Backbone solution structures of proteins using residual dipolar couplings: application to a novel structural genomics target

Abstract

Structural genomics (or proteomics) activities are critically dependent on the availability of high-throughput structure determination methodology. Development of such methodology has been a particular challenge for NMR based structure determination because of the demands for isotopic labeling of proteins and the requirements for very long data acquisition times. We present here a methodology that gains efficiency from a focus on determination of backbone structures of proteins as opposed to full structures with all sidechains in place. This focus is appropriate given the presumption that many protein structures in the future will be built using computational methods that start from representative fold family structures and replace as many as 70% of the sidechains in the course of structure determination. The methodology we present is based primarily on residual dipolar couplings (RDCs), readily accessible NMR observables that constrain the orientation of backbone fragments irrespective of separation in space. A new software tool is described for the assembly of backbone fragments under RDC constraints and an application to a structural genomics target is presented. The target is an 8.7 kDa protein from Pyrococcus furiosus, PF1061, that was previously not well annotated, and had a nearest structurally characterized neighbor with only 33% sequence identity. The structure produced shows structural similarity to this sequence homologue, but also shows similarity to other proteins, which suggests a functional role in sulfur transfer. Given the backbone structure and a possible functional link this should be an ideal target for development of modeling methods.

Figure 1

E.COSY HNCA spectra for PF1061 under (a) isotropic and (b) C12E5 aligned conditions. Boxed peaks show the change in coupling upon alignment, which was used to measure RDCs as well as to connect fragments. The samples were prepared as described in the NMR Sample Preparation section of Materials and Methods.

Figure 2

Structural overlay of fragment 3 with the appropriate section of the structure modeled from 1JSB with backbone rmsd of 0.82 Å. This fragment is a direct outcome of REDcRAFT before any minimization. This structure exhibits the hair-pin like structure very well.

Figure 3

Structural overlay of fragment 1 with the appropriate section of the structure modeled from 1JSB win backbone rmsd of 3.4 Å. This fragment is a direct outcome of REDcRAFT before any minimization. Although the dihedral angles of each strand correspond to that of a β strand, the relative orientation of the two strands is not recognized as a β sheet in MolMol.

Figure 4

Resolving orientational ambiguities using two media. (A) possible relative orientations for a helix and hairpin fragment using C12E5-CTAB as a medium. (B) possible relative orientations for the same helix and hairpin fragments using C12E5 as a medium. All structures in (B) have been rotated to superimpose helix orientations. Only the relative orientation shown as the first hairpin structure is seen in both media.

Figure 5

Final structure of PF1061. The structure was produced with the MolMol program. The mixed α/β character is apparent as is the C-terminal tail which exhibits substantial internal motion.