Thermodynamic stability and aggregation kinetics of EF helix and EF loop variants of transthyretin

Abstract

Misfolding and aggregation of transthyretin (TTR) is linked to amyloid disease. Amyloidosis occurs when the TTR homotetramer dissociates into aggregation-prone monomers that self-assemble into amyloid. In familial transthyretin amyloidosis, hereditary amino acid substitutions destabilize TTR and promote aggregation. In the present work, we used ¹⁹F-NMR to determine the effect of mutations, in the EF helix (Y78F, K80D, K80E, and A81T) and EF loop (G83R and I84S) on the aggregation kinetics and stability of the TTR tetramer and monomer. The EF region acts as a scaffold that stabilizes interactions in both the strong and weak dimer interfaces of the tetramer and is the site of a cluster of pathogenic mutations. K80D and K80E are non-natural mutants that destabilize the EF helix and yield an equilibrium mixture of tetramer and monomer at neutral pH, providing a unique opportunity to determine the thermodynamic parameters for tetramer assembly under non-denaturing conditions. Of the pathogenic mutants studied, only A81T formed appreciable monomer at neutral pH. Real-time ¹⁹F NMR measurements showed that the pathogenic Y78F mutation accelerates aggregation by destabilizing both the tetrameric and monomeric species. The pathogenic mutations A81T, G83R, and I84S destabilize the monomer and increase its aggregation rate by disrupting a Schellman helix C-capping motif. These studies provide new insights into the mechanism by which relatively subtle mutations that affect tetramer or monomer stability promote entry of TTR into the dissociation-aggregation pathway.

Graphical Abstract

Introduction

Transthyretin is a homotetrameric protein secreted into the blood and the cerebrospinal fluid (CSF) where it acts, respectively, as the secondary and primary carrier of thyroxin.¹ Misfolding and aggregation of TTR is linked to amyloid diseases such as familial associated polyneuropathy and cardiomyopathy (FAP and FAC) and senile systemic amyloidosis (SSA; now called wild-type transthyretin amyloidosis, (ATTRwt). In all cases amyloidosis occurs in two steps. Firstly, in the rate limiting step, the tetramer dissociates into an aggregation-prone monomer,^2–7 which then misfolds and aggregates via a downhill polymerization pathway,⁸ to form oligomers and amyloid. In ATTRwt, it is the wild type (WT) protein that undergoes amyloidosis, a disease that may affect up to 20% of people over the age of 80.⁹ In FAP and FAC, hereditary amino acid point mutations are responsible for destabilizing the tetramer and typically result in an earlier onset of disease than ATTRwt,¹⁰ depending on the denaturation energetics of the mutation.¹¹ Many of these point mutants exhibit tissue-selective depositions and pathologies.¹²

The TTR tetramer consists of a dimer of dimers. Each protomer contains 8 β-strands, designated A-H, that are arranged in two β-sheets (formed from strands DAGH and CBEF) that form a β-sandwich,^13,14 shown schematically together with the amino acid sequence in Figure S1. Hydrogen bonds between the H and H’ strands of adjacent protomers form the strong dimer interface (Figure S1C). Interactions between the F and F’ strands also contribute to this interface. Two strong dimers pack to form the tetramer via hydrophobic and hydrogen bonding interactions involving the AB and GH loops and the C-terminal region of strand H. This interface is referred to as the weak dimer interface (Figure S1D). The importance of hydrophobic packing for TTR tetramer formation and stability is demonstrated by the observation that the subunit exchange rates increase at lower temperatures.¹⁵

Dissociation of the tetramer is necessary but not sufficient to drive TTR amyloidosis; partial unfolding of the monomer is also a prerequisite for progression along the aggregation pathway.^7,16 The factors that destabilize the tetramer and lead to formation of amyloidogenic monomers are only partially understood. Dissociation of the tetramer and entry into the aggregation pathway can be initiated in vitro at mildly acidic pH.^2,6 An X-ray structure of the WT TTR tetramer at pH 4.0 revealed structural perturbations in the EF helix (residues D74-L82) and EF loop (residues G83-E89) relative to the pH 7 structure.¹⁷ This region is of particular interest since it acts as a scaffold that stabilizes interactions that form both the strong and weak dimer interfaces, has been implicated in amyloidogenesis, and is the locus of multiple pathogenic mutations (D74H, S77F/Y, Y78F, A81T/V, G83R, I84N/S/T). Maximum amyloid formation of WT TTR occurs at pH 4.4,³ and NMR resonances of residues in the EF helix of the TTR monomer are broadened at pH 4.4, indicating increased conformational fluctuations under conditions where fibril formation is maximal.¹⁸ The EF helix undergoes structural changes in the amyloid state, refolding to form two short β-strands in patient-derived fibrils.^19,20 Previous studies have shown that TTR is highly sensitive to subtle structural changes in the EF region; substitution of W79 by 6-fluorotryptophan destabilizes the tetramer and increases its aggregation propensity, even though the fluorine atom has an atomic radius only 0.3 Å larger than the hydrogen it replaces.²¹

