RESEARCH ARTICLE Disease-associated mutations impacting BC-loop flexibility trigger long-range transthyretin tetramer destabilization and aggregation Received for publication, April 6, 2021, and in revised form, July 30, 2021 Published, Papers in Press, July 31, 2021, https://doi.org/10.1016/j.jbc.2021.101039 Sebastián A. Esperante1, *, Nathalia Varejao1 , Francisca Pinheiro1, Ricardo Sant’Anna1, Juan Román Luque-Ortega2, Carlos Alfonso3 , Valentina Sora4,5, Elena Papaleo4,5, Germán Rivas3, David Reverter1, and Salvador Ventura1, * From the 1Institut de Biotecnologia i de Biomedicina and Departament de Bioquímica i de Biologia Molecular, Universitat Autònoma de Barcelona, Bellaterra, Barcelona, Spain; 2Molecular Interactions Facility and 3Systems Biochemistry of Bacterial Division Laboratory, Centro de Investigaciones Biológicas Margarita Salas, CSIC, Madrid, Spain; 4Computational Biology Laboratory, Danish Cancer Society Research Center, Copenhagen, Denmark; and 5Cancer Systems Biology, Health and Technology Department, Section for Bioinformatics, Technical University of Denmark, Lyngby, Denmark Edited by Wolfgang Peti Hereditary transthyretin amyloidosis (ATTR) is an autosomal Amyloid diseases constitute a heterogeneous group of dominant disease characterized by the extracellular deposition clinical disorders characterized by protein misfolding, aggre- of the transport protein transthyretin (TTR) as amyloid fibrils. gation, and the deposition of insoluble fibrils in a variety of Despite the progress achieved in recent years, understanding tissues and organs (1, 2). Although amyloidogenic proteins are why different TTR residue substitutions lead to different clin- derived from different and unrelated sources, they all converge ical manifestations remains elusive. Here, we studied the mo- to form similar structures made by cross-β sheets, termed lecular basis of disease-causing missense mutations affecting amyloid fibrils. Amyloid deposition can be confined to a residues R34 and K35. R34G and K35T variants cause vitreous particular organ or tissue, leading to localized amyloidosis (e.g., amyloidosis, whereas R34T and K35N mutations result in am- Alzheimer’s disease, prion diseases, type 2 diabetes mellitus). yloid polyneuropathy and restrictive cardiomyopathy. All vari- Conversely, protein deposition can also occur systemically ants are more sensitive to pH-induced dissociation and amyloid throughout the body, leading to systemic amyloidosis (e.g., formation than the wild-type (WT)-TTR counterpart, specif- amyloid light chain (AL), amyloid-A (AA), and hemodialysis- ically in the variants deposited in the eyes amyloid formation associated (Aβ2m), amyloidosis) (3). occurs close to physiological pHs. Chemical denaturation ex- Transthyretin (TTR) is one of more than 30 amyloidogenic periments indicate that all the mutants are less stable than WT- proteins associated with amyloid diseases (4). TTR is an TTR, with the vitreous amyloidosis variants, R34G and K35T, extracellular soluble nonglycosylated tetrameric protein syn- being highly destabilized. Sequence-induced stabilization of the thesized mainly in the liver, retinal pigment epithelium, dimer–dimer interface with T119M rendered tetramers con- pancreas, and choroid plexus. TTR transports holoretinol- taining R34G or K35T mutations resistant to pH-induced ag- binding protein and is a minor carrier of thyroxine (T4) in gregation. Because R34 and K35 are among the residues more the blood, whereas in the cerebrospinal fluid, it is the primary distant to the TTR interface, their impact in this region is carrier of T4 (5). Extracellular misfolding and misassembly of therefore theorized to occur at long range. The crystal struc- TTR lead to the formation and accumulation of amyloid fibrils tures of double mutants, R34G/T119M and K35T/T119M, in a variety of tissues, which gives rise to distinct progressive together with molecular dynamics simulations indicate that clinical syndromes known as TTR-related amyloidosis. The their strong destabilizing effect is initiated locally at the BC propensity of TTR to form amyloids is associated with aging loop, increasing its flexibility in a mutation-dependent manner. and/or mutations that destabilize the native state. Deposition Overall, the present findings help us to understand the of wild-type (WT) TTR within the extracellular matrix of the sequence-dynamic-structural mechanistic details of TTR amy- heart and other tissues causes senile systemic amyloidosis loid aggregation triggered by R34 and K35 variants and to link (SSA), a late-onset sporadic cardiomyopathy, affecting as much the degree of mutation-induced conformational flexibility to as 25% of the population over 80 years of age (6, 7). protein aggregation propensity. More than 130 mutations within the TTR gene have been associated with autosomal dominant familial forms of amyloidosis, which typically present earlier onset and are often severe (8). Most of the pathogenic variants identified so far * For correspondence: Sebastián A. Esperante, sesperante@infant.org.ar; correspond to missense mutations and display tissue-selective Salvador Ventura, salvador.ventura@uab.es. amyloid deposition and pathology. Familial amyloid poly- Present address for Sebastián A. Esperante: INFANT Foundation and CONI- CET, Buenos Aires, Argentina. neuropathy (FAP) and familial amyloid cardiomyopathy (FAC) J. Biol. Chem. (2021) 297(3) 101039 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/). High TTR BC-loop dynamics causes high destabilization are the most common forms of hereditary amyloidosis, results in amyloidosis. However, this generic mechanism does affecting the peripheral nervous system and the heart, not explain why the R34G- and K35T-TTR variants were respectively. The V30M variant is the most common mutation identified in patients with vitreous amyloidosis (24, 25), associated with FAP (9), whereas V122I and I68L are pre- whereas R34T and K35N mutations were described in families dominantly associated with FAC (10, 11). Highly destabilizing with amyloid polyneuropathy and restrictive cardiomyopathy mutations, such as A25T (12) and D18G (13), cause selective (26, 27). amyloidosis restricted to the central nervous system. These In the present study, we investigate the pathogenic mecha- variants are synthesized in the liver, but the misfolded proteins nism of missense mutations affecting residues R34 and K35 of are retained in the endoplasmic reticulum and degraded TTR, using computational, biophysical, and structural ap- intracellularly by the endoplasmic reticulum-associated proaches. The results contribute to understand why different degradation (ERAD) mechanism (14). Contrarily, when the mutations of the same TTR amino acid and the same substi- synthesis occurs in the choroid plexus due to the high avail- tution in adjacent residues in a given structural element might ability of T4, this small molecule acts as a metabolite chap- elicit different disease manifestations. erone, stabilizing the tetramer, which evades the proteostasis ERAD mechanism. Once secreted, dissociation of T4 promotes Results the destabilized TTR tetramer to dissociate, misfold, and pH-induced amyloidogenicity of missense mutations at aggregate. residues 34 and 35 of TTR Even though most of the mutations destabilize the TTR tetramer and favor amyloid formation, some disease protective WT-TTR is converted into amyloid fibrils by pH-mediated mutations have also been identified. Heterozygotes harboring tetramer dissociation. The dissociation is linked to tertiary V30M and T119M mutations were protected from developing structural changes resulting in the formation of a monomeric FAP (15). The presence of T119M in the heterotetramers re- amyloidogenic intermediate. The optimal pH to induce WT- duces its dissociation rate and, consequently, the aggregation TTR tetramer dissociation and the concomitant monomeric rate and amyloidogenesis progression. This mechanism, structural rearrangement is pH 4.4, whereas at pHs closer to known as kinetic stabilization, laid the foundation for thera- physiological values, WT-TTR is unable to form amyloid fi- peutic strategies to ameliorate TTR amyloidosis (16). Several brils (21). We started investigating the effect of missense small molecules that bind and kinetically stabilize the non- mutations involving residues R34 and K35 of TTR on pH- amyloidogenic tetramer have been identified. For instance, the induced amyloidogenicity. The dependence of aggregation benzoxazole tafamidis (17) has shown safety and efficacy in and amyloid formation of TTR variants on pH was monitored slowing disease progression of polyneuropathy (18) and car- by measuring turbidity at 330 nm and ThT binding, respec- diomyopathy (19), and it was approved by different regulatory tively (Fig. 1, A and B) and compared with those of WT-TTR agencies worldwide. Tolcapone is another promising potent and the L55P-TTR mutant (28). Both aggregation and amyloid kinetic stabilizer of TTR that crosses the blood–brain barrier formation increase, as we move from pH 7.0 to pH 4.4 in all and might find therapeutic application in CNS amyloidosis the studied variants. Surprisingly, at pH 5.6 and 6.0, the R34G- (9, 11, 20). and K35T-TTR mutants displayed similar turbidity and ThT- The sequence-dependent factors that lead to TTR aggre- binding ability than that of