Transthyretin Binding Mode Dichotomy of Fluorescent trans-Stilbene Ligands

The orientations of ligands bound to the transthyretin (TTR) thyroxine (T4) binding site are difficult to predict. Conflicting binding modes of resveratrol have been reported. We previously reported two resveratrol based trans-stilbene fluorescent ligands, (E)-4-(2-(naphthalen-1-yl)vinyl)benzene-1,2-diol (SB-11) and (E)-4-(2-(naphthalen-2-yl)vinyl)benzene-1,2-diol (SB-14), that bind native and misfolded protofibrillar TTR. The binding orientations of these two analogous ligands to native tetrameric TTR were predicted to be opposite. Herein we report the crystal structures of these TTR:ligand complexes. Opposite binding modes were verified but were different than predicted. The reverse binding mode (SB-14) placing the naphthalene moiety toward the opening of the binding pocket renders the fluorescent ligand pH sensitive due to changes in Lys15 amine protonation. Conversely, the forward binding mode (SB-11) placing the naphthalene inward mediates a stabilizing conformational change, allowing intersubunit H-bonding between Ser117 of different monomers across the dimer interface. Our structures of TTR complexes answer important questions in ligand design and interpretation of trans-stilbene binding modes to the TTR T4 binding site.

■ INTRODUCTION TTR in Transport, Chaperone and Amyloid Processes. Transthyretin (TTR) is a plasma and cerebrospinal fluid (CSF) protein synthesized by the liver, choroid plexus, eye and pancreas. TTR is the main transporter of T4 hormone in CSF and the primary carrier of retinol (Vitamin A) binding protein in blood plasma. TTR appears to function as a molecular chaperone in preventing amyloid formation of Aβ in the brain 1−4 and islet amyloid polypeptide (IAPP) in the pancreas. 5 TTR is inherently amyloidogenic, and the wildtype protein accounts for a significant number of cardiac amyloidosis cases in elderly people 6 as well as diverse mutation-dependent amyloid disease phenotypes in familial diseases such as familial amyloid polyneuropathy (FAP). 7 Collectively this group of diseases are known as transthyretin amyloidosis with deposited ATTR. TTR research to understand protein misfolding diseases is of considerable interest due to the impact of TTR amyloidosis on the afflicted ATTR patient and society. The small molecule ligand Tafamidis is the active component of Vyndaqel and Vyndamax, which are approved drugs for treating FAP and wild-type ATTR cardiomyopathy. Tafamidis works as a kinetic stabilizer by binding to the T4 binding site, preventing tetramer dissociation, and thereby subsequent misfolding and fibril formation is suppressed. 8 Tafamidis binding also imposes longrange allosteric conformational changes 9 which may in addition to preventing tetramer dissociation inhibit aberrant proteolysis, which is a putative fibrillation initiation mechanism. 10 Resveratrol Is a Health Promoting Natural Compound. trans-Resveratrol is a trans-stilbene polyphenolic antioxidant found predominantly in plants such as grapes. Resveratrol can reach significant concentrations in red wine. There are reports of resveratrol being anti-inflammatory, anticancer, neuroprotective, and anti-aging, similar to effects of calorie restriction. 11 The latter activity is largely attributed to resveratrol isoform selective activation and inhibition of the deacetylase activity of sirtuins. 12 The class of trans-stilbenes have been of considerable interest for TTR amyloidosis to find natural products as alternatives to kinetic stabilizers of TTR tetramers such as tafamidis. 8 trans-Stilbene Ligands for Amyloid Proteins. TTR is a homo tetramer with a total mass of 55 kDa composed of 127 amino acids in each subunit. The structure of TTR was originally solved by Blake in 1978, 13 and numerous structures of TTR in complex with ligands have been published since. With all this knowledge there has been controversy regarding certain binding modes of ligands to TTR. Studies of TTRtargeting molecules are of significant interest due to TTR association with age-dependent amyloidosis. TTR is a rather promiscuous binder of aromatic compounds resembling T4 in its dual T4 binding pockets. Resveratrol has been reported to bind to the T4 binding pocket (T4BP) with rather high affinity. Interestingly, it appears that the binding modes of several molecules to TTR are also promiscuous regarding preferred binding orientation. Crystal structures of the TTR:resveratrol complex report contrasting binding modes. 14,15 That resveratrol can be used as a TTR tetramer-sensitive fluorescent ligand when binding to the T4BP was shown many years ago. 16 Furthermore, the trans-stilbene chemical motif is of particular interest for amyloid targeting because it is present in several amyloid fibril-specific ligands including fluorescent ligands X-34, Methoxy-XO4, and PET ligands florbetaben (18F) and florbetapir (18F). 17 Our rationale for the current study is of general interest for understanding the selectivity and binding modes of TTR-binding trans-stilbene ligands and in particular for resveratrol and resveratrol analogues of amyloid fibril probes.
