Uniform isotope labeling of a eukaryotic seven-transmembrane helical protein in yeast enables high-resolution solid-state NMR studies in the lipid environment

Overexpression of isotope-labeled multi-spanning eukaryotic membrane proteins for structural NMR studies is often challenging. On the one hand, difficulties with achieving proper folding, membrane insertion, and native-like post-translational modifications frequently disqualify bacterial expression systems. On the other hand, eukaryotic cell cultures can be prohibitively expensive. One of the viable alternatives, successfully used for producing proteins for solution NMR studies, is yeast expression systems, particularly Pichia pastoris. We report on successful implementation and optimization of isotope labeling protocols, previously used for soluble secreted proteins, to produce homogeneous samples of a eukaryotic seven-transmembrane helical protein, rhodopsin from Leptosphaeria maculans. Even in shake-flask cultures, yields exceeded 5 mg of purified uniformly ¹³C,¹⁵N-labeled protein per liter of culture. The protein was stable (at least several weeks at 5°C) and functionally active upon reconstitution into lipid membranes at high protein-to-lipid ratio required for solid-state NMR. The samples gave high-resolution ¹³C and ¹⁵N solid-state magic angle spinning NMR spectra, amenable to a detailed structural analysis. We believe that similar protocols can be adopted for challenging mammalian targets, which often resist characterization by other structural methods.     

Solid-state NMR structures of integral membrane proteins

Abstract

Solid-state NMR is unique for its ability to obtain three-dimensional structures and to measure atomic-resolution structural and dynamic information for membrane proteins in native lipid bilayers. An increasing number and complexity of integral membrane protein structures have been determined by solid-state NMR using two main methods. Oriented sample solid-state NMR uses macroscopically aligned lipid bilayers to obtain orientational restraints that define secondary structure and global fold of embedded peptides and proteins and their orientation and topology in lipid bilayers. Magic angle spinning (MAS) solid-state NMR uses unoriented rapidly spinning samples to obtain distance and torsion angle restraints that define tertiary structure and helix packing arrangements. Details of all current protein structures are described, highlighting developments in experimental strategy and other technological advancements. Some structures originate from combining solid- and solution-state NMR information and some have used solid-state NMR to refine X-ray crystal structures. Solid-state NMR has also validated the structures of proteins determined in different membrane mimetics by solution-state NMR and X-ray crystallography and is therefore complementary to other structural biology techniques. By continuing efforts in identifying membrane protein targets and developing expression, isotope labelling and sample preparation strategies, probe technology, NMR experiments, calculation and modelling methods and combination with other techniques, it should be feasible to determine the structures of many more membrane proteins of biological and biomedical importance using solid-state NMR. This will provide three-dimensional structures and atomic-resolution structural information for characterising ligand and drug interactions, dynamics and molecular mechanisms of membrane proteins under physiological lipid bilayer conditions.     

Solution structure of tRNA<Superscript>Val</Superscript> from refinement of homology model against residual dipolar coupling and SAXS data

A procedure is presented for refinement of a homology model of E. coli tRNA^Val, originally based on the X-ray structure of yeast tRNA^Phe, using experimental residual dipolar coupling (RDC) and small angle X-ray scattering (SAXS) data. A spherical sampling algorithm is described for refinement against SAXS data that does not require a globbic approximation, which is particularly important for nucleic acids where such approximations are less appropriate. Substantially higher speed of the algorithm also makes its application favorable for proteins. In addition to the SAXS data, the structure refinement employed a sparse set of NMR data consisting of 24 imino N–H^N RDCs measured with Pf1 phage alignment, and 20 imino N–H^N RDCs obtained from magnetic field dependent alignment of tRNA^Val. The refinement strategy aims to largely retain the local geometry of the 58% identical tRNA^Phe by ensuring that the atomic coordinates for short, overlapping segments of the ribose-phosphate backbone and the conserved base pairs remain close to those of the starting model. Local coordinate restraints are enforced using the non-crystallographic symmetry (NCS) term in the XPLOR-NIH or CNS software package, while still permitting modest movements of adjacent segments. The RDCs mainly drive the relative orientation of the helical arms, whereas the SAXS restraints ensure an overall molecular shape compatible with experimental scattering data. The resulting structure exhibits good cross-validation statistics (jack-knifed Q _free = 14% for the Pf1 RDCs, compared to 25% for the starting model) and exhibits a larger angle between the two helical arms than observed in the X-ray structure of tRNA^Phe, in agreement with previous NMR-based tRNA^Val models. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Determination of structural topology of a membrane protein in lipid bilayers using polarization optimized experiments (POE) for static and MAS solid state NMR spectroscopy