L55P-TTR, the most pathogenic gation have been extensively studied over the last 25 years (14, TTR mutant described so far. It should be noted that at pH 7.0, 21, 22). WT-TTR homotetramers or heterotetramers all TTR variants exhibited low turbidity, but R34G-, K35T-, comprised of mutant and WT subunits dissociate into and L55P-TTR displayed a 20-fold increase in ThT-binding monomers, which can rapidly misfold and subsequently self- fluorescence intensity compared with WT-TTR. This finding assemble into amyloid fibrils. It is well established that TTR suggests some extent of detectable amyloid formation at tetramer dissociation is the rate-limiting step for amyloid fibril neutral pH, as previously described for L55P-TTR (28). The formation. In this scenario, pathogenic mutations can desta- morphological properties of the aggregates formed upon in- bilize the tetramer, decreasing intersubunit affinity and/or cubation at pH 4.4 were further evaluated by transmission increasing the dissociation rate. Once the tetramer is dissoci- electron microscopy (TEM) that confirmed the presence of ated, mutations can influence the monomer thermodynamic protofibrilar aggregates in all cases (Fig. 1C). Thus, the posi- stability, thus increasing the population of the aggregation- tively charged residues at positions 34 and 35 are critical for prone misfolded monomer. However, we still do not under- preserving the native, nonaggregation prone conformation of stand why a mutation in a given residue and especially why TTR. Strikingly, it is an interplay between the mutated position different substitutions of the same residue may lead to and the substitution type, which seems to determine the TTR different clinical manifestations. This is exemplified by the variant aggregation behavior. R34G, R34T, K35N, and K35T mutations identified in patients with familial amyloidosis. R34 and K35 are two positively Conformational stability of TTR variants charged residues residing at the C-terminus of TTR β-strand TTR pathogenic mutations reported so far affect either B. We have shown that, in vitro, these sequentially adjacent tetramer stability or thermodynamics of misfolding of mono- positive charges contribute to protect TTR against aggregation meric TTR. To understand the pathogenic mechanism of (23), which may justify why the loss of one of these charges missense mutations involving residues 34 and 35, we 2 J. Biol. Chem. (2021) 297(3) 101039 High TTR BC-loop dynamics causes high destabilization Figure 1. pH dependence on aggregation propensity and amyloid protofibril formation of TTR disease associated variants. 3.5 μM of WT and pathogenic TTR variants were incubated in 100 mM Tris.HCl, 50 mM MES, 50 mM sodium acetate, and 0.1 M KCl at different pHs (ranging from 4.4 to 7.0), for 72 h at 37  C prior measurements. The amount of protofibril formation as a function of pH was determined by measuring the suspension turbidity at 330 nm (A) and the increase in fluorescence intensity at 482 nm in ThT-binding assays (B). Each experiment was performed in triplicates. C, TEM images of the aggregates of TTR variants, formed at pH 4.4 at 37  C. Scale bars values are indicated. investigated the effects of these variants on TTR tertiary and resistant to urea denaturation. Thus, two types of TTR quaternary structural stabilities. We probed the tertiary tetramer populations likely coexist, the nonstabilized tetra- structure’s conformational stability by urea denaturation ex- mers, which dissociate and unfold with a transition midpoint periments monitored by tryptophan fluorescence spectroscopy Cm1 = 3.2 M (at 1.5 μM protein concentration), and stabilized (Fig. 2). Tryptophan fluorescence emission spectra of TTR tetramers, which cannot be completely unfolded on the variants in native conditions revealed that the emission experimental timescale. maximum wavelength of K35T, and especially K35N, was red- The R34-TTR variants exhibit two separated transitions, the shifted relative to R34 variants and WT-TTR, suggesting a first one with an estimated midpoint of Cm1 = 1.7 M for R34G distinct tryptophan environment (Fig. S1). The four studied and Cm1 = 2.2 M for R34T. At the urea concentration range variants, together with WT- and L55P-TTRs, were analyzed to from 4.0 to 6.0 M, the 355/335 intensity signal stabilizes, while compare their dissociation/unfolding mechanisms. It is above 6.0 M, a second transition is observed with midpoints of important to note that after 96 h incubation, WT-TTR did not Cm2 = 5.0 M and Cm2 = 7.0 M for R34G and R34T, respectively reach a stable unfolded state baseline at high urea concen- (Fig. 2). Noticeably, the first transition of R34T is coincident trations, indicating that the protein was not fully unfolded. with the single broad transition displayed by the L55P-TTR Completely unfolding was achieved using Gdm.HCl as dena- mutant. Altogether, the specific residue at position 34 affects turant (Fig. S2). Remarkably, the 355/335 ratio and the fluo- both transitions differently, with the glycine residue exerting rescence center of spectral mass value of the Gdm.HCl the largest destabilization effect, as evidenced by the shift of denaturation midpoint coincide with that of 5.0 M urea both transition midpoints to lower urea concentrations. (Fig. S2). This finding suggests that, in our assay conditions, The urea denaturation curves of the missense mutants at after 96 h incubation in 5.0 M urea, 50% of WT-TTR re- residue K35 also displayed two transitions, the first one with an mains tetrameric. This urea denaturation biphasic conforma- estimated Cm1 of 2.2 M for K35N and 2.1 M for K35T, and a tional behavior was previously reported (29) and attributed to second transition midpoint with Cm2 = 6.0 M for K35T and a fraction of anion-stabilized tetrameric TTR, which is highly Cm2 = 7.0 M for K35N. The urea denaturation curves for R34 J. Biol. Chem. (2021) 297(3) 101039 3 High TTR BC-loop dynamics causes high destabilization Figure 2. Urea equilibrium denaturation of TTR disease associated variants.1.5 μM of WT- and pathogenic TTR variants were incubated for 96 h at 25  C in 50 mM sodium phosphate (pH 7.4) containing 0.1 M KCl with increasing urea concentrations prior measurements of tryptophan fluorescence emission spectra. The tryptophan fluorescence emission intensity ratio displayed is defined as the ratio of the tryptophan emission intensity at 355 nm (unfolded state) to the tryptophan emission intensity at 335 nm (folded state) and is used as a measure of foldedness as was previously described (22). The missense variants in residues 34 (A) and 35 (B) are plotted in the same graph with the denaturation curves of WT-TTR and L55P TTR to compare the denaturation mechanism. and K35 mutants suggest a similar dissociation/unfolding tetramers turned to be less stable than their R34T and R35N mechanism. Compared with WT-TTR, both transitions are counterparts, all the mutants being destabilized relative to shifted to lower urea concentrations, suggesting destabilization WT-TTR. of both tetramer populations mentioned above. Urea denaturation curves of TTR variants at increasing protein concentrations revealed an increase in the first transi- Quaternary structure of TTR urea denaturation intermediates tion midpoint as the protein concentration increases, indicative TTR variants were incubated at 5 μM and 15 μM in 2.5 M of a dissociation process. The second transitions were nearly urea for 96 h and subjected to sedimentation velocity analytical coincident, irrespective of the protein concentration (Fig. S3). ultracentrifugation (AUC) studies to characterize the nature of Resveratrol binding is an ideal probe to assess TTR qua- the denaturation intermediates. The results obtained are ternary structural changes, as the tetramer can bind resvera- summarized in Tables 1 and 2. Under the assayed conditions, trol, whereas the monomer cannot (22). Quaternary structural sedimentation velocity experiments showed WT-TTR and changes induced by increasing urea concentration were T119M-TTR at 5 μM behaving as a single species with an monitored spectroscopically by measuring resveratrol fluo- experimental sedimentation coefficient of 3.5 S (s20,w= 4.1 S) rescence. Accordingly, two transitions were also observed, compatible with the globular tetrameric form of the protein overlapping with the tryptophan fluorescence center of spec- (Fig. 3 and Table 1). When assayed at 15 μM, both WT- and tral mass (Fig. S4). Resveratrol-binding estimates of the T119M-TTR proteins behave as tetramers, in equilibrium with tetramer percentage as a function of urea concentration reveal a small fraction of higher-order soluble oligomers, accounting that the first posttransition baseline corresponds to 30–50% of for 4.3% and 2.5% of total protein, respectively (Table 2). As tetramers (Fig. S4). previously reported, to trigger its dissociation to monomer, To better understand the chemical denaturation mecha- T119M-TTR was subjected to freeze–thawing cycles after the nism, equilibrium denaturation experiments were also per- 96 h incubation with urea (30). Under these conditions, be- formed using Gdm.HCl as a denaturant. All TTR variants sides the peak corresponding to the T119M-TTR tetramer showed a single transition, displaying an estimated Cm = 3.5 M for both R34G and K35T, a Cm = 4.5 for K35N and Cm = 4.7 for R34T-TTR (Fig. S5). WT-TTR also displayed a single Table 1 transition with a Cm = 5.1 M Gdm.HCl, indicating that the Normalized sedimentation coefficient in water at 20  C (sw20) variants destabilized the tetramer to different extents. Size- 2.5 M urea c(sw20) monomer c(sw20) tetramer exclusion chromatography performed at 2.0 M Gdm.HCl WT-TTR - 4.12 ± 0.07 L55P-TTR 1.41 ± 0.11 4.08 ± 0.04 revealed that all the variants remain tetrameric in the pre- R34T-TTR 1.30 ± 0.07 4.06 ± 0.13 transition region (Fig. S5), suggesting a simple two-state K35N-TTR 1.47 ± 0.01 4.07 ± 0.02 K35T-TTR 1.40 ± 0.08 4.18 ± 0.07 mechanism (tetramer to unfolded monomer) or a three-state R34G-TTR 1.49 ± 0.01 4.07 ± 0.05 mechanism with a single transition (tetramer dissociation to T119M-TTRa 1.66 ± 0.08 4.10 ± 0.06 monomer and monomer unfolding linked). Samples were incubated prior measurements for 96 h at 5.0 μM tetramer concentration in 50 mM sodium phosphate (pH 7.4), 0.1 M KCl and 2.5 M urea. Importantly, independent of the employed denaturant and a T119M-TTR was subjected to freeze–thawing cycles to induce tetramer dissociation probe, the more amyloidogenic R34G and K35T TTR to monomer. 