Within the last decade, MAS (magic angle spinning) solid-state NMR evolved into a powerful technique which allows to determine the structure and dynamics of uniformly isotopically enriched, non-soluble biological macromolecules (Castellani et al. 2002; Wasmer et al. 2008). However, low sensitivity limits many solid-state NMR applications. NMR signal intensities can be enhanced by up to a factor of 660 by making use of dynamic nuclear polarization (DNP) (Maly et al. 2008; Barnes et al. 2009). DNP enhancement is achieved by microwave (MW) irradiation inducing a transfer of polarization from unpaired electrons to the nuclei (Hovav et al. 2011). This signal enhancement allows to investigate systems of biological interest which are not accessible otherwise (Reggie et al. 2011; Renault et al. 2011; Sergeyev et al. 2011; Jacso et al. 2012). Cryogenic temperatures are a pre-requisite for DNP. Low temperatures, on the other hand, yield considerable line broadening in many cases, which is due to suppression of conformational averaging as a result of freezing, or due to non-uniform ordering of water molecules at the glass transition temperature (Ngai et al. 2008; Doster 2010). This line broadening is the major limitation for a wide-spread application of DNP solid-state NMR for the investigation of biological systems. Despite large efforts in improving DNP solid-state NMR, including hardware design (Griffin and Prisner 2010; Rosay et al. 2010) and progress in theory (Hovav et al. 2010; Hovav et al. 2011; Hu et al. 2011), it remains unclear if this technique is suitable to study uniformly isotopically enriched protein samples. We present here field dependent spectra of an uniformly isotopically enriched fibril sample of the Alzheimer’s disease amyloid peptide Aβ. We find that resolution significantly improves at higher external magnetic fields, indicating that conformational heterogeneity is not dominating the achieveable resolution.

Aggregation of monomeric amyloid–β peptides (Aβ) into soluble oligomers and insoluble fibrils is one of the major pathological hallmarks of Alzheimer’s disease (AD) (Hardy and Selkoe 2002; Walsh et al. 2002). In fact, fibrils formed by the Aβ peptides (36-43 residues) are the primary components of the plaques found in human brains affected by Alzheimer’s disease (Cummings and Cotman 1995). Aβ fibrils are considered a difficult system for solid-state NMR as it is assumed that they can adopt a large variety of different polymorphous forms which in turn might result in line broadening (Fandrich et al. 2009). Fibrils employed in this study are uniformly isotopically enriched and grown using a seeding protocol (Lopez del Amo et al. 2012). An electron microscopic (EM) image of the fibrils obtained by this procedure is shown in Fig. 1a. Figure 1b shows a 1D CPMAS 13C spectra obtained at room temperature, at 96 K and at 96 K with microwave irradiation. A gain in intensity of around fourfold is observed by lowering the temperature only, due to a more favourable Boltzmann distribution. With DNP, a total signal enhancement of a factor of 20 was obtained. In principle, signal enhancements of ε ≈ 660 are possible. Typically, enhancements in biological systems are smaller (Mak-Jurkauskas et al. 2008; Bajaj et al. 2009a), in agreement with the values observed here. Partial sample deuteration can yield significantly larger enhancements on the order of ε = 120–140 (Akbey et al. 2010).

Fig. 1
figure 1

a EM image of the Aβ40 fibrils employed in the MAS solid-state NMR studies. b 1D-13C spectra of Alzheimer’s disease β-amyloid fibrils obtained under DNP conditions, with and without microwave irradiation (MW = on, MW = off), and at room temperature. In total, an enhancement factor of 20 is obtained. c 2D 15N,13C correlation spectrum of β-amyloid fibrils under DNP conditions (100 K). Side chain resonances of lysine, arginine and histidines are well resolved. d DNP 2D 15N,13C correlation spectrum of the histidine side chain region highlighted in b. All spectra were acquired at a magnetic field strength of 9.4 T

Full size image

In the NCO and NCA experiments recorded for Aβ fibrils (Fig. 1b), most of the backbone resonances are overlapped due to the intrinsic low resolution at 9.4 T. However, well resolved resonances are observed for sidechains. This is especially obvious for the histidine spectral region (Fig. 1c). Here, two spin systems can be identified (labeled as 1 and 2) which are presumably due to His13 and His14 in Aß40. The 13C/15N line widths of the resolved Cδ1-Nε1 correlation peak amounts to 140 Hz in the 15N dimension, and ~150 Hz in the 13C dimension. These values are very similar to the line width which is observed at room temperature (Lopez del Amo et al. 2012). This indicates that structural heterogeneity of the sample is not dominating the achievable resolution at 100 K, but rather the inherent lack of resolution at low magnetic fields (9.4 T).

