protein residues. The structure buries a number of compensating fragments, but instead it consists of two structurally ordered seg- charge–charge interactions, for example, at residues Asp25 and ments (Ala9–Arg49 and Arg60/Asn68–Val107) that are linked by a Arg28, Arg28 and Asp84, Lys38 and Asp81, Glu80 and Arg90 structurally heterogeneous region (Lys50–Asp59/Gly67). (Fig. 2c and Supplementary Fig. 5b), as well as an acidic moiety, The fibril protein shows β-strand conformation at residues which is not fully charge compensated. This moiety is formed by Val10–Leu14, Thr17–Gln23, Asp25–Ser26, Arg28–Ser31, residues Glu80, Asp81, Glu82, and Asp84 (Fig. 2d), resembling Trp34–Gln37, Pro43–Ile47, Leu72–Thr75, Ala79–Glu82, the partially uncompensated acidic moiety in the previously Tyr85–Tyr86, Asn88–Asp91, Asn95–Gln97, and Thr103–Thr106 described λ1 fibril structure . In contrast to the previous λ1 fibril, in both fibrils (Fig. 2a). We refer to these segments as β1to β12. however, there is no water-filled cavity around the acidic moiety Structure A contains two additional β-strands in a segment that is in our fibril. disordered in structure B (Arg60–Gly67). These strands are formed by residues Arg60–Ser62 and Ser64–Ser65 and are termed Location of aggregation-prone segments and mutations. The β6′ and β6″ because they are in between the strands β6 and β7. All mutagenic changes of the amyloidogenic LCs compared with the strands form cross-β-sheets with parallel, hydrogen bonded GL sequences are widely believed to trigger amyloidosis in the strand–strand interactions (Fig. 2b and Supplementary Fig. 5a). 9,17 respective patients . However, analysis of the mutated positions The protein fold is compact and devoid of large internal cavities. within our structure does not readily offer an explanation for their pathogenicity. Some mutations, such as Asn51Ser and Molecular interactions defining the fibril structure. The fibril Val135Gly, lie within a part of the precursor protein that is dis- proteins interact along the fibril axis through backbone hydrogen ordered or cleaved off in the fibril (Fig. 2a). In addition, none of NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 β4 a b β3 β1 β2 β3 β4 β5 Native β2 β9 β5 β1 β2 β3 β4 β4 β5 β6 Fibril A β10 β1 β8 β1 β2 β3 β4 β4 β5 β6 Fibril B β11 2 10 20 30 40 50 β7 β12 SELTQDPAVSVALGQTVRITCQGDSLRSYSASWYQQKPGQAPVLVIFRK β6 β6'' Native β6 β7 β8 β9 β10 β6' Fibril A β6' β6'' β7 β8 β9 β10 β11 β12 Fibril B β7 β8 β9 β10 β11 β12 60 70 80 90 100 110 116 SNRPSGIPDRFSGSSSGNTASLTITGAQAEDEADYYCNSRDSSANHQVFGGGTKLTVLGQPKAAPS 44 F P A Q I Y 38 72 A 70 L 60 74 A S S W I 34 E T Y 80 N 62 30 G 84 E 82 T 94 68 S D 92 28 Y 90 N R G 86 66 R S 24 Y N S hydrophobic (A, F, I, L, V, W) Q 100 64 D H G G S C 14 22 S 16 98 polar (C, N, Q, S, T, Y) 26 C Q G G V Q I V basic (K, R, H) T A K acidic (D, E) glycine proline Fig. 2 Secondary structure and folding of the fibril proteins. a Schematic representation of the secondary structure of the fibril proteins A and B, and of a crystal structure of the refolded fibril protein (native, PDB: 5L6Q ). Arrows indicate β-strands and cylinders α-helical conformations. Continuous lines indicate ordered conformation, dotted lines indicate disordered segments. The definition of secondary structural elements follows the definition in the respective manuscripts. b Ribbon diagram of a stack of six fibril proteins (conformation A). β-strands have been colored in rainbow palette from the N- to the C-terminus. c Schematic representation of the amino acid positions in conformation A. d Electrostatic surface representation of the fibril protein conformation A. Red indicates negative charge, blue positive, and white neutral. Supplementary Figure 5 shows the corresponding images for fibril conformation B. the mutations affecting the fibril core adds an obviously favorable analysis. Support for this view comes from a recent study in interaction. The nonconservative Gly49Arg mutation even leads which the rather counterintuitive observation was reported that a to a buried charge that is not compensated by a nearby opposite conservative leucine to valine mutation on the surface of a charge (Fig. 3a), suggesting that this mutation may even be patient-derived V domain is strongly destabilizing to the native unfavorable to the fibril structure. Moreover, analysis of the protein structure, and promotes the formation of amyloid fibrils location of the mutations within the native LC does not readily in vitro . provide evidence that they might be destabilizing to the native protein conformation (Supplementary Fig. 7a, b). The mutations do not remove an obviously stabilizing interaction and do not Conformations A and B coexist within the same fibrils. Finally, affect internal residues that might be considered to be crucial for we sought to determine whether the two reconstructed 3D maps protein stability. Instead, all mutations are located on the surface A and B represent two different fibril morphologies, or whether of the globularly folded LC (Supplementary Fig. 7a, b). the two structures coexist within the same fibril particle. By visual Computer-based predictions of the aggregation propensity of inspection of the cryo-EM micrographs and measurement of the FOR005 LC identified the highest aggregation score in the V global parameters, such as fibril width or crossover distance, we domain at residues Val44–Arg49 (Fig. 3b and Supplementary could not categorize the fibrils in our sample into separate Fig. 8). These residues form a hydrophobic patch on the fibril structures A and B. Also, the reconstructed 3D maps have surface, which is decorated with the extra density region identical helical parameters, such as fibril symmetry, polarity, described above (Fig. 1c, d, blue star). The three disordered axial rise, and twist value, as well as a fibril pitch of 155 nm protein segments in the fibril protein (Ser2–Pro8, Lys50–Gly67, (Supplementary Table 1). The difference between the two struc- and Leu108–Ser116) correlate with regions having low aggrega- tures could only be revealed when the fibril images were cropped tion scores (Supplementary Fig. 8). However, comparing the into segments that were then aligned independently of their aggregation score of the FOR005 LC to that of the GL protein structural context during 3D classification. sequence (Supplementary Fig. 8) does not reveal any clear trend Analyzing the origin of the fibril segments in the respective whether the FOR005 LC or its putative GL precursor is more data sets producing reconstructions A and B, we would have aggregation prone (Supplementary Fig. 8). We conclude that the expected, for separate morphologies, that each fibril contains effect of mutation is subtle and not readily evident by the above segments belonging to only one of the two data sets A or B 4 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE a b Y48F CDR2 G40 N51S R49 Y31S E80 CDR1 G49R Y30 A70 R60 G94A R90 V97Q G100 I20 CDR3 V10 A9 C V107 Fig. 3 Location of the mutational positions and aggregation-prone regions in the fibril structure. a Mutations with respect to the GL protein sequence (purple) and CDRs (black) marked in fibril structure A (green). The dotted line represents the intermedial disordered segment. b Molecular model of conformation A colored according to aggregation score. (except for minor classification errors). Surprisingly, however, we the intramolecular disulfide bond of the native V domain, found that most fibrils in our sample showed a mixture of A and indicating that LC misfolding happens in an oxidative environ- B segments (Fig. 4a and Supplementary Fig. 9a). Fractions of type ment, such as the extracellular space or an endocytic compart- B segments per fibril varied almost continuously from 0 to 1 ment, that retains the disulfide bond of the native V domain. (Fig. 4b and Supplementary Fig. 9b). The assignment of segments The fibril proteins show an antiparallel N-to-C orientation at the to the fibril protein conformations A and B could not readily be disulfide that is flipped by 180° relative to the native state. The correlated with certain positions on the fibril helix on the cryo- fibril protein conformations differ substantially from the natively EM micrographs, for example, the crossover or the segment in folded LC, demonstrating that a global structural rearrangement between two crossovers, which might have suggested problems in and/or unfolding reaction takes place during the conversion of their alignment. Moreover, the segments are not randomly the native LC, or of a LC fragment, into a fibril. The fibrils are distributed across the fibrils, but mostly separated into distinct decorated by blurry density of uncertain origin that may arise regions along the fibril axis, in which all segments correspond to from fibril protein segments outside the ordered core, or cellular either structure A or B (Fig. 4a). These results were obtained factors attached to the fibril surface. consistently across different data sets, including the data set Patient-specific features of the fibrils include the exact fold of resulting from the initial 3D classification (Fig. 4a, b), as well as the fibril protein (Fig. 5), and the location of the β-strands and the data set from the final reconstructions (Supplementary Fig. 9a, disordered segments within the sequence (Supplementary b). In conclusion, the fibrils in our data set cannot simply be Fig. 10). The current λ3- and the previous λ1-derived fibril pro- divided into two fibril morphologies A and B. Instead, the two teins possess solvent-exposed and conformationally disordered fibril structures A and B occur simultaneously within a fibril N-termini, while the N-terminal segment of the λ6-derived fibril protein stack. This observation indicates that there are structural is buried in the fibril core and part of a β-strand (Fig. 5 and breaks at the interface of fibril regions corresponding to Supplementary Fig. 10). The C-termini are disordered in each of structures A or B (Fig. 4c, d). these fibril proteins. Our current fibril structures and the previous λ6 fibril structure each contain an internal, disordered segment interrupting the fibril protein fold. In contrast, the fold of the λ1 Discussion fibril protein is continuous (Fig. 5 and Supplementary Fig. 10). The λ1 fibril possesses three large channels, two of which are We here present the cryo-EM structures of two amyloid fibrils (A and B) that were extracted from the explanted heart of a patient thought to be water-filled, while the third one contains an apolar molecular inclusion . No such channels or inclusions were (FOR005) with systemic AL amyloidosis. The spatial resolutions are 3.2 Å for fibril structure A and 3.4 Å for fibril structure B identified in the other fibril structures. A feature unique to the current λ3 fibrils is a well-resolved density island attached to the (Supplementary Table 1). These resolutions are sufficient to establish the overall fibril topology and the fibril protein fold. fibril core (Supplementary Fig. 4a, b). However, uncertainty remains in the exact conformation of the Our structures also differ from a number of studies which used nuclear magnetic resonance (NMR) spectroscopy to investigate backbone and side chains. This problem is further exacerbated by the known artifacts of cryo-EM structures, such as a loss of side the structure of LC-derived fibrils formed in vitro. These fibrils were formed from V domain constructs and include murine κ chain density due to beam damage , which could be relevant in 20 21 22 our reconstructions, e.g., at residues Glu80 or Lys104 (Fig. 1c, d). (ref. ), human κ1 (ref. ), as well as human λ3 (ref. ) and λ6 sequences . Particularly relevant in this case is the compar- Systemic AL amyloidosis is a patient-specific disease . Identi- fication of common structural features in different patient- ison of our structures to the NMR analysis of recombinant FOR005 V domain fibrils . These fibrils were seeded in vitro derived amyloid fibrils is potentially informative about common steps in the misfolding pathways across patient cases. Based on with amyloid fibrils that were extracted from the heart of the patient (FOR005) with the aim to propagate the ex vivo fibril the available cryo-EM structures of ex vivo fibrils from systemic AL amyloidosis (Fig. 5 and Supplementary Fig. 10), the following structure in the in vitro seeded fibrils . Comparison of the in vitro seeded fibrils with our cryo-EM structures of patient