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

Learn More →

Molecular mechanism of amyloidogenic mutations in hypervariable regions of antibody light chains

Molecular mechanism of amyloidogenic mutations in hypervariable regions of antibody light chains RESEARCH ARTICLE EDITORS’ PICK Molecular mechanism of amyloidogenic mutations in hypervariable regions of antibody light chains Received for publication, September 3, 2020, and in revised form, January 14, 2021 Published, Papers in Press, January 26, 2021, https://doi.org/10.1016/j.jbc.2021.100334 1 1 1 1 1 Georg J. Rottenaicher , Benedikt Weber , Florian Rührnößl , Pamina Kazman , Ramona M. Absmeier , 2 2 1, Manuel Hitzenberger , Martin Zacharias , and Johannes Buchner * 1 2 From the Center for Integrated Protein Science Munich at the Department Chemie, Center for Integrated Protein Science Munich at the Physik-Department, Technische Universität München, Garching, Germany Edited by Ursula Jakob Systemic light chain (AL) amyloidosis is a fatal protein linked by disulfide bridges. Each of the LCs is made up of an misfolding disease in which excessive secretion, misfolding, N-terminal variable (V ) domain and a C-terminal constant and subsequent aggregation of free antibody light chains (C ) domain (5). In AL amyloidosis, malignant monoclonal eventually lead to deposition of amyloid plaques in various plasma cells overproduce and secrete LCs into the blood organs. Patient-specific mutations in the antibody V domain stream leading to very high concentrations of circulating LCs are closely linked to the disease, but the molecular mechanisms (6, 7). The malignant plasma cells often emerge in the course by which certain mutations induce misfolding and amyloid of an underlying plasma cell dyscrasia (e.g., multiple myeloma) aggregation of antibody domains are still poorly understood. (7). During the complex maturation process responsible for Here, we compare a patient V domain with its non- the creation of antibody binding diversity, these LCs acquire amyloidogenic germline counterpart and show that, out of the amyloidogenic point mutations mostly in the V domain, five mutations present, two of them strongly destabilize the which is the main constituent of fibrils in AL amyloidosis protein and induce amyloid fibril formation. Surprisingly, the (8–10). These point mutations and (in many cases) proteolytic decisive, disease-causing mutations are located in the highly cleavage of the LC to produce the free V domain are key variable complementarity determining regions (CDRs) but factors for disease onset and progression (11–14). Yet, the exhibit a strong impact on the dynamics of conserved core exact mechanism by which certain mutations favor amyloid regions of the patient V domain. This effect seems to be based formation of antibody domains remains largely elusive. Recent on a deviation from the canonical CDR structures of CDR2 and cryo-EM and solid-state NMR studies of AL amyloid fibrils CDR3 induced by the substitutions. The amyloid-driving mu- show that the fold of the fibril core is strikingly different from tations are not necessarily involved in propagating fibril for- the native Ig fold suggesting that a complete structural rear- mation by providing specific side chain interactions within the rangement has to occur in the process of fibril formation fibril structure. Rather, they destabilize the V domain in a (15–20). For a large number of AL cases, it has been shown specific way, increasing the dynamics of framework regions, that the decrease in thermodynamic stability of the V domain which can then change their conformation to form the fibril is a decisive factor for amyloidogenicity (21–25). However, also core. These findings reveal unexpected influences of CDR- nondestabilizing mutations in the V domain can induce fibril framework interactions on antibody architecture, stability, formation, thus other factors such as LC dimerization, struc- and amyloid propensity. tural changes, and conformational dynamics also need to be taken into account (26–30). A major enigma in this context is how different mutations Amyloidoses comprise a family of protein misfolding dis- shift V domains toward the fibrillary pathway, especially if eases in which disease-specific precursor proteins aggregate these mutations are not in the conserved framework but in the into highly ordered amyloid fibrils (1). These fibrils form variable antigen binding regions called complementarity amyloid plaques, which are deposited either systemically or in determining regions (CDRs). These are solvent-exposed loops an organ-specific manner causing severe damage (2). The most connecting β-strands (31). In order to recognize a large variety common systemic disease in this context is amyloid light chain of antigens, CDRs need to tolerate a high sequence variability, (AL) amyloidosis, in which an antibody light chain (LC) acts as which in turn suggests that CDR point mutations should not the precursor protein that eventually forms amyloid fibers (3, strongly affect the thermodynamic stability and aggregation 4). In healthy individuals, plasma cells secrete IgG antibodies, propensity of the antibody domain (32–35). In 2017, however, which consist of two LCs and two heavy chains covalently Annamalai et al. (36) reported the crystal structure and the fibril morphology of a V domain (FOR005-PT) obtained from an AL amyloidosis patient with mainly cardiac involvement in This article contains supporting information. * which four out of the five mutations are located in the CDRs For correspondence: Johannes Buchner, johannes.buchner@tum.de. Present address for Benedikt Weber: Roche Diagnostics GmbH, Nonnenwald (according to the Kabat/Chothia domain numbering). Since 2, Penzberg 82377, Germany. J. Biol. Chem. (2021) 296 100334 1 © 2021 THE AUTHORS. Published by Elsevier Inc on behalf of American Society for Biochemistry and Molecular Biology. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis current models cannot explain the amyloidogenic character of segment containing the mutations G49R and N51S. The fifth this variant, we set out to determine which of these mutations mutation, G94A, is located in the hypervariable CDR3 loop. drive amyloid aggregation and found that specifically two of Residue conservation analysis with Consurf revealed the lowest the CDR mutations are causative for fibril formation. Our degree of conservation (Consurf score = 1) for all four CDR findings further show that seemingly minor side chain alter- mutations and an average conservation degree (Consurf ations, even in poorly conserved CDRs, can destabilize the score = 5) for the framework residue F48 in the patient entire V domain and drive it toward misfolding and amyloid sequence (Fig. S1)(45). We further assessed aggregation-prone aggregation. regions and individual mutational effects by applying the prediction tools AmylPred2, MetAmyl, and ZipperDB (46–48). Results Amyloidogenicity predictions by these tools did not suggest Sequence and structure analysis significant alterations of the amyloid aggregation or steric zipper propensity of the patient V sequence in comparison In 2017, Annamalai et al. (36) reported the cDNA sequence with its corresponding germline sequence (Fig. 1A). To explain and crystal structure (PDB: 5L6Q) of an amyloid forming V structural effects of mutations in more detail, we created a domain (FOR005-PT) derived from a patient with cardiac LC homology model of FOR005-GL based on the template amyloidosis. We used IgBLAST, IMGT, and abYsis to deter- structure 5BV7, which exhibits 98.2% sequence identity, using mine the corresponding germline sequence (FOR005-GL) with the SWISS-MODEL web server (49). Structural alignments of the highest possible protein sequence identity for this amy- the homology model with the crystal structure of FOR005-PT loidogenic V (37–40). FOR005-PT belongs to the λ3l LC (PDB: 5L6Q) showed that the overall structure was conserved, subfamily (gene segments: IGLV3-19/IGLJ2). The related although the conformations of the CDR2 and CDR3 loops are germline λ3r has been reported to be associated with AL altered (Fig. 1B). However, one needs to take into account that amyloidosis (41, 42). Five point mutations were identified in homology models are merely an approximation of the actual the patient-derived V domain (Y31S, Y48F, G49R, N51S, native protein structure. G94A) compared with the germline sequence (Fig. 1A), but it was not clear which mutation causes amyloid aggregation. FOR005-PT and FOR005-GL differ substantially in fibril Four of them are located in the hypervariable CDRs according formation propensity and thermodynamic stability to the Kabat and Chothia numbering systems (43, 44). The mutation Y31S is located in the CDR1 region, Y48F lies in the To test the biophysical properties of the proteins directly, conserved framework 2 region (FR2) right next to the begin- we produced patient and germline V domains recombinantly ning of CDR2. The CDR2 comprises a short, protruding loop in E.coli and purified them to homogeneity. The far-UV Figure 1. Sequence and structural analyses of FOR005-PT and FOR005-GL. A, sequence alignment of patient and germline V shows the five point mutations highlighted in red and the variable CDR loops in cyan (CDR1), blue (CDR2), and green (CDR3). Predictions of amyloidogenic regions by three different tools overlap well indicating that the point mutations do not introduce new amyloid driving segments. Aggregation-prone positions are indicated by asteriks. The sequence numbering as derived from Annamalai et al. starts with the first serine residue, Ser1. The N-terminal glycine in our sequence results from using NcoI during subcloning of the FOR005 gene constructs and is, therefore, numbered as Gly0. B, structural alignment of FOR005-PT shown in black (PDB: 5L6Q) with the homology model derived for FOR005-GL depicted in gray. The homology model was created using the SWISS-MODEL server and the template structure 5BV7. CDRs are colored according to the sequence alignment in A. Mutated positions are shown in red on the patient V and light red on the germline V domain with side chains depicted as sticks. For F48 on the patient structure two rotamers are shown. The structural comparison suggests rearrangements of loop conformations in CDR2 and CDR3. 2 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis circular dichroism (CD) spectra of the purified proteins thiazol-based fluorescent dye Thioflavin T (ThT), which showed that both are properly folded and possess the typical β- specifically binds to the characteristic cross-β motif in amy- sheet-rich immunoglobulin fold as indicated by the minimum loid fibrils (51). ThT-binding kinetics showed that the patient at around 218 nm in the far-UV region (Fig. S2)(50). Near-UV V domain starts to form amyloid fibrils in vitro after CD spectra, which represent a specific tertiary structure approximately 3 days, whereas the corresponding germline fingerprint, were highly similar for the two proteins. Thus, protein does not engage in amyloid aggregation (Fig. 2A, FOR005-PT and FOR005-GL seem to have nearly identical Table 1). To obtain direct evidence for the presence of fibrils tertiary structure and topology. Additionally, analytical ultra- in the samples, we performed transmission electron micro- centrifugation (AUC) was performed to assess the quaternary scopy (TEM). The TEM micrographs showed fibrils only in structure. As indicated by sedimentation coefficients of 1.52 the patient V sample and not in the germline control and 1.59 S, respectively, both the patient and germline V (Fig. 2B). Thus, the patient V behaved as expected and the L L domains are monomeric in solution (Fig. S2). germline protein does not show amyloidogenic behavior. To test whether the two V domains differ in their fibril FOR005 fibrils isolated from patient tissue contained only the formation propensities, we incubated the proteins in phos- V domain (35). Since the role of proteolytic cleavage of phate buffered saline (PBS) at pH 7.4 and 37 C under precursor LCs in amyloidosis is still only poorly understood continuous shaking and monitored fibril formation via the (14), we purified full-length LCs of the patient and germline Figure 2. Fibril formation propensity of the patient V domain. A, thioflavin T-binding kinetics of FOR005-PT (black), -GL (red), -PLC (light green), and -GLC (light purple) obtained at 37 C and pH 7.4 under continuous shaking. The increase in fluorescence shows that the patient V domain is the only protein engaging in amyloid fibril formation after approximately 3 days. Connecting the patient V domain with the C domain to form a full-length LC completely L L inhibited fibril formation. All kinetic curves were normalized to a fluorescence start value of 1. B, TEM micrographs of samples from finished ThT assays were recorded after negative stain with uranyl acetate. The amyloid fibers of FOR005-PT can be seen in the upper left panel, the scale bar represents 200 nm. C, for chemical unfolding transitions, 1 μM protein was equilibrated with increasing concentrations of urea over night at room temperature. Fluorescence spectra (λ = 280 nm/λ = 300–400 nm) were recorded at 25 C in a 96-well plate. The transition of FOR005-PT is shown with black dots, the data for FOR005-GL is ex em shown as red dots. The black and red sigmoidal lines represent the individual fit functions. D, thermal unfolding transitions of FOR005-PT and -GL were obtained by recording CD signal at 205 nm while applying a temperature gradient from 20 to 90 C with a heating rate of 1 C/min. Sample concentration was 10 μM in PBS and the measurement was performed in a 1 mm quartz cuvette. J. Biol. Chem. (2021) 296 100334 3 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Table 1 Stability parameters, unfolding cooperativity, and fibril formation midpoints of FOR005 constructs. T C mt pH 7.4 t pH 6.4 m m 50 50 −1 −1 V domain C M Urea kJ mol M dd FOR005-PT 43.5 ± 0.14 1.90 ± 0.02 5.99 ± 0.33 3.8 2.8 FOR005-GL 56.3 ± 0.11 4.28 ± 0.04 4.33 ± 0.31 - - GL Y31S 56.1 ± 0.12 4.20 ± 0.03 5.12 ± 0.47 - - GL Y48F 55.4 ± 0.13 4.15 ± 0.03 4.56 ± 0.32 - - GL G49R 52.2 ± 0.10 2.91 ± 0.01 5.55 ± 0.13 - - GL N51S 54.4 ± 0.10 3.44 ± 0.02 5.01 ± 0.27 - - GL G94A 50.5 ± 0.17 3.25 ± 0.05 4.95 ± 0.99 - - GL Y31S/G94A 50.3 ± 0.15 3.31 ± 0.03 5.17 ± 0.58 - - GL Y48F/G94A 51.9 ± 0.10 3.22 ± 0.02 5.32 ± 0.28 - - GL G49R/G94A 47.0 ± 0.09 2.09 ± 0.02 6.69 ± 0.41 8.9 4.4 GL N51S/G94A 48.6 ± 0.08 2.56 ± 0.02 5.98 ± 0.33 - 11.5 Thermal transitions were obtained by recording the CD signal at 205 nm between 20 to 90 C at a heating rate of 1 C/min. Chemical unfolding transitions were obtained by fluorescence spectroscopy using 1 μM of each V domain with increasing concentrations of urea. Since both thermal and chemical unfoldings are irreversible, the stability parameters T and C represent apparent values. Transition midpoints and standard deviations were derived from a Boltzmann fit. Chemical unfolding data was also subjected to m m app a two-state unfolding fit model to determine cooperativity and ΔG values (Table S1). Fibril formation assays were carried out at 37 C, pH 7.4 or 6.4 under continouos shaking un in a Tecan Genios platereader. The t values represent the time point at which fibril formation is 50% completed. variants (FOR005-PLC and FOR005-GLC, respectively) to Mutations in hypervariable regions affect domain stability determine whether the patient LC is also amyloidogenic. We and aggregation performed fibril formation assays and transition electron While the results described above show that the patient- microscopy and found that both LCs did not engage in the specific mutations affect conformational stability and fibril amyloidogenic pathway (Fig. 2). formation, it was not possible to rationalize which of the To determine differences between the two V domains mutations are responsible for amyloidogenesis. To determine regarding their thermodynamic propertiesinmoredetail, we the specific effects of each of the five point mutations, we investigated their stabilities by chemical and thermal dena- replaced them individually in the germline sequence by the turation experiments. Unfolding transitions in the presence respective patient residues (Y31S, Y48F, G49R, N51S, and of increasing urea concentrations were performed to assess G94A). CD spectroscopy and AUC analysis of the mutants the chemical domain stability and unfolding cooperativity of showed that all point mutants adopted the conserved β-sheet the V domains (Fig. 2, Table 1). The patient V domain L L structure and were monomeric in solution (Fig. S3). Addi- showed a midpoint of unfolding at a urea concentration (C ) tionally, highly similar near-UV CD spectra suggest that the of 1.90 M, whereas for the germline domain, the midpoint is amino acid substitutions have only minor effects on the global at 4.28 M urea (Fig. 2, Table 1). We assessed the reversibility tertiary structure of the antibody domain (Fig. S3). of urea-induced unfolding by fluorescence spectroscopy and Thermal unfolding experiments of the germline V domain found that both V domains cannot be completely refolded constructs containing the individual patient mutations showed into their native structure within 24 h at room temperature the largest stability decrease for the G94A mutant with a T (Fig. S2). It should be noted, however, that the germline V value of 50.5 C and the second largest effect for the G49R exhibits a higher degree of unfolding reversibility than the mutant with a transition temperature of 52.2 C. The thermal patient variant (Fig. S2). We applied a two-state fitmodel to stabilities of the remaining mutants Y31S, Y48F, and N51S our transition data to calculate unfolding free energies were only slightly decreased with transition midpoints tem- (ΔG ). However, this is in principle only possible if un peratures of 56.1 C, 55.4 C, and 54.4 C, respectively unfolding is completely reversible. Since this is not the case (Table 1, Fig. S3). In the case of chemical unfolding, the under the conditions used (Fig. S2), these results do not strongest decrease in stability was observed for the G49R represent true ΔG values but are rather apparent unfolding un mutant with a C value of 2.91 M urea, whereas the G94A app free energies (ΔG )(Table S1). un variant unfolded at a concentration of 3.25 M urea. Both the app In thermal denaturation experiments, the patient-derived G49R and G94A variant show comparable ΔG values of un protein exhibited a melting temperature (T ) of 43.5 C, 16.49 kJ/mol and 16.18 kJ/mol, respectively (Table S1). Again, whereas the germline protein showed a melting temperature of the transition midpoints of the Y31S (4.2 M) and Y48F 56.3 C. The T values correspond to the temperatures at (4.15 M) mutants lie only slightly below that