Measurement of (1)H-(15)N and (1)H-(13)C residual dipolar couplings in nucleic acids from TROSY intensities

Analogous to the recently introduced ARTSY method for measurement of one-bond (1)H-(15)N residual dipolar couplings (RDCs) in large perdeuterated proteins, we introduce methods for measurement of base (13)C-(1)H and (15)N-(1)H RDCs in protonated nucleic acids. Measurements are based on quantitative analysis of intensities in (1)H-(15)N and (13)C-(1)H TROSY-HSQC spectra, and are illustrated for a 71-nucleotide adenine riboswitch. Results compare favorably with those of conventional frequency-based measurements in terms of completeness and convenience of use. The ARTSY method derives the size of the coupling from the ratio of intensities observed in two TROSY-HSQC spectra recorded with different dephasing delays, thereby minimizing potential resonance overlap problems. Precision of the RDC measurements is limited by the signal-to-noise ratio, S/N, achievable in the 2D TROSY-HSQC reference spectrum, and is approximately given by 30/(S/N) Hz for (15)N-(1)H and 65/(S/N) Hz for (13)C-(1)H. The signal-to-noise ratio of both (1)H-(15)N and (1)H-(13)C spectra greatly benefits when water magnetization during the experiments is not perturbed, such that rapid magnetization transfer from bulk water to the nucleic acid, mediated by rapid amino and hydroxyl hydrogen exchange coupled with (1)H-(1)H NOE transfer, allows for fast repetition of the experiment. RDCs in the mutated helix 1 of the riboswitch are compatible with nucleotide-specifically modeled, idealized A-form geometry and a static orientation relative to the helix 2/3 pair, which differs by ca 6° relative to the X-ray structure of the native riboswitch.

Whole-body rocking motion of a fusion peptide in lipid bilayers from size-dispersed 15N NMR relaxation

Biological membranes present a highly fluid environment, and integration of proteins within such membranes is itself highly dynamic: proteins diffuse laterally within the plane of the membrane and rotationally about the normal vector of this plane. We demonstrate that whole-body motions of proteins within a lipid bilayer can be determined from NMR (15)N relaxation rates collected for different-sized bicelles. The importance of membrane integration and interaction is particularly acute for proteins and peptides that function on the membrane itself, as is the case for pore-forming and fusion-inducing proteins. For the influenza hemagglutinin fusion peptide, which lies on the surface of membranes and catalyzes the fusion of membranes and vesicles, we found large-amplitude, rigid-body wobbling motions on the nanosecond time scale relative to the lipid bilayer. This behavior complements prior analyses where data were commonly interpreted in terms of a static oblique angle of insertion for the fusion peptide with respect to the membrane. Quantitative disentanglement of the relative motions of two interacting objects by systematic variation of the size of one is applicable to a wide range of systems beyond protein-membrane interactions.

Structural discrimination in small molecules by accurate measurement of long-range proton-carbon NMR residual dipolar couplings

Accurate measurement of long-range CH residual dipolar couplings (RDCs) (2D CH and 3D CH) by a new selective J -scaled HSQC experiment significantly improves the structural discrimination power of RDCs in small molecules with multiple stereocenters. The extraction of the long-range couplings is clean and straightforward, and in most cases yields the sign of the RDC too. The experiment is demonstrated with 10-epi-8-deoxycumambrin B, a tricyclic natural compound with five chiral centers.

Determination of the structures of symmetric protein oligomers from NMR chemical shifts and residual dipolar couplings

Symmetric protein dimers, trimers, and higher-order cyclic oligomers play key roles in many biological processes. However, structural studies of oligomeric systems by solution NMR can be difficult due to slow tumbling of the system and the difficulty in identifying NOE interactions across protein interfaces. Here, we present an automated method (RosettaOligomers) for determining the solution structures of oligomeric systems using only chemical shifts, sparse NOEs, and domain orientation restraints from residual dipolar couplings (RDCs) without a need for a previously determined structure of the monomeric subunit. The method integrates previously developed Rosetta protocols for solving the structures of monomeric proteins using sparse NMR data and for predicting the structures of both nonintertwined and intertwined symmetric oligomers. We illustrated the performance of the method using a benchmark set of nine protein dimers, one trimer, and one tetramer with available experimental data and various interface topologies. The final converged structures are found to be in good agreement with both experimental data and previously published high-resolution structures. The new approach is more readily applicable to large oligomeric systems than conventional structure-determination protocols, which often require a large number of NOEs, and will likely become increasingly relevant as more high-molecular weight systems are studied by NMR.

