|Home | About | Journals | Submit | Contact Us | Français|
PISEMO, a separated local field experiment that can be performed with either direct 15N (or 13C) detection or indirect 1H detection, is demonstrated on a single crystal of a model peptide. The 1H signals modulated by 1H-15N heteronuclear dipole-dipole couplings are observed stroboscopically in the windows of a multiple-pulse sequence used to attenuate the 1H-1H homonuclear dipole-dipole couplings. 1H-detection yields spectra with about 2.5 times the signal to noise ratio observed with 15N-detection under equivalent conditions. Resolution in both the 15N chemical shift and 1H-15N heteronuclear dipole-dipole coupling dimensions is similar to that observed with PISEMA, however, since only on-resonance pulses are utilized, the bandwidth is better.
The sensitivity advantages of indirect 1H-detection in double-resonance solid-state NMR experiments are well known (1). However, its implementation has lagged, largely because of the difficulty in devising experiments that combine continuous detection of high resolution 1H chemical shift spectra with the high sensitivity required for studies of proteins at high fields. In a way, this reinforces the motivation for the original development of proton-enhanced nuclear induction spectroscopy (2), where the direct detection of 1H decoupled 15N (or 13C) spectra obviated the difficulties of multiple-pulse line narrowing (3) required to observe high resolution 1H chemical shift spectra. Many aspects of 1H-detection of 15N signals in solid-state NMR experiments have been discussed as part of applications of MAS solid-state NMR to polycrystalline peptides and proteins (4 – 9). An initial application of 1H-detection to a stationary powder sample provides essential background, although an extra time domain was required in order to implement point-by-point sampling of spin-locked magnetization (10). In this Communication we demonstrate a multiple-pulse based SLF (separated local field) (11) experiment that can be performed with either direct 15N-detection or indirect 1H-detection. In spectra obtained in the same amount of time and under equivalent experimental conditions on the same sample, the 1H-detected signals are about 2.5 times larger than the 15N-detected signals with the same noise levels. We refer to this experiment as PISEMO (polarization inversion spin exchange-modulated observation). Its resolution and sensitivity are similar to those of PISEMA (polarization inversion spin exchange at the magic angle) (12), but it has improved bandwidth, which is an important advantage for high field experiments.
SLF is among the most elegant NMR methods (11); it yields two-dimensional NMR spectra where frequency splittings that are resolved by differences in the chemical shift frequencies of 15N (or 13C) nuclei provide direct measurements of the heteronuclear 1H-15N (or 1H-13C) dipole-dipole couplings. In the initial experimental implementations of SLF spectroscopy (13, 14), the heteronuclear dipole-dipole couplings of interest evolve under conditions where the interfering homonuclear (1H/1H) dipole-dipole couplings are suppressed with multiple-pulse (3) irradiations prior to the direct detection of 1H decoupled 15N (or 13C) signals. PISEMA relies on off-resonance irradiation to precisely define the “magic angle” (15) for homonuclear 1H/1H decoupling, therefore, it is inherently quite sensitive to frequency offsets among the 1H resonances, and these are larger (in kHz) at high fields due to the greater spread of chemical shift frequencies. To address this issue, we developed SAMMY (16) and SAMPI4 (17) as magic-sandwich (18, 19) based pulse sequences for SLF spectroscopy with on-resonance irradiations that are less susceptible to offset effects. This is now an active research area, and a variety of high resolution SLF experiments have been developed by other groups (20–22) and have been reviewed (23). It is now possible to select among experiments that are optimized for single crystal, powder, liquid crystal, or aligned biopolymer samples. However, all of these experiments have in common the detection of 1H decoupled 15N (or 13C) signals, which provides chemical shift frequencies in the direct dimension, as in the original double-resonance and SLF experiments.
Figure 1 compares the timing diagrams for three SLF pulse sequences and the corresponding two-dimensional spectra obtained under identical experimental conditions. The sample is a single crystal of 13Cα, 15N labeled N-acetyl-leucine; since there are four uniquely aligned molecules in each unit cell, the spectra have four resonances for each labeled 15N amide site. Because this sample has directly bonded 13C and 15N nuclei, 13C irradiation is applied in the appropriate intervals to yield fully decoupled spectra; this is not needed for experiments performed on samples where only the nitrogen sites are isotopically labeled. Complete two-dimensional spectra are shown in the figures to illustrate the experimental results from application of the pulse sequences. However, the unique spectroscopic information is contained in half of each spectrum, which is symmetric about the zero frequency in the heteronuclear dipole-dipole coupling dimension.