In the present work, we used our recently-reported ¹⁹F NMR probe²² to obtain new insights into the effect of pathogenic and non-natural mutations in the EF loop and helix on TTR tetramer stability and aggregation kinetics. Disease-causing mutations were introduced into the EF helix (Y78F and A81T) and EF loop (G83R and I84S). To probe the role of the EF helix in stabilizing the TTR tetramer, we introduced the non-natural, helix disrupting mutation K80D,²³ with the less disruptive mutation K80E as a control. K80 is fully solvent exposed (Figure 1) so that mutations at this site would not be expected to disrupt tertiary contacts within the TTR structure. We hypothesized that K80 helps stabilize the helix through favorable interactions with the negative end of the helix dipole. Introduction of a charge reversal at this site provides an opportunity to test the effects of subtle changes in EF helix stability on the stability and aggregation kinetics of the TTR tetramer.

Figure 1:

Overall architecture of homotetrameric human TTR (PDBID 2ROX²⁴). The four subunits are shown in different colors. The EF helix and loop are highlighted in magenta on the yellow protomer. The top left panel shows the hydrogen bond between the side chains of Y78 and D18 in the AB loop. The top right panel shows the location of K80 at the C-terminal end of the helix with the side chain fully solvent exposed. The bottom left panel shows the Schellman motif residues (labeled using the notation of Aurora et. al.²⁵). Hydrophobic interactions between W79 (C3) and I84 (C”) are important for helix stability. The bottom right panel shows the location of naturally occurring pathogenic mutations in the EF helix and loop.

Materials and Methods:

Protein expression, purification, and labelling

The C10S/S85C variant of human TTR²² was used as the template for all subsequent mutations discussed in this paper. All constructs were encoded in a pET29a plasmid that includes an N-terminal methionine. Mutants were generated using the QuikChange kit (Agilent), according to the manufacturer’s instructions. Proteins were expressed in E. coli BL21 Star (DE3) in minimal medium for isotopic labeling and purified by a previously described method²² except that dithiothreitol (DTT) was used instead of tris(2-carboxyethyl)phosphine (TCEP), the medium contained 30 μg.ml⁻¹ kanamycin, and LB medium was used for samples that did not require isotopic enrichment. ¹⁵N labeled samples for NMR chemical shift perturbation (CSP) analysis were grown in M9 media with 1g/L (¹⁵NH₄)₂SO₄, and 3g/L ¹²C-glucose.

Purified TTR samples were labeled with 3-bromo-1,1,1-trifluoroacetone (BTFA) as described previously.²² The various BTFA-labeled TTR constructs are designated using the notation TTR^F (shorthand for C10S/S85C-BTFA), K80D^F (C10S/ K80D/S85C-BTFA), etc. Unless otherwise noted, NMR samples were prepared in buffer containing 10 mM potassium phosphate, 100 mM KCl, at pH 7.0 (NMR buffer), and 10% D₂O. The NMR buffer was extensively degassed with argon and NMR tubes were purged with argon before use. Spectra in low salt solution were acquired in buffer containing 10 mM potassium phosphate, pH 7.0, and 10% D₂O and no added KCl (0 mM KCl buffer).

NMR experiments

¹H, ¹⁵N-TROSY-HSQC spectra²⁶ were collected on a Bruker Avance 900 spectrometer equipped with a 5mm room temperature TXI probe with a triple axis gradient. Spectra were acquired with 2K complex points and 256 increments. The ¹⁵N carrier offset was at 118 ppm, the ¹H carrier offset was at 4.7 ppm, spectral width 16 ppm, and the interscan relaxation delay was 1s. One-dimensional ¹⁹F NMR spectra were recorded on Bruker Avance 601 or Avance 700 spectrometers equipped with ¹H/¹⁹F-¹³C/¹⁵N cryoprobes with z-gradients. Spectra were acquired with 4K complex points using a ¹⁹F pulse length of 10 μs, sweep width of 40 ppm, carrier offset at −84.2 ppm, and interscan relaxation delay of 1 s. At sample concentrations of 1.25 μM, the relaxation delay was shortened to 0.4 s to decrease the time required to acquire a spectrum. All NMR data were acquired using XWinNMR/TopSpin NMR, processed using NMRPipe,²⁷ and analyzed and visualized by MATLAB and Sparky.²⁸ For 1D ¹⁹F-NMR data processing, a 1-Hz exponential line-broadening factor was applied to the free induction decay which was then zero-filled to 16k before Fourier transformation. Unless otherwise stated, populations were calculated using peak areas and all uncertainties in NMR populations were calculated as one standard deviation of 50 bootstrapped data sets.

