Research

Many of our pulse sequence and data analysis developments have been catalyzed by the discovery that the immunoglobulin B1 binding domain of protein G (GB1) is perhaps the best model protein so far for MAS SSNMR. Although the protein G family (including GB3) is a well known subject of solution NMR, protein folding and other biophysical studies, our report of the sample preparation protocol and ¹³C and ¹⁵N chemical shift assignments (Franks et al., JACS 2005) was its first use for SSNMR, and this study has been cited already 24 times in two years. We have more recently reported several new polymorphs, all of which yield high-resolution spectra with distinct sets of chemical shifts (Frericks et al., submitted). Beyond the practical advantages of high thermostability, high production yield, a rigid backbone, high microscopic order and excellent chemical shift dispersion, GB1 provides opportunities for fundamental studies of protein structure and dynamics that can be readily compared with other experimental techniques. Moreover, GB1 is sufficiently small that rigorous computational modeling can and has been performed. Among those scientists with broader interest in our fundamental spectroscopic results are solution NMR spectroscopists developing improved spin relaxation analysis methods, crystallographers attempting to distinguish static from dynamic disorder, and theoreticians validating calculations of electronic structure.

Our pulse sequence design efforts have included both the quantitative measurement of tensor parameters in highly-¹³C,¹⁵N-enriched proteins, as well as new chemical shift assignment methods. Tensor measurements have extended from existing methods to developing 3D versions of pulse sequences to correlate two isotropic chemical shift dimensions with a third, anisotropic dimension. We employed this approach to report the first chemical shift tensor measurements systematically through a protein by chemical shift lineshape (Wylie et al., JACS 2005, JPC B 2006) and sideband analysis (Wylie et al., JACS 2007), as well as dipole vector orientations (Franks et al., JACS 2006) that have proved especially valuable for refining high-resolution structure (Franks et al., submitted). In terms of chemical shift assignment, we have developed numerous new methods for more effective selective pulses (Li et al., JMR 2006), homonuclear 3D correlation schemes with high digital resolution (Zhou et al., J. Biomol. NMR 2006), scalar-based correlation methods (Chen et al., JACS 2006 and JACS 2007) and the first 4D chemical shift correlation methods applied to solid proteins (Franks et al., J. Biomol. NMR 2007).   

An ongoing theme of these efforts is to establish new benchmarks for high-resolution structure determination. We set out to establish whether any fundamental issues impede SSNMR from solving structures at a resolution comparable to a good crystal or solution NMR structure. Although the ability of SSNMR to address disordered systems has been well recognized, it was less clear whether SSNMR could solve structures of full proteins truly at atomic resolution. We have solved the structure of GB1 at better than 0.4 Å RMSD backbone resolution, the highest resolution protein structure yet determined by SSNMR, signifying that SSNMR can be competitive with x-ray diffraction and solution NMR at least in terms of resolution (Franks et al., submitted).