Search

A series of NH2-terminal truncated forms of human cysteine proteinase inhibitor, cystatin A, was prepared by genetic engineering using Escherichia coli harboring mutated genes. Each variant of cystatin A was efficiently expressed as a fused protein with porcine adenylate kinase and released by CNBr degradation after exchange of the sole inner Met to Leu. The mutant cystatin A lacking an amino-terminal Met residue (called standard variant starting from Ile2, CystA2-98(M65L) showed the same inhibitory activity as authentic one isolated from human epidermis. Two-residue truncation scarcely influenced the activity, but further truncations deleting Pro3 and beyond conservative Gly4 and Gly5 caused a remarkable decrease of their inhibitory activity. But little effect was observed by a substitution of Pro3 with Leu. The loss of the activity by amino-terminal truncation was compensated slightly by engineered substitution of Gly75 with His on a second loop. In the two-dimensional 15N-1H HSQC NMR spectrum, four-residue truncation was found to cause changes in the chemical shifts of Val47 and Val48, which locate on a first loop and consist of a conservative QVVAG sequence. Furthermore, the truncation led to a change in fluorescence spectroscopic behavior of Trp75, which was introduced as a probe on the second loop. Fluorescence intensity of the Trp of the truncated (5-98) form was more affected by heating than the active standard variant. Conversely, fluorescence of Trp75 in 2-98 form was more quenched by acrylamide than the 5-98 variant. Thus, the amino-terminal region of cystatin A is essential for the expression of its inhibitory activity.(ABSTRACT TRUNCATED AT 250 WORDS)

Protein N-glycosylation occurs in the three domains of life. Oligosaccharyltransferase (OST) transfers glycan to asparagine in the N-glycosylation sequon. The catalytic subunit of OST is called STT3 in eukaryotes, AglB in archaea, and PglB in eubacteria. The genome of a hyperthermophilic archaeon, Archaeoglobus fulgidus, encodes three AglB paralogs. Two of them are the shortest AglBs across all domains of life. We determined the crystal structure of the C-terminal globular domain of the smallest AglB to identify the minimal structural unit. The Archaeoglobus AglB lacked a β-barrel-like structure, which had been found in other AglB and PglB structures. In agreement, the deletion in a larger Pyrococcus AglB confirmed its dispensability for the activity. By contrast, the Archaeoglobus AglB contains a kinked helix bearing a conserved motif, called DK/MI motif. The lysine and isoleucine residues in the motif participate in the Ser/Thr recognition in the sequon. The Archaeoglobus AglB structure revealed that the kinked helix contained an unexpected insertion. A revised sequence alignment based on this finding identified a variant type of the DK motif with the insertion. A mutagenesis study of the Archaeoglobus AglB confirmed the contribution of this particular type of the DK motif to the activity. When taken together with our previous results, this study defined the classification of OST: one group consisting of eukaryotes and most archaea possesses the DK-type Ser/Thr pocket, and the other group consisting of eubacteria and the remaining archaea possesses the MI-type Ser/Thr pocket. This classification provides a useful framework for OST studies.

A synthetic replacement study of the amino acid residues of mu-conotoxin GIIIA, a peptide blocker for muscle sodium channels, has recently shown that the conformation formed by three disulfide bridges and the molecular basicity, especially the one around the Arg13 residue, are important for blocking activity. In the present study, we determined the three-dimensional structure of an inactive analog, [Ala13]mu-conotoxin GIIIA, and refined that of the native toxin by NMR spectroscopy combined with simulated annealing calculations. The atomic root-mean-square difference of the mutant from the native conotoxin was 0.62 A for the backbone atoms (N, C alpha, C') of all residues except for the two terminal residues. The observation that the replacement of Arg13 by Ala13 does not significantly change the molecular conformation suggests that the loss of activity is not due to the conformational change but to the direct interaction of the essential Arg13 residue with the sodium channel molecules. In the determined structure, important residues for the activity, Arg13, Lys16, Hyp(hydroxyproline)17, and Arg19, are clustered on one side of the molecule, an observation which suggests that this face of the molecule associates with the receptor site of sodium channels. The hydroxyl group of Hyp17 is suggested to interact with the channel site with which the essential hydroxyl groups of tetrodotoxin and saxitoxin interact.