¹⁹F-NMR aggregation assay.

Real-time ¹⁹F-NMR aggregation assays were performed at 10 μM protomer concentration following a previously reported protocol.²² Briefly, 50 μl of a 100 μM solution of TTR^F mutant in NMR buffer was mixed with 50μl D₂O followed by 400μl 50 mM sodium acetate, 100 mM KCl at pH 4.4 to initiate aggregation. Corrections were made for the measurement dead time (<10 min, required for mixing, matching, tuning, and shimming the probe), so that t=0 in the kinetics data analyses corresponds to the time when the pH was lowered from 7.0 to 4.4. A series of 1D ¹⁹F NMR spectra was acquired to determine the time course of aggregation. Aggregation was also assayed by monitoring changes in optical density at 330 nm (OD₃₃₀) according to.²²

Fitting tetramer dissociation data

Concentration-dependent dissociation of the K80D^F and K80E^F tetramers was monitored by ¹⁹F NMR. The equilibrium constant for the T ⇌ 4M dissociation is:

$$K_{\text{d}}=\frac{{\left[M\right]}^4}{\left[T\right]}$$

By mass conversion we have:

$$4\left[T\right]+\left[M\right]=c_{\mathrm{t}}$$

where c_t is the total concentration of TTR, expressed as the monomer concentration and quantified by an extinction coefficient of 18450 M⁻¹ cm^-1.

We then define the monomer molar fraction f_M = [M]/c_t, which is equivalent to the ¹⁹F peak area of monomer relative to the total ¹⁹F peak areas. By rearranging these three equations, we arrive at:

$$\frac{4}{K_{\text{d}}}{f_{\mathrm{M}}}^4+\frac{1}{c_{\mathrm{t}}^3}{f}_{\mathrm{M}}-\frac{1}{c_{\mathrm{t}}^3}=0$$

This quartic equation was numerically solved by the roots function and the residuals were minimized by the fminsearch function in MATLAB. The error was calculated as 1 standard deviation from 50 bootstrapped datasets.

Results

Effect of Helix Destabilizing Mutations on Tetramer Stability

To destabilize the EF helix, the solvent-exposed K80 was replaced by aspartate, a known helix breaker. The K80D mutation is expected to destabilize the helix both through interactions of the short carboxyl side chain with the helix backbone and, by virtue of its location on the last turn of the helix (Figure 1), through unfavorable interactions with the negative end of the helix dipole.²³ As a control, K80 was also replaced by Glu, a helix favoring amino acid; Glu at this site would also interact unfavorably with the helix dipole but has the potential to form a stabilizing salt bridge with K76.

To determine the effect of the K80D/K80E mutations, we collected ¹⁹F NMR spectra of the BTFA-labeled constructs K80D^F and K80E^F in NMR buffer (pH7.0, 100mM KCl) at 298K, with [TTR]_total = 25 μM and 1.25 μM, in the typical concentration ranges of TTR in the blood and in the CSF, respectively²⁹ (Figure 2A). When [TTR]_total = 25 μM, K80D^F and K80E^F have 5.0% and 3.0% population of monomer, respectively. However, when the concentration is lowered to 1.25 μM, the population of monomer increases to 40% and 20% for K80D^F and K80E^F, respectively. No monomer is detectable for TTR^F at these concentrations. Thus, even a relatively conservative K80E substitution, which changes the charge at the C-terminus without affecting the stability of the helix, results in thermodynamic destabilization of the tetramer.

Figure 2:

(A) ¹⁹F-NMR spectra of BTFA labelled K80D^F (blue), K80E^F (red) and TTR^F (black) at [TTR]_total = 1.25 μM or 25 μM in NMR buffer (pH 7.0 at 298K). Intensity is normalized to the highest peak at the stated concentration.