In order to probe the field dependence of the resonance line widths at low temperatures, MAS solid-state NMR experiments were acquired at 14.1 T and 20.0 T. Figure 2a shows 2D 13C,13C correlation spectra of Aβ40 fibrils that were recorded at low temperatures (~100 K) at the indicated magnetic field strengths. Obviously, the spectral resolution is improving at higher magnetic fields. The significantly overlapped spectrum recorded at 9.4 T is very well resolved at 20.0 T yielding linewidths for individual crosspeaks that are on the order of 140–160 Hz. Figure 2b shows a comparison of the Aβ fibril 2D-13C,13C correlation spectrum recorded at low temperatures (20.0 T) and the spectrum recorded at room temperature (21.1 T). Methyl resonances are typically weak at a temperature of 100 K (Bajaj et al. 2009b). The observed temperature dependent chemical shifts indicate temperature dependent conformational changes that can potentially be probed by NMR. In comparison to the room temperature spectrum, the resolution is only slightly decreased at 100 K. The line width for the resolved 13Cβ serine resonance increases to 160 Hz, in comparison to 130 Hz at room temperature. Structural heterogeneity of the sample is thus not dominating the achievable resolution. We expect similar results for other amyloidogenic peptides and proteins. DNP solid-state NMR was applied recently for the structural characterization of a partially labelled nanocrystallin and a fibrillar sample of the Sup35 derived peptide GNNQQNY (Debelouchina et al. 2010), as well as for the study of acid induced fibrils formed by the PI3 K-SH3 domain (Bayro et al. 2011). These experiments were carried out at 9.4 T and 100 K, and show that line-broadening is in general not an issue for measurements at low temperatures.

Fig. 2
figure 2

a Field dependence of 2D-13C,13C correlation spectra of Aβ40 fibrils at low temperatures (100 K). Spectra were acquired and processed using the same parameters in all cases. In each plot, the lowest contour was drawn at 20 % of the maximum intensity of the isoleucine Cα–Cδ correlation signals. Spectra plotted at different contour levels for each field are shown as part of the Supporting Figure 1, together with 1D traces for selected peaks (Supporting Figure 3). b Superposition of the Aβ fibril 2D-13C,13C correlation spectra recorded at cryogenic (850 MHz, red) and at room temperature (900 MHz, black). The resolution is clearly not compromised at low temperature. A comparison of 2D-13C,13C correlation spectra recorded at cryogenic temperatures (600 and 850 MHz) is shown in Supporting Figure 3

Full size image

We demonstrate here that homogeneously labelled Alzheimer’s Disease amyloid Aβ40 fibrils retain their spectral resolution at cryogenic temperatures, and are amenable for assignments and a structural characterization in case samples are measured at high enough magnetic field strengths. In the future, signal enhancements of several orders of magnitude combined with high resolution solid-state NMR will be extremely valuable for the characterization of membrane proteins and amyloid structures in complex biological systems. In situ studies of Aβ peptides interacting with native membranes for example might allow to elucidate the mechanism of toxicity involved in Alzheimer’s Disease. Such experiments were impossible in the past due to the low sensitivity as a consequence of the low amount of material that can be packed into an NMR rotor. High-field DNP might be suitable to overcome such problems in the future, with high potential for applications in structural biology research.

Experimental details

The Aβ peptide was prepared as described previously (Dasari et al. 2011). For one Aβ fibril sample, approximately 8 mg of uniformly 13C and 15N labelled Aβ40 was fibrilized using a seeding protocol: Seeds were obtained after multiple rounds of sonification and fibril growth in order to select a single amyloid morphology (Lopez del Amo et al. 2012). Aβ fibrils were doped using 30 mM of the biradical TOTAPOL (1-(TEMPO-4-oxy)-3-(TEMPO-4-amino)-propan-2-ol) (Hu et al. 2004). The material was concentrated by centrifugation and packed into a 3.2 mm zirconia NMR rotor. All DNP-NMR experiments were performed using a commercial Bruker DNP spectrometer operating at a 1H frequency of 400 MHz and microwave frequency of 263 GHz. Signal enhancement was achieved in situ, directly within the sample. The microwave power was generated by a Bruker gyrotron oscillator. High field low temperature MAS-NMR experiments were acquired at 600 MHz and 850 MHz spectrometers (Bruker Rheinstetten and Max-Planck-Insitut für Biophysikalische Chemie Göttingen, respectively). The reference room temperature spectrum was acquired at 900 MHz (FMP Berlin). In all cases, spectra were recorded using a triple resonance, low-temperature, HCN probe employing conventional 3.2 mm zirconia rotors. Cryogenic temperatures were achieved and controlled with Bruker low-temperature MAS accessory. All spectra were recorded by setting the MAS frequency to 10 kHz, and employing a 1H/13C CP contact time of 1.5 ms. High-power proton decoupling (ωrf = 85 kHz) was applied during acquisition using SPINAL-64 (Fung et al. 2000). 13C,13C transfers were achieved using a PDSD sequence, setting the mixing time to 15 ms. Magnetization was allowed to evolve for 12 ms in the direct and indirect dimensions, respectively.