We have developed a method for the prediction of an amino acid sequence that is compatible with a three-dimensional backbone structure. Using only a backbone structure of a protein as input, the algorithm is capable of designing sequences that closely resemble natural members of the protein family to which the template structure belongs. In general, the predicted sequences are shown to have multiple sequence profile scores that are dramatically higher than those of random sequences, and sometimes better than some of the natural sequences that make up the superfamily. As anticipated, highly conserved but poorly predicted residues are often those that contribute to the functional rather than structural properties of the protein. Overall, our analysis suggests that statistical profile scores of designed sequences are a novel and