Biomacromolecules that are challenging for the usual structural techniques can be studied with atomic resolution by solid-state nuclear magnetic resonance. However, the paucity of >5 Å distance restraints, traditionally derived from measurements of magnetic dipole-dipole couplings between protein nuclei, is a major bottleneck that hampers such structure elucidation efforts. Here we describe a general approach that enables the rapid determination of global protein fold in the solid phase via measurements of nuclear paramagnetic relaxation enhancements (PREs) in several analogs of the protein of interest containing covalently-attached paramagnetic tags, without the use of conventional internuclear distance restraints. The method is demonstrated using six cysteine-EDTA-Cu2+ mutants of the 56-residue B1 immunoglobulin-binding domain of protein G, for which ~230 longitudinal backbone 15N PREs corresponding to ~10-20 Å distances were obtained. The mean protein fold determined in this manner agrees with the X-ray structure with a backbone atom root-mean-square deviation of 1.8 Å.

We describe a condensed data collection approach that facilitates rapid acquisition of multidimensional magic-angle spinning solid-state nuclear magnetic resonance (SSNMR) spectra of proteins by combining rapid sample spinning, optimized low-power radio frequency pulse schemes and covalently attached paramagnetic tags to enhance protein (1)H spin-lattice relaxation. Using EDTA-Cu(2+)-modified K28C and N8C mutants of the B1 immunoglobulin binding domain of protein G as models, we demonstrate that high resolution and sensitivity 2D and 3D SSNMR chemical shift correlation spectra can be recorded in as little as several minutes and several hours, respectively, for samples containing approximately 0.1-0.2 micromol of (13)C,(15)N- or (2)H,(13)C,(15)N-labeled protein. This mode of data acquisition is naturally suited toward the structural SSNMR studies of paramagnetic proteins, for which the typical (1)H longitudinal relaxation time constants are inherently a factor of at least approximately 3-4 lower relative to their diamagnetic counterparts. To illustrate this, we demonstrate the rapid site-specific determination of backbone amide (15)N longitudinal paramagnetic relaxation enhancements using a pseudo-3D SSNMR experiment based on (15)N-(13)C correlation spectroscopy, and we show that such measurements yield valuable long-range (15)N-Cu(2+) distance restraints which report on the three-dimensional protein fold.

Many structures of the proteins and protein assemblies that play central roles in fundamental biological processes and disease pathogenesis are not readily accessible via the conventional techniques of single-crystal X-ray diffraction and solution-state nuclear magnetic resonance (NMR). On the other hand, many of these challenging biological systems are suitable targets for atomic-level structural and dynamic analysis by magic-angle spinning (MAS) solid-state NMR spectroscopy, a technique that has far less stringent limitations on the molecular size and crystalline state. Over the past decade, major advances in instrumentation and methodology have prompted rapid growth in the field of biological solid-state NMR. However, despite this progress, one challenge for the elucidation of three-dimensional (3D) protein structures via conventional MAS NMR methods is the relative lack of long-distance data. Specifically, extracting unambiguous interatomic distance restraints larger than ∼5 Å from through-space magnetic dipole-dipole couplings among the protein (1)H, (13)C, and (15)N nuclei has proven to be a considerable challenge for researchers. It is possible to circumvent this problem by extending the structural studies to include several analogs of the protein of interest, intentionally modified to contain covalently attached paramagnetic tags at selected sites. In these paramagnetic proteins, the hyperfine couplings between the nuclei and unpaired electrons can manifest themselves in NMR spectra in the form of relaxation enhancements of the nuclear spins that depend on the electron-nucleus distance. These effects can be significant for nuclei located up to ∼20 Å away from the paramagnetic center. In this Account, we discuss MAS NMR structural studies of nitroxide and EDTA-Cu(2+) labeled variants of a model 56 amino acid globular protein, B1 immunoglobulin-binding domain of protein G (GB1), in the microcrystalline solid phase. We used a set of six EDTA-Cu(2+)-tagged GB1 mutants to rapidly determine the global protein fold in a de novo fashion. Remarkably, these studies required quantitative measurements of only approximately four or five backbone amide (15)N longitudinal paramagnetic relaxation enhancements per residue, in the complete absence of the usual internuclear distance restraints. Importantly, this paramagnetic solid-state NMR methodology is general and can be directly applied to larger proteins and protein complexes for which a significant fraction of the signals can be assigned in standard 2D and 3D MAS NMR chemical shift correlation spectra.