commonalities can now be identified: the extracted fibril samples contain a dominant fibril morphology that consists of a single, fibrils revealed several differences. First, the ex vivo fibrils possess polar protofilament. The fibril core is formed by the V domain of a stable β-strand at residues Thr103–Thr106 that are outside the ordered core of the in vitro seeded fibrils. Second, the in vitro the precursor λ-LC in all cases. The C domain is structurally disordered and/or lost by proteolysis. The fibril proteins retain seeded fibrils contain a salt bridge between residues Arg49 and NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 a b 4500 All fibrils Fibrils containg at least 5 segments 4000 Fibrils containg at least 10 segments 30 Fibrils containg at least 20 segments 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Fraction of segments in each fibril in conformation B R49 I74 β7 N68 N68 β7 I74 R60 β6'' S64 A9 V107 N C Fig. 4 Evidence for structural breaks in FOR005 fibrils. a Representative cryo-EM micrograph showing the location of segments classified after the first 3D classification as fibril structure A (green) and fibril structure B (magenta). Scale bar: 100 nm. Data were collected from 1,378 micrographs from one fibril sample. Supplementary Fig. 9a shows the same image, highlighting only the segments used in the final fibril reconstruction. b Histogram of the fraction of segments classified as fibril structure B, per fibril, after the first 3D classification. Different thresholds were chosen for the minimum number of segments per fibril, resulting in four categories: all fibrils (11,194 fibrils), fibrils containing a minimum of 5 segments (7,738 fibrils), fibrils containing a minimum of 10 segments (4,278 fibrils), and fibrils containing a minimum of 20 segments (951 fibrils). The percentages were normed using the total number of fibrils in each category. Colored points show the absolute number of fibrils in each category and group (fraction in conformation B). In total, the data set (n = 101,319) contained 64,652 segments classified as fibril structure A and 36,667 as fibril structure B. Supplementary Fig. 9b shows an analogous histogram, but including only the segments used for the final reconstructions. c Stack of three fibril proteins in conformation B (magenta) on top of three fibril proteins in conformation A (green), illustrating the presence of structural breaks within the patient amyloid fibrils. d Detailed view of a structural break, including side chains. Asp25 (ref. ), which are far apart in the protein fold of the replication of the fibril protein fold, this replication of the seed patient fibril (Fig. 2c). Third, the in vitro seeded fibrils show an protein structure may not necessarily occur during heterogeneous electrostatic interaction between Lys50 and Asp81 (ref. ), seeding. Indeed, the observation of density islands on the outside whereas in the ex vivo fibrils, Lys50 is in the internal disordered of the FOR005 fibril core structure (Fig. 1c, d, red star and region, and Asp81 is far away from this segment (Fig. 2c) and has Supplementary Fig. 4a, b) suggests that fibril proteins may have an electrostatic interaction with Lys38 (Supplementary Fig. 6). attached on the fibril surface, but do not fully replicate the fold of These data demonstrate that the in vitro fibrils are structurally the fibril protein. Hence, it is important to use patient-derived different from the ex vivo fibrils analyzed here with cryo-EM. In fibrils when investigating the structural basis of disease. A similar vitro seeding with ex vivo FOR005 fibrils did not propagate, in conclusion was obtained previously when FOR005 fibril protein this case, the seed structure to the daughter fibrils, although it was extracted from the patient’s heart, denatured in guanidine, modified the fibril structure compared with unseeded fibrils . refolded, and fibrillated in vitro (without seeds). These in vitro These observations imply that the seeding mechanism did not fibrils also showed a different morphology than the fibrils that involve a replication of the seed structure. Early work with sickle were purified from FOR005 patient tissue , as judged by trans- cell hemoglobin identified two possible seeding mechanisms: mission electron microscopy. homogeneous and heterogeneous nucleation. Homogeneous A particularly interesting finding in the present study is the nucleation involves the attachment of the soluble fibril precursor observation of structural breaks. So far, it has been part of our proteins to the fibril tip, while heterogeneous nucleation involves general understanding of amyloid fibril structures that these are the nucleation of new fibrils on the lateral cylindrical surface of an conformationally uniform along the fibril axis. Occasionally, existing fibril . While attachment of the fibril precursor protein fibrils were reported that differed morphologically at its two 25–28 to the tip of an amyloid fibril would be expected to lead to a ends . Some of these cases could be attributed to a fibril cross- 6 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications % of fibrils within category Number of fibrils NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE λ1 (PDB: 6IC3) λ3 (PDB: 6Z10) λ6 (PDB: 6HUD) R49 R60 G105 C87 C91 C89 N1 C22 C22 C22 S66 Y37 A9 V107 G15 T105 S2 S116 S118 V3 Fig. 5 Comparison of the available cryo-EM structures of ex vivo AL amyloid fibrils. Ribbon diagrams of a λ1 fibril (PDB: 6IC3 ), the current λ3 fibril (conformation A, PDB: 6Z10), and a λ6 fibril (PDB: 6HUD ). The fibrils are shown in a cross-sectional view. For all structures, the location of the disulfide bond forming cysteine residues is marked. Disordered segments are represented as dotted gray lines and depicted in an arbitrary conformation. The first and the last residue of the ordered segments, as well as the first and the last residue of the fibril protein are indicated, if known. seeding, that is, the attachment of a different fibril precursor While more work is necessary to discriminate between these protein to a fibril tip, or to a splintering of a multi-protofilament two mechanisms, our observations lead to an important change in fibril into fibril morphologies with a smaller number, or a dif- our understanding of the assembly of polypeptide chains into ferent arrangement of protofilaments. In other cases, it was amyloid fibrils. They demonstrate that these linear aggregates are unclear whether two fibril morphologies may have annealed after not as perfectly regular and uniform as has generally been their formation. In our samples, however, the majority of fibrils assumed by most previous studies. Considering that the breaks show a mixture of conformations, and show multiple, seemingly were revealed in the FOR005 fibril samples only at an advanced arbitrary switching between the conformations A and B in each stage of the analysis, we would predict that they will be observed fibril. Therefore, short segments, possibly down to a single protein more frequently in the future, as the methods of structural biol- layer, may be able to adopt a conformation different from that of ogy become more powerful and will be able to resolve such fine the surrounding layers. The fibril breaks and the two fibril details more routinely. Structural breaks and other structural structures defining the breaks emerged at the 3D classification defects in cross-β-sheets could have significant ramifications for stage in our analysis and resolved a previously blurry density the biological properties of amyloid fibrils. Examples hereof region into two distinct density paths (Supplementary Fig. 1b). include the fragility and the loss of torsional coherence of 32,33 Unresolved density regions resulting from one or more dis- amyloid fibrils , the branching of amyloid fibrils during fibril ordered segments of the protein chain are reported for the outgrowth and the ability of molecular chaperones to bind to, to majority of cryo-EM structures of in vitro and ex vivo amyloid sever, and to break down amyloid aggregates . fibrils . Therefore, structural heterogeneity such as described here could be relevant to other fibril structures as well. Further- Methods more, it is possible that our fibrils contain fibril protein structures Source of AL fibrils. Heart tissue was collected from a female patient (FOR005) at which we were unable to resolve so far. We originally extracted the age of 50, suffering from AL amyloidosis and consequent advanced heart 194,502 fibril segments from the cryo-electron micrographs and failure. A monoclonal gammopathy was the underlying condition. The patient was used only 11,003 (A) and 12,122 (B) of these segments for the treated within the heart transplant program of the University Hospital Heidelberg. final 3D reconstructions (Supplementary Table 1). The explanted heart tissue was stored at −80 °C. The study was approved by the ethical committees of the University of Heidelberg (123/2006) and of Ulm Uni- Two possible scenarios can be envisioned to explain the versity (203/18). Informed consent was obtained from the patient for the analysis mechanism of the formation of structural breaks. One scenario is of the amyloid deposits. that they appear during fibril assembly due to an imperfect replication of the seed structure, as a new molecule attaches to the fibril end (Fig. 6). Consistent with this idea, real-time microscopy Fibril extraction from patient tissue. Applying a previously established proto- studies explained the stop-and-go kinetics during fibril growth col for fibril extraction, 250 mg of patient heart tissue were diced finely and 0.5 mL of ice-cold Tris calcium buffer (20 mM Tris, 138 mM NaCl, 2 mM CaCl , with irregularities in the addition of molecules to the tip of a 0.1 (w/v) % NaN , pH 8.0) added. The sample was homogenized using a Kontes 30 3 growing fibril . The other scenario is that breaks emerge after Pellet Pestle, after which it was centrifuged for 5 min at 3100 × g at 4 °C. The fibril assembly, for example, because initially disordered segments washing step was repeated five times and each supernatant was stored for −1 adopt different stable conformations, which then proliferate along further analysis. Afterward, 1 mL of freshly prepared 5 mg mL Clostridium his- tolyticum collagenase (Sigma) in Tris calcium buffer with ethylenediaminete- the fibril axis (Fig. 6). Support for the latter mechanism is pro- traacetic acid (EDTA)-free protease inhibitor (Roche) were added and the pellet vided by the fact that the type A and B fibril proteins are mostly resuspended. Overnight incubation at 37 °C was followed by a 30 min centrifuge identical, and that the differences are confined to a small segment cycle at 3100 × g. Ten further washing steps with 20 mM Tris, 140 mM NaCl, that lies in the vicinity of an unstructured region. 10 mM EDTA, and 0.1 % (w/v) NaN and ten subsequent steps with ice-cold water NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 During fibril extension After fibril formation Fig. 6 The origin of structural breaks: two possible scenarios. Schematic representation of a stack of fibril proteins, illustrating two different hypotheses on how structural breaks form: during fibril extension (left) or after fibril formation (right). Conformation A is represented by two β-sheets (green). Conformation B is represented by one β-sheet (magenta). Disordered segments are represented by dotted gray lines. Gray arrows represent the immature fibril proteins, before the mature conformations A and B are fully adopted. were then performed on the pellet, using a pipette for homogenization. One of the crystallographic symmetry and secondary structure restraints imposed. This pro- water supernatants was selected for cryo-EM. cess was repeated until a satisfactory map to model fit was obtained. This model was subsequently used to perform model based automated sharpening of the map 42,43 (phenix.auto_sharpen ). The sharpened map was used for improving the model Cryo-EM. Holey carbon-coated grids (C-flat 1.2/1.3 400 mesh) were glow- further. The final refined model was evaluated for its quality using the MolProb- discharged using 40 mA for 40 s. Using a Vitrobot (Thermo Fisher Scientific), 44 ity validation report. 