4 J. Biol. Chem. (2021) 297(3) 101039 High TTR BC-loop dynamics causes high destabilization Table 2 Percentage of species determined by analytical ultracentrifugation after 96 h incubation in 50 mM sodium phosphate (pH 7.4), 0.1 M KCl, and 2.5 M urea Monomer Tetramer Higher-order oligomers 2.5 M urea 15 μM 5 μM 15 μM 5 μM 15 μM 5 μM WT-TTR 0.0% 0.0% 95.7% 100.0% 4.3% 0.0% L55P-TTR 10.4% 16.6% 50.8% 44.4% 38.8% 39.0% R34T-TTR 3.2% 8.1% 81.9% 79.5% 14.9% 12.4% K35N-TTR 9.2% 15.9% 63.7% 57.9% 27.1% 26.2% K35T-TTR 12.2% 20.1% 46.0% 43.7% 41.8% 36.2% R34G-TTR 21.0% 29.3% 59.5% 59.9% 19.5% 10.8% V30M-TTR 15.2% 33.7% 59.5% 54.3% 25.3% 12.0% T119M-TTR 0.0% 0.0% 94.8% 100.0% 2.5% 0.0% (s20,w = 4.1 S), a second peak was detected at 1.41 S (s20,w = Therefore, we speculated that these missense variants trigger a 1.66 S) compatible with a globular monomer (frictional ratio f/ long-range destabilization of both populations of TTR tetra- f0 = 1.22) (Fig. 3 and Table 1). mers. To test this, the dimer–dimer interface was stabilized In the case of L55P-TTR and R34-and K35-TTR variants, with the well-characterized T119M mutation (22, 30, 31) by the sedimentation profile, regardless of the concentration, generating the double mutants R34G/T119M- and K35T/ included a main peak at 3.5 S (s20,w = 4.1 S) corresponding to T119M-TTRs. This should allow determining the impact of the globular tetramer, a second peak at 1.2 S (s20,w = 1.4 S) the missense mutations on a homogeneous population of compatible with a slightly elongated monomer (f/f0 = 1.4), and kinetically stabilized tetramers. The proteins were expressed, a variable amount of higher-order oligomers beyond 5 S (Fig. 3 purified, and subjected to urea equilibrium denaturation ex- and table 2). Thus, all pathogenic variants exhibited a signifi- periments. As is shown in Figure 4A, the first urea-induced cant fraction of partially unfolded monomer, absent in the transition is completely abolished, and the single transition stable WT and T119M forms, and a larger extent of higher- observed starts at a urea concentration similar to that char- order oligomers than these proteins. These two species are acteristic of the second transition observed in the single mu- likely interconnected since, for each particular missense mu- tants (see Fig. 4 for comparison). This indicates that both tation, the fraction of monomer decreases as the concentration tetramers and stabilized tetramers dissociate and unfold be- increases, whereas that of soluble oligomers follows the tween 5.0 to 8.0 M urea in these sequence-stabilized variants. opposite trend. Most interestingly, intermediate species that Figure 4B represents the double mutants plotted in the same populates the urea denaturation reaction between 1.0 and graph together with T119M-TTR, which is kinetically stabi- 3.5 M urea displayed ThT-binding ability (Fig. S6). In the lized and does not reach equilibrium after 96 h incubation 1.0–2.0 M urea range, the R34- and K35-TTR variants (22). R34G or K35T missense mutations destabilize the TTR exhibited the highest fluorescence emission, equal to or higher tetramer harboring the T119M mutation, promoting a single than L55P-TTR and much higher than WT-TTR. dissociation/unfolding transition shifted to lower urea con- centrations (Fig. 4B). Gdm.HCl-induced equilibrium denaturation revealed that Effect of R34G and K35T mutations in the context of a the double mutants exhibited lower stabilities relative to kinetically stabilized T119M-TTR tetramer T119M-TTR, all displaying a single transition, with midpoints The TTR R34 and K35 residues are located on the opposite Cm= 4.3 M for T119M, Cm = 2.6 M for R34G/T119M, and side of the dimer–dimer interface and more distant from the Cm = 3.2 M for K35T/T119M (Figs. 4C and S7). Noteworthy, central cavity than L55. Despite this, R34G and K35T variants the Gdm.HCl denaturation midpoint of T119M-TTR was are as amyloidogenic as L55P at neutral and acidic pH. lower than WT-TTR (Cm 4.3 vs. 5.1, respectively). In addition, 0.3 V30M-TTR WT-TTR R34G-TTR K35T-TTR 1.6 0.6 L55P-TTR T119M-TTR * R34T-TTR K35N-TTR 0.2 0.4 1.2 c(s) 0.4 0.8 0.2 0.1 0.2 0.4 0.0 0.0 0.0 0.0 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 0 2 4 6 8 10 Figure 3. Quaternary structure of urea denaturation intermediates of TTR variants. Sedimentation coefficient distribution of TTR variants determined by a sedimentation velocity experiment. TTR variants at 5.0 μM were incubated for 96 h at 25  C in 50 mM sodium phosphate (pH 7.4), 0.1 M KCl, and 2.5 M urea. *T119M-TTR was subjected to freeze–thawing cycles to induce dissociation to monomer (see Experimental procedures). The TTR variants tested are indicated in each panel. J. Biol. Chem. (2021) 297(3) 101039 5 High TTR BC-loop dynamics causes high destabilization Figure 4. TTR double mutants chemical equilibrium denaturation.1.5 μM of T119M-TTR or double mutants R34G/T119M and K35T/T119M TTR were incubated for 96 h at 25  C in 50 mM sodium phosphate (pH 7.4), 0.1 M KCl with increasing urea or Gdm.HCl concentrations prior measurements of tryptophan fluorescence emission spectra. A, comparison of urea equilibrium denaturation curves of R34G with R34G/T119M TTR (upper panel) and K35 with K35T/T119 TTR (lower panel). B, urea denaturation curves of T119M, K35T/T119M and R34G/T119M. C, Gdm.HCl denaturation curves of T119M, K35T/T119M, and R34G/T119M. double mutants’ transition midpoints were also lower than impact of the mutations on the TTR protomers, the frequency those of single mutants (see Fig. S7 for comparison). of side chain residue–residue contacts was evaluated along the Overall, K35T and R34G shifted the urea and Gdm.HCl simulation (see experimental procedures). In agreement with single transition midpoints in the context of T119M (Fig. 4, B the RMSF profiles, the contact analysis for the BC loop region and C), indicating a robust long-range destabilization of the revealed that R34G establishes less persistent contacts in otherwise kinetically stabilized tetrameric interface. subunit A, whereas K35T shows a marked decrease in the frequency of contacts in subunit B (Fig. S8). Similar findings were observed for the DE loop region, with K35T causing a Molecular dynamic simulations of the TTR variants involving marked reduction in the frequency of contacts in subunit B, residues 34 and 35 while R34G diminishes the contact frequencies of subunit A To gain insights into the pathogenic mechanism, we studied (Fig. S9). The molecular reasons behind these different be- the impact of mutations at positions 34 and 35 by molecular haviors are not evident, but what it comes evident is that, dynamics (MD) simulations. All-atom explicit solvent MD according to the MD simulations, the impact of the mutations simulations were carried out to evaluate the changes in the is restricted to the BC and DE loops surroundings. dimer’s flexibility on the short 10-ns timescale. The flexibility profiles were computed as per-residue Cα Root Mean Square Fluctuation (RMSF). The RMSF profiles of both WT-TTR Structural basis for TTR destabilization induced by variants at monomers (chain A and B) were almost perfectly superim- residues 34 and 35 posable with the only exception of the region 96–104 (Fig. 5), The crystal structures of a significant number of TTR which is a solvent-exposed loop not in direct contact with the pathogenic mutants, including L55P, have been solved (32, 33). monomer–monomer interface. The TTR variants structures However, at neutral pH, they are all virtually identical to that were generated by virtual mutagenesis and their RMSF profiles of WT-TTR. To decipher the structural determinants ac- compared with the reference WT structure. Regions with counting for the high amyloidogenicity of the R34G and K35T higher flexibility were mapped on the 3D structure (Fig. 5). TTR variants, we decided to explore the conformation of the The RMSF profiles of each monomer for WT- and TTR var- region surrounding the mutation sites under destabilizing iants were compared individually to appreciate specific dif- conditions using X-ray diffraction. R34G and K35T TTR ferences better. In good agreement with the thermodynamic mutants readily aggregate at mildly acidic pH, and therefore data, the