Helical hairpin structure of influenza hemagglutinin fusion peptide stabilized by charge-dipole interactions between the N-terminal amino group and the second helix

The fusion domain of the influenza coat protein hemagglutinin HA2, bound to dodecyl phosphocholine micelles, was recently shown to adopt a structure consisting of two antiparallel α-helices, packed in an exceptionally tight hairpin configuration. Four interhelical H(α) to C═O aliphatic H-bonds were identified as factors stabilizing this fold. Here, we report evidence for an additional stabilizing force: a strong charge-dipole interaction between the N-terminal Gly(1) amino group and the dipole moment of helix 2. pH titration of the amino-terminal (15)N resonance, using a methylene-TROSY-based 3D NMR experiment, and observation of Gly(1 13)C' show a strongly elevated pK = 8.8, considerably higher than expected for an N-terminal amino group in a lipophilic environment. Chemical shifts of three C-terminal carbonyl carbons of helix 2 titrate with the protonation state of Gly(1)-N, indicative of a close proximity between the N-terminal amino group and the axis of helix 2, providing an optimal charge-dipole stabilization of the antiparallel hairpin fold. pK values of the side-chain carboxylate groups of Glu(11) and Asp(19) are higher by about 1 and 0.5 unit, respectively, than commonly seen for solvent-exposed side chains in water-soluble proteins, indicative of dielectric constants of ε = ∼30 (Glu(11)) and ∼60 (Asp(19)), placing these groups in the headgroup region of the phospholipid micelle.

Improved fitting of solution X-ray scattering data to macromolecular structures and structural ensembles by explicit water modeling

A new procedure, AXES, is introduced for fitting small-angle X-ray scattering (SAXS) data to macromolecular structures and ensembles of structures. By using explicit water models to account for the effect of solvent, and by restricting the adjustable fitting parameters to those that dominate experimental uncertainties, including sample/buffer rescaling, detector dark current, and, within a narrow range, hydration layer density, superior fits between experimental high resolution structures and SAXS data are obtained. AXES results are found to be more discriminating than standard Crysol fitting of SAXS data when evaluating poorly or incorrectly modeled protein structures. AXES results for ensembles of structures previously generated for ubiquitin show improved fits over fitting of the individual members of these ensembles, indicating these ensembles capture the dynamic behavior of proteins in solution.

Measuring rapid hydrogen exchange in the homodimeric 36 kDa HIV-1 integrase catalytic core domain

Measurements of rapid hydrogen exchange (HX) of water with protein amide sites contain valuable information on protein structure and function, but current NMR methods for measuring HX rates are limited in their applicability to large protein systems. An alternate method for measuring rapid HX is presented that is well-suited for larger proteins, and we apply the method to the deuterated, homodimeric 36 kDa HIV-1 integrase catalytic core domain (CCD). Using long mixing times for water-amide magnetization exchange at multiple pH values, HX rates spanning more than four orders of magnitude were measured, as well as NOE cross-relaxation rates to nearby exchangeable protons. HX protection factors for the CCD are found to be large (>10(4)) for residues along the dimer interface, but much smaller in many other regions. Notably, the catalytic helix (residues 152-167) exhibits low HX protection at both ends, indicative of fraying at both termini as opposed to just the N-terminal end, as originally thought. Residues in the LEDGF/p75 binding pocket also show marginal stability, with protection factors in the 10-100 range (∼1.4-2.7 kcal/mol). Additionally, elevated NOE cross-relaxation rates are identified and, as expected, correspond to proximity of the amide proton to a rapidly exchanging proton, typically from an OH side chain. Indirect NOE transfer between H(2) O and the amide proton of I141, a residue in the partially disordered active site of the enzyme, suggests its proximity to the side chain of S147, an interaction seen in the DNA-bound form for a homologous integrase.