The low sensitivity inherent to both the static and magic angle spinning techniques of solid-state NMR (ssNMR) spectroscopy has thus far limited the routine application of multidimensional experiments to determine the structure of membrane proteins in lipid bilayers. Here, we demonstrate the advantage of using a recently developed class of experiments, polarization optimized experiments, for both static and MAS spectroscopy to achieve higher sensitivity and substantial time-savings for 2D and 3D experiments. We used sarcolipin, a single pass membrane protein, reconstituted in oriented bicelles (for oriented ssNMR) and multilamellar vesicles (for MAS ssNMR) as a benchmark. The restraints derived by these experiments are then combined into a hybrid energy function to allow simultaneous determination of structure and topology. The resulting structural ensemble converged to a helical conformation with a backbone RMSD ~0.44 Å, a tilt angle of 24° ± 1°, and an azimuthal angle of 55° ± 6°. This work represents a crucial first step toward obtaining high-resolution structures of large membrane proteins using combined multidimensional oriented solid-state NMR and magic angle spinning solid-state NMR.     

Analysis of the amide <Superscript>15</Superscript>N chemical shift tensor of the C<Subscript>α</Subscript> tetrasubstituted constituent of membrane-active peptaibols,the α-aminoisobutyric acid residue,compared to those of di- and tri-substituted proteinogenic amino acid residues

In protein NMR spectroscopy the chemical shift provides important information for the assignment of residues and a first structural evaluation of dihedral angles. Furthermore, angular restraints are obtained from oriented samples by solution and solid-state NMR spectroscopic approaches. Whereas the anisotropy of chemical shifts, quadrupolar couplings and dipolar interactions have been used to determine the structure, dynamics and topology of oriented membrane polypeptides using solid-state NMR spectroscopy similar concepts have been introduced to solution NMR through the measurements of residual dipolar couplings. The analysis of ¹⁵N chemical shift spectra depends on the accuracy of the chemical shift tensors. When investigating alamethicin and other peptaibols, i.e. polypeptides rich in α-aminoisobutyric acid (Aib), the ¹⁵N chemical shift tensor of this C_α-tetrasubstituted amino acid exhibits pronounced differences when compared to glycine, alanine and other proteinogenic residues. Here we present an experimental investigation on the ¹⁵N amide Aib tensor of N-acetyl-Aib-OH and for the Aib residues within peptaibols. Furthermore, a statistical analysis of the tensors published for di- (glycine) and tri-substituted residues has been performed, where for the first time the published data sets are compiled using a common reference. The size of the isotropic chemical shift and main tensor elements follows the order di- < tri- < tetra-substituted amino acids. A ¹⁵N chemical shift-¹H-¹⁵N dipolar coupling correlation NMR spectrum of alamethicin is used to evaluate the consequences of variations in the main tensor elements for the structural analysis of this membrane peptide.     

Characterisation of Schiff base and chromophore in green proteorhodopsin by solid-state NMR

The proteorhodopsin family consists of hundreds of homologous retinal containing membrane proteins found in bacteria in the photic zone of the oceans. They are colour tuned to their environment and act as light-driven proton pumps with a potential energetic and regulatory function. Precise structural details are still unknown. Here, the green proteorhodopsin variant has been selected for a chemical shift analysis of retinal and Schiff base by solid-state NMR. Our data show that the chromophore exists in mainly all-trans configuration in the proteorhodopsin ground state. The optical absorption maximum together with retinal and Schiff base chemical shifts indicate a strong interaction network between chromophore and opsin. Electronic supplementary material The online version of this article (doi:) contains supplementary material, which is available to authorized users. Mark Lorch and Andreas C. Woerner contributed equally to this work.     

