Low solubility is a major stumbling block in the detailed structural and functional characterization of many proteins and isolated protein domains. The production of some proteins in a soluble form may only be possible through alteration of their sequences by mutagenesis. The feasibility of this approach has been demonstrated in a number of cases where amino acid substitutions were shown to increase protein solubility without altering structure or function. However, identifying residues to mutagenize to increase solubility is difficult, especially in the absence of structural knowledge. For this reason, we have developed a method by which soluble mutants of an insoluble protein can be easily distinguished in vivo in Escherichia coli. This method is based on our observation that cells expressing fusions of an insoluble protein to chloramphenicol acetyltransferase (CAT) exhibit decreased resistance to chloramphenicol compared to fusions with soluble proteins. We found that a soluble mutant of an insoluble protein fused to CAT could be selected by plating on high levels of chloramphenicol.

The gene thiI encodes a protein (ThiI) that plays a role in the transfer of sulfur from cysteine to both thiamin and 4-thiouridine, but the reaction catalyzed by ThiI remains undetermined. Based upon sequence alignments, ThiI shares a unique “P-loop” motif with the PPi synthetase family, four enzymes that catalyze adenylation and subsequent substitution of carbonyl oxygens. To test whether or not this motif is critical for ThiI function, the Asp in the motif was converted to Ala (D189A), and a screen for in vivo 4-thiouridine production revealed the altered enzyme to be inactive. Further scrutiny of sequence data and the crystal structures of two members of the PPi synthetase family implicated Lys321 in the proposed adenylation function of ThiI, and the critical nature of Lys321 has been demonstrated by site-directed mutagenesis and genetic screening. Our results, then, indicate that ThiI catalyzes the adenylation of a substrate at the expense of ATP, a narrowing of possible reactions that provides a strong new basis for deducing the early steps in the transfer of sulfur from cysteine to both thiamin and 4-thiouridine.

The ETS family members display specific DNA binding site preferences. As an example, PU.1 and ETS-1 recognize different DNA sequences with a core element centered over 5′-GGAA-3′ and 5′-GGAA/T-3′, respectively. To understand the molecular basis of this recognition, we carried out site-directed mutagenesis experiments followed by DNA binding studies that use electrophoretic mobility shift assay (EMSA) and surface plasmon resonance methods. EMSA experiments identified amino acid changes A231S and/or N236Y as being important for PU.1 recognition of both 5′-GGAA-3′ and 5′-GGAT-3′ containing oligonucleotides. To confirm these data and obtain accurate binding parameters, we performed kinetic studies using surface plasmon resonance on these mutants. The N236Y substitution revealed a weak protein-DNA interaction with the 5′-GGAA-3′ containing oligonucleotide caused by a faster release of the protein from the DNA (koff tenfold higher than the wild-type protein). With the double mutant A231S-N236Y, we obtained an increase in binding affinity and stability toward both 5′-GGAA-3′ and 5′-GGAT-3′ containing oligonucleotides. We propose that substitution of alanine for serine introduces an oxygen atom that can accept hydrogen and interact with potential water molecules or other atoms to make an energetically favorable hydrogen bond with both 5′-GGAA-3′ and 5′-GGAT-3′ oligonucleotides. The free energy of dissociation for the double mutant A231S-N236Y with 5′-GGAA-3′ (ΔΔG((A231S-N236Y) − (N236Y)) = −1.2 kcal mol−1) confirm the stabilizing effect of this mutant in the protein-DNA complex formation. We conclude that N236Y mutation relaxes the specificity toward 5′-GGAA-3′ and 5′-GGAT-3′ sequences, while A231S mutation modulates the degree of specificity toward 5′-GGAA-3′ and 5′GGAT-3′ sequences. This study explains why wild-type PU.1 does not recognize 5′-GGAT-3′ sequences and in addition broadens our understanding of 5′-GGAA/T-3′ recognition by ETS protein family members.