The impact of hydrogen bonding on amide 1H chemical shift anisotropy studied by cross-correlated relaxation and liquid crystal NMR spectroscopy

Site-specific ¹H chemical shift anisotropy (CSA) tensors have been derived for the well-ordered backbone amide moieties in the B3 domain of protein G (GB3). Experimental input data include residual chemical shift anisotropy (RCSA), measured in six mutants that align differently relative to the static magnetic field when dissolved in a liquid crystalline Pf1 suspension, and cross-correlated relaxation rates between the ¹H^N CSA tensor and either the ¹H−¹⁵N, the ¹H−¹³C′, or the ¹H−¹³C^α dipolar interactions. Analyses with the assumption that the ¹H^N CSA tensor is symmetric with respect to the peptide plane (three-parameter fit) or without this premise (five-parameter fit) yield very similar results, confirming the robustness of the experimental input data, and that, to a good approximation, one of the principal components orients orthogonal to the peptide plane. ¹H^N CSA tensors are found to deviate strongly from axial symmetry, with the most shielded tensor component roughly parallel to the N−H vector, and the least shielded component orthogonal to the peptide plane. DFT calculations on pairs of N-methyl acetamide and acetamide in H-bonded geometries taken from the GB3 X-ray structure correlate with experimental data and indicate that H-bonding effects dominate variations in the ¹H^N CSA. Using experimentally derived ¹H^N CSA tensors, the optimal relaxation interference effect needed for narrowest ¹H^N TROSY line widths is found at ∼1200 MHz.

Facile measurement of ¹H-¹5N residual dipolar couplings in larger perdeuterated proteins

We present a simple method, ARTSY, for extracting ¹J(NH) couplings and ¹H-¹⁵N RDCs from an interleaved set of two-dimensional ¹H-¹⁵N TROSY-HSQC spectra, based on the principle of quantitative J correlation. The primary advantage of the ARTSY method over other methods is the ability to measure couplings without scaling peak positions or altering the narrow line widths characteristic of TROSY spectra. Accuracy of the method is demonstrated for the model system GB3. Application to the catalytic core domain of HIV integrase, a 36 kDa homodimer with unfavorable spectral characteristics, demonstrates its practical utility. Precision of the RDC measurement is limited by the signal-to-noise ratio, S/N, achievable in the 2D TROSY-HSQC spectrum, and is approximately given by 30/(S/N) Hz.

SPARTA+: a modest improvement in empirical NMR chemical shift prediction by means of an artificial neural network

NMR chemical shifts provide important local structural information for proteins and are key in recently described protein structure generation protocols. We describe a new chemical shift prediction program, SPARTA+, which is based on artificial neural networking. The neural network is trained on a large carefully pruned database, containing 580 proteins for which high-resolution X-ray structures and nearly complete backbone and (13)C(beta) chemical shifts are available. The neural network is trained to establish quantitative relations between chemical shifts and protein structures, including backbone and side-chain conformation, H-bonding, electric fields and ring-current effects. The trained neural network yields rapid chemical shift prediction for backbone and (13)C(beta) atoms, with standard deviations of 2.45, 1.09, 0.94, 1.14, 0.25 and 0.49 ppm for delta(15)N, delta(13)C', delta(13)C(alpha), delta(13)C(beta), delta(1)H(alpha) and delta(1)H(N), respectively, between the SPARTA+ predicted and experimental shifts for a set of eleven validation proteins. These results represent a modest but consistent improvement (2-10%) over the best programs available to date, and appear to be approaching the limit at which empirical approaches can predict chemical shifts.

Major groove width variations in RNA structures determined by NMR and impact of 13C residual chemical shift anisotropy and 1H-13C residual dipolar coupling on refinement

Ribonucleic acid structure determination by NMR spectroscopy relies primarily on local structural restraints provided by (1)H- (1)H NOEs and J-couplings. When employed loosely, these restraints are broadly compatible with A- and B-like helical geometries and give rise to calculated structures that are highly sensitive to the force fields employed during refinement. A survey of recently reported NMR structures reveals significant variations in helical parameters, particularly the major groove width. Although helical parameters observed in high-resolution X-ray crystal structures of isolated A-form RNA helices are sensitive to crystal packing effects, variations among the published X-ray structures are significantly smaller than those observed in NMR structures. Here we show that restraints derived from aromatic (1)H- (13)C residual dipolar couplings (RDCs) and residual chemical shift anisotropies (RCSAs) can overcome NMR restraint and force field deficiencies and afford structures with helical properties similar to those observed in high-resolution X-ray structures.

