The accelerated pace of genomic sequencing has increased the demand for structural models of gene products. Improved quantitative methods are needed to study the many systems (e.g., macromolecular assemblies) for which data are scarce. Here, we describe a new molecular dynamics method for protein structure determination and molecular modeling. An energy function, or database potential, is derived from distributions of interatomic distances obtained from a database of known structures. X-ray crystal structures are refined by molecular dynamics with the new energy function replacing the Van der Waals potential. Compared to standard methods, this method improved the atomic positions, interatomic distances, and side-chain dihedral angles of structures randomized to mimic the early stages of refinement. The greatest enhancement in side-chain placement was observed for groups that are characteristically buried. More accurate calculated model phases will follow from improved interatomic distances. Details usually seen only in high-resolution refinements were improved, as is shown by an R-factor analysis. The improvements were greatest when refinements were carried out using X-ray data truncated at 3.5 Å. The database potential should therefore be a valuable tool for determining X-ray structures, especially when only low-resolution data are available.

Self-association of ClpB (a mixture of 95- and 80-kDa subunits) has been studied with gel filtration chromatography, analytical ultracentrifugation, and electron microscopy. Monomeric ClpB predominates at low protein concentration (0.07 mg/mL), while an oligomeric form is highly populated at >4 mg/mL. The oligomer formation is enhanced in the presence of 2 mM ATP or adenosine 5′-O-thiotriphosphate (ATPγS). In contrast, 2 mM ADP inhibits full oligomerization of ClpB. The apparent size of the ATP- or ATPγS-induced oligomer, as determined by gel filtration, sedimentation velocity and electron microscopy image averaging, and the molecular weight, as determined by sedimentation equilibrium, are consistent with those of a ClpB hexamer. These results indicate that the oligomerization reactions of ClpB are similar to those of other Hsp100 proteins.

The crystal structure of tyrosine aminotransferase (TAT) from the parasitic protozoan Trypanosoma cruzi, which belongs to the aminotransferase subfamily Iγ, has been determined at 2.5 Å resolution with the R-value R = 15.1%. T. cruzi TAT shares less than 15% sequence identity with aminotransferases of subfamily Iα but shows only two larger topological differences to the aspartate aminotransferases (AspATs). First, TAT contains a loop protruding from the enzyme surface in the larger cofactor-binding domain, where the AspATs have a kinked α-helix. Second, in the smaller substrate-binding domain, TAT has a four-stranded antiparallel β-sheet instead of the two-stranded β-sheet in the AspATs. The position of the aromatic ring of the pyridoxal-5′-phosphate cofactor is very similar to the AspATs but the phosphate group, in contrast, is closer to the substrate-binding site with one of its oxygen atoms pointing toward the substrate. Differences in substrate specificities of T. cruzi TAT and subfamily Iα aminotransferases can be attributed by modeling of substrate complexes mainly to this different position of the cofactor-phosphate group. Absence of the arginine, which in the AspATs fixes the substrate side-chain carboxylate group by a salt bridge, contributes to the inability of T. cruzi TAT to transaminate acidic amino acids. The preference of TAT for tyrosine is probably related to the ability of Asn17 in TAT to form a hydrogen bond to the tyrosine side-chain hydroxyl group.

Desulforedoxin (Dx), isolated from the sulfate reducing bacterium Desulfovibrio gigas, is a small homodimeric (2 × 36 amino acids) protein. Each subunit contains a high-spin iron atom tetrahedrally bound to four cysteinyl sulfur atoms, a metal center similar to that found in rubredoxin (Rd) type proteins. The simplicity of the active center in Dx and the possibility of replacing the iron by other metals make this protein an attractive case for the crystallographic analysis of metal-substituted derivatives. This study extends the relevance of Dx to the bioinorganic chemistry field and is important to obtain model compounds that can mimic the four sulfur coordination of metals in biology. Metal replacement experiments were carried out by reconstituting the apoprotein with In3+, Ga3+, Cd2+, Hg2+, and Ni2+ salts. The In3+ and Ga3+ derivatives are isomorphous with the iron native protein; whereas Cd2+, Hg2+, and Ni2+ substituted Dx crystallized under different experimental conditions, yielding two additional crystal morphologies; their structures were determined by the molecular replacement method. A comparison of the three-dimensional structures for all metal derivatives shows that the overall secondary and tertiary structures are maintained, while some differences in metal coordination geometry occur, namely, bond lengths and angles of the metal with the sulfur ligands. These data are discussed in terms of the entatic state theory.