The interaction of mouse epidermal growth factor (mEGF) with micelles of a phospholipid analogue, perdeuterated dodecylphosphocholine (DPC), was investigated by two-dimensional {sup 1}H NMR. Sequence-specific resonance assignments of the micelle-bound mEGF have been made, and the chemical shifts were compared with those in the absence of DPC. DPC induced large chemical shift changes of the resonances from the residues in the C-terminal tail (residues 46-53) but little perturbation on the residues in the main core (residues 1-45). Starting from the three-dimensional structure in the absence of DPC, micelle-bound structures were calculated using the program XPLOR with interproton distance data obtained from NOESY spectra recorded in the presence of DPC. The C-terminal tail of mEGF was found to change conformation to form an amphiphilic structure when bound to the micelles. It is possible that induced fit in the C-terminal tail of mEGF occurs upon binding to a putative hydrophobic pocket of the EGF receptor.

The pH titration shifts for all the nonlabile proton resonances in a 53-residue protein (mouse epidermal growth factor) were measured in the p2H range 1.5-9 with two-dimensional (2D) 1H NMR. The 2D NMR pH titration experiment made it possible to determine the pK values for all the ionizable groups which were titrated in the pH range 1.5-9 in the protein. The pK values of the nine ionizable groups (alpha-amino group, four Asp, two Glu, one His, and alpha-carboxyl group) were found to be near their normal values. The 2D titration experiment also provided a detailed description of the pH-dependent behavior of the proton chemical shifts and enabled us to characterize the pH-dependent changes of protein conformation. Analysis of the pH-dependent shifts of ca. 200 proton resonances offered evidence of conformational changes in slightly basic pH solution: The deprotonation of the N-terminal alpha-amino group induced a widespread conformational change over the beta-sheet structure in the protein, while the effects of deprotonation of the His22 imidazole group were relatively localized. We found that the 2D NMR pH titration experiment is a powerful tool for investigating the structural and dynamic properties of proteins.

We have isolated four novel oligosaccharides with the sialyl-Lea structure from human milk using a monoclonal antibody, MSW 113. These oligosaccharides were purified by affinity chromatography on a column of the immobilized monoclonal antibody and by high-performance liquid chromatography. The results of structural analyses, i.e., 500-MHz 1H NMR spectroscopy, fast atom bombardment mass spectrometry, and binding to specific anticarbohydrate antibodies, are consistent with the following structures. (formula; see text)

Aromatic proton and high field shifted methyl proton resonances of RNase T1 complexed with Guo, 2'GMP, 3'GMP, or 5'GMP were assigned to specific amino acid residues by analyses of the two-dimensional NMR spectra in comparison with the crystal structure of the RNase T1-2'GMP complex. These assignments were subsequently correlated to those of free RNase T1 [Hoffmann & Ruterjans (1988) Eur. J. Biochem. 177, 539-560]. The spatial proximities of amino acid residues as elucidated by NOESY spectra were found to be quite similar among free RNase T1 and the inhibitor complexes, showing that large conformational changes did not occur upon complex formation. However, small but appreciable conformational changes were induced, which were reflected by the systematic chemical shift changes of some amino acid residues in the active site. Furthermore, we confirmed that RNase T1 contains two specific binding sites, one for the guanine base and the other for the phosphate moiety. The inhibitors are forced to adapt their conformations to fit the guanine base and the phosphate moiety to each binding site on the enzyme. This is consistent with our previous studies that 2'GMP and 3'GMP take the syn form as a bound conformation, while 5'GMP takes the anti conformation around glycosidic bonds [Inagaki et al. (1985) Biochemistry 24, 1013-1020]. The slow-exchange process between free and bound forms involving Tyr42 and Tyr45 was found to be specific to the recognition of the guanine base.