We previously hypothesized on the binding mode of two structurally homologous, amyloid-sensitive ligands SB-11 and SB-14. 18 Based on fluorescence spectroscopy, we speculated that they display opposite binding orientations in the T4BP of TTR. 18 We determined the structures of these compounds bound to TTR by X-ray crystallography to verify these findings. Here we present the crystal structures of wild-type TTR in complex with three trans-stilbene compounds TTR:SB-11, TTR:SB-14, and TTR:resveratrol. The data confirm our previous findings that these ligands bind in opposite directions despite being analogues and explain their distinct activities in terms of fluorescence and as stabilizing inhibitors to prevent fibril formation. Highresolution X-ray crystal structures of TTR in complex with SB-11, SB-14, and resveratrol were obtained by cocrystallization with ligands coincubated with the protein for at least 2 h at room temperature. Comparisons were done with apo-TTR crystallized under identical conditions. All four crystal structures belong to the P2 1 22 1 space group; the protein structure remains unchanged upon ligand binding, as evidenced by the low deviation in Cα positions between the aligned monomers (<0.35 Å rmsd). The dimer AB is found in the asymmetric unit and the second dimer (A′B′) to form the tetramer can be obtained by rotation along the crystallographic 2-fold c-axis ( Figure 1A). The inner β-sheets of the dimer− dimer (AB−A′B′) interface form two T4BP cavities referred to as sites AA′ and BB′, respectively. The symmetric binding sites within each AA′ and BB′ dimer are composed of three socalled halogen binding pockets (HBPs). HBP1 is in the outer cavity close to the protein surface and HPB3 is in the inner cavity closest to the center of the tetramer, with HPB2 between these two cavities. The T4BP is predominantly hydrophobic, though some of the constituent amino acids have polar side chains that allow hydrophilic interactions.
Since the crystallographic 2-fold axis crosses the T4BPs, symmetry-related ligands are found superposed at this special position as already reported. 15,19 The electron densities for all TTR:ligand complexes presented enabled unambiguous placement of ligands ( Figure  Structural Consequences of Different Binding Modes. Binding of resveratrol (in both orientations) and SB-14 with their hydroxyl groups pointing into HBP3 positions these within distance for H-bonding with Ser117 at the base of the T4BP (Figures 1−3). In comparison, SB-11 binds in the opposing forward mode, burying its hydrophobic naphthalene moiety in HBP3 (Figures 1, 3). Notably, in apo-TTR, the side chain of Ser117 adopts several alternative conformations and H-bonds with water molecules occupying the binding cavity ( Figure 4A). Upon ligand binding, the conformation of Ser117 becomes more constrained. A common conformation for Ser117 is imposed when its hydroxyl group H-bonds with a polar ligand as observed for resveratrol (in either pose) and the asymmetric SB-14 ( Figure 4B, D). In contrast, SB-11 induces intersubunit H-bonding interactions between two neighboring Ser117 (A:B and A′:B′) ( Figure 4C). This latter pose is suggestive of a tetramer-stabilizing conformation with H-bonds across the dimer interface. Furthermore, in the SB-11 structure, a specific central water molecule appears to engage Ser117 by H-bonds from all four monomers across the dimer−dimer interface ( Figure 4C).
