In the study of biomolecular systems, magic angle spinning (MAS) solid state NMR is emerging as a powerful complementary tool to X-ray crystallography and solution state NMR. Making use of the distance and torsion angle constraints extracted in the solid state the structural characteristics of many biologically interesting systems have been elucidated via MAS solid state NMR recently (Castellani et al. 2002; Jaroniec et al. 2002; Rienstra et al. 2002; Luca et al. 2003; Tycko 2003; Krabben et al. 2004; Zech et al. 2005; Iwata et al. 2006; Egawa et al. 2007; Goldbourt et al. 2007). Homo- and heteronuclear distances involving 13C and 15N nuclei with well resolved isotropic chemical shifts are commonly used in MAS solid state NMR based structural studies. Although 1H resonances are generally broad due to strong homonuclear dipolar couplings, the possibilities for extracting short range 1H–1H distance estimates from fully protonated (13C, 15N) labelled peptide/protein samples have been demonstrated recently (de Boer et al. 2002; Lange et al. 2002, 2003, 2005; Tycko and Ishii 2003; Reif et al. 2003). This approach exploits the improved spectral resolution seen in 15N and 13C spectra and involves 1H–1H dipolar coupling mediated chemical shift correlation of the low γ nuclei. Cross-peak intensities seen in such data, commonly referred to as CHHC, CHHN, NHHN and NHHC spectra, are related to the spatial proximity of the protons that are directly attached to the corresponding nuclei observed in the two dimensions. Our recent studies (Riedel et al. 2005a, 2006) indicate that NHHN, CHHC and NHHC type experiments also hold considerable potential in the structural studies of RNAs that play a critical role in many biological processes and exhibit a variety of secondary and tertiary structural features. Duplex regions arising from consecutive formation of hydrogen bonded base pairs are commonly found in RNA. Hence, the identification of the hydrogen bonded base-pairs and the characterisation of the underlying hydrogen bonding patterns is of critical importance in the study of RNA. As different canonical and non-canonical base-pairing schemes encountered in nucleic acids are characterised by topologically different networks of strong proton–proton dipolar couplings, it has been demonstrated that the characterisation of the hydrogen bonding networks in RNAs can be effectively carried out via NHHN and NHHC type of experiments (Riedel et al. 2005a). In addition, it has also been shown that 1H–1H dipolar coupling mediated 13C–13C chemical shift correlation experiments can facilitate the characterisation of the glycosidic torsion angle χ, the sugar pucker and the helical regions of RNAs (Riedel et al. 2006). Hence, data from NHHN, CHHC and NHHC type experiments are critically important in RNA structural studies. Currently, the different proton–proton dipolar coupling mediated 15N/13C chemical shift correlation experiments are carried out individually and, hence, considerable amount of spectrometer time is required to generate data with good signal-to-noise ratio. In this communication an efficient approach for the simultaneous collection of these different MAS solid state NMR data sets is presented. The efficacy of the approach is demonstrated using an RNA composed of 97 CUG repeats, (CUG)97, a system that is under investigation in our laboratory (Riedel et al. 2004, 2005a, 2005b, 2006; Riedel 2007, Leppert et al. 2004).

The RF pulse sequence employed for generating 1H–1H dipolar coupling mediated chemical shift correlation of 13C/15N nuclei is shown in Fig. 1. The initial transverse 13C/15N magnetisation prepared by the first cross-polarisation (CP) step is allowed to evolve under 1H decoupling during the t 1 period. The second CP step transfers the t 1 modulated 13C/15N magnetisation back to the protons. The proton magnetisation is then flipped to the z axis and longitudinal 1H magnetisation exchange mediated by proton–proton dipolar couplings is allowed to take place during the proton spin diffusion (Lange et al. 2003; Riedel et al. 2005a, 2006) period τmix. The proton magnetisation at the end of τmix is rotated back to the transverse plane. The final CP transfers the polarisation from the protons to the 13C/15N nuclei for simultaneous detection in t 2 using the multiple receiver capabilities in the current generation of spectrometers. The experiment is typically carried out with a very short CP contact time and proton spin diffusion mixing time τmix to minimise relayed magnetisation transfers during CP and τmix. Cross-peaks with appreciable intensities are expected only between proton-attached 13C/15N sites that are connected by 1H–1H distances of less than ~3 Å. It is worth mentioning that when the different experiments are carried out individually the HC and HN transfer times can be optimised independently at any given spinning speed. Typically, the HC transfer requires a shorter CP contact time compared to the HN transfer. At the spinning speed of 25,000 Hz employed in this study, a CP contact time of 200 μs was found experimentally to be optimal to achieve satisfactory transfer to both the 15N and 13C nuclei. With such short CP contact times employed in this work, we do not observe any major signal losses during simultaneous H -> N, C cross-polarisations. The {(90)x − (k − 1)t 1 − (90)ϕ4} sandwich of pulses and delay applied in the 13C channel is eliminated when the required 15N and 13C ω1 spectral widths are identical. When the 15N and 13C ω1 spectral width requirements are different, the chemical shift evolution of 15N is started first, as the 15N spectral width is typically smaller, and during the time interval (k − 1)t 1, when the 15N magnetisation is evolving, the 13C magnetisation is kept in the z state. The ratio of the 15N and 13C ω1 spectral widths is controlled by the parameter k. Standard phase cycling procedures were employed to select the desired coherence transfer pathway and quadrature detection in the t 1 dimension is achieved via the States procedure (States et al. 1982). For each increment in t 1, four sets of data were acquired and processed, as described in Table 1, to separate the signals arising from 15N and 13C chemical shift evolutions in t 1 (Sorenson 1990; Farmer II 1991; Ramachandran et al. 1996). The RNA sample was prepared using appropriately labelled ribonucleoside triphosphates as described earlier (Leppert et al. 2004). A hydrated, undiluted {15N, 13C} labelled RNA sample of (CUG)97 (~7 mg) was used in the experiments carried out at a spinning speed of 25,000 Hz and with cooling air kept at a temperature of ~−50°C on a 500 MHz wide-bore Bruker Avance III solid state NMR spectrometer equipped with a 2.5 mm triple resonance probe. The processing of the NMR data was carried out with the NMRpipe software (Delaglio et al. 1995).