of the germline which 50% of the protein is unfolded. Since thermal unfolding reference, while the N51S variant unfolded at 3.44 M urea of both FOR005-PT and FOR005-GL is also irreversible (Table 1, Fig. S3). (Fig. S2), the obtained transition midpoints represent apparent Among the five single mutations, G49R and G94A exerted melting temperatures (Table 1). These data show that the the strongest destabilizing effect on the germline V domain. patient V domain has a significantly decreased thermody- Since G94A is a small, conservative mutation located in the namic stability compared with its germline counterpart hypervariable CDR3 loop, these results were unexpected. (Table 1). Therefore, we created double mutants by individually 4 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis combining G94A with the remaining four mutations yielding germline counterpart pointing toward a higher degree of the double mutations Y31S/G94A, Y48F/G94A, G49R/G94A, conformational dynamics (Fig. 4A, Fig. S5). Further, the single- and N51S/G94A. The largest effect on thermal stability was point mutants G49R and G94A behave similarly to the observed for the double mutants G49R/G94A and N51S/ germline V domain and exhibit overall slow degradation ki- G94A. These exhibited severely decreased thermal stabilities netics. However, G94A is processed faster and to a greater with melting temperatures of 47.0 C and 48.6 C, respectively. extent than G49R and FOR005-GL. The double mutant N51S/ Accordingly, also in terms of chemical stability, the mutations G94A is also cleaved much faster than the germline and the G49R/G94A and N51S/G94A had the most significant effect observed single mutants, yet not as fast as the patient-derived with transition midpoints of 2.09 M and 2.56 M urea, V domain. Interestingly, the double mutant G49R/G94A is respectively (Table 1, Fig. S3). degraded even more readily than the patient V domain Furthermore, ThT-binding kinetics and TEM micrographs FOR005-PT (Fig. 4A, Fig. S5). revealed that G49R/G94A is the only mutant that forms H/DX-MS was applied to gain more detailed insights into amyloid fibrils in vitro at pH 7.4 and 37 C(Fig. 3A). The the conformational dynamics of both the patient and germ- N51S/G94A mutant, however, did not form fibrils within line V domain. This method is based on the enhanced sol- 2 weeks at pH 7.4, despite exhibiting significantly decreased vent exchange rates of backbone amide hydrogens in flexible thermodynamic stability, similarly to G49R/G94A. It has protein regions from which peptide-resolved dynamic infor- been shown that destabilization is not necessarily the only mation can be derived after pepsin cleavage and mass spec- driving force in the amyloid formation pathway and that trometric analyses (60). Fractional deuterium uptake was protein dynamics and population of nonnative intermediate determined for FOR005-PT, FOR005-GL, G49R, G94A, as states can play important roles, too (26, 27, 52). To further well as the double mutants G49R/G94A and N51S/G94A. investigate the involvement of these molecular traits, addi- The fold change in fractional uptake was calculated by tional fibril formation assays were performed at pH 6.4, since dividing uptake ratios of the investigated mutants by the acidification can lead to a decrease in stability and population uptake ratios of the germline V domain (Fig. 4B). A value of alternatively folded intermediate states (53–55). As ex- below 1 indicates that the germline exhibits higher deuterium pected, in ThT assays carried out at pH 6.4, fibril formation uptake, whereas a value above 1 shows increased deuterium was accelerated for FOR005-PT and G49R/G94A, but also for uptake for the observed mutant. Conformational dynamics N51S/G94A fibril formation was observed after approxi- are especially pronounced for residues 12 to 20, residues 65 mately 10 days (Fig 3B). Thepresenceofamyloid fibrils in the to 85, and residues 97 to 105 in the case of FOR005-PT and ThT assay was confirmed by TEM micrographs (Fig. 3C). for the double mutant G49R/G94A (Fig. 4B). The double These findings imply that the amyloid aggregation of mutation, however, still exhibits slightly lower flexibility FOR005-PT relies on a mechanism in which domain desta- compared with the actual patient V that contains all five bilization is an important, yet not the only decisive bio- substitutions. Interestingly, residues 50 to 60 including the physical factor. CDR2 loop are moredynamic in thegermlineV domain and Overall, the results for the single and double mutations in G49R/G94A. The double mutant N51S/G94A also exhibits show that out of the five mutations two are mainly responsible lightly increased dynamic behavior, especially for residues 80 for the significant loss in thermodynamic stability and the gain to 105, whereas the single mutants G49R and G94A, in in amyloid formation propensity. Remarkably, the conservative comparison, do not impose a strong increase in conforma- G94A mutation in the exposed CDR3 loop has a strong impact tional flexibility. Notably, G94A has a slightly larger effect on on the biophysical properties of the V domain, despite being overall dynamics than G49R (Fig. 4B). To better visualize located in the most variable part of the protein. which parts of the patient-derived V domain experience enhanced dynamics in comparison with the germline, the change in fractional uptake was plotted onto the crystal Conformational dynamics are linked to decreased protein structure of FOR005-PT (PDB: 5L6Q). Structurally, the most stability and amyloid formation affected regions correspond to β-strands A2 and B and the Previous studies have demonstrated that there is a causal small loop connecting them (residues 12–20), β-strands E link between conformational dynamics and aggregation pro- and F including the small helical segment between them pensity, as well as cellular toxicity of prefibrillar species (27, (residues 65–85), and the C-terminal β-strands G1 and G2 56–58). Therefore, we set out to investigate the structural (Fig. 4C). In summary, our results show that there is a clear dynamics and flexibility of patient and germline V domains by connection between conformational dynamics and amyloid limited proteolysis and hydrogen/deuterium exchange mass aggregation. Remarkably, we observe the strongest increase spectrometry (H/DX-MS). in dynamics in conserved framework regions rather than the Limited proteolysis allows obtaining information about segments where the point mutations are located. These structural flexibility since proteolytic degradation is increased findings suggest that small mutation-induced changes in due to enhanced protein dynamics and local unfolding (59). CDR loop conformations might propagate through the entire When we carried out limited proteolysis experiments with the domain architecture and thereby lead to increased dynamics proteases trypsin or proteinase K, we found that the patient- in framework regions, which causes lower stability and derived V domain was degraded much faster than its enhanced aggregation propensity. J. Biol. Chem. (2021) 296 100334 5 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Unfavorable main chain conformations in the CDR2 and CDR3 Whether proteolytic processing of LC precursors is a pre- loops destabilize the V domain requisite or a consequence of amyloid formation still remains enigmatic (12, 14, 30, 61). It has been shown that the C To gain further insight into the flexibility of the patient V domain can exert a protective function in vitro and that full- domain, molecular dynamics (MD) simulations were per- length LCs do not readily aggregate into amyloid fibrils formed in explicit solvent on the FOR005-GL, the GL G49R/ (56, 62, 63). In the case of FOR005, the V domain was G94A, the GL N51S/G94A, the FOR005-PT V and a PT identified as the sole component of amyloid deposits in the variant containing the R49G and A94G double substitution. On patient’s tissue (36). Accordingly, in vitro only the patient- the timescale of 1 μs, the variants were stable during the MD derived V domain but not the corresponding full-length LC simulations and exhibited only small and similar deviations formed amyloid fibrils, thus implying a protective role of the from the start structure (Fig. S6). Comparison of root-mean- C domain. square fluctuations (RMSF) indicated the lowest fluctuations FOR005 is an interesting case, as the V domain contains for FOR005-GL, slightly enhanced fluctuations for GL N51S/ four CDR mutations and only one framework mutation G94A, and significantly increased fluctuations especially compared with its germline counterpart. Of note, the exact around residue 49 and 94 in case of the GL G49R/G94A variant location of the CDRs depends on which domain numbering (Fig. 5A). Slightly larger conformational fluctuations on the MD system is used. The three systems according to Kabat, Chothia, timescale were also observed for the FOR005-PT variant and IMGT are the most common ones (38, 43, 44). When the compared with the PT R49G/A94G mutation (Fig. S6A). IMGT numbering scheme is applied to FOR005, the substi- The inspection of the peptide backbone dihedral angles in tution N51S would be considered a framework mutation the loop regions near residue 49 and 94 revealed the sampling rather than a CDR2 mutation. However, the Kabat and Cho- of the left-handed helical regimes in the Ramachandran plots thia classifications identify this residue as belonging to the of residues 49 and 50 as well as 94 and 95 (but not for residues CDR and this coincides with the Consurf residue conservation 51 or 52, Fig. 5, B–D). This regime is sterically favorable in case analysis (Fig. S1). Furthermore, the identification of a suitable of the glycine but less so for nonglycine residues. Hence, R49 germline sequence for a given V domain can yield different or A94 creates steric strain in the loop structure, whereas G49 results depending on which method/database is used. We or G94 relaxes this strain. Also, in the case of FOR005-PT, the applied abYsis, IgBLAST, and IMGT to identify a V domain MD simulations reveal sampling of sterically unfavorable with highest possible amino acid sequence identity (38–40). peptide backbone states that—in a relaxed peptide structure— The most important practical test for the germline sequence of are typically only adopted by glycine residues (Fig. S6B). choice is whether it forms fibrils, as this allows to identify and Notably, these unfavorable backbone states are also observed test the effect of the patient mutations concerning their in the crystal structure of the patient protein. Hence, in the amyloidogenic potential (56). patient structure, the loop forces its residues at least partially Up to now, mostly framework mutations have been reported into an energetically unfavorable backbone structure upon as key factors in LC amyloid aggregation (22, 24, 26, 56, 62, folding. The “germline” substitutions R49G and A94G can 64–68). Regarding the only framework mutation in FOR005- relax this strain because now the glycine residues at positions PT—Y48F—it has already been shown for a different V 49 and 94 are better compatible with the required backbone domain that this particular mutation has little to no influence structure. Interestingly, the substitutions can also have an ef- on domain stability and aggregation propensity (22). There- fect on neighboring residues and partially modify their back- fore, we hypothesized that only the CDR mutations play a bone sampling (Fig. 5, B–D, Fig. S6, B and C). Energetically, crucial role in the case of FOR005, which was confirmed by the the stress of enforcing unfavorable backbone conformation of experimental results. CDR loops are not only involved in an- residues 49 and 94 can amount to several kcal/mol and hence tigen recognition, they have also been shown to play important could be the reason for the much lower stability of FOR005-PT structural roles in antibody domain architecture and V /V H L and of the variants with nonglycine residues at positions 49 domain association. Various experimental and computational and 94. For residues 51 and 52 approximately the same sam- studies on V domains demonstrated that CDRs can have a pling of favorable backbone states was observed (Fig. 5C) with strong influence on the folding pathway, stability, and no significant effect of the N51S substitution. conformation of the protein (69–72). The involvement of a Discussion CDR mutation in LC amyloidogenicity has been shown for a Systemic LC amyloidosis is a highly complex protein mis- proline residue in the CDR3 loop of an amyloidogenic V folding disease because of the enormous sequence variability of domain. Its deletion resulted in enhanced stability and delayed the soluble precursor protein—the antibody LC. This renders fibril formation kinetics (73). Furthermore, nonconservative the mechanistic understanding of the amyloid aggregation mutations in the V domains of AL and multiple myeloma process a very challenging task. Different, case-dependent or- (MM) patients—also encompassing the CDR3 loops—were gan involvement and a wide spectrum of symptoms further reported to affect the kinetic stability of the LCs (74). Addi- complicate analysis and treatment of this rare disorder (3, 4). tionally, it has been shown that CDR1 can act as a hotspot for Additionally, there are a number of other factors that can aggregation and that a peptide based on part of a CDR3 affect fibril formation, disease onset, and progression, segment can drive amyloid fibril formation due to enhanced including proteolytic processing of the precursor LCs (7, 10). steric zipper propensity (75, 76). However, a detailed 6 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis mechanistic understanding of the effects caused by specific conformation. The changes caused by unfavorable CDR loop CDR residues in the context of the disease is still lacking (77). conformations seem to propagate through the entire protein Multiple studies on substitutions in the V domain inducing increased flexibility, which leads to the enhanced demonstrate that misfolding and amyloid aggregation depend population of partially unfolded, aggregation-competent states on the thermodynamic/kinetic stability, structural dynamics or (52, 78). Therefore, an altered interplay of hypervariable loops partial unfolding, LC dimerization, and local conformational and conserved framework can play a key role in stability and alterations of the native fold (10, 24, 27–29, 78). Thermody- amyloidogenicity of V domains (69, 73). In this context, namic and kinetic stabilities have widely been thought of as the FOR005-PT represents the first case where the onset of fibril major driving force in the misfolding and aggregation pathway formation is directly and mechanistically correlated to the (26, 56, 58, 79). In the case of FOR005, a synergistic combi- substitution of two distinct amino acids in CDR loops. Sur- nation of thermodynamic destabilization and altered confor- prisingly, one of these two decisive substitutions is the small, mational dynamics appears to determine the pathway of the conservative G94A mutation in the surface-exposed CDR3 soluble V monomer toward amyloid fibrils. FOR005-PT and loop. FOR005-GL show a pronounced difference in stability with a Pradhan et al. (19) have recently shown that the R49 res- ΔT of 12.8 C and the mutations G49R and G94A have the idue in FOR005-PT plays a key role in stabilizing the fibril largest impact on domain stability. However, fibril formation core. With this information, it becomes plausible that muta- kinetics and thermodynamic data of the FOR005 double mu- tions in amyloid-forming LCs can serve different purposes. tants suggest that destabilization through CDR mutations is The G94A mutation leads to a conformational change in the not the only driving force in the amyloid formation process, CDR3 loop, which thereby adopts a structure that differs from since the severely destabilized N51S/G94A mutant only forms the canonical CDR class. This conformational change results fibrils after prolonged incubation at lower pH. Additionally, an in enhanced framework dynamics and decreased overall increase in conformational dynamics—mediated by the two domain stability. To illustrate this concept, a structural decisive CDR mutations G49R and G94A—is necessary to alignment of FOR005-PT was performed with three highly induce the amyloid aggregation. Remarkably, the strongest similar, nonamyloidogenic V domains taken from the PDB increase in dynamics is observed in conserved protein core (Fig. S7). The mutation-induced changes in CDR loop regions rather than the loop segments in which the mutations conformation depict the described deviation from the ca- are located. MD simulations indicate that the loop residues 49, nonical CDR class. In the final core structure of FOR005-PT 50, 94, and 95 sample mostly backbone conformations that are fibrils, however, A94 does not play an important role. Seem- energetically unfavorable for nonglycine residues, which ingly, its only effect lies in the destabilization of the precursor lowers the overall stability of the folded structure. This co- V domain. The CDR2 mutation G49R, on the other hand, incides with reports that glycine is structurally preferable at drives amyloidogenesis both by altering CDR2 loop confor- positions with certain u/ψ angles (22, 53). Hence, the interplay mation and by providing a stabilizing side chain interaction in of the CDRs with the framework enforces an energetically the fibril core (19). Yet, as our data show G94A mediates a unfavorable conformation of the loops. These strained loop larger increase in conformational dynamics than G49R, structures affect framework dynamics and are the likely reason especially in the framework 3 region and the C-terminal part for the lower stability of variants with a nonglycine residue at of the domain (Fig. 4). Further, the CDR2 mutation N51S is the corresponding positions. also capable of inducing fibril formation. Thus, the primary This is at first glance counterintuitive, as one might assume role of G49R in the fibril formation pathway of FOR005- that the basic traits of CDRs are their sequence diversity and PT appears to lie in stabilizing the final product of the conformational flexibility that allow them to adapt to the pathway—the core of the amyloid fiber. Individually, however, structure of the antigen upon interaction. However, five out of the two point mutations do not induce fibril formation the six CDRs in an antibody F only adopt a limited number in vitro. Yet in combination, the two CDR mutations G49R ab of backbone conformations, known as canonical classes, with and G94A act synergistically as the obtained stabilities and the heavy chain CDR3 (CDRH3) being the only exception apparent free energies imply (Table 