Computer-based database searching and protein multiple sequence alignment has identified a novel clan of zinc metallopeptidases, which, by phylogenetic analysis, has been shown to contain six subfamilies. The family is characterized by four common transmembrane segments and three conserved sequence motifs. The combination of topology analysis and motif identification has detected three potential Zn2+ coordinating residues. Only two of the sequences of this novel zinc metallopeptidase clan possess any functional annotation, one of which is able to cleave its substrate within a cytosol/transmembrane segment junction. A number of observations suggest that the remaining members of this novel clan may also cleave their substrates within transmembrane segments.

The complete covalent structure of a novel boar DQH sperm surface protein resistant to many classical procedures of enzymatic fragmentation was determined. The relative molecular mass of the major form of this protein determined by ESI-MS and MALDI-MS was 13,065.2 ± 1.0 and 13,065.1, respectively. However, additional peaks differing by 162 Da (i.e., minus hexose), 365 Da (i.e., minus hexose and N-acetylhexosamine), 146 Da (i.e., plus deoxyhexose), and 291 Da (i.e., plus sialic acid) indicated the heterogeneity due to differences in glycosylation. The complete covalent structure of the protein was determined using automated Edman degradation, MALDI-MS, and post-source decay (PSD) MALDI-MS, and shown to consist of N-terminal O-glycosylated peptide followed by two fibronectin type II repeats. The carbohydrates are O-glycosidically linked to threonine 10, as confirmed by PSD MALDI-MS of the isolated N-terminal glycopeptide. Eight cysteine residues of the protein form four disulfide bridges, the positions of which were assigned from MALDI-MS and Edman degradation data. We conclude that mass spectral techniques provide an indispensable tool for the detailed analysis of the covalent structure of proteins, especially those that are refractory to standard approaches of protein chemistry.

Protein SRP54 is an essential component of eukaryotic signal recognition particle (SRP). The methionine-rich M-domain (SRP54M or 54M) interacts with the SRP RNA and is also involved in the binding to signal peptides of secretory proteins during their targeting to cellular membranes. To gain insight into the molecular details of SRP-mediated protein targeting, we studied the human 54M polypeptide. The recombinant human protein was expressed successfully in Escherichia coli and was purified to homogeneity. Our studies determined the sites that were susceptible to limited proteolysis, with the goal to design smaller functional mutant derivatives that lacked nonessential amino acid residues from both termini. Of the four polypeptides produced by V8 protease or chymotrypsin, 54MM-2 was the shortest (120 residues; Mr = 13,584.8), but still contained the conserved amino acids suggested to associate with the signal peptide or the SRP RNA. 54MM-2 was cloned, expressed, purified to homogeneity, and was shown to bind human SRP RNA in the presence of protein SRP19, indicating that it was functional. Highly reproducible conditions for the crystallization of 54MM-2 were established. Examination of the crystals by X-ray diffraction showed an orthorhombic unit cell of dimensions a = 29.127 Å, b = 63.693 Å, and c = 129.601 Å, in space group P212121, with reflections extending to at least 2.0 Å.

Computer simulations of protein models on lattices have been widely used as an aid in the study of protein folding process. Following the suggestion of Raghunathan and Jernigan (1997, Protein Sci 6:2072–2083) that the cubo-octahedral lattice can allow a more realistic representation of proteins than other lattices, we propose here the use of a new set of internal coordinates θ for the description of a protein model on this lattice. An easy procedure for the conversion of the θ coordinates to the Cartesian coordinates is also described. When the Contact Interaction approach, already proposed by us for simulations on square or cubic lattices, was applied to the cubo-octahedral lattice, the system obeyed the correct thermodynamics derived from the definition of energy. Thus, lattice simulations of protein models, in which secondary structure elements such as α-helices or β-strands can be easily identifiable, can be performed.