High-resolution 3D structure determination of kaliotoxin by solid-state NMR spectroscopy

High-resolution solid-state NMR spectroscopy can provide structural information of proteins that cannot be studied by X-ray crystallography or solution NMR spectroscopy. Here we demonstrate that it is possible to determine a protein structure by solid-state NMR to a resolution comparable to that by solution NMR. Using an iterative assignment and structure calculation protocol, a large number of distance restraints was extracted from (1)H/(1)H mixing experiments recorded on a single uniformly labeled sample under magic angle spinning conditions. The calculated structure has a coordinate precision of 0.6 A and 1.3 A for the backbone and side chain heavy atoms, respectively, and deviates from the structure observed in solution. The approach is expected to be applicable to larger systems enabling the determination of high-resolution structures of amyloid or membrane proteins.     

A Practical Implicit Membrane Potential for NMR Structure Calculations of Membrane Proteins

The highly anisotropic environment of the lipid bilayer membrane imposes significant constraints on the structures and functions of membrane proteins. However, NMR structure calculations typically use a simple repulsive potential that neglects the effects of solvation and electrostatics, because explicit atomic representation of the solvent and lipid molecules is computationally expensive and impractical for routine NMR-restrained calculations that start from completely extended polypeptide templates. Here, we describe the extension of a previously described implicit solvation potential, eefxPot, to include a membrane model for NMR-restrained calculations of membrane protein structures in XPLOR-NIH. The key components of eefxPot are an energy term for solvation free energy that works together with other nonbonded energy functions, a dedicated force field for conformational and nonbonded protein interaction parameters, and a membrane function that modulates the solvation free energy and dielectric screening as a function of the atomic distance from the membrane center, relative to the membrane thickness. Initial results obtained for membrane proteins with structures determined experimentally in lipid bilayer membranes show that eefxPot affords significant improvements in structural quality, accuracy, and precision. Calculations with eefxPot are straightforward to implement and can be used to both fold and refine structures, as well as to run unrestrained molecular-dynamics simulations. The potential is entirely compatible with the full range of experimental restraints measured by various techniques. Overall, it provides a useful and practical way to calculate membrane protein structures in a physically realistic environment.     

Solid-State NMR-Restrained Ensemble Dynamics of a Membrane Protein in Explicit Membranes

Solid-state NMR has been used to determine the structures of membrane proteins in native-like lipid bilayer environments. Most structure calculations based on solid-state NMR observables are performed using simulated annealing with restrained molecular dynamics and an energy function, where all nonbonded interactions are represented by a single, purely repulsive term with no contributions from van der Waals attractive, electrostatic, or solvation energy. To our knowledge, this is the first application of an ensemble dynamics technique performed in explicit membranes that uses experimental solid-state NMR observables to obtain the refined structure of a membrane protein together with information about its dynamics and its interactions with lipids. Using the membrane-bound form of the fd coat protein as a model membrane protein and its experimental solid-state NMR data, we performed restrained ensemble dynamics simulations with different ensemble sizes in explicit membranes. For comparison, a molecular dynamics simulation of fd coat protein was also performed without any restraints. The average orientation of each protein helix is similar to a structure determined by traditional single-conformer approaches. However, their variations are limited in the resulting ensemble of structures with one or two replicas, as they are under the strong influence of solid-state NMR restraints. Although highly consistent with all solid-state NMR observables, the ensembles of more than two replicas show larger orientational variations similar to those observed in the molecular dynamics simulation without restraints. In particular, in these explicit membrane simulations, Lys⁴⁰, residing at the C-terminal side of the transmembrane helix, is observed to cause local membrane curvature. Therefore, compared to traditional single-conformer approaches in implicit environments, solid-state NMR restrained ensemble simulations in explicit membranes readily characterize not only protein dynamics but also protein-lipid interactions in detail.     

Structure determination of membrane proteins in five easy pieces

Rotational Alignment (RA) solid-state NMR provides the basis for a general method for determining the structures of membrane proteins in phospholipid bilayers under physiological conditions. Membrane proteins are high priority targets for structure determination, and are challenging for existing experimental methods. Because membrane proteins reside in liquid crystalline phospholipid bilayer membranes it is important to study them in this type of environment. The RA solid-state NMR approach we have developed can be summarized in five steps, and incorporates methods of molecular biology, biochemistry, sample preparation, the implementation of NMR experiments, and structure calculations. It relies on solid-state NMR spectroscopy to obtain high-resolution spectra and residue-specific structural restraints for membrane proteins that undergo rotational diffusion around the membrane normal, but whose mobility is otherwise restricted by interactions with the membrane phospholipids. High resolution spectra of membrane proteins alone and in complex with other proteins and ligands set the stage for structure determination and functional studies of these proteins in their native, functional environment.     