1, Table S1). In summary, (43, 80–82). Therefore, it seems plausible for some CDR the decisive, amyloid-driving mutations are not necessarily mutations in LCs to induce unfavorable loop conformations, involved in propagating fibril formation by providing specific which represent a deviation from the canonical CDR class and side chain interactions within the fibril structure. Rather, they thereby put structural strain on the framework. This deviation destabilize the V domain in a specific way, increasing the can be seen by aligning the crystal structure of the amyloid- dynamics of framework regions, which upon structural tran- forming FOR005-PT with the structures of similar, non- sitions form the conformationally rearranged fibril core. Thus, amyloidogenic V domains (Fig. S7). A similar canonical class the relationship of the mutations and fibril formation can be alteration has been observed for the CDR1 loop of some topologically indirect as seen by the effects of the G94A amyloidogenic λ6 LCs (83). Nonetheless, especially conserva- mutation in FOR005. tive mutations in exposed loops were not expected to drasti- In conclusion, our findings add further proof to the concept cally alter protein structure and stability (33). Yet, the CDR that thermodynamic stability is an important, yet not the only mutations in FOR005—especially the conservative G94A crucial molecular determinant in the fibril formation pathway substitution—strongly affect V domain stability and of LCs and that conformational dynamics play an important J. Biol. Chem. (2021) 296 100334 7 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Figure 3. The effects of point mutations on fibril formation propensity. A, fibril formation kinetics at 37 C, pH 7.4, and continuous shaking show that G49R/G94A (dark yellow) is the only one of the nine investigated mutants that forms amyloid fibrils in vitro. All ThT kinetics were normalized to a fluo- rescence start value of 1. B, at pH 6.4, fibril formation of FOR005-PT (black) and G49R/G94A (dark yellow) is accelerated and also amyloid aggregation of N51S/G94A (pale blue) can be observed after approximately 10 days. Despite strong thermodynamic destabilization, N51S/G94A needs additional acidic conditions to form fibrils. C, TEM micrographs of all FOR005 variants were obtained after 2 weeks of incubation at 37 C, pH 7.4 or 6.4, and continuous shaking in a Tecan Genios Platereader. Samples were stained using uranyl acetate. The four panels show the only samples that exhibited an increase in ThT fluorescence signal in the ThT assays depicted in A and B. The scale bar represents 200 nm. part. Additionally, we show that different mutations can be AmylPred2 (46), MetAmyl (47), ZipperDB (48), and SWISS- important in amyloid formation by either destabilizing the MODEL (49) were used. precursor protein or stabilizing the final fibril core structure or even both. Furthermore, our study provides detailed mecha- Cloning, mutagenesis, protein expression, and purification nistic information on the limitations of CDR flexibility, on Synthetic DNA constructs of FOR005-PT/GL and antibody domain architecture, and how mutations in the hy- FOR005-PLC/GLC in pET28b(+) were obtained from Invi- pervariable CDRs can have a major impact on V domain trogen. Variants were produced by site-directed mutagenesis integrity and induce fibril formation. using primers designed with NEBaseChanger. Primers were synthesized by Eurofins Genomics. Q5-Polymerase chain Experimental procedures reactions and subsequent KLD enzyme reactions were per- All chemicals were purchased from Sigma-Aldrich or VWR formed according to the manufacturer’s protocol. Plasmid unless stated otherwise. sequencing was performed by Eurofins Genomics. Plasmids were transformed into E.coli BL21 (DE3)-star cells and the Sequence and structure analysis proteins were expressed as insoluble inclusion bodies at 37 The cDNA sequence of FOR005-PT was previously re- C over night after induction with 1 mM IPTG. Cells were ported by Annamalai et al.(36)(https://www.ncbi.nlm.nih. harvested and inclusion bodies prepared as previously gov/nuccore/KX290463). The corresponding germline described (85). Inclusion bodies were solubilized in 50 mM sequence, FOR005-GL, was determined using IgBLAST Tris/HCl, 8 M urea, 0.1% β-mercapto ethanol, pH 8.0 at (https://www.ncbi.nlm.nih.gov/igblast/), the international im- room temperature for 4 to 8 h and then dialyzed against an munogenetics information system (http://www.imgt.org/), and excess of 50 mM Tris, 5 M urea, pH 8.0 at 10 C over the abYsis database (http://www.abysis.org/abysis/). The night. The solubilized protein was then subjected to anion GenBank accession code for the germline V domain is exchange chromatography using Q-Sepharose (GE Health- AAZ13705.1. For bioinformatic analyses of the protein se- care, Uppsala, Sweden). Protein-containing fractions were quences and structures Clustal Omega (84), Consurf (45), pooled and diluted to 0.5 mg/ml protein or below. The 8 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Figure 4. Conformational dynamics play a major role in the fibril formation of FOR005-PT. A, limited proteolysis of FOR005 constructs with trypsin was carried out in triplicates at room temperature using a protein/protease ratio of 15/1 (w/w). FOR005-PT is shown in black, GL in red, G49R in green, G94A in violet, G49R/G94A in dark yellow, and N51S/G94A in pale blue. Increased susceptibility to proteolytic degradation implies enhanced structural dynamics. B, fractional deuterium was detected after 2 h incubation with D O by ESI-TOF/TOF mass spectrometry to give peptide-resolved information on protein backbone dynamics. The fold change in fractional uptake compared to the germline V was calculated by dividing the uptake values of the respective mutants by the uptake values of FOR005-GL. Therefore, a fold change value below 1 means lower flexibility than the germline, a value above 1 indicates enhanced dynamics in comparison. The data sets for the mutants are colored according to A. The dashed red line at a value of 1 represents the germline V . C, the fold change in uptake of FOR005-PT was plotted onto the crystal structure of the patient V domain. Red color indicates strongly enhanced dynamics in the patient-derived V domain, blue color indicates increased dynamics of the germline protein. Residues colored in black could not be analyzed in the H/ DX-MS experiments. The most strongly affected segments lie in the β-sheet framework, especially in structural regions close to the C terminus of the V domain. diluted protein was dialysed against an excess of 50 mM from 260 nm to 320 nm using 50 μM protein in PBS. Thermal Tris, 3 M urea, pH 8.5 at 10 C over night. Afterwards, the transitions were recorded from 20 to 90 C at 205 nm using a protein was dialyzed against PBS pH 7.4 for approximately heating rate of 1 C/min. 24 h at 10 C. As a polishing step, the refolded protein was purified by size-exclusion chromatography using a Super- Analytical ultracentrifugation dex75 column (GE Healthcare, Uppsala, Sweden) running in PBS. Protein quality was checked by SDS-PAGE and ESI-ion For AUC measurements, a ProteomLab XL-I centrifuge trap mass spectrometry. (Beckman) equipped with absorbance optics was used. The protein concentration for the measurements was 40 μMin Circular dichroism spectroscopy PBS. The assembled cells were loaded with 350 μl of sample solution. The cells are equipped with quartz windows and 12- CD measurements were carried out on a Chirascan spec- mm-path-length charcoal-filled epon double-sector center- tropolarimeter (Applied Photophysics, Surrey, UK) and on a pieces. An eight-hole Beckman-Coulter AN50-ti rotor was JASCO J-1500 CD spectrometer (JASCO). Far-UV spectra used for all measurements, which were carried out at were recorded in a 1 mm quartz cuvette at 20 C from 260 nm 42,000 rpm and 20 C. Sedimentation was continuously to 200 nm using 10 μM protein diluted in PBS. Near-UV scanned with a radial resolution of 30 mm and monitored at spectra were recorded in a 2 mm quartz cuvette at 20 C J. Biol. Chem. (2021) 296 100334 9 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Figure 5. MD simulations show energetically unfavorable backbone conformations in CDR2 and CDR3. A, root-mean-square fluctuations (RMSF) observed in MD simulations (1 μs, at 310 K) along the residue sequence for the FOR005-GL V variant (red line), the FOR005-GL G49R/G94A (black line), and the GL N51S/G94A (blue line) substitutions. B, sampled backbone dihedral angles phi and psi plotted as Ramachandran plots for residues 48 to 50 (same color code as in A). Favorable regions for non-Gly residues are indicated by a green dashed boundary in the Ramachandran plots and a regime favorable for Gly but less for other amino acids is indicated in orange with a blue boundary. C, same as in B but for residues 50 to 52. D, same as in B but for residues 93 to 95. 280 nm. For data analysis, SEDFIT with continuous c(S) dis- Chemical unfolding transitions were carried out in tripli- tribution mode was used (86, 87). cates by incubating 1 μM of protein with increasing concen- trations of urea in a sample volume of 200 μl in reaction tubes. Fluorescence spectroscopy After incubation over night at room temperature, the samples Reversibilty of unfolding was checked by incubating were transferred into a 96-well Greiner UV-star plate (Greiner 10 μM native patient and germline V domain (each in L Bio-One, Kremsmünster Austria) and intrinsic tryptophan triplicates) with 6 M urea for 2 h at room temperature. Then fluorescence was monitored at 25 C in a Tecan Infinite M the samples were diluted 1:9 with PBS pH 7.4 for over night Nano+ plate reader (Tecan Group Ltd). The excitation wave- refolding yielding a final protein concentration of 1 μMand length was 280 nm and emission spectra were recorded from a final urea concentration of 0.6 M. For comparison, 1 μM 300 to 400 nm. Transition curves were obtained by plotting native V domains were incubated with 0.6 M and 6 M urea normalized fluorescence intensities at the wavelength at which for 24 h. Fluorescence spectra were recorded at 25 Con a native and unfolded state shows the largest signal difference Horiba FluoroMax4 spectrofluorimeter (Horiba Jobin Yvon) against the concentration of urea. The transition curves with an excitation wavelength of 280 nm and emission from represent triplicates that were averaged and normalized. Sub- 300 to 400 nm. Excitation and emission slits were set to sequently, data was analyzed with Origin by applying a 5 nm, and for every spectrum two accumulations were Boltzmann fit and a two-state unfolding fit model to obtain app averaged. ΔG and cooperativity values (88). un 10 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Thioflavin T-binding kinetics Waters AQUITY UPLC BEH C18 column (1.7 mm, 1.0 × 100 mm) by an H O to acetonitrile gradient with both eluents Prior to sample preparation, protein stock solutions were containing 0.1% formic acid (v/v). Eluting peptides were centrifuged in an Optima MAX-E ultracentrifuge (Beckman) directly subjected to the Synapt TOF mass spectrometer by for 3 to 4 h at 40,000 rpm in order to remove aggregates. electrospray ionization. Prior to fragmentation and mass Additionally, all assay components were filtered through a detection, peptides were additonally seperated by drift time. 0.22 μm filter (Merck) before the samples were prepared. For Samples were pipetted by a LEAP autosampler (HTS PAL; all measurements, 200 μl of each sample was incubated in 96- Leap Technologies, NC). Data analysis was performed with the well Nunc plates (Nunc, Thermo Fisher) sealed with Crystal Waters Protein Lynx Global Server PLGs (version 3.0.3) and Clear PP sealing foil (HJ-Bioanalytik GmbH). Thioflavin T the DynamX (Version 3.0) software package. assays were carried out in triplicates with 15 μM protein, 7.5 μM ThT, 0.05% sodium azide, pH 7.4 or 6.4, at 37 C under Molecular dynamics simulations continuous orbital shaking in a Tecan Genios platereader with the shaking intensity set to high (Tecan Group Ltd). For MD simulations were carried out and analyzed using the determining the ThT fluorescence of the samples, the excita- Amber18 simulation package (91). Simulations were per- tion wavelength was 440 nm, the emission wavelength was formed starting from the FOR005-PT V variant for which a 480 nm, and the gain was set to 70 to 75. Values of midpoint crystal structure is available (PDB: 5L6Q) and on the in silico amyloid fibril formation (t ) were determined using a Boltz- generated variants with the R49G und A94G substitutions, the mann fit. FOR005GL (wild-type sequence), the GL G49R/G94A, and the GL N51S/G94A sequence variants. Each protein was solvated Transmission electron microscopy in TIP3P water in a periodic octahedral box with a minimum distance of protein atoms to the box boundary of 10 Å (92). Activated copper grids (200 mesh) were loaded with 10 μlof + − The ff14SB force field was employed and Na and Cl ions sample from finished ThT assays for 1 min. The grids were were added to neutralize the system and reach an ion con- washed with 20 μlH O and stained with 8 μl of a 1.5% uranyl centration of 0.15 M. Energy minimization of each system was acetate solution for 1 min. Excess solutions were removed performed with the sander module of Amber18 (2500 mini- from the grids with filter paper. TEM micrographs were mization cycles). The systems were heated in steps of 100 K recorded at 120 kV on a JEOL JEM 1400-plus transmission (50 ps per step) to a final temperature of 310 K with the solute electron microscope (JEOL Germany GmbH). nonhydrogen atoms harmonically restraint to the start struc- ture. All bonds involving hydrogen atoms were kept at optimal Limited proteolysis length. In additional four steps, the harmonic restraints were The V domains were diluted to 0.3 mg/ml in 100 mM Tris, removed stepwise. For the subsequent production simulations, 100 mM NaCl, 10 mM CaCl , pH 7.8 and incubated at room hydrogen mass repartitioning (HMR) was employed allowing a temperature with trypsin using substrate/enzyme ratio of 15/1 time step of 4 fs (instead of 2 fs used during heating and (w/w) or with proteinase K at a substrate/enzyme ratio of 150/ equilibration). Unrestrained production simulations were 1 (w/w). At defined time points, samples were taken from the extended to 1 μs for each system. Coordinates were saved reaction and mixed with PMSF (final concentration 2 mM) every 8 ps. Root mean square deviation (RMSD), root mean and Lämmli buffer to stop the proteolytic degradation. After- square fluctuations (RMSF), and analysis of dihedral angle ward, the samples were run on a SERVA Prime 4 to 20% SDS distributions were performed using the cpptraj module of gel, and protein ratios were subsequently analyzed using NIH Amber18. ImageJ (89). Hydrogen/deuterium exchange mass spectrometry (H/DX-MS) Data availability For all H/DX-MS experiments, a fully automated system All data are contained within the article. equipped with a Leap robot (HTS PAL; Leap Technologies, NC), a Waters ACQUITY M-Class UPLC, an H/DX manager Acknowledgments—This study was performed in the framework of (Waters Corp), and a Synapt G2-S mass spectrometer (Waters the research unit FOR2969 on systemic light chain amyloidosis Corp) were used as previously described (90). Protein samples (German Research Foundation DFG; Project SP03). We thank Prof. with a concentration of 30 μM were diluted in a ratio of 1:20 Dr Sevil Weinkauf and Dr Carsten Peters for help with TEM data with PBS buffer (pH 7.4) containing deuterium oxide. The acquisition. We also thank Sonja Engler and Jacqueline Bambach samples were incubated with D O for 0 s, 10 s, 1 min, 10 min, for help with protein purification and biophysical experiments. 30 min, or 2 h. The exchange was stopped by diluting the labeled protein 1:1 in quenching buffer (200 mM Na HPO ×2 2 4 Author contributions—G. J. R. and J. B. conceptualization; G. J. R., F. H O, 200 mM NaH PO ×2H O, 250 mM Tris (2- 2 2 4 2 R., R. M. A., M. H., and M. Z. data curation; G. J. R., F. R., R. M. A., carboxyethyl)phosphine, 3 M GdmCl, pH 2.2) at 1 C. Pro- P. K., and M. H., formal analysis; G. J. R., B. W., F. R., P. K., R. M. A., teolytic online digestion was performed using an immobilized and M. H., investigation; G. J. R. and J. B. writing original draft; G. J. Waters Enzymate BEH Pepsin Column (2.1 × 30 mm) at 20 C. R., P. K., R. M. A., M. H., and M. Z. writing-review and editing; M. Z. The resulting peptides were trapped and separated at 0 Cona and J. B., supervision; M. Z. and J. B. funding acquisition. J. Biol. Chem. (2021) 296 100334 11 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Conflict of interest—The authors declare no conflict of interest. structural breaks in a patient-derived amyloid fibril from systemic AL amyloidosis. Nat. Commun. 12, 875 21. Kim, Y., Wall, J. S., Meyer, J., Murphy, C., Randolph, T. W., Manning, M. C., Solomon, A., and Carpenter, J. F. (2000) Thermodynamic modulation References of light chain amyloid fibril formation. J. Biol. Chem. 275, 1570–1574 1. Knowles, T. P. J., Vendruscolo, M., and Dobson, C. M. (2014) The am- 22. Hurle, M. R., Helms, L. R., Li, L., Chan, W., and Wetzel, R. (1994) A role yloid state and its association with protein misfolding diseases. Nat. Rev. for destabilizing amino acid replacements in light-chain amyloidosis. Proc. Mol. Cell Biol. 15, 384–396 Natl. Acad. Sci. U. S. A. 91, 5446–5450 2. Chiti, F., and Dobson, C. M. (2017) Protein misfolding, amyloid forma- 23. Wall, J., Schell, M., Murphy, C., Hrncic, R., Stevens, F. J., and Solomon, A. tion, and human disease: A summary of progress over the last decade. (1999) Thermodynamic instability of human lambda 6 light chains: Annu. Rev. Biochem. 86,27–68 Correlation with fibrillogenicity. Biochemistry 38, 14101–14108 3. Gertz, M. A. (2018) Immunoglobulin light chain amyloidosis: 2018 Update 24. Baden, E. M., Randles, E. G., Aboagye, A. K., Thompson, J. R., and on diagnosis, prognosis, and treatment. Am. J. Hematol. 93,1169–1180 Ramirez-Alvarado, M. (2008) Structural insights into the role of muta- 4. Merlini, G. (2017) AL amyloidosis: From molecular mechanisms to targeted tions in amyloidogenesis. J. Biol. Chem. 283, 30950–30956 therapies. Hematology Am. Soc. Hematol. Educ. Program 2017,1–12 25. Nokwe, C. N., Zacharias, M., Yagi, H., Hora, M., Reif, B., Goto, Y., and 5. Murphy, K., and Weaver, C. (2016) Janeway’s Immunobiology, W.W. Buchner, J. (2014) A residue-specific shift in stability and amyloidoge- Norton & Company, New York, NY nicity of antibody variable domains. J. Biol. Chem. 289, 26829–26846 6. Weiss, B. M., Hebreo, J., Cordaro, D. V., Roschewski, M. J., Baker, T. P., 26. Nokwe, C. N., Hora, M., Zacharias, M., Yagi, H., Peschek, J., Reif, B., Abbott, K. C., and Olson, S. W. (2014) Increased serum free light chains Goto, Y., and Buchner, J. (2016) A stable mutant predisposes antibody domains to amyloid formation through specific non-native interactions. precede the presentation of immunoglobulin light chain amyloidosis. J. J. Mol. Biol. 428, 1315–1332 Clin. Oncol. 