(B) and (C) Changes in tetramer population as a function of the protomer concentration at 298 K (see Table 1). Data are shown for K80D^F(B) and K80E^F(C) in NMR buffer and in 10 mM potassium phosphate buffer, pH 7.0, with 0 mM KCl

Measurement of Tetramer Stability by ¹⁹F NMR

The stability of the K80D^F and K80E^F tetramers was determined quantitatively by monitoring the tetramer and monomer populations in ¹⁹F NMR spectra recorded over a range of TTR concentrations (Figure 2B, C). Only tetramer and monomer resonances were observed; no peaks that could potentially arise from a dimeric intermediate were observed. Given the location of the BTFA probe near both the strong and weak dimer interfaces, a new ¹⁹F resonance would be expected if appreciable quantities of dimer accumulated. Spectra were acquired from samples in buffer containing 100 mM or 0 mM KCl at 298K and the populations of tetramer and monomer were determined from peak areas. The concentration dependence of the tetramer and monomer populations was fitted to two different dissociation models. In the first model, it was assumed that tetramer dissociates to dimer which immediately dissociates into monomer.⁵ The data were fitted to a single K_d for the equilibrium T ⇌ 2D, where [D] = [M]/2. This model does not fit the experimental ¹⁹F data well in the physiological concentration range (Figure S2). In the second model, the NMR data were fitted to a T ⇌ 4M equilibrium, as described in Materials and Methods. This model fitted the data with the lowest root mean squared error (fits shown in Figure 2B, C), giving K_d values of 1.3×10⁻¹⁹ M³ and 2.0×10⁻²⁰ M³ for the K80D^F and K80E^F tetramers, respectively, in NMR buffer at 298K (Table 1). As expected,¹⁵ lowering the salt concentration of the NMR buffer to 0 mM KCl further destabilized the tetramer and led to an increased population of monomer (blue data points and fitted curves in Figure 2B, C), with K_d values of 1.6×10⁻¹⁵ M³ and 1.2×10⁻¹⁶ M³ for K80D^F and K80E^F respectively.

Table 1.

Tetramer dissociation constants and thermodynamic parameters determined from concentration dependence and temperature dependence of ¹⁹F NMR spectra.

Data source	Parameter	WTe	K80DF	K80DF	K80EF	K80EF
[KCl]			100 mM	0 mM	100 mM	0 mM
Concentration Titrationa	Kd (M3)	9 × 10−25	1.3 × 10−19	1.6 × 10−15	2.0 × 10−20	1.2 × 10−16
	ΔG (kcal.mol−1)	32.8	25.7	20.1	26.8	21.6
van’t Hoff plotb	ΔH (kcal.mol−1)		−16.6 ± 0.3	−17 ± 2	−18 ± 2	−17 ± 1
	TΔS (kcal.mol−1)c		−41.8 ± 0.3	−37 ± 2	−44 ± 2	−39 ± 1
	ΔG (kcal.mol−1)d		25.2	20.3	25.9	21.8

^aConcentration titration performed in NMR Buffer (10mM sodium phosphate (pH7.0), 100mM KCl) at 298K.

^bUncertainties in ΔH and TΔS are standard deviations from 3 measurements except for K80D^F (100mM KCl) where the uncertainty is the standard error of two measurements.

^cTΔS at 298K

^dΔG = ΔH - TΔS

^eValues from ³⁰ measured from urea denaturation experiments.

Temperature Dependence of Tetramer Population

Since TTR folding and tetramer assembly are heavily dependent on the hydrophobic effect, we predicted that the population of tetramer would increase when the temperature is raised. Figure S3A shows the population of tetramer as a function of temperature for both K80D^F and K80E^F, for samples at [TTR]_total = 30 μM in NMR buffer containing 0 mM or 100 mM KCl. In all cases the population of monomer increased as the temperature decreased. In the absence of salt, both mutant proteins were destabilized and were more sensitive to changes in temperature. In terms of stability: K80E^F(salt)>K80D^F(salt)>K80E^F(no salt)>K80D^F(no salt).

A van’t Hoff plot indicates that formation of monomer from tetramer is an exothermic process and is entropically disfavored (Figure S3B). The linearity of this plot suggests that enthalpy and entropy are constant over the temperature range (277 – 310K) studied. The thermodynamic parameters determined from the van’t Hoff plot are summarized in Table 1. Overall, a decrease in temperature favors the exothermic process of dissociation to monomers, and lowers its entropic cost (less negative TΔS), in accord with the well-known destabilization of the TTR tetramer at lower temperatures.¹⁵

Structural Perturbations Caused by K80D/E Mutations

To obtain insights into potential structural perturbations that may lead to tetramer destabilization, ¹H,¹⁵N -HSQC spectra of K80D^F and K80E^F were acquired (Figures S4 and S5) to determine backbone amide chemical shift perturbations (CSP) relative to the parent TTR^F (Figure S6). Changes in chemical shift are generally very small, indicating that the overall TTR structure is not significantly perturbed. The largest CSPs are in the immediate vicinity of the mutation site, revealing subtle changes in structure or hydrogen bonding at the C-terminal end of the helix and beginning of the EF loop, with the Asp substitution causing slightly larger chemical shift changes than Glu. No significant chemical shift changes are observed at distant sites, indicating that the effects of the K80D and K80E substitutions are entirely local.