Solution conformation and dynamics of the HIV-1 integrase core domain

The human immunodeficiency virus type 1 (HIV-1) integrase (IN) is a critical enzyme involved in infection. It catalyzes two reactions to integrate the viral cDNA into the host genome, 3' processing and strand transfer, but the dynamic behavior of the active site during catalysis of these two processes remains poorly characterized. NMR spectroscopy can reveal important structural details about enzyme mechanisms, but to date the IN catalytic core domain has proven resistant to such an analysis. Here, we present the first NMR studies of a soluble variant of the catalytic core domain. The NMR chemical shifts are found to corroborate structures observed in crystals, and confirm prior studies suggesting that the alpha4 helix extends toward the active site. We also observe a dramatic improvement in NMR spectra with increasing MgCl(2) concentration. This improvement suggests a structural transition not only near the active site residues but also throughout the entire molecule as IN binds Mg(2+). In particular, the stability of the core domain is linked to the conformation of its C-terminal helix, which has implications for relative domain orientation in the full-length enzyme. (15)N relaxation experiments further show that, although conformationally flexible, the catalytic loop of IN is not fully disordered in the absence of DNA. Indeed, automated chemical shift-based modeling of the active site loop reveals several stable clusters that show striking similarity to a recent crystal structure of prototype foamy virus IN bound to DNA.

Site-specific backbone amide (15)N chemical shift anisotropy tensors in a small protein from liquid crystal and cross-correlated relaxation measurements

Site-specific (15)N chemical shift anisotropy (CSA) tensors have been derived for the well-ordered backbone amide (15)N nuclei in the B3 domain of protein G (GB3) from residual chemical shift anisotropy (RCSA) measured in six different mutants that retain the native structure but align differently relative to the static magnetic field when dissolved in a liquid crystalline Pf1 suspension. This information is complemented by measurement of cross-correlated relaxation rates between the (15)N CSA tensor and either the (15)N-(1)H or (15)N-(13)C' dipolar interaction. In agreement with recent solid state NMR measurements, the (15)N CSA tensors exhibit only a moderate degree of variation from averaged values, but have larger magnitudes in alpha-helical (-173 +/- 7 ppm) than in beta-sheet (-162 +/- 6 ppm) residues, a finding also confirmed by quantum computations. The orientations of the least shielded tensor component cluster tightly around an in-peptide-plane vector that makes an angle of 19.6 +/- 2.5 degrees with the N-H bond, with the asymmetry of the (15)N CSA tensor being slightly smaller in alpha-helix (eta = 0.23 +/- 0.17) than in beta-sheet (eta = 0.31 +/- 0.11). The residue-specific (15)N CSA values are validated by improved agreement between computed and experimental (15)N R(1rho) relaxation rates measured for (15)N-{(2)H} sites in GB3, which are dominated by the CSA mechanism. Use of residue-specific (15)N CSA values also results in more uniform generalized order parameters, S(2), and predicts considerable residue-by-residue variations in the magnetic field strengths where TROSY line narrowing is most effective.

De novo structure generation using chemical shifts for proteins with high-sequence identity but different folds

Proteins with high-sequence identity but very different folds present a special challenge to sequence-based protein structure prediction methods. In particular, a 56-residue three-helical bundle protein (GA(95)) and an alpha/beta-fold protein (GB(95)), which share 95% sequence identity, were targets in the CASP-8 structure prediction contest. With only 12 out of 300 submitted server-CASP8 models for GA(95) exhibiting the correct fold, this protein proved particularly challenging despite its small size. Here, we demonstrate that the information contained in NMR chemical shifts can readily be exploited by the CS-Rosetta structure prediction program and yields adequate convergence, even when input chemical shifts are limited to just amide (1)H(N) and (15)N or (1)H(N) and (1)H(alpha) values.

