Retinol-Binding Protein Interferes with Transthyretin-Mediated β-Amyloid Aggregation Inhibition

Abstract

β-Amyloid (Aβ) aggregation is causally linked to Alzheimer’s disease. On the basis of in vitro and transgenic animal studies, transthyretin (TTR) is hypothesized to provide neuroprotection against Aβ toxicity by binding to Aβ and inhibiting its aggregation. TTR is a homotetrameric protein that circulates in blood and cerebrospinal fluid; its normal physiological role is as a carrier for thyroxine and retinol- binding protein (RBP). RBP forms a complex with retinol, and the holoprotein (hRBP) binds with high affinity to TTR. In this study, the role of TTR ligands in TTR-mediated inhibition of Aβ aggregation was investigated. hRBP strongly reduced the ability of TTR to inhibit Aβ aggregation. The effect was not due to competition between Aβ and hRBP for binding to TTR, as Aβ bound equally well to TTR–hRBP complexes and TTR. hRBP is known to stabilize the TTR tetrameric structure. We show that Aβ partially destabilizes TTR and that hRBP counteracts this destabilization. Taken together, our results support a mechanism wherein TTR-mediated inhibition of Aβ aggregation requires not only TTR–Aβ binding but also destabilization of TTR quaternary structure.

Graphical Abstract

The characteristic features of Alzheimer’s disease (AD) include extracellular senile plaques, intraneuronal neurofibrillary tangles, and extensive neuronal cell death. The plaques contain insoluble amyloid deposits composed predominantly of the peptide β-amyloid (Aβ), produced by cleavage of the transmembrane amyloid precursor protein (APP).¹ Upon release from APP, Aβ monomers spontaneously self-assemble into soluble oligomers and insoluble fibrillar aggregates,^2,3 a process that is believed to be causally linked to neurotoxicity.^4,5

Transgenic mice expressing the Swedish mutation APP_Sw (Tg2576) produce high levels of Aβ and develop amyloid deposits.⁶ However, these mice lack neurofibrillary tangles and, although gliosis and dystrophic neuritis are observed, there is no neuronal cell loss,⁷ seemingly contradicting the hypothesis that Aβ aggregation is responsible for AD pathology. This result was explained by the demonstration of an increased level of transthyretin (TTR) expression in Tg2576 mice compared to controls.⁸ Administration of the anti-TTR antibody led to increased levels of tau phosphorylation and neuronal cell death.⁹ This result gave rise to the hypothesis that an increased level of TTR expression was a protective response against high levels of Aβ, a hypothesis that is supported by other animal studies.^10–15 TTR’s ability to abrogate Aβ toxicity in cell culture has been demonstrated in numerous studies.^16–19 Beyond cellular and animal studies, there is evidence for a role of TTR in human AD; for example, amyloid deposits in AD brain are strongly stained by TTR.¹⁸ TTR levels in cerebrospinal fluid (CSF) fluctuate with disease state,^20–27 suggesting that TTR levels respond dynamically to Aβ. Early in disease, TTR may be upregulated in response to loss of Aβ homeostasis, but as the disease progresses, TTR levels decrease as the protein is trapped in amyloid deposits or as cell death leads to a reduced level of synthesis.

TTR is a homotetrameric soluble protein (55 kDa) that is synthesized in both the liver and the choroid plexus and circulates in blood (3–7 μM) and CSF (0.1–0.4 μM).^28,29 The protein consists of four identical 127-residue subunits, each with two four-stranded antiparallel β-sheets and a short α-helix (Figure 1).^30,31 Extensive hydrogen bonding between β-strands H and F of two monomers forms the dimer. The packing of two dimers produces a hydrophobic channel in which the thyroid hormone thyroxine (T4) coordinates to residues in the inner β- sheet.³² Two T4 molecules may bind per TTR tetramer. The first T4 binds with high affinity (K_d ~ 10 nM) to TTR, but the second has lower affinity due to negative cooperativity; the TTR-T4 complex predominantly exists with a 1:1 stoichiometry in vivo.³³ TTR serves as the primary carrier of T4 in CSF and as a minor carrier in blood.^31,34 TTR also facilitates the delivery of retinol from blood to peripheral tissue through its interaction with retinol-binding protein (RBP; 21 kDa; 2–4 μM).³⁵ Most RBP (~95%) in circulation is complexed with TTR to prevent glomerular filtration and renal catabolism of RBP.³⁶ Holo-RBP (hRBP) coordinates with three subunits of TTR, making contact with the EF helix loop, AB loop, and GH loop in subunits B and C as well as the FG loop in subunit A (Figure 1).^37,38 Although two hRBPs can bind per TTR tetramer, the predominant species in vivo is a 1:1 complex.³⁹ The affinity of hRBP for TTR (K_d ~ 0.20–0.35 μM)^37,40,41 decreases upon loss of retinol.⁴⁰ In human serum, approximately half of TTR is complexed with hRBP and 10–20% with T4. The RBP-binding sites on TTR are orthogonally positioned to the nonoverlapping thyroxine-binding sites;^32,37,38 no competition has been observed.⁴² RBP is also synthesized in the choroid plexus and circulates in CSF (6–12 nM), where it likely plays a role in retinol uptake and transport in the brain.^43,44

Figure 1.

Ribbon structure of 1:1 binding of human serum hRBP to human TTR. The TTR tetramer (light gray, with subunits A—D) and hRBP (dark gray) coordinate by predominantly hydrophobic interactions. Residues on TTR (blue) and hRBP (orange) involved in binding are highlighted. Retinol (green) is buried in the hydrophobic cavity of RBP’s β-barrel with the alcohol moiety pointing toward the EF loop of TTR. Thyroxine binds in the central cavity (not shown). The putative weak Aβ-binding domain on the EF α-helix and the strong binding domain on β-strand G on each TTR monomeric subunit are colored red. Generated from Protein Data Bank entry 3BSZ.

Multiple investigators have shown that TTR binds directly to Aβ and inhibits Aβ aggregation.^{17–19,45,46} Furthermore, TTR- mediated inhibition of Aβ toxicity is directly linked to its inhibition of aggregation.^19,47 We and others have shown that binding of Aβ to TTR involves residues in β-strands G and H in the “inner” fi-sheet of TTR (Figure 1) in or near the thyroxine-binding sites.^46–48 There is some evidence supporting the existence of a weak secondary interaction between Aβ and TTR’s solvent-exposed EF α-helix (Figure 1).^19,47,49 TTR binds significantly more Aβ oligomers than Aβ monomers,^19,50 a factor that can explain how TTR is effective at inhibiting aggregation and toxicity at well below stoichiometric TTR:Aβ ratios.