Effect of pathogenic mutations in EF helix and loop

While K80D and K80E provide new insights into the importance of the helix in determining tetramer stability, these mutants do not occur naturally. We therefore generated four TTR^F variants, Y78F^F, A81T^F, G83R^F, and I84S^F, with naturally-occurring pathogenic mutations in the EF helix and loop. ¹⁹F NMR spectra at [TTR]_total = 25.0 μM and 1.25 μM are shown in Figure 3. With the exception of A81T^F, which contained 17% population of monomer at 1.25 μM, the variant TTRs were fully tetrameric at both concentrations.

Figure 3.

¹⁹F spectra of pathogenic EF helix mutants at 298K in pH 7.0 NMR buffer containing 100 mM KCl. Spectra were recorded at 25 μM (upper panel) and 1.25 μM (lower panel) TTR concentration. The scale of each spectrum is normalized to the highest peak at the stated concentration.

Kinetics of aggregation

Since the ¹⁹F NMR spectra show that the K80D^F and K80E^F tetramers are destabilized relative to the parent TTR^F and spontaneously form 20 – 40% monomer upon dilution to 1.25 μM, we investigated their aggregation kinetics using our recently developed ¹⁹F-NMR aggregation assay²² and turbidity measurements at 330 nm. Our expectation was that the mutants would exhibit increased aggregation rates since the tetrameric structure has been destabilized. We also used ¹⁹F NMR to measure the aggregation kinetics of the pathogenic Y78F^F, A81T^F, G83R^F, and I84S^F variants.

For each TTR variant, aggregation was initiated by diluting the protein to 10 μM protomer concentration in pH 4.4 buffer. Aggregation was then monitored by real-time ¹⁹F NMR, as illustrated in Figure 4B and S7A for experiments performed at 310K and Figure 4A and S7B for those performed at 298K.

Figure 4.

Monitoring aggregation kinetics of A81T^F at pH 4.4 (A) at 298K and (B) at 310K, using ¹⁹F-NMR and OD₃₃₀ measurements. The left panels show the overlaid time series of ¹⁹F-NMR spectra with raw peak heights. The right panels show normalized ¹⁹F-NMR and OD₃₃₀ data. The ¹⁹F-NMR peak areas (left black axis) and OD₃₃₀ (right gold axis) are plotted as a function of aggregation time. The data points are colored blue for tetramer (T). At 298K, growth and decay of a well-separated monomer peak (M) is observed (red data points), which is much less populated at 310K, thus not quantified. Green points represent the missing NMR signal amplitude for the ensemble of aggregates (A) and are derived by subtracting normalized T (at 310K) or T+M intensity (at 298K) from 1. The maximal OD₃₃₀ is normalized to the maximum of the A signal for convenience of comparison. The black lines in (A) show the fits of the NMR kinetic data based on the two-step reversible kinetic scheme while those in (B) are single exponential fits. The orange connecting lines in (A) are for guidance of the eye only and those in (B) are single exponential fits to the OD signal. Kinetic data are shown in Table 2. Data for the remaining 5 mutants are shown in Figure S7.

The total ¹⁹F signal intensity decreases with time due to formation of aggregated species that cannot be observed by NMR. At 298K the time-dependent changes in ¹⁹F peak area were fitted to a reversible two-step kinetic scheme:

$$\text{T}\underset{k_{-1}}{\overset{k_1}{\rightleftarrows }}\text{M}\underset{k_{-2}}{\overset{k_2}{\rightleftarrows }}\text{A}$$

where T denotes tetramer, M is the monomer, and A represents an ensemble of NMR-invisible aggregated species, as described in detail elsewhere.²² The fitted kinetic parameters are summarized in Table 2. At 310K all variants were fitted to a single exponential rate, k_slow (Table 2). Although small amounts of monomer were observed for K80D^F, K80E^F, and A81T^F, the populations were too small for accurate quantitation and the data were fitted to the dominant single exponential decay.