Differential phospholipid binding of alpha-synuclein variants implicated in Parkinson's disease revealed by solution NMR spectroscopy

Three familial variants of the presynaptic protein α-synuclein (αS), A30P, E46K, and A53T, correlate with rare inherited Parkinson’s disease (PD), while wild-type αS is implicated in sporadic PD. The classic manifestation of both familiar and sporadic PD is the formation of fibrillar structures of αS which accumulate as the main component in intraneuronal Lewy bodies. At presynaptic termini, the partitioning of αS between disordered cytosolic and membrane-bound states likely mediates its proposed role in regulation of reserve pools of synaptic vesicles. Previously, we reported on multiple distinct phospholipid binding modes of αS with slow binding kinetics. Here, we report the phospholipid binding properties of the disease variants, viewed by solution NMR in a residue-specific manner. Our results agree qualitatively with previous biophysical studies citing overall decreased lipid affinity for the A30P mutation, comparable affinity for A53T, and an increased level of binding of E46K, relative to wild-type αS. Additionally, our NMR results describe the distribution of lipid-bound states for αS: the population of the SL1 binding mode (residues 3−25 bound as a helix) is augmented by each of the disease variants, relative to wild-type αS. We propose that the SL1 binding mode, which anchors the N-terminus of αS in the lipoprotein complex while the hydrophobic NAC region remains dynamically disordered, is prone to intermolecular interactions which progress toward disease-associated oligomers and fibrils. The elevation of the SL1 binding mode, unchecked by a proportionate population of binding modes incorporating the full N-terminal domain, may well account for the increased toxicity of the A30P, E46K, and A53T disease variants of αS.

Prediction of Xaa-Pro peptide bond conformation from sequence and chemical shifts

We present a program, named Promega, to predict the Xaa-Pro peptide bond conformation on the basis of backbone chemical shifts and the amino acid sequence. Using a chemical shift database of proteins of known structure together with the PDB-extracted amino acid preference of cis Xaa-Pro peptide bonds, a cis/trans probability score is calculated from the backbone and (13)C(beta) chemical shifts of the proline and its neighboring residues. For an arbitrary number of input chemical shifts, which may include Pro-(13)C(gamma), Promega calculates the statistical probability that a Xaa-Pro peptide bond is cis. Besides its potential as a validation tool, Promega is particularly useful for studies of larger proteins where Pro-(13)C(gamma) assignments can be challenging, and for on-going efforts to determine protein structures exclusively on the basis of backbone and (13)C(beta) chemical shifts.

Chemical shift anisotropy of imino 15N nuclei in Watson-Crick base pairs from magic angle spinning liquid crystal NMR and nuclear spin relaxation

Knowledge of (15)N chemical shift anisotropy is prerequisite both for quantitative interpretation of nuclear spin relaxation rates in terms of local dynamics and for the use of residual chemical shift anisotropy (RCSA) as a constraint in structure determination. Accurate measurement of the very small RCSA from the difference in (15)N chemical shift under isotropic and weakly aligning liquid crystalline conditions is very sensitive to minute differences in sample conditions, such as pH or ionic strength. For this reason, chemical shifts were measured for the same solution, under static liquid crystalline alignment, and under magic angle spinning conditions where alignment relative to the magnetic field is removed. Measurements were made for 14 well-resolved G-N(1) and 6 U-N(3) (15)N nuclei in a sample of tRNA(Val). Fitting these RCSA data together with (15)N-(1)H dipole-CSA cross-correlated relaxation measurements to the recently refined structural model of tRNA(Val) yields the magnitude, asymmetry, and orientation of the (15)N CSA tensors.

TALOS+: a hybrid method for predicting protein backbone torsion angles from NMR chemical shifts

NMR chemical shifts in proteins depend strongly on local structure. The program TALOS establishes an empirical relation between 13C, 15N and 1H chemical shifts and backbone torsion angles phi and psi (Cornilescu et al. J Biomol NMR 13 289-302, 1999). Extension of the original 20-protein database to 200 proteins increased the fraction of residues for which backbone angles could be predicted from 65 to 74%, while reducing the error rate from 3 to 2.5%. Addition of a two-layer neural network filter to the database fragment selection process forms the basis for a new program, TALOS+, which further enhances the prediction rate to 88.5%, without increasing the error rate. Excluding the 2.5% of residues for which TALOS+ makes predictions that strongly differ from those observed in the crystalline state, the accuracy of predicted phi and psi angles, equals +/-13 degrees . Large discrepancies between predictions and crystal structures are primarily limited to loop regions, and for the few cases where multiple X-ray structures are available such residues are often found in different states in the different structures. The TALOS+ output includes predictions for individual residues with missing chemical shifts, and the neural network component of the program also predicts secondary structure with good accuracy.