larger changes in dynamics were observed for the their structural properties cannot be accessed in these proa- R34G and K35T variants. A local effect, resulting in increased myloidogenic conditions. We took profit of the fact that R34G/ flexibility of the BC loop (residues 36–40), was evident for the T119M- and K35T/T119M-TTR double mutants do not four variants. Also, a longer-range effect impacting the flexi- aggregate significantly at acidic pH (Fig. S10) to solve their bility of the DE loop region comprising residues 55–64 was crystal structures at pH 5.5 at 1.5 and 1.4 Å resolutions, observed; this might explain why the impact of some of these respectively. We also solved the crystal structure of the non- mutations resembles that of L55P in terms of aggregation and amyloidogenic variant T119M-TTR at 1.3 Å resolution at the amyloid propensity. Surprisingly, R34G impacted mostly the same pH as a control. first monomer’s dynamic (chain A), whereas K35T essentially As expected, no major structural differences between the affected the second one (chain B). To confirm this differential pathogenic mutants and T119M-TTR were observed, with 6 J. Biol. Chem. (2021) 297(3) 101039 High TTR BC-loop dynamics causes high destabilization A 36-40 B 36-40 R34 K35 0.24 R34 WT chain A K35 55-64 0.20 WT chain B RMSF (nm) 0.16 55-64 0.12 0.08 0.04 0.00 0 20 40 60 80 100 120 Chain B Residue number Chain A C 0.24 R34G R34T K35T K35N 0.20 WT WT WT WT 0.16 RMSF (nm) Chain A 0.12 0.08 0.04 0.00 0.20 RMSF (nm) 0.16 Chain B 0.12 0.08 0.04 0.00 0 20 40 60 80 100 120 20 40 60 80 100 120 20 40 60 80 100 120 20 40 60 80 100 120 Residue number Figure 5. Effect of TTR mutations on protein dynamics. A, location on the 3D structure of the regions with different flexibility patterns. Side chains of residues R34 (red), 35 (green), and loop regions 36–40 (cyan) and 55–64 (yellow) are shown. B, the per-residue Cα RMSF profiles comparison for the WT-TTR chain A and chain B. C, the per-residue Cα RMSF profiles for each pairwise comparison between WT and mutant variants for chain A (upper panel) and chain B (lower panel) are represented. overall Root Mean Square Deviation (RMSD) values for the Cα tetrameric interface of double mutants, which consists of hy- of 0.24 Å for R34G/T119M-TTR and 0.15 Å for K35T/ drophobic interactions between the AB- (residues 18–28) and T119M-TTR (Fig. S11). The crystallographic parameters for the GH loop (residues 105–122), was preserved. The presence both variants are shown in Table S1. As observed for other of Met119 keeps all those interaction residues virtually at the TTR crystals, the complete polypeptide chains could be traced, same positions, which allow the formation of key contacts in except for the terminal residues, 1–9 and 126–127, disordered the presence of the second amyloidogenic mutation. The and not defined in the electron density. We also could not dimerization interfaces were also preserved on all crystal trace the residues 101–102 from chain B in all three structures. structures. The well-reported aggregation-prone CD-, DE-, EF-, and FG- Overall, the crystallographic data suggest that BC-loop loop regions (residues: 49–53, 56–66, 74–90, and 98–104, fluctuations constitute the triggering mechanism for R34G respectively) were not affected by R34G and K35T mutations. and R35T tetramer dissociation and subsequent aggregation. On the other hand, in good agreement with the molecular simulations, the BC-loop regions (residues 36–40) were remarkably affected in the double mutants, in one or both Discussion asymmetrical chains. Indeed, putty-style cartoon representa- Pathogenic mechanisms of distinct disease-associated mu- tions based on mean Cα β-factors show much greater values in tations of TTR have been extensively studied for more than these regions, pointing them as the most dynamic in K35T/ 25 years (22, 34–36). It is well accepted that they affect the T119M and R34G/T119M TTR (Figs. 6 and S12). It is worth tetramer affinity, tetramer dissociation rates (rate-determining mentioning that the higher flexibility of the BC loop in both step for amyloid fibril formation), and/or the thermodynamics chains of R34G matches with its higher instability revealed by of misfolding of monomeric TTR (21). Nevertheless, it is still the thermodynamic analysis. The pattern and the distances of unclear why a given mutation is associated with TTR deposi- main-chain H-bonds that pack the B strand to C and E strands tion in a specific tissue or organ and connected to a particular were well conserved between the double mutants and T119M clinical manifestation. This is best illustrated by the R34G, TTR, except for the one established by Lys35 and Ile68, which R34T, K35N, and K35T TTR familial mutations we study here. become more sparse: T119M (2.81, 2.86), K35T/T119M (2.87, R34G- and K35T-TTR variants cause vitreous amyloidosis (24, 3.07) to R34G/T119M (2.93, 3.26). The contact region at the 25), whereas R34T and K35N mutations result in amyloid J. Biol. Chem. (2021) 297(3) 101039 7 High TTR BC-loop dynamics causes high destabilization Figure 6. Structural features of T119M, R34G/T119M, and K35T/T119M TTRs. A, stick representation showing the BC-loop regions of TTR dimers with electro densities maps countered at 1.5σ. B, Putty cartoon of B-factor variation on the A/B dimers, colored from low to high (blue to red). Side chains of mutated residues are depicted as sticks. Rendered using PyMOL. polyneuropathy and restrictive cardiomyopathy (26, 27). these TTR residues are among the more distant from the Therefore, in both cases, a mutation in the same position re- dimer–dimer interface, where the initial tetramer dissociation sults in two different clinical phenotypes. Besides, the muta- necessarily starts. tion of a positive residue to the same amino acid (R34T, K35T) The thermodynamic stability of disease-associated TTR also gives rise to different disease manifestations, even if these variants has been often assessed by equilibrium urea dena- positions are adjacent in the sequence. We address the turation curves, comparing a single transition midpoint of conformational origin of these differences here. dissociation/unfolding relative to the WT-TTR (Cm = 3.2 at Several in vitro approaches have been used to evaluate the 1.5 μM) (35). In most cases, they are thermodynamically effect of single-point mutations on TTR conformational sta- destabilized, displaying a Cm < 3.2 (29, 33). In the case of WT- bility. One of them is acid-mediated denaturation, as lowering TTR, tetramer dissociation and monomer unfolding have been pH induces tetramer dissociation and triggers the conforma- proposed to be thermodynamically linked. Supporting evi- tional arrangement needed for fibril formation (21). The dence for this mechanism comes from studying an engineered optimal pH to form amyloid fibrils in-vitro for WT-TTR is pH nonamyloidogenic monomeric variant of TTR (M-TTR: 4.4, while disease-causing variants usually dissociate and form F87M:L110M), harboring identical secondary and tertiary fibrils at higher pHs and/or to a much greater extent. The R34- structures to that of tetrameric TTR at pH 7.0 (36). Urea and and K35- variants studied here are more sensitive than WT- Gdm.HCl denaturation curves of M-TTR and tetrameric TTR TTR to pH-induced dissociation. Remarkably, the pH range followed by tryptophan fluorescence showed nearly identical over which dissociation and amyloid formation occurred was transitions, beginning to dissociate-unfold above 2.5 M urea, close to physiological pH, particularly in the case of the eye and considered to be fully unfolded beyond 4.0 to 5.0 M urea. deposited R34G- and K35T-TTR variants, more aggregation- In our case, a detailed inspection of the conformational sta- prone than the R34T- K35N-TTR polyneuropathic/cardiac bility of WT-TTR as measured by urea denaturation revealed variants in all assayed conditions. These findings suggest that that a second transition starts above 8.0 M urea, without R34G- and K35T-TTR could dissociate and form amyloid fi- reaching a stable unfolded baseline even beyond 9.0 M urea brils near physiological conditions. Indeed, these two mutants’ after 96 h incubation. Several factors have been reported to behavior resembles the one of L55P-TTR (28), the most increase the kinetic barrier of dissociation, making tetramer pathogenic mutant described so far. The high susceptibility to dissociation slower. Two Lys residues (Lys 15 and Lys 150 ) aggregation promoted by these mutations is surprising since from neighboring subunits, projecting into the dimer–dimer 8 J. Biol. Chem. (2021) 297(3) 101039 High TTR BC-loop dynamics causes high destabilization interface, repel each other destabilizing the tetramer. The binding of negatively charged ions to these positively charged Lys residues drastically increases tetramer stability (29). It was shown that increasing concentrations of KCl stabilizes the tetramer, rendering biphasic denaturation curves similar to the KD 2- U one we obtained for WT-TTR (29). Our last purification step A K Higher order consisted of a size-exclusion chromatography containing oligomers 0.1 M KCl, and the urea equilibrium denaturation was per- KD KU formed in the presence of 0.1 M KCl. Because we consistently Unfolded obtained biphasic WT-TTR denaturation curves, one possible Partially folded monomer Non-stabilized monomer explanation is that a subpopulation of TTR is loaded with an -U Tetramer D K K IN anion (Cl−) along the purification process. However, based on T the present data, we cannot rule out another mechanism that leads to the stabilization of tetrameric TTR. The single-point mutations studied in the present work uncovered a complex urea denaturation pathway, showing two StabilizedTetramer transitions that were affected differently depending on the variant (Fig. 7). As mentioned above, we assume that two 1.0 Normalized 355/335 different populations of TTR tetramers coexist; nonstabilized tetramers and stabilized tetramers. The first equilibrium transition represents nonstabilized tetramer dissociation, monomer partial unfolding, and monomer unfolding, whereas 0.5 R34G-TTR the second transition would correspond to stabilized tetramers R34T-TTR that dissociate very slow in urea and do not reach equilibrium K35T-TTR within 96 h incubation. Hence the second transition would be K35N-TTR an apparent transition and not an equilibrium transition. 