Table 2:

Aggregation kinetics determined by real-time ¹⁹F NMR.

Mutant	Temp (K)	k1 (h−1)	k−1 (h−1)	k2 (h−1)	k−2 (h−1)	γ2b (h−1)
TTRF (a)	298	0.13 ± 0.02	0.63 ± 0.10	0.06 ± 0.01	0.01 ± 0.01	0.02 ± 0.01
	310	γ2 = 0.06 ± 0.01
K80DF	298	0.14 ± 0.01	0.8 ± 0.1	0.44 ± 0.04	0.03 ± 0.01	0.07 ± 0.01
	310c	kslow = 0.14 ± 0.01
K80EF	298	0.17 ± 0.08	0.7 ± 0.3	0.05 ± 0.01	0.01 ± 0.01	0.02 ± 0.01
	310c	kslow = 0.06 ± 0.01
Y78FF	298	0.49 ± 0.07	0.7 ± 0.1	0.64 ± 0.03	0.05 ± 0.01	0.23 ± 0.01
	310c	kslow = 0.30 ± 0.02
A81TF	298	0.12 ± 0.04	0.23 ± 0.09	0.15 ± 0.01	0.01 ± 0.01	0.05 ± 0.01
	310c	kslow = 0.06 ± 0.01
G83RF	298	0.14 ± 0.02	0.8 ± 0.2	0.39 ± 0.07	0.01 ±0.01	0.06 ± 0.01
	310c	kslow = 0.06 ± 0.01
I84SF	298	0.20 ± 0.04	0.6 ± 0.2	0.39 ± 0.02	0.02 ± 0.01	0.09 ± 0.01
	310c	kslow = 0.08 ± 0.01

^aData for TTR^F from ²²

^bγ₂ is the slower relaxation rate defined in ²² and represents the rate of increase of A in the aggregation scheme used here.

^cThe population of monomer intermediate is small or unobservable and the decay of the tetramer peak was fitted to a single exponential with rate k_slow which approximates the relaxation rate γ₂.²²

Discussion

Previous studies have established that entry into and progression along the transthyretin aggregation pathway requires both dissociation of the tetramer and partial unfolding of the resulting monomer.^2–5,7,16 More than 120 single-site mutations that are associated with familial amyloid disease have been identified (http://amyloidosismutations.com/mut-attr.php), but insights into the mechanisms by which these mutations drive amyloidosis are mostly lacking. Elucidation of the general molecular determinants of TTR tetramer and monomer stability, and how stability is modulated by pathogenic mutations, is therefore of great interest and importance. In the present work, we have examined the effects of pathogenic and non-natural mutations in the EF helix and EF loop, an area distant from the subunit interfaces, on the stability and aggregation kinetics of TTR. Our results reveal that the TTR tetramer is poised on the threshold of stability, such that even a modest charge reversal, K80E, near the C-terminal end of the EF helix is sufficient to thermodynamically destabilize the tetramer.

The K80E substitution, at a site that is one turn from the C-terminus of the EF helix, has an effect on helix stability that varies with the side chain charge. At pH 7, where the Glu carboxyl is deprotonated, the K80E mutation destabilizes the helix by placing a negative charge close to the negative end of the helix macrodipole.³¹ As a consequence, the tetramer is destabilized and dissociates to form a substantial population of monomer as the K80E^F concentration is decreased (Figure 2C). The monomer population is dependent upon the salt concentration and is decreased in the presence of 100 mM KCl, which partially screens the unfavorable charge-helix dipole interaction and stabilizes the tetramer by screening electrostatic repulsion between K15 side chains on opposing subunits.¹⁵ The K80D substitution is even more destabilizing than K80E due to the proximity of the carboxylate group to the helix backbone²³ and, under the same conditions, a higher monomer population is observed for K80D^F (Figure 2B). The pK_as of Asp and Glu side chains in the C-terminal region of a helix are increased by interactions with the negative end of the helix dipole.^23,31 At pH 4.4, where the E80 and D80 side chains are likely to be fully or partially protonated, the destabilizing charge-dipole interactions are relieved and the apparent tetramer dissociation constants (= k₁/k_-1, Table 2) of K80E^F and K80D^F are similar to that of the parent TTR^F (0.26, 0.19 and 0.21, respectively).