Fig. 1
figure 1

RF pulse sequence employed in this work. For the case where the 15N and 13C ω1 spectral width requirements are different, the RF pulse and the receiver phases were cycled as: ϕ1 = (y, −y); ϕ2 = (y, y, −y, −y); ϕ3 = (y, y, −y, −y); ϕ4 = (16* x, 16* −x); ϕ5 = (16* −y, 16* y); ϕ6 = (4* −y, 4* y); ϕ7 = ϕ8 = (8* −y, 8* y); (ϕR)N = (ϕR)C = (x, −x, −x, x, −x, x, x, −x, −x, x, x, −x, x, −x, −x, x, −x, x, x, −x, x, −x, −x, x, x, −x, -x, x, −x, x, x, −x). For the case where the 15N and 13C ω1 spectral widths are identical, the {(90)x − (k − 1)t 1 − (90)ϕ4} sandwich in the 13C channel is not applied and the RF pulse and the receiver phases were cycled as: ϕ5 = −y; (ϕR)N = (ϕR)C = (x, −x, −x, x, −x, x, x, −x, −x, x, x, −x, x, −x, −x, x) with the other RF pulse phases defined as before. High power 1H decoupling via the SPINAL 64 (Fung et al. 2000; De Paepe et al. 2003) scheme was employed in t 1 and t 2 time periods and phase sensitive 2D spectra were generated as described in the literature (States et al. 1982). The 13C and 15N signals were acquired in t 2 using the corresponding receivers in the two channels

Full size image
Table 1 Phase cycling used for quadrature detection in t 1 and for seperating signals arising from 13C and 15N chemical shift evolutions in t 1
Full size table

1H–1H dipolar coupling mediated isotropic 15N/13C chemical shift correlation spectra of the uniformly {15N, 13C} labelled sample of (CUG)97 obtained at a spinning speed of 25,000 Hz using a short proton spin diffusion mixing time of 160 μs and a 2.5 mm MAS rotor are shown in Fig. 2a–d. The resonance assignments indicated are based on our recent studies (Riedel et al. 2005b). The spectral characteristics observed are as expected for a helical RNA with GC base-pairs and glycosidic torsion angle χ in anti and are consistent with the data presented earlier (Riedel et al. 2005a, 2006). Due to the high spinning speed employed, the structurally important C8/C6 (aromatic) ↔ C2′/C3′ (ribose) cross-peaks are seen with good signal-to-noise ratio in the CHHC spectrum (Fig. 2a). It is worth mentioning that it is possible to extract structurally meaningful conclusions, from data such as the CHHC spectrum, not only by the presence but also from the absence of certain cross-peaks. For example, the absence of cross-peaks between the aromatic and sugar C1′ carbons directly indicates a χ angle, describing the relative orientation of the base and the sugar moiety across the glycosidic bond, in the anti range (Riedel et al. 2006). The cross-peak between the guanine imino and cytosine amino nitrogens arising from the spatial proximity of the corresponding protons in a GC base-pair is seen in the NHHN spectrum (Fig. 2d). The spectral characteristics of CHHN (Fig. 2b) and NHHC (Fig. 2c) data are also as expected. With the χ angle in the anti range, the intra-nucleotide distance between the guanine H1′ and the non-hydrogen bonded guanine amino proton is larger than 4 Å, hence, an intra-nucleotide cross-peak between the guanine amino nitrogen and the ribose C1′ carbon is not expected in the NHHC/CHHN spectra. However, it is well known from solution state NMR studies that an A-form helix leads to the spatial proximity of the non-hydrogen-bonded G amino proton with the ribose H1′ proton of the n + 1 neighbouring nucleotide in the same strand (Heus and Pardi 1991). Hence, the presence of a cross-peak between the guanine amino nitrogen and the ribose C1′ carbon in the NHHC and CHHN spectra of (CUG)97 is consistent with the A-form helical conformation of this RNA (Riedel et al. 2006). The spatial proximity between the non-hydrogen bonded amino proton of the cytosine and the cytosine aromatic H5 proton results in the appearance of the strong cross-peak observed between the corresponding amino nitrogen and the C5 carbon in the NHHC/CHHN spectra, permitting the assignment of these resonances. The signal intensities seen in the CHHN spectrum, with 15N detection in the t 2 dimension, are typically weaker than that observed in the NHHC spectrum. Although cross-peaks in the CHHC, CHHN, NHHN and NHHC spectra should arise in principle between heteronuclei attached to spatially proximal protons, the NHHC spectrum shows a number of cross-peaks involving the imino and amino nitrogens and carbons with no attached protons. Such correlations arise due to the fact that during the final CP step the cross-polarisation to the quaternary carbons predominantly arises from protons that are not strongly coupled to a 13C spin, such as the imino and amino protons (van Rossum et al. 2000). With short CP contact time typically employed in these studies, such cross-peaks originate mainly from short range 2-bond intra-nucleotide heteronuclear dipolar interactions (Fig. 2c) and, hence, provide a convenient alternative approach for the assignment of many of the important quaternary carbon resonances in RNAs. A short HC transfer time after proton mixing can lead to a reduction in the transfer of proton polarisation to non-protonated 13C nuclei. However, at the spinning speed of 25,000 Hz it is seen that using a very short HC transfer time leads to a reduction in the intensities of other signals of interest. An optimised value of 200 μs was employed for the CP contact times. Cross-peaks arising from inter-strand heteronuclear dipolar interactions and reflecting the presence of hydrogen-bonded base-pairs can be clearly seen in experiments carried out with longer contact time for the third CP step (data not shown).