32, 2699–2704 27. Oberti, L., Rognoni, P., Barbiroli, A., Lavatelli, F., Russo, R., Maritan, M., 7. Zhang, C., Huang, X., and Li, J. (2017) Light chain amyloidosis: Where are Palladini, G., Bolognesi, M., Merlini, G., and Ricagno, S. (2017) Con- the light chains from and how they play their pathogenic role? Blood Rev. current structural and biophysical traits link with immunoglobulin light 31, 261–270 chains amyloid propensity. Sci. Rep. 7, 16809 8. Tonegawa, S. (1983) Somatic generation of antibody diversity. Nature 28. Baden, E. M., Owen, B. A. L., Peterson, F. C., Volkman, B. F., Ramirez- 302, 575–581 Alvarado, M., and Thompson, J. R. (2008) Altered dimer interface de- 9. Abraham, R. S., Geyer, S. M., Ramírez-Alvarado, M., Price-Troska, T. L., Gertz, M. A., and Fonseca, R. (2004) Analysis of somatic hypermutation creases stability in an amyloidogenic protein. J. Biol. Chem. 283, 15853– and antigenic selection in the clonal B cell in immunoglobulin light chain 15860 amyloidosis (AL). J. Clin. Immunol. 24, 340–353 29. DiCostanzo, A. C., Thompson, J. R., Peterson, F. C., Volkman, B. F., and 10. Blancas-Mejia, L. M., Misra, P., Dick, C. J., Cooper, S. A., Redhage, K. R., Ramirez-Alvarado, M. (2012) Tyrosine residues mediate fibril formation Bergman, M. R., Jordan, T. L., Maar, K., and Ramirez-Alvarado, M. (2018) in a dynamic light chain dimer interface. J. Biol. Chem. 287, 27997–28006 Immunoglobulin light chain amyloid aggregation. Chem. Commun. 30. Rennella, E., Morgan, G. J., Kelly, J. W., and Kay, L. E. (2019) Role of (Camb.) 54, 10664–10674 domain interactions in the aggregation of full-length immunoglobulin 11. Buxbaum, J. N. (2004) The systemic amyloidoses. Curr. Opin. Rheumatol. light chains. Proc. Natl. Acad. Sci. U. S. A. 116, 854–863 16,67–75 31. Bork, P., Holm, L., and Sander, C. (1994) The immunoglobulin fold. 12. Morgan, G. J., and Kelly, J. W. (2016) The kinetic stability of a full-length Structural classification, sequence patterns and common core. J. Mol. antibody light chain dimer determines whether endoproteolysis can Biol. 242, 309–320 release amyloidogenic variable domains. J. Mol. Biol. 428, 4280–4297 32. el Hawrani, A. S., Moreton, K. M., Sessions, R. B., Clarke, A. R., and Hol- 13. Enqvist, S., Sletten, K., Stevens, F. J., Hellman, U., and Westermark, P. brook, J. J. (1994) Engineering surface loops of proteins–a preferred strategy (2007) Germ line origin and somatic mutations determine the target for obtaining new enzyme function. Trends Biotechnol. 12,207–211 tissues in systemic AL-amyloidosis. PLoS One 2, e981 33. Chang, H.-J., Jian, J.-W., Hsu, H.-J., Lee, Y.-C., Chen, H.-S., You, J.-J., 14. Enqvist, S., Sletten, K., and Westermark, P. (2009) Fibril protein frag- Hou, S.-C., Shao, C.-Y., Chen, Y.-J., Chiu, K.-P., Peng, H.-P., Lee, K. H., mentation pattern in systemic AL-amyloidosis. J. Pathol. 219, 473–480 and Yang, A.-S. (2014) Loop-sequence features and stability determinants 15. Lecoq, L., Wiegand, T., Rodriguez-Alvarez, F. J., Cadalbert, R., Herrera, in antibody variable domains by high-throughput experiments. Structure G. A., Del Pozo-Yauner, L., Meier, B. H., and Böckmann, A. (2019) 22,9–21 34. Rouet, R., Lowe, D., and Christ, D. (2014) Stability engineering of the A substantial structural conversion of the native monomer leads to in- human antibody repertoire. FEBS Lett. 588, 269–277 register parallel amyloid fibril formation in light-chain amyloidosis. 35. Davis, M. M. (2004) The evolutionary and structural ‘logic’ of antigen Chembiochem 20, 1027–1031 receptor diversity. Semin. Immunol. 16, 239–243 16. Radamaker, L., Lin, Y.-H., Annamalai, K., Huhn, S., Hegenbart, U., 36. Annamalai, K., Liberta, F., Vielberg, M.-T., Close, W., Lilie, H., Gührs, K.- Schönland, S. O., Fritz, G., Schmidt, M., and Fändrich, M. (2019) Cryo- H., Schierhorn, A., Koehler, R., Schmidt, A., Haupt, C., Hegenbart, U., EM structure of a light chain-derived amyloid fibril from a patient with Schönland, S., Schmidt, M., Groll, M., and Fändrich, M. (2017) Common systemic AL amyloidosis. Nat. Commun. 10, 1103 fibril structures imply systemically conserved protein misfolding path- 17. Swuec, P., Lavatelli, F., Tasaki, M., Paissoni, C., Rognoni, P., Maritan, M., Brambilla, F., Milani, P., Mauri, P., Camilloni, C., Palladini, G., Merlini, G., ways in vivo. Angew. Chem. Int. Ed. Engl. 56, 7510–7514 Ricagno, S., and Bolognesi, M. (2019) Cryo-EM structure of cardiac 37. Brochet, X., Lefranc, M.-P., and Giudicelli, V. (2008) IMGT/V-QUEST: amyloid fibrils from an immunoglobulin light chain AL amyloidosis pa- The highly customized and integrated system for IG and TR stan- tient. Nat. Commun. 10, 1269 dardized V-J and V-D-J sequence analysis. Nucleic Acids Res. 36, 18. Hora, M., Sarkar, R., Morris, V., Xue, K., Prade, E., Harding, E., Buchner, W503–W508 J., and Reif, B. (2017) MAK33 antibody light chain amyloid fibrils are 38. Lefranc, M.-P., Giudicelli, V., Duroux, P., Jabado-Michaloud, J., Folch, G., similar to oligomeric precursors. PLoS One 12, e0181799 Aouinti, S., Carillon, E., Duvergey, H., Houles, A., Paysan-Lafosse, T., 19. Pradhan, T., Annamalai, K., Sarkar, R., Huhn, S., Hegenbart, U., Schön- Hadi-Saljoqi, S., Sasorith, S., Lefranc, G., and Kossida, S. (2014) IMGT®, land, S., Fändrich, M., and Reif, B. (2020) Seeded fibrils of the germline the international ImMunoGeneTics information system® 25 years on. variant of human λ-III immunoglobulin light chain FOR005 have a Nucleic Acids Res. 43, D413–D422 similar core as patient fibrils with reduced stability. J. Biol. Chem. 295, 39. Swindells, M. B., Porter, C. T., Couch, M., Hurst, J., Abhinandan, K. R., 18474–18484 Nielsen, J. H., Macindoe, G., Hetherington, J., and Martin, A. C. R. (2017) 20. Radamaker, L., Baur, J., Huhn, S., Haupt, C., Hegenbart, U., Schönland, S., abYsis: Integrated antibody sequence and structure-management, anal- Bansal, A., Schmidt, M., and Fändrich, M. (2021) Cryo-EM reveals ysis, and prediction. J. Mol. Biol. 429, 356–364 12 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis 40. Ye, J., Ma, N., Madden, T. L., and Ostell, J. M. (2013) IgBLAST: An 59. Fontana, A., de Laureto, P. P., de Filippis, V., Scaramella, E., and Zam- immunoglobulin variable domain sequence analysis tool. Nucleic Acids bonin, M. (1999) Limited proteolysis in the study of protein conforma- Res. 41,W34–W40 tion. In: Sterchi, E. E., Stöcker, W., eds. Proteolytic Enzymes: Tools and 41. Perfetti, V., Casarini, S., Palladini, G., Vignarelli, M. C., Klersy, C., Diegoli, Targets, Springer Berlin Heidelberg, Berlin, Heidelberg: 253–280 M., Ascari, E., and Merlini, G. (2002) Analysis of V(lambda)-J(lambda) 60. Englander, S. W. (2006) Hydrogen exchange and mass spectrometry: A expression in plasma cells from primary (AL) amyloidosis and normal historical perspective. J. Am. Soc. Mass Spectrom. 17, 1481–1489 bone marrow identifies 3r (lambdaIII) as a new amyloid-associated 61. Klimtchuk, E. S., Gursky, O., Patel, R. S., Laporte, K. L., Connors, L. H., germline gene segment. Blood 100, 948–953 Skinner, M., and Seldin, D. C. (2010) The critical role of the constant 42. Villalba, M. I., Canul-Tec, J. C., Luna-Martínez, O. D., Sánchez-Alcalá, R., region in thermal stability and aggregation of amyloidogenic immuno- Olamendi-Portugal, T., Rudiño-Piñera, E., Rojas, S., Sánchez-López, R., globulin light chain. Biochemistry 49, 9848–9857 Fernández-Velasco, D. A., and Becerril, B. (2015) Site-directed muta- 62. Weber, B., Hora, M., Kazman, P., Göbl, C., Camilloni, C., Reif, B., and genesis reveals regions implicated in the stability and fiber formation of Buchner, J. (2018) The antibody light-chain linker regulates domain human λ3r light chains. J. Biol. Chem. 290, 2577–2592 orientation and amyloidogenicity. J. Mol. Biol. 430, 4925–4940 43. Chothia, C., and Lesk, A. M. (1987) Canonical structures for the hyper- 63. Nokwe, C. N., Hora, M., Zacharias, M., Yagi, H., John, C., Reif, B., Goto, variable regions of immunoglobulins. J. Mol. Biol. 196, 901–917 Y., and Buchner, J. (2015) The antibody light-chain linker is important for 44. Wu, T. T., and Kabat, E. A. (1970) An analysis of the sequences of the domain stability and amyloid formation. J. Mol. Biol. 427, 3572–3586 variable regions of Bence Jones proteins and myeloma light chains and 64. Luna-Martínez, O. D., Hernández-Santoyo, A., Villalba-Velázquez, M. I., their implications for antibody complementarity. J. Exp. Med. 132, 211– Sánchez-Alcalá,R., Fernández-Velasco, D. A.,and Becerril,B.(2017) Stabi- 250 lizing an amyloidogenic λ6 light chain variable domain. FEBS J. 284, 3702–3717 45. Ashkenazy, H., Abadi, S., Martz, E., Chay, O., Mayrose, I., Pupko, T., and 65. Del Pozo-Yauner, L., Wall, J. S., González Andrade, M., Sánchez-López, Ben-Tal, N. (2016) ConSurf 2016: An improved methodology to estimate R., Rodríguez-Ambriz, S. L., Pérez Carreón, J. I., Ochoa-Leyva, A., and and visualize evolutionary conservation in macromolecules. Nucleic Acids Fernández-Velasco, D. A. (2014) The N-terminal strand modulates Res. 44, W344–W350 immunoglobulin light chain fibrillogenesis. Biochem. Biophys. Res. Com- 46. Tsolis, A. C., Papandreou, N. C., Iconomidou, V. A., and Hamodrakas, S. mun. 443, 495–499 J. (2013) A consensus method for the prediction of ’aggregation-prone’ 66. González-Andrade, M., Becerril-Luján, B., Sánchez-López, R., Ceceña- peptides in globular proteins. PLoS One 8, e54175 Álvarez, H., Pérez-Carreón, J. I., Ortiz, E., Fernández-Velasco, D. A., and 47. Emily, M., Talvas, A., and Delamarche, C. (2013) MetAmyl: A METa- Del Pozo-Yauner, L. (2013) Mutational and genetic determinants of λ6 predictor for AMYLoid proteins. PLoS One 8, e79722 light chain amyloidogenesis. FEBS J. 280, 6173–6183 48. Goldschmidt, L., Teng, P. K., Riek, R., and Eisenberg, D. (2010) Identifying 67. Hernández-Santoyo, A., del Pozo Yauner, L., Fuentes-Silva, D., Ortiz, E., the amylome, proteins capable of forming amyloid-like fibrils. Proc. Natl. Rudiño-Piñera, E., Sánchez-López, R., Horjales, E., Becerril, B., and Acad. Sci. U. S. A. 107, 3487 Rodríguez-Romero, A. (2010) A single mutation at the sheet switch re- 49. Waterhouse, A., Bertoni, M., Bienert, S., Studer, G., Tauriello, G., gion results in conformational changes favoring lambda6 light-chain Gumienny, R., Heer, F. T., de Beer, T. A. P., Rempfer, C., Bordoli, L., fibrillogenesis. J. Mol. Biol. 396, 280–292 Lepore, R., and Schwede, T. (2018) SWISS-MODEL: Homology model- 68. Rennella, E., Morgan, G. J., Yan, N., Kelly, J. W., and Kay, L. E. (2019) The ling of protein structures and complexes. Nucleic Acids Res. 46, W296– role of protein thermodynamics and primary structure in fibrillogenesis of W303 variable domains from immunoglobulin light chains. J. Am. Chem. Soc. 50. Woody, R. W. (1995) [4] Circular Dichroism. Biochemical Spectroscopy, 141, 13562–13571 Elsevier, Amsterdam, Netherlands 69. Helms, L. R., and Wetzel, R. (1995) Destabilizing loop swaps in the CDRs 51. Gade Malmos, K., Blancas-Mejia, L. M., Weber, B., Buchner, J., Ram- of an immunoglobulin VL domain. Protein Sci. 4, 2073–2081 irez-Alvarado, M., Naiki, H., and Otzen, D. (2017) ThT 101: A primer 70. Gąciarz, A., and Ruddock, L. W. (2017) Complementarity determining on the use of thioflavin T to investigate amyloid formation. Amyloid 24, regions and frameworks contribute to the disulfide bond independent 1–16 folding of intrinsically stable scFv. PLoS One 12, e0189964 52. Khurana, R., Gillespie, J. R., Talapatra, A., Minert, L. J., Ionescu-Zanetti, 71. Herold, E. M., John, C., Weber, B., Kremser, S., Eras, J., Berner, C., C., Millett, I., and Fink, A. L. (2001) Partially folded intermediates as Deubler, S., Zacharias, M., and Buchner, J. (2017) Determinants of critical precursors of light chain amyloid fibrils and amorphous aggre- the assembly and function of antibody variable domains. Sci. Rep. 7, gates. Biochemistry 40, 3525–3535 12276 53. Creighton, T. E. (1993) Proteins: Structures and Molecular Properties, 72. van der Kant, R., Bauer, J., Karow-Zwick, A. R., Kube, S., Garidel, P., Macmillan, New York, NY Blech, M., Rousseau, F., and Schymkowitz, J. (2019) Adaption of human 54. Feige, M. J., Hendershot, L. M., and Buchner, J. (2010) How antibodies antibody λ and κ light chain architectures to CDR repertoires. Protein fold. Trends Biochem. Sci. 35, 189–198 Eng. Des. Sel. 32, 109–127 55. Buchner, J., Renner, M., Lilie, H., Hinz, H. J., Jaenicke, R., Kiefhabel, T., 73. Blancas-Mejía, L. M., Tischer, A., Thompson, J. R., Tai, J., Wang, L., and Rudolph, R. (1991) Alternatively folded states of an immunoglobulin. Auton, M., and Ramirez-Alvarado, M. (2014) Kinetic control in protein Biochemistry 30, 6922–6929 folding for light chain amyloidosis and the differential effects of somatic 56. Kazman, P., Vielberg, M.-T., Pulido Cendales, M. D., Hunziger, L., mutations. J. Mol. Biol. 426, 347–361 Weber, B., Hegenbart, U., Zacharias, M., Köhler, R., Schönland, S., Groll, 74. Blancas-Mejía, L. M., Horn, T. J., Marin-Argany, M., Auton, M., Tischer, M., and Buchner, J. (2020) Fatal amyloid formation in a patient’s antibody A., and Ramirez-Alvarado, M. (2015) Thermodynamic and fibril forma- light chain is caused by a single point mutation. eLife 9, e52300 tion studies of full length immunoglobulin light chain AL-09 and its 57. Le Marchand, T., de Rosa, M., Salvi, N., Sala, B. M., Andreas, L. B., germline protein using scan rate dependent thermal unfolding. Biophys. Barbet-Massin, E., Sormanni, P., Barbiroli, A., Porcari, R., Sousa Mota, C., Chem. 207,13–20 de Sanctis, D., Bolognesi, M., Emsley, L., Bellotti, V., Blackledge, M., et al. 75. Ruiz-Zamora, R. A., Guillaumé, S., Al-Hilaly, Y. K., Al-Garawi, Z., (2018) Conformational dynamics in crystals reveal the molecular bases Rodríguez-Alvarez, F. J., Zavala-Padilla, G., Pérez-Carreón, J. I., Rodrí- for D76N beta-2 microglobulin aggregation propensity. Nat. Commun. 9, guez-Ambriz, S. L., Herrera, G. A., Becerril-Luján, B., Ochoa-Leyva, A., 1658 Melendez-Zajgla, J., Serpell, L., and Del Pozo-Yauner, L. (2019) The 58. Maritan, M., Romeo, M., Oberti, L., Sormanni, P., Tasaki, M., Russo, R., CDR1 and other regions of immunoglobulin light chains are hot spots for Ambrosetti, A., Motta, P., Rognoni, P., Mazzini, G., Barbiroli, A., Palla- amyloid aggregation. Sci. Rep. 9, 3123 dini, G., Vendruscolo, M., Diomede, L., Bolognesi, M., et al. (2020) 76. Brumshtein,B., Esswein, S. R.,Sawaya, M. R.,Rosenberg,G., Ly,A. Inherent biophysical properties modulate the toxicity of soluble amyloi- T.,Landau, M.,and Eisenberg, D. S. (2018) Identification of two dogenic light chains. J. Mol. Biol. 432, 845–860 principal amyloid-driving segments in variable domains of Ig light J. Biol. Chem. (2021) 296 100334 13 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis chains in systemic light-chain amyloidosis. J. Biol. Chem. 293, andHiggins,D.G.(2011) Fast,scalable generation of high-quality 19659–19671 protein multiple sequence alignments using Clustal Omega. Mol. 77. Poshusta, T. L., Sikkink, L. A., Leung, N., Clark, R. J., Dispenzieri, A., and Syst. Biol. 7,539 Ramirez-Alvarado, M. (2009) Mutations in specific structural regions of 85. Thies, M. J. W., and Pirkl, F. (2000) Chromatographic purification of the immunoglobulin light chains are associated with free light chain levels in CH2 domain of the monoclonal antibody MAK33. J. Chromatogr. B patients with AL amyloidosis. PLoS One 4, e5169 Biomed. Sci. Appl. 737,63–69 78. Feige, M. J., Groscurth, S., Marcinowski, M., Yew, Z. T., Truffault, V., 86. Brown, P. H., and Schuck, P. (2006) Macromolecular size-and-shape Paci, E., Kessler, H., and Buchner, J. (2008) The structure of a folding distributions by sedimentation velocity analytical ultracentrifugation. intermediate provides insight into differences in immunoglobulin amy- Biophys. J. 90, 4651–4661 loidogenicity. Proc. Natl. Acad. Sci. U. S. A. 105, 13373–13378 87. Schuck, P. (2000) Size-distribution analysis of macromolecules by sedi- 79. Marin-Argany, M., Güell-Bosch, J., Blancas-Mejía, L. M., Villegas, S., and mentation velocity ultracentrifugation and Lamm equation modeling. Ramirez-Alvarado, M. (2015) Mutations can cause light chains to be too Biophys. J. 78, 1606–1619 stable or too unstable to form amyloid fibrils. Protein Sci. 24, 1829–1840 88. Santoro, M. M., and Bolen, D. W. (1988) Unfolding free energy changes 80. Chothia, C., Lesk, A. M., Tramontano, A., Levitt, M., Smith-Gill, S. J., determined by the linear extrapolation method. 1. Unfolding of phenyl- Air, G., Sheriff, S., Padlan, E. A., Davies, D., Tulip, W. R., Colman, P. M., methanesulfonyl alpha-chymotrypsin using different denaturants. Spinelli, S., Alzari, P. M., and Poljak, R. J. (1989) Conformations of Biochemistry 27, 8063–8068 immunoglobulin hypervariable regions. Nature 342, 877–883 89. Abràmoff, M. D., Magalhães, P. J., and Ram, S. J. (2004) Image processing 81. Al-Lazikani, B., Lesk, A. M., and Chothia, C. (1997) Standard confor- with ImageJ. Biophotonics Int. 11,36–42 mations for the canonical structures of immunoglobulins. J. Mol. Biol. 90. Weber, B., Brandl, M. J., Pulido Cendales, M. D., Berner, C., Pradhan, T., 273, 927–948 Feind, G. M., Zacharias, M., Reif, B., and Buchner, J. (2018) A single 82. Dondelinger, M., Filée, P., Sauvage, E., Quinting, B., Muyldermans, S., residue switch reveals principles of antibody domain integrity. J. Biol. Galleni, M., and Vandevenne, M. S. (2018) Understanding the signifi- Chem. 293, 17107–17118 cance and implications of antibody numbering and antigen-binding sur- 91. Case, D. A., Ben-Shalom, I. Y., Brozell, S. R., Cerutti, D. S., Cheatham, T. face/residue definition. Front. Immunol. 9, 2278 E., III, Cruzeiro, V. W. D., Darden, T. A., Duke, R. E., Ghoreishi, D., 83. del Pozo Yauner, L., Ortiz, E., and Becerril, B. (2006) The CDR1 of the Gilson, M. K., Gohlke, H., Goetz, A. W., Greene, D., Harris, R., Homeyer, human lambdaVI light chains adopts a new canonical structure. Proteins N., et al. (2018) Amber 18, University of California, San Francisco, CA 62, 122–129 92. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., and 84. Sievers, F.,Wilm,A., Dineen,D., Gibson,T.J., Karplus, K.,Li, W., Klein, M. L. (1983) Comparison of simple potential functions for simu- Lopez, R.,McWilliam, H., Remmert,M., Söding,J., Thompson,J. D., lating liquid water. J. Chem. Phys. 79, 926–935 14 J. Biol. Chem. (2021) 296 100334 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Biological Chemistry Unpaywall

Molecular mechanism of amyloidogenic mutations in hypervariable regions of antibody light chains

Loading next page...