We use both combinatorial and site-directed mutagenesis to explore the consequences of surface hydrophobic substitutions for the folding of two small single domain proteins, the src SH3 domain, and the IgG binding domain of Peptostreptococcal protein L. We find that in almost every case, destabilizing surface hydrophobic substitutions have much larger effects on the rate of unfolding than on the rate of folding, suggesting that nonnative hydrophobic interactions do not significantly interfere with the rate of core assembly.

The topology of a protein structure is a highly simplified description of its fold including only the sequence of secondary structure elements, and their relative spatial positions and approximate orientations. This information can be embodied in a two-dimensional diagram of protein topology, called a TOPS cartoon. These cartoons are useful for the understanding of particular folds and making comparisons between folds. Here we describe a new algorithm for the production of TOPS cartoons, which is more robust than those previously available, and has a much higher success rate. This algorithm has been used to produce a database of protein topology cartoons that covers most of the data bank of known protein structures.

Two human Mu class glutathione transferases, hGST M1-1 and hGST M2-2, with high sequence identity (84%) exhibit a 100-fold difference in activities with the substrates aminochrome, 2-cyano-1,3-dimethyl-1-nitrosoguanidine (cyanoDMNG), and 1,2-dichloro-4-nitrobenzene (DCNB), hGST M2-2 being more efficient. A sequence alignment with the rat Mu class GST M3-3, an enzyme also showing high activities with aminochrome and DCNB, demonstrated an identical structural cluster of residues 164–168 in the α6-helices of rGST M3-3 and hGST M2-2, a motif unique among known sequences of human, rat, and mouse Mu class GSTs. A putative electrostatic network Arg107-Asp161-Arg165-Glu164(-Gln167) was identified based on the published three-dimensional structure of hGST M2-2. Corresponding variant residues of hGSTM1-1 (Leu165, Asp164, and Arg167) as well as the active site residue Ser209 were targeted for point mutations, introducing hGST M2-2 residues to the framework of hGST M1-1, to improve the activities with substrates characteristic of hGST M2-2. In addition, chimeric enzymes composed of hGST M1-1 and hGST M2-2 sequences were analyzed. The activity with 1-chloro-2,4-dinitrobenzene (CDNB) was retained in all mutant enzymes, proving that they were catalytically competent, but none of the point mutations improved the activities with hGST M2-2 characteristic substrates. The chimeric enzymes showed that the structural determinants of these activities reside in domain II and that residue Arg165 in hGST M2-2 appears to be important for the reactions with cyanoDMNG and DCNB. A mutant, which contained all the hGST M2-2 residues of the putative electrostatic network, was still lacking one order of magnitude of the activities with the characteristic substrates of wild-type hGST M2-2. It was concluded that a limited set of point mutations is not sufficient, but that indirect secondary structural affects also contribute to the hGST M2-2 characteristic activities with aminochrome, cyanoDMNG, and DCNB.

GRASS (Graphical Representation and Analysis of Structures Server), a new web-based server, is described. GRASS exploits many of the features of the GRASP program and is designed to provide interactive molecular graphics and quantitative analysis tools with a simple interface over the World-Wide Web. Using GRASS, it is now possible to view many surface features of biological macromolecules on either standard workstations used in macromolecular analysis or personal computers. The result is a World-Wide Web-based, platform-independent, easily used tool for macromolecular visualization and structure analysis.

All proteins of Saccharomyces cerevisiae have been compared to determine how frequently segments from one protein are present in other proteins. Proteins that are recently evolutionarily related were excluded. The most frequently present protein segments are long, tandem repetitions of a single amino acid. For some of these segments, up to 14% of all proteins in the genome were found to have similar peptides within them. These peptide segments may not be functional protein domains. Although they are the most common shared feature of yeast proteins, their ubiquity and simplicity argue that their probable function may be to simply serve as spacers between other protein motifs.

