Low-E Probes for Biological Solid State NMR

Introduction

Solid state nuclear magnetic resonance (NMR) spectroscopy at high magnetic fields is rapidly developing into an important technique for protein structure determination.  It is especially valuable since it can be applied to a vast array of insoluble proteins for which neither solution NMR nor X-ray crystallography is appropriate.  However, the high radio-frequency (RF) fields needed for solid state NMR techniques can very easily heat the protein samples, distorting the spectra and even destroying the samples.

To solve this problem, we have developed new "low-E" RF probes at the Magnet Lab to minimize these sample heating effects.  Probes for PISEMA spectroscopy of mechanically aligned samples are available to the users of the Magnet Lab NMR facilities at fields from 400 to 900 MHz.  Also, a ¹⁹F—¹H probe for aligned samples at 600 MHz, bicelle sample probes for 600, 900 MHz, and a low-E ¹H—X magic angle spinning (MAS) probe for 4 mm samples at 750 MHz have been constructed and can be arranged for.

The Sample Heating Dillemma

Fig. 1. RF loss in saline sample inside 400 MHz solenoid [2]. "ESR" stands for effective series resistance of the saline.

An RF electromagnetic excitation field is an essential part of the NMR experiment.  The RF magnetic field B1 interacts with the nuclear spin magnetic moments, flipping them back and forth as needed for a given experiment.  However, any time-varying magnetic field carries with it an electric field, as described by Faraday's law rot E1 = —dB1/dt.  In an aqueous sample, charge carriers respond according to Ohm's law to form a current distribution J = s(w)E1, where s(w) is the frequency dependent conductivity of the sample.  These currents circle around the parallel lines of magnetic induction B1 and cause "inductive loss" that can be represented as an effective series resistance (ESR) RI ~ sw² [1].  A spectroscopist can reduce the heating effects of inductive loss by using the least conductive sample possible, either by reducing its hydration or salt concentration.  However, there is not much that an RF probe can do to reduce this E1 field without changing B1 as well.

As if inductive loss were not bad enough, solenoidal sample coils also create a strong electric field along their principal axis.  This field is due to the large voltage, up to several kilovolts, required to force sufficient current through the large inductance of a multi-turn solenoid.  This axial field is often described as conservative, meaning that the motion of a charge in this field conserves energy, so we give it the symbol EC: rot EC = 0.  The loss RD associated with EC depends upon the distributed capacitance between the coil and the sample.  Unlike RI, RD does not generally have a simple dependence on conductivity and frequency.  To determine the relative inductive and dielectric loss, the total ESR of a series of samples, identical except for their salinity, was measured in an NMR probe at 400 MHz.  Figure 1 illustrates that, for samples with salinity in the typical range, dielectric loss RD dominated inductive loss was the dominant effect [2].  Fortunately, the conservative field is not tied to the magnetic field, and there are a number of ways to suppress it and its associated dielectric loss.  Faraday shields inside a solenoid can be used at lower frequencies, but since they tend to reduce the self-resonance of the solenoid they are not really appropriate for high frequencies.  Directly reducing the inductance of the coil will reduce heating too.  Scroll coil is one elegant example of such approach [3].  Unfortunately, low self-resonance limits scroll use to smaller sample sizes (or to lower fields) and its low-inductance can also reduce the sensitivity for the low-frequency ¹⁵N and ¹³C channels in the multi-tuned matching network.  Different approach is needed to reduce heating while preserving sensitivity and sample size.

The Low-E Approach

In order to increase low-gamma sensitivity and reduce sample heating at the same time, we separated resonators for the high and low frequency channels, allowing each to be optimized for its individual purpose.  We combined an inner solenoid for the ¹³C or ¹⁵N channels with an outer single turn ¹H solenoid otherwise known as a loop-gap resonator (LGR).  Crossed-coil probes with good sample heating characteristics were in fact available commercially for magic angle spinning [4], but with reversed order of coils: the detection solenoid was on the outside.  In our design the observe coil is wound tightly around the sample to maximize sensitivity of detection in already dilute preparations.  An example of our approach is shown in Figure 2A. This probe affords the study of aligned membrane protein samples in a large 7.5 x 5.5 mm rectangular glass tube at 900 MHz.  The inner coil, a multi-turn rectangular solenoid, is designed to give excellent sensitivity for the 91 MHz ¹⁵N detection channel in PISEMA experiments.  The solenoid is placed inside a LGR that produces the ¹H decoupling field with excellent homogeneity and minimal conservative electric field EC (hence the name "low-E").  These probes have been fully tested at 600 and 900 MHz and have been shown to reduce sample heating by approximately 10-fold.  They are available to external scientists through the NMR user program.  Recent publications describing the design [5] and applications of these probes to protein structure determination [6-10] will be of interest to potential users.



Fig. 2.  Low-E cross coil assemblies: (A) 900 MHz ¹⁵N—¹H flat coil PISEMA probe; (B) 600 MHz ¹⁵N—¹H probe for 5 mm bicelle samples; (C) 600 MHz ¹⁹F—¹H flat coil probe for oriented samples; (D) 750 MHz 4mm MAS probe using Revolution NMR stator.

References

[1] D.G. Gadian, J. Magn. Reson. 34, 449 (1979)
[2] C. Li et al., J. Magn. Reson. 180, 51 (2006), PDF
[3] J.A. Stringer et al., J. Magn. Reson. 173, 40 (2005)
[4] F.D. Doty et al., J. Magn. Reson. 182, 239 (2006)
[5] P.L. Gor'kov et al., J. Magn. Reson. 185, 77 (2007), PDF
[6] E.Y. Chekmenev et al., J. Am. Chem. Soc. 128, 5308 (2006), PDF
[7] E.Y. Chekmenev et al., Biochim. Biophys. Acta 1758, 1359 (2006), PDF
[8] J.J. Buffy et al., Biochemistry 45, 10939 (2006), PDF
[9] N.J. Traaseth et al., Biochemistry 45, 13827 (2006)
[10] C. Li et al., J. Am. Chem. Soc. 129 (17), 5304 (2007), PDF