3.5 μL of the extracted fibril sample were incubated on each grid for 30 s at a humidity of >95%, the excess fluid was blotted off and the grid plunged into liquid Sequence analysis. The amino acid sequence of the LC investigated in this study ethane, then transferred to a grid box. After plunging, one grid from each grid box was taken from the gene bank entry KX290463 (ref. ), which was obtained by (containing four grids) was checked using a 200 kV Jeol JEM 2100 F electron cDNA sequencing of FOR005 patient material. The residue numbering throughout microscope (Ulm University). The remaining grids in the grid boxes were kept at this article refers to the precursor LC sequence (GenBank ANN81988.1) from the liquid nitrogen temperature. Cryo-electron microscopic image acquisition of one patient, starting with the first residue (Ser1) after the signal sequence, which is selected grid was performed using a Titan Krios transmission electron microscope cleaved off in the fibril protein. All mutations in this manuscript are represented in (Thermo Fisher Scientific) at 300 kV equipped with a K2-Summit detector (Gatan) the direction GL to FOR005 fibril protein. The sequence elements were defined as in counting mode. A Gatan imaging filter with a 20 eV slit was applied. The follows. First, the patient cDNA was translated to the putative amino acid sequence software package SerialEM v3.7 was used for data collection. The data acquisition of the fibril precursor protein. Then, the cDNA of the patient and the corre- parameters can be found in Supplementary Table 1. Global parameters of the fibril 36 sponding amino acid sequence were analyzed to determine the most probable GL morphologies were measured using Fiji . No clearly identifiable second mor- 45 46 segments, using the vbase2 (ref. ) and BLAST/BLAT search tools (http://www. phology was found and the occurrence of the main morphology was estimated at ensembl.org). This analysis yielded several hits for possible GL segments. The over 95%. cDNA sequences and corresponding amino acid sequences of these V, J, and C GL 45 47 segments were retrieved from the vbase2 (ref. ), ENSEMBL , and http://www. imgt.org databases, and genetic distances to the patient sequence were calculated Helical reconstruction. Helical reconstruction was performed using Relion 2.1 37 38 (ref. ). The raw data were converted from TIFF to mrcs format using IMOD . by maximum composite likelihood to confirm the most probable V, J, and C GL segments. Finally, the cDNA sequence of the patient and the corresponding amino Motion and gain corrections, as well as dose-weighting were performed using acid sequence were aligned to these GL segments (fit > 80%), using the MEGA Motioncor2 (ref. ). The non-dose-weighted micrographs were used to determine 40 49 multiple sequencing alignment tool “MUSCLE” and Clustal Omega (https:// the defocus values with Gctf , whereas the dose-weighted micrographs were used www.ebi.ac.uk/Tools/msa/clustalo/) to identify patient-specific mutations. The for all subsequent steps. Fibrils were manually selected from 1,964 micrographs and segments were extracted with a box size of 300 pixels (312 Å) with an inter-box definitions of the CDRs were taken from a previous manuscript . LC aggregation propensities for every residue were calculated using the distance of 33.6 Å (~11%). Two rounds of reference-free 2D classification were 51 52 53 54 performed with a regularization value of T = 2. Class averages which showed the programs TANGO , Foldamyloid , Aggrescan , and PASTA 2.0 (ref. ). For every prediction tool standard parameters were chosen. For Tango (Version 2.1), helical repeat along the fibril axis were selected, whereas classes showing artifacts and noise were discarded, resulting in a selection of 101,319 particles. The selection every amino acid which scores above 5% was counted as a hit. For Foldamyloid, the “triple hybrid” scale was used, meaning that five successive amino acids with a was confirmed by manually arranging the class averages into a full 2D fibril side view. An initial model from a previous reconstruction was filtered to 60 Å resulting score above 21.4 were counted as hits. PASTA 2.0 calculated aggregation with 90% sensitivity and predicted hits if the energy cutoff fell below −2.8 PASTA Energy in a rod-like structure. This initial model was used as a reference to create a first 3D map using 3D classification with 553 particles picked from nine micrographs. The Units. For Aggrescan, all amino acids with values above −0.02 were counted as hot spots. In this study, an aggregation score of 0 means that none of these programs resulting map was used as a reference for 3D classification with six classes and T = identified the corresponding residue as aggregation prone. An aggregation score of 3 using the 101,319 particles obtained from the 2D classification selection. Three of 4 means that all four programs identified the corresponding residue as prone to the six classes showed a clear backbone and revealed the presence of two different aggregation. conformations. These classes were selected (60,044 particles) and further rounds of 3D classification with four classes and step-wise increasing T-values from 3 to 20 were performed. Approximately 11,003 particles were selected for conformation A Protein structure representation. UCSF Chimera was used for creating the and 12,122 particles for conformation B. Final round of 3D classification with images of the density maps and protein models. The structure of the refolded increasing T-values from 80 to 200, followed by 3D auto-refinement and post- FOR005 fibril protein was obtained previously using protein X-ray crystallography processing with a soft-edged mask and an estimated sharpening B-factor of and has the protein data bank (PDB) entry 5L6Q . The native C domain is the −33.1013 Ų (conformation A) and −55.1958 Ų (conformation B) for each of the IGLC2 segment of PDB 4EOW , where the side chain of residue Val135 was not two conformations led to the post-processed maps. Helical z-percentages used for shown in order to represent the mutated residue Gly135 of the fibril protein. The 3D classification and 3D refinement varied between 0.1 and 0.3. The twist and rise following previously published coordinates were used in Fig. 5: 6IC3 and 6HUD. were 1.11 and 4.79, respectively, for both conformations. These values agree with the measured crossover distances on the motion-corrected cryo-EM micrographs, Reporting summary. Further information on research design is available in the Nature and the corresponding power spectra. A left-handed twist was assumed. The Research Reporting Summary linked to this article. resolution of each map was estimated from the value of the FSC curve for two independently refined half-maps at 0.143. Data availability The reconstructed cryo-EM maps were deposited in the Electron Microscopy Data Bank Model building and refinement. The model was manually built using the software 41 with accession codes EMD-11031 (structure A) and EMD-11030 (structure B). The Coot . The process was initiated by tracing a poly-Ala chain along the 3D map. coordinates of the fitted atomic model were deposited in the PDB under the accession The alanine residues were mutated to the LC sequence as determined previously . codes 6Z10 (structure A) and 6Z1I (structure B). The cryo-EM data were deposited on The model obtained was subjected to manual, as well as automated refinement 42,43 EMPIAR with the accession code EMPIAR-10457. The data sets used during the current using phenix.real_space_refine tool, as implemented in phenix with non- 8 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE study are available from the corresponding author on reasonable request. The unique 27. Zhang, W. et al. Heparin-induced tau filaments are polymorphic and differ materials used in the current study are available from the authors on reasonable from those in Alzheimer’s and Pick’s diseases. Elife 8, e43584 (2019). request. Source data are provided with this paper. 28. Kollmer, M. et al. Cryo-EM structure and polymorphism of Aβ amyloid fibrils purified from Alzheimer’s brain tissue. Nat. Commun. 10.1,1–8 (2019). 29. Gallardo, R., Ranson, N. E. & Radford, S. E. Amyloid structures: much more Received: 5 August 2020; Accepted: 21 December 2020; than just a cross-β fold. Curr. Opin. Struct. Biol. 60,7–16 (2020). 30. Ferkinghoff-Borg, J. et al. Stop-and-go kinetics in amyloid fibrillation. Phys. Rev. E 82, 010901 (2010). 31. Ndlovu, H., Ashcroft, A. E., Radford, S. E. & Harris, S. A. Molecular dynamics simulations of mechanical failure in polymorphic arrangements of amyloid fibrils containing structural defects. Beilstein J. Nanotechnol. 4, 429–440 (2013). References 32. Knowles, T. P., Smith, J. F., Craig, A., Dobson, C. M. & Welland, M. E. Spatial 1. Quock, T. P., Yan, T., Chang, E., Guthrie, S. & Broder, M. S. Epidemiology of persistence of angular correlations in amyloid fibrils. Phys. Rev. Lett. 96, AL amyloidosis: a real-world study using US claims data. Blood Adv. 2, 238301 (2006). 1046–1053 (2018). 33. Knowles, T. P., Smith, J. F., Devlin, G. L., Dobson, C. M. & Welland, M. E. 2. Shimazaki, C. Current diagnosis and treatment of AL amyloidosis in Japan: a Analysis of structural order in amyloid fibrils. Nanotechnology 18, 044031 nationwide epidemiological survey. Rinsho Ketsueki 60.8, 973–978 (2019). (2006). 3. Wisniowski, B. et al. The epidemiology of amyloidosis in Queensland, 34. Andersen, C. B. et al. Branching in amyloid fibril growth. Biophys. J. 96, Australia. Br. J. Haematol. 186.6, 829–836 (2019). 1529–1536 (2009). 4. Merlini, G. et al. Systemic immunoglobulin light chain amyloidosis. Nat. Rev. 35. Gao, X. et al. Human Hsp70 disaggregase reverses Parkinson’s-linked α- Dis. Prim. 4,1–19 (2018). synuclein amyloid fibrils. Mol. Cell 59, 781–793 (2015). 5. Blancas-Mejía, L. M. & Ramirez-Alvarado, M. Systemic amyloidoses. Annu. 36. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Rev. Biochem 82, 745–774 (2013). Nat. Methods 9, 676–682 (2012). 6. Perfetti, V. et al. The repertoire of lambda light chains causing predominant 37. He, S. & Scheres, S. H. W. Helical reconstruction in RELION. J. Struct. Biol. amyloid heart involvement and identification of a preferentially involved 198, 163–176 (2017). germline gene, IGLV1-44. Blood 119, 144–150 (2012). 38. Kremer, J. R., Mastronarde, D. N. & McIntosh, J. R. Computer visualization of 7. Kourelis, T. V. et al. Clarifying immunoglobulin gene usage in systemic and three-dimensional image data using IMOD. J. Struct. Biol. 116,71–76 (1996). localized immunoglobulin light-chain amyloidosis by mass spectrometry. 39. Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced Blood 129, 299–306 (2017). motion for improved cryo-electron microscopy. Nat. Methods 14, 331–332 8. Vaxman, I., Dispenzieri, A., Muchtar, E. & Gertz, M. New developments in (2017). diagnosis, risk assessment and management in systemic amyloidosis. Blood 40. Zhang, K. Gctf: Real-time CTF determination and correction. J. Struct. Biol. Rev. 40, 100636 (2020). 193,1–12 (2016). 9. Bellotti, V., Mangione, P. & Merlini, G. Review: immunoglobulin light chain 41. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development amyloidosis—the archetype of structural and pathogenic variability. J. Struct. of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010). Biol. 130, 280–289 (2000). 42. Afonine, P. V. et al. New tools for the analysis and validation of cryo-EM 10. Annamalai, K. et al. Common fibril structures imply systemically conserved maps and atomic models. Acta Crystallogr. D Struct. Biol. 74, 814–840 (2018). protein misfolding pathways in vivo. Angew. Chem. Int. Ed. 56.26, 7510–7514 43. Afonine, P. V. et al. Real-space refinement in PHENIX for cryo-EM and (2017). crystallography. Acta Crystallogr. D Struct. Biol. 74, 531–544 (2018). 11. Annamalai, K. et al. Polymorphism of amyloid fibrils in vivo. Angew. Chem. 44. Williams, C. J. et al. MolProbity: more and better reference data for improved Int. Ed. 55.15, 4822–4825 (2016). all atom structure validation. Protein Sci. 27.1, 293–315 (2018). 12. Glenner, G. G., Harbaugh, J., Ohms, J. I., Harada, M. & Cuatrecasas, P. An 45. Retter, I., Althaus, H. H., Munch, R. & Muller, W. VBASE2, an integrative V amyloid protein: the amino-terminal variable fragment of an immunoglobulin gene database. Nucleic Acids Res. 33, D671–D674 (2005). light chain. Biochem. Biophys. Res. Commun. 41, 1287–1289 (1970). 46. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local 13. Radamaker, L. et al. Cryo-EM structure of a light chain-derived amyloid fibril alignment search tool. J. Mol. Biol. 215, 403–410 (1990). from a patient with systemic AL amyloidosis. Nat. Commun. 