TTR Tetramer Stability and Inhibition of Fibril Formation. The tetramer-stabilizing activity of the ligands was assessed by differential scanning fluorimetry (DSF). At neutral pH TTR has a midpoint of thermal denaturation (T m ) close to 100°C and an increase in stability is therefore not easily assayed. Consequently, we selected pH 5.0 as the pH for this assay where the T m for TTR is 92.3°C. Resveratrol and

ACS Chemical Neuroscience
pubs.acs.org/chemneuro Letter SB-14 provided a rather modest +1.1°C and +1.3°C thermal shift, respectively. Interestingly, we observed that SB-11 markedly elevated the T m more (+3.0°C) than SB-14 (Table 1). In addition, in line with this thermal stabilization, SB-11 was a better fibrillation inhibitor at pH 4.4 than resveratrol and SB-14 ( Table 1). The same trend of SB-11 activity outperforming resveratrol and SB-14 was also true for the TTR FAP mutation V30M both regarding thermal stability and fibril inhibition ( Table 1). The stabilizing effect of SB-11 is consistent with the conformational rigidity of Ser117 and differences in the hydrogen-bonding networks observed when comparing our structures. Although the ligands in this study are not negatively charged at the pH of our experiments, the forward binding mode of SB-11 orienting the hydrophobic naphthalene toward HBP3 and polar groups toward the exposed Lys15 is consistent with previous structures for flufenamic acid, 20 N-phenyl phenoxazines, 21 and Tafamidis. 8 Naphthalene Positioning Governs the pH-Dependence of Fluorescence Properties. That the binding mode appeared to affect ligand properties was evident from our previous pH-dependent fluorescence studies 18 ( Figure 5A). Interestingly, this property was limited to SB-14 when bound to TTR, showing a suppression of blue emission (390 nm) versus green emission (500 nm) illustrated in Figure 5A. Now, our structural data reveals that the reverse binding mode of SB-14 aligns the naphthalene moiety facing Lys15−Lys15′ at the entrance of the T4BP ( Figure 5A). The pH sensitivity of SB-14 bound to TTR likely reflects the protonation state of the amine headgroup of Lys15 as described previously 18 (apparent pK a of 8.5) in proximity of the naphthalene moiety of SB-14 Averaging of the density at the special position could explain the lack of electron density for the asymmetrically superposing naphthalene ring in this orientation. In panels C−F, the 2mFo-DFc electron density contoured at 1.5σ is shown as blue mesh and the mFo-DFc difference density contoured at 3σ in green (pos)/red (neg).
( Figure 5C). SB-11's fluorescence ratio was instead rather unaffected by pH in the range pH 5−10 ( Figure 5A). The binding mode observed in the TTR:SB-11 structure positions the naphthalene moiety inside HBP3, now explaining its reduced pH sensitivity ( Figure 5B). While the different positioning explains the different pH sensitivity of SB-14 versus SB-11, it does not explain the green fluorescence (500 nm) which interestingly was unique for TTR-T4BP binding for both ligands compared to fibril binding. 18

■ CONCLUSIONS
Misfolding and aggregation of TTR is associated with numerous gain-of-toxic function amyloid diseases called TTR amyloidoses with deposited ATTR. Today, there are many treatment options for ATTR diseases including liver transplantation, siRNA, and antisense oligonucleotides (ASOs) for modifying TTR expression as well as small molecule kinetic stabilizers to avoid tetramer dissociation and TTR misfolding and amyloid formation. 22 There is interest in generating new and improved treatment options, but early diagnosis is even more urgent, which is the key for effective treatment. We have  previously identified fluorescent ligands to report on native tetrameric and misfolded protofibrillar TTR. 18 Ligands were based on the trans-stilbene resveratrol. We concluded based on pH-dependent fluorescence spectroscopy experiments that two structural analogues of trans-stilbenes bound with opposing directions within the T4BP. Our high-resolution crystal structures of these complexes presented in this study now confirm that the binding modes are indeed opposite, albeit contrary to our prediction. The consequences of the different ligand binding modes to TTR are important, both for fluorescence and stabilization. Our data provides the structural basis for two critical parameters to facilitate the design of TTR fluorescent ligands and inhibitors based on trans-stilbene scaffolds.