 
/lp/unpaywall/molecular-mechanism-of-amyloidogenic-mutations-in-hypervariable-a5SZmUdU3V

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

Publisher
Unpaywall
ISSN
0021-9258
DOI
10.1016/j.jbc.2021.100334
Publisher site
See Article on Publisher Site

Abstract

RESEARCH ARTICLE EDITORS’ PICK Molecular mechanism of amyloidogenic mutations in hypervariable regions of antibody light chains Received for publication, September 3, 2020, and in revised form, January 14, 2021 Published, Papers in Press, January 26, 2021, https://doi.org/10.1016/j.jbc.2021.100334 1 1 1 1 1 Georg J. Rottenaicher , Benedikt Weber , Florian Rührnößl , Pamina Kazman , Ramona M. Absmeier , 2 2 1, Manuel Hitzenberger , Martin Zacharias , and Johannes Buchner * 1 2 From the Center for Integrated Protein Science Munich at the Department Chemie, Center for Integrated Protein Science Munich at the Physik-Department, Technische Universität München, Garching, Germany Edited by Ursula Jakob Systemic light chain (AL) amyloidosis is a fatal protein linked by disulfide bridges. Each of the LCs is made up of an misfolding disease in which excessive secretion, misfolding, N-terminal variable (V ) domain and a C-terminal constant and subsequent aggregation of free antibody light chains (C ) domain (5). In AL amyloidosis, malignant monoclonal eventually lead to deposition of amyloid plaques in various plasma cells overproduce and secrete LCs into the blood organs. Patient-specific mutations in the antibody V domain stream leading to very high concentrations of circulating LCs are closely linked to the disease, but the molecular mechanisms (6, 7). The malignant plasma cells often emerge in the course by which certain mutations induce misfolding and amyloid of an underlying plasma cell dyscrasia (e.g., multiple myeloma) aggregation of antibody domains are still poorly understood. (7). During the complex maturation process responsible for Here, we compare a patient V domain with its non- the creation of antibody binding diversity, these LCs acquire amyloidogenic germline counterpart and show that, out of the amyloidogenic point mutations mostly in the V domain, five mutations present, two of them strongly destabilize the which is the main constituent of fibrils in AL amyloidosis protein and induce amyloid fibril formation. Surprisingly, the (8–10). These point mutations and (in many cases) proteolytic decisive, disease-causing mutations are located in the highly cleavage of the LC to produce the free V domain are key variable complementarity determining regions (CDRs) but factors for disease onset and progression (11–14). Yet, the exhibit a strong impact on the dynamics of conserved core exact mechanism by which certain mutations favor amyloid regions of the patient V domain. This effect seems to be based formation of antibody domains remains largely elusive. Recent on a deviation from the canonical CDR structures of CDR2 and cryo-EM and solid-state NMR studies of AL amyloid fibrils CDR3 induced by the substitutions. The amyloid-driving mu- show that the fold of the fibril core is strikingly different from tations are not necessarily involved in propagating fibril for- the native Ig fold suggesting that a complete structural rear- mation by providing specific side chain interactions within the rangement has to occur in the process of fibril formation fibril structure. Rather, they destabilize the V domain in a (15–20). For a large number of AL cases, it has been shown specific way, increasing the dynamics of framework regions, that the decrease in thermodynamic stability of the V domain which can then change their conformation to form the fibril is a decisive factor for amyloidogenicity (21–25). However, also core. These findings reveal unexpected influences of CDR- nondestabilizing mutations in the V domain can induce fibril framework interactions on antibody architecture, stability, formation, thus other factors such as LC dimerization, struc- and amyloid propensity. tural changes, and conformational dynamics also need to be taken into account (26–30). A major enigma in this context is how different mutations Amyloidoses comprise a family of protein misfolding dis- shift V domains toward the fibrillary pathway, especially if eases in which disease-specific precursor proteins aggregate these mutations are not in the conserved framework but in the into highly ordered amyloid fibrils (1). These fibrils form variable antigen binding regions called complementarity amyloid plaques, which are deposited either systemically or in determining regions (CDRs). These are solvent-exposed loops an organ-specific manner causing severe damage (2). The most connecting β-strands (31). In order to recognize a large variety common systemic disease in this context is amyloid light chain of antigens, CDRs need to tolerate a high sequence variability, (AL) amyloidosis, in which an antibody light chain (LC) acts as which in turn suggests that CDR point mutations should not the precursor protein that eventually forms amyloid fibers (3, strongly affect the thermodynamic stability and aggregation 4). In healthy individuals, plasma cells secrete IgG antibodies, propensity of the antibody domain (32–35). In 2017, however, which consist of two LCs and two heavy chains covalently Annamalai et al. (36) reported the crystal structure and the fibril morphology of a V domain (FOR005-PT) obtained from an AL amyloidosis patient with mainly cardiac involvement in This article contains supporting information. * which four out of the five mutations are located in the CDRs For correspondence: Johannes Buchner, johannes.buchner@tum.de. Present address for Benedikt Weber: Roche Diagnostics GmbH, Nonnenwald (according to the Kabat/Chothia domain numbering). Since 2, Penzberg 82377, Germany. J. Biol. Chem. (2021) 296 100334 1 © 2021 THE AUTHORS. Published by Elsevier Inc on behalf of American Society for Biochemistry and Molecular Biology. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis current models cannot explain the amyloidogenic character of segment containing the mutations G49R and N51S. The fifth this variant, we set out to determine which of these mutations mutation, G94A, is located in the hypervariable CDR3 loop. drive amyloid aggregation and found that specifically two of Residue conservation analysis with Consurf revealed the lowest the CDR mutations are causative for fibril formation. Our degree of conservation (Consurf score = 1) for all four CDR findings further show that seemingly minor side chain alter- mutations and an average conservation degree (Consurf ations, even in poorly conserved CDRs, can destabilize the score = 5) for the framework residue F48 in the patient entire V domain and drive it toward misfolding and amyloid sequence (Fig. S1)(45). We further assessed aggregation-prone aggregation. regions and individual mutational effects by applying the prediction tools AmylPred2, MetAmyl, and ZipperDB (46–48). Results Amyloidogenicity predictions by these tools did not suggest Sequence and structure analysis significant alterations of the amyloid aggregation or steric zipper propensity of the patient V sequence in comparison In 2017, Annamalai et al. (36) reported the cDNA sequence with its corresponding germline sequence (Fig. 1A). To explain and crystal structure (PDB: 5L6Q) of an amyloid forming V structural effects of mutations in more detail, we created a domain (FOR005-PT) derived from a patient with cardiac LC homology model of FOR005-GL based on the template amyloidosis. We used IgBLAST, IMGT, and abYsis to deter- structure 5BV7, which exhibits 98.2% sequence identity, using mine the corresponding germline sequence (FOR005-GL) with the SWISS-MODEL web server (49). Structural alignments of the highest possible protein sequence identity for this amy- the homology model with the crystal structure of FOR005-PT loidogenic V (37–40). FOR005-PT belongs to the λ3l LC (PDB: 5L6Q) showed that the overall structure was conserved, subfamily (gene segments: IGLV3-19/IGLJ2). The related although the conformations of the CDR2 and CDR3 loops are germline λ3r has been reported to be associated with AL altered (Fig. 1B). However, one needs to take into account that amyloidosis (41, 42). Five point mutations were identified in homology models are merely an approximation of the actual the patient-derived V domain (Y31S, Y48F, G49R, N51S, native protein structure. G94A) compared with the germline sequence (Fig. 1A), but it was not clear which mutation causes amyloid aggregation. FOR005-PT and FOR005-GL differ substantially in fibril Four of them are located in the hypervariable CDRs according formation propensity and thermodynamic stability to the Kabat and Chothia numbering systems (43, 44). The mutation Y31S is located in the CDR1 region, Y48F lies in the To test the biophysical properties of the proteins directly, conserved framework 2 region (FR2) right next to the begin- we produced patient and germline V domains recombinantly ning of CDR2. The CDR2 comprises a short, protruding loop in E.coli and purified them to homogeneity. The far-UV Figure 1. Sequence and structural analyses of FOR005-PT and FOR005-GL. A, sequence alignment of patient and germline V shows the five point mutations highlighted in red and the variable CDR loops in cyan (CDR1), blue (CDR2), and green (CDR3). Predictions of amyloidogenic regions by three different tools overlap well indicating that the point mutations do not introduce new amyloid driving segments. Aggregation-prone positions are indicated by asteriks. The sequence numbering as derived from Annamalai et al. starts with the first serine residue, Ser1. The N-terminal glycine in our sequence results from using NcoI during subcloning of the FOR005 gene constructs and is, therefore, numbered as Gly0. B, structural alignment of FOR005-PT shown in black (PDB: 5L6Q) with the homology model derived for FOR005-GL depicted in gray. The homology model was created using the SWISS-MODEL server and the template structure 5BV7. CDRs are colored according to the sequence alignment in A. Mutated positions are shown in red on the patient V and light red on the germline V domain with side chains depicted as sticks. For F48 on the patient structure two rotamers are shown. The structural comparison suggests rearrangements of loop conformations in CDR2 and CDR3. 2 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis circular dichroism (CD) spectra of the purified proteins thiazol-based fluorescent dye Thioflavin T (ThT), which showed that both are properly folded and possess the typical β- specifically binds to the characteristic cross-β motif in amy- sheet-rich immunoglobulin fold as indicated by the minimum loid fibrils (51). ThT-binding kinetics showed that the patient at around 218 nm in the far-UV region (Fig. S2)(50). Near-UV V domain starts to form amyloid fibrils in vitro after CD spectra, which represent a specific tertiary structure approximately 3 days, whereas the corresponding germline fingerprint, were highly similar for the two proteins. Thus, protein does not engage in amyloid aggregation (Fig. 2A, FOR005-PT and FOR005-GL seem to have nearly identical Table 1). To obtain direct evidence for the presence of fibrils tertiary structure and topology. Additionally, analytical ultra- in the samples, we performed transmission electron micro- centrifugation (AUC) was performed to assess the quaternary scopy (TEM). The TEM micrographs showed fibrils only in structure. As indicated by sedimentation coefficients of 1.52 the patient V sample and not in the germline control and 1.59 S, respectively, both the patient and germline V (Fig. 2B). Thus, the patient V behaved as expected and the L L domains are monomeric in solution (Fig. S2). germline protein does not show amyloidogenic behavior. To test whether the two V domains differ in their fibril FOR005 fibrils isolated from patient tissue contained only the formation propensities, we incubated the proteins in phos- V domain (35). Since the role of proteolytic cleavage of phate buffered saline (PBS) at pH 7.4 and 37 C under precursor LCs in amyloidosis is still only poorly understood continuous shaking and monitored fibril formation via the (14), we purified full-length LCs of the patient and germline Figure 2. Fibril formation propensity of the patient V domain. A, thioflavin T-binding kinetics of FOR005-PT (black), -GL (red), -PLC (light green), and -GLC (light purple) obtained at 37 C and pH 7.4 under continuous shaking. The increase in fluorescence shows that the patient V domain is the only protein engaging in amyloid fibril formation after approximately 3 days. Connecting the patient V domain with the C domain to form a full-length LC completely L L inhibited fibril formation. All kinetic curves were normalized to a fluorescence start value of 1. B, TEM micrographs of samples from finished ThT assays were recorded after negative stain with uranyl acetate. The amyloid fibers of FOR005-PT can be seen in the upper left panel, the scale bar represents 200 nm. C, for chemical unfolding transitions, 1 μM protein was equilibrated with increasing concentrations of urea over night at room temperature. Fluorescence spectra (λ = 280 nm/λ = 300–400 nm) were recorded at 25 C in a 96-well plate. The transition of FOR005-PT is shown with black dots, the data for FOR005-GL is ex em shown as red dots. The black and red sigmoidal lines represent the individual fit functions. D, thermal unfolding transitions of FOR005-PT and -GL were obtained by recording CD signal at 205 nm while applying a temperature gradient from 20 to 90 C with a heating rate of 1 C/min. Sample concentration was 10 μM in PBS and the measurement was performed in a 1 mm quartz cuvette. J. Biol. Chem. (2021) 296 100334 3 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Table 1 Stability parameters, unfolding cooperativity, and fibril formation midpoints of FOR005 constructs. T C mt pH 7.4 t pH 6.4 m m 50 50 −1 −1 V domain C M Urea kJ mol M dd FOR005-PT 43.5 ± 0.14 1.90 ± 0.02 5.99 ± 0.33 3.8 2.8 FOR005-GL 56.3 ± 0.11 4.28 ± 0.04 4.33 ± 0.31 - - GL Y31S 56.1 ± 0.12 4.20 ± 0.03 5.12 ± 0.47 - - GL Y48F 55.4 ± 0.13 4.15 ± 0.03 4.56 ± 0.32 - - GL G49R 52.2 ± 0.10 2.91 ± 0.01 5.55 ± 0.13 - - GL N51S 54.4 ± 0.10 3.44 ± 0.02 5.01 ± 0.27 - - GL G94A 50.5 ± 0.17 3.25 ± 0.05 4.95 ± 0.99 - - GL Y31S/G94A 50.3 ± 0.15 3.31 ± 0.03 5.17 ± 0.58 - - GL Y48F/G94A 51.9 ± 0.10 3.22 ± 0.02 5.32 ± 0.28 - - GL G49R/G94A 47.0 ± 0.09 2.09 ± 0.02 6.69 ± 0.41 8.9 4.4 GL N51S/G94A 48.6 ± 0.08 2.56 ± 0.02 5.98 ± 0.33 - 11.5 Thermal transitions were obtained by recording the CD signal at 205 nm between 20 to 90 C at a heating rate of 1 C/min. Chemical unfolding transitions were obtained by fluorescence spectroscopy using 1 μM of each V domain with increasing concentrations of urea. Since both thermal and chemical unfoldings are irreversible, the stability parameters T and C represent apparent values. Transition midpoints and standard deviations were derived from a Boltzmann fit. Chemical unfolding data was also subjected to m m app a two-state unfolding fit model to determine cooperativity and ΔG values (Table S1). Fibril formation assays were carried out at 37 C, pH 7.4 or 6.4 under continouos shaking un in a Tecan Genios platereader. The t values represent the time point at which fibril formation is 50% completed. variants (FOR005-PLC and FOR005-GLC, respectively) to Mutations in hypervariable regions affect domain stability determine whether the patient LC is also amyloidogenic. We and aggregation performed fibril formation assays and transition electron While the results described above show that the patient- microscopy and found that both LCs did not engage in the specific mutations affect conformational stability and fibril amyloidogenic pathway (Fig. 2). formation, it was not possible to rationalize which of the To determine differences between the two V domains mutations are responsible for amyloidogenesis. To determine regarding their thermodynamic propertiesinmoredetail, we the specific effects of each of the five point mutations, we investigated their stabilities by chemical and thermal dena- replaced them individually in the germline sequence by the turation experiments. Unfolding transitions in the presence respective patient residues (Y31S, Y48F, G49R, N51S, and of increasing urea concentrations were performed to assess G94A). CD spectroscopy and AUC analysis of the mutants the chemical domain stability and unfolding cooperativity of showed that all point mutants adopted the conserved β-sheet the V domains (Fig. 2, Table 1). The patient V domain L L structure and were monomeric in solution (Fig. S3). Addi- showed a midpoint of unfolding at a urea concentration (C ) tionally, highly similar near-UV CD spectra suggest that the of 1.90 M, whereas for the germline domain, the midpoint is amino acid substitutions have only minor effects on the global at 4.28 M urea (Fig. 2, Table 1). We assessed the reversibility tertiary structure of the antibody domain (Fig. S3). of urea-induced unfolding by fluorescence spectroscopy and Thermal unfolding experiments of the germline V domain found that both V domains cannot be completely refolded constructs containing the individual patient mutations showed into their native structure within 24 h at room temperature the largest stability decrease for the G94A mutant with a T (Fig. S2). It should be noted, however, that the germline V value of 50.5 C and the second largest effect for the G49R exhibits a higher degree of unfolding reversibility than the mutant with a transition temperature of 52.2 C. The thermal patient variant (Fig. S2). We applied a two-state fitmodel to stabilities of the remaining mutants Y31S, Y48F, and N51S our transition data to calculate unfolding free energies were only slightly decreased with transition midpoints tem- (ΔG ). However, this is in principle only possible if un peratures of 56.1 C, 55.4 C, and 54.4 C, respectively unfolding is completely reversible. Since this is not the case (Table 1, Fig. S3). In the case of chemical unfolding, the under the conditions used (Fig. S2), these results do not strongest decrease in stability was observed for the G49R represent true ΔG values but are rather apparent unfolding un mutant with a C value of 2.91 M urea, whereas the G94A app free energies (ΔG )(Table S1). un variant unfolded at a concentration of 3.25 M urea. Both the app In thermal denaturation experiments, the patient-derived G49R and G94A variant show comparable ΔG values of un protein exhibited a melting temperature (T ) of 43.5 C, 16.49 kJ/mol and 16.18 kJ/mol, respectively (Table S1). Again, whereas the germline protein showed a melting temperature of the transition midpoints of the Y31S (4.2 M) and Y48F 56.3 C. The T values correspond to the temperatures at (4.15 M) mutants lie only slightly