Using low-E resonators to reduce RF heating in biological samples for static solid-state NMR up to 900 MHz

RF heating of solid-state biological samples is known to be a destabilizing factor in high-field NMR experiments that shortens the sample lifetime by continuous dehydration during the high-power cross-polarization and decoupling pulses. In this work, we describe specially designed, large volume, low-E 15N-1H solid-state NMR probes developed for 600 and 900 MHz PISEMA studies of dilute membrane proteins oriented in hydrated and dielectrically lossy lipid bilayers. The probes use an orthogonal coil design in which separate resonators pursue their own aims at the respective frequencies, resulting in a simplified and more efficient matching network. Sample heating at the 1H frequency is minimized by a loop-gap resonator which produces a homogeneous magnetic field B1 with low electric field E. Within the loop-gap resonator, a multi-turn solenoid closely matching the shape of the sample serves as an efficient observe coil. We compare power dissipation in a typical lossy bilayer sample in the new low-E probe and in a previously reported 15N-1H probe which uses a double-tuned 4-turn solenoid. RF loss in the sample is measured in each probe by observing changes in the 1H 360 degrees pulse lengths. For the same values of 1H B1 field, sample heating in the new probe was found to be smaller by an order of magnitude. Applications of the low-E design to the PISEMA study of membrane proteins in their native hydrated bilayer environment are demonstrated at 600 and 900 MHz.     

Low-E probe for (19)F-(1)H NMR of dilute biological solids

Sample heating induced by radio frequency (RF) irradiation presents a significant challenge to solid state NMR experiments in proteins and other biological systems, causing the sample to dehydrate which may result in distorted spectra and a damaged sample. In this work we describe a large volume, low-E (19)F-(1)H solid state NMR probe, which we developed for the 2D (19)F CPMG studies of dilute membrane proteins in a static and electrically lossy environment at 600MHz field. In (19)FCPMG and related multi-pulse (19)F-(1)H experiments the sample is heated by the conservative electric fields E produced in the sample coil at both (19)F and (1)H frequencies. Instead of using a traditional sample solenoid, our low-E (19)F-(1)H probe utilizes two orthogonal loop-gap resonators in order to minimize the conservative electric fields responsible for sample heating. Absence of the wavelength effects in loop-gap resonators results in homogeneous RF fields and enables the study of large sample volumes, an important feature for the dilute protein preparations. The orthogonal resonators also provide intrinsic isolation between the (19)F and (1)H channels, which is another major challenge for the (19)F-(1)H circuits where Larmor frequencies are only 6% apart. We detail steps to reduce (19)F background signals from the probe, which included careful choice of capacitor lubricants and manufacture of custom non-fluorinated coaxial cables. Application of the probe for two-dimensional (19)F CPMG spectroscopy in oriented lipid membranes is demonstrated with Flufenamic acid (FFA), a non-steroidal anti-inflammatory drug.     

Analysis of RF heating and sample stability in aligned static solid-state NMR spectroscopy

Sample instability during solid-state NMR experiments frequently arises due to RF heating in aligned samples of hydrated lipid bilayers. A new, simple approach for estimating sample temperature is used to show that, at 9.4 T, sample heating depends mostly on (1)H decoupling power rather than on (15)N irradiation in PISEMA experiments. Such heating for different sample preparations, including lipid composition, salt concentration and hydration level was assessed and the hydration level was found to be the primary parameter correlated with sample heating. The contribution to RF heating from the dielectric loss appears to be dominant under our experimental conditions. The heat generated by a single scan was approximately calculated from the Q values of the probe, to be a 1.7 degrees C elevation per single pulse sequence iteration under typical sample conditions. The steady-state sample temperature during PISEMA experiments can be estimated based on the method presented here, which correlates the loss factor with the temperature rise induced by the RF heating of the sample.     