TTR’s efficacy at binding Aβ, inhibiting aggregation, and/or inhibiting toxicity is inversely correlated with TTR tetramer stability.^46,47,50,51 For example, Aβ binding, inhibition of aggregation, or inhibition of toxicity was strongest with an engineered TTR mutant that is a stable monomer (mTTR), moderate for wild-type (wt) human TTR (huTTR), and weakest for wt murine TTR (muTTR), which forms more stable tetramers than huTTR does.^46,47,50 Intriguingly, Li et al. reported a modest (K_d ~ 20 μM) affinity of Aβ monomers for huTTR but could detect no evidence of monomeric Aβ binding to mTTR,⁴⁶ a result supported by a more recent investigation.⁵²

The putative Aβ-binding sites on TTR overlap with those of the natural ligands, T4 and RBP. T4 and RBP stabilize the TTR tetramer against dissociation at acidic pH,^42,53 and RBP stabilizes TTR tetramers against dissociation or subunit exchange under physiological conditions.^54,55 To date, in vitro investigations of TTR—Aβ interactions have been conducted in the absence of these ligands. In this study, we examined whether either ligand interferes with or alters TTR’s ability to protect against Aβ aggregation.

MATERIALS AND METHODS

Transthyretin Production, Purification, and Modification.

Wild-type human recombinant transthyretin (TTR) was produced and purified as previously described.^47,51,56 An extinction coefficient of 77600 M^—1 cm^–1 at 280 nm and a molecular weight of 55044 Da were used to determine the TTR tetramer concentration.⁵⁷

Transthyretin was fluorescently labeled with AlexaFluor 594 (AF594) C₅ Maleimide (ThermoFisher) for FRET binding assays. The conjugation scheme followed a maleimide reaction to TTR’s solvent-exposed cysteine residue (maximum of one per TTR monomer). AF594 was dissolved in anhydrous N,N-dimethylformamide (DMF, Sigma) to a concentration of 2 mM and then diluted into TTR (6 μM) in PBSA [10 mM Na₂HPO₄/NaH₂PO₄, 150 mM NaCl, and 0.02% (w/v) NaN₃ (pH 7.4)] to a final concentration of 200 μM [8% (v/v) DMF]. The TTR/AF594 mixture vial was wrapped in aluminum foil and gently shaken for 2 h at room temperature for dye conjugation. The mixture was dialyzed against PBSA for 48 h at 4°C to remove excess unconjugated AF594. The TTR tetramer conjugation efficiency was 65%. The protein concentration was determined from the absorbance at 280 and 588 nm using extinction coefficients of 77600 and 96000 M^—1 cm^–1 (Thermo Fisher) for TTR and AF594, respectively. A correction factor of 0.51 was used to account for the 280 nm absorbance contribution of AF594. TTR—AF594 conjugates were concentrated to 24 μM in PBSA and stored at 4 °C wrapped in foil.

Preparation of the Transthyretin—Ligand Complex.

A purified native human plasma apo-retinol-binding protein 4 (Fitzgerald Industries International, Acton, MA) stock was diluted to 30 μM in PBSA. The apo-RBP (aRBP) concentration was measured by the absorbance at 280 nm using an extinction coefficient of 40400 M^—1 cm^–1 and a molecular weight of 21000 Da.⁵⁸ Synthetic all-trans-retinol (Sigma-Aldrich) was dissolved in ethanol at a concentration of approximately 3 mM, and the concentration was determined by the absorbance at 325 nm using an extinction coefficient of 52480 M^—1 cm⁻¹ (Sigma-Aldrich). Holo-RBP (hRBP, 12 μM) was prepared by diluting retinol into aRBP in PBSA to final concentrations of 12 μM aRBP and 15 μM retinol (1:1.25 molar ratio); the mixture was incubated for 4 h at RT in the dark (final concentration of ethanol of <0.3%). The absorbance at 330 nm confirmed retinol binding (data not shown). The A₃₃₀/A₂₈₀ ratio of hRBP was approximately 0.80, which corresponds to 85% saturation of RBP with retinol.⁵⁹ Retinol binds aRBP with a 1:1 stoichiometry and an affinity of 50–80 nM.⁶⁰ RBP samples were filtered (0.22 μm) before being used. All retinol-containing samples were shielded from light, and fresh batches were regularly produced to minimize retinol oxidation. L-Thyroxine (T4, Sigma-Aldrich) was dissolved in 0.22 μm filtered 15 mM NaOH; the concentration was measured using an extinction coefficient of 6180 M^–1 cm^–1 at 325 nm.⁶¹

Transthyretin—ligand complexes (6 μM) were prepared by diluting aRBP, hRBP, or T4 into TTR at a 1:1 molar ratio and incubating the mixture at room temperature overnight in the dark. Binding of retinol to aRPB and hRBP to TTR was confirmed by FRET (Figure S1). The fluorescence anisotropy (monitoring 460 nm emission at 330 nm excitation) of hRBP and that of TTR with hRBP (1:1) were 0.19 ± 0.02 and 0.26 ± 0.02, respectively, further confirming binding of hRBP to TTR.⁴¹ After preparation, all samples were stored at 4 °C wrapped in foil.

β-Amyloid Preparation.

Lyophilized Aβ(1—40) (Aβ40, American Peptide) was dissolved in 0.22 μm filtered 50% acetonitrile to a concentration of 1 mg/mL, frozen at —80 °C, and relyophilized. Aβ40 was redissolved in filtered (0.22 μm) denaturing buffer [8 M urea and 100 mM glycine-NaOH buffer (pH 10)] to a final concentration of 2.8 mM. Aβ40 aliquots of 4 or 8 μL were snap-frozen and stored at —80 °C. To further disaggregate Aβ40 immediately before each experiment, aliquots were thawed and incubated with a final concentration of 80 mM NaOH for 15 min at RT. Aβ40 was diluted into 0.22 μm filtered PBS [10 mM Na₂HPO₄/NaH₂PO₄ and 150 mM NaCl (pH 7.4)] or PBSA to initiate aggregation. Aβ40 monomers have a molecular weight of 4330 Da and an extinction coefficient of 1490 M^—1 cm^–1 at 280 nm.⁶² All Aβ40 concentrations in this study are reported in terms of equivalent monomer molar concentrations. To prepare Aβ40 oligomers, Aβ40 monomers (80 μM) in PBSA were incubated at RT for 24 h and diluted to experimental concentrations prior to use.⁶³

HiLyte Fluor 488-labeled Aβ40 (Aβ−488, Anaspec) was reconstituted in 0.22 μm filtered denaturing buffer to a final concentration of 1.6 mM. Small volumes were aliquoted, snap-frozen, and stored at —80 °C. Oligomeric Aβ40 with fluorescent tracers was produced by diluting an aliquot of Aβ−488 into a stock of unlabeled Aβ40 monomers in PBSA at a 1:36 molar ratio of Aβ−488 to unlabeled Aβ40 and then incubating at RT for 24 h wrapped in foil. Aβ−488 was used only in FRET binding assays with TTR-594.

