TiO₂ Nanoparticles Catalyze Oxidation of Huntingtin Exon 1-Derived Peptides Impeding Aggregation: A Quantitative NMR Study of Binding Kinetics

Abstract

Polyglutamine expansion within the N-terminal region of the huntingtin protein results in the formation of intracellular aggregates responsible for Huntington’s disease, a fatal neurodegenerative condition. The interaction between TiO₂ nanoparticles and huntingtin peptides comprising the N-terminal amphiphilic domain without (htt^NT) or with (htt^NTQ₁₀) a ten-residue C-terminal polyglutamine tract, is investigated by NMR spectroscopy. TiO₂ nanoparticles decrease aggregation of htt^NTQ₁₀ by catalyzing the oxidation of Met⁷ to a sulfoxide, resulting in an aggregation-incompetent peptide. The oxidation agent is hydrogen peroxide generated on the surface of the TiO₂ nanoparticles either by UV irradiation or at low steady-state levels in the dark. The binding kinetics of nonaggregating htt^NT to TiO₂ nanoparticles is characterized by quantitative analysis of ¹⁵N dark state exchange saturation transfer and lifetime line broadening NMR data. Binding involves a sparsely populated intermediate that experiences hindered rotational diffusion relative to the free state. Catalysis of methionine oxidation within the N-terminal domain of the huntingtin protein may potentially provide a strategy for delaying the onset of Huntington’s disease.

Polyglutamine expansion within the N-terminal region of the huntingtin protein (corresponding to exon-1) favors aggregation and is responsible for Huntington’s disease, a fatal neurodegenerative condition.¹ The polyglutamine domain lies downstream of the 16-residue N-terminal amphiphilic domain (htt^NT). Peptides comprising htt^NT with as few as 10 glutamines (htt^NTQ₁₀) aggregate rapidly in solution and form polymorphic fibrils.² We recently observed that oxidation of the Met⁷ side-chain to a sulfoxide (Met⁷) prevents aggregation of httNTQ₁₀.³ In general, adsorption of fibril forming proteins and peptides on the surface of nanoparticles enhances aggregation and fibril formation by increasing the local peptide concentration and hence the probability of forming a critical initiation nucleus.^4,5 Titanium oxide nanoparticles (TiO₂ NPs) are unique as they possess photocatalytic properties that generate reactive oxygen species upon UV irradiation.^6,7 Here we study the interaction of htt^NT and htt^NTQ₁₀ with TiO₂ NPs by NMR, and show that TiO₂ NPs specifically catalyze the oxidation of Met7, thereby preventing fibril formation by reducing the concentration of the aggregation-competent, native reduced form of htt^NTQ₁₀.

Spontaneous aggregation of htt^NTQ₁₀ occurs over time as monitored both by an increase in Thioflavin T (ThT) fluorescence emission (Figure 1A), attributed to the formation of β-rich amyloid-like structures, and by a decrease in the intensity of the amide proton envelope of the NMR signal (Figure 1B; see SI). Reduced and oxidized monomeric htt^NTQ₁₀ are NMR visible, while aggregates of reduced htt^NTQ₁₀ are broadened beyond detection, resulting in a decrease in the observable amide proton envelope intensity. Addition of photoexcited TiO₂ NPs (see SI and Figure S1) dramatically reduces the extent of aggregation. At 10 °C, only 8% of a 300 μM ¹⁵N-htt^NTQ₁₀ sample remains NMR visible after 100 h; in the presence of photoexcited TiO₂ NPs, however, aggregation plateaus out with 33% of the sample remaining NMR visible (Figure 1B). The apparent aggregation t_1/2 (∼23−25 h), however, is not affected by the TiO₂ NPs (Figure 1B). The reduction in the fraction of aggregating species can be attributed to TiO₂ NP catalyzed oxidation of the side chain of Met⁷ to a sulfoxide. Several reactive oxygen species are formed on the surface of TiO₂ NPs upon UV irradiation (Figure S2),^6,7 but only H₂O₂ is stable with significant amounts generated both upon UV irradiation and in the dark (Figures 2A and S3).

Figure 1.