X-ray crystallography has been a useful tool in the development of site-directed spin labeling by resolving rotamers of the nitroxide spin-label side chain in a variety of α-helical environments. In this work, the crystal structure of a doubly spin-labeled N8C/K28C mutant of the B1 immunoglobulin-binding domain of protein G (GB1) was solved. The double mutant formed a domain-swapped dimer under crystallization conditions. Two rotameric states of the spin-label were resolved at the solvent-exposed α-helical site, at residue 28; these are in good agreement with rotamers previously reported for helical structures. The second site, at residue 8 on an interior β-strand, shows the presence of three distinct solvent-exposed side-chain rotamers. One of these rotamers is rarely observed within crystal structures of R1 sites and suggests that the H(α) and S(δ) hydrogen bond that is common to α-helical sites is absent at this interior β-strand residue. Variable temperature continuous wave (CW) experiments of the β-strand site showed two distinct components that were correlated to the rotameric states observed in crystallography. Interestingly, the CW data at room temperature could be fit without the use of an order parameter, which is consistent with the lack of the H(α) and S(δ) interaction. Additionally, double electron electron resonance (DEER) spectroscopy was performed on the GB1 double mutant in its monomeric form and yielded a most probable interspin distance of 25 ± 1 Å. In order to evaluate the accuracy of the measured DEER distance, the rotamers observed in the crystal structure of the domain-swapped GB1 dimer were modeled into a high-resolution structure of the wild type monomeric GB1. The distances generated in the resulting GB1 structural models match the most probable DEER distance within ~2 Å. The results are interesting as they indicate by direct experimental measurement that the rotameric states of R1 found in this crystal provide a very close match to the most probable distance measured by DEER.

Magic-angle spinning solid-state NMR measurements of (15)N longitudinal paramagnetic relaxation enhancements (PREs) in (13)C,(15)N-labeled proteins modified with Cu(2+)-chelating tags can yield multiple long-range electron-nucleus distance restraints up to ~20 Å (Nadaud et al. in J Am Chem Soc 131:8108-8120, 2009). Using the EDTA-Cu(2+) K28C mutant of B1 immunoglobulin binding domain of protein G (GB1) as a model, we investigate the effects on such measurements of intermolecular electron-nucleus couplings and intrinsic metal binding sites, both of which may potentially complicate the interpretation of PRE data in terms of the intramolecular protein fold. To quantitatively assess the influence of intermolecular (15)N-Cu(2+) interactions we have determined a nearly complete set of longitudinal (15)N PREs for a series of microcrystalline samples containing ~10, 15 and 25 mol percent of the (13)C,(15)N-labeled EDTA-Cu(2+)-tagged protein diluted in a matrix of diamagnetic natural abundance GB1. The residual intermolecular interactions were found to be minor on the whole and account for only a fraction of the relatively small but systematic deviations observed between the experimental (15)N PREs and corresponding values calculated using protein structural models for residues furthest removed from the EDTA-Cu(2+) tag. This suggests that these deviations are also caused in part by other factors not related to the protein structure, such as the presence in the protein of intrinsic secondary sites capable of binding Cu(2+) ions. To probe this issue we performed a Cu(2+) titration study for K28C-EDTA GB1 monitored by 2D (15)N-(1)H solution-state NMR, which revealed that while for Cu(2+):protein molar ratios of ≤ 1.0 Cu(2+) binds primarily to the high-affinity EDTA tag, as anticipated, at even slightly super-stoichiometric ratios the Cu(2+) ions can also associate with side-chains of aspartate and glutamate residues. This in turn is expected to lead to enhanced PREs for residues located in the vicinity of the secondary Cu(2+) binding sites, and indeed many of these residues were ones found to display the elevated longitudinal (15)N PREs in the solid phase.