The serum albumin-binding domain of streptococcal protein G is a three-helical bundle: a heteronuclear NMR study.

Streptococcal protein G (SPG) is a cell surface receptor protein with a multiple domain structure containing tandem repeats of serum albumin-binding domains (ABD) and immunoglobulin-binding domains (IgBD). In this paper, we have analysed the fold of ABD. Far-UV circular dichroism analysis of ABD indicates high helical content (56%). Based on an analysis of nuclear magnetic resonance 13C secondary chemical shifts, sequential and short-range NOEs, and a few key nuclear Overhauser effects, we conclude that the ABD is a three-helix bundle. The structure of the ABD is, thus, quite different from the IgBD of protein G [Gronenborn, A.M. et al. (1991) Science 253, 657-661]. This strongly suggests that the ABD and the IgBD of SPG have evolved independently from each other. However, the fold of ABD is similar to that of the IgBD of staphylococcal protein A, possibly indicating a common evolutionary ancestor, despite the lack of sequence homology.

Protein three-dimensional structure determination and sequence-specific assignment of 13C and 15N-separated NOE data. A novel real-space ab initio approach.

The sequence-specific assignment of resonances is considered to be a requirement for the determination of the three-dimensional (3D) structure of a protein in solution by nuclear magnetic resonance methods. The main source of structural information is the nuclear Overhauser effect spectroscopy (NOESY) spectrum, which contains information about spatially close pairs of protons. Currently, various J-correlated spectra must be recorded in order to obtain the sequence-specific assignments necessary to interpret the NOESY spectra. In this work, a novel procedure to determine the 3D structure and the sequence-specific assignments of a protein using only data from 13C and 15N-separated multidimensional NOESY spectra is described. No information from J-correlated spectra is required. The algorithm is called ANSRS (Assignment of NOESY Spectra in Real Space) and is based on an inversion of the traditional strategy. A 3D real-space structure of detected, but unassigned, 1H spins is calculated from the nuclear Overhauser effect (NOE) distance restraints using a dynamical simulated annealing procedure. The sequence-specific assignments are then determined by searching among the 1H spins in the 3D real-space structure for plausible residue assignments. The search uses a Monte Carlo simulated annealing algorithm based on assignment probabilities derived from the 1H, 15N and 13C chemical shifts, various spatial constraints, and the known sequence of the protein. The procedure has been tested on semi-synthetic data sets comprising published experimental chemical shifts and NOE distance restraints derived from the known 3D structures of the two proteins GAL4 (residues 9 to 41) and bovine pancreatic trypsin inhibitor. The ANSRS procedure was able to determine the sequence-specific assignments for more than 95% of the spins, and was fairly robust with respect to missing NOE data. The potential of the ANSRS approach with respect to automated assignment, reduction of the number of NMR spectra required for a structure determination, assignment of homologous and mutant proteins, and the possibility of analysing spectra recorded at high pH is discussed.

Solution structure and dynamics of ras p21.GDP determined by heteronuclear three- and four-dimensional NMR spectroscopy.