Backbone structure of a small helical integral membrane protein: A unique structural characterization

The structural characterization of small integral membrane proteins pose a significant challenge for structural biology because of the multitude of molecular interactions between the protein and its heterogeneous environment. Here, the three‐dimensional backbone structure of Rv1761c from Mycobacterium tuberculosis has been characterized using solution NMR spectroscopy and dodecylphosphocholine (DPC) micelles as a membrane mimetic environment. This 127 residue single transmembrane helix protein has a significant (10 kDa) C‐terminal extramembranous domain. Five hundred and ninety distance, backbone dihedral, and orientational restraints were employed resulting in a 1.16 Å rmsd backbone structure with a transmembrane domain defined at 0.40 Å. The structure determination approach utilized residual dipolar coupling orientation data from partially aligned samples, long‐range paramagnetic relaxation enhancement derived distances, and dihedral restraints from chemical shift indices to determine the global fold. This structural model of Rv1761c displays some influences by the membrane mimetic illustrating that the structure of these membrane proteins is dictated by a combination of the amino acid sequence and the protein's environment. These results demonstrate both the efficacy of the structural approach and the necessity to consider the biophysical properties of membrane mimetics when interpreting structural data of integral membrane proteins and, in particular, small integral membrane proteins.     

Direct methods and residue type specific isotope labeling in NMR structure determination and model-driven sequential assignment

Direct methods in NMR based structure determination start from an unassigned ensemble of unconnected gaseous hydrogen atoms. Under favorable conditions they can produce low resolution structures of proteins. Usually a prohibitively large number of NOEs is required, to solve a protein structure ab-initio, but even with a much smaller set of distance restraints low resolution models can be obtained which resemble a protein fold. One problem is that at such low resolution and in the absence of a force field it is impossible to distinguish the correct protein fold from its mirror image. In a hybrid approach these ambiguous models have the potential to aid in the process of sequential backbone chemical shift assignment when ¹³C^β and ¹³C′ shifts are not available for sensitivity reasons. Regardless of the overall fold they enhance the information content of the NOE spectra. These, combined with residue specific labeling and minimal triple-resonance data using ¹³C^α connectivity can provide almost complete sequential assignment. Strategies for residue type specific labeling with customized isotope labeling patterns are of great advantage in this context. Furthermore, this approach is to some extent error-tolerant with respect to data incompleteness, limited precision of the peak picking, and structural errors caused by misassignment of NOEs. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

High-resolution membrane protein structure by joint calculations with solid-state NMR and X-ray experimental data

X-ray diffraction and nuclear magnetic resonance spectroscopy (NMR) are the staple methods for revealing atomic structures of proteins. Since crystals of biomolecular assemblies and membrane proteins often diffract weakly and such large systems encroach upon the molecular tumbling limit of solution NMR, new methods are essential to extend structures of such systems to high resolution. Here we present a method that incorporates solid-state NMR restraints alongside of X-ray reflections to the conventional model building and refinement steps of structure calculations. Using the 3.7 Å crystal structure of the integral membrane protein complex DsbB-DsbA as a test case yielded a significantly improved backbone precision of 0.92 Å in the transmembrane region, a 58% enhancement from using X-ray reflections alone. Furthermore, addition of solid-state NMR restraints greatly improved the overall quality of the structure by promoting 22% of DsbB transmembrane residues into the most favored regions of Ramachandran space in comparison to the crystal structure. This method is widely applicable to any protein system where X-ray data are available, and is particularly useful for the study of weakly diffracting crystals.     