Human ribonuclease inhibitor (hRI) is a cytosolic protein that protects cells from the adventitious invasion of pancreatic-type ribonucleases. hRI has 32 cysteine residues. The oxidation of these cysteine residues to form disulfide bonds is a rapid, cooperative process that inactivates hRI. The most proximal cysteine residues in native hRI are two pairs that are adjacent in sequence: Cys94 and Cys95, and Cys328 and Cys329. A cystine formed from such adjacent cysteine residues would likely contain a perturbing cis peptide bond within its eight-membered ring, which would disrupt the structure of hRI and could facilitate further oxidation. We find that replacing Cys328 and Cys329 with alanine residues has little effect on the affinity of hRI for bovine pancreatic ribonuclease A (RNase A), but increases its resistance to oxidation by 10- to 15-fold. Similar effects are observed for the single variants, C328A hRI and C329A hRI, suggesting that oxidation resistance arises from the inability to form a Cys328–Cys329 disulfide bond. Replacing Cys94 and Cys95 with alanine residues increases oxidation resistance to a lesser extent, and decreases the affinity of hRI for RNase A. The C328A, C329A, and C328A/C329A variants are likely to be more useful than wild-type hRI for inhibiting pancreatic-type ribonucleases in vitro and in vivo. We conclude that replacing adjacent cysteine residues can confer oxidation resistance in a protein.

Hsp16.3, the α-crystallin-related small heat shock protein of Mycobacterium tuberculosis that is maximally expressed during the stationary phase and is a major membrane protein, has been reported to form specific trimer-of-trimers structure and to act as an effective molecular chaperone (Chang Z et al., 1996, J Biol Chem 271:7218–7223). However, little is known about its action mechanism. In this study, Hsp16.3 conformational intermediates with dramatically increased chaperone activities were detected after treatment with very low concentrations of guanidine hydrochloride (0.05 M), urea (0.3 M), or mild heating (30°C). The intermediates showed a significant increase in their capacity to bind the hydrophobic probe 1-anilino-8-naphthalene sulfonate (ANS), indicating an increased exposure of hydrophobic surfaces. Interestingly, the greatest chaperone activities of Hsp16.3 were observed in the presence of 0.3 M guanidine HCl or when heated to 35°C. CD spectroscopy studies revealed no significant changes in protein secondary and tertiary structures at these mild treatments. Our in vitro studies also indicate that long-time-heated Hsp16.3, heated even to temperatures as high as 85°C, has almost the same, if not a slightly greater, chaperone activities as the native protein when cooled to room temperature and its secondary structures also almost recovered. Together, these results suggest that Hsp16.3 modulates its chaperone activity by exposing hydrophobic surfaces and that the protein structure is highly stable and flexible, thus highly adapted for its function.

The DNA repair protein RecA of Mycobacterium tuberculosis contains an intein, a self-splicing protein element. We have employed this Mtu recA intein to create a selection system for successful intein splicing by inserting it into a kanamycin-resistance gene so that functional antibiotic resistance can only be restored upon protein splicing. We then proceeded to develop an ORFTRAP, i.e., a selection system for the cloning of open reading frames (ORFs). The ORFTRAP exploits the self-splicing properties of inteins (which depend on full-length in-frame translation of a precursor protein) by allowing protein splicing to occur when DNA fragments encoding ORFs are inserted into the Mtu recA intein, whereas DNA fragments containing non-ORFs are selected against. Regions of the Mtu recA intein that tolerate the insertion of additional amino acids were identified by Bgl II linker scanning mutagenesis, and a respective construct was chosen as the ORFTRAP. To test the maximum insert size that could be cloned into ORFTRAP, DNA fragments of increasing length from the Listeria monocytogenes hly gene as well as a genomic library of Haemophilus influenzae were inserted and it was found that the longest permissive inserts were 425 bp and 251 bp, respectively. The H. influenzae ORFTRAP library also demonstrated the strength (strong selection power) and weakness (insertion of very small fragments) of the system. Further modifications should make the ORFTRAP useful for protein expression, epitope mapping, and antigen screening.

The monomer-trimer equilibrium of the ectodomain of SIV gp41 (residues 27–149, e-gp41) has been characterized by analytical ultracentrifugation, circular dichroism (CD), and NMR spectroscopy. Based on analytical ultracentrifugation experiments performed at different rotor speeds and protein concentrations, the equilibrium association constant for the SIV e-gp41 trimer is 3.1 × 1011 M−2. The presence of intermolecular nuclear Overhauser effects in a mixture of 12C and 13C-labeled e-gp41 prepared under nondenaturing conditions unambiguously demonstrates that there is a dynamic equilibrium between the monomer and trimer. The CD spectra taken as a function of SIV e-gp41 concentration suggest that the helical content of the monomeric state does not change significantly relative to that of the trimeric state. The relevance of the monomer-trimer equilibrium is discussed with respect to gp41 function and the inhibitory properties of gp41 peptides.