10.1,1–8 (2019). 47. Hunt, S. E. et al. Ensembl variation resources. Database 2018, bay119 (2018). 14. Swuec, P. et al. Cryo-EM structure of cardiac amyloid fibrils from an 48. Ruiz, M. et al. IMGT, the international ImMunoGeneTics database. Nucleic immunoglobulin light chain AL amyloidosis patient. Nat. Commun. 10.1,1–9 Acids Res. 28, 219–221 (2000). (2019). 49. Tamura, K., Stecher, G., Peterson, D., Filipski, A. & Kumar, S. MEGA6: 15. Boyer, D. R. et al. Structures of fibrils formed by α-synuclein hereditary molecular evolutionary genetics analysis version 6.0. Mol. Biol. Evol. 30, disease mutant H50Q reveal new polymorphs. Nat. Struct. Mol. Biol. 26, 2725–2729 (2013). 1044–1052 (2019). 50. Sievers, F. et al. Fast, scalable generation of high-quality protein multiple 16. Cao, Q., Boyer, D. R., Sawaya, M. R., Ge, P. & Eisenberg, D. S. Cryo-EM sequence alignments using Clustal Omega. Mol. Syst. Biol. 7, 539 (2011). structures of four polymorphic TDP-43 amyloid cores. Nat. Struct. Mol. Biol. 51. Fernandez-Escamilla, A. M., Rousseau, F., Schymkowitz, J. & Serrano, L. 26, 619–627 (2019). Prediction of sequence-dependent and mutational effects on the aggregation 17. Blancas-Mejia, L. M. et al. Immunoglobulin light chain amyloid aggregation. of peptides and proteins. Nat. Biotechnol. 22, 1302–1306 (2004). Chem. Commun. 54.76, 10664–10674 (2018). 52. Garbuzynskiy, S. O., Lobanov, M. Y. & Galzitskaya, O. V. FoldAmyloid: a 18. Kazman, P. et al. Fatal amyloid formation in a patient’s antibody light chain is method of prediction of amyloidogenic regions from protein sequence. caused by a single point mutation. Elife 9, e52300 (2020). Bioinformatics 26, 326–332 (2010). 19. Wang, J. & Moore, P. B. On the interpretation of electron microscopic maps of 53. Conchillo-Solé, O. et al. AGGRESCAN: a server for the prediction and evaluation biological macromolecules. Protein Sci. 26.1, 122–129 (2017). of “hot spots” of aggregation in polypeptides. BMC Bioinformatics 8, 65 (2007). 20. Hora, M. et al. MAK33 antibody light chain amyloid fibrils are similar to 54. Walsh, I., Seno, F., Tosatto, S. C. & Trovato, A. PASTA 2.0: an improved server oligomeric precursors. PLoS ONE 12.7, e0181799 (2017). for protein aggregation prediction. Nucleic Acids Res. 42, W301–W307 (2014). 21. Piehl, D. W., Blancas-Mejía, L. M., Ramirez-Alvarado, M. & Rienstra, C. M. 55. Pettersen, E. F. et al. UCSF Chimera—a visualization system for exploratory Solid-state NMR chemical shift assignments for AL-09 V L immunoglobulin research and analysis. J. Comput. Chem. 25, 1605–1612 (2004). light chain fibrils. Biomol. NMR Assign. 11,45–50 (2017). 56. Blech, M. et al. Molecular structure of human GM-CSF in complex with a 22. Pradhan, T. et al. Seeded fibrils of the germline variant of human λ-III disease-associated anti-human GM-CSF autoantibody and its potential immunoglobulin light chain FOR005 have a similar core as patient fibrils with biological implications. Biochem. J. 447.2, 205–215 (2012). reduced stability. J. Biol. Chem. 295, 18474–18484 (2020). 23. Lecoq, L. et al. A substantial structural conversion of the native monomer leads to in-register parallel amyloid fibril formation in light-chain amyloidosis. Chembiochem 20.8, 1027–1031 (2019). Acknowledgements 24. Ferrone, F. A., Hofrichter, J. & Eaton, W. A. Kinetics of sickle hemoglobin We would like to thank the Deutsche Forschungsgemeinschaft for funding of the polymerization: I. Studies using temperature-jump and laser photolysis Research Unit FOR 2969, projects FA 456/27, HA 7138/3, HE 8472/1-1, HU 2400/1-1, techniques. J. Mol. Biol. 183, 591–610 (1985). and SCHO 1364/2-1. We are grateful to Prof. Dr. Bernd Reif (Technical University 25. Crowther, R. A. Straight and paired helical filaments in Alzheimer disease have Munich) for helpful discussions and to the students who participated in the practical a common structural unit. Proc. Natl Acad. Sci. USA 88, 2288–2292 (1991). course “Protein Biochemistry and Structural Biology” in the winter semester 2019, for 26. Makarava, N., Ostapchenko, V. G., Savtchenko, R. & Baskakov, I. V. assisting in the manual picking of the fibril segments. All cryo-EM data were collected at Conformational switching within individual amyloid fibrils. J. Biol. Chem. the European Molecular Biology Laboratory, Heidelberg (Germany), funded by iNEXT 284, 14386–14395 (2009). (Horizon 2020, European Union). NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 Author contributions Reprints and permission information is available at http://www.nature.com/reprints L.R. and J.B. carried out experiments. L.R, J.B., S.H., C.H., A.B., M.S., and M.F. analyzed data. U.H. and S.S. contributed tools and reagents. M.F. designed research. L.R. and M.F. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in wrote the paper. published maps and institutional affiliations. Funding Open Access This article is licensed under a Creative Commons Open Access funding enabled and organized by Projekt DEAL. Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give Competing interests appropriate credit to the original author(s) and the source, provide a link to the Creative The authors declare no competing interests. Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the Additional information article’s Creative Commons license and your intended use is not permitted by statutory Supplementary information The online version contains supplementary material regulation or exceeds the permitted use, you will need to obtain permission directly from available at https://doi.org/10.1038/s41467-021-21126-2. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Correspondence and requests for materials should be addressed to M.F. licenses/by/4.0/. Peer review information Nature Communications thanks Luis del Pozo-Yauner and the © The Author(s) 2021 other, anonymous, reviewer(s) for their contribution to the peer review of this work. 10 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Nature Communications Springer Journals

Cryo-EM reveals structural breaks in a patient-derived amyloid fibril from systemic AL amyloidosis

Loading next page...
 
/lp/springer-journals/cryo-em-reveals-structural-breaks-in-a-patient-derived-amyloid-fibril-Fo7i2kJYws

References (65)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2021
eISSN
2041-1723
DOI
10.1038/s41467-021-21126-2
Publisher site
See Article on Publisher Site

Abstract

ARTICLE https://doi.org/10.1038/s41467-021-21126-2 OPEN Cryo-EM reveals structural breaks in a patient- derived amyloid fibril from systemic AL amyloidosis 1 1 2 1 3 3 Lynn Radamaker , Julian Baur , Stefanie Huhn , Christian Haupt , Ute Hegenbart , Stefan Schönland , 1 1 1 Akanksha Bansal , Matthias Schmidt & Marcus Fändrich Systemic AL amyloidosis is a debilitating and potentially fatal disease that arises from the misfolding and fibrillation of immunoglobulin light chains (LCs). The disease is patient- specific with essentially each patient possessing a unique LC sequence. In this study, we present two ex vivo fibril structures of a λ3 LC. The fibrils were extracted from the explanted heart of a patient (FOR005) and consist of 115-residue fibril proteins, mainly from the LC variable domain. The fibril structures imply that a 180° rotation around the disulfide bond and a major unfolding step are necessary for fibrils to form. The two fibril structures show highly similar fibril protein folds, differing in only a 12-residue segment. Remarkably, the two structures do not represent separate fibril morphologies, as they can co-exist at different z-axial positions within the same fibril. Our data imply the presence of structural breaks at the interface of the two structural forms. 1 2 Institute of Protein Biochemistry, Ulm University, Ulm, Germany. Medical Department V, Section of Multiple Myeloma, Heidelberg University Hospital, 3 ✉ Heidelberg, Germany. Medical Department V, Amyloidosis Center, Heidelberg University Hospital, Heidelberg, Germany. email: marcus.faendrich@uni-ulm.de NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 1 1234567890():,; ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ystemic AL amyloidosis is a protein misfolding disease that IGLV3-19*01, IGLJ2*01, and IGLC2, which agrees with previous is newly diagnosed in 4–15 persons per one million per year data showing that the IGLV3-19 GL segment is linked to heart 1 6,7 Sin the United States of America and other parts of the involvement . The amino acid sequence of the fibril protein 2,3 world . The amyloid deposits underlying this disease frequently differs from the protein sequence of the translated GL segments occur in heart and kidneys, with cardiac involvement being the in several positions, probably as a result of the B-cell clone-spe- most important prognostic factor for the patient survival . The cific somatic hypermutation. The fibril protein sequence contains LC amino acid sequence is highly variable, as a consequence of five mutations with respect to the GL protein sequence within the the recombination of different variable (V), joining (J), and IGLV3-19 segment (Tyr31Ser, Tyr48Phe, Gly49Arg, Asn51Ser, constant (C) germ line (GL) segments, as well as somatic and Gly94Ala), one within IGLJ2*01 (Val97Gln), and one within hypermutation . Out of the resulting LC variants, the subtypes λ1, IGLC2 (Val135Gly). C mutations are rarely reported for patients λ2, λ3, λ6, and κ1 are in particular associated with AL with AL amyloidosis, possibly because the cDNA-based gene 6,7 amyloidosis . sequencing of the fibril protein precursor is often confined to the It is well established that amyloid fibrils and other LC aggre- V domain. 4,5 gates play a defining role in the pathogenicity of this disease . However, except for the chemotherapeutic removal of the pathogenic plasma cell clone, no pharmacological treatment Observation of two fibril structures in the fibril extracts. The options exist which directly prevent fibril formation or reverse extracted fibrils were subjected to cryo-EM and imaged at 300 kV fibril-induced organ damage . One reason for this paucity of (Supplementary Fig. 1a). Visual inspection of the recorded images treatment options is the lack of knowledge about the mechanism revealed, despite evidence for polymorphism , one apparently of LC misfolding and the structure of pathogenic amyloid fibrils dominant fibril morphology that corresponded to >95% of fibrils in vivo. Another reason is the patient-specific nature of systemic visible in the fibril extracts. Picking this fibril morphology for 3D AL amyloidosis, with each patient presenting an essentially reconstruction and performing two-dimensional (2D) and 3D unique LC precursor and fibril protein . classification resulted in two 3D classes showing two different To provide insight into the fibril structure and LC misfolding fibril structures, termed here A and B (Fig. 1 and Supplementary mechanism in vivo, we recently set up a research strategy in Fig. 1b). The corresponding reconstructions were refined to spatial which AL amyloid fibrils were extracted from diseased tissue and resolutions of 3.2 Å for fibril structure A and 3.4 Å for B (Sup- 10,11 subjected to biochemical analysis . The fibril proteins are plementary Table 1), based on the 0.143 Fourier shell correlation mainly derived from the LC variable (V ) domain of the fibril (FSC) criterion (Supplementary Fig. 2a). Their local resolution 10,11 12 protein precursor , consistent with earlier observations . varied in the fibril cross sections, with higher resolution occurring They contain the intramolecular disulfide bond that is also pre- at the fibril center and lower resolution toward the edges (Sup- 10,13 sent within the natively folded V domain . The fibrils are plementary Fig. 2b). Additional rounds of 3D classification did not polymorphic , but consistent amyloid fibril morphologies are further subdivide the data sets in a meaningful fashion. found in different organs/deposition sites within the same After the initial 3D classification, the data set contained patient . Different AL patients present different fibril 64,652 segments classified as fibril structure A and 36,667 as fibril 10,11 morphologies , suggesting that the variability of the LC structure B. The final reconstructions contained 11,003 segments sequence leads to different, or even patient-specific fibril for fibril structure A and 12,122 for fibril structure B (Supplemen- structures. tary Table 1). We interpreted the two reconstructions with To obtain insight into their molecular conformations, we and molecular models (Fig. 1c, d) and obtained model resolutions of others recently started to employ cryo-electron microscopy (cryo- 3.1 Å for reconstruction A and 3.2 Å for B (Supplementary Table 1 EM) combined with three-dimensional (3D) reconstruction. So and Supplementary Fig. 2c). 