Thioflavin T Fluorescence.

Thioflavin T (ThT; Fisher Scientific, Fair Lawn, NJ) at 11 μM in PBSA (0.22 μM filtered) was prepared using an extinction coefficient of 26.6 mM^—1 cm^—1 (416 nm in ethanol).⁶⁴ Aβ40 (28 μM) was incubated alone or with TTR, aRBP, hRBP, T4, TTR aRBP, TTR+hRBP, or TTR +T4 (3.6 μM) in PBSA at 37 °C +for 1 or 48 h. Aβ40 (28 μM) was also incubated with 0.5, 1.0, or 1.8 μM TTR in PBSA at 37 °C for 1 or 48 h. Immediately prior to each measurement on a QuantaMaster spectrofluorometer, 10 μL of protein was mixed with 130 μL of 11 μM ThT and excited at 440 nm with emission recorded from 455 to 500 nm. Retinol does not absorb between 440 and 500 nm.⁴¹ The mean and standard deviation of triplicate measurements at 480 nm minus the background of ThT alone in PBSA were used to estimate the relative mass content of Aβ fibrils in each sample. Experiments were repeated multiple times with similar results. In separate experiments, Aβ40 (28 μM) was incubated alone or with TTR or TTR+hRBP (3.6 μM) in PBSA at 37 °C for ≤60 h; aliquots were removed at several time points, and the ThT fluorescence intensity was measured as described. The kinetic data were fitted to a logistic sigmoidal growth equation to obtain parameters a, k_app, and t₅₀, which represent the maximum relative ThT signal, the apparent rate constant of Aβ fibril growth, and the time to reach 50% of the maximum relative ThT signal, respectively.^65–67

Dynamic Light Scattering.

Aβ40 (28 μM final concentration) was diluted into TTR, hRBP, or TTR+hRBP (3.6 μM final concentration) in filtered (0.02 μm) PBSA and then immediately filtered (0.22 μm) directly into an extensively cleaned light scattering cuvette and positioned in a bath of index-matching solvent decahydronaphthalene temperature controlled at approximately 37 °C. Using a Brookhaven BI-200SM system (Brookhaven Instruments Corp., Holtsville, NY) and an Innova 90C-5 argon laser (Coherent, Santa Clara, CA) operating at 488 nm and 150 mW, light scattering data were collected at 90°. The z-averaged hydrodynamic diameter (⟨d_h⟩_z) for each sample was calculated from the autocorrelation function using the method of cumulants.⁶⁸ ⟨d_h⟩_z includes contributions from all species in solution (Aβ40 monomer, Aβ40 aggregates, TTR, hRBP, or TTR—hRBP complexes) but is weighted more heavily toward the larger species. DLS kinetic data were fitted to a power-law equation to determine the rate of Aβ40 aggregate size growth (see the Supporting Information).

Nanoparticle Tracking Analysis.

Nanoparticle tracking analysis (NTA) measurements were collected using a Nanosight LM10 instrument (Nanosight, Amesbury, U.K.) equipped with a 405 nm laser. Aβ40 (28 μM) alone or with TTR, hRBP, or TTR+hRBP (3.6 μM) in PBSA was filtered (0.02 μm) and injected into the sample chamber using a syringe. Experiments were conducted at RT (~20 °C). One 90 s video was captured at multiple time points between 0 and 2 h after sample preparation with the camera level set to the maximal value. The experiment was stopped when the number of particles reached the upper limit recommended by the instrument manufacturer. The data were recorded and analyzed using NTA version 2.3. TTR, hRBP, or buffer (PBSA) alone contained negligible particle counts (data not shown).

Enzyme-Linked Immunosorbent Assay (ELISA).

Protein stocks were prepared at concentrations of 5—10 μM and diluted immediately prior to use. ELISA plates (Corning, Inc., Corning, NY) were coated with 50 nM TTR, hRBP, or TTR+hRBP (100 μL/well) in coating buffer [10 mM sodium carbonate, 30 mM sodium bicarbonate, and 0.05% NaN₃ (pH 9.6)] overnight at 4 °C protected from light. The plate was washed at least three times with wash buffer (200 μL per well per wash; PBS with 0.05% Tween 20) and incubated with protein-free blocking buffer (300 μL/well; Pierce, Rockford, IL) for 2 h at RT. As negative controls, TTR+hRBP were not coated but wells were still blocked. Aβ40 (86 μM in PBS) was incubated at RT for 1 day to form oligomers, then diluted to 250, 400, 600, or 750 nM in PBS, and immediately added to protein-coated or negative control wells (50 μL/well). The plate was incubated at 37 °C for 1 h. After washing, anti-Aβ antibody 6E10 (Covance, Princeton, NJ) in wash buffer (1:3000) was added to each well (100 μL/ well) and incubated at RT for 1 h with gentle shaking. After washing, the anti-mouse HRP antibody (1:3000 dilution; Pierce) was added to each well (100 μL/well) and incubated for 1 h at RT with gentle shaking. After several washes, 100 μL of a 3,3′,5,5′-tetramethylbenzidine (TMB) substrate solution (Pierce) was added to each well. Color development (15—30 min incubation at RT) was stopped by addition of 100 μL of 2 M sulfuric acid per well. The absorbance was measured at 450 nm with an EL800 Universal Microplate Reader (Biotek Instruments, Inc., Winooski, VT). Aβ40 binding was calculated as the mean of four replicate wells minus the mean of the negative control (Aβ40 without TTR, hRBP, or TTR and hRBP) absorbance.

Experiments measuring binding of aRBP and hRBP (50 nM) to Aβ40 monomers, oligomers, and fibrils were conducted similarly as described above, except for the preparation of Aβ. Aβ40 (80 μM) was freshly prepared (monomers) or incubated at RT for 24 h to form oligomers and at 37 °C for 24 h to form fibrils.⁶³ Prior to being added to wells, Aβ40 was diluted to a concentration of200 nM in PBS. Aβ40 binding was calculated as the mean of three replicate wells minus the mean of the negative control (Aβ40 monomers, oligomers, or fibrils without aRBP or hRBP) absorbance.