0.0 Physical evidence by AUC experiments revealed that in 2.5 M 0 1 2 3 4 5 6 7 8 9 10 urea, the main species populating the first posttransition re- [Urea] (M) gion are tetramers and extended monomers. Interestingly, Figure 7. Model for urea denaturation mechanism of TTR variants at higher-order oligomers were also detected, raising the possi- residues R34 and K35. Urea denaturation curves followed by tryptophan bility that partially folded monomers can self-assemble into fluorescence of TTR variants display two transitions. The first one takes place oligomers, even in the presence of urea. It is very likely that the between 1.0 and 3.0 M urea and corresponds mainly to nonstabilized tetramer dissociation (KD) to partially folded monomers. This misfolded low concentration of urea at which tetramer dissociation oc- monomeric species can either self-assemble (KA) into higher-order oligo- curs is insufficient to trigger monomer unfolding and allows mers or unfold (KU) at increasing urea concentrations. The second apparent transition occurs between 5.0 and 8.0 M urea and likely represents a pop- the accumulation of misfolded monomers that arrange into ulation of kinetically stabilized tetramers that do not reach equilibrium on oligomers. Additionally, we found that in the urea concen- the experimental timescale. Higher-order oligomer disassembly and tration range at which the first transition occurs, the mutant monomer unfolding could also occur within this urea range (KD2-U). Inter- conversion between nonstabilized and stabilized tetramers (KINT) by sub- proteins bind Th-T, indicating the presence of intermediates units exchange would be very slow due to the high kinetic barrier of with amyloid-like properties, which likely correspond to the dissociation of stabilized tetramer. observed high-order oligomers. In agreement with our results, the TTR acid denaturation Gdm.HCl, only one transition is observed, corresponding to a pathways of WT-, V30M-, and L55P-TTR variants were pre- linked dissociation-unfolding process, as previously reported viously characterized by AUC (37). A monomer with an (38). In this case, chloride ions saturate their binding sites, alternatively folded tertiary structure was identified, which resulting in a homogeneous population of highly stable tetra- self-assembles into a ladder of quaternary structural in- mers populating the pretransition region (29). termediates to form amyloid fibrils (37). In addition, In urea denaturation experiments, R34- and K35- mutations resveratrol-binding experiments support the coexistence of significantly decrease the Cm for both the first and the second two populations of tetramers, with different stabilities. In transitions, with R34G- and K35T-TTR exerting the most concordance with our findings, a similar biphasic urea curve substantial destabilizing effect, as confirmed by the Cm of the was described for V30M TTR, with denaturation intermediates single transitions observed in Gdm.HCl denaturation experi- populating the first posttransition region able to bind resver- ments, and in excellent agreement with the relative proag- atrol (35). Interestingly, after incubation of V30M TTR for gregational effect of the different mutations at mildly acidic 2 weeks in urea, Hurshman Babbes et al. showed that the pH. Strikingly, the R34G-TTR first transition occurs at a lower amplitude of the first transition increases, whereas the urea concentration than the single broad transition of L55P- amplitude of the second transition decreases, compatible with TTR, indicating that this vitreous amyloidosis-associated our proposed model of two populations of tetramers coexist- mutant is highly destabilized. Indeed, AUC experiments indi- ing (35). When chemical denaturation was performed using cate that R34-and K35- variants and L55P-TTR all dissociate J. Biol. Chem. (2021) 297(3) 101039 9 High TTR BC-loop dynamics causes high destabilization significantly to a misfolded aggregation-prone monomer under crystallographic experiments with respect to MD simulations, conditions where WT-TTR still exhibits a stable tetrameric where the system is in solution and not conformationally structure. constrained in crystal form. Notably, the relative destabilizing Despite the R34 and K35 residues at the C-terminus of TTR effect of the different 34 and 35 variants cannot be read β-strand B are distant from the dimer–dimer interface, the data directly from the static TTR structure at physiological pH indicated that their mutation should impact the stability of the without considering dynamics since structure-based compu- intermolecular contacts in this TTR tetramer region. To tational tools wrongly classified K34T-TTR to be the most confirm this long-range destabilizing effect, we exploited the destabilized of the mutants (43). T119M mutation. T119M (22, 30), R104H (39), and A108V Overall, our data indicate that the specific residue at posi- (40) are among the best characterized natural TTR tetramer tion 34 or 35 determines the BC-loop flexibility. An increase in stabilizing mutations, all acting by increasing the kinetic barrier this TTR region dynamism triggers long-range destabilization of dissociation (41). When the T119M mutation was intro- of the dimer–dimer interface, which in turn promotes duced together with either R34G or K35T, the first transition at tetramer dissociation and the accumulation of an aggregation- low urea concentration was completely abolished. We assume prone misfolded monomer that assembles into oligomers. that a homogeneous population of sequence-induced stabilized Other TTR variants were shown to destabilize the native tetramers populate the pretransition region. However, even in tetrameric structure by enhancing loop flexibility. For instance, the highly stabilized T119M context, the introduction of the single-point mutations G53A and S52P reduce the thermo- R34G and K35T mutations strongly destabilizes the tetramer, dynamic stability by perturbing the CD loop (44). shifting the midpoint transitions to lower urea or Gdm.HCl The degree of BC-loop dynamism induced by TTR muta- concentrations, again with a higher impact of the R34G sub- tions at positions 34 and 35 is associated with their destabi- stitution, confirming in both cases a long-range destabilizing lizing impact and the aggregation propensity of the variant. impact that facilitates tetramer dissociation. It is worth According to our data, for the studied TTR variants, vitreous mentioning that the urea range at which anion-stabilized tet- amyloidosis occurs when the combination of the mutated ramers and T119M stabilized tetramers dissociate fairly co- position and the identity of the substituent residue results in a incides (Fig. 4A). By using distinct approaches, several reports high dynamism and destabilization, which endorses the variant agreed that T119M-TTR tetramers and monomers have ther- with increased amyloidogenicity. modynamic stabilities similar to their WT-TTR counterparts Importantly, the introduction of the T119M mutation in the (30, 42). Counterintuitively, by Gdm.HCl denaturation exper- context of R34 and R35-TTR variants prevents aggregation. iments, we found that tetramers harboring Met at position 119 Thus, it is expected that kinetics stabilizers that mimic the were less stable than those with Thr (see Fig. S7). One plausible T119M stabilizing effect, such as Tafamidis (17, 45), Diflunisal explanation for this behavior is that the specific residue at (46), Tolcapone (20, 47), or AG10 (48, 49), would become position 119 may differently affect tetramer stabilization effective drugs to treat the TTR amyloidosis associated with mediated by Cl−, although this effect should be further BC-loop fluctuations. confirmed. In our hands and in agreement with previous re- ports, Met at position 119 protects TTR from acid and urea- Experimental procedures induced denaturation, by increasing the kinetic barrier of tetramer dissociation in these conditions (Figs. 4B and S10). Protein expression and purification The kinetically stabilized double mutants were crystallized WT-TTR was cloned into pET28A vector without any at pH 5.5, allowing to rationalize the structural basis for pH- fusion tag. R34G-; R34T-; K35N-; K35T-; T119M-; R34G/ induced tetramer destabilization. Although the crystallo- T119M-; and K35T/T119M- TTR