2D projections of the models far, two AL amyloid fibrils were analyzed with this combination correspond well to the 2D class averages of the original segments of methods, one derived from a λ1 (ref. ), termed hereafter (Supplementary Fig. 3). Both models depict polar fibrils with C1 FOR006, and one from a λ6LC . The two fibril proteins showed helical symmetry (Supplementary Table 1), consisting of only one markedly different folds, and their conformations differed fun- protofilament and a single stack of fibril proteins (Fig. 1b). All damentally from natively folded LCs. In the present study, we peptide bonds of the fibril proteins, including the two X-Pro bonds, analyze the structure of fibrils that were purified from the heart are modeled as trans isomers. muscle tissue of a patient (FOR005) with λ3 LC-derived amyloid Both 3D maps contain diffuse density decorating the ordered fibrils. Using cryo-EM, we obtained two different fibril structures, fibril core (Fig. 1c, d, blue star), reminiscent of the two previously 13,14 termed here A and B. The two structures coexist at different z- reported cryo-EM structures of ex vivo AL amyloid fibrils . axial positions within the same fibril, which implies the presence These diffuse density regions may represent disordered parts of of structural breaks in these patient-derived amyloid fibrils. the fibril protein or non-fibril components. In addition, there is a well-defined density feature (Fig. 1c, d, red star) that appears to stem from a peptide segment in β-sheet conformation, owing Results to the zigzag pattern and a 4.8 Å rise along the fibril axis Extraction of the fibril protein and sequence analysis. The (Supplementary Fig. 4a, b). Similar well-defined density islands analyzed AL amyloid fibrils were extracted from the explanted were previously observed with in vitro formed fibril 15,16 heart of a female patient (FOR005) with systemic AL amyloidosis. structures . One study suggested that the density islands were The patient suffered from severe cardiomyopathy and underwent formed from a segment of the fibril protein that was protruding heart transplantation at the age of 50 years. We previously from the main fibril core . In another study, the density island obtained the amino acid sequence of the fibril protein by protein originated from a peripherally attached fibril protein that adopted sequencing, and the nucleotide sequence of the precursor LC by a single, short cross-β-strand but that was otherwise conforma- 10 16 cDNA sequencing . The tissue-deposited fibril protein consists tionally disordered . As all segments outside the FOR005 fibril of residues Ser2–Ser116 of the λ3 precursor LC, which corre- core are too short to reach our density islands (Supplementary spond to the V domain and a few residues (Gly109–Ser116) of Fig. 4c), we conclude that non-covalently attached fibril proteins, the LC constant (C ) domain. Bioinformatic analysis of the LC or fibril protein fragments, are the most plausible explanation of cDNA sequence indicated that it originates from the GL segments the density islands in the FOR005 fibril structure. 2 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE Fig. 1 Two different fibril protein conformations underlie the FOR005 amyloid fibrils. a Side views of the 3D maps of fibril structures A and B (left, gray), and corresponding molecular models (right, green/magenta). b Side view of longer segments of the two molecular models. c, d Cross-sectional views of the fibril protein conformations A (c) (EMD-11031) and B (d) (EMD-11030). Blue asterisk: region with blurry density surrounding the fibril core. Red asterisk: extra density decorating the fibril core, indicating an ordered peptide conformation. e Overlay of the molecular models of fibril structures A (PDB: 6Z10) and B (PDB: 6Z1I). The N- and C-terminal residues of the model are highlighted. Comparison of the fibril protein conformations A and B.The bonds that extend between the strands of the cross-β-sheets. In two fibril structures arise from similar but slightly different protein addition, there are side chain–side chain interactions, such as conformations. The fibril proteins are essentially indistinguishable polar ladders of asparagine or glutamine residues, or stacked at residues Ala9–Arg49 and Leu72–Val107. Residues Ser2–Pro8 and hydrophobic or aromatic groups. These features are shown for Leu108–Ser116 could not be assigned to any well-defined density in residues Gln37 and Phe48 in the Supplementary Fig. 6a, b. The the 3D map, which implies that these segments are structurally fibril backbones show axial height changes of 7.5 Å (fibril A) and heterogeneous or disordered. The main difference between the two 7.0 Å (fibril B), which lead to polar fibril topologies and sterically structures lies in the segment Arg60–Ser71 (Fig. 1e). In fibril A, interlock the fibril layers. Each fibril protein layer interacts only residues Arg60–Ser71 are in a stable conformation encompassing with the layers above and below, e.g., Lys38 from layer i interacts an arch, while residues Lys50–Asp59 are not well defined in the 3D with Asp81 from layer i + 1 (Supplementary Fig. 6c). map (Fig. 1c). In fibril B, residues Asn68–Ser71 are in a relatively The surfaces of both fibrils are rich in charged and polar amino extended conformation and the region of structural disorder occurs acids (Fig. 2c ad Supplementary Fig. 5b). The fibril cores contain between residues Lys50–Gly67 (Fig. 1d).Importantly,massspec- small hydrophobic patches, such as the one formed by residues trometry previously demonstrated the fibril protein to be con- Val10, Val12, Leu14, Val98, Phe99, Leu105, and Val107 (Fig. 2c tinuous and to extend from Ser2 to Ser116 (ref. ). Thus, the fibril and Supplementary Fig. 5b), as well as patches of buried polar core as seen in our 3D map is not made up of two fibril protein residues. The structure buries a number of compensating fragments, but instead it consists of two structurally ordered seg- charge–charge interactions, for example, at residues Asp25 and ments (Ala9–Arg49 and Arg60/Asn68–Val107) that are linked by a Arg28, Arg28 and Asp84, Lys38 and Asp81, Glu80 and Arg90 structurally heterogeneous region (Lys50–Asp59/Gly67). (Fig. 2c and Supplementary Fig. 5b), as well as an acidic moiety, The fibril protein shows β-strand conformation at residues which is not fully charge compensated. This moiety is formed by Val10–Leu14, Thr17–Gln23, Asp25–Ser26, Arg28–Ser31, residues Glu80, Asp81, Glu82, and Asp84 (Fig. 2d), resembling Trp34–Gln37, Pro43–Ile47, Leu72–Thr75, Ala79–Glu82, the partially uncompensated acidic moiety in the previously Tyr85–Tyr86, Asn88–Asp91, Asn95–Gln97, and Thr103–Thr106 described λ1 fibril structure . In contrast to the previous λ1 fibril, in both fibrils (Fig. 2a). We refer to these segments as β1to β12. however, there is no water-filled cavity around the acidic moiety Structure A contains two additional β-strands in a segment that is in our fibril. disordered in structure B (Arg60–Gly67). These strands are formed by residues Arg60–Ser62 and Ser64–Ser65 and are termed Location of aggregation-prone segments and mutations. The β6′ and β6″ because they are in between the strands β6 and β7. All mutagenic changes of the amyloidogenic LCs compared with the strands form cross-β-sheets with parallel, hydrogen bonded GL sequences are widely believed to trigger amyloidosis in the strand–strand interactions (Fig. 2b and Supplementary Fig. 5a). 9,17 respective patients . However, analysis of the mutated positions The protein fold is compact and devoid of large internal cavities. within our structure does not readily offer an explanation for their pathogenicity. Some mutations, such as Asn51Ser and Molecular interactions defining the fibril structure. The fibril Val135Gly, lie within a part of the precursor protein that is dis- proteins interact along the fibril axis through backbone hydrogen ordered or cleaved off in the fibril (Fig. 2a). In addition, none of NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 β4 a b β3 β1 β2 β3 β4 β5 Native β2 β9 β5 β1 β2 β3 β4 β4 β5 β6 Fibril A β10 β1 β8 β1 β2 β3 β4 β4 β5 β6 Fibril B β11 2 10 20 30 40 50 β7 β12 SELTQDPAVSVALGQTVRITCQGDSLRSYSASWYQQKPGQAPVLVIFRK β6 β6'' Native β6 β7 β8 β9 β10 β6' Fibril A β6' β6'' β7 β8 β9 β10 β11 β12 Fibril B β7 β8 β9 β10 β11 β12 60 70 80 90 100 110 116 SNRPSGIPDRFSGSSSGNTASLTITGAQAEDEADYYCNSRDSSANHQVFGGGTKLTVLGQPKAAPS 44 F P A Q I Y 38 72 A 70 L 60 74 A S S W I 34 E T Y 80 N 62 30 G 84 E 82 T 94 68 S D 92 28 Y 90 N R G 86 66 R S 24 Y N S hydrophobic (A, F, I, L, V, W) Q 100 64 D H G G S C 14 22 S 16 98 polar (C, N, Q, S, T, Y) 26 C Q G G V Q I V basic (K, R, H) T A K acidic (D, E) glycine proline Fig. 2 Secondary structure and folding of the fibril proteins. a Schematic representation of the secondary structure of the fibril proteins A and B, and of a crystal structure of the refolded fibril protein (native, PDB: 5L6Q ). Arrows indicate β-strands and cylinders α-helical conformations. Continuous lines indicate ordered conformation, dotted lines indicate disordered segments. The definition of secondary structural elements follows the definition in the respective manuscripts. b Ribbon diagram of a stack of six fibril proteins (conformation A). β-strands have been colored in rainbow palette from the N- to the C-terminus. c Schematic representation of the amino acid positions in conformation A. d Electrostatic surface representation of the fibril protein conformation A. Red indicates negative charge, blue positive, and white neutral. Supplementary Figure 5 shows the corresponding images for fibril conformation B. the mutations affecting the fibril core adds an obviously favorable analysis. Support for this view comes from a recent study in interaction. The nonconservative Gly49Arg mutation even leads which the rather counterintuitive observation was reported that a to a buried charge that is not compensated by a nearby opposite conservative leucine to valine mutation on the surface of a charge (Fig. 3a), suggesting that this mutation may even be patient-derived V domain is strongly destabilizing to the native unfavorable to the fibril structure. Moreover, analysis of the protein structure, and promotes the formation of amyloid fibrils location of the mutations within the native LC does not readily in vitro . provide evidence that they might be destabilizing to the native protein conformation (Supplementary Fig. 7a, b). The mutations do not remove an obviously stabilizing interaction and do not Conformations A and B coexist within the same fibrils. Finally, affect internal residues that might be considered to be crucial for we sought to determine whether the two reconstructed 3D maps protein stability. Instead, all mutations are located on the surface A and B represent two different fibril morphologies, or whether of the globularly folded LC (Supplementary Fig. 7a, b). the two structures coexist within the same fibril particle. By visual Computer-based predictions of the aggregation propensity of inspection of the cryo-EM micrographs and measurement of the FOR005 LC identified the highest aggregation score in the V global parameters, such as fibril width or crossover distance, we domain at residues Val44–Arg49 (Fig. 3b and Supplementary could not categorize the fibrils in our sample into separate Fig. 8). These residues form a hydrophobic patch on the fibril structures A and B. Also, the reconstructed 3D maps have surface, which is decorated with the extra density region identical helical parameters, such as fibril symmetry, polarity, described above (Fig. 1c, d, blue star). The three disordered axial rise, and twist value, as well as a fibril pitch of 155 nm protein segments in the fibril protein (Ser2–Pro8, Lys50–Gly67, (Supplementary Table 1). The difference between the two struc- and Leu108–Ser116) correlate with regions having low aggrega- tures could only be revealed when the fibril images were cropped tion scores (Supplementary Fig. 8). However, comparing the into segments that were then aligned independently of their aggregation score of the FOR005 