below that of the germline which 50% of the protein is unfolded. Since thermal unfolding reference, while the N51S variant unfolded at 3.44 M urea of both FOR005-PT and FOR005-GL is also irreversible (Table 1, Fig. S3). (Fig. S2), the obtained transition midpoints represent apparent Among the five single mutations, G49R and G94A exerted melting temperatures (Table 1). These data show that the the strongest destabilizing effect on the germline V domain. patient V domain has a significantly decreased thermody- Since G94A is a small, conservative mutation located in the namic stability compared with its germline counterpart hypervariable CDR3 loop, these results were unexpected. (Table 1). Therefore, we created double mutants by individually 4 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis combining G94A with the remaining four mutations yielding germline counterpart pointing toward a higher degree of the double mutations Y31S/G94A, Y48F/G94A, G49R/G94A, conformational dynamics (Fig. 4A, Fig. S5). Further, the single- and N51S/G94A. The largest effect on thermal stability was point mutants G49R and G94A behave similarly to the observed for the double mutants G49R/G94A and N51S/ germline V domain and exhibit overall slow degradation ki- G94A. These exhibited severely decreased thermal stabilities netics. However, G94A is processed faster and to a greater with melting temperatures of 47.0 C and 48.6 C, respectively. extent than G49R and FOR005-GL. The double mutant N51S/ Accordingly, also in terms of chemical stability, the mutations G94A is also cleaved much faster than the germline and the G49R/G94A and N51S/G94A had the most significant effect observed single mutants, yet not as fast as the patient-derived with transition midpoints of 2.09 M and 2.56 M urea, V domain. Interestingly, the double mutant G49R/G94A is respectively (Table 1, Fig. S3). degraded even more readily than the patient V domain Furthermore, ThT-binding kinetics and TEM micrographs FOR005-PT (Fig. 4A, Fig. S5). revealed that G49R/G94A is the only mutant that forms H/DX-MS was applied to gain more detailed insights into amyloid fibrils in vitro at pH 7.4 and 37 C(Fig. 3A). The the conformational dynamics of both the patient and germ- N51S/G94A mutant, however, did not form fibrils within line V domain. This method is based on the enhanced sol- 2 weeks at pH 7.4, despite exhibiting significantly decreased vent exchange rates of backbone amide hydrogens in flexible thermodynamic stability, similarly to G49R/G94A. It has protein regions from which peptide-resolved dynamic infor- been shown that destabilization is not necessarily the only mation can be derived after pepsin cleavage and mass spec- driving force in the amyloid formation pathway and that trometric analyses (60). Fractional deuterium uptake was protein dynamics and population of nonnative intermediate determined for FOR005-PT, FOR005-GL, G49R, G94A, as states can play important roles, too (26, 27, 52). To further well as the double mutants G49R/G94A and N51S/G94A. investigate the involvement of these molecular traits, addi- The fold change in fractional uptake was calculated by tional fibril formation assays were performed at pH 6.4, since dividing uptake ratios of the investigated mutants by the acidification can lead to a decrease in stability and population uptake ratios of the germline V domain (Fig. 4B). A value of alternatively folded intermediate states (53–55). As ex- below 1 indicates that the germline exhibits higher deuterium pected, in ThT assays carried out at pH 6.4, fibril formation uptake, whereas a value above 1 shows increased deuterium was accelerated for FOR005-PT and G49R/G94A, but also for uptake for the observed mutant. Conformational dynamics N51S/G94A fibril formation was observed after approxi- are especially pronounced for residues 12 to 20, residues 65 mately 10 days (Fig 3B). Thepresenceofamyloid fibrils in the to 85, and residues 97 to 105 in the case of FOR005-PT and ThT assay was confirmed by TEM micrographs (Fig. 3C). for the double mutant G49R/G94A (Fig. 4B). The double These findings imply that the amyloid aggregation of mutation, however, still exhibits slightly lower flexibility FOR005-PT relies on a mechanism in which domain desta- compared with the actual patient V that contains all five bilization is an important, yet not the only decisive bio- substitutions. Interestingly, residues 50 to 60 including the physical factor. CDR2 loop are moredynamic in thegermlineV domain and Overall, the results for the single and double mutations in G49R/G94A. The double mutant N51S/G94A also exhibits show that out of the five mutations two are mainly responsible lightly increased dynamic behavior, especially for residues 80 for the significant loss in thermodynamic stability and the gain to 105, whereas the single mutants G49R and G94A, in in amyloid formation propensity. Remarkably, the conservative comparison, do not impose a strong increase in conforma- G94A mutation in the exposed CDR3 loop has a strong impact tional flexibility. Notably, G94A has a slightly larger effect on on the biophysical properties of the V domain, despite being overall dynamics than G49R (Fig. 4B). To better visualize located in the most variable part of the protein. which parts of the patient-derived V domain experience enhanced dynamics in comparison with the germline, the change in fractional uptake was plotted onto the crystal Conformational dynamics are linked to decreased protein structure of FOR005-PT (PDB: 5L6Q). Structurally, the most stability and amyloid formation affected regions correspond to β-strands A2 and B and the Previous studies have demonstrated that there is a causal small loop connecting them (residues 12–20), β-strands E link between conformational dynamics and aggregation pro- and F including the small helical segment between them pensity, as well as cellular toxicity of prefibrillar species (27, (residues 65–85), and the C-terminal β-strands G1 and G2 56–58). Therefore, we set out to investigate the structural (Fig. 4C). In summary, our results show that there is a clear dynamics and flexibility of patient and germline V domains by connection between conformational dynamics and amyloid limited proteolysis and hydrogen/deuterium exchange mass aggregation. Remarkably, we observe the strongest increase spectrometry (H/DX-MS). in dynamics in conserved framework regions rather than the Limited proteolysis allows obtaining information about segments where the point mutations are located. These structural flexibility since proteolytic degradation is increased findings suggest that small mutation-induced changes in due to enhanced protein dynamics and local unfolding (59). CDR loop conformations might propagate through the entire When we carried out limited proteolysis experiments with the domain architecture and thereby lead to increased dynamics proteases trypsin or proteinase K, we found that the patient- in framework regions, which causes lower stability and derived V domain was degraded much faster than its enhanced aggregation propensity. J. Biol. Chem. (2021) 296 100334 5 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Unfavorable main chain conformations in the CDR2 and CDR3 Whether proteolytic processing of LC precursors is a pre- loops destabilize the V domain requisite or a consequence of amyloid formation still remains enigmatic (12, 14, 30, 61). It has been shown that the C To gain further insight into the flexibility of the patient V domain can exert a protective function in vitro and that full- domain, molecular dynamics (MD) simulations were per- length LCs do not readily aggregate into amyloid fibrils formed in explicit solvent on the FOR005-GL, the GL G49R/ (56, 62, 63). In the case of FOR005, the V domain was G94A, the GL N51S/G94A, the FOR005-PT V and a PT identified as the sole component of amyloid deposits in the variant containing the R49G and A94G double substitution. On patient’s tissue (36). Accordingly, in vitro only the patient- the timescale of 1 μs, the variants were stable during the MD derived V domain but not the corresponding full-length LC simulations and exhibited only small and similar deviations formed amyloid fibrils, thus implying a protective role of the from the start structure (Fig. S6). Comparison of root-mean- C domain. square fluctuations (RMSF) indicated the lowest fluctuations FOR005 is an interesting case, as the V domain contains for FOR005-GL, slightly enhanced fluctuations for GL N51S/ four CDR mutations and only one framework mutation G94A, and significantly increased fluctuations especially compared with its germline counterpart. Of note, the exact around residue 49 and 94 in case of the GL G49R/G94A variant location of the CDRs depends on which domain numbering (Fig. 5A). Slightly larger conformational fluctuations on the MD system is used. The three systems according to Kabat, Chothia, timescale were also observed for the FOR005-PT variant and IMGT are the most common ones (38, 43, 44). When the compared with the PT R49G/A94G mutation (Fig. S6A). IMGT numbering scheme is applied to FOR005, the substi- The inspection of the peptide backbone dihedral angles in tution N51S would be considered a framework mutation the loop regions near residue 49 and 94 revealed the sampling rather than a CDR2 mutation. However, the Kabat and Cho- of the left-handed helical regimes in the Ramachandran plots thia classifications identify this residue as belonging to the of residues 49 and 50 as well as 94 and 95 (but not for residues CDR and this coincides with the Consurf residue conservation 51 or 52, Fig. 5, B–D). This regime is sterically favorable in case analysis (Fig. S1). Furthermore, the identification of a suitable of the glycine but less so for nonglycine residues. Hence, R49 germline sequence for a given V domain can yield different or A94 creates steric strain in the loop structure, whereas G49 results depending on which method/database is used. We or G94 relaxes this strain. Also, in the case of FOR005-PT, the applied abYsis, IgBLAST, and IMGT to identify a V domain MD simulations reveal sampling of sterically unfavorable with highest possible amino acid sequence identity (38–40). peptide backbone states that—in a relaxed peptide structure— The most important practical test for the germline sequence of are typically only adopted by glycine residues (Fig. S6B). choice is whether it forms fibrils, as this allows to identify and Notably, these unfavorable backbone states are also observed test the effect of the patient mutations concerning their in the crystal structure of the patient protein. Hence, in the amyloidogenic potential (56). patient structure, the loop forces its residues at least partially Up to now, mostly framework mutations have been reported into an energetically unfavorable backbone structure upon as key factors in LC amyloid aggregation (22, 24, 26, 56, 62, folding. The “germline” substitutions R49G and A94G can 64–68). Regarding the only framework mutation in FOR005- relax this strain because now the glycine residues at positions PT—Y48F—it has already been shown for a different V 49 and 94 are better compatible with the required backbone domain that this particular mutation has little to no influence structure. Interestingly, the substitutions can also have an ef- on domain stability and aggregation propensity (22). There- fect on neighboring residues and partially modify their back- fore, we hypothesized that only the CDR mutations play a bone sampling (Fig. 5, B–D, Fig. S6, B and C). Energetically, crucial role in the case of FOR005, which was confirmed by the the stress of enforcing unfavorable backbone conformation of experimental results. CDR loops are not only involved in an- residues 49 and 94 can amount to several kcal/mol and hence tigen recognition, they have also been shown to play important could be the reason for the much lower stability of FOR005-PT structural roles in antibody domain architecture and V /V H L and of the variants with nonglycine residues at positions 49 domain association. Various experimental and computational and 94. For residues 51 and 52 approximately the same sam- studies on V domains demonstrated that CDRs can have a pling of favorable backbone states was observed (Fig. 5C) with strong influence on the folding pathway, stability, and no significant effect of the N51S substitution. conformation of the protein (69–72). The involvement of a Discussion CDR mutation in LC amyloidogenicity has been shown for a Systemic LC amyloidosis is a highly complex protein mis- proline residue in the CDR3 loop of an amyloidogenic V folding disease because of the enormous sequence variability of domain. Its deletion resulted in enhanced stability and delayed the soluble precursor protein—the antibody LC. This renders fibril formation kinetics (73). Furthermore, nonconservative the mechanistic understanding of the amyloid aggregation mutations in the V domains of AL and multiple myeloma process a very challenging task. Different, case-dependent or- (MM) patients—also encompassing the CDR3 loops—were gan involvement and a wide spectrum of symptoms further reported to affect the kinetic stability of the LCs (74). Addi- complicate analysis and treatment of this rare disorder (3, 4). tionally, it has been shown that CDR1 can act as a hotspot for Additionally, there are a number of other factors that can aggregation and that a peptide based on part of a CDR3 affect fibril formation, disease onset, and progression, segment can drive amyloid fibril formation due to enhanced including proteolytic processing of the precursor LCs (7, 10). steric zipper propensity (75, 76). However, a detailed 6 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis mechanistic understanding of the effects caused by specific conformation. The changes caused by unfavorable CDR loop CDR residues in the context of the disease is still lacking (77). conformations seem to propagate through the entire protein Multiple studies on substitutions in the V domain inducing increased flexibility, which leads to the enhanced demonstrate that misfolding and amyloid aggregation depend population of partially unfolded, aggregation-competent states on the thermodynamic/kinetic stability, structural dynamics or (52, 78). Therefore, an altered interplay of hypervariable loops partial unfolding, LC dimerization, and local conformational and conserved framework can play a key role in stability and alterations of the native fold (10, 24, 27–29, 78). Thermody- amyloidogenicity of V domains (69, 73). In this context, namic and kinetic stabilities have widely been thought of as the FOR005-PT represents the first case where the onset of fibril major driving force in the misfolding and aggregation pathway formation is directly and mechanistically correlated to the (26, 56, 58, 79). In the case of FOR005, a synergistic combi- substitution of two distinct amino acids in CDR loops. Sur- nation of thermodynamic destabilization and altered confor- prisingly, one of these two decisive substitutions is the small, mational dynamics appears to determine the pathway of the conservative G94A mutation in the surface-exposed CDR3 soluble V monomer toward amyloid fibrils. FOR005-PT and loop. FOR005-GL show a pronounced difference in stability with a Pradhan et al. (19) have recently shown that the R49 res- ΔT of 12.8 C and the mutations G49R and G94A have the idue in FOR005-PT plays a key role in stabilizing the fibril largest impact on domain stability. However, fibril formation core. With this information, it becomes plausible that muta- kinetics and thermodynamic data of the FOR005 double mu- tions in amyloid-forming LCs can serve different purposes. tants suggest that destabilization through CDR mutations is The G94A mutation leads to a conformational change in the not the only driving force in the amyloid formation process, CDR3 loop, which thereby adopts a structure that differs from since the severely destabilized N51S/G94A mutant only forms the canonical CDR class. This conformational change results fibrils after prolonged incubation at lower pH. Additionally, an in enhanced framework dynamics and decreased overall increase in conformational dynamics—mediated by the two domain stability. To illustrate this concept, a structural decisive CDR mutations G49R and G94A—is necessary to alignment of FOR005-PT was performed with three highly induce the amyloid aggregation. Remarkably, the strongest similar, nonamyloidogenic V domains taken from the PDB increase in dynamics is observed in conserved protein core (Fig. S7). The mutation-induced changes in CDR loop regions rather than the loop segments in which the mutations conformation depict the described deviation from the ca- are located. MD simulations indicate that the loop residues 49, nonical CDR class. In the final core structure of FOR005-PT 50, 94, and 95 sample mostly backbone conformations that are fibrils, however, A94 does not play an important role. Seem- energetically unfavorable for nonglycine residues, which ingly, its only effect lies in the destabilization of the precursor lowers the overall stability of the folded structure. This co- V domain. The CDR2 mutation G49R, on the other hand, incides with reports that glycine is structurally preferable at drives amyloidogenesis both by altering CDR2 loop confor- positions with certain u/ψ angles (22, 53). Hence, the interplay mation and by providing a stabilizing side chain interaction in of the CDRs with the framework enforces an energetically the fibril core (19). Yet, as our data show G94A mediates a unfavorable conformation of the loops. These strained loop larger increase in conformational dynamics than G49R, structures affect framework dynamics and are the likely reason especially in the framework 3 region and the C-terminal part for the lower stability of variants with a nonglycine residue at of the domain (Fig. 4). Further, the CDR2 mutation N51S is the corresponding positions. also capable of inducing fibril formation. Thus, the primary This is at first glance counterintuitive, as one might assume role of G49R in the fibril formation pathway of FOR005- that the basic traits of CDRs are their sequence diversity and PT appears to lie in stabilizing the final product of the conformational flexibility that allow them to adapt to the pathway—the core of the amyloid fiber. Individually, however, structure of the antigen upon interaction. However, five out of the two point mutations do not induce fibril formation the six CDRs in an antibody F only adopt a limited number in vitro. Yet in combination, the two CDR mutations G49R ab of backbone conformations, known as canonical classes, with and G94A act synergistically as the obtained stabilities and the heavy chain CDR3 (CDRH3) being the only exception apparent free