Evaluation of cross-correlation effects and measurement of one-bond couplings in proteins with short transverse relaxation times

Various strategies are described and compared for measurement of one-bond J(NH) and J(NC') splittings in larger proteins. In order to evaluate the inherent resolution obtainable in the various experiments, relaxation rates of (15)N-(1)H(N) coupled and heteronuclear decoupled resonances were measured at 600- and 800-MHz field strengths for both perdeuterated and protonated proteins. A comparison of decay rates for the two (15)N-?H(N)? doublet components shows average ratios of 4.8 and 3.5 at 800- and 600-MHz (1)H frequency, respectively, in the perdeuterated proteins. For the protonated proteins these ratios are 3.2 (800 MHz) and 2.4 (600 MHz). Relative to the regular HSQC experiment, the enhancement in TROSY (15)N resolution is 2.6 (perdeuterated; 800 MHz), 2.0 (perdeuterated; 600 MHz), 2.1 (protonated; 800 MHz), and 1.7 (protonated; 600 MHz). For the (1)H dimension, the upfield (1)H(N)-?(15)N? component on average relaxes slower than the downfield (1)H(N)-?(15)N? component by a factor of 1.8 (perdeuterated; 800 MHz) and 1.6 (perdeuterated; 600 MHz). The poor resolution for the upfield (15)N-?(1)H? doublet component in slowly tumbling proteins makes it advantageous to derive the J(NH) splitting from the difference in frequency between the narrow downfield (15)N doublet component and either the (1)H-decoupled (15)N resonance or the peak position in an experiment which J-scales the frequency of the upfield doublet component but maintains some of the advantages of the TROSY experiment.     

14N Polarization Inversion Spin Exchange at Magic Angle (PISEMA)

Polarization Inversion Spin Exchange at Magic Angle (PISEMA) is a powerful experiment for determining peptide orientation in uniformly aligned samples such as planar membranes. In this paper, we present (14)N-PISEMA experiment which correlates (14)N quadrupolar coupling and (14)N-(1)H dipolar coupling. (14)N-PISEMA enables the use of (14)N quadrupolar coupling tensor as an ultra sensitive probe for peptide orientation and can be carried out without the need of isotope enrichment. The experiment is based on selective spin-exchange between a proton and a single-quantum transition of (14)N spins. The spin-exchange dynamics is described and the experiment is demonstrated with a natural abundant N-acetyl valine crystal sample.     

The effect of Hartmann-Hahn mismatching on polarization inversion spin exchange at the magic angle

The effect of the Hartmann-Hahn mismatch delta = omega(eff)-omega(1S) during polarization inversion spin exchange at the magic angle (PISEMA) has been investigated, where omega(eff) and omega(1S) represent the amplitudes of the 1H effective spin-locking field at the magic angle and the 15N RF spin-locking field, respectively. During the PISEMA evolution period, the exact Hartmann-Hahn match condition (i.e., delta = 0) yields a maximum dipolar scaling factor of 0.816 for PISEMA experiments, while any mismatch results in two different effective fields for the first and second half of each frequency switched Lee-Goldburg (FSLG) cycle. The mismatch effect on the scaling factor depends strongly on the transition angle from one effective field to the other within each FSLG cycle as well as on the cycle time. At low RF spin-lock amplitudes in which the FSLG cycle time is relatively long, the scaling factor rapidly becomes smaller as omega(1S) becomes greater than omega(eff). On the other hand, when omega(1S) < omega(eff), there is relatively little effect on the scaling factor with variation in delta. As a result, the presence of RF inhomogeneities may significantly broaden the line-width in the dipolar dimension because of the mismatch effect. Higher RF spin-lock amplitudes result in a relatively small variation for the scaling factor. Furthermore, ramped amplitude of the 15N RF spin-lock field in synchronization with the flip-flop of the FSLG sequence minimizes the transition angle between the two effective fields within the FSLG cycle. It is shown experimentally that such a ramped amplitude not only gives rise to the same scaling factor but also results in a narrower dipolar line-width in comparison with the rectangular amplitude.     