Fig. 2
figure 2

Experimental CHHC (a), CHHN (b), NHHC (c) and NHHN (d) MAS solid state NMR spectra of (CUG)97 RNA generated via the pulse sequence given in Fig. 1 employing 15N and 13C ω1 spectral widths of 8,333.3 Hz and 25,000 Hz, respectively, a spinning speed of 25,000 Hz, data acquisition in the direct dimension of 10 ms for 13C and 12 ms for 15N, ramp CP contact times of 200 μs, a recycle time of 2 s, 40 t 1 increments (corresponding to t 1 acquisition times of 1.6 ms and 4.8 ms, respectively, in the 13C and 15N dimensions) with 256 transients per t 1 increment and a τmix of 160 μs. The assignments of the different resonances are also indicated in the 1D spectra collected with short CP contact times. A schematic representation of the double stranded (CUG)97 employed in this study and the GC Watson–Crick base-pairing scheme are also shown. Additional details of spectral processing and representative 1D slices taken from the 2D spectra are given in the Supplementary material

Full size image

Simultaneous collection of different 1H–1H dipolar coupling mediated chemical shift correlation data sets of (13C, 15N) labelled RNAs provides in a single experiment a variety of structural information and resonance assignments. The availability of such complementary data sets may also be required for undertaking a detailed quantitative analysis of the observed cross-peak intensities as a function of the mixing time. For example, an analysis of the observed cross-peak intensities in the NHHN spectrum (Fig. 2d) considering a dipolar network involving only protons connected to 15N nuclei can lead to erroneous results as the NHHC spectrum (Fig. 2c) clearly reveals the presence of a much larger homo- and heteronuclear dipolar spin network involving protons connected to both 15N and 13C nuclei. Hence, the simultaneous evaluation of all the relevant cross-peak intensities in the different data sets may be necessary for extracting 1H–1H distances quantitatively. The possibilities for reducing data acquisition times by the simultaneous collection of the signals arising from different nuclei in the direct dimension t 2 was demonstrated recently in the context of solution state NMR studies (Kupce et al. 2006). The present work, dealing with MAS solid state NMR studies, extends this approach further and illustrates how several multidimensional correlation spectra can be much more efficiently collected. It makes use of the capabilities in the current generation of NMR spectrometers for simultaneously collecting the signals from different nuclear species in the acquisition dimension. Additionally, suitable RF pulse phase cycling procedures are employed for generating the phase- sensitive data sets such that the signals arising from 15N and 13C evolutions in t 1 are also simultaneously obtained. It is worth mentioning that it is often necessary to work with small quantities of fully labelled materials as experiments carried out at high MAS frequencies to minimise the deleterious effects of CSAs, e.g. on the aromatic carbons of the nucleotide bases, would necessarily require the use of small volume rotors. Considering the fact that the sample has to be also fully hydrated to get the proper RNA fold (Leppert et al. 2004), the amount of labelled sample that can be used in these experiments will be very limited. Under these circumstances, the approach presented leads to considerable savings in time in collecting different proton–proton dipolar coupling mediated 15N/13C chemical shift correlation data with good signal-to-noise ratio. Although demonstrated in the context of RNA, the MAS solid state NMR method outlined here can be equally employed to reduce spectral data acquisition times in the study of peptides/proteins.