Tryptophan residues have been introduced into two domains of the human U1A protein to probe solution dynamics. The full length protein contains 282 residues, separated into three distinct domains: the N-terminal RBD1 (RNA Binding Domain I), consisting of amino acids 1–101; the C-terminal RBD2, residues 202–282; and the intervening linker region. Tryptophan residues have been substituted for specific phenylalanine residues on the surface of the β-sheet of either RBD1 or RBD2, thus introducing a single solvent exposed tryptophan as a fluorescence reporter. Both steady-state and time-resolved fluorescence measurements of the isolated RBD domains show that each tryptophan experiences a unique environment on the β-sheet surface. The spectral properties of each tryptophan in RBD1 and RBD2 are preserved in the context of the U1A protein, indicating these domains do not interact with each other or with the linker region. The rotational correlation times of the isolated RBDs and the whole U1A, determined by dynamic polarization measurements, show that the linker region is highly flexible such that each RBD exhibits uncorrelated motion.

Two different recombinant constructs of the N-terminal domain in Escherichia coli DnaJ were uniformly labeled with nitrogen-15 and carbon-13. One, DnaJ(1–78), contains the complete “J-domain,” and the other, DnaJ(1–104), contains both the “J-domain” and a conserved “G/F” extension at the C-terminus. The three-dimensional structures of these proteins have been determined by heteronuclear NMR experiments. In both proteins the “J-domain” adopts a compact structure consisting of a helix-turn-helix-loop-helix-turn-helix motif. In contrast, the “G/F” region in DnaJ(1–104) does not fold into a well-defined structure. Nevertheless, the “G/F” region has been found to have an effect on the packing of the helices in the “J-domain” in DnaJ(1–104). Particularly, the interhelical angles between Helix IV and other helices are significantly different in the two structures. In addition, there are some local conformational changes in the loop region connecting the two central helices. These structural differences in the “J-domain” in the presence of the “G/F” region may be related to the observation that DnaJ(1–78) is incapable of stimulating the ATPase activity of the molecular chaperone protein DnaK despite evidence that sites mediating the binding of DnaJ to DnaK are located in the 1–78 segment.

Proteolytic studies have enabled two of the three putative domains of the fibrinolytic protein streptokinase to be isolated and characterized (Conejero-Lara F et al., 1996, Protein Sci 5:2583–2591). The N-terminal domain, however, could not be isolated in these experiments because of its susceptibility to proteolytic cleavage. To complete the biophysical characterization of the domain structure of streptokinase we have overexpressed, purified, and characterized the N-terminal region of the protein, residues 1–146. The results show this is cooperatively folded with secondary structure content and overall stability closely similar to those of the equivalent region in the intact protein.

Triad1 was recently identified as a nuclear RING finger protein, which is up-regulated during retinoic acid induced granulocytic differentiation of acute leukemia cells. Here we show that a cysteine-rich domain (C6HC), present in Triad1, is conserved in at least 24 proteins encoded by various eukaryotes. The C6HC consensus pattern C-x(4)-C-x(14–30)-C-x(1–4)-C-x(4)-C-x(2)-C-x(4)-H-x(4)-C defines this structure as the fourth family member of the zinc-binding RING, LIM, and LAP/PHD fingers. Strikingly, in 22 of 24 proteins the C6HC domain is flanked by two RING finger structures. We have termed the novel C6HC motif DRIL (double RING finger linked). The strong conservation of the larger tripartite TRIAD (two RING fingers and DRIL) structure indicates that the three subdomains are functionally linked and identifies a novel class of proteins.