The solution structure of a human cystatin A variant, cystatin A2-98 M65L, which maintains the full inhibitory activity of the wild-type protein, was determined at pH 3.8 by sD/3D heteronuclear double- and triple-resonance NMR spectroscopy. The structure is based on a total of 1343 experimental restraints, comprising 1139 distance, 154 phi and chi 1 torsion angle restraints, and 50 distance constraints for 25 backbone hydrogen bonds. A total of 15 structures was calculated using the YASAP protocol with X-PLOR, and the atomic rms distribution about the mean coordinate positions for residues 8-93 was 0.55 +/- 0.10 A for the backbone atoms and 1.05 +/- 0.11 A for all heavy atoms. The structure consists of five antiparallel beta-sheets and two short alpha-helices. Comparison with the X-ray structure of cystatin B in the papain complex shows that the conformation of the first binding loop is quite similar to that of cystatin A, with an rms deviation of 0.78 A for the backbone atoms in the 43-53 region (cystatin A numbering). The second binding loop, however, is significantly different in the two structures, with an rms deviation greater than 2 A. There are some other significant differences, especially for the N-terminal and alpha-helix regions. The overall structure of cystatin A is also compared with the recently reported NMR structure of the wild-type cystatin A (stefin A) at pH 5.5 (Martin et al., 1995) and reveals the following features. that differ in our structure from the previous one: (1) the N-terminal segment, which was unstructured in the previous report, folds over in close vicinity to the C-terminus, as revealed by the distinctive NOEs between those segments; (2) two discrete short alpha-helices linked by a type II reverse turn were found, instead of the continuous single alpha-helix with a slight kink shown in the previous structure; (3) the second binding loop, which was not well converged in the previous study at pH 5.5, is determined very well in our structure. The effect of the N-terminal truncation on the cystatin A structure was examined by comparing the 1H-15N HSQC spectrum of cystatin A2-98 with that of the cystatin A5-98 variant, which lacks the anti-papain activity, revealing significant chemical shift differences in the residual N-terminal segment and the first binding loop, together with small shifts in the other parts.(ABSTRACT TRUNCATED AT 400 WORDS)

The three-dimensional structures of epidermal growth factors (EGF) previously reported were all in acidic solutions (pH 2.0-3.2), at which pHs EGF cannot bind to the receptor. Here we studied the structure of mouse EGF at pH 6.8, where EGF is physiologically active, and compared it with the structure at pH 2.0 by CD and NMR. From pH dependence of CD spectra and a comparison between the chemical shifts of the proton resonances at pH 6.8 and 2.0, the conformations at two pHs were found to be nearly identical except for the C-terminal tail region. The three-dimensional structures at pH 6.8 and 2.0 were determined independently by a combination of two-dimensional 1H NMR and stimulated annealing calculations using the program XPLOR. The calculations were based on 261 distance constraints at pH 6.8 and 355 distance and 24 torsion angle constraints at pH 2.0. The conformational difference of the C-terminal domain (residues 33-50) was detected between the two structures, which were supported by CD and the chemical shift comparison. The positions of the side chains of Leu47, Arg48, Trp49, and Trp50 are changed probably by the effect of the deprotonation of Asp46. Considering the fact that Leu47 is essential in EGF binding to the receptor, this conformational difference may be important in receptor recognition.

The {sup 1}H NMR spectrum of melittin bound to perdeuterated dodecylphosphocholine micelles was analyzed completely by the sequential assignment method with use of two-dimensional NMR techniques. On the basis of the complete sequence-specific resonance assignments, the three-dimensional structure of melittin bound to micelles was determined by distance geometry calculation. Melittin forms an {alpha}-helical rod with a kink of 120-160{degree} at Thr11 and Gly12. This rod binds to the micelles with the {alpha}-helical axis parallel to the lipid-water interface. The present result is consistent with the wedge model, where melittin binds to the outer surface of membranes and destabilizes the membrane organization, leading lysis. However, the above parallel arrangement does not exclude the possibility that melittin is reoriented to a transmembrane position to form a pore when a membrane potential is applied.