A high-resolution solution structure of the GDP form of a truncated version of the ras p21 protein (residues 1-166) has been determined using NMR spectroscopy. Ras p21 is the product of the human ras protooncogene and a member of a ubiquitous eukaryotic gene family which is highly conserved in evolution. A virtually complete assignment (13C, 15N, and 1H), including stereospecific assignments of 54 C beta methylene protons and 10 C gamma methyl protons of valine residues, was obtained by analysis of three- and four-dimensional (3D and 4D) heteronuclear NMR spectra using a newly developed 3D/4D version of the ANSIG software. A total of 40 converged structures were computed from 3369 experimental restraints consisting of 3,167 nuclear Overhauser effect (NOE) derived distances, 14 phi and 54 chi 1 torsion angle restraints, 109 hydrogen bond distance restraints, and an additional 25 restraints derived from literature data defining interactions between the GDP ligand, the magnesium ion, and the protein. The structure in the region of residues 58-66 (loop L4), and to a lesser degree residues 30-38 (loop L2), is ill-defined. Analysis of the dynamics of the backbone 15N nuclei in the protein showed that residues within the regions 58-66, 107-109, and, to a lesser degree, 30-38 are dynamically mobile on the nanosecond time scale. The root mean square (rms) deviations between the 40 solution structures and the mean atomic coordinates are 0.78 A for the backbone heavy atoms and 1.29 A for all non-hydrogen atoms if all residues (1-166) are included in the analysis. If residues 30-38 and residues 58-66 are excluded from the analysis, the rms deviations are reduced to 0.55 and 1.00 A, respectively. The structure was compared to the most highly refined X-ray crystal structure of ras p21.GDP (1-189) [Milburn, M. V., Tong, L., de Vos, A. M., Brünger, A. T., Yamaizumi, Z., Nishimura, S., & Kim, S.-H. (1990) Science 24, 939-945]. The structures are very similar except in the regions found to be mobile by NMR spectroscopy. In addition, the second alpha-helix (helix-2) has a slightly different orientation. The rms deviation between the average of the solution structures and the X-ray crystal structure is 0.94 A for the backbone heavy atoms if residues 31-37 and residues 59-73 are excluded from the analysis.

Structure of the HMG box motif in the B-domain of HMG1.

The conserved, abundant chromosomal protein HMG1 consists of two highly homologous, folded, basic DNA-binding domains, each of approximately 80 amino acid residues, and an acidic C-terminal tail. Each folded domain represents an 'HMG box', a sequence motif recently recognized in certain sequence-specific DNA-binding proteins and which also occurs in abundant HMG1-like proteins that bind to DNA without sequence specificity. The HMG box is defined by a set of highly conserved residues (most distinctively aromatic and basic) and appears to define a novel DNA-binding structural motif. We have expressed the HMG box region of the B-domain of rat HMG1 (residues 88-164 of the intact protein) in Escherichia coli and we describe here the determination of its structure by 2D 1H-NMR spectroscopy. There are three alpha-helices (residues 13-29, 34-48 and 50-74), which together account for approximately 75% of the total residues and contain many of the conserved basic and aromatic residues. Strikingly, the molecule is L-shaped, the angle of approximately 80 degrees between the two arms being defined by a cluster of conserved, predominantly aromatic, residues. The distinctive shape of the HMG box motif, which is distinct from hitherto characterized DNA-binding motifs, may be significant in relation to its recognition of four-way DNA junctions.

Structure of the DNA-binding domain of zinc GAL4.

The yeast transcriptional activator GAL4 binds co-operatively to four related 17-base-pair sequences within an upstream activating sequence (UASG) to activate transcription of the GAL1 and GAL10 genes. It belongs to a class of gene regulatory proteins which all contain a highly conserved cysteine-rich region within their DNA-binding domains. This region binds zinc and it has been proposed that the cysteine residues coordinate the zinc, creating a structure analogous to one of the 'zinc fingers' of the transcription factor TFIIIA (ref. 8). Using 1H-113Cd two-dimensional nuclear magnetic resonance spectra of the cadmium form of the domain, we previously showed that the protein contains a Cd2Cys6 cluster where cysteines 11 and 28 act as bridging ligands. A similar study of a fragment of GAL4 has recently been published. We report here the solution structure of the DNA binding domain of GAL4; two helix-turn-strand motifs pack around a Zn2Cys6 cluster in a novel pseudo-symmetrical arrangement. The results show that the GAL4 zinc-binding domain differs significantly from both the TFIIIA-type zinc finger and the steroid hormone receptor DNA-binding domains.

The solution conformation of the antibacterial peptide cecropin A: a nuclear magnetic resonance and dynamical simulated annealing study.

