Robust structure-based resonance assignment for functional protein studies by NMR

Abstract

High-throughput functional protein NMR studies, like protein interactions or dynamics, require an automated approach for the assignment of the protein backbone. With the availability of a growing number of protein 3D structures, a new class of automated approaches, called structure-based assignment, has been developed quite recently. Structure-based approaches use primarily NMR input data that are not based on J-coupling and for which connections between residues are not limited by through bonds magnetization transfer efficiency. We present here a robust structure-based assignment approach using mainly H(N)-H(N) NOEs networks, as well as (1)H-(15) N residual dipolar couplings and chemical shifts. The NOEnet complete search algorithm is robust against assignment errors, even for sparse input data. Instead of a unique and partly erroneous assignment solution, an optimal assignment ensemble with an accuracy equal or near to 100% is given by NOEnet. We show that even low precision assignment ensembles give enough information for functional studies, like modeling of protein-complexes. Finally, the combination of NOEnet with a low number of ambiguous J-coupling sequential connectivities yields a high precision assignment ensemble. NOEnet will be available under: http://www.icsn.cnrs-gif.fr/download/nmr.

Fig. 1

Threshold function T _CS(n) of the chemical shift filter. Only current assignments with RMSD_a < T _CS(n) are accepted during the search for the assignment ensemble. The ‘upper’ threshold u _CS is usually fixed, while the minimum threshold m _CS and the decay constant c _CS are optimized. The threshold function T _RDC(n) of the RDC filter has the same shape. See text for more details

Fig. 2

Flowchart of the parameter optimization protocol. First are optimized the NOE thresholds d _max^theo (maximum ¹ H ^N–¹ H ^N distance in 3D structure, one value for each NOE intensity class), T _NOE (number of allowed outliers) and Δd (outlier range in Å). Then are optimized the CS thresholds c _CS (decay constant of RMSD filter) and m _CS (minimum RMSD threshold) and finally the RDC thresholds c _RDC and m _RDC. See text for further explanations

Fig. 3

Assignment results on lysozyme using experimental data obtained by (Schwalbe et al. 2001). a Results represented on the NMR structure 1E8L (Schwalbe et al. 2001), using NOE data only (left, case 1 of Table 2), NOE + CS data (middle, case 2 of Table 2), NOE + CS + RDC data (right, case 3 of Table 2). The black lines on the left structure correspond to the experimental NOEs network. Proline residues are shown in gray. The color code represents the spatial assignment range (SAR), as depicted in the colorbar below the structures. Unique assignments are shown in black. b Spatial assignment range (SAR) for each peak and for each case. The peaks are ordered by increasing SAR values. The SAR-value of 10 Å that has been chosen as maximum for the class of exploitable peaks is depicted (dashed red line)

Fig. 4

Error detection success in % for lysozyme. The impact of errors in the classification of NOEs is tested here. Erroneous NOEs were generated by adding to experimental data randomly simulated NOEs corresponding to medium or long distances in the lysozyme X-ray structure 193L and classified incorrectly as short or medium NOEs, respectively. The number of erroneous NOEs introduced is shown on the x-axis. The error detection success, shown on the y-axis, is the ratio between the number of runs for which holes occurred in the assignment list and the total number of runs for which the erroneous NOEs caused the removal of correct assignment possibilities. One run consists in the random generation of erroneous NOEs, as described above, and the search for assignment possibilities by NOEnet

Fig. 5

The response of NOEnet to erroneous constraints is shown on the corrupted NOE data set introduced in Fig. 4. All runs that were done for Fig. 4 are taken together, independent of the number of erroneous NOE-constraints, as their amount is not known in advance in real situations. The gray bar shows the number of runs for which the presence of erroneous NOE-constraints is detected successfully through the appearance of holes in the assignment ensemble. The black bars show the number of runs for which the detection is not successful, as no hole occurred. The distribution of assignment accuracies of these runs is shown in form of a histogram. a The NOE-only data set. b The NOE+CS+RDC data set

Fig. 6

The assignment errors of uniquely assigned peaks are quantified here by the maximum spatial assignment error (SAR_max), i.e. the maximum distance to the correct residue among all uniquely, but erroneously assigned peaks of one assignment ensemble. All runs are taken together, independent of the number of introduced erroneous NOE-constraints, like in Fig. 5. a NOE-only data set. b NOE+CS+RDC data set

Fig. 7

Interaction site estimation of EIN-Hpr using ¹⁵ N–¹ H chemical shift perturbation (CSP) data (Garrett et al. 1997b) with a the correct assignment and b the assignment ensemble obtained by NOEnet (Case 4 of Table 3). The NMR structure of the complex EIN-Hpr (PDB 3EZA (Garrett et al. 1999)) is shown here. EIN is shown by its backbone ribbon and Hpr is shown in yellow by its solvent accessible surface (including side chains). a Using the correct assignment, the corresponding residues of the perturbated peaks are colored in red. b The ensemble of assignment possibilities of the same perturbed peaks are colored in red and in black. The unique assignments are colored in black, while the assignment possibilities of the perturbed peaks with a SAR value below 30 Å are colored in red

Fig. 8

Docking results using HADDOCK on EIN-HPR. The scoring results of the final 200 water-refined EIN-HPR complex models are shown here. The interaction zone (active residues) is defined using ¹⁵ N–¹ H chemical shift perturbation (CSP) data on EIN(Garrett et al. 1997b) and HPR (van Nuland et al. 1995). a The correct assignment of EIN and HPR is used to define the active residues from the CSP data (see Fig. 7a). b The assignment ensemble of EIN, obtained by NOEnet using NOE-data and ¹⁵ N–¹ H chemical shifts (Case 4 of Table 3), is used to define the active residues on EIN (see Fig. 7b)