Steady State Fluorescence Resonance Energy Transfer.

All measurements were taken using a QuantaMaster spectrofluorometer. Afi40 oligomers binding fluorescently tagged TTR followed the FRET scheme illustrated in panels A and B of Figure S2. Aβ40 oligomers (28 μM) with fluorescent tracers (36:1 molar ratio of unlabeled Aβ40 to Aβ−488) were incubated with varying concentrations of TTR-594 with or without hRBP (0, 100, 250, 500, 1000, 1500, 2000, 3500, and 5000 nM) in PBSA for 2 h at 37 °C wrapped in foil. Fluorescence spectra of all samples were measured using an excitation of 504 nm and emission scan from 525 to 700 nm. Background fluorescence spectra of TTR-594 with or without hRBP in PBSA at each concentration were subtracted using an excitation of 504 nm. The emission of the donor (Aβ−488) at 544 nm was used to determine the FRET efficiency (E):

$$\text{E}=1-\frac{{\text{F}}_{\text{DA}}}{{\text{F}}_{\text{D}}}$$ (1)

where F_DA and F_D are the emissions at 544 nm in the presence and absence of the acceptor (TTR-594), respectively. The mean and standard deviation (triplicate experiments) of FRET efficiency as a function of acceptor concentration were reported. Affinities (K_d) for binding of Aβ−488 oligomers to TTR-594 tetramers were determined by fitting to a 1:1 binding model (see the Supporting Information).⁶⁹ Binding of TTR-594 and hRBP was confirmed by an increase in fluorescence anisotropy compared to that of hRBP alone (data not shown).

FRET of intrinsic aromatic residues of hRBP and/or TTR (donor) to bound retinol (acceptor) followed the schemes illustrated in panels C—F of Figure S. Aβ40 monomers (80 μM in PBSA) were incubated for 24 h at RT to form Aβ40 oligomers, then diluted in PBSA (5 or 15 μM Aβ40) alone or with either hRBP or TTR and hRBP (1 μm), and incubated at 37 °C for 1 h. FRET was monitored by measuring the fluorescence spectra using an excitation of 280 nm and emission scan from 295 to 550 nm. The retinol fluorescence was monitored using an excitation of 330 nm and emission scan from 350 to 550 nm. The mean of triplicate spectra minus the background (PBSA or Aβ40 alone) was reported. Unbound, free retinol exhibits negligible fluorescence in PBSA; the fluorophore alone is quenched in aqueous solvents.

Gel Electrophoresis (SDS—PAGE) of TTR—hRBP—Aβ Complexes.

For monitoring Aβ-induced TTR tetramer destabilization, TTR or TTR with hRBP (3.6 μM) was mixed with Aβ40 monomers (60 μM) in PBSA and then incubated for 0–50 h at 37 °C to induce Aβ40 aggregation. Samples were diluted into Tris-glycine SDS sample buffer (Novex) and briefly centrifuged. Control TTR and TTR/hRBP samples were boiled for 10 min to use as RBP and monomeric TTR standards on the gel. Samples were loaded in a 10 to 20% polyacrylamide gradient gel (Novex), electrophoresed in Tris-glycine SDS running buffer (Novex), and stained with Coomassie blue. Destained gels were photocopied, and the densitometry of each protein bands was quantified using ImageJ. The percentage of total TTR tetramer dissociation was determined by the density ratio between the amount of monomeric TTR and total TTR protein. It was reported as the mean and standard deviation of three replicate gel assay experiments.

RESULTS

hRBP Interferes with TTR-Mediated Inhibition of Aβ Aggregation.

We and others have previously demonstrated that TTR inhibits Aβ aggregation and toxicity in vitro.^{16,18,19,46,47,50,70} These experiments were conducted in the absence of TTR’s natural ligands, hRBP and T4. RBP’s binding to TTR involves residues in the EF α-helix’ overlapping with a possible Aβ-binding domain (Figure 1).^37,38,47 T4 binds in the hydrophobic cavity of the TTR tetramer, interacting with several residues known to be important for Aβ-TTR binding.^32,47 We therefore inquired if either RBP or T4 would affect the protective ability of TTR against Aβ aggregation.

The Aβ40 fibril content was monitored by ThT, a fluorescent dye that is widely used as a measure of the mass of amyloid fibrils.^71,72 Aβ40 alone was initially ThT-negative but developed a strong ThT-positive signal by 48 h, as expected (Figure 2A). TTR (7:1 Aβ40:TTR molar ratio) decreased the ThT intensity to 25% of that of Aβ40 alone at 48 h, consistent with previously published data.⁵¹ In contrast, a 1:1 mixture of TTR and hRBP decreased the ThT intensity to only 60% of that of Aβ40 alone at 48 h, indicating that TTR with hRBP is much less effective at inhibiting Aβ40 fibrillation than TTR alone (p < 0.001). On the other hand, a 1:1 TTR and aRBP mixture was nearly as effective at inhibiting Aβ40 fibrillation as TTR alone.

Figure 2.

TTR-mediated inhibition of Aβ fibril formation after 48 h. (A) TTR alone (3.6 μM) or complexed with aRBP, hRBP, or T4 (3.6 μM) was incubated with Afi40 (28 μM) at 37 °C for 1 or 48 h. (B) Aβ40 (28 μM) alone or with TTR (3.6, 1.8, 1.0, or 0.5 μM) was incubated at 37 °C for 1 or 48 h. The Aβ40 fibril content was detected by ThT fluorescence. Data shown are the mean ± SD of three replicates. TTR, aRBP, hRBP, and T4 did not exhibit ThT fluorescence (data not shown). An asterisk indicates the value differs from the value for 48 h Aβ alone (p < 0.001). 0 indicates the value differs from the value for 48 h TTR (3.6 μM) and Aβ (p < 0.001). w indicates the value differs from the value for 48 h Aβ alone (p < 0.01). A number sign indicates the value differs from the value for 48 h TTR (3.6 μM) and Aβ(p < 0.005).