One-dimensional slices through the 15N chemical shift and 1H-15N heteronuclear dipole-dipole coupling frequency dimensions for all four resonances are shown in Figure 2. All of the frequency axes are adjusted for the scaling factors of the pulse sequences so that the frequencies and linewidths can be directly compared. The order of the individual slices in each panel in Figure 2 corresponds to the order of resonances with decreasing 1H-15N heteronuclear dipole-dipole coupling frequencies in Figure 1B, D, F. Panels A, B, and C in Figure 2 are 15N chemical shift slices and Panels D, E, and F are heteronuclear dipole-dipole coupling slices taken from the same data sets that are presented as two-dimensional contour plots in Figure 1.
For comparison, the 15N-detected PISEMA pulse sequence (Figure 1A), which yields spectra with the narrowest linewidths in the heteronuclear dipolar coupling frequency dimension, was used to obtain the experimental spectrum shown in Figure 1B. Only three out of the four expected resonances are present in this two-dimensional contour plot because of the broader linewidth and correspondingly reduced intensity of one of the resonances (bottom slices in Figure 2A and D). This is a typical issue with PISEMA spectra because the strong offset dependence of the Lee-Goldburg irradiation used to suppress the homonuclear 1H/1H dipole-dipole couplings during the heteronuclear spin-exchange interval makes it impossible to have optimal performance for all of the sites when the 1H resonances are spread over a broad range of chemical shift frequencies.
The pulse sequence for 15N-detected PISEMO (Figure 1C) differs from that for PISEMA (Figure 1A) in that it uses semi-windowless WaHuHa instead of continuous flip-flop Lee-Goldburg off-resonance irradiation (24) to effect 1H/1H homonuclear decoupling during heteronuclear 1H-15N spin-exchange.
The four pulses labeled P1, P2, P3, and P4 constitute a semi-windowless WaHuHa (3) cycle. The phase P1 is the same as that of the preceding spin-lock irradiation, and the phase of P5 differs by 180°. The phases applied during the even numbered t1 intervals are the inverse of the phases in the odd numbered t1 intervals. The durations of the pulses correspond to a nominal nutation angle of 116° (25). In practice, the phases of P1, P2, P3, P4, P5, P6 are Y, X, −X, −Y, −Y, Y, respectively, for the odd numbered t1 cycles; and −Y, −X, X, Y, Y, −Y, respectively, for the even numbered t1 cycles. To suppress the effects of probe ringing in 15N-detected PISEMO, the phase of the first 1H 90° pulse is alternated between X and −X, and the corresponding receiver phase is alternated between X and −X. In the 1H detected version, the phase of the first 1H 90° pulse is kept constant while the phase of P0 is alternated between X and −X, and the alternation of the phase of PS between X and Y is used to achieve quadrature detection.
The homonuclear decoupling pulse sequence used for 1H detection must fulfill two basic criteria. First, it must have at least one detection window without RF irradiation, which rules out FSLG (24), BLEW12 (26), and other windowless sequences, such as those used in PISEMA (12) and HIMSELF (22) experiments. It is possible to introduce very short detection windows in PISEMA sequences (23), however, in our experience line broadening and loss of sensitivity occur even with the shortest windows (3 µsec) that allow sampling on our instruments. Second, the 1H magnetization must be in the transverse plane during the period of the window. The 1H magnetization is along the Z axis during the windows of SAMMY (16) or SAMPI4 (17), which would otherwise be suitable for this purpose. WaHuHa is effective in this role, since the 1H magnetization is in the transverse plane during the window used to detect the signals as they evolve under the influence of the heteronuclear dipole-dipole coupling. The timing diagrams for the 1H-detected (Figure 1E) and 15N-detected (Figure 1C) versions of PISEMO differ only in the placement of the 15N chemical shift evolution period (t1 in Figure 1E vs. t2 in Figure 1C). Heteronuclear spin-exchange occurs during the t2 period of 1H-detected PISEMO; the homonuclear decoupling required for high resolution is effected by a multiple-pulse cycle that incorporates windows without irradiation between two of the four pulses. One of the windows is used for data acquisition. The 1H signals are sampled stroboscopically once per cycle to monitor the evolution of the exchange of magnetization between 15N and 1H. The experimental spectra obtained with 1H-detected PISEMO and 15N-detected PISEMO are essentially identical, except that the 1H-detected version yields signals whose individual signal to noise ratios are between 2.2 and 2.7 fold higher in spectra obtained in the same amount of time. This is shown in the comparison of the one-dimensional slices in Figure 2.