The K80D and K80E mutations provide a unique opportunity to directly determine thermodynamic parameters for the tetramer – monomer equilibrium under non-denaturing conditions using ¹⁹F NMR. To date, the stability of WT and mutant TTR tetramers could be determined only under denaturing conditions, by fitting concentration-dependent urea denaturation data or by denaturing the tetramer at high pressure and low temperature.^30,32 The high sensitivity of the ¹⁹F NMR spectrum allows us to directly measure the relative populations of tetramer and monomer over a range of K80D^F and K80E^F concentrations from 1 – 300 μM (Figure 2). In buffer containing 100 mM KCl, the K80D^F and K80E^F tetramers are 7.0 and 5.9 kcal.mol⁻¹ less stable than the WT TTR tetramer, respectively (Table 1). By measuring the temperature dependence of the K80D^F and K80E^F NMR spectra, we obtained the first insights into the thermodynamics of tetramer formation (Table 1). Tetramer formation is enthalpically unfavorable but is strongly favored by the product of temperature and entropy (ΔH = 16.6 kcal.mol⁻¹, TΔS = 41.8 kcal.mol⁻¹ for formation of the K80D^F tetramer in 100 mM KCl at 298K), highlighting the importance of hydrophobic packing in stabilization of the tetramer at high temperatures compared to low temperatures.^15,33 The large increase in entropy upon association of the subunits is probably associated with release of hydration water from the exposed hydrophobic surfaces of the monomers. The well-known effect of salt in stabilization of the TTR tetramer is seen to be entirely entropy driven: increasing the KCl concentration from 0 to 100 mM increases TΔS for tetramer formation by ~5 kcal.mol⁻¹ for both K80D^F and K80E^F while there is no change in enthalpy within the experimental uncertainties (Table 1). These results are in accord with molecular dynamics simulations which show that high-charge-density ions enhance hydrophobic interactions between surfaces by a purely entropic effect.³⁴

Aggregation kinetics at pH 4.4 and 310K were measured both by real-time ¹⁹F NMR and by optical density (Figure 4B and S7A). At this temperature, loss of intensity of the tetramer peak and increase in OD₃₃₀ is monophasic and dominated by the slow step in the aggregation pathway (Table 2). At 298K, however, formation of a transient monomeric intermediate can be observed directly by ¹⁹F NMR (Figure 4A and S7B), providing important insights into the detailed mechanisms by which EF helix and loop mutations promote TTR aggregation. Although the TTR tetramer is thermodynamically destabilized by the glutamate side chain of K80E at physiological pH, the aggregation kinetics of K80E^F at pH 4.4, where the Glu carboxyl would be largely protonated, are indistinguishable from those of TTR^F (Table 2). In contrast, K80D^F aggregates more rapidly than TTR^F; while both tetramers have similar stability (similar values of k₁ and k_-1) at 298K, the monomeric intermediate is destabilized by the mutation and proceeds more rapidly to NMR-invisible aggregates (k₂ = 0.44 h⁻¹ for K80D^F versus 0.06 h⁻¹ for TTR^F) (Table 2). The behavior of the K80E and K80D substitutions at low pH can be understood from their differing effects on helix stability;³⁵ the uncharged Glu side chain is helix stabilizing while Asp, even in its uncharged state, is helix destabilizing and promotes aggregation of the monomer.

The EF helix (residues T75 – L82) and EF loop (residues G83 – E89) are the locus of a cluster of pathogenic mutations (http://amyloidosismutations.com/mut-attr.php). We used real-time ¹⁹F NMR to elucidate the mechanism by which representative mutations in this region destabilize TTR, accelerate aggregation, and cause amyloid disease. Only one of the studied pathogenic variants (A81T) led to substantial thermodynamic destabilization of the tetramer at physiological pH in 100 mM KCl, exhibiting a 17% population of monomer at 1.25 μM concentration (Figure 3). However, at pH 4.4 and 298K, all of the variants aggregate via a two-step pathway involving a monomeric intermediate (Figure 4A and S7B).

The Y78F mutant is highly amyloidogenic and is associated with peripheral neuropathy and cardiomyopathy.^36,37 In WT TTR, Y78 helps anchor the EF helix to the TTR core by way of hydrophobic interactions and a hydrogen bond between the tyrosine hydroxyl group and the carboxyl group of D18 in the AB loop (Figure 1). D18 plays a critical role in stabilization of the weak dimer interface; substitution by Gly in the pathogenic D18G variant destabilizes the AB loop, disrupts the weak dimer interface, and renders the protein monomeric.^38,39 The structure of the Y78F variant is almost indistinguishable from that of WT TTR.⁴⁰ However, our measurements of aggregation kinetics by real-time ¹⁹F NMR show that disruption of the hydrogen bond to D18 destabilizes both the tetramer and monomer, increasing the rate of tetramer dissociation (k₁ = 0.49 h⁻¹ for Y78F^F vs 0.13 h⁻¹ for TTR^F at 298K) and of monomer unfolding and aggregation (k₂ = 0.64 h⁻¹ for Y78F^F vs 0.06 h⁻¹ for TTR^F) (Table 2). These observations underscore the importance of the Y78 – D18 hydrogen bond for maintaining the structural integrity of the AB loop and weak dimer interface and for stabilizing the EF helix packing.