Using a cross-coil to reduce RF heating by an order of magnitude in triple-resonance multinuclear MAS at high fields

Four different coil designs for use with MAS in triple-resonance multi-nuclear experiments at high fields are compared, using a combination of finite element analysis (FEA) software and NMR experiments, with respect to RF field strength per unit power and relative sample heating, as governed by mean E/B(1) within the sample region. A commercial FEA package, Microwave Studio 5.1 by Computer Simulation Technology (CST) is shown to obtain remarkably accurate agreement with the experiments in Q(L), L, B, E, and mode frequencies in all cases. A simplified treatment of RF heating in NMR MAS samples is derived and shown to agree with the NMR experimental results within about 10% for two representative stator designs. The coil types studied include: (1) a variable-pitch solenoid outside a ceramic coilform, (2) a conventional solenoid very closely spaced to the MAS rotor, (3) a scroll coil, and (4) a segmented saddle cross coil (XC) for (1)H with an additional solenoid over it for the two lower-frequency channels. The XC/solenoid is shown to offer substantial advantages in reduced decoupler heating, improved S/N, and improved compatibility with multinuclear tuning and high-power decoupling. This seems largely because the division of labor between two orthogonal coils allows them each, and their associated circuitry, to be separately optimized for their respective regimes.     

Design of small volume HX and triple-resonance probes for improved limits of detection in protein NMR experiments

Three- and four-frequency nuclear magnetic-resonance probes have been designed for the study of small amounts of protein. Both "HX" (1H, X, and 2H channels) and "triple-resonance" (1H, 15N, 13C, and 2H) probes were implemented using a single transmit/receive coil and multiple-frequency impedance matching circuits. The coil used was a six-turn solenoid with an observe volume of 15 microl. A variable pitch design was used to improve the B1 homogeneity of the coil. Two-dimensional HSQC spectra of approximately 1mM single labeled 15N- and double labeled 15N/13C-proteins were acquired in experimental times of approximately 2h. Triple-resonance capability of the small-volume triple-resonance probe was demonstrated by acquiring three-dimensional HNCO spectra from the same protein samples. In addition to enabling very small quantities of protein to be used, the extremely short pulse widths (1H = 4, 15N = 4, and 13C = 2 micros) of this particular design result in low power decoupling and wide-bandwidth coverage, an important factor for the ever-higher operating frequencies used for protein NMR studies.     

Optimal electric fields for different sample shapes in high resolution NMR spectroscopy

For many applications, reducing sample resistance, rather than increasing probe Q or filling factor, is the only way to further improve the signal-to-noise ratio of cryogenically cooled NMR probes. In this paper, bounds are calculated for the minimum sample resistance that can be achieved for various sample geometries. The sample resistance of 100 mM NaCl in H(2)O in 5 mm sample tubes was measured on a 600 MHz cold probe to be within 14% of the optimum value. The minimum sample resistance can however be lowered by altering the tube cross section. Rectangular tubes oriented with the long axis along the RF magnetic field are particularly favourable.     

Reduction of RF-induced sample heating with a scroll coil resonator structure for solid-state NMR probes

Heating due to high power 1H decoupling limits the experimental lifetime of protein samples for solid-state NMR (SSNMR). Sample deterioration can be minimized by lowering the experimental salt concentration, temperature or decoupling fields; however, these approaches may compromise biological relevance and/or spectroscopic resolution and sensitivity. The desire to apply sophisticated multiple pulse experiments to proteins therefore motivates the development of probes that utilize the RF power more efficiently to generate a high ratio of magnetic to electric field in the sample. Here a novel scroll coil resonator structure is presented and compared to a traditional solenoid. The scroll coil is demonstrated to be more tolerant of high sample salt concentrations and cause less RF-induced sample heating. With it, the viable experimental lifetime of a microcrystalline ubiquitin sample has been extended by more than an order of magnitude. The higher B1 homogeneity and permissible decoupling fields enhance polarization transfer efficiency in 15N-13C correlation experiments employed for protein chemical shift assignments and structure determination.     