Effect of TiO₂ NPs on aggregation of htt^NTQ₁₀ monitored by NMR at 10 °C. Time course of (A) ThT emission of 25 μM htt^NTQ₁₀ in the absence of TiO₂ NPs, and (B) of the intensity of the amide proton envelope of 300 μM ¹⁵N-labeled htt^NTQ₁₀ in the absence (blue circles) and presence (red circles) of 5 g·L⁻¹ TiO₂ NPs photoexcited by exposure to UV light prior to addition to the peptide solution. The solid lines represent best fits where the disappearance of reduced, NMR visible httNTQ₁₀ (k_agg ∼ 0.03 h−1) due to aggregation competes with oxidation to aggregation-incompetent Met⁷O-htt^NTQ₁₀ (see Scheme 1). The normalized intensities are corrected for sample to sample variations in the amounts of Met⁷O-htt^NTQ₁₀ present at t = 0 (see SI).

Figure 2.

TiO₂-catalyzed oxidation of httNT. (A) Amplex Red assay for H₂O₂ generated by 5 g·L⁻¹ TiO₂ NPs in the dark (blue, ∼4 μM H₂O₂) and upon UV exposure for 3 h (red, NP suspension, ∼76 μM H₂O₂; green, supernatant after removal of NPs, ∼11 μM H₂O₂). (B) Time course of Met⁷ oxidation of 300 μM ¹⁵N-labeled htt^NT following addition of (B) 100 μM H₂O₂ and (C) UV-irradiated TiO₂ NPs (5 g·L⁻¹, 3 h UV exposure)monitored by the reduction and corresponding increase in intensities of Ala⁹ and Phe¹⁰ cross-peaks arising from reduced (blue) and Met⁷O oxidized (red) htt^NT, respectively, in a series of ¹H−15N HSQC spectra. The insets in (C) show the cross-peaks corresponding to the reduced (upfield) and oxidized (downfield) states at t = 0 (purple) and 80 (green) hours. The experimental data in panels B and C are shown as circles and the best-fit curves obtained by nonlinear optimization and integration of the differential equations (eq S1) describing the oxidation process in Scheme 1 are represented by solid lines. The different ratios of oxidized to reduced htt^NT at time zero in panels (B) and (C) reflect sample to sample variations. Data were collected at 10 °C and a spectrometer frequency of 600 MHz.

The oxidation kinetics of htt^NTQ_n huntingtin peptides (where n = number of glutamines in the polyglutamine tract) in the presence of TiO₂ NPs can be described by three parallel reactions (Scheme 1): two second-order processes involving oxidation of httNTQ_n (P_red) to Met⁷O-httNTQ_n (P_oxi) by H₂O₂, generated upon UV irradiation, either dissolved in solution (k₁) or on the surface of the TiO₂ NPs (k₂), and a pseudo-first-order process $\left(k_3^{\text{dark}}=k_2{\left[{\text{H}}_2{\text{O}}_2\right]}_{\text{dark}}\right)$ occurring in the dark that involves a low steady-state (i.e., continuously generated) level of H₂O₂ on the surface of the TiO₂ NPs.⁸ These reactions occur concomitantly with aggregation of P_red which, for simplicity, is described as a unimolecular process with a rate constant k_agg. Our treatment of the kinetics of all oxidative processes assumes that oxidation proceeds on a time-scale much slower than that of binding to TiO₂ NPs, as is amply confirmed experimentally (see below).

Scheme 1.

Parallel Reactions Describing Aggregation and TiO₂ NP-Catalyzed Oxidation of htt^NTQ_n Peptides

