John L. Battiste

To test whether distances derived from paramagnetic broadening of (15)N heteronuclear single quantum coherence (HSQC) resonances could be used to determine the global fold of a large, perdeuterated protein, we used site-directed spin-labeling of 5 amino acids on the surface of (15)N-labeled eukaryotic translation initiation factor 4E (eIF4E). eIF4E is a 25 kDa translation initiation protein, whose solution structure was previously solved in a 3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate hydrate (CHAPS) micelle of total molecular mass approximately 45-50 kDa. Distance-dependent line broadening consistent with the three-dimensional structure of eIF4E was observed for all spin-label substitutions. The paramagnetic broadening effects (PBEs) were converted into distances for modeling by a simple method comparing peak heights in (15)N-HSQC spectra before and after reduction of the nitroxide spin label with ascorbic acid. The PBEs, in combination with HN-HN nuclear Overhauser effects (NOEs) and chemical shift index (CSI) angle restraints, correctly determined the global fold of eIF4E with a backbone precision of 2.3 A (1.7 A for secondary structure elements). The global fold was not correctly determined with the HN-HN NOEs and CSI angles alone. The combination of PBEs with simulated restraints from another nuclear magnetic resonance (NMR) method for global fold determination of large proteins (methyl-protonated, highly deuterated samples) improved the quality of calculated structures. In addition, the combination of the two methods simulated from a crystal structure of an all alpha-helical protein (40 kDa farnesyl diphoshphate synthase) correctly determined the global fold where neither method individually was successful. These results show the potential feasibility of obtaining medium-resolution structures for proteins in the 40-100 kDa range via NMR.

The solution structure of a human immunodeficiency virus type-1 (HIV-1) Rev peptide bound to stem-loop IIB of the Rev response element (RRE) RNA was solved by nuclear magnetic resonance spectroscopy. The Rev peptide has an alpha-helical conformation and binds in the major groove of the RNA near a purine-rich internal loop. Several arginine side chains make base-specific contacts, and an asparagine residue contacts a G.A base pair. The phosphate backbone adjacent to a G.G base pair adopts an unusual structure that allows the peptide to access a widened major groove. The structure formed by the two purine-purine base pairs of the RRE creates a distinctive binding pocket that the peptide can use for specific recognition.

The Rev responsive element (RRE) is an RNA secondary structural element within the env gene of HIV and is the binding site for the viral Rev protein. Formation of the Rev-RRE complex is involved in regulation of splicing and transport of mRNA from the nucleus. To understand the structural basis for the specific recognition of RRE by Rev, we have studied a model system for this interaction using NMR. We have obtained a specific 1:1 complex between an RNA derived from stem IIB of RRE, which contains the highest affinity Rev binding site, and a modified Rev34-50 peptide, which binds the RRE as an alpha-helix [Tan, R., et al. (1993) Cell 73, 1031-1040]. Binding of the peptide was accompanied by a conformational change in the RNA, which resulted in the formation of additional base pairs not present in the free RNA. Two of these induced base pairs are purine-purine pairs within the internal loop of RRE, which had been previously proposed on the basis of biochemical experiments [Bartel, D.P., et al. (1991) Cell 67, 529-536]. The formation of non-Watson-Crick base pairs, interactions in the major groove, and protein-induced conformational changes may prove to be common characteristics of RNA recognition of proteins.