Design, construction, and validation of a 1-mm triple-resonance high-temperature-superconducting probe for NMR

We report a 600-MHz 1-mm triple-resonance high-temperature-superconducting (HTS) probe for nuclear magnetic resonance spectroscopy. The probe has a real sample volume of about 7.5 microl, an active volume of 6.3 microl, and appears to have the highest mass sensitivity at any field strength. The probe is constructed with four sets of HTS coils that are tuned to 1H, 2H, 13C, and 15N, and there is a z-axis gradient. The coils are cooled with a conventional Bruker CryoPlatform to about 20 K, and the sample chamber can be regulated above or below room temperature over a moderate range using a Bruker variable temperature unit. The absolute S/N for 0.1% ethylbenzene is approximately 1/3 that of a conventional 5mm probe with just 1/70 of the sample volume. We demonstrate the utility of this probe for small molecules and proteins with 2D spectra of just 1.7 microg of ibuprofen and 400 microM 15N-labeled ubiquitin.     

NMR spin locking of proton magnetization under a frequency-switched Lee-Goldburg pulse sequence

The spin dynamics of NMR spin locking of proton magnetization under a frequency-switched Lee-Goldburg (FSLG) pulse sequence is investigated for a better understanding of the line-narrowing mechanism in PISEMA experiments. For the sample of oriented 15N(1,3,5,7)-labeled gramicidin A in hydrated DMPC bilayers, it is found that the spin-lattice relaxation time T(1rho)(H) in the tilted rotating frame is about five times shorter when the 1H magnetization is spin locked at the magic angle by the FSLG sequence compared to the simple Lee-Goldburg sequence. It is believed that the rapid phase alternation of the effective fields during the FSLG cycles results in averaging of the spin lock field so that the spin lock becomes less efficient. A FSLG supercycle has been suggested here to slow the phase alternation. It has been demonstrated experimentally that a modified PISEMA pulse sequence with such supercycles gives rise to about 30% line narrowing in the dipolar dimension in the PISEMA spectra compared to a standard PISEMA pulse sequence.     

Measurement of high intensity focused ultrasound fields by a fiber optic probe hydrophone

The acoustic fields of a high intensity focused ultrasound (HIFU) transducer operating either at its fundamental (1.1 MHz) or third harmonic (3.3 MHz) frequency were measured by a fiber optic probe hydrophone (FOPH). At 1.1 MHz when the electric power applied to the transducer was increased from 1.6 to 125 W, the peak positive/negative pressures at the focus were measured to be p(+) = 1.7-23.3 MPa and p(-) = -1.2(-) -10.0 MPa. The corresponding spatial-peak pulse-average (I(SPPA)) and spatial-average pulse-average (I(SAPA)) intensities were I(SPPA) =77-6000 W/cm2 and I(SAPA) = 35-4365 W/cm2. Nonlinear propagation with harmonics generation was dominant at high intensities, leading to a reduced -6 dB beam size (L x W) of the compressional wave (11.5 x 1.8-8.8 1.04 mm) but an increased beam size of the rarefactional wave (12.5 x 1.6-13.2 x 2.0 mm). Enhancement ratio of absorbed power density in water increased from 1.0 to 3.0. In comparison, the HIFU transducer working at 3.3 MHz produced higher peak pressures (p(+) = 3.0-35.1 MPa and p(-) = -2.5(-) - 13.8 MPa) with smaller beam size (0.5 x 4 mm). Overall, FOPH was found to be a convenient and reliable tool for HIFU exposimetry measurement.     