LC to that of the GL protein structural context during 3D classification. sequence (Supplementary Fig. 8) does not reveal any clear trend Analyzing the origin of the fibril segments in the respective whether the FOR005 LC or its putative GL precursor is more data sets producing reconstructions A and B, we would have aggregation prone (Supplementary Fig. 8). We conclude that the expected, for separate morphologies, that each fibril contains effect of mutation is subtle and not readily evident by the above segments belonging to only one of the two data sets A or B 4 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE a b Y48F CDR2 G40 N51S R49 Y31S E80 CDR1 G49R Y30 A70 R60 G94A R90 V97Q G100 I20 CDR3 V10 A9 C V107 Fig. 3 Location of the mutational positions and aggregation-prone regions in the fibril structure. a Mutations with respect to the GL protein sequence (purple) and CDRs (black) marked in fibril structure A (green). The dotted line represents the intermedial disordered segment. b Molecular model of conformation A colored according to aggregation score. (except for minor classification errors). Surprisingly, however, we the intramolecular disulfide bond of the native V domain, found that most fibrils in our sample showed a mixture of A and indicating that LC misfolding happens in an oxidative environ- B segments (Fig. 4a and Supplementary Fig. 9a). Fractions of type ment, such as the extracellular space or an endocytic compart- B segments per fibril varied almost continuously from 0 to 1 ment, that retains the disulfide bond of the native V domain. (Fig. 4b and Supplementary Fig. 9b). The assignment of segments The fibril proteins show an antiparallel N-to-C orientation at the to the fibril protein conformations A and B could not readily be disulfide that is flipped by 180° relative to the native state. The correlated with certain positions on the fibril helix on the cryo- fibril protein conformations differ substantially from the natively EM micrographs, for example, the crossover or the segment in folded LC, demonstrating that a global structural rearrangement between two crossovers, which might have suggested problems in and/or unfolding reaction takes place during the conversion of their alignment. Moreover, the segments are not randomly the native LC, or of a LC fragment, into a fibril. The fibrils are distributed across the fibrils, but mostly separated into distinct decorated by blurry density of uncertain origin that may arise regions along the fibril axis, in which all segments correspond to from fibril protein segments outside the ordered core, or cellular either structure A or B (Fig. 4a). These results were obtained factors attached to the fibril surface. consistently across different data sets, including the data set Patient-specific features of the fibrils include the exact fold of resulting from the initial 3D classification (Fig. 4a, b), as well as the fibril protein (Fig. 5), and the location of the β-strands and the data set from the final reconstructions (Supplementary Fig. 9a, disordered segments within the sequence (Supplementary b). In conclusion, the fibrils in our data set cannot simply be Fig. 10). The current λ3- and the previous λ1-derived fibril pro- divided into two fibril morphologies A and B. Instead, the two teins possess solvent-exposed and conformationally disordered fibril structures A and B occur simultaneously within a fibril N-termini, while the N-terminal segment of the λ6-derived fibril protein stack. This observation indicates that there are structural is buried in the fibril core and part of a β-strand (Fig. 5 and breaks at the interface of fibril regions corresponding to Supplementary Fig. 10). The C-termini are disordered in each of structures A or B (Fig. 4c, d). these fibril proteins. Our current fibril structures and the previous λ6 fibril structure each contain an internal, disordered segment interrupting the fibril protein fold. In contrast, the fold of the λ1 Discussion fibril protein is continuous (Fig. 5 and Supplementary Fig. 10). The λ1 fibril possesses three large channels, two of which are We here present the cryo-EM structures of two amyloid fibrils (A and B) that were extracted from the explanted heart of a patient thought to be water-filled, while the third one contains an apolar molecular inclusion . No such channels or inclusions were (FOR005) with systemic AL amyloidosis. The spatial resolutions are 3.2 Å for fibril structure A and 3.4 Å for fibril structure B identified in the other fibril structures. A feature unique to the current λ3 fibrils is a well-resolved density island attached to the (Supplementary Table 1). These resolutions are sufficient to establish the overall fibril topology and the fibril protein fold. fibril core (Supplementary Fig. 4a, b). However, uncertainty remains in the exact conformation of the Our structures also differ from a number of studies which used nuclear magnetic resonance (NMR) spectroscopy to investigate backbone and side chains. This problem is further exacerbated by the known artifacts of cryo-EM structures, such as a loss of side the structure of LC-derived fibrils formed in vitro. These fibrils were formed from V domain constructs and include murine κ chain density due to beam damage , which could be relevant in 20 21 22 our reconstructions, e.g., at residues Glu80 or Lys104 (Fig. 1c, d). (ref. ), human κ1 (ref. ), as well as human λ3 (ref. ) and λ6 sequences . Particularly relevant in this case is the compar- Systemic AL amyloidosis is a patient-specific disease . Identi- fication of common structural features in different patient- ison of our structures to the NMR analysis of recombinant FOR005 V domain fibrils . These fibrils were seeded in vitro derived amyloid fibrils is potentially informative about common steps in the misfolding pathways across patient cases. Based on with amyloid fibrils that were extracted from the heart of the patient (FOR005) with the aim to propagate the ex vivo fibril the available cryo-EM structures of ex vivo fibrils from systemic AL amyloidosis (Fig. 5 and Supplementary Fig. 10), the following structure in the in vitro seeded fibrils . Comparison of the in vitro seeded fibrils with our cryo-EM structures of patient commonalities can now be identified: the extracted fibril samples contain a dominant fibril morphology that consists of a single, fibrils revealed several differences. First, the ex vivo fibrils possess polar protofilament. The fibril core is formed by the V domain of a stable β-strand at residues Thr103–Thr106 that are outside the ordered core of the in vitro seeded fibrils. Second, the in vitro the precursor λ-LC in all cases. The C domain is structurally disordered and/or lost by proteolysis. The fibril proteins retain seeded fibrils contain a salt bridge between residues Arg49 and NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 a b 4500 All fibrils Fibrils containg at least 5 segments 4000 Fibrils containg at least 10 segments 30 Fibrils containg at least 20 segments 0 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Fraction of segments in each fibril in conformation B R49 I74 β7 N68 N68 β7 I74 R60 β6'' S64 A9 V107 N C Fig. 4 Evidence for structural breaks in FOR005 fibrils. a Representative cryo-EM micrograph showing the location of segments classified after the first 3D classification as fibril structure A (green) and fibril structure B (magenta). Scale bar: 100 nm. Data were collected from 1,378 micrographs from one fibril sample. Supplementary Fig. 9a shows the same image, highlighting only the segments used in the final fibril reconstruction. b Histogram of the fraction of segments classified as fibril structure B, per fibril, after the first 3D classification. Different thresholds were chosen for the minimum number of segments per fibril, resulting in four categories: all fibrils (11,194 fibrils), fibrils containing a minimum of 5 segments (7,738 fibrils), fibrils containing a minimum of 10 segments (4,278 fibrils), and fibrils containing a minimum of 20 segments (951 fibrils). The percentages were normed using the total number of fibrils in each category. Colored points show the absolute number of fibrils in each category and group (fraction in conformation B). In total, the data set (n = 101,319) contained 64,652 segments classified as fibril structure A and 36,667 as fibril structure B. Supplementary Fig. 9b shows an analogous histogram, but including only the segments used for the final reconstructions. c Stack of three fibril proteins in conformation B (magenta) on top of three fibril proteins in conformation A (green), illustrating the presence of structural breaks within the patient amyloid fibrils. d Detailed view of a structural break, including side chains. Asp25 (ref. ), which are far apart in the protein fold of the replication of the fibril protein fold, this replication of the seed patient fibril (Fig. 2c). Third, the in vitro seeded fibrils show an protein structure may not necessarily occur during heterogeneous electrostatic interaction between Lys50 and Asp81 (ref. ), seeding. Indeed, the observation of density islands on the outside whereas in the ex vivo fibrils, Lys50 is in the internal disordered of the FOR005 fibril core structure (Fig. 1c, d, red star and region, and Asp81 is far away from this segment (Fig. 2c) and has Supplementary Fig. 4a, b) suggests that fibril proteins may have an electrostatic interaction with Lys38 (Supplementary Fig. 6). attached on the fibril surface, but do not fully replicate the fold of These data demonstrate that the in vitro fibrils are structurally the fibril protein. Hence, it is important to use patient-derived different from the ex vivo fibrils analyzed here with cryo-EM. In fibrils when investigating the structural basis of disease. A similar vitro seeding with ex vivo FOR005 fibrils did not propagate, in conclusion was obtained previously when FOR005 fibril protein this case, the seed structure to the daughter fibrils, although it was extracted from the patient’s heart, denatured in guanidine, modified the fibril structure compared with unseeded fibrils . refolded, and fibrillated in vitro (without seeds). These in vitro These observations imply that the seeding mechanism did not fibrils also showed a different morphology than the fibrils that involve a replication of the seed structure. Early work with sickle were purified from FOR005 patient tissue , as judged by trans- cell hemoglobin identified two possible seeding mechanisms: mission electron microscopy. homogeneous and heterogeneous nucleation. Homogeneous A particularly interesting finding in the present study is the nucleation involves the attachment of the soluble fibril precursor observation of structural breaks. So far, it has been part of our proteins to the fibril tip, while heterogeneous nucleation involves general understanding of amyloid fibril structures that these are the nucleation of new fibrils on the lateral cylindrical surface of an conformationally uniform along the fibril axis. Occasionally, existing fibril . While attachment of the fibril precursor protein fibrils were reported that differed morphologically at its two 25–28 to the tip of an amyloid fibril would be expected to lead to a ends . Some of these cases could be attributed to a fibril cross- 6 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications % of fibrils within category Number of fibrils NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE λ1 (PDB: 6IC3) λ3 (PDB: 6Z10) λ6 (PDB: 6HUD) R49 R60 G105 C87 C91 C89 N1 C22 C22 C22 S66 Y37 A9 V107 G15 T105 S2 S116 S118 V3 Fig. 5 Comparison of the available cryo-EM structures of ex vivo AL amyloid fibrils. Ribbon diagrams of a λ1 fibril (PDB: 6IC3 ), the current λ3 fibril (conformation A, PDB: 6Z10), and a λ6 fibril (PDB: 6HUD ). The fibrils are shown in a cross-sectional view. For all structures, the location of the disulfide bond forming cysteine residues is marked. Disordered segments are represented as dotted gray lines and depicted in an arbitrary conformation. The first and the last residue of the ordered segments, as well as the first and the last residue of the fibril protein are indicated, if known. seeding, that is, the attachment of a different fibril precursor While more work is necessary to discriminate between these protein to a fibril tip, or to a splintering of a multi-protofilament two mechanisms, our observations lead to an important change in fibril into fibril morphologies with a