Determining the depth of insertion of dynamically invisible membrane peptides by gel-phase 1H spin diffusion heteronuclear correlation NMR

Solid-state NMR determination of the depth of insertion of membrane peptides and proteins has so far utilized ¹H spin diffusion and paramagnetic relaxation enhancement experiments, which are typically conducted in the liquid-crystalline phase of the lipid bilayer. For membrane proteins or peptide assemblies that undergo intermediate-timescale motion in the liquid-crystalline membrane, these approaches are no longer applicable because the protein signals are broadened beyond detection. Here we show that the rigid-solid HETCOR experiment, with an additional spin diffusion period, can be used to determine the depth of proteins in gel-phase lipid membranes, where the proteins are immobilized to give high-intensity solid-state NMR spectra. Demonstration on two membrane peptides with known insertion depths shows that well-inserted peptides give rise to high lipid cross peak intensities and low water cross peaks within a modest spin diffusion mixing time, while surface-bound peptides have higher water than lipid cross peaks. Furthermore, well-inserted membrane peptides have nearly identical ¹H cross sections as the lipid chains, indicating equilibration of the peptide and lipid magnetization. Using this approach, we measured the membrane topology of the α-helical fusion peptide of the paramyxovirus, PIV5, in the anionic POPC/POPG membrane, in which the peptide undergoes intermediate-timescale motion at physiological temperature. The gel-phase HETCOR spectra indicate that the α-helical fusion peptide is well inserted into the POPC/POPG bilayer, spanning both leaflets. This insertion motif gives insight into the functional role of the α-helical PIV5 fusion peptide in virus-cell membrane fusion.     

Structure and Dynamics of the Membrane-Bound Form of Pf1 Coat Protein: Implications of Structural Rearrangement for Virus Assembly

The three-dimensional structure of the membrane-bound form of the major coat protein of Pf1 bacteriophage was determined in phospholipid bilayers using orientation restraints derived from both solid-state and solution NMR experiments. In contrast to previous structures determined solely in detergent micelles, the structure in bilayers contains information about the spatial arrangement of the protein within the membrane, and thus provides insights to the bacteriophage assembly process from membrane-inserted to bacteriophage-associated protein. Comparisons between the membrane-bound form of the coat protein and the previously determined structural form found in filamentous bacteriophage particles demonstrate that it undergoes a significant structural rearrangement during the membrane-mediated virus assembly process. The rotation of the transmembrane helix (Q16-A46) around its long axis changes dramatically (by 160°) to obtain the proper alignment for packing in the virus particles. Furthermore, the N-terminal amphipathic helix (V2-G17) tilts away from the membrane surface and becomes parallel with the transmembrane helix to form one nearly continuous long helix. The spectra obtained in glass-aligned planar lipid bilayers, magnetically aligned lipid bilayers (bicelles), and isotropic lipid bicelles reflect the effects of backbone motions and enable the backbone dynamics of the N-terminal helix to be characterized. Only resonances from the mobile N-terminal helix and the C-terminus (A46) are observed in the solution NMR spectra of the protein in isotropic q > 1 bicelles, whereas only resonances from the immobile transmembrane helix are observed in the solid-state ¹H/¹⁵N-separated local field spectra in magnetically aligned bicelles. The N-terminal helix and the hinge that connects it to the transmembrane helix are significantly more dynamic than the rest of the protein, thus facilitating structural rearrangement during bacteriophage assembly.     

Assigning large proteins in the solid state: a MAS NMR resonance assignment strategy using selectively and extensively <Superscript>13</Superscript>C-labelled proteins

In recent years, solid-state magic-angle spinning nuclear magnetic resonance spectroscopy (MAS NMR) has been growing into an important technique to study the structure of membrane proteins, amyloid fibrils and other protein preparations which do not form crystals or are insoluble. Currently, a key bottleneck is the assignment process due to the absence of the resolving power of proton chemical shifts. Particularly for large proteins (approximately >150 residues) it is difficult to obtain a full set of resonance assignments. In order to address this problem, we present an assignment method based upon samples prepared using [1,3-¹³C]- and [2-¹³C]-glycerol as the sole carbon source in the bacterial growth medium (so-called selectively and extensively labelled protein). Such samples give rise to higher quality spectra than uniformly [¹³C]-labelled protein samples, and have previously been used to obtain long-range restraints for use in structure calculations. Our method exploits the characteristic cross-peak patterns observed for the different amino acid types in ¹³C-¹³C correlation and 3D NCACX and NCOCX spectra. An in-depth analysis of the patterns and how they can be used to aid assignment is presented, using spectra of the chicken α-spectrin SH3 domain (62 residues), αB-crystallin (175 residues) and outer membrane protein G (OmpG, 281 residues) as examples. Using this procedure, over 90% of the Cα, Cβ, C′ and N resonances in the core domain of αB-crystallin and around 73% in the flanking domains could be assigned (excluding 24 residues at the extreme termini of the protein). Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Yeast-expressed human membrane protein aquaporin-1 yields excellent resolution of solid-state MAS NMR spectra