Given an estimate for K_d of 0.35 for binding of hRBP to TTR at a 1:1 stoichiometry,³⁷ free hRBP and TTR will be in equilibrium with the TTR-hRBP complex and must be accounted for. We calculated that 27% of TTR (1 μM) is not bound to hRBP at equilibrium under our experimental conditions. At 1 μM, TTR decreased the ThT intensity to 60% of that of Aβ40 alone (Figure 2B), similar to the result with the 3.6 μM TTR+hRBP mixture (Figure 2A). aRBP binds more weakly to TTR than hRBP does (K_d ~ 1.20 μM).⁴⁰ Under our experimental conditions, the 1:1 TTR+aRBP mixture would contain 43% unbound TTR (~1.6 μM) at equilibrium. For comparison, TTR alone at a similar concentration (1.8 μM) (Figure 2b) was not as effective as the 3.6 μM TTR+aRBP mixture at inhibiting Aβ40 fibrillation (Figure 2A). This suggests that, unlike the case for hRBP, when aRBP is complexed to TTR, it does not affect TTR’s ability to hinder Aβ fibrillogenesis. Taken together, these data indicate that the decrease in ThT intensity with the TTR+hRBP mixture is likely due to the inhibitory activity of uncomplexed TTR and that the TTR–hRBP complex is ineffective at inhibiting Aβ fibrillation.

TTR with bound thyroxine was also less effective against Aβ40 fibrillation than uncomplexed TTR was (Figure 2A). However, because the population of uncomplexed TTR is very low given the strong binding affinity (K_d = 10 nM),³³ we conclude that the TTR-T4 complex still exhibits inhibitory activity against Aβ40 fibrillation, albeit modestly less than free TTR. Because the effect of hRBP was much stronger, we chose to focus our remaining work on hRBP.

Next, ThT was used to measure fibrillogenesis kinetics for Aβ40 alone or co-incubated with TTR or TTR+hRBP (Figure 3). The data were fit to a simple sigmoidal growth curve (eq 2),^65–67

$$\text{y}=\frac{\text{a}}{1+\exp \left[-{\text{k}}_{\text{ app }}\left(\text{t}-{\text{t}}_{50}\right)\right]}$$ (2)

where y is the ThT signal at time t, a is the maximum ThT signal (normalized to Aβ40 alone), k_app is the apparent rate constant of fibril growth, and t₅₀ is the time to reach 50% of the maximum signal. For all three curves, there was a lag of ~5 h before observation of the ThT-positive signal. Aβ alone reached a plateau at 48 h. Upon co-incubation with TTR and hRBP, fibrillation was slower, but at long times, the ThT signal approached 70% of that of Aβ alone. In contrast, co-incubation with TTR resulted in a ThT signal that plateaued at only 20% of that of Aβ, indicating that TTR partially prevents, not simply delays, Aβ fibrillogenesis. k_app was lower when Aβ40 was incubated with TTR compared to Aβ40 alone (p < 0.05). For Aβ40 with TTR and hRBP, k_app was intermediate between those of Aβ40 alone and Aβ40 with TTR (Table S1). Values of t₅₀ for the three samples ranged from 19 to 27 h but were not statistically different (Table S1).

Figure 3.

TTR-mediated inhibition of Aβ40 fibril growth kinetics. Aβ40 (28 μM, ♦ ) alone or with TTR (3.6 μM, ♦) or TTR+hRBP (3.6 μM, ■ ) was incubated at 37 °C for 0–60 h. Aβ40 fibril growth was monitored by ThT fluorescence. Data shown are the means ± SD of three replicates. The curves show the fit to a sigmoidal growth equation (see the Supporting Information).^65–67

Further confirmation that hRBP impairs TTR’s efficacy as an anti-aggregation agent was sought from light scattering experiments. The size of Aβ aggregates and the rate of growth were monitored by dynamic light scattering (DLS). While ThT fluorescence reports on the mass of amyloid fibrils, DLS provides information about the size of particles in solution. The technique is particularly useful at detecting aggregation at early times, prior to the onset of the ThT signal, because of its high sensitivity to very low levels of aggregates. The mean hydrodynamic diameter (⟨d_h⟩_z) of Aβ40 (28 μM) in solution increased rapidly over time (Figure 4). The hydrodynamic diameter at time zero was reduced when either TTR or TTR +hRBP (3.6 μM) were added, because ⟨d_h⟩_z includes both Aβ40 aggregates (150 nm) and TTR or TTR and hRBP (6 or 8 nm, respectively). The increase in aggregate size was measured, and the data were fit to a power-law equation

$${\langle {\text{d}}_{\text{h}}\rangle }_{\text{z}}={\langle {\text{d}}_{\text{h}}\rangle }_{\text{z}0}+{\text{k}}_{\text{obs}}{\text{t}}^{\mathcal{v}}$$ (3)

where ⟨d_h⟩_z0 is ⟨d_h⟩_z at time zero, k_obs is the observed rate constant for growth, and v provides some indication of the primary mechanism of growth (see the Supporting Information). TTR significantly slowed Afi40 growth (Figure 4), consistent with ThT results (Figure 3) as well as previous reports.^19,51 In contrast, TTR and hRBP did not slow the growth of Aβ40 aggregates, as seen by comparison of k_obs values (Table S2).

Figure 4.

Effect of TTR or TTR with hRBP on Aβ40 aggregate growth kinetics. The mean hydrodynamic diameter of Aβ40 alone (28 μM, ♦) or Aβ40 incubated with 3.6 μM TTR (□) or TTR+hRBP (■) at 37 °C was monitored by DLS. Curves show fits of data to eq 3 (see the Supporting Information). Data shown are from one representative run. The experiment was repeated two or three times with similar results.

We further analyzed the effect of TTR+hRBP on Aβ40 aggregate size using nanoparticle tracking analysis (NTA), a particle-by-particle scattering technique, in which both the total number of particles and the particle size distribution can be ascertained. Only ≳30 nm particles are reliably detected.^73,74 TTR, hRBP, and TTR–hRBP complexes are smaller than 30 nm and therefore invisible with this technique, as are the Aβ40 monomer and smaller oligomers. Aβ40 (28 μM) was incubated alone, with 3.6 μM TTR, or with TTR+hRBP for 2 h. Data are reported as the particle number concentration as a function of time (Figure 5A) or as the particle size distribution 10 or 60 min after preparation (panel B or C of Figure 5, respectively). TTR reduced both the number concentration of large Aβ40 aggregates (Figure 5A) and the aggregate size (figure 5B,C), indicating that the protein suppressed the formation of new Aβ aggregate particles as well as the growth of preexisting aggregates. In contrast, the total number of Aβ40 aggregate particles was essentially unaffected by TTR+hRBP (Figure 5A). Furthermore, size distributions were shifted toward larger Aβ40 aggregates upon incubation with TTR+hRBP as compared to those in their absence (Figure 5B,C). The increase in aggregate size with TTR+hRBP could be explained by hypothesizing that TTR–hRBP complexes bind to Aβ40 aggregates but do not inhibit their growth. If the TTR–hRBP complex did not interact with Aβ40 at all, one would expect to see the same size distribution for Aβ40 alone compared to that with the mixture of Aβ40, hRBP, and TTR because TTR–hRBP complexes are too small to be detected by NTA. In all cases, the aggregates that did form were fibrillar, as imaged by TEM (Figure S3).