Protein B23 is an abundant, multifunctional nucleolar phosphoprotein whose activities are proposed to play a role in ribosome assembly. Szebeni et al. (1997) showed stimulation of nuclear import in vitro by protein B23 and suggested that this effect was due to a molecular chaperone-like activity. Protein B23 was tested for chaperone activities using several protein substrates. The temperature-dependent and -independent aggregation of the HIV-1 Rev protein was measured using a zero angle light scattering (turbidity) assay. Protein B23 inhibited the aggregation of the Rev protein, with the amount of inhibition proportional to the concentration of B23 added. This activity was saturable with nearly complete inhibition when the molar ratio of B23:Rev was slightly above one. Protein B23 also protected liver alcohol dehydrogenase (LADH), carboxypeptidase A, citrate synthase, and rhodanese from aggregation during thermal denaturation and preserved the enzyme activity of LADH under these conditions. In addition, protein B23 was able to promote the restoration of activity of LADH previously denatured with guanidine-HCl. Protein B23 preferentially bound denatured substrates and exposed hydrophobic regions when complexed with denatured proteins. Thus, by several criteria, protein B23 behaves like a molecular chaperone; these activities may be related to its role in ribosome biogenesis.

The consequences of amino acid substitutions at the dimer interface for the strength of the interactions between the monomers and for the catalytic function of the dimeric enzyme alkaline phosphatase from Escherichia coli have been investigated. The altered enzymes R10A, R10K, R24A, R24K, T59A, and R10A/R24A, which have amino acid substitutions at the dimer interface, were characterized using kinetic assays, ultracentrifugation, and transverse urea gradient gel electrophoresis. The kinetic data for the wild-type and altered alkaline phosphatases show comparable catalytic behavior with kcat values between 51.3 and 69.5 s−1 and Km values between 14.8 and 26.3 μM. The ultracentrifugation profiles indicate that the wild-type enzyme is more stable than all the interface-modified enzymes. The wild-type enzyme is dimeric in the pH range of pH 4.0 and above, and disassembled at pH 3.5 and below. All the interface-modified enzymes, however, are apparently monomeric at pH 4.0, begin assembly at pH 5.0, and are not fully assembled into the dimeric form until pH 6.0. The results from transverse urea gradient gel electrophoresis show clear and reproducible differences both in the position and the shape of the unfolding patterns; all these modified enzymes are more sensitive to the denaturant and begin to unfold at urea concentrations between 1.0 and 1.5 M; the wild-type enzyme remains in the folded high mobility form beyond 2.5 M urea. Alkaline phosphatase H370A, modified at the active site and not at the dimer interface, resembles the wild-type enzyme both in ultracentrifugation and electrophoresis studies. The results obtained suggest that substitution of a single amino acid at the interface sacrifices not only the integrity of the assembled dimer, but also the stability of the monomer fold, even though the activity of the enzyme at optimal pH remains unaffected and does not appear to depend on interface stability.

The YER057c/YIL051c/YjgF protein family is a set of 24 full-length homologs, each ∼130 residues in length, and each with no known function or relationship to proteins of known structure. To determine the function of this family, the structure of one member—the YjgF protein from Escherichia coli—was solved and refined at a resolution of 1.2 Å. The YjgF molecule is a homotrimer with exact threefold symmetry. Its tertiary and quaternary structures are related to that of Bacillus subtilis chorismate mutase, although their active sites are completely different. The YjgF protein has an active site curiously similar to protein tyrosine phosphatases, including a covalently modified cysteine, but it is unlikely to be functionally related. The lessons learned from this attempt to deduce function from structure may be useful to future projects in structural genomics.

Using techniques from optimization theory, we have developed a computer program that approximates a desired probability distribution for amino acids by imposing a probability distribution on the four nucleotides in each of the three codon positions. These base probabilities allow for the generation of biased codons for use in mutational studies and in the design of biologically encoded libraries. The dependencies between codons in the genetic code often makes the exact generation of the desired probability distribution for amino acids impossible. Compromises are often necessary. The program, therefore, not only solves for the “optimal” approximation to the desired distribution (where the definition of “optimal” is influenced by several types of parameters entered by the user), but also solves for a number of “sub-optimal” solutions that are classified into families of similar solutions. A representative of each family is presented to the program user, who can then choose the type of approximation that is best for the intended application. The Combinatorial Codons program is available for use over the web from http://www.wi.mit.edu/kim/computing.html.