In this work with ab initio computations, we describe relevant interactions between protein active sites and ligands, using as a test case arthropod hemocyanins. A computational analysis of models corresponding to the oxygenated and deoxygenated forms of the hemocyanin active site is performed using the Density Functional Theory approach. We characterize the electron density distribution of the binding site with and without bound oxygen in relation to the geometry, which stems out of the crystals of three hemocyanin proteins, namely the oxygenated form from the horseshoe crab Limulus polyphemus, and the deoxygenated forms, respectively, from the same source and from another arthropod, the spiny lobster Panulirus interruptus. Comparison of the three available crystals indicate structural differences at the oxygen binding site, which cannot be explained only by the presence and absence of the oxygen ligand, since the geometry of the ligand site of the deoxygenated Panulirus hemocyanin is rather similar to that of the oxygenated Limulus protein. This finding was interpreted in the frame of a mechanism of allosteric regulation for oxygen binding. However, the cooperative mechanism, which is experimentally well documented, is only partially supported by crystallographic data, since no oxygenated crystal of Panulirus hemocyanin is presently available. We address the following question: is the local ligand geometry responsible for the difference of the dicopper distance observed in the two deoxygenated forms of hemocyanin or is it necessary to advocate the allosteric regulation of the active site conformations in order to reconcile the different crystal forms? We find that the difference of the dicopper distance between the two deoxygenated hemocyanins is not due to the small differences of ligand geometry found in the crystals and conclude that it must be therefore stabilized by the whole protein tertiary structure.

A method to produce αβ T-cell receptors (TCRs) in a soluble form suitable for biophysical analysis was devised involving in vitro refolding of a TCR fusion protein. Polypeptides corresponding to the variable and constant domains of each chain of a human and a murine receptor, fused to a coiled coil heterodimerization motif from either c-Jun (alpha) or v-Fos (beta), were overexpressed separately in Escherichia coli. Following recovery from inclusion bodies, the two chains of each receptor were denatured, and then refolded together in the presence of denaturants. For the human receptor, which is specific for the immunodominant influenza A HLA–A2-restricted matrix epitope (M58-66), a heterodimeric protein was purified in milligram yields and found to be homogeneous, monomeric, antibody-reactive, and stable at concentrations lower than 1 μM. Using similar procedures, analogous results were obtained with a murine receptor specific for an influenza nucleoprotein epitope (366–374) restricted by H2-Db. Production of these receptors has facilitated a detailed analysis of viral peptide–Major Histocompatibility Complex (peptide–MHC) engagement by the TCR using both surface plasmon resonance (SPR) and, in the case of the human TCR, isothermal titration calorimetry (ITC) (Willcox et al., 1999). The recombinant methods described should enable a wide range of TCR–peptide–MHC interactions to be studied and may also have implications for the production of other heterodimeric receptor molecules.

The structure of a bifunctional 5,10-methylene-tetrahydrofolate dehydrogenase/cyclohydrolase from Escherichia coli has been determined at 2.5 Å resolution in the absence of bound substrates and compared to the NADP-bound structure of the homologous enzyme domains from a trifunctional human synthetase enzyme. Superposition of these structures allows the identification of a highly conserved cluster of basic residues that are appropriately positioned to serve as a binding site for the poly-γ-glutamyl tail of the tetrahydrofolate substrate. Modeling studies and molecular dynamic simulations of bound methylene-tetrahydrofolate and NADP shows that this binding site would allow interaction of the nicotinamide and pterin rings in the dehydrogenase active site. Comparison of these enzymes also indicates differences between their active sites that might allow the development of inhibitors specific to the bacterial target.