To study the kinetics of TiO₂ catalyzed oxidation of huntingtin peptides in the absence of aggregation (k_agg = 0 in Scheme 1), we made use of a peptide comprising only the N-terminal amphiphilic domain, htt^NT, which remains stable in the presence of TiO₂ NPs for many days (Figure S4). Oxidation of htt^NT by H₂O₂ free in solution makes an insignificant contribution to the kinetics of htt^NT oxidation in the presence of 5 g·L⁻¹ TiO₂ NPs as the concentration of dissolved H₂O₂ generated upon UV irradiation (∼11 μM, Figure 2A) is 10-fold lower than that used in the experiment with added H₂O₂ shown in Figure 2B. The time course of oxidation in the presence of UV irradiated TiO₂ is biphasic (Figure 2C). The first phase arises from second-order (k₂ ∼ 500 M⁻¹ h⁻¹) oxidation of Met⁷ by the substantial amount of H₂O₂ located on the surface of the TiO₂ NPs generated by UV irradiation. Oxidation on the TiO NP surface is accelerated ∼45-fold relative to that free in solution. Once H₂O₂ generated by UV irradiation is consumed, a slower apparent first order oxidation process $\left(k_3^{\text{dark}}=k_2{\left[{\text{H}}_2{\text{O}}_2\right]}_{\text{dark}}~0.002{\text{h}}^{-1}\right)$ occurs owing to the steady-state level of H₂O₂ present on the TiO₂ NP surface in the dark. With these values of k₂ and $k_3^{\text{dark}}$, Scheme 1 quantitatively accounts for the disappearance of NMR visible htt^NTQ₁₀ in the presence of photoactivated TiO₂ NPs with k_agg ∼ 0.03 h⁻¹ which remains unchanged in the absence of TiO₂ NPs (Figure 1B).

The efficiency of heterogeneous catalysis is dependent upon the strength of interaction between the adsorbate (htt^NT) and the surface of the catalyst.⁹ We therefore characterized the kinetic and mechanistic details of ¹⁵N-labeled htt^NT adsorption on the surface of TiO₂ NPs using ¹⁵N dark state exchange saturation transfer (DEST) and lifetime line broadening (ΔR₂) which enable one to quantitatively analyze exchange processes between an NMR visible species and very high molecular weight, NMR invisible “dark” states.^10–13 Although binding of httNTQ₁₀ to TiO₂ NPs cannot be studied quantitatively owing to httNTQ₁₀ aggregation during the time course of the DEST experiment (several days), ΔR₂ measurements are sufficiently fast (a few hours) to demonstrate significant ¹⁵N lifetime line broadening (∼10 s−1), and hence binding, of httNTQ₁₀ in the presence of TiO NPs (Figure S5).

Examples of ¹⁵N-DEST profiles and a plot of ¹⁵N-ΔR₂ as a function of residue for reduced htt^NT in the presence of TiO₂ NPs in the dark are shown in Figure 3A,B, respectively. These conditions correspond to a quasi-equilibrium state where Met⁷ oxidation by the small amount of steady-state H₂O₂ generated on the TiO₂ NP surface in the dark proceeds very slowly (t_1/2 ∼ 350 h). The shapes of the ¹⁵N-DEST profiles are unusual and characterized by a narrow component superimposed on a broad one (Figure 3A) indicative of a three-state exchanging system in which the slow exchange process due to htt^NT binding to the TiO₂ NP surface is accompanied by a second exchange process occurring on a much faster time scale. The ¹⁵N-DEST and R₂ data were fit simultaneously to the three-state exchange model shown in red in Figure 3B via propagation of a set of Bloch-McConnell¹⁴ equations.

Figure 3.

Kinetics of htt^NT binding to TiO₂ NPs characterized by relaxation-based NMR. (A) Examples of ¹⁵N-DEST profiles and (B) ¹⁵N-ΔR₂ as a function of residue observed for 300 μM ¹⁵N-labeled htt^NT in the presence of 5 g·L⁻¹ TiO₂ NPs in the dark. Circles represent experimental data, and solid lines are best fits to a three-state exchange model. Residues requiring a separate treatment are shown in gray. Data were recorded at 600 MHz and 10 °C.

Exchange between free $\left({\text{P}}_{\text{F}}^{\text{red}}\right)$ and TiO₂ NP-bound $\left({\text{P}}_{\text{B}}^{\text{red}}\right)$ htt^NT monomer proceeds via an intermediate $\left({\text{P}}_{\text{R}}^{\text{red}}\right)$ whose rotational diffusion is only partially restricted, presumably due to initial binding to the hydration layer surrounding the TiO₂ NP; the final bound state experiences the same rotational diffusion as the TiO₂ NP itself (Figure 4). This binding mechanism is similar to that observed for the adsorption of cholic acid to ceria NPs.¹⁵

Figure 4.

Overall kinetic scheme for the binding htt^NT to TiO₂ NPs coupled with oxidation of htt^NT to the Met⁷ sulfoxide form either in solution or on the NP surface. States in contact with TiO₂ NPs are shaded.