energies imply (Table 1, Table S1). In summary, (43, 80–82). Therefore, it seems plausible for some CDR the decisive, amyloid-driving mutations are not necessarily mutations in LCs to induce unfavorable loop conformations, involved in propagating fibril formation by providing specific which represent a deviation from the canonical CDR class and side chain interactions within the fibril structure. Rather, they thereby put structural strain on the framework. This deviation destabilize the V domain in a specific way, increasing the can be seen by aligning the crystal structure of the amyloid- dynamics of framework regions, which upon structural tran- forming FOR005-PT with the structures of similar, non- sitions form the conformationally rearranged fibril core. Thus, amyloidogenic V domains (Fig. S7). A similar canonical class the relationship of the mutations and fibril formation can be alteration has been observed for the CDR1 loop of some topologically indirect as seen by the effects of the G94A amyloidogenic λ6 LCs (83). Nonetheless, especially conserva- mutation in FOR005. tive mutations in exposed loops were not expected to drasti- In conclusion, our findings add further proof to the concept cally alter protein structure and stability (33). Yet, the CDR that thermodynamic stability is an important, yet not the only mutations in FOR005—especially the conservative G94A crucial molecular determinant in the fibril formation pathway substitution—strongly affect V domain stability and of LCs and that conformational dynamics play an important J. Biol. Chem. (2021) 296 100334 7 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Figure 3. The effects of point mutations on fibril formation propensity. A, fibril formation kinetics at 37 C, pH 7.4, and continuous shaking show that G49R/G94A (dark yellow) is the only one of the nine investigated mutants that forms amyloid fibrils in vitro. All ThT kinetics were normalized to a fluo- rescence start value of 1. B, at pH 6.4, fibril formation of FOR005-PT (black) and G49R/G94A (dark yellow) is accelerated and also amyloid aggregation of N51S/G94A (pale blue) can be observed after approximately 10 days. Despite strong thermodynamic destabilization, N51S/G94A needs additional acidic conditions to form fibrils. C, TEM micrographs of all FOR005 variants were obtained after 2 weeks of incubation at 37 C, pH 7.4 or 6.4, and continuous shaking in a Tecan Genios Platereader. Samples were stained using uranyl acetate. The four panels show the only samples that exhibited an increase in ThT fluorescence signal in the ThT assays depicted in A and B. The scale bar represents 200 nm. part. Additionally, we show that different mutations can be AmylPred2 (46), MetAmyl (47), ZipperDB (48), and SWISS- important in amyloid formation by either destabilizing the MODEL (49) were used. precursor protein or stabilizing the final fibril core structure or even both. Furthermore, our study provides detailed mecha- Cloning, mutagenesis, protein expression, and purification nistic information on the limitations of CDR flexibility, on Synthetic DNA constructs of FOR005-PT/GL and antibody domain architecture, and how mutations in the hy- FOR005-PLC/GLC in pET28b(+) were obtained from Invi- pervariable CDRs can have a major impact on V domain trogen. Variants were produced by site-directed mutagenesis integrity and induce fibril formation. using primers designed with NEBaseChanger. Primers were synthesized by Eurofins Genomics. Q5-Polymerase chain Experimental procedures reactions and subsequent KLD enzyme reactions were per- All chemicals were purchased from Sigma-Aldrich or VWR formed according to the manufacturer’s protocol. Plasmid unless stated otherwise. sequencing was performed by Eurofins Genomics. Plasmids were transformed into E.coli BL21 (DE3)-star cells and the Sequence and structure analysis proteins were expressed as insoluble inclusion bodies at 37 The cDNA sequence of FOR005-PT was previously re- C over night after induction with 1 mM IPTG. Cells were ported by Annamalai et al.(36)(https://www.ncbi.nlm.nih. harvested and inclusion bodies prepared as previously gov/nuccore/KX290463). The corresponding germline described (85). Inclusion bodies were solubilized in 50 mM sequence, FOR005-GL, was determined using IgBLAST Tris/HCl, 8 M urea, 0.1% β-mercapto ethanol, pH 8.0 at (https://www.ncbi.nlm.nih.gov/igblast/), the international im- room temperature for 4 to 8 h and then dialyzed against an munogenetics information system (http://www.imgt.org/), and excess of 50 mM Tris, 5 M urea, pH 8.0 at 10 C over the abYsis database (http://www.abysis.org/abysis/). The night. The solubilized protein was then subjected to anion GenBank accession code for the germline V domain is exchange chromatography using Q-Sepharose (GE Health- AAZ13705.1. For bioinformatic analyses of the protein se- care, Uppsala, Sweden). Protein-containing fractions were quences and structures Clustal Omega (84), Consurf (45), pooled and diluted to 0.5 mg/ml protein or below. The 8 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Figure 4. Conformational dynamics play a major role in the fibril formation of FOR005-PT. A, limited proteolysis of FOR005 constructs with trypsin was carried out in triplicates at room temperature using a protein/protease ratio of 15/1 (w/w). FOR005-PT is shown in black, GL in red, G49R in green, G94A in violet, G49R/G94A in dark yellow, and N51S/G94A in pale blue. Increased susceptibility to proteolytic degradation implies enhanced structural dynamics. B, fractional deuterium was detected after 2 h incubation with D O by ESI-TOF/TOF mass spectrometry to give peptide-resolved information on protein backbone dynamics. The fold change in fractional uptake compared to the germline V was calculated by dividing the uptake values of the respective mutants by the uptake values of FOR005-GL. Therefore, a fold change value below 1 means lower flexibility than the germline, a value above 1 indicates enhanced dynamics in comparison. The data sets for the mutants are colored according to A. The dashed red line at a value of 1 represents the germline V . C, the fold change in uptake of FOR005-PT was plotted onto the crystal structure of the patient V domain. Red color indicates strongly enhanced dynamics in the patient-derived V domain, blue color indicates increased dynamics of the germline protein. Residues colored in black could not be analyzed in the H/ DX-MS experiments. The most strongly affected segments lie in the β-sheet framework, especially in structural regions close to the C terminus of the V domain. diluted protein was dialysed against an excess of 50 mM from 260 nm to 320 nm using 50 μM protein in PBS. Thermal Tris, 3 M urea, pH 8.5 at 10 C over night. Afterwards, the transitions were recorded from 20 to 90 C at 205 nm using a protein was dialyzed against PBS pH 7.4 for approximately heating rate of 1 C/min. 24 h at 10 C. As a polishing step, the refolded protein was purified by size-exclusion chromatography using a Super- Analytical ultracentrifugation dex75 column (GE Healthcare, Uppsala, Sweden) running in PBS. Protein quality was checked by SDS-PAGE and ESI-ion For AUC measurements, a ProteomLab XL-I centrifuge trap mass spectrometry. (Beckman) equipped with absorbance optics was used. The protein concentration for the measurements was 40 μMin Circular dichroism spectroscopy PBS. The assembled cells were loaded with 350 μl of sample solution. The cells are equipped with quartz windows and 12- CD measurements were carried out on a Chirascan spec- mm-path-length charcoal-filled epon double-sector center- tropolarimeter (Applied Photophysics, Surrey, UK) and on a pieces. An eight-hole Beckman-Coulter AN50-ti rotor was JASCO J-1500 CD spectrometer (JASCO). Far-UV spectra used for all measurements, which were carried out at were recorded in a 1 mm quartz cuvette at 20 C from 260 nm 42,000 rpm and 20 C. Sedimentation was continuously to 200 nm using 10 μM protein diluted in PBS. Near-UV scanned with a radial resolution of 30 mm and monitored at spectra were recorded in a 2 mm quartz cuvette at 20 C J. Biol. Chem. (2021) 296 100334 9 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Figure 5. MD simulations show energetically unfavorable backbone conformations in CDR2 and CDR3. A, root-mean-square fluctuations (RMSF) observed in MD simulations (1 μs, at 310 K) along the residue sequence for the FOR005-GL V variant (red line), the FOR005-GL G49R/G94A (black line), and the GL N51S/G94A (blue line) substitutions. B, sampled backbone dihedral angles phi and psi plotted as Ramachandran plots for residues 48 to 50 (same color code as in A). Favorable regions for non-Gly residues are indicated by a green dashed boundary in the Ramachandran plots and a regime favorable for Gly but less for other amino acids is indicated in orange with a blue boundary. C, same as in B but for residues 50 to 52. D, same as in B but for residues 93 to 95. 280 nm. For data analysis, SEDFIT with continuous c(S) dis- Chemical unfolding transitions were carried out in tripli- tribution mode was used (86, 87). cates by incubating 1 μM of protein with increasing concen- trations of urea in a sample volume of 200 μl in reaction tubes. Fluorescence spectroscopy After incubation over night at room temperature, the samples Reversibilty of unfolding was checked by incubating were transferred into a 96-well Greiner UV-star plate (Greiner 10 μM native patient and germline V domain (each in L Bio-One, Kremsmünster Austria) and intrinsic tryptophan triplicates) with 6 M urea for 2 h at room temperature. Then fluorescence was monitored at 25 C in a Tecan Infinite M the samples were diluted 1:9 with PBS pH 7.4 for over night Nano+ plate reader (Tecan Group Ltd). The excitation wave- refolding yielding a final protein concentration of 1 μMand length was 280 nm and emission spectra were recorded from a final urea concentration of 0.6 M. For comparison, 1 μM 300 to 400 nm. Transition curves were obtained by plotting native V domains were incubated with 0.6 M and 6 M urea normalized fluorescence intensities at the wavelength at which for 24 h. Fluorescence spectra were recorded at 25 Con a native and unfolded state shows the largest signal difference Horiba FluoroMax4 spectrofluorimeter (Horiba Jobin Yvon) against the concentration of urea. The transition curves with an excitation wavelength of 280 nm and emission from represent triplicates that were averaged and normalized. Sub- 300 to 400 nm. Excitation and emission slits were set to sequently, data was analyzed with Origin by applying a 5 nm, and for every spectrum two accumulations were Boltzmann fit and a two-state unfolding fit model to obtain app averaged. ΔG and cooperativity values (88). un 10 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Thioflavin T-binding kinetics Waters AQUITY UPLC BEH C18 column (1.7 mm, 1.0 × 100 mm) by an H O to acetonitrile gradient with both eluents Prior to sample preparation, protein stock solutions were containing 0.1% formic acid (v/v). Eluting peptides were centrifuged in an Optima MAX-E ultracentrifuge (Beckman) directly subjected to the Synapt TOF mass spectrometer by for 3 to 4 h at 40,000 rpm in order to remove aggregates. electrospray ionization. Prior to fragmentation and mass Additionally, all assay components were filtered through a detection, peptides were additonally seperated by drift time. 0.22 μm filter (Merck) before the samples were prepared. For Samples were pipetted by a LEAP autosampler (HTS PAL; all measurements, 200 μl of each sample was incubated in 96- Leap Technologies, NC). Data analysis was performed with the well Nunc plates (Nunc, Thermo Fisher) sealed with Crystal Waters Protein Lynx Global Server PLGs (version 3.0.3) and Clear PP sealing foil (HJ-Bioanalytik GmbH). Thioflavin T the DynamX (Version 3.0) software package. assays were carried out in triplicates with 15 μM protein, 7.5 μM ThT, 0.05% sodium azide, pH 7.4 or 6.4, at 37 C under Molecular dynamics simulations continuous orbital shaking in a Tecan Genios platereader with the shaking intensity set to high (Tecan Group Ltd). For MD simulations were carried out and analyzed using the determining the ThT fluorescence of the samples, the excita- Amber18 simulation package (91). Simulations were per- tion wavelength was 440 nm, the emission wavelength was formed starting from the FOR005-PT V variant for which a 480 nm, and the gain was set to 70 to 75. Values of midpoint crystal structure is available (PDB: 5L6Q) and on the in silico amyloid fibril formation (t ) were determined using a Boltz- generated variants with the R49G und A94G substitutions, the mann fit. FOR005GL (wild-type sequence), the GL G49R/G94A, and the GL N51S/G94A sequence variants. Each protein was solvated Transmission electron microscopy in TIP3P water in a periodic octahedral box with a minimum distance of protein atoms to the box boundary of 10 Å (92). Activated copper grids (200 mesh) were loaded with 10 μlof + − The ff14SB force field was employed and Na and Cl ions sample from finished ThT assays for 1 min. The grids were were added to neutralize the system and reach an ion con- washed with 20 μlH O and stained with 8 μl of a 1.5% uranyl centration of 0.15 M. Energy minimization of each system was acetate solution for 1 min. Excess solutions were removed performed with the sander module of Amber18 (2500 mini- from the grids with filter paper. TEM micrographs were mization cycles). The systems were heated in steps of 100 K recorded at 120 kV on a JEOL JEM 1400-plus transmission (50 ps per step) to a final temperature of 310 K with the solute electron microscope (JEOL Germany GmbH). nonhydrogen atoms harmonically restraint to the start struc- ture. All bonds involving hydrogen atoms were kept at optimal Limited proteolysis length. In additional four steps, the harmonic restraints were The V domains were diluted to 0.3 mg/ml in 100 mM Tris, removed stepwise. For the subsequent production simulations, 100 mM NaCl, 10 mM CaCl , pH 7.8 and incubated at room hydrogen mass repartitioning (HMR) was employed allowing a temperature with trypsin using substrate/enzyme ratio of 15/1 time step of 4 fs (instead of 2 fs used during heating and (w/w) or with proteinase K at a substrate/enzyme ratio of 150/ equilibration). Unrestrained production simulations were 1 (w/w). At defined time points, samples were taken from the extended to 1 μs for each system. Coordinates were saved reaction and mixed with PMSF (final concentration 2 mM) every 8 ps. Root mean square deviation (RMSD), root mean and Lämmli buffer to stop the proteolytic degradation. After- square fluctuations (RMSF), and analysis of dihedral angle ward, the samples were run on a SERVA Prime 4 to 20% SDS distributions were performed using the cpptraj module of gel, and protein ratios were subsequently analyzed using NIH Amber18. ImageJ (89). Hydrogen/deuterium exchange mass spectrometry (H/DX-MS) Data availability For all H/DX-MS experiments, a fully automated system All data are contained within the article. equipped with a Leap robot (HTS PAL; Leap Technologies, NC), a Waters ACQUITY M-Class UPLC, an H/DX manager Acknowledgments—This study was performed in the framework of (Waters Corp), and a Synapt G2-S mass spectrometer (Waters the research unit FOR2969 on systemic light chain amyloidosis Corp) were used as previously described (90). Protein samples (German Research Foundation DFG; Project SP03). We thank Prof. with a concentration of 30 μM were diluted in a ratio of 1:20 Dr Sevil Weinkauf and Dr Carsten Peters for help with TEM data with PBS buffer (pH 7.4) containing deuterium oxide. The acquisition. We also thank Sonja Engler and Jacqueline Bambach samples were incubated with D O for 0 s, 10 s, 1 min, 10 min, for help with protein purification and biophysical experiments. 30 min, or 2 h. The exchange was stopped by diluting the labeled protein 1:1 in quenching buffer (200 mM Na HPO ×2 2 4 Author contributions—G. J. R. and J. B. conceptualization; G. J. R., F. H O, 200 mM NaH PO ×2H O, 250 mM Tris (2- 2 2 4 2 R., R. M. A., M. H., and M. Z. data curation; G. J. R., F. R., R. M. A., carboxyethyl)phosphine, 3 M GdmCl, pH 2.2) at 1 C. Pro- P. K., and M. H., formal analysis; G. J. R., B. W., F. R., P. K., R. M. A., teolytic online digestion was performed using an immobilized and M. H., investigation; G. J. R. and J. B. writing original draft; G. J. Waters Enzymate BEH Pepsin Column (2.1 × 30 mm) at 20 C. R., P. K., R. M. A., M. H., and M. Z. writing-review and editing; M. Z. The resulting peptides were trapped and separated at 0 Cona and J. B., supervision; M. Z. and J. B. funding acquisition. J. Biol. Chem. (2021) 296 100334 11 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis Conflict of interest—The authors declare no conflict of interest. structural breaks in a patient-derived amyloid fibril from systemic AL amyloidosis. Nat. Commun. 12, 875 21. Kim, Y., Wall, J. S., Meyer, J., Murphy, C., Randolph, T. W., Manning, M. C., Solomon, A., and Carpenter, J. F. (2000) Thermodynamic modulation References of light chain amyloid fibril formation. J. Biol. Chem. 275, 1570–1574 1. Knowles, T. P. J., Vendruscolo, M., and Dobson, C. M. (2014) The am- 22. Hurle, M. R., Helms, L. R., Li, L., Chan, W., and Wetzel, R. (1994) A role yloid state and its association with protein misfolding diseases. Nat. Rev. for destabilizing amino acid replacements in light-chain amyloidosis. Proc. Mol. Cell Biol. 15, 384–396 Natl. Acad. Sci. U. S. A. 91, 5446–5450 2. Chiti, F., and Dobson, C. M. (2017) Protein misfolding, amyloid forma- 23. Wall, J., Schell, M., Murphy, C., Hrncic, R., Stevens, F. J., and Solomon, A. tion, and human disease: A summary of progress over the last decade. (1999) Thermodynamic instability of human lambda 6 light chains: Annu. Rev. Biochem. 86,27–68 Correlation with fibrillogenicity. Biochemistry 38, 14101–14108 3. Gertz, M. A. (2018) Immunoglobulin light chain amyloidosis: 2018 Update 24. Baden, E. M., Randles, E. G., Aboagye, A. K., Thompson, J. R., and on diagnosis, prognosis, and treatment. Am. J. Hematol. 93,1169–1180 Ramirez-Alvarado, M. (2008) Structural insights into the role of muta- 4. Merlini, G. (2017) AL amyloidosis: From molecular mechanisms to targeted tions in amyloidogenesis. J. Biol. Chem. 283, 30950–30956 therapies. Hematology Am. Soc. Hematol. Educ. Program 2017,1–12 25. Nokwe, C. N., Zacharias, M., Yagi, H., Hora, M., Reif, B., Goto, Y., and 5. Murphy, K., and Weaver, C. (2016) Janeway’s Immunobiology, W.W. Buchner, J. (2014) A residue-specific shift in stability and amyloidoge- Norton & Company, New York, NY nicity of antibody variable domains. J. Biol. Chem. 289, 26829–26846 6. Weiss, B. M., Hebreo, J., Cordaro, D. V., Roschewski, M. J., Baker, T. P., 26. Nokwe, C. N., Hora, M., Zacharias, M., Yagi, H., Peschek, J., Reif, B., Abbott, K. C., and Olson, S. W. (2014) Increased serum free light chains Goto, Y., and Buchner, J. (2016) A stable mutant predisposes antibody domains to amyloid formation through specific non-native interactions. precede the presentation of immunoglobulin light chain amyloidosis. J. J. Mol. Biol. 428, 1315–1332 Clin. Oncol. 