One of the biggest challenges in solid-state NMR studies of membrane proteins is to obtain a homogeneous natively folded sample giving high spectral resolution sufficient for structural studies. Eukaryotic membrane proteins are especially difficult and expensive targets in this respect. Methylotrophic yeast Pichia pastoris is a reliable producer of eukaryotic membrane proteins for crystallography and a promising economical source of isotopically labeled proteins for NMR. We show that eukaryotic membrane protein human aquaporin 1 can be doubly (¹³C/¹⁵N) isotopically labeled in this system and functionally reconstituted into phospholipids, giving excellent resolution of solid-state magic angle spinning NMR spectra.     

Solid-state NMR studies of the structure and mechanisms of proteins

Magic-angle spinning solid-state NMR experiments are well suited to investigating the structures and mechanisms of important proteins that are inaccessible to X-ray crystallography and solution NMR spectroscopy, including membrane proteins and disease-related protein aggregates. Good progress has been made in the development of methods for the complete structure determination of small (<20 kDa) solid proteins using uniformly 13C, 15N-labeled samples. Studies of selectively labeled proteins focusing on labeled active sites have yielded insights into the mechanisms of enzymes and of membrane proteins involved in energy and signal transduction. Studies of selectively labeled synthetic peptides have yielded structural models for biomedically important systems, including amyloid fibrils and surface-associated peptides involved in biomineralization and cell adhesion. Novel NMR and biochemical methods are being developed to target solid-state NMR experiments within large proteins and whole cells. These approaches are being used to investigate mechanisms of transmembrane signaling by membrane receptors and to characterize binding interactions between antibiotics and bacterial cell walls. Thus, solid-state NMR is proving to be a valuable biophysical tool for probing structure and dynamics in a wide range of biomolecules.     

Investigating molecular recognition and biological function at interfaces using piscidins, antimicrobial peptides from fish

We studied amidated and non-amidated piscidins 1 and 3, amphipathic cationic antimicrobial peptides from fish, to characterize functional and structural similarities and differences between these peptides and better understand the structural motifs involved in biological activity and functional diversity among amidated and non-amidated isoforms. Antimicrobial and hemolytic assays were carried out to assess their potency and toxicity, respectively. Site-specific high-resolution solid-state NMR orientational restraints were obtained from ¹⁵N-labeled amidated and non-amidated piscidins 1 and 3 in the presence of hydrated oriented lipid bilayers. Solid-state NMR and circular dichroism results indicate that the peptides are α-helical and oriented parallel to the membrane surface. This orientation was expected since peptide-lipid interactions are enhanced at the water-bilayer interface for amphipathic cationic antimicrobial peptides. ¹⁵N solid-state NMR performed on oriented samples demonstrate that piscidin experiences fast, large amplitude backbone motions around an axis parallel to the bilayer normal. Under the conditions tested here, piscidin 1 was confirmed to be more antimicrobially potent than piscidin 3 and antimicrobial activity was not affected by amidation. In light of functional and structural similarities between piscidins 1 and 3, we propose that their topology and fast dynamics are related to their mechanism of action.     

De novo protein structure generation from incomplete chemical shift assignments

NMR chemical shifts provide important local structural information for proteins. Consistent structure generation from NMR chemical shift data has recently become feasible for proteins with sizes of up to 130 residues, and such structures are of a quality comparable to those obtained with the standard NMR protocol. This study investigates the influence of the completeness of chemical shift assignments on structures generated from chemical shifts. The Chemical-Shift-Rosetta (CS-Rosetta) protocol was used for de novo protein structure generation with various degrees of completeness of the chemical shift assignment, simulated by omission of entries in the experimental chemical shift data previously used for the initial demonstration of the CS-Rosetta approach. In addition, a new CS-Rosetta protocol is described that improves robustness of the method for proteins with missing or erroneous NMR chemical shift input data. This strategy, which uses traditional Rosetta for pre-filtering of the fragment selection process, is demonstrated for two paramagnetic proteins and also for two proteins with solid-state NMR chemical shift assignments. Electronic supplementary material  The online version of this article (doi:) contains supplementary material, which is available to authorized users.