The principal reason for performing 1H-detected experiments is to gain sensitivity. Following Hong and Yamaguchi (10), we use equation (1) to predict the signal-to-noise ratio improvement of 1H-detection compared to 15N-detection in solid-state NMR experiments.
ζ is the signal enhancement factor; f is the CP transfer efficiency; γ and W are the gyromagnetic ratios and the line widths in Hz, respectively, for the indicated nuclei (H and N); Q and η are the probe quality factor and the sample filling factor, respectively; SWH is the 1H spectral width and FWH is the filter bandwidth. The factor u accounts for 15N-detected PISEMO starting with full 15N magnetization and the 1H-detected version starting with no 1H magnetization. This factor has the effect of reducing the signal enhancement by 0.5. The factors (γH/γN)3/2, (WX/WH)1/2 and f are 31.6, 1, and 1 respectively since the linewidths in the 1H-15N dipole-dipole coupling frequency dimension are about 150 Hz compared to the 3 ppm linewidths observed in the 15N chemical shift frequency dimension; cross-polarization transfer is part of the detection procedure and does not affect the relative sensitivity.
The factor (SWH/FWH)1/2 is determined, in practice, by the recovery time of the probe following a pulse, and how many data points can be sampled during the duration of the window. The bandwidth of the filter can be adjusted experimentally for a particular probe and window length, or during post processing with data sampled as fast as possible inside the available window after discarding distorted data points at the start and end of the period. The experimental spectra in Figures 1 and and22 resulted from sampling for 2.8 µs out of the total cycle time of 34.8 µs, and the factor (SWH/FWH)1/2 is calculated as (2.8 / 34.8)½ = 0.28.
The factors containing Q and η reflect the performance of the two channels of a triple-resonance probe, and can be evaluated using the principle of reciprocity. The 90° pulse length is a direct measure of the signal-to-noise ratio obtainable from a single coil that is used for both transmission and receiving. The magnitude of the B1 field is described by equation (2) (27):
The power required to achieve the Hartmann-Hahn match condition can be readily measured. In our homebuilt probe, which utilizes a single solenoid coil triple-tuned for 1H at 500 MHz, and 13C and 15N at their corresponding resonance frequencies; 17 Watts in the 1H channel and 511 Watts in the 15N channel provides B1 fields of 55.6 kHz. The calculated probe performance difference is (QHVN/QNVH)1/2 = (PNνN/PHνH)1/2 = 1.8. This is essentially the same as that based on the QH of 220 and QN of 70 measured with a network analyzer. For other types of probes, such as those utilizing cross-coils, measuring the Q alone will not be sufficient to determine the probe-related factors affecting sensitivity (28). In our experience, measuring the B1 field is a simple and reliable approach to characterizing the probe.
Combining all the factors that affect the signal-to-noise ratio, we predict an enhancement factor ζ= 2.8 = 0.5 * 0.35 * 31.6 * 1.0 * 1.8 * 0.28, which is in remarkably good agreement with the experimentally observed enhancement factors for the four individual resonances between 2.2 and 2.7.
The 1H-detected PISEMO spectrum of a single crystal in Figure 1E provides a significant improvement in sensitivity compared to the equivalent 15N-detected experiment. The signal to noise ratio can be further improved through the use of a preamplifier and radiofrequency switching system that is optimized for observing the 1H frequencies; additional improvements may come from an increase of the number of data points acquired in the windows used for stroboscopic detection and optimization of both the analog and digital filters in the spectrometer. Further improvements in the sensitivity of solid-state NMR experiments performed on samples of proteins and other biopolymers in lossy aqueous solutions can result from the increased efficiency and homogeneity of the 1H channel and reduced sample heating that result from the use of “low-E” coils (29–32). Because these probes often sacrifice efficiency on the low frequency channels, their optimal implementation may be for 1H-detected experiments at high fields.
Not only does 1H-detected PISEMO yield SLF spectra with improved sensitivity compared to the equivalent 15N-detected experiment, but also it provides a building block for more sophisticated pulse sequences (33) because of its favorable bandwidth for heteronuclear spin-exchange and the added flexibility that results from the ability to detect free induction decays that reflect the evolution of dipole-dipole coupling rather than chemical shift frequencies.
We thank A.A. Nevzorov and C.V. Grant for helpful discussions, and N. Sinha for providing the sample. This research was supported by grants RO1EB001966 and RO1GM075877 from the National Institute of Health, and utilized the Resource of NMR Molecular Imaging of Proteins, which is supported by grant P41EB002031.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.