The solution conformation of the antibacterial polypeptide cecropin A from the Cecropia moth is investigated by nuclear magnetic resonance (NMR) spectroscopy under conditions where it adopts a fully ordered structure, as judged by previous circular dichroism studies [Steiner, H. (1982) FEBS Lett. 137, 283-287], namely, 15% (v/v) hexafluoroisopropyl alcohol. By use of a combination of two-dimensional NMR techniques the 1H NMR spectrum of cecropin A is completely assigned. A set of 243 approximate interproton distance restraints is derived from nuclear Overhauser enhancement (NOE) measurements. These, together with 32 distance restraints for the 16 intrahelical hydrogen bonds identified on the basis of the pattern of short-range NOEs, form the basis of a three-dimensional structure determination by dynamical simulated annealing [Nilges, M., Clore, G.M., & Gronenborn, A.M. (1988) FEBS Lett. 229, 317-324]. The calculations are carried out starting from three initial structures, an alpha-helix, an extended beta-strand, and a mixed alpha/beta structure. Seven independent structures are computed from each starting structure by using different random number seeds for the assignments of the initial velocities. All 21 calculated structures satisfy the experimental restraints, display very small deviations from idealized covalent geometry, and possess good nonbonded contacts. Analysis of the 21 converged structure indicates that there are two helical regions extending from residues 5 to 21 and from residues 24 to 37 which are very well defined in terms of both atomic root mean square differences and backbone torsion angles. For the two helical regions individually the average backbone rms difference between all pairs of structures is approximately 1 A. The long axes of the two helices lie in two planes, which are at an angle of 70-100 degrees to each other. The orientation of the helices within these planes, however, cannot be determined due to the paucity of NOEs between the two helices.

Three-dimensional NMR spectroscopy of a protein in solution.

The geometric information used to solve three-dimensional (3D) structures of proteins by NMR spectroscopy resides in short (less than 5 A) interproton-distance data. To obtain these distances, the 1H-NMR spectrum must first be assigned using correlation and nuclear Overhauser effect (NOE) experiments to demonstrate through-bond (scalar) and through-space connectivities, respectively. Because the NOE is proportional to r-6, distance information can then be derived. The increased resolution afforded by extending NMR experiments into a second dimension enables one to detect and interpret effects that would not be possible in one dimension owing to extensive spectral overlap and much reduced information. A number of small protein structures have previously been solved in this way. Extending this methodology to larger proteins, however, requires yet an additional improvement in resolution as overlap of cross-peaks in the two-dimensional (2D) NMR spectra present a major barrier to their unambiguous identification. One way of increasing the resolution is to extend the 2D-NMR experiments into a third dimension. We report here the applicability of three-dimensional NMR to macromolecules using the 46-residue protein alpha 1-purothionin as an example.

Determination of three-dimensional protein structures from nuclear magnetic resonance data using fragments of known structures.

A method to build a three-dimensional protein model from nuclear magnetic resonance (NMR) data using fragments from a data base of crystallographically determined protein structures is presented. The interproton distances derived from the nuclear Overhauser effect (NOE) data are compared to the precalculated distances in the known protein structures. An efficient search algorithm is used, which arranges the distances in matrices akin to a C alpha diagonal distance plot, and compares the NOE distance matrices for short sequential zones of the protein to the data base matrices. After cluster analysis of the fragments found in this way, the structure is built by aligning fragments in overlapping zones. The sequentially long-range NOEs cannot be used in the initial fragments search but are vital to discriminate between several possible combinations of different groups of fragments. The method has been tested on one simulated NOE data set derived from a crystal structure and one experimental NMR data set. The method produces models that have good local structure, but may contain larger global errors. These models can be used as the starting point for further refinement, e.g., by restrained molecular dynamics or interactive graphics.

The structure of beta-lactoglobulin and its similarity to plasma retinol-binding protein.

Since its first isolation, bovine beta-lactoglobulin (BLG) has been an enigma: although it is abundant in the whey fraction of milk, its function is still not clear. The results of the many physicochemical studies on the protein need a structural interpretation. We report here the structure of the orthorhombic crystal form of cow BLG at pH 7.6, at a resolution of 2.8 A. It has an unusual protein fold, composed of two slabs of antiparallel beta-sheet, which shows a remarkable similarity to plasma retinol-binding protein. A possible binding site for retinol in BLG has been identified by model-building. This suggests a role for BLG in vitamin A transport and we have discovered specific receptors for the BLG-retinol complex in the intestine of neonate calves.