32, 2699–2704 27. Oberti, L., Rognoni, P., Barbiroli, A., Lavatelli, F., Russo, R., Maritan, M., 7. Zhang, C., Huang, X., and Li, J. (2017) Light chain amyloidosis: Where are Palladini, G., Bolognesi, M., Merlini, G., and Ricagno, S. (2017) Con- the light chains from and how they play their pathogenic role? Blood Rev. current structural and biophysical traits link with immunoglobulin light 31, 261–270 chains amyloid propensity. Sci. Rep. 7, 16809 8. Tonegawa, S. (1983) Somatic generation of antibody diversity. Nature 28. Baden, E. M., Owen, B. A. L., Peterson, F. C., Volkman, B. F., Ramirez- 302, 575–581 Alvarado, M., and Thompson, J. R. (2008) Altered dimer interface de- 9. Abraham, R. S., Geyer, S. M., Ramírez-Alvarado, M., Price-Troska, T. L., Gertz, M. A., and Fonseca, R. (2004) Analysis of somatic hypermutation creases stability in an amyloidogenic protein. J. Biol. Chem. 283, 15853– and antigenic selection in the clonal B cell in immunoglobulin light chain 15860 amyloidosis (AL). J. Clin. Immunol. 24, 340–353 29. DiCostanzo, A. C., Thompson, J. R., Peterson, F. C., Volkman, B. F., and 10. Blancas-Mejia, L. M., Misra, P., Dick, C. J., Cooper, S. A., Redhage, K. R., Ramirez-Alvarado, M. (2012) Tyrosine residues mediate fibril formation Bergman, M. R., Jordan, T. L., Maar, K., and Ramirez-Alvarado, M. (2018) in a dynamic light chain dimer interface. J. Biol. Chem. 287, 27997–28006 Immunoglobulin light chain amyloid aggregation. Chem. Commun. 30. Rennella, E., Morgan, G. J., Kelly, J. W., and Kay, L. E. (2019) Role of (Camb.) 54, 10664–10674 domain interactions in the aggregation of full-length immunoglobulin 11. Buxbaum, J. N. (2004) The systemic amyloidoses. Curr. Opin. Rheumatol. light chains. Proc. Natl. Acad. Sci. U. S. A. 116, 854–863 16,67–75 31. Bork, P., Holm, L., and Sander, C. (1994) The immunoglobulin fold. 12. Morgan, G. J., and Kelly, J. W. (2016) The kinetic stability of a full-length Structural classification, sequence patterns and common core. J. Mol. antibody light chain dimer determines whether endoproteolysis can Biol. 242, 309–320 release amyloidogenic variable domains. J. Mol. Biol. 428, 4280–4297 32. el Hawrani, A. S., Moreton, K. M., Sessions, R. B., Clarke, A. R., and Hol- 13. Enqvist, S., Sletten, K., Stevens, F. J., Hellman, U., and Westermark, P. brook, J. J. (1994) Engineering surface loops of proteins–a preferred strategy (2007) Germ line origin and somatic mutations determine the target for obtaining new enzyme function. Trends Biotechnol. 12,207–211 tissues in systemic AL-amyloidosis. PLoS One 2, e981 33. Chang, H.-J., Jian, J.-W., Hsu, H.-J., Lee, Y.-C., Chen, H.-S., You, J.-J., 14. Enqvist, S., Sletten, K., and Westermark, P. (2009) Fibril protein frag- Hou, S.-C., Shao, C.-Y., Chen, Y.-J., Chiu, K.-P., Peng, H.-P., Lee, K. H., mentation pattern in systemic AL-amyloidosis. J. Pathol. 219, 473–480 and Yang, A.-S. (2014) Loop-sequence features and stability determinants 15. Lecoq, L., Wiegand, T., Rodriguez-Alvarez, F. J., Cadalbert, R., Herrera, in antibody variable domains by high-throughput experiments. Structure G. A., Del Pozo-Yauner, L., Meier, B. H., and Böckmann, A. (2019) 22,9–21 34. Rouet, R., Lowe, D., and Christ, D. (2014) Stability engineering of the A substantial structural conversion of the native monomer leads to in- human antibody repertoire. FEBS Lett. 588, 269–277 register parallel amyloid fibril formation in light-chain amyloidosis. 35. Davis, M. M. (2004) The evolutionary and structural ‘logic’ of antigen Chembiochem 20, 1027–1031 receptor diversity. Semin. Immunol. 16, 239–243 16. Radamaker, L., Lin, Y.-H., Annamalai, K., Huhn, S., Hegenbart, U., 36. Annamalai, K., Liberta, F., Vielberg, M.-T., Close, W., Lilie, H., Gührs, K.- Schönland, S. O., Fritz, G., Schmidt, M., and Fändrich, M. (2019) Cryo- H., Schierhorn, A., Koehler, R., Schmidt, A., Haupt, C., Hegenbart, U., EM structure of a light chain-derived amyloid fibril from a patient with Schönland, S., Schmidt, M., Groll, M., and Fändrich, M. (2017) Common systemic AL amyloidosis. Nat. Commun. 10, 1103 fibril structures imply systemically conserved protein misfolding path- 17. Swuec, P., Lavatelli, F., Tasaki, M., Paissoni, C., Rognoni, P., Maritan, M., Brambilla, F., Milani, P., Mauri, P., Camilloni, C., Palladini, G., Merlini, G., ways in vivo. Angew. Chem. Int. Ed. Engl. 56, 7510–7514 Ricagno, S., and Bolognesi, M. (2019) Cryo-EM structure of cardiac 37. Brochet, X., Lefranc, M.-P., and Giudicelli, V. (2008) IMGT/V-QUEST: amyloid fibrils from an immunoglobulin light chain AL amyloidosis pa- The highly customized and integrated system for IG and TR stan- tient. Nat. Commun. 10, 1269 dardized V-J and V-D-J sequence analysis. Nucleic Acids Res. 36, 18. Hora, M., Sarkar, R., Morris, V., Xue, K., Prade, E., Harding, E., Buchner, W503–W508 J., and Reif, B. (2017) MAK33 antibody light chain amyloid fibrils are 38. Lefranc, M.-P., Giudicelli, V., Duroux, P., Jabado-Michaloud, J., Folch, G., similar to oligomeric precursors. PLoS One 12, e0181799 Aouinti, S., Carillon, E., Duvergey, H., Houles, A., Paysan-Lafosse, T., 19. Pradhan, T., Annamalai, K., Sarkar, R., Huhn, S., Hegenbart, U., Schön- Hadi-Saljoqi, S., Sasorith, S., Lefranc, G., and Kossida, S. (2014) IMGT®, land, S., Fändrich, M., and Reif, B. (2020) Seeded fibrils of the germline the international ImMunoGeneTics information system® 25 years on. variant of human λ-III immunoglobulin light chain FOR005 have a Nucleic Acids Res. 43, D413–D422 similar core as patient fibrils with reduced stability. J. Biol. Chem. 295, 39. Swindells, M. B., Porter, C. T., Couch, M., Hurst, J., Abhinandan, K. R., 18474–18484 Nielsen, J. H., Macindoe, G., Hetherington, J., and Martin, A. C. R. (2017) 20. Radamaker, L., Baur, J., Huhn, S., Haupt, C., Hegenbart, U., Schönland, S., abYsis: Integrated antibody sequence and structure-management, anal- Bansal, A., Schmidt, M., and Fändrich, M. (2021) Cryo-EM reveals ysis, and prediction. J. Mol. Biol. 429, 356–364 12 J. Biol. Chem. (2021) 296 100334 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis 40. Ye, J., Ma, N., Madden, T. L., and Ostell, J. M. (2013) IgBLAST: An 59. Fontana, A., de Laureto, P. P., de Filippis, V., Scaramella, E., and Zam- immunoglobulin variable domain sequence analysis tool. Nucleic Acids bonin, M. (1999) Limited proteolysis in the study of protein conforma- Res. 41,W34–W40 tion. In: Sterchi, E. E., Stöcker, W., eds. Proteolytic Enzymes: Tools and 41. Perfetti, V., Casarini, S., Palladini, G., Vignarelli, M. C., Klersy, C., Diegoli, Targets, Springer Berlin Heidelberg, Berlin, Heidelberg: 253–280 M., Ascari, E., and Merlini, G. (2002) Analysis of V(lambda)-J(lambda) 60. Englander, S. W. (2006) Hydrogen exchange and mass spectrometry: A expression in plasma cells from primary (AL) amyloidosis and normal historical perspective. J. Am. Soc. Mass Spectrom. 17, 1481–1489 bone marrow identifies 3r (lambdaIII) as a new amyloid-associated 61. Klimtchuk, E. S., Gursky, O., Patel, R. S., Laporte, K. L., Connors, L. H., germline gene segment. Blood 100, 948–953 Skinner, M., and Seldin, D. C. (2010) The critical role of the constant 42. Villalba, M. I., Canul-Tec, J. C., Luna-Martínez, O. D., Sánchez-Alcalá, R., region in thermal stability and aggregation of amyloidogenic immuno- Olamendi-Portugal, T., Rudiño-Piñera, E., Rojas, S., Sánchez-López, R., globulin light chain. Biochemistry 49, 9848–9857 Fernández-Velasco, D. A., and Becerril, B. (2015) Site-directed muta- 62. Weber, B., Hora, M., Kazman, P., Göbl, C., Camilloni, C., Reif, B., and genesis reveals regions implicated in the stability and fiber formation of Buchner, J. (2018) The antibody light-chain linker regulates domain human λ3r light chains. J. Biol. Chem. 290, 2577–2592 orientation and amyloidogenicity. J. Mol. Biol. 430, 4925–4940 43. Chothia, C., and Lesk, A. M. (1987) Canonical structures for the hyper- 63. Nokwe, C. N., Hora, M., Zacharias, M., Yagi, H., John, C., Reif, B., Goto, variable regions of immunoglobulins. J. Mol. Biol. 196, 901–917 Y., and Buchner, J. (2015) The antibody light-chain linker is important for 44. Wu, T. T., and Kabat, E. A. (1970) An analysis of the sequences of the domain stability and amyloid formation. J. Mol. Biol. 427, 3572–3586 variable regions of Bence Jones proteins and myeloma light chains and 64. Luna-Martínez, O. D., Hernández-Santoyo, A., Villalba-Velázquez, M. I., their implications for antibody complementarity. J. Exp. Med. 132, 211– Sánchez-Alcalá,R., Fernández-Velasco, D. A.,and Becerril,B.(2017) Stabi- 250 lizing an amyloidogenic λ6 light chain variable domain. FEBS J. 284, 3702–3717 45. Ashkenazy, H., Abadi, S., Martz, E., Chay, O., Mayrose, I., Pupko, T., and 65. Del Pozo-Yauner, L., Wall, J. S., González Andrade, M., Sánchez-López, Ben-Tal, N. (2016) ConSurf 2016: An improved methodology to estimate R., Rodríguez-Ambriz, S. L., Pérez Carreón, J. I., Ochoa-Leyva, A., and and visualize evolutionary conservation in macromolecules. Nucleic Acids Fernández-Velasco, D. A. (2014) The N-terminal strand modulates Res. 44, W344–W350 immunoglobulin light chain fibrillogenesis. Biochem. Biophys. Res. Com- 46. Tsolis, A. C., Papandreou, N. C., Iconomidou, V. A., and Hamodrakas, S. mun. 443, 495–499 J. (2013) A consensus method for the prediction of ’aggregation-prone’ 66. González-Andrade, M., Becerril-Luján, B., Sánchez-López, R., Ceceña- peptides in globular proteins. PLoS One 8, e54175 Álvarez, H., Pérez-Carreón, J. I., Ortiz, E., Fernández-Velasco, D. A., and 47. Emily, M., Talvas, A., and Delamarche, C. (2013) MetAmyl: A METa- Del Pozo-Yauner, L. (2013) Mutational and genetic determinants of λ6 predictor for AMYLoid proteins. PLoS One 8, e79722 light chain amyloidogenesis. FEBS J. 280, 6173–6183 48. Goldschmidt, L., Teng, P. K., Riek, R., and Eisenberg, D. (2010) Identifying 67. Hernández-Santoyo, A., del Pozo Yauner, L., Fuentes-Silva, D., Ortiz, E., the amylome, proteins capable of forming amyloid-like fibrils. Proc. Natl. Rudiño-Piñera, E., Sánchez-López, R., Horjales, E., Becerril, B., and Acad. Sci. U. S. A. 107, 3487 Rodríguez-Romero, A. (2010) A single mutation at the sheet switch re- 49. Waterhouse, A., Bertoni, M., Bienert, S., Studer, G., Tauriello, G., gion results in conformational changes favoring lambda6 light-chain Gumienny, R., Heer, F. T., de Beer, T. A. P., Rempfer, C., Bordoli, L., fibrillogenesis. J. Mol. Biol. 396, 280–292 Lepore, R., and Schwede, T. (2018) SWISS-MODEL: Homology model- 68. Rennella, E., Morgan, G. J., Yan, N., Kelly, J. W., and Kay, L. E. (2019) The ling of protein structures and complexes. Nucleic Acids Res. 46, W296– role of protein thermodynamics and primary structure in fibrillogenesis of W303 variable domains from immunoglobulin light chains. J. Am. Chem. Soc. 50. Woody, R. W. (1995) [4] Circular Dichroism. Biochemical Spectroscopy, 141, 13562–13571 Elsevier, Amsterdam, Netherlands 69. Helms, L. R., and Wetzel, R. (1995) Destabilizing loop swaps in the CDRs 51. Gade Malmos, K., Blancas-Mejia, L. M., Weber, B., Buchner, J., Ram- of an immunoglobulin VL domain. Protein Sci. 4, 2073–2081 irez-Alvarado, M., Naiki, H., and Otzen, D. (2017) ThT 101: A primer 70. Gąciarz, A., and Ruddock, L. W. (2017) Complementarity determining on the use of thioflavin T to investigate amyloid formation. Amyloid 24, regions and frameworks contribute to the disulfide bond independent 1–16 folding of intrinsically stable scFv. PLoS One 12, e0189964 52. Khurana, R., Gillespie, J. R., Talapatra, A., Minert, L. J., Ionescu-Zanetti, 71. Herold, E. M., John, C., Weber, B., Kremser, S., Eras, J., Berner, C., C., Millett, I., and Fink, A. L. (2001) Partially folded intermediates as Deubler, S., Zacharias, M., and Buchner, J. (2017) Determinants of critical precursors of light chain amyloid fibrils and amorphous aggre- the assembly and function of antibody variable domains. Sci. Rep. 7, gates. Biochemistry 40, 3525–3535 12276 53. Creighton, T. E. (1993) Proteins: Structures and Molecular Properties, 72. van der Kant, R., Bauer, J., Karow-Zwick, A. R., Kube, S., Garidel, P., Macmillan, New York, NY Blech, M., Rousseau, F., and Schymkowitz, J. (2019) Adaption of human 54. Feige, M. J., Hendershot, L. M., and Buchner, J. (2010) How antibodies antibody λ and κ light chain architectures to CDR repertoires. Protein fold. Trends Biochem. Sci. 35, 189–198 Eng. Des. Sel. 32, 109–127 55. Buchner, J., Renner, M., Lilie, H., Hinz, H. J., Jaenicke, R., Kiefhabel, T., 73. Blancas-Mejía, L. M., Tischer, A., Thompson, J. R., Tai, J., Wang, L., and Rudolph, R. (1991) Alternatively folded states of an immunoglobulin. Auton, M., and Ramirez-Alvarado, M. (2014) Kinetic control in protein Biochemistry 30, 6922–6929 folding for light chain amyloidosis and the differential effects of somatic 56. Kazman, P., Vielberg, M.-T., Pulido Cendales, M. D., Hunziger, L., mutations. J. Mol. Biol. 426, 347–361 Weber, B., Hegenbart, U., Zacharias, M., Köhler, R., Schönland, S., Groll, 74. Blancas-Mejía, L. M., Horn, T. J., Marin-Argany, M., Auton, M., Tischer, M., and Buchner, J. (2020) Fatal amyloid formation in a patient’s antibody A., and Ramirez-Alvarado, M. (2015) Thermodynamic and fibril forma- light chain is caused by a single point mutation. eLife 9, e52300 tion studies of full length immunoglobulin light chain AL-09 and its 57. Le Marchand, T., de Rosa, M., Salvi, N., Sala, B. M., Andreas, L. B., germline protein using scan rate dependent thermal unfolding. Biophys. Barbet-Massin, E., Sormanni, P., Barbiroli, A., Porcari, R., Sousa Mota, C., Chem. 207,13–20 de Sanctis, D., Bolognesi, M., Emsley, L., Bellotti, V., Blackledge, M., et al. 75. Ruiz-Zamora, R. A., Guillaumé, S., Al-Hilaly, Y. K., Al-Garawi, Z., (2018) Conformational dynamics in crystals reveal the molecular bases Rodríguez-Alvarez, F. J., Zavala-Padilla, G., Pérez-Carreón, J. I., Rodrí- for D76N beta-2 microglobulin aggregation propensity. Nat. Commun. 9, guez-Ambriz, S. L., Herrera, G. A., Becerril-Luján, B., Ochoa-Leyva, A., 1658 Melendez-Zajgla, J., Serpell, L., and Del Pozo-Yauner, L. (2019) The 58. Maritan, M., Romeo, M., Oberti, L., Sormanni, P., Tasaki, M., Russo, R., CDR1 and other regions of immunoglobulin light chains are hot spots for Ambrosetti, A., Motta, P., Rognoni, P., Mazzini, G., Barbiroli, A., Palla- amyloid aggregation. Sci. Rep. 9, 3123 dini, G., Vendruscolo, M., Diomede, L., Bolognesi, M., et al. (2020) 76. Brumshtein,B., Esswein, S. R.,Sawaya, M. R.,Rosenberg,G., Ly,A. Inherent biophysical properties modulate the toxicity of soluble amyloi- T.,Landau, M.,and Eisenberg, D. S. (2018) Identification of two dogenic light chains. J. Mol. Biol. 432, 845–860 principal amyloid-driving segments in variable domains of Ig light J. Biol. Chem. (2021) 296 100334 13 EDITORS’ PICK: Role of CDR mutations in light chain amyloidosis chains in systemic light-chain amyloidosis. J. Biol. Chem. 293, andHiggins,D.G.(2011) Fast,scalable generation of high-quality 19659–19671 protein multiple sequence alignments using Clustal Omega. Mol. 77. Poshusta, T. L., Sikkink, L. A., Leung, N., Clark, R. J., Dispenzieri, A., and Syst. Biol. 7,539 Ramirez-Alvarado, M. (2009) Mutations in specific structural regions of 85. Thies, M. J. W., and Pirkl, F. (2000) Chromatographic purification of the immunoglobulin light chains are associated with free light chain levels in CH2 domain of the monoclonal antibody MAK33. J. Chromatogr. B patients with AL amyloidosis. PLoS One 4, e5169 Biomed. Sci. Appl. 737,63–69 78. Feige, M. J., Groscurth, S., Marcinowski, M., Yew, Z. T., Truffault, V., 86. Brown, P. H., and Schuck, P. (2006) Macromolecular size-and-shape Paci, E., Kessler, H., and Buchner, J. (2008) The structure of a folding distributions by sedimentation velocity analytical ultracentrifugation. intermediate provides insight into differences in immunoglobulin amy- Biophys. J. 90, 4651–4661 loidogenicity. Proc. Natl. Acad. Sci. U. S. A. 105, 13373–13378 87. Schuck, P. (2000) Size-distribution analysis of macromolecules by sedi- 79. Marin-Argany, M., Güell-Bosch, J., Blancas-Mejía, L. M., Villegas, S., and mentation velocity ultracentrifugation and Lamm equation modeling. Ramirez-Alvarado, M. (2015) Mutations can cause light chains to be too Biophys. J. 78, 1606–1619 stable or too unstable to form amyloid fibrils. Protein Sci. 24, 1829–1840 88. Santoro, M. M., and Bolen, D. W. (1988) Unfolding free energy changes 80. Chothia, C., Lesk, A. M., Tramontano, A., Levitt, M., Smith-Gill, S. J., determined by the linear extrapolation method. 1. Unfolding of phenyl- Air, G., Sheriff, S., Padlan, E. A., Davies, D., Tulip, W. R., Colman, P. M., methanesulfonyl alpha-chymotrypsin using different denaturants. Spinelli, S., Alzari, P. M., and Poljak, R. J. (1989) Conformations of Biochemistry 27, 8063–8068 immunoglobulin hypervariable regions. Nature 342, 877–883 89. Abràmoff, M. D., Magalhães, P. J., and Ram, S. J. (2004) Image processing 81. Al-Lazikani, B., Lesk, A. M., and Chothia, C. (1997) Standard confor- with ImageJ. Biophotonics Int. 11,36–42 mations for the canonical structures of immunoglobulins. J. Mol. Biol. 90. Weber, B., Brandl, M. J., Pulido Cendales, M. D., Berner, C., Pradhan, T., 273, 927–948 Feind, G. M., Zacharias, M., Reif, B., and Buchner, J. (2018) A single 82. Dondelinger, M., Filée, P., Sauvage, E., Quinting, B., Muyldermans, S., residue switch reveals principles of antibody domain integrity. J. Biol. Galleni, M., and Vandevenne, M. S. (2018) Understanding the signifi- Chem. 293, 17107–17118 cance and implications of antibody numbering and antigen-binding sur- 91. Case, D. A., Ben-Shalom, I. Y., Brozell, S. R., Cerutti, D. S., Cheatham, T. face/residue definition. Front. Immunol. 9, 2278 E., III, Cruzeiro, V. W. D., Darden, T. A., Duke, R. E., Ghoreishi, D., 83. del Pozo Yauner, L., Ortiz, E., and Becerril, B. (2006) The CDR1 of the Gilson, M. K., Gohlke, H., Goetz, A. W., Greene, D., Harris, R., Homeyer, human lambdaVI light chains adopts a new canonical structure. Proteins N., et al. (2018) Amber 18, University of California, San Francisco, CA 62, 122–129 92. Jorgensen, W. L., Chandrasekhar, J., Madura, J. D., Impey, R. W., and 84. Sievers, F.,Wilm,A., Dineen,D., Gibson,T.J., Karplus, K.,Li, W., Klein, M. L. (1983) Comparison of simple potential functions for simu- Lopez, R.,McWilliam, H., Remmert,M., Söding,J., Thompson,J. D., lating liquid water. J. Chem. Phys. 79, 926–935 14 J. Biol. Chem. (2021) 296 100334

Journal

Journal of Biological ChemistryUnpaywall

Published: Jan 1, 2021

There are no references for this article.