Proton-detected separated local field spectroscopy

PISEMO, a separated local field experiment that can be performed with either direct (15)N (or (13)C) detection or indirect (1)H detection, is demonstrated on a single crystal of a model peptide. The (1)H signals modulated by (1)H-(15)N heteronuclear dipole-dipole couplings are observed stroboscopically in the windows of the multiple-pulse sequence used to attenuate (1)H-(1)H homonuclear dipole-dipole couplings. (1)H-detection yields spectra with about 2.5 times the signal to noise ratio observed with (15)N-detection under equivalent conditions. Resolution in both the (15)N chemical shift and (1)H-(15)N heteronuclear dipole-dipole coupling dimensions is similar to that observed with PISEMA, however, since only on-resonance pulses are utilized, the bandwidth is better.     

Very fast magic angle spinning (1)H-(14)N 2D solid-state NMR: sub-micro-liter sample data collection in a few minutes

Substantial resolution and sensitivity enhancements of solid-state (1)H detected (14)N HMQC NMR spectra at very fast MAS rates up to 80 kHz, in a 1mm MAS rotor, are presented. Very fast MAS enhances the (1)H transverse relaxation time and efficiently decouples the (1)H-(14)N interactions, both effects leading to resolution enhancement. The micro-coil contributes to the sensitivity increase via strong (14)N rf fields and high sensitivity per unit volume. (1)H-(14)N HMQC 2D spectra of glycine and glycyl-L-alanine at 70 kHz MAS at 11.7 T are observed in a few minutes with a sample volume of 0.8 μL.     

H,N,C-triple resonance NMR of very large systems at 900 MHz

We provide quantitative signal to noise data and feasibility study at 900 MHz for 1H-15N-13C triple resonance backbone assignment pulse sequences obtained from a medium sized 2H, 13C, 15N labeled protein slowed down in glycerol-water solution to mimic relaxation and spectroscopic properties of a much larger protein system with macromolecular tumbling correlation time of 52 and 80 ns, respectively, at 296 and 283 K (corresponding to molecular weights of 130 and 250 kDa). Comparisons of several different schemes for transferring magnetization from proton to nitrogen and back to proton confirms Yang and Kay's 1999 prediction that avoiding the unfavorable relaxation properties of 1H-15N multiple quantum coherence in the TROSY phase cycle of the final 15N-1H transfer before acquisition is crucial for maximal sensitivity from these very large molecular weight systems. We also show results which confirm some predictions regarding the superiority of TROSY at 900 MHz vs. 800 MHz especially as the molecular weights become very large.     

1H,15N,13C-triple resonance NMR of very large systems at 900 MHz

We provide quantitative signal to noise data and feasibility study at 900 MHz for 1H-15N-13C triple resonance backbone assignment pulse sequences obtained from a medium sized 2H, 13C, 15N labeled protein slowed down in glycerol-water solution to mimic relaxation and spectroscopic properties of a much larger protein system with macromolecular tumbling correlation time of 52 and 80 ns, respectively, at 296 and 283 K (corresponding to molecular weights of 130 and 250 kDa). Comparisons of several different schemes for transferring magnetization from proton to nitrogen and back to proton confirms Yang and Kay's 1999 prediction that avoiding the unfavorable relaxation properties of 1H-15N multiple quantum coherence in the TROSY phase cycle of the final 15N-1H transfer before acquisition is crucial for maximal sensitivity from these very large molecular weight systems. We also show results which confirm some predictions regarding the superiority of TROSY at 900 MHz vs. 800 MHz especially as the molecular weights become very large.     

Ultra-broadband NMR probe: numerical and experimental study of transmission line NMR probe