smaller number, or a dif- our understanding of the assembly of polypeptide chains into ferent arrangement of protofilaments. In other cases, it was amyloid fibrils. They demonstrate that these linear aggregates are unclear whether two fibril morphologies may have annealed after not as perfectly regular and uniform as has generally been their formation. In our samples, however, the majority of fibrils assumed by most previous studies. Considering that the breaks show a mixture of conformations, and show multiple, seemingly were revealed in the FOR005 fibril samples only at an advanced arbitrary switching between the conformations A and B in each stage of the analysis, we would predict that they will be observed fibril. Therefore, short segments, possibly down to a single protein more frequently in the future, as the methods of structural biol- layer, may be able to adopt a conformation different from that of ogy become more powerful and will be able to resolve such fine the surrounding layers. The fibril breaks and the two fibril details more routinely. Structural breaks and other structural structures defining the breaks emerged at the 3D classification defects in cross-β-sheets could have significant ramifications for stage in our analysis and resolved a previously blurry density the biological properties of amyloid fibrils. Examples hereof region into two distinct density paths (Supplementary Fig. 1b). include the fragility and the loss of torsional coherence of 32,33 Unresolved density regions resulting from one or more dis- amyloid fibrils , the branching of amyloid fibrils during fibril ordered segments of the protein chain are reported for the outgrowth and the ability of molecular chaperones to bind to, to majority of cryo-EM structures of in vitro and ex vivo amyloid sever, and to break down amyloid aggregates . fibrils . Therefore, structural heterogeneity such as described here could be relevant to other fibril structures as well. Further- Methods more, it is possible that our fibrils contain fibril protein structures Source of AL fibrils. Heart tissue was collected from a female patient (FOR005) at which we were unable to resolve so far. We originally extracted the age of 50, suffering from AL amyloidosis and consequent advanced heart 194,502 fibril segments from the cryo-electron micrographs and failure. A monoclonal gammopathy was the underlying condition. The patient was used only 11,003 (A) and 12,122 (B) of these segments for the treated within the heart transplant program of the University Hospital Heidelberg. final 3D reconstructions (Supplementary Table 1). The explanted heart tissue was stored at −80 °C. The study was approved by the ethical committees of the University of Heidelberg (123/2006) and of Ulm Uni- Two possible scenarios can be envisioned to explain the versity (203/18). Informed consent was obtained from the patient for the analysis mechanism of the formation of structural breaks. One scenario is of the amyloid deposits. that they appear during fibril assembly due to an imperfect replication of the seed structure, as a new molecule attaches to the fibril end (Fig. 6). Consistent with this idea, real-time microscopy Fibril extraction from patient tissue. Applying a previously established proto- studies explained the stop-and-go kinetics during fibril growth col for fibril extraction, 250 mg of patient heart tissue were diced finely and 0.5 mL of ice-cold Tris calcium buffer (20 mM Tris, 138 mM NaCl, 2 mM CaCl , with irregularities in the addition of molecules to the tip of a 0.1 (w/v) % NaN , pH 8.0) added. The sample was homogenized using a Kontes 30 3 growing fibril . The other scenario is that breaks emerge after Pellet Pestle, after which it was centrifuged for 5 min at 3100 × g at 4 °C. The fibril assembly, for example, because initially disordered segments washing step was repeated five times and each supernatant was stored for −1 adopt different stable conformations, which then proliferate along further analysis. Afterward, 1 mL of freshly prepared 5 mg mL Clostridium his- tolyticum collagenase (Sigma) in Tris calcium buffer with ethylenediaminete- the fibril axis (Fig. 6). Support for the latter mechanism is pro- traacetic acid (EDTA)-free protease inhibitor (Roche) were added and the pellet vided by the fact that the type A and B fibril proteins are mostly resuspended. Overnight incubation at 37 °C was followed by a 30 min centrifuge identical, and that the differences are confined to a small segment cycle at 3100 × g. Ten further washing steps with 20 mM Tris, 140 mM NaCl, that lies in the vicinity of an unstructured region. 10 mM EDTA, and 0.1 % (w/v) NaN and ten subsequent steps with ice-cold water NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 During fibril extension After fibril formation Fig. 6 The origin of structural breaks: two possible scenarios. Schematic representation of a stack of fibril proteins, illustrating two different hypotheses on how structural breaks form: during fibril extension (left) or after fibril formation (right). Conformation A is represented by two β-sheets (green). Conformation B is represented by one β-sheet (magenta). Disordered segments are represented by dotted gray lines. Gray arrows represent the immature fibril proteins, before the mature conformations A and B are fully adopted. were then performed on the pellet, using a pipette for homogenization. One of the crystallographic symmetry and secondary structure restraints imposed. This pro- water supernatants was selected for cryo-EM. cess was repeated until a satisfactory map to model fit was obtained. This model was subsequently used to perform model based automated sharpening of the map 42,43 (phenix.auto_sharpen ). The sharpened map was used for improving the model Cryo-EM. Holey carbon-coated grids (C-flat 1.2/1.3 400 mesh) were glow- further. The final refined model was evaluated for its quality using the MolProb- discharged using 40 mA for 40 s. Using a Vitrobot (Thermo Fisher Scientific), 44 ity validation report. 3.5 μL of the extracted fibril sample were incubated on each grid for 30 s at a humidity of >95%, the excess fluid was blotted off and the grid plunged into liquid Sequence analysis. The amino acid sequence of the LC investigated in this study ethane, then transferred to a grid box. After plunging, one grid from each grid box was taken from the gene bank entry KX290463 (ref. ), which was obtained by (containing four grids) was checked using a 200 kV Jeol JEM 2100 F electron cDNA sequencing of FOR005 patient material. The residue numbering throughout microscope (Ulm University). The remaining grids in the grid boxes were kept at this article refers to the precursor LC sequence (GenBank ANN81988.1) from the liquid nitrogen temperature. Cryo-electron microscopic image acquisition of one patient, starting with the first residue (Ser1) after the signal sequence, which is selected grid was performed using a Titan Krios transmission electron microscope cleaved off in the fibril protein. All mutations in this manuscript are represented in (Thermo Fisher Scientific) at 300 kV equipped with a K2-Summit detector (Gatan) the direction GL to FOR005 fibril protein. The sequence elements were defined as in counting mode. A Gatan imaging filter with a 20 eV slit was applied. The follows. First, the patient cDNA was translated to the putative amino acid sequence software package SerialEM v3.7 was used for data collection. The data acquisition of the fibril precursor protein. Then, the cDNA of the patient and the corre- parameters can be found in Supplementary Table 1. Global parameters of the fibril 36 sponding amino acid sequence were analyzed to determine the most probable GL morphologies were measured using Fiji . No clearly identifiable second mor- 45 46 segments, using the vbase2 (ref. ) and BLAST/BLAT search tools (http://www. phology was found and the occurrence of the main morphology was estimated at ensembl.org). This analysis yielded several hits for possible GL segments. The over 95%. cDNA sequences and corresponding amino acid sequences of these V, J, and C GL 45 47 segments were retrieved from the vbase2 (ref. ), ENSEMBL , and http://www. imgt.org databases, and genetic distances to the patient sequence were calculated Helical reconstruction. Helical reconstruction was performed using Relion 2.1 37 38 (ref. ). The raw data were converted from TIFF to mrcs format using IMOD . by maximum composite likelihood to confirm the most probable V, J, and C GL segments. Finally, the cDNA sequence of the patient and the corresponding amino Motion and gain corrections, as well as dose-weighting were performed using acid sequence were aligned to these GL segments (fit > 80%), using the MEGA Motioncor2 (ref. ). The non-dose-weighted micrographs were used to determine 40 49 multiple sequencing alignment tool “MUSCLE” and Clustal Omega (https:// the defocus values with Gctf , whereas the dose-weighted micrographs were used www.ebi.ac.uk/Tools/msa/clustalo/) to identify patient-specific mutations. The for all subsequent steps. Fibrils were manually selected from 1,964 micrographs and segments were extracted with a box size of 300 pixels (312 Å) with an inter-box definitions of the CDRs were taken from a previous manuscript . LC aggregation propensities for every residue were calculated using the distance of 33.6 Å (~11%). Two rounds of reference-free 2D classification were 51 52 53 54 performed with a regularization value of T = 2. Class averages which showed the programs TANGO , Foldamyloid , Aggrescan , and PASTA 2.0 (ref. ). For every prediction tool standard parameters were chosen. For Tango (Version 2.1), helical repeat along the fibril axis were selected, whereas classes showing artifacts and noise were discarded, resulting in a selection of 101,319 particles. The selection every amino acid which scores above 5% was counted as a hit. For Foldamyloid, the “triple hybrid” scale was used, meaning that five successive amino acids with a was confirmed by manually arranging the class averages into a full 2D fibril side view. An initial model from a previous reconstruction was filtered to 60 Å resulting score above 21.4 were counted as hits. PASTA 2.0 calculated aggregation with 90% sensitivity and predicted hits if the energy cutoff fell below −2.8 PASTA Energy in a rod-like structure. This initial model was used as a reference to create a first 3D map using 3D classification with 553 particles picked from nine micrographs. The Units. For Aggrescan, all amino acids with values above −0.02 were counted as hot spots. In this study, an aggregation score of 0 means that none of these programs resulting map was used as a reference for 3D classification with six classes and T = identified the corresponding residue as aggregation prone. An aggregation score of 3 using the 101,319 particles obtained from the 2D classification selection. Three of 4 means that all four programs identified the corresponding residue as prone to the six classes showed a clear backbone and revealed the presence of two different aggregation. conformations. These classes were selected (60,044 particles) and further rounds of 3D classification with four classes and step-wise increasing T-values from 3 to 20 were performed. Approximately 11,003 particles were selected for conformation A Protein structure representation. UCSF Chimera was used for creating the and 12,122 particles for conformation B. Final round of 3D classification with images of the density maps and protein models. The structure of the refolded increasing T-values from 80 to 200, followed by 3D auto-refinement and post- FOR005 fibril protein was obtained previously using protein X-ray crystallography processing with a soft-edged mask and an estimated sharpening B-factor of and has the protein data bank (PDB) entry 5L6Q . The native C domain is the −33.1013 Ų (conformation A) and −55.1958 Ų (conformation B) for each of the IGLC2 segment of PDB 4EOW , where the side chain of residue Val135 was not two conformations led to the post-processed maps. Helical z-percentages used for shown in order to represent the mutated residue Gly135 of the fibril protein. The 3D classification and 3D refinement varied between 0.1 and 0.3. The twist and rise following previously published coordinates were used in Fig. 5: 6IC3 and 6HUD. were 1.11 and 4.79, respectively, for both conformations. These values agree with the measured crossover distances on the motion-corrected cryo-EM micrographs, Reporting summary. Further information on research design is available in the Nature and the corresponding power spectra. A left-handed twist was assumed. The Research Reporting Summary linked to this article. resolution of each map was estimated from the value of the FSC curve for two independently refined half-maps at 0.143. Data availability The reconstructed cryo-EM maps were deposited in the Electron Microscopy Data Bank Model building and refinement. The model was manually built using the software 41 with accession codes EMD-11031 (structure A) and EMD-11030 (structure B). The Coot . The process was initiated by tracing a poly-Ala chain along the 3D map. coordinates of the fitted atomic model were deposited in the PDB under the accession The alanine residues were mutated to the LC sequence as determined previously . codes 6Z10 (structure A) and 6Z1I (structure B). The cryo-EM data were deposited on The model obtained was subjected to manual, as well as automated refinement 42,43 EMPIAR with the accession code EMPIAR-10457. The data sets used during the current using phenix.real_space_refine tool, as implemented in phenix with non- 8 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 ARTICLE study are available from the corresponding author on reasonable request. The unique 27. Zhang, W. et al. Heparin-induced tau filaments are polymorphic and differ materials used in the current study are available from the authors on reasonable from those in Alzheimer’s and Pick’s diseases. Elife 8, e43584 (2019). request. Source data are provided with this paper. 