Multiple tight phospholipid-binding modes of alpha-synuclein revealed by solution NMR spectroscopy

'In dopaminergic neurons, alpha-synuclein (alphaS) partitions between a disordered cytosolic state and a lipid-bound state. Binding of alphaS to membrane phospholipids is implicated in its functional role in synaptic regulation, but also impacts fibril formation associated with Parkinson's disease. We describe here a solution NMR study in which alphaS is added to small unilamellar vesicles of a composition mimicking synaptic vesicles; the results provide evidence for multiple distinct phospholipid-binding modes of alphaS. Exchange between the free state and the lipid-bound alphaS state, and between different bound states is slow on the NMR timescale, being in the range of 1-10 s(-1). Partitioning of the binding modes is dependent on lipid/alphaS stoichiometry, and tight binding with slow-exchange kinetics is observed at stoichiometries as low as 2:1. In all lipid-bound states, a segment of residues starting at the N-terminus of alphaS adopts an alpha-helical conformation, while succeeding residues retain the characteristics of a random coil. The 40 C-terminal residues remain dynamically disordered, even at high-lipid concentrations, but can also bind to lipids to an extent that appears to be determined by the fraction of cis X-Pro peptide bonds in this region. While lipid-bound alphaS exhibits dynamic properties that preclude its direct observation by NMR, its exchange with the NMR-visible free form allows for its indirect characterization. Rapid amide-amide nuclear Overhauser enhancement buildup points to a large alpha-helical conformation, and a distinct increase in fluorescence anisotropy attributed to Tyr39 indicates an ordered environment for this "dark state." Titration of alphaS with increasing amounts of lipids suggests that the binding mode under high-lipid conditions remains qualitatively similar to that in the low-lipid case. The NMR data appear incompatible with the commonly assumed model where alphaS lies in an alpha-helical conformation on the membrane surface and instead suggest that considerable remodeling of the vesicles is induced by alphaS.

Improved accuracy of 15N-1H scalar and residual dipolar couplings from gradient-enhanced IPAP-HSQC experiments on protonated proteins

The presence of dipole-dipole cross-correlated relaxation as well as unresolved E.COSY effects adversely impacts the accuracy of (1)J(NH) splittings measured from gradient-enhanced IPAP-HSQC spectra. For isotropic samples, the size of the systematic errors caused by these effects depends on the values of (2)J(NHalpha), (3)J(NHbeta) and (3)J(HNHalpha). Insertion of band-selective (1)H decoupling pulses in the IPAP-HSQC experiment eliminates these systematic errors and for the protein GB3 yields (1)J(NH) splittings that agree to within a root-mean-square difference of 0.04 Hz with values measured for perdeuterated GB3. Accuracy of the method is also highlighted by a good fit to the GB3 structure of the (1)H-(15)N RDCs extracted from the minute differences in (1)J(NH) splitting measured at 500 and 750 MHz (1)H frequencies, resulting from magnetic susceptibility anisotropy. A nearly complete set of (2)J(NHalpha) couplings was measured in GB3 in order to evaluate whether the impact of cross-correlated relaxation is dominated by the (15)N-(1)H(alpha) or (15)N-(1)H(beta) dipolar interaction. As expected, we find that (2)J(NHalpha) < or = 2 Hz, with values in the alpha-helix (0.86 +/- 0.52 Hz) slightly larger than in beta-sheet (0.66 +/- 0.26 Hz). Results indicate that under isotropic conditions, N-H(N)/N-H(beta) cross-correlated relaxation often dominates. Unresolved E.COSY effects under isotropic conditions involve (3)J(HNHalpha) and J(NHalpha), but when weakly aligned any aliphatic proton proximate to both N and H(N) can contribute.

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.