Figure 5.

Total aggregate number concentrations and particle size distributions measured by NTA. (A) Aβ40 alone (28 μM, ♦) or with 3.6 μM TTR (□) or TTR+hRBP (■) was incubated at RT. The total Aβ40 aggregate number concentration was reported as a function of time. (B and C) Particle size distributions of Aβ40 aggregates alone (black) or incubated with TTR (gray) or TTR+hRBP (white) at RT for 10 and 60 min, respectively. Data shown are from one representative run. The experiment was repeated three times with similar results.

Because an estimated 27% of hRBP remains dissociated from TTR under our experimental conditions, we tested whether hRBP in the absence of TTR affected Aβ40 aggregation. By all measures (ThT, DLS, and NTA), we observed that hRBP modestly suppressed Aβ40 aggregation (Figures S4 and S5). Thus, although either hRBP or TTR by itself can mildly (hRBP) or strongly (TTR) inhibit Aβ40 aggregation, the TTR–hRBP complex is a poor inhibitor of aggregation.

The Transthyretin–hRBP Complex Binds β-Amyloid.

We next asked why TTR–hRBP complexes were less effective than TTR at inhibiting Aβ40 aggregation. TTR must bind to Aβ to inhibit its aggregation,¹⁹ so direct interference of hRBP with binding of Aβ to TTR could explain the loss of efficacy. To determine whether hRBP interfered with binding of Aβ40 to TTR, we conducted a solution-based fluorescence resonance energy transfer (FRET) assay. Briefly, Aβ40 was mixed with HiLyte-488-labeled Aβ40 (36:1 Aβ:Aβ−488 molar ratio, total Aβ40 concentration of 28 μM), incubated at RT for 24 h to form oligomers, and then added to TTR labeled with AlexaFluor 594 (TTR-594) or TTR-594 and hRBP. Binding was measured by an increase in FRET efficiency as a function of TTR-594 concentration (Figure 6A). The data were fit to a simple 1:1 binding model to obtain estimates of pseudo K_d of 2.6 ± 0.8 μM for Aβ oligomers binding to TTR tetramers and 1.9 ± 0.8 μM for Aβ oligomers binding to TTR and hRBP (Figure S6 and Table S3).⁶⁹ These are pseudo K_d’s because the binding model does not account for the heterogeneous and oligomeric nature of the Aβ preparation. There was no statistical difference between the two affinities, indicating no effect of hRBP on binding of Aβ40 to TTR.

Figure 6.

Binding of Aβ to TTR and the TTR–hRBP complex. (A) Oligomeric Aβ40 with fluorescent tracers (HiLyte-488) at 28 μM (36:1 Aβ:Aβ−488 molar ratio) were incubated with increasing amounts of TTR-594 without (□) and with hRBP bound (■). The FRET efficiency was measured as a function of FRET acceptor concentration (TTR-594). Data shown are the mean ± SD of three replicates. (B) TTR (□), hRBP (∘), and TTR+hRBP (■) were adsorbed to a 96-well plate. Oligomeric Aβ40 (0, 250, 400, 600, or 700 nM) was added to each well and allowed to bind for 1 h. Unbound Aβ40 was removed by washing, and the amount of Aβ40 bound was measured by an ELISA. Data shown are the mean ± SD of four replicates. These experiments were repeated multiple times with similar results.

We confirmed this finding by an ELISA. In brief, TTR or TTR with hRBP was immobilized on ELISA plates, Aβ40 oligomers were added to each well, and binding was detected by anti-Aβ antibodies. There was no statistical difference in the amount of Aβ40 bound to TTR versus the TTR–hRBP complex (Figure 6B).

As shown in Figure 6B and Figure S7, hRBP alone also bound Aβ40, although less than TTR or the TTR–hRBP complex. This is consistent with the weak inhibition of Aβ40 aggregation (Figures S4 and S5). We asked next whether Aβ binds directly to TTR or to hRBP on the TTR–hRBP complex. To do this, we utilized FRET from intrinsic aromatic residues (donor) to bound retinol (acceptor). With hRBP alone, excitation at 280 nm resulted in FRET and retinol emission at ~460 nm. Addition of Aβ40 oligomers to hRBP increased the retinol emission intensity (Figure 7A,B and Figure S7B), which we attributed to binding of Aβ40 to hRBP (possibly due to additional FRET from Aβ40’s tyrosine, change in hRBP aromatic residue emission due to Aβ40 binding, or enhancement of retinol’s quantum yield by further secluding bound retinol from the solvent via Aβ40 binding). In contrast, addition of a 15-fold excess of oligomeric Aβ40 did not affect the emission of retinol from TTR and hRBP (Figure 7C); no significant change was observed via direct excitation of the acceptor either (Figure 7D). Because binding of Aβ40 directly to hRBP caused an increase in retinol emission, we interpreted these data to indicate that, on the TTR–hRBP complex, Aβ40 binds to TTR at a site distant from hRBP and not to hRBP. hRBP contains a solvent-accessible hydrophobic surface that is buried upon complexation with TTR, which could explain why Aβ40 binds to hRBP differently than to the TTR–hRBP complex.

Figure 7.

Fluorescence emission spectra of hRBP and the TTR–hRBP complex with Aβ40 oligomers. Emission was recorded with excitation at (A and C) 280 nm or (B and D) 330 nm. (A and B) hRBP (1 μM) was incubated alone or with oligomeric Aβ40 (5 or 15 μM). (C and D) TTR+hRBP (1 μM) were incubated alone or with oligomeric Aβ40 (15 μM). Data shown are the mean of three replicate spectra minus the background spectra of buffer or Aβ40 alone.

Taken together, these results demonstrate that oligomeric Aβ40 binds equally well to the TTR–hRBP complex and to TTR. Therefore, the loss of TTR-mediated inhibition of Aβ40 aggregation by formation of the TTR–hRBP complex cannot be attributed to competition between hRBP and Aβ40 oligomers for binding to TTR.

hRBP Stabilizes TTR against Aβ-Mediated Destabilization.