28. Kollmer, M. et al. Cryo-EM structure and polymorphism of Aβ amyloid fibrils purified from Alzheimer’s brain tissue. Nat. Commun. 10.1,1–8 (2019). 29. Gallardo, R., Ranson, N. E. & Radford, S. E. Amyloid structures: much more Received: 5 August 2020; Accepted: 21 December 2020; than just a cross-β fold. Curr. Opin. Struct. Biol. 60,7–16 (2020). 30. Ferkinghoff-Borg, J. et al. Stop-and-go kinetics in amyloid fibrillation. Phys. Rev. E 82, 010901 (2010). 31. Ndlovu, H., Ashcroft, A. E., Radford, S. E. & Harris, S. A. Molecular dynamics simulations of mechanical failure in polymorphic arrangements of amyloid fibrils containing structural defects. Beilstein J. Nanotechnol. 4, 429–440 (2013). References 32. Knowles, T. P., Smith, J. F., Craig, A., Dobson, C. M. & Welland, M. E. Spatial 1. Quock, T. P., Yan, T., Chang, E., Guthrie, S. & Broder, M. S. Epidemiology of persistence of angular correlations in amyloid fibrils. Phys. Rev. Lett. 96, AL amyloidosis: a real-world study using US claims data. Blood Adv. 2, 238301 (2006). 1046–1053 (2018). 33. Knowles, T. P., Smith, J. F., Devlin, G. L., Dobson, C. M. & Welland, M. E. 2. Shimazaki, C. Current diagnosis and treatment of AL amyloidosis in Japan: a Analysis of structural order in amyloid fibrils. Nanotechnology 18, 044031 nationwide epidemiological survey. Rinsho Ketsueki 60.8, 973–978 (2019). (2006). 3. Wisniowski, B. et al. The epidemiology of amyloidosis in Queensland, 34. Andersen, C. B. et al. Branching in amyloid fibril growth. Biophys. J. 96, Australia. Br. J. Haematol. 186.6, 829–836 (2019). 1529–1536 (2009). 4. Merlini, G. et al. Systemic immunoglobulin light chain amyloidosis. Nat. Rev. 35. Gao, X. et al. Human Hsp70 disaggregase reverses Parkinson’s-linked α- Dis. Prim. 4,1–19 (2018). synuclein amyloid fibrils. Mol. Cell 59, 781–793 (2015). 5. Blancas-Mejía, L. M. & Ramirez-Alvarado, M. Systemic amyloidoses. Annu. 36. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Rev. Biochem 82, 745–774 (2013). Nat. Methods 9, 676–682 (2012). 6. Perfetti, V. et al. The repertoire of lambda light chains causing predominant 37. He, S. & Scheres, S. H. W. Helical reconstruction in RELION. J. Struct. Biol. amyloid heart involvement and identification of a preferentially involved 198, 163–176 (2017). germline gene, IGLV1-44. Blood 119, 144–150 (2012). 38. Kremer, J. R., Mastronarde, D. N. & McIntosh, J. R. Computer visualization of 7. Kourelis, T. V. et al. Clarifying immunoglobulin gene usage in systemic and three-dimensional image data using IMOD. J. Struct. Biol. 116,71–76 (1996). localized immunoglobulin light-chain amyloidosis by mass spectrometry. 39. Zheng, S. Q. et al. MotionCor2: anisotropic correction of beam-induced Blood 129, 299–306 (2017). motion for improved cryo-electron microscopy. Nat. Methods 14, 331–332 8. Vaxman, I., Dispenzieri, A., Muchtar, E. & Gertz, M. New developments in (2017). diagnosis, risk assessment and management in systemic amyloidosis. Blood 40. Zhang, K. Gctf: Real-time CTF determination and correction. J. Struct. Biol. Rev. 40, 100636 (2020). 193,1–12 (2016). 9. Bellotti, V., Mangione, P. & Merlini, G. Review: immunoglobulin light chain 41. Emsley, P., Lohkamp, B., Scott, W. G. & Cowtan, K. Features and development amyloidosis—the archetype of structural and pathogenic variability. J. Struct. of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501 (2010). Biol. 130, 280–289 (2000). 42. Afonine, P. V. et al. New tools for the analysis and validation of cryo-EM 10. Annamalai, K. et al. Common fibril structures imply systemically conserved maps and atomic models. Acta Crystallogr. D Struct. Biol. 74, 814–840 (2018). protein misfolding pathways in vivo. Angew. Chem. Int. Ed. 56.26, 7510–7514 43. Afonine, P. V. et al. Real-space refinement in PHENIX for cryo-EM and (2017). crystallography. Acta Crystallogr. D Struct. Biol. 74, 531–544 (2018). 11. Annamalai, K. et al. Polymorphism of amyloid fibrils in vivo. Angew. Chem. 44. Williams, C. J. et al. MolProbity: more and better reference data for improved Int. Ed. 55.15, 4822–4825 (2016). all atom structure validation. Protein Sci. 27.1, 293–315 (2018). 12. Glenner, G. G., Harbaugh, J., Ohms, J. I., Harada, M. & Cuatrecasas, P. An 45. Retter, I., Althaus, H. H., Munch, R. & Muller, W. VBASE2, an integrative V amyloid protein: the amino-terminal variable fragment of an immunoglobulin gene database. Nucleic Acids Res. 33, D671–D674 (2005). light chain. Biochem. Biophys. Res. Commun. 41, 1287–1289 (1970). 46. Altschul, S. F., Gish, W., Miller, W., Myers, E. W. & Lipman, D. J. Basic local 13. Radamaker, L. et al. Cryo-EM structure of a light chain-derived amyloid fibril alignment search tool. J. Mol. Biol. 215, 403–410 (1990). from a patient with systemic AL amyloidosis. Nat. Commun. 10.1,1–8 (2019). 47. Hunt, S. E. et al. Ensembl variation resources. Database 2018, bay119 (2018). 14. Swuec, P. et al. Cryo-EM structure of cardiac amyloid fibrils from an 48. Ruiz, M. et al. IMGT, the international ImMunoGeneTics database. Nucleic immunoglobulin light chain AL amyloidosis patient. Nat. Commun. 10.1,1–9 Acids Res. 28, 219–221 (2000). (2019). 49. Tamura, K., Stecher, G., Peterson, D., Filipski, A. & Kumar, S. MEGA6: 15. Boyer, D. R. et al. Structures of fibrils formed by α-synuclein hereditary molecular evolutionary genetics analysis version 6.0. Mol. Biol. Evol. 30, disease mutant H50Q reveal new polymorphs. Nat. Struct. Mol. Biol. 26, 2725–2729 (2013). 1044–1052 (2019). 50. Sievers, F. et al. Fast, scalable generation of high-quality protein multiple 16. Cao, Q., Boyer, D. R., Sawaya, M. R., Ge, P. & Eisenberg, D. S. Cryo-EM sequence alignments using Clustal Omega. Mol. Syst. Biol. 7, 539 (2011). structures of four polymorphic TDP-43 amyloid cores. Nat. Struct. Mol. Biol. 51. Fernandez-Escamilla, A. M., Rousseau, F., Schymkowitz, J. & Serrano, L. 26, 619–627 (2019). Prediction of sequence-dependent and mutational effects on the aggregation 17. Blancas-Mejia, L. M. et al. Immunoglobulin light chain amyloid aggregation. of peptides and proteins. Nat. Biotechnol. 22, 1302–1306 (2004). Chem. Commun. 54.76, 10664–10674 (2018). 52. Garbuzynskiy, S. O., Lobanov, M. Y. & Galzitskaya, O. V. FoldAmyloid: a 18. Kazman, P. et al. Fatal amyloid formation in a patient’s antibody light chain is method of prediction of amyloidogenic regions from protein sequence. caused by a single point mutation. Elife 9, e52300 (2020). Bioinformatics 26, 326–332 (2010). 19. Wang, J. & Moore, P. B. On the interpretation of electron microscopic maps of 53. Conchillo-Solé, O. et al. AGGRESCAN: a server for the prediction and evaluation biological macromolecules. Protein Sci. 26.1, 122–129 (2017). of “hot spots” of aggregation in polypeptides. BMC Bioinformatics 8, 65 (2007). 20. Hora, M. et al. MAK33 antibody light chain amyloid fibrils are similar to 54. Walsh, I., Seno, F., Tosatto, S. C. & Trovato, A. PASTA 2.0: an improved server oligomeric precursors. PLoS ONE 12.7, e0181799 (2017). for protein aggregation prediction. Nucleic Acids Res. 42, W301–W307 (2014). 21. Piehl, D. W., Blancas-Mejía, L. M., Ramirez-Alvarado, M. & Rienstra, C. M. 55. Pettersen, E. F. et al. UCSF Chimera—a visualization system for exploratory Solid-state NMR chemical shift assignments for AL-09 V L immunoglobulin research and analysis. J. Comput. Chem. 25, 1605–1612 (2004). light chain fibrils. Biomol. NMR Assign. 11,45–50 (2017). 56. Blech, M. et al. Molecular structure of human GM-CSF in complex with a 22. Pradhan, T. et al. Seeded fibrils of the germline variant of human λ-III disease-associated anti-human GM-CSF autoantibody and its potential immunoglobulin light chain FOR005 have a similar core as patient fibrils with biological implications. Biochem. J. 447.2, 205–215 (2012). reduced stability. J. Biol. Chem. 295, 18474–18484 (2020). 23. Lecoq, L. et al. A substantial structural conversion of the native monomer leads to in-register parallel amyloid fibril formation in light-chain amyloidosis. Chembiochem 20.8, 1027–1031 (2019). Acknowledgements 24. Ferrone, F. A., Hofrichter, J. & Eaton, W. A. Kinetics of sickle hemoglobin We would like to thank the Deutsche Forschungsgemeinschaft for funding of the polymerization: I. Studies using temperature-jump and laser photolysis Research Unit FOR 2969, projects FA 456/27, HA 7138/3, HE 8472/1-1, HU 2400/1-1, techniques. J. Mol. Biol. 183, 591–610 (1985). and SCHO 1364/2-1. We are grateful to Prof. Dr. Bernd Reif (Technical University 25. Crowther, R. A. Straight and paired helical filaments in Alzheimer disease have Munich) for helpful discussions and to the students who participated in the practical a common structural unit. Proc. Natl Acad. Sci. USA 88, 2288–2292 (1991). course “Protein Biochemistry and Structural Biology” in the winter semester 2019, for 26. Makarava, N., Ostapchenko, V. G., Savtchenko, R. & Baskakov, I. V. assisting in the manual picking of the fibril segments. All cryo-EM data were collected at Conformational switching within individual amyloid fibrils. J. Biol. Chem. the European Molecular Biology Laboratory, Heidelberg (Germany), funded by iNEXT 284, 14386–14395 (2009). (Horizon 2020, European Union). NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-21126-2 Author contributions Reprints and permission information is available at http://www.nature.com/reprints L.R. and J.B. carried out experiments. L.R, J.B., S.H., C.H., A.B., M.S., and M.F. analyzed data. U.H. and S.S. contributed tools and reagents. M.F. designed research. L.R. and M.F. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in wrote the paper. published maps and institutional affiliations. Funding Open Access This article is licensed under a Creative Commons Open Access funding enabled and organized by Projekt DEAL. Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give Competing interests appropriate credit to the original author(s) and the source, provide a link to the Creative The authors declare no competing interests. Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the Additional information article’s Creative Commons license and your intended use is not permitted by statutory Supplementary information The online version contains supplementary material regulation or exceeds the permitted use, you will need to obtain permission directly from available at https://doi.org/10.1038/s41467-021-21126-2. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Correspondence and requests for materials should be addressed to M.F. licenses/by/4.0/. Peer review information Nature Communications thanks Luis del Pozo-Yauner and the © The Author(s) 2021 other, anonymous, reviewer(s) for their contribution to the peer review of this work. 10 NATURE COMMUNICATIONS | (2021) 12:875 | https://doi.org/10.1038/s41467-021-21126-2 | www.nature.com/naturecommunications

Journal

Nature CommunicationsSpringer Journals

Published: Feb 8, 2021

There are no references for this article.