The population $\left(p_{\text{B}}^{\text{red}}=1.7\pm 0.1\%\right)$ of TiO₂ NP-bound reduced htt^NT, the rate constant k_BR (97 ± 5 s⁻¹) for the transition from fully bound to partially restricted states, and the transverse relaxation rates in the bound state $\left({<}^{15}\text{N-}{R}_2^{\text{bound}}>=31000\pm 300{\text{s}}^{-1}\right)$ are well-defined from the fits. The population of the intermediate state $\left(p_{\text{R}}^{\text{red}}\right)$, however, cannot be determined with certainty without assumptions regarding the magnitude of its transverse spin relaxation rate, ${}^{15}\text{N-}{R}_2^{\text{restricted}}$. This is due to fast interconversion between $\left({\text{P}}_{\text{F}}^{\text{red}}\right)$ and $\left({\text{P}}_{\text{R}}^{\text{red}}\right)$ on the relaxation time-scale $\left(k_{\text{FR}}+k_{\text{RF}}\gg R_2^{\text{restricted}}\right)$, so that extraction of k_FR, k_RF and ${}^{15}\text{N-}{R}_2^{\text{restricted}}$ is not possible without assumptions regarding these parameters. Thus, only ranges of ${}^{15}\text{N-}{R}_2^{\text{restricted}}$ and $\left(p_{\text{R}}^{\text{red}}\right)$ could be established (see SI): for ${<}^{15}{\text{N-R}}_2^{\text{resticted}}>=200\text{\hspace{1em}to\hspace{1em}}750{\text{s}}^{-1}$, $p_{\text{R}}^{\text{red}}$ varies from 0.5 to 0.2%. Although populated at less than 1%, the inclusion of $\left({\text{P}}_{\text{R}}^{\text{red}}\right)$ into the analysis is essential to reproduce the experimental ¹⁵N-DEST profiles.

The ¹⁵N-DEST/ΔR₂ data for Leu³ and the C-terminal three residues of htt^NT required a separate treatment and were fitted to the three-state model shown in gray in Figure 3B. Although formally the same overall three-state exchange model is used for these residues $\left({\text{P}}_{\text{F}}^{\text{red}}\leftrightarrow {\text{P}}_{\text{R}}^{\text{red}}\leftrightarrow {\text{P}}_{\text{T}}^{\text{red}}\right)$, the exchange process subsumes initial binding to the NP surface followed by reversible detachment from the surface (state $\left({\text{P}}_{\text{T}}^{\text{red}}\right)$, where “T” denotes “tethered”).¹³ The rate constants k_BT and k_TB were calculated a-posteriori (see SI and Figure 4) and fall in the ranges 40 to 75 s⁻¹ and 160 to 240 s⁻¹, respectively, depending on the assumed population of state $\left({\text{P}}_{\text{R}}^{\text{red}}\right)$. It follows that reversible detachment of these 4 residues from the NP surface occurs ∼3-fold slower than the binding event proper $\left(P_{\text{R}}^{\text{red}}\leftrightarrow P_{\text{B}}^{\text{red}}\right)$. While the central residues of htt^NT are likely to form an ordered helical structure when bound to TiO₂ NPs, as observed for htt^NTQ_n peptides bound to lipid micelles,³ the parts of the peptide that remain unstructured in the bound state retain the flexibility of the free peptide and can transiently detach from the NP surface (i.e., they are “tethered”). The same phenomenon was also observed in the interaction of htt^NTQ_n with small unilamellar lipid vesicles.¹³

The binding of Met⁷O-htt^NT to TiO₂ NPs was also characterized by analysis of ¹⁵N-DEST and ΔR₂ data (Figure S6). The population of the bound state $P_{\text{B}}^{\text{oxi}}$ is ∼0.6%, ∼3-fold lower than that of the reduced form, indicating that oxidation of Met⁷ reduces the binding affinity to TiO₂ NPs, in agreement with previous studies on the interaction of htt^NTQ₇ with lipid micelles.³

In conclusion, the current work provides a mechanistic basis for understanding the interaction of huntingtin peptides with TiO₂ NPs coupled with surface-catalyzed oxidation, and suggests that targeted catalysis of Met⁷ oxidation within the htt^NT domain of the huntingtin protein may provide a strategy for delaying the onset of Huntington’s disease.