The binding modes of inhibitors to ribonuclease T1 (RNase T1) were studied by the analyses of 270-MHz proton NMR spectra. The chemical shift changes upon binding of phosphate, guanosine, 2'-GMP, 3'-GMP, 5'-GMP, and guanosine 3',5'-bis(phosphate) were observed as high field shifted methyl proton resonances of RNase T1. One methyl resonance was shifted upon binding of phosphate and guanosine nucleotides but not upon binding of guanosine. Four other methyl resonances were shifted upon binding of guanosine and guanosine nucleotides but not upon binding of phosphate. From the analyses of nuclear Overhauser effects for the pair of H8 and H1' protons, together with the vicinal coupling constants for the pair of H1' and H2' protons, the conformation of the guanosine moiety as bound to RNase T1 is found to be C3'-endo-syn for 2'-GMP and 3'-GMP and C3'-endo-anti for 5'-GMP and guanosine 3',5'-bis(phosphate). These observations suggest that RNase T1 probably has specific binding sites for the guanine base and 3'-phosphate group (P1 site) but not for the 5'-phosphate group (PO site) or the ribose ring. The weak binding of guanosine 3',5'-bis(phosphate) and 5'-GMP to RNase T1 is achieved by taking the anti form about the glycosyl bond. The productive binding to RNase T1 probably requires the syn form of the guanosine moiety of RNA substrates.

The 1H NMR spectrum of human transforming growth factor alpha (TGF-alpha) was analyzed almost completely by the sequential assignment method using two-dimensional NMR techniques. On the basis of the nearly complete sequence-specific resonance assignment, secondary and tertiary structures of human TGF-alpha in solution (pH 4.9, 28 degrees C) were determined to satisfy the upper limits of proton-proton distances derived from nuclear Overhauser effect experiments. Although human TGF-alpha and mouse epidermal growth factor (EGF) share 27% homology in amino acid sequence, the backbone chain folds in the two growth factors are quite similar. The structure and function of TGF-alpha is well characterized by the "mitten model" previously proposed for mouse EGF. The gross shape of the TGF-alpha molecule resembles a mitten. TGF-alpha interacts with the receptor as a mitten would grasp an object. However, there is an appreciable structural difference between the two growth factors in the back of the mitten that is formed by the N-terminal polypeptide segment. This is consistent with the evidence that the backs of these molecules are not involved in the receptor binding.

An advanced method was developed for lanthanide-probe analyses of the conformations of flexible biomolecules such as nucleotides. The new method is to determine structure parameters (such as internal-rotation angles) and population parameters for local conformational equilibria of flexible sites, together with standard deviations of these parameters. As the prominent advantage of this method, the interrelations among local conformations of flexible sites may be quantitatively elucidated from the experimental data of lanthanide-induced shifts and relaxations and vicinal coupling constants. As a structural unit of ribonucleic acids, the molecular conformations and conformational equilibria of uridine 3'-monophosphate in aqueous solution were analyzed. The stable local conformers about the C3'-O3' bond are the G+ (phi' = 281 +/- 11 degrees) and G- (phi' = 211 +/- 8 degrees) forms. The internal rotation about the C3'-O3' bond and the ribose-ring puckering are interrelated; 97 +/- 5% of the C3'-endo ribose ring is associated with the G- form while 70 +/- 22% o the C2'-endo ribose ring is associated with the G+ form. An interdependency also exists between the internal rotation about the C4'-C5' bond and the ribose-ring puckering. These short-range conformational interrelations are probably important in controlling the dynamic aspects of ribonucleic acid structures.