■ MATERIALS AND METHODS
Chemicals. Chemicals were purchased from Sigma−Aldrich; columns and resins were from Cytiva. All chemicals used in the experiments were of reagent grade quality. Synthesis of ligands SB-11 and SB-14 was as previously reported. 18 Recombinant Expression and Purification of Wild-Type TTR and the TTR V30M Mutant. Expression and purification of TTR were carried out as described previously. 23 The TTR wild-type and the V30M mutation were expressed and purified the same way. Briefly, Escherichia coli BL21 (DE3) cells were transformed with the different plasmids and grown at 37°C. At an OD 600 of 0.6, the temperature was lowered to 20°C and TTR expression was induced by addition of 0.4 mM IPTG. After 18 h, cells were harvested, resuspended in 20 mM Tris-HCl, pH 8.0, 100 mM NaCl, and disrupted by sonication. After clarification the supernatant was heated to 60°C for 30 min. Subsequently, the precipitated material was removed by centrifugation and the supernatant was filtered (0.45 μm) followed by anion exchange (Source-15Q 10/10) and size exclusion chromatography (HiPrep 16/60 Superdex75) in 10 mM Naphosphate buffer, 100 mM KCl pH 7.6 at 20°C. Fractions containing pure TTR were collected, pooled, and concentrated to 5.2 mg/mL, prior to flash-freezing in liquid nitrogen and storage at −80°C until use.
Crystallization of the TTR:Ligand Complexes. The protein was crystallized using the vapor-diffusion hanging drop method at room temperature as described previously. 23 Briefly, purified TTR (5.2 mg/mL) was cocrystallized with either 500 μM SB-14 or SB-11 or 2 mM resveratrol (added from DMSO stock solutions (10 mM)). Drops containing 3 μL protein solution were mixed with 3 μL precipitant and equilibrated against 1 mL reservoir solution containing 1.3−1.6 M sodium citrate pH 5.5 and 3.5% v/v glycerol. Crystals grew to 0.1 × 0.1 × 0.4 mm after 5−7 days. Crystals were transferred into drops containing the same concentration of ligand and protein and incubated for 3 days. Crystals were cryo-protected with mother liquor supplemented with 12.5% v/v glycerol and ligand.
Structure Determination. Diffraction data of the "apo", and SB-11, SB-14, and resveratrol complex forms of TTR were collected under cryogenic conditions at BioMax (MAXIV), Sweden, at a wavelength of 0.97993 Å. Data were processed to a resolution of 1.15 Å, 1.45 Å, 1.45 Å, 1.35 Å, respectively, using XDS 24 and AIMLESS 25 from the CCP4 software suite. 26 Data collection statistics are summarized in Table 2.
Phasing was by molecular replacement using Phaser 27 with a search model derived from the published coordinates 1F41 28 (omitting terminal residues and a known flexible region). Ligands and solvent were placed in density after 1 to 2 initial rounds of rebuilding the protein model with COOT 29 and refinement using REFMAC. 30 After placing the ligands, a further 2 to 3 iterations of rebuild/REFMAC refinement were performed with ligand occupancy increasing when appropriate.
Ligand atoms were initially placed at 0.3 occupancy if clearly visible in the electron density; otherwise, occupancy was set at 0.1. Occupancy was increased in line with (i) developing density and (ii) consistency with B-factors of surrounding atoms/residues. Due to the positioning of compounds (on the special position) the maximum occupancy is 0.5.
Because of the lack of interpretable electron densities in the final map, nine N-terminal (residues 1−9) and two or three C-terminal residues were not included in the final model. A summary of the crystallographic analyses is presented in Table 2.
Fibril Formation Assay. TTR (2.0 mg/mL, in PBS buffer) was preincubated with or without 2 equimolar concentrations of inhibitor