The reaction mechanism of the catalytic phosphoryl transfer of cAMP-dependent protein kinase (cAPK) was investigated by semi-empirical AM1 molecular orbital computations of an active site model system derived from the crystal structure of the catalytic subunit of the enzyme. The activation barrier is calculated as 20.7 kcal mol−1 and the reaction itself to be exothermic by 12.2 kcal mol−1. The active site residue Asp166, which was often proposed to act as a catalytic base, does not accept a proton in any of the reaction steps. Instead, the hydroxyl hydrogen of serine is shifted to the simultaneously transferred phosphate group of ATP. Although the calculated transition state geometry indicates an associative phosphoryl transfer, no concentration of negative charge is found. To study the influence of protein mutations on the reaction mechanism, we compared two-dimensional energy hypersurfaces of the protein kinase wild-type model and a corresponding mutant in which Asp166 was replaced by alanine. Surprisingly, they show similar energy profiles despite the experimentally known decrease of catalytic activity for corresponding mutants. Furthermore, a model structure was examined, where the charged NH3 group of Lys168 was replaced by a neutral methyl group. The energetic hypersurface of this hypothetical mutant shows two possible pathways for phosphoryl transfer, which both require significantly higher activation energies than the other systems investigated, while the energetic stabilization of the reaction product is similar in all systems. As the position of the amino acid side chains and the substrate peptide is virtually unchanged in all model systems, our results suggest that the exchange of Asp166 by other amino acid is less important to the phosphoryl transfer itself, but crucial to maintain the configuration of the active site in vivo. The positively charged side chain of Lys168, however, is necessary to stabilize the intermediate reaction states, particularly the side chain of the substrate peptide.

A popular approach to the computational modeling of ligand/receptor interactions is to use an empirical free energy like model with adjustable parameters. Parameters are learned from one set of complexes, then used to predict another set. To improve these empirical methods requires an independent way to study their inherent errors. We introduce a toy model of ligand/receptor binding as a workbench for testing such errors. We study the errors incurred from the two state binding assumption—the assumption that a ligand is either bound in one orientation, or unbound. We find that the two state assumption can cause large errors in free energy predictions, but it does not affect rank order predictions significantly. We show that fitting parameters using data from high affinity ligands can reduce two state errors; so can using more physical models that do not use the two state assumption. We also find that when using two state models to predict free energies, errors are more severe on high affinity ligands than low affinity ligands. And we show that two state errors can be diagnosed by systematically adding new binding modes when predicting free energies: if predictions worsen as the modes are added, then the two state assumption in the fitting step may be at fault.

An apparent conservative mutation, Leu to Val, at the second residue of the rat liver mitochondrial aldehyde dehydrogenase (ALDH) presequence resulted in a precursor protein that was not imported into mitochondria. Additional mutants were made to substitute various amino acids with nonpolar side chains for Leu2. The Ile, Phe, and Trp mutants were imported to an extent similar to that of the native precursor, but the Ala mutant was imported only about one-fourth as well. It was shown that the N-terminal methionine was removed from the L2V mutant in a reaction catalyzed by methionine aminopeptidase. The N-terminal methionine of native pALDH and the other mutant presequences was blocked, presumably by acetylation. Because of the difference in co-translational modification, the L2V mutant sustained a significant loss in the available hydrophobic surface of the presequence. Import competence was restored to the L2V mutant when it was translated using a system that did not remove Met1. The removal of an Arg-Gly-Pro helix linker segment (residues 11–14) from the L2V mutant, which shifted three leucine residues toward the N-terminus, also restored import competence. These results lead to the conclusion that a minimum amount of hydrophobic surface area near the N-termini of mitochondrial presequences is an essential property to determine their ability to be imported. As a result, both electrostatic and hydrophobic components must be considered when trying to understand the interactions between precursor proteins and proteins of the mitochondrial import apparatus.

We have identified a ∼140 amino acid domain that is shared by a variety of proteins in budding and fission yeast, nematode, rat, mouse, frog, oat, and man. Typically, this domain is located within 20 residues of the N-terminus of the various proteins. The percent identity among the domains in the 12 proteins ranges from 42 to 93%, with 16 absolutely conserved residues: N-x11–13-V-x2-A-T-x34–36-R-x7–8-W-R-x3-K-x12-G-x-E-x15-L-x11–12-D-x-G-R-x11-D-x7-R. Even though these proteins share little beyond their segment of homology, data are emerging that several of the proteins are involved in endocytosis and or regulation of cytoskeletal organization. We have named this protein segment the ENTH domain, for Epsin N-terminal Homology domain, and hypothesize that it is a candidate for binding specific ligands and/or enzymatic activity in the cell.

Denaturation with guanidine hydrochloride (GdnHCl) or urea is one of the primary ways of measuring the conformational stability of proteins and comparing the stability of mutant proteins. Despite the widespread use of these two denaturants to provide quantitative data for the free energies of unfolding, the mode of action of these agents is not well understood. We are not even certain whether the action of these agents on proteins is direct and can be regarded as ligand binding, or indirect and involves a change in the properties of solvent (water) in the presence of GdnHCl and urea. In this paper, an extensive kinetic study of the inhibition of ribonuclease A and papain by urea has been performed. The results suggest that the effect of urea on activities of these enzymes can be well described by the denaturant binding model. The binding constants of urea determined by the present method are nearly identical to that determined from a variety of different studies on model compounds and proteins.