At pH 4.4 and 298K, where aggregation proceeds via an observable monomeric intermediate, the A81T^F, G83R^F, and I84S^F tetramers dissociate at similar rates to the parent TTR^F (Table 2). However, for each of these variants, the monomers are destabilized by the mutations and aggregate 2- to 6-fold faster than the TTR^F monomer (k₂ = 0.15 h⁻¹ for A81T^F, 0.39 h⁻¹ for G83R^F, and 0.39 h⁻¹ for I84S^F vs 0.06 h⁻¹ for TTR^F). Thus, these EF loop variants have little effect on the rate of dissociation of the tetramer, but they accelerate oligomerization by destabilization of the monomer. Although the A81T mutation does not change the rate of tetramer dissociation, it does decrease the rate of re-association of the monomer (k_-1 = 0.23 h⁻¹ for A81T^F vs 0.63 h⁻¹ for TTR^F), consistent with the observed destabilization of the A81T^F tetramer at pH 7 (Figure 3).

The mechanism by which pathogenic mutations in the EF loop promote aggregation can be understood in terms of their effects on capping of the EF helix. The EF loop functions as a classic Schellman helix C-capping motif.²⁵ In the notation of Aurora et al.,²⁵ the C-terminal residues of the helix and capping residues are designated:

$$\mathrm{W}79\left(\mathrm{C}3\right)–\mathrm{K}80\left(\mathrm{C}2\right)–\mathrm{A}81\left(\mathrm{C}1\right)–\mathrm{L}82\left(\mathrm{C}\mathrm{c}\mathrm{a}\mathrm{p}\right)–\mathrm{G}83\left(\mathrm{C}’\right)–\mathrm{I}84\left(\mathrm{C}”\right)$$

where C3, C2, and C1 belong to the EF helix, the Ccap residue is at the boundary between the helix and the EF loop, and C’ and C” are EF loop residues that cap and stabilize the helix (Figure 1). The key sites of the Schellman motif are C3, C’, and C”; substitution of disallowed residues at these sites is highly destabilizing. Hydrophobic interactions between C3 (W79 in TTR) and C”(I84) are critical for helix stabilization;²⁵ the known pathogenic mutations at C” (I84S, I84N, I84T) would abrogate these interactions and thereby destabilize the EF helix. Indeed, the importance of a bulky hydrophobic residue at position 84 is confirmed by X-ray structures of the I84S and I84A variants; at pH 7.5, the structures are virtually identical to WT TTR but at pH 4.6 the EF helix is unwound and the loop conformation rearranged in two subunits of the tetramer.⁴¹ The C’ residue of the Schellman motif must be in the α_L backbone conformation, with a positive φ angle; Gly is strongly conserved at this position²⁵ and substitution of G83 by Arg, for which the α_L conformation is unfavorable, is highly destabilizing and leads to a 6-fold increase in the rate of aggregation of the monomer (k₂ = 0.39 s⁻¹ vs 0.06 s⁻¹ for TTR^F, Table 2). Finally, Ala is strongly favored at position C1 of the helix and substitution by branched side chains in the pathogenic variants A81T and A81V destabilizes the monomer and enhances aggregation.

Conclusions

The present study shows clearly that residues in the EF helix and EF loop are involved in critical interactions that contribute to the structural stability of both the TTR tetramer and monomer and that mutations that perturb these interactions predispose TTR towards aggregation. Mutations that destabilize the EF helix by disrupting packing interactions with the core (Y78F) or by perturbing capping interactions or helical propensity (K80D/E, A81T, G83R, I84S) accelerate monomer unfolding and aggregation. Remodeling of the EF helix and loop is an essential step towards fibril formation^19,20 and mutations that destabilize the native conformation probably increase k₂ by lowering the energy barrier for helix unfolding. Amongst the pathogenic variants studied in the present work, only Y78F, which disrupts a critical hydrogen bond to the AB loop, destabilizes the tetramer directly and increases its rate of dissociation. The present work provides important new insights into the mechanisms by which pathogenic mutations destabilize transthyretin and promote fibrillogenesis and amyloid disease.