In previous work, we showed that binding of Aβ to TTR is required for TTR-mediated inhibition of aggregation, that Aβ binding appears to destabilize TTR tetramers, and that destabilized TTR mutants are more effective than the wt as inhibitors of Aβ aggregation.^19,47,49,51 hRBP is known to stabilize TTR tetramers against dissociation or subunit exchange at physiological pH.^42,55 Because Aβ binds to the TTR–hRBP complexes, we postulated that hRBP counteracts Aβ-mediated destabilization of TTR tetramers and subsequently prevents TTR from inhibiting Aβ aggregation.

To test this hypothesis, we used an SDS–PAGE assay that we previously developed to observe the stability of TTR tetramers at physiological pH and temperature. Briefly, TTR or the TTR–hRBP complex was mixed with fresh Aβ40 and then incubated for 0–20 h at 37 °C to allow for Aβ40 aggregation. These samples along with TTR or the TTR–hRBP complex without Aβ40 were diluted into 2% SDS, applied to a gel without boiling, and subjected to electrophoresis. Samples are not cross-linked, so this assay measures the resistance of the protein complex to SDS-induced dissociation. Under these conditions, TTR is stable as a tetramer and does not dissociate into its monomeric subunits,¹⁹ while TTR–hRBP complexes dissociate into their constituent proteins (tetrameric TTR and RBP). To facilitate identification, TTR and the TTR–hRBP complex were boiled and then applied to the same gel. Boiling dissociates TTR into its monomeric subunits (Figure 8A). When TTR was incubated with Aβ40, a new band corresponding to the TTR monomer appeared (Figure 8A), as reported previously.¹⁹ By densitometry, we quantified the appearance of the TTR monomer as a function of time. The TTR monomer was detected after co-incubation with Aβ40 for ~10 h, with 5 ± 1% of the total TTR dissociated after 20 h (Figure 8B). The appearance of a measurable amount of TTR monomer corresponded to a decrease in the intensity of the Aβ40 monomer/dimer band. This time scale is consistent with the kinetics of Aβ40 fibrillogenesis as measured by ThT (Figure 3). This suggests that the association of Aβ40 into SDS-resistant, ThT-positive aggregates may trigger TTR destabilization. When Afi40 was co-incubated with TTR and hRBP, no TTR monomer band was detected (Figure 8), showing that hRBP hinders Aβ-induced dissociation of the TTR tetramer to TTR monomers. Because we have established that Aβ40 binds to the TTR–hRBP complex, this result supports the hypothesis that hRBP stabilizes TTR, counteracting Aβ-induced destabilization.

Figure 8.

Effect of hRBP on Aβ-mediated TTR tetramer destabilization. TTR or TTR with hRBP (3.6 μM) was mixed with Aβ40 monomers (60 μM) and then incubated for 0,5, 10, or 20 h at 37 °C to induce Aβ40 aggregation. (A) SDS-PAGE gel of TTR or TTR with hRBP without (—) or with (+) Aβ40. For comparison, TTR and TTR with hRBP without Aβ40 was incubated at 37 °C for 20h. TTR and TTR with RBP were also boiled, providing molecular weight markers for hRBP and TTR monomers. (B) Percent of Aβ-induced TTR tetramer dissociation in the absence (□) and presence of hRBP (■) determined by densitometric measurement of the ratio between the amount of monomeric TTR and total TTR protein. The results shown represent the mean ± SD of three independent experiments. With longer co-incubation periods with Aβ40, the TTR tetramer bands are shifted toward slightly lower molecular weights (Figure S8).

DISCUSSION

Over the past two decades, evidence that the circulating transport protein TTR may play a neuroprotective role in Alzheimer’s disease has accumulated. The earliest reports of interaction between Aβ and TTR were published more than 20 years ago.^70,75 The biological implications of this interaction became clearer when studies demonstrated that TTR provides protection against amyloid toxicity in transgenic APP mice.^9,13,14,18 The relevance of these observations in animal models to human disease has been supported by proteomic studies showing a complex, dynamic relationship between Aβ and TTR levels in AD.²⁴

These intriguing findings have motivated more detailed investigations of the molecular basis for the interactions between Aβ and TTR. The primary binding site has been identified as involving residues on TTR’s inner β-sheet, facing the interior hydrophobic cavity.^19,46,47,49 There may be a secondary Aβ- binding site at or near the EF helix,^19,47,49 although that finding was not confirmed.⁴⁶

Because the putative Aβ-binding domains on TTR overlap with binding sites for T4 and hRBP, we examined the effect of these natural ligands on TTR’s ability to inhibit Aβ aggregation. In our initial assessment, we observed that hRBP has a greater adverse impact than T4 on TTR-mediated inhibition of aggregation, so we chose to focus on the role of hRBP. TTR–hRBP complexes are significantly less effective at inhibiting Aβ40 aggregation than is TTR, as demonstrated by ThT fluorescence, dynamic light scattering, and nanoparticle tracking analysis (Figures 2–5 and Tables S1 and S2). In addition, any inhibition of Afi40 aggregation with TTR+hRBP mixtures can be attributed to free TTR (Figure 2). Although by themselves hRBP modestly (Figures S4 and S5) and TTR strongly suppressed Aβ aggregation, the TTR–hRBP complex is a poor inhibitor of aggregation.

If hRBP blocked binding of Aβ40 to TTR, this could explain the loss of TTR-mediated inhibition of Aβ40 aggregation. However, Aβ40 binds equally well to TTR–hRBP complexes and TTR alone (Figure 6), ruling out competition between Aβ40 and hRBP for binding to TTR. Because there are two hRBP-binding sites on TTR but only one is typically occupied (Figure 1), this result does not contradict the possibility of an Aβ-binding site at or near the EF helix.

To explain the lack of inhibition of aggregation despite the evidence of binding, we considered the fact that hRBP binding is known to stabilize TTR tetramers.^42,55 If tetramer destabilization is important for inhibition of Aβ aggregation, then hRBP may prevent this from occurring. We used a gel electrophoresis assay, which detects resistance to SDS as a measure of stability, to test this hypothesis. We observed that Aβ40 destabilized TTR tetramers over time and that hRBP prevented this destabilization (Figure 8). Taken together, these data demonstrate that binding of Aβ and inhibition of its aggregation are separate events: binding of Aβ to TTR is necessary but not sufficient for inhibition of Aβ aggregation. Further support for the hypothesis that TTR–hRBP complexes can bind to Aβ40 aggregates without inhibiting aggregation comes from NTA experiments, where we observed a measurable shift of the size distribution toward larger aggregates without a change in the number concentration (Figure 5). In addition, aRBP does not stabilize TTR tetramers,⁵⁵ unlike hRBP, and the TTR–aRBP complex is nearly as effective as TTR at inhibiting Aβ40 aggregation (Figure 2).