The aspartyl dyad of free HIV-1 protease has apparent pKas of ∼3 and ∼6, but recent NMR studies indicate that the aspartyl dyad is fixed in the doubly protonated form over a wide pH range when cyclic urea inhibitors are bound, and in the monoprotonated form when the inhibitor KNI-272 is bound. We present computations and measurements related to these changes in protonation and to the thermodynamic linkage between protonation and inhibition. The Poisson–Boltzmann model of electrostatics is used to compute the apparent pKas of the aspartyl dyad in the free enzyme and in complexes with four different inhibitors. The calculations are done with two parameter sets. One assigns ε = 4 to the solute interior and uses a detailed model of ionization; the other uses ε = 20 for the solute interior and a simplified representation of ionization. For the free enzyme, both parameter sets agree well with previously measured apparent pKas of ∼3 and ∼6. However, the calculations with an internal dielectric constant of 4 reproduce the large pKa shifts upon binding of inhibitors, but the calculations with an internal dielectric constant of 20 do not. This observation has implications for the accurate calculation of pKas in complex protein environments.

Because binding of a cyclic urea inhibitor shifts the pKas of the aspartyl dyad, changing the pH is expected to change its apparent binding affinity. However, we find experimentally that the affinity is independent of pH from 5.5 to 7.0. Possible explanations for this discrepancy are discussed.

A protein structure comparison method is described that allows the generation of large populations of high-scoring alternate alignments. This was achieved by incorporating a random element into an iterative double dynamic programming algorithm. The maximum scores from repeated comparisons of a pair of structures converged on a value that was taken as the global maximum. This lay 15% over the score obtained from the single fixed (unrandomized) calculation. The effect of the gap penalty was observed through the shift of the alignment populations, characterized by their alignment length and root-mean-square deviation (RMSD). The best (lowest RMSD) values found in these populations provided a base-line against which other methods were compared.

Periodate oxidized CTP (oCTP) was used to investigate the importance of lysine residues in the CTP binding site of the cytidine 5′-monophosphate N-acetylneuraminic acid (CMP-NeuAc) synthetase (EC 2.7.7.43) from Haemophilus ducreyi. The reaction of oCTP with the enzyme follows pseudo-first-order saturation kinetics, giving a maximum rate of inactivation of 0.6 min−1 and a KI of 6.0 mM at pH 7.1. Mass spectrometric analysis of the modified enzyme provided data that was consistent with β-elimination of triphosphate after the reaction of oCTP with the enzyme. A fully reduced enzyme-oCTP conjugate, retaining the triphosphate moiety, was obtained by inclusion of NaBH3CN in the reaction solution. The β-elimination product of oCTP reacted several times more rapidly with the enzyme compared to equivalent concentrations of oCTP. This compound also formed a stable reduced morpholino adduct with CMP-NeuAc synthetase when the reaction was conducted in the presence of NaBH3CN, and was found to be a useful lysine modifying reagent. The substrate CTP was capable of protecting the enzyme to a large degree from inactivation by oCTP and its β-elimination product. Lys19, a residue conserved in CMP-NeuAc synthetases, was identified as being labeled with the β-elimination product of oCTP.

Three peptides covering the sequence regions corresponding to the first two (CspB-1), the first three (CspB-2), and the last two (CspB-3) β-strands of CspB, the major cold shock protein of Bacillus subtilis, have been synthesized and analyzed for their conformations in solution and for their precipitation behavior. The peptides are nearly insoluble in water, but highly soluble in aqueous solutions containing 50% acetonitrile (pH 4.0). Upon shifts of the solvent condition toward lower or higher acetonitrile concentrations, the peptides all form fibrils resembling those observed in amyloid associated diseases. These fibrils have been identified and characterized by electron microscopy, binding of the dye congo red, and X-ray fiber diffraction. Characterization of the peptides in solution by circular dichroism and NMR spectroscopy shows that the formation of these fibrils does not require specific preformed secondary structure in the solution state species. While the majority of the soluble fraction of each peptide is monomeric and unstructured, different types of structures including α-helical, β-sheet, and random coil conformations are observed under conditions that eventually lead to fibril formation. We conclude that the absence of tertiary contacts under solution conditions where binding interactions between peptide units are still favorable is a crucial requirement for amyloid formation. Thus, fragmentation of a sequence, like partial chemical denaturation or mutation, can enhance the capacity of specific protein sequences to form such fibrils.