We have reinvestigated a transmission line NMR probe first published by Lowe and co-workers in 1970s [Rev. Sci. Instrum. 45 (1974) 631; 48 (1977) 268] numerically and experimentally. The probe is expected to be ultra-broadband, thus might enable new types of solid-state NMR experiments. The NMR probe consists of a coil and capacitors which are connected to the coil at regular intervals. The circuit is the same as a cascaded LC low-pass filter, except there are nonzero mutual inductances between different coil sections. We evaluated the mutual inductances by Neumann's formula and calculated the electrical characteristics of the probe as a function of a carrier frequency. We found that they were almost the same as those of a cascaded LC low-pass filter, when the inductance L of a section was estimated from the inductance of the whole coil divided by the number of the sections, and if C was set to the capacitance in a section. For example, the characteristic impedance of a transmission line coil is given by Z=(L/C)(1/2). We also calculated the magnitude and the distribution of RF magnetic field inside the probe. The magnitude of RF field decreases when the carrier frequency is increased because the phase delay between neighboring sections is proportional to the carrier frequency. For cylindrical coils, the RF field is proportional to (pinu/2nu(d))(1/2)exp(-nu/nu(d)), where the decay frequency nu(d) is determined by the dimensions of the coil. The observed carrier frequency thus must be much smaller than the decay frequency. This condition restricts the size of transmission line coils. We made a cylindrical coil for a 1H NMR probe operating below 400 MHz. It had a diameter 2.3mm and a pitch 1.2mm. Five capacitors of 6pF were connected at every three turns. The RF field strength was 40 and 60 kHz at the input RF power 100 W by a calculation and by experiments, respectively. The calculations showed that the RF field inhomogeneity along the coil axis was caused by a standing wave of current, which arose from the reflections at the coil ends. The calculation showed that the homogeneity could be improved by decreasing the pitch near the both ends and making their impedance close to that at the center.     

Frequency offset refocused PISEMA-type sequences

The popular PISEMA experiment is highly sensitive to the 1H chemical shift dispersion and the choice of the 1H carrier frequency. This is due to the off-resonance 1H irradiation in the FSLG-CP sequence employed during the dipolar evolution period. In the modified approach described in this work, the interfering frequency offset terms are suppressed. In the new pulse schemes, conventional FSLG-CP is intercalated with 180 degrees pulses applied simultaneously to both frequency channels, and with phases set orthogonal to those of the spin-lock fields. The technique is demonstrated on a nematic liquid-crystalline sample. Extensions to amplitude-modulated FSLG-CP recoupling under MAS are also presented.     

Estimating localized oscillatory tissue motion for assessment of the underlying mechanical modulus

The technique of harmonic motion imaging (HMI) uses the localized stimulus of the oscillatory ultrasonic radiation force as produced by two overlapping beams of distinct frequencies, and estimates the resulting harmonic displacement in the tissue in order to assess its underlying mechanical properties. In this paper, we studied the relationship between measured displacement and stiffness in gels and tissues in vitro. Two focused ultrasound transducers with a 100 mm focal length were used at frequencies of 3.7500 MHz and either 3.7502 (or 3.7508 MHz), respectively, in order to produce an oscillatory motion at 200 Hz in the gel or tissue. A 1.1 MHz diagnostic transducer (Imasonics, Inc.) was also focused at 100 mm and acquired 5 ms RF signals (pulse repetition frequency (PRF)=3.5 kHz) at 100 MHz sampling frequency during radiation force application. First, three 50x50 mm(2) acrylamide gels were prepared at concentrations of 4%, 8% and 16%. The resulting displacement was estimated using crosscorrelation techniques between successively acquired RF signals with a 2 mm window and 80% window overlap at 1260 W/cm(2). A normal 1-D indentation instrument (TeMPeST) applied oscillatory loads at 0.1-200 Hz with a 5 mm-diameter flat indenter. Then, 12 displacement measurements in 6 porcine muscle specimens (two measurements/case, as above) were made in vitro, before and after ablation which was performed for 10 s at 1260 W/cm(2). In all gel cases, the harmonic displacement was found to linearly increase with intensity and exponentially decrease with gel concentration. The TeMPeST measurements showed that the elastic moduli for the 4%, 8% and 16% gels equaled 3.93+/-0.06, 17.1+/-0.2 and 75+/-2 kPa, respectively, demonstrating that the HMI displacement estimate depends directly on the gel stiffness. Finally, in the tissues samples, the mean displacement amplitude showed a twofold decrease between non-ablated and ablated tissue, demonstrating a correspondence between the HMI response and an increase in stiffness measured with the TeMPeST instrument.