The details of the mechanism of TTR-mediated inhibition of Aβ aggregation are still a matter of debate. Several groups have established that Aβ oligomers and fibrils bind to TTR to a greater extent than do Aβ monomers^49,50 and that TTR’s efficacy at binding Aβ and inhibiting aggregation is inversely correlated with TTR tetramer stability.^46,49–51 Li et al.⁴⁶ proposed that the dominant in vivo mechanism of TTR-mediated inhibition of Aβ aggregation was binding of Aβ monomers, and possibly oligomers, to TTR tetramers, with binding in or near the site occupied by T4. In this model, TTR- Aβ binding decreases the concentration of Aβ and thus decreases the rate of initiation of aggregation. We have proposed a somewhat different mechanism. Specifically, our group has hypothesized that binding of Aβ oligomers to TTR, possibly at the EF α-helix, triggers partial destabilization of TTR tetramers, opening the interior hydrophobic cavity for further Aβ binding, and that the small fraction of TTR that is naturally monomeric may play an outsized role in sequestering Aβ oligomers.^19,47,49,51 In agreement with this, monomeric TTR was shown to be a better inhibitor of curli amyloid formation than is the tetramer.⁷⁶ The data collected in this study support the hypothesis that both binding and tetramer destabilization are needed for inhibition of aggregation.

Both TTR and RBP are synthesized in the choroid plexus and are presumably important for transport of T4 and retinol across the blood–brain and blood–CSF barriers as well as within the CSF and brain interstitial fluids.^44–77 Using reported concentrations in human CSF of TTR (0.1–0.4 μM),^28,29 RBP (6–12 nM),^43,78 and retinol (6–12 nM),^43,79 we estimate that only ~2% of CSF TTR is complexed with RBP.⁴⁰ Thus, hRBP- mediated stabilization of TTR is not a factor on the brain side. Furthermore, RBP levels are reportedly lower in AD.^80,81 Estimates of the concentration of Aβ in these fluids are typically 1–2 nM in CSF and 100-fold lower in plasma (equivalent monomer molar concentrations).^46,82–84 Li et al.⁴⁶ reported an estimated K_d of ~20 μM for TTR tetramers binding to Aβ monomers. Using this value of K_d along with 2 nM Aβ and 0.2 μM TTR, we calculate that only 1% of the Aβ in CSF would be bound to TTR. Thus, it is highly unlikely that binding of Aβ monomers to TTR tetramers could account for any measurable inhibition of Aβ aggregation under physiologically relevant conditions. Alternatively, several lines of evidence support the notion that Aβ oligomers bind more than monomers, and in fact, the stable monomeric mutant mTTR reportedly does not bind Aβ monomers at all^46,52 yet clearly binds Aβ oligomers well.^49,50,52,63 On the basis of estimates of the TTR tetramermonomer equilibrium, ~5% of TTR in CSF (10 nM) is dissociated to monomers.^28,29,51 We calculate that TTR monomers are in molar excess over Aβ in CSF (5-fold excess if Aβ is a monomer, more so if Aβ is oligomerized), so even if the concentration of TTR monomers is only a fraction of that of their TTR tetramer counterpart, it is not negligible. The problem is one of affinity rather than of stoichiometry. We estimated a pseudo K_d of 2 μM for a heterogeneous mix of Aβ40 monomers and oligomers to TTR tetramers (Figure S6 and Table S3) and argue that the affinity for Aβ oligomers and TTR monomers would be stronger. K_d values of 10—100 nM would lead to an estimate of 10—50% Aβ oligomers bound to the TTR monomer. It should be pointed out that TTR has been detected in amyloid plaques in AD,¹⁸ but there is no direct evidence for TTR—Aβ complexes in CSF. Furthermore, TTR is synthesized and secreted by neurons¹⁸ as well as by the choroid plexus. Thus, the biologically relevant site for TTR-mediated protection against Aβ could lie in the tissue and brain interstitial fluid as well as or instead of in the CSF.

Our data strongly support the hypothesis that binding of Aβ to TTR is insufficient for inhibition of aggregation and indicates that destabilization of the TTR tetramer is needed. Furthermore, it is Aβ binding that appears to trigger TTR destabilization. Not yet known is exactly how destabilization works to enhance inhibition of Aβ aggregation. One possibility is that TTR monomeric subunits are released, which then bind to Aβ oligomers. Given the slow release of the monomer in the gel assay (Figure 8) and the close correlation in time between the onset of Aβ fibril formation (Figure 3) and that of tetramer destabilization (Figure 8), this possibility is unlikely. A second possibility is that the tetrameric structure is simply “loosened” by binding of Aβ oligomers, weakening steric constraints on access of Aβ oligomers to the interior cavity. A third is that, by increasing the rate of TTR monomer subunit exchange, Aβ could more readily compete with TTR reassembly.⁴⁹ Because TTR by itself can misfold into amyloid fibrils and TTR destabilization is an early step in its amyloidogenesis pathway,⁸⁵ any TTR-based therapeutic intervention in AD should not proceed via destabilization of TTR other than when it is directly coupled to Aβ binding. A strategy based on developing stable mimics of the TTR monomer is more likely to succeed.

ABBREVIATIONS

Aβ

β-amyloid

Aβ−488

HiLyte Fluor 488-labeled Aβ(1—40)

AD

Alzheimer’s disease

AF594

AlexaFluor 594 C₅ maleimide

APP

amyloid precursor protein

aRBP

apo-retinol-binding protein

CSF

cerebrospinal fluid

DLS

dynamic light scattering

DMF

N,N-dimethylformamide

ELISA

enzyme-linked immunosorbent assay

FRET

fluorescence resonance energy transfer

hRBP

holo-retinol-binding protein

mTTR

engineered double-mutant monomeric transthyretin

NTA

nanoparticle tracking analysis

PBS

phosphate-buffered saline

PBSA

phosphate- buffered saline with azide

RBP

retinol-binding protein

RT

room temperature

SD

standard deviation

SDS—PAGE

sodium dodecyl sulfate—polyacrylamide gel electrophoresis

T4

thyroxine

TEM

transmission electron microscopy

ThT

thioflavin T

TMB

3,3′,5,5′-tetramethylbenzidine

TS-FRET

time-resolved fluorescence resonance energy transfer

TTR

transthyretin

TTR-594

TTR tetramer conjugated with AlexaFluor 594

wt

wild type