variants were prepared by graphic structures of the double mutant variants were very standard site-directed mutagenesis protocols using WT-TTR similar to that of T119M-TTR at the same pH, greater flexi- pET28A as template. TTR mutations were confirmed by bility of the BC-loop (higher Cα β-factors) was observed in the Sanger sequencing. L55P and V30M TTR expression vectors presence of the K35T and, especially, the R34G mutations. were kindly provided by Prof. Debora Foguel. All TTR variants This region becomes the most dynamic in the double mutant were recombinantly expressed in BL21(DE3) as soluble pro- structures, in excellent agreement with the individual mutants teins and purified from cell supernatants following previously studied by MDs simulations, suggesting that transient local described procedures (20). Briefly, soluble fractions were unfolding events at the BC strands are likely. In addition to the treated by two consecutive steps of ammonium sulfate pre- crystallographic and MD evidences of a higher BC loop cipitation (50% and 90%, respectively). The 90% precipitate dynamism triggered by R34 and K35 mutations, the simula- was solubilized in 25 mM Tris.HCl (pH 8.0) and extensively tions also suggest that the variants may indirectly impact the dialyzed against the same buffer. Samples were loaded onto a DE loop region flexibility, including L55, which might explain Hi-Trap Q HP column (GE Healthcare, Chicago, IL, USA) why the most aggressive of these mutants resemble L55P-TTR equilibrated in 25 mM Tris HCl (pH 8.0) and eluted with a ten regarding their aggregation and amyloid propensity in vitro. CV linear gradient from 0 to 0.5 M NaCl. The TTR-enriched This dynamic signature of the DE loop was not observable in fractions were precipitated in 90% ammonium sulfate, and the static-X ray structure, which might respond to the drastically precipitate was dissolved in 10 ml of 25 mM Tris.HCl (pH 8.0), different environment that the system experiences in 100 mM KCl. TTR enriched samples were finally purified by 10 J. Biol. Chem. (2021) 297(3) 101039 High TTR BC-loop dynamics causes high destabilization gel filtration chromatography onto a HiLoad 26/600 Superdex The binding of resveratrol to the TTR tetramers was per- 75 prep-grade column (GE Healthcare) equilibrated in 20 mM formed to quantitatively assess their quaternary structure Tris.HCl (pH 8.0), 100 mM KCl. The concentrations of all stabilities as a function of urea concentration, as previously protein solutions were determined spectrophotometrically at reported (22). After measuring the fluorescence emission 280 nm, using a molar extinction coefficient of λ280 nm = spectra, the samples at 1 μM were mixed with 18 μM resver- 77.600 M−1 cm−1 for the TTR tetramer. atrol and fluorescence emission spectra were obtained with excitation and emission wavelengths of 320 and 394 nm, Aggregation and fibril formation assays respectively. To quantify the concentration of TTR tetramer as a function of urea concentration, resveratrol-binding curves Two approaches were used to evaluate amyloid fibril for- were performed for each TTR variant (0–2.0 μM TTR mation: (1) Turbidity assay, which measures the presence of tetramer) by using 18 μM of resveratrol (see Fig. S4). The large insoluble aggregates. (2) Thioflavin T (ThT)-binding fluorescence intensity at 394 nm was plotted versus the con- assay that measures the presence of either amyloid fibrils or centration of TTR, exhibiting a linear fit as was previously soluble amyloid-like small oligomers. TTR variants (3.5 μM) reported (22). were incubated under quiescent conditions in a broad range buffer (100 mM Tris.HCl, 50 mM MES, 50 mM sodium ace- tate, 0.1 M KCl) ranging from pHs 4.0 to 7.0 for 72 h at 37  C. Sedimentation velocity assays (SV) The ionic strength of the buffer is kept constant throughout TTR samples at 5 and 15 μM were incubated during 96 h in the pH range tested (48). After the incubation the samples 50 mM sodium phosphate pH 7.0, 100 mM KCl, and 2.5 M were vortexed and optical density was measured at 330 nm on urea. After incubation, samples were centrifuged at 12.100g for a UVI-Vis Carry spectrophotometer (Shimadzu). Fibril for- 15 min to remove insoluble aggregates. The remaining soluble mation was also assessed by ThT binding incubating 1 μM of protein was quantified spectrophotometrically at 280 nm and the sample (vortexed to achieve homogeneity) with 25 μM the percentage of proteins that remain in solution after ThT in 100 mM Tris⋅HCl (pH 8.0), 0.1 M KCl. The mixed centrifugation ranged from 90 to 100%. TTR samples were sample was then excited at 440 nm and emission at 482 nm then loaded (400 μl) into analytical ultracentrifugation cells. was recorded on a Jasco 8200 spectrofluorometer. The experiments were carried out at 25  C and 48,000 rpm in a XL-A analytical ultracentrifuge (Beckman-Coulter Inc) Transmission electron microscopy (TEM) equipped with an UV-VIS absorbance detection system, using an An-50Ti rotor, and 12 mm Epon-charcoal standard double- Aggregated suspensions were absorbed onto 200-mesh sector centerpieces. Sedimentation profiles were recorded at carbon-coated copper grids for 5 min and then blotted to 280 nm. Differential sedimentation coefficient distributions remove excess material. Negative staining was performed by were calculated by least-squares boundary modeling of sedi- adding 5 μl of 2% (w/v) uranyl acetate. Samples were dried on mentation velocity data using the continuous distribution c(s) air for 3 min. The grids were imaged with a Jeol 1200 electron Lamm equation model as implemented by SEDFIT (50). These microscope (Jeol Ltd) operating at a60 kV acceleration voltage. s values were corrected to standard conditions (water, 20  C, and infinite dilution) (51) using the program SEDNTERP (52) Chemical equilibrium denaturation of TTR variants to get the corresponding standard s values (s20,w). Urea and guanidinium chloride (Gdm.HCl) denaturation curves were performed incubating 1.5 μM of TTR tetramer in Crystallography and structure determination 50 mM sodium phosphate (pH 7.4), 0.1 M KCl with varying concentrations of chaotropes. Samples were incubated at room Crystals of R34G/T119M, K35T/T119M and T119M TTR temperature for 96 h before fluorescence measurements. variants were obtained at 18  C by hanging-drop vapor diffu- Tryptophan fluorescence was used to monitor TTR tertiary sion methods after purification and concentration. The structural changes as a function of denaturant concentration. reservoir solution contained between 7% glycerol, 1.3 M so- The samples (25  C) were excited at 295 nm, and three ac- dium citrate, pH 5.5. Single crystals appeared after 3 days from cumulations of the fluorescence emission spectra were taken equal volumes of protein solution (0.1 mM in 50 mM Tris⋅HCl from 310 to 410 nm on a Jasco 8200 spectrofluorometer. pH 8.0, 100 mM KCl, and 1 mM EDTA) and reservoir solu- Fluorescence emission data were analyzed by first subtracting tion. Crystals were cryo-protected in reservoir buffer con- the buffer background at the appropriate denaturant concen- taining 12% glycerol and directly flash-frozen in liquid tration, and the center of spectral mass of the emission spec- nitrogen prior to diffraction analysis. Diffraction data were trum was quantified as follows: recorded from cryo-cooled crystals (100 K) at the BL13- XALOC beamline from ALBA synchrotron (53). Data were X .X integrated and merged using XDS (54) and scaled, reduced, CM ðcm−1 Þ ¼ ðυi:FiÞ F i (1) and further analyzed using CCP4 (55) (Table S1). The struc- tures of TTR variants were determined from the X-ray data by where F i is the fluorescence emission at wave number υi, and molecular replacement using a former TTR structure (PDB the summation is carried out over the range of measured 1F41) as a model using the program Phaser (56). Model values of F. refinement and rebuilding were performed with (57, 58). J. Biol. Chem. (2021) 297(3) 101039 11 High TTR BC-loop dynamics causes high destabilization Refinement and data statistics are provided in Table S1, and 6FXU. The MD data are available in the OSF (https://osf. structural representations were rendered using Pymol Soft- io/az7vu/) and GitHub repositories (https://github.com/ ware Package. The structures have been deposited with PDB ELELAB/TTR_gatekeeper) associated with this publication. codes 6FWD, 6FZL, and 6FXU. Additional data that support the findings of this study are contained within the manuscript and supporting information. Molecular dynamics (MD) simulations Any additional information or data are available upon request. Contact information: e-mail: salvador.ventura@uab.es. The X-ray structure of wild-type TTR (PDB 1F41) has been used as starting structure for 150-ns explicit solvent MD simulations using the CHARMM22* force field (59) and Supporting information—This article contains supporting GROMACS software. We simulated the dimeric assembly of information. the protein. The starting structures for TTR variants (R34G, R34T, K35N, and K35T) have been achieved upon in silico Acknowledgments—The authors thank to Damián Alvarez-Paggi mutagenesis with Pymol. We mutated in silico the wild-type and Alana Byrne for reading the manuscript and helpful discussions. X-ray experiments were performed at BL-13 Xaloc structure to be able to assess the perturbation induced upon beamline at ALBA Synchrotron with the collaboration of ALBA the mutations on the native structure dynamics. Periodic staff. boundary conditions were employed for the simulations and the initial structures were embedded in a dodecahedral box of Author contributions—S. A. E. and S. V. conceptualization; N. V., F. TIP3P water molecules so that all the protein atoms were at a P., V. S., E. P., G. R., D. R., and S. V. data curation; S. A. E., N. V., F. distance equal or greater than 15 Å from the box edges. To P., R. S., J. R. L.-O., C. A., V. S., E. P., G. R., D. R., and S. V. formal neutralize the overall charge of the system, a number of water analysis; C. A. and S. V. funding acquisition; S. A. E., N. V., F. P., R. molecules equal to the protein net charge were replaced by S., J. R. L.-O., C. A., G. R., D. R., and S. V. investigation; S. A. E., F. P., counterions and a concentration of 150 mM NaCl has been R. S., J. R. L.-O., C. A., V. S., and S. V. methodology; S. V. project administration; S. V. resources; S. V. supervision; S. A. E. and S. V. used. The preparation steps have been carried out as previ- writing–original draft; S. A. E. and S. V. writing–review and editing. ously described (60) and productive MD simulations were performed in the NVT ensemble at 300 K and 1 bar using an Funding and additional information—This work was funded by the external Berendsen bath with thermal and pressure coupling of Spanish Ministry of Science and Innovation PID2019-105017RB- 0.1 and 1 ps, respectively. The LINCS algorithm was used to 100 to S. V., PID2019-104544GB-I00 to C. A., and by ICREA, constrain heavy-atom bonds, allowing for a 2-fs time step. ICREA-Academia 2020 to S. V. and Danmarks Grundforsknings- Long-range electrostatic interactions were calculated using the fond (DNRF125) and Carlsbergfondet Distinguished Fellowship Particle-Mesh-Ewald (PME) summation scheme. van der (CF18-0314) to E. P. Waals and short-range Coulomb interactions were truncated Conflict of interest—S. A. E. is a member of the scientific and at 10 Å. The nonbonded pair list was updated every ten steps technological research career of CONICET. The authors declare and conformations were stored every 4 ps. that they have no conflicts of interest with the contents of this The main-chain RMSD was computed using the initial article. structure for MD simulations as a reference to assess the stability of the productive simulations. To evaluate the effects Abbreviations—The abbreviations used are: ERAD, endoplasmic induced by the mutations on the monomer–monomer inter- reticulum-associated degradation; FAC, familial amyloid cardio- face, we calculated an average per-residue C-α RMSF profile myopathy; FAP, familial amyloid polyneuropathy; MD, molecular dynamics; RMSD, root mean square deviation; RMSF, root mean using time windows of 10 ns for the averaging. We then square fluctuation; TEM, transmission electron microscopy; TTR, compared pairwise the RMSF profiles of the WT and mutant transthyretin; WT, wild-type. TTR variants and mapped the regions with enhanced flexibility upon mutation on the 3D structure. The contact analysis was carried out using a custom Python script, and two residues References were considered in contact if they had at least a pair of side 1. Aminoff Michael, J., and Daroff Robert, B. (2014) Encyclopedia of the chain atoms lying within 4.5 Åfrom each other and a sequence Neurological Sciences, 2nd Ed., Academic Press, Elsevier, London, UK 2. Eisenberg, D., and Jucker, M. (2012) The amyloid state of proteins in distance greater than 1 (i.e., they were separated by at least one human diseases. Cell 148, 1188–1203 residue along the sequence). The persistence of each contact 3. Sipe, J. D., Benson, M. D., Buxbaum, J. N., Ikeda, S.-I., Merlini, G., Saraiva, was calculated by dividing the number of frames of the M. J. M., and Westermark, P. (2016) Amyloid fibril proteins and amyloid- simulation where the contact was present by the total number osis: Chemical identification and clinical classification International Society of frames. The MD data are available in the OSF (https://osf. of amyloidosis 2016 Nomenclature Guidelines. Amyloid 23, 209–213 4. Ankarcrona, M., Winblad, B., Monteiro, C., Fearns, C., Powers, E. T., io/az7vu/) and GitHub repositories (https://github.com/ Johansson, J., Westermark, G. T., Presto, J., Ericzon, B.-G., and Kelly, J. ELELAB/TTR_gatekeeper) associated with this publication. W. (2016) Current and future treatment of amyloid diseases. J. Intern. Med. 280, 177–202 5. Hagen, G. A., and Elliott, W. J. (1973) Transport of thyroid hormones in Data availability serum and cerebrospinal fluid. J. Clin. Endocrinol. Metab. 37, 415–422 The crystal structures have been deposited in Protein Data 6. Christmanson, L., Betsholtz, C., Gustavsson, A., Johansson, B., Sletten, K., Bank (https://www.rcsb.org/) with PDB codes 6FWD, 6FZL, and Westermark, P. (1991) The transthyretin cDNA sequence is normal 12 J. Biol. Chem. (2021) 297(3) 101039 High TTR BC-loop dynamics causes high destabilization in transthyretin-derived senile systemic amyloidosis. FEBS Lett. 281, Implications for transthyretin-related amyloidoses. J. Biol. Chem. 289, 177–180 28324–28337 7. Buxbaum, J. N., and Reixach, N. (2009) Transthyretin: The servant of 24. Long, D., Zeng, J., Wu, L. Q., Tang, L. S., Wang, H. L., and Wang, H. many masters. Cell. Mol. Life Sci. 66, 3095–3101 (2012) Vitreous amyloidosis in two large mainland Chinese kindreds 8. Finsterer, J., Iglseder, S., Wanschitz, J., Topakian, R., Löscher, W. N., and resulting from transthyretin variant Lys35Thr and Leu55Arg. Ophthalmic Grisold, W. (2019) Hereditary transthyretin-related amyloidosis. Acta Genet. 33, 28–33 Neurol. Scand. 139, 92–105 25. Levy, J., Hawkins, P. N., Rowczenio, D., Godfrey, T., Stawell, R., and 9. Coelho, T. (1996) Familial amyloid polyneuropathy: New developments in Zamir, E. (2012) Familial amyloid polyneuropathy associated with the genetics and treatment. Curr. Opin. Neurol. 9, 355–359 novel transthyretin variant Arg34Gly. Amyloid 19, 201–203 10. Jiang, X., Buxbaum, J. N., and Kelly, J. W. (2001) The V122I cardiomy- 26. Patrosso, M. C., Salvi, F., De Grandis, D., Vezzoni, P., Jacobson, D. R., and opathy variant of transthyretin increases the velocity of rate-limiting Ferlini, A. (1998) Novel transthyretin missense mutation (Thr34) in an tetramer dissociation, resulting in accelerated amyloidosis. Proc. Natl. Italian family with hereditary amyloidosis. Am. J. Med. Genet. 77, 135– Acad. Sci. U. S. A. 98, 14943–14948 138 11. Almeida, M. R., Hesse, A., Steinmetz, A., Maisch, B., Altland, K., Linke, R. 27. Reilly, M. M., Adams, D., Booth, D. R., Davis, M. B., Said, G., Laubriat- P., Gawinowicz, M. A., and Saraiva, M. J. (1991) Transthyretin Leu 68 in a Bianchin, M., Pepys, M. B., Thomas, P. K., and Harding, A. E. (1995) form of cardiac amyloidosis. Basic Res. Cardiol. 86, 567–571 Transthyretin gene analysis in European patients with suspected familial 12. Azevedo, E. P. C., Pereira, H. M., Garratt, R. C., Kelly, J. W., Foguel, D., amyloid polyneuropathy. Brain 118, 849–856 and Palhano, F. L. (2011) Dissecting the structure, thermodynamic sta- 28. Lashuel, H. A., Wurth, C., Woo, L., and Kelly, J. W. (1999) The most bility, and aggregation properties of the A25T transthyretin (A25T-TTR) pathogenic transthyretin variant, L55P, forms amyloid fibrils under acidic variant involved in leptomeningeal amyloidosis: Identifying protein conditions and protofilaments under physiological conditions. Biochem- partners that co-aggregate during A25T-TTR fibrillogenesis in cerebro- istry 38, 13560–13573 spinal fluid. Biochemistry 50, 11070–11083 29. Hammarström, P., Jiang, X., Deechongkit, S., and Kelly, J. W. (2001) 13. Hammarström, P., Sekijima, Y., White, J. T., Wiseman, R. L., Lim, A., Anion shielding of electrostatic repulsions in transthyretin modulates Costello, C. E., Altland, K., Garzuly, F., Budka, H., and Kelly, J. W. (2003) stability and amyloidosis: Insight into the chaotrope unfolding dichotomy. D18G transthyretin is monomeric, aggregation prone, and not detectable Biochemistry 40, 11453–11459 in plasma and cerebrospinal fluid: A prescription for central nervous 30. Palhano, F. L., Leme, L. P., Busnardo, R. G., and Foguel, D. (2009) system amyloidosis? Biochemistry 42, 6656–6663 Trapping the monomer of a non-amyloidogenic variant of transthyretin: 14. Sekijima, Y., Wiseman, R. L., Matteson, J., Hammarström, P., Miller, S. R., Exploring its possible use as a therapeutic strategy against transthyretin Sawkar, A. R., Balch, W. E., and Kelly, J. W. (2005) The biological and amyloidogenic diseases. J. Biol. Chem. 284, 1443–1453 chemical basis for tissue-selective amyloid disease. Cell 121, 73–85 31. Almeida, M. R., Alves, I. L., Terazaki, H., Ando, Y., and Saraiva, M. J. 15. Coelho, T., Chorão, R., Sousa, A., Alves, I., Torres, M. F., and Saraiva, M. (2000) Comparative studies of two transthyretin variants with protective J. M. (1996) Compound heterozygotes of transthyretin Met30 and effects on familial amyloidotic polyneuropathy: TTR R104H and TTR transthyretin Met119 are protected from the devastating effects of fa- T119M. Biochem. Biophys. Res. Commun. 270, 1024–1028 milial amyloid polyneuropathy. Neuromuscul. Disord. 6, S20 32. Palaninathan, S. K. (2012) Nearly 200 X-ray crystal structures of trans- 16. Johnson, S. M., Connelly, S., Fearns, C., Powers, E. T., and Kelly, J. W. thyretin: What do they tell us about this protein and the design of drugs (2012) The transthyretin amyloidoses: From delineating the molecular for TTR amyloidoses? Curr. Med. Chem. 19, 2324–2342 mechanism of aggregation linked to pathology to a regulatory-agency- 33. Cendron, L., Trovato, A., Seno, F., Folli, C., Alfieri, B., Zanotti, G., and approved drug. J. Mol. Biol. 421, 185–203 Berni, R. (2009) Amyloidogenic potential of transthyretin variants: In- 17. Bulawa, C. E., Connelly, S., Devit, M., Wang, L., Weigel, C., Fleming, J. A., sights from structural and computational analyses. J. Biol. Chem. 284, Packman, J., Powers, E. T., Wiseman, R. L., Foss, T. R., Wilson, I. A., 25832–25841 Kelly, J. W., and Labaudinière, R. (2012) Tafamidis, a potent and selective 34. Colon, W., and Kelly, J. W. (1992) Partial denaturation of transthyretin is transthyretin kinetic stabilizer that inhibits the amyloid cascade. Proc. sufficient for amyloid fibril formation in vitro. Biochemistry 31, 8654– Natl. Acad. Sci. U. S. A. 109, 9629–9634 8660 18. Coelho, T., Maia, L. F., da Silva, A. M., Cruz, M. W., Planté-Bordeneuve, 35. Hurshman Babbes, A. R., Powers, E. T., and Kelly, J. W. (2008) Quanti- V., Suhr, O. B., Conceiçao, I., Schmidt, H. H.-J., Trigo, P., Kelly, J. W., fication of the thermodynamically linked quaternary and tertiary struc- Labaudinière, R., Chan, J., Packman, J., and Grogan, D. R. (2013) Long- tural stabilities of transthyretin and its disease-associated variants: The term effects of tafamidis for the treatment of transthyretin familial am- relationship between stability and amyloidosis. Biochemistry 47, 6969– yloid polyneuropathy. J. Neurol. 260, 2802–2814 6984 19. Maurer, M. S., Schwartz, J. H., Gundapaneni, B., Elliott, P. M., Merlini, G., 36. Jiang, X., Smith, C. S., Petrassi, H. M., Hammarström, P., White, J. T., Waddington-Cruz, M., Kristen, A. V., Grogan, M., Witteles, R., Damy, T., Sacchettini, J. C., and Kelly, J. W. (2001) An engineered transthyretin Drachman, B. M., Shah, S. J., Hanna, M., Judge, D. P., Barsdorf, A. I., et al. monomer that is nonamyloidogenic, unless it is partially denatured. (2018) Tafamidis treatment for patients with transthyretin amyloid car- Biochemistry 40, 11442–11452 diomyopathy. N. Engl. J. Med. 379, 1007–1016 37. Lashuel, H. A., Lai, Z., and Kelly, J. W. (1998) Characterization of the 20. Pinheiro, F., Varejão, N., Esperante, S., Santos, J., Velázquez-Campoy, A., transthyretin acid denaturation pathways by analytical ultracentrifuga- Reverter, D., Pallarès, I., and Ventura, S. (2020) Tolcapone, a potent ag- tion: Implications for wild-type, V30M, and L55P amyloid fibril forma- gregation inhibitor for the treatment of familial leptomeningeal tion. Biochemistry 37, 17851–17864 amyloidosis. FEBS J. https://doi.org/10.1111/febs.15339 38. Lai, Z., McCulloch, J., Lashuel, H. A., and Kelly, J. W. (1997) Guanidine 21. Lai, Z., Colón, W., and Kelly, J. W. (1996) The acid-mediated denatur- hydrochloride-induced denaturation and refolding of transthyretin ex- ation pathway of transthyretin yields a conformational intermediate that hibits a marked hysteresis: Equilibria with high kinetic barriers. can self-assemble into amyloid. Biochemistry 35, 6470–6482 Biochemistry 36, 10230–10239 22. Hammarström, P., Jiang, X., Hurshman, A. R., Powers, E. T., and Kelly, J. 39. Sekijima, Y., Dendle, M. T., Wiseman, R. L., White, J. T., D’Haeze, W., and W. (2002) Sequence-dependent denaturation energetics: A major deter- Kelly, J. W. (2006) R104H may suppress transthyretin amyloidogenesis by minant in amyloid disease diversity. Proc. Natl. Acad. Sci. U. S. A. 99 thermodynamic stabilization, but not by the kinetic mechanism charac- Suppl 4, 16427–16432 terizing T119 interallelic trans-suppression. Amyloid 13, 57–66 23. Sant’Anna, R., Braga, C., Varejão, N., Pimenta, K. M., Graña-Montes, R., 40. Sant’Anna, R., Almeida, M. R., Varejao, N., Gallego, P., Esperante, S., Alves, A., Cortines, J., Cordeiro, Y., Ventura, S., and Foguel, D. (2014) The Ferreira, P., Pereira-Henriques, A., Palhano, F. L., de Carvalho, M., importance of a gatekeeper residue on the aggregation of transthyretin: Foguel, D., Reverter, D., Saraiva, M. J., and Ventura, S. (2017) Cavity J. Biol. Chem. (2021) 297(3) 101039 13 High TTR BC-loop dynamics causes high destabilization filling mutations at the thyroxine-binding site dramatically increase 50. Schuck, P. (2000) Size-distribution analysis of macromolecules by sedi- transthyretin stability and prevent its aggregation. Sci. Rep. 7, 44709 mentation velocity ultracentrifugation and lamm equation modeling. 41. Foss, T. R., Kelker, M. S., Wiseman, R. L., Wilson, I. A., and Kelly, J. W. Biophys. J. 78, 1606–1619 (2005) Kinetic stabilization of the native state by protein engineering: 51. Van Holde, K. E., Johnson, C., and Ho, P. S. (2005) Principles of Physical Implications for inhibition of transthyretin amyloidogenesis. J. Mol. Biol. Biochemistry, 2 Ed., Prentice Hall, Upper Saddle River, NJ 347, 841–854 52. Harding, S. E., Rowe, A. J., and Horton, J. C. (1992) Analytical Ultra- 42. Hammarström, P., Schneider, F., and Kelly, J. W. (2001) Trans-suppres- centrifugation in Biochemistry and Polymer Science, Royal Society of sion of misfolding in an amyloid disease. Science 293, 2459–2462 Chemistry, Cambridge, UK 43. Srinivasan, E., Natarajan, N., and Rajasekaran, R. (2020) TTRMDB: A 53. Juanhuix, J., Gil-Ortiz, F., Cuní, G., Colldelram, C., Nicolás, J., Lidón, J., database for structural and functional analysis on the impact of SNPs over Boter, E., Ruget, C., Ferrer, S., and Benach, J. (2014) Developments in transthyretin (TTR) using bioinformatic tools. Comput. Biol. Chem. 87, optics and performance at BL13-XALOC, the macromolecular crystal- 107290 lography beamline at the ALBA synchrotron. J. Synchrotron Radiat. 21, 44. Dasari, A. K. R., Arreola, J., Michael, B., Griffin, R. G., Kelly, J. W., and 679–689 Lim, K. H. (2020) Disruption of the CD loop by enzymatic cleavage 54. Kabsch, W. (2010) XDS. Acta Crystallogr. D Biol. Crystallogr. 66, 125– promotes the formation of toxic transthyretin oligomers through a 132 common transthyretin misfolding pathway. Biochemistry 59, 2319–2327 55. Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., 45. Coelho, T., Merlini, G., Bulawa, C. E., Fleming, J. A., Judge, D. P., Kelly, J. Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., W., Maurer, M. S., Planté-Bordeneuve, V., Labaudinière, R., Mundayat, McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., R., Riley, S., Lombardo, I., and Huertas, P. (2016) Mechanism of action Powell, H. R., et al. (2011) Overview of the CCP4 suite and current de- and clinical application of tafamidis in hereditary transthyretin amyloid- velopments. Acta Crystallogr. D Biol. Crystallogr. 67, 235–242 osis. Neurol. Ther. 5, 1–25 56. McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., 46. Sekijima, Y., Dendle, M. A., and Kelly, J. W. (2006) Orally administered Storoni, L. C., and Read, R. J. (2007) Phaser crystallographic software. J. diflunisal stabilizes transthyretin against dissociation required for amy- Appl. Crystallogr. 40, 658–674 loidogenesis. Amyloid 13, 236–249 57. Adams, P. D., Afonine, P. V., Bunkóczi, G., Chen, V. B., Davis, I. W., 47. Sant’Anna, R., Gallego, P., Robinson, L. Z., Pereira-Henriques, A., Fer- Echols, N., Headd, J. J., Hung, L.-W., Kapral, G. J., Grosse-Kunstleve, R. reira, N., Pinheiro, F., Esperante, S., Pallares, I., Huertas, O., Almeida, M. W., McCoy, A. J., Moriarty, N. W., Oeffner, R., Read, R. J., Richardson, D. R., Reixach, N., Insa, R., Velazquez-Campoy, A., Reverter, D., Reig, N., C., et al. (2010) PHENIX: A comprehensive Python-based system for et al. (2016) Repositioning tolcapone as a potent inhibitor of transthyretin macromolecular structure solution. Acta Crystallogr. D Biol. Crystallogr. amyloidogenesis and associated cellular toxicity. Nat. Commun. 7, 10787 66, 213–221 48. Miller, M., Pal, A., Albusairi, W., Joo, H., Pappas, B., Haque Tuhin, M. T., 58. Emsley, P., Lohkamp, B., Scott, W. G., and Cowtan, K. (2010) Features Liang, D., Jampala, R., Liu, F., Khan, J., Faaij, M., Park, M., Chan, W., and development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486– Graef, I., Zamboni, R., et al. (2018) Enthalpy-driven stabilization of 501 transthyretin by AG10 mimics a naturally occurring genetic variant that 59. Piana, S., Lindorff-Larsen, K., and Shaw, D. E. (2011) How robust are protects from transthyretin amyloidosis. J. Med. Chem. 61, 7862–7876 protein folding simulations with respect to force field parameterization? 49. Penchala, S. C., Connelly, S., Wang, Y., Park, M. S., Zhao, L., Baranczak, Biophys. J. 100, L47–L49 A., Rappley, I., Vogel, H., Liedtke, M., Witteles, R. M., Powers, E. T., 60. Nygaard, M., Terkelsen, T., Vidas Olsen, A., Sora, V., Salamanca Viloria, Reixach, N., Chan, W. K., Wilson, I. A., Kelly, J. W., et al. (2013) AG10 J., Rizza, F., Bergstrand-Poulsen, S., Di Marco, M., Vistesen, M., Tiberti, inhibits amyloidogenesis and cellular toxicity of the familial amyloid M., Lambrughi, M., Jäättelä, M., Kallunki, T., and Papaleo, E. (2016) The cardiomyopathy-associated V122I transthyretin. Proc. Natl. Acad. Sci. U. mutational landscape of the oncogenic MZF1 SCAN domain in cancer. S. A. 110, 9992–9997 Front. Mol. Biosci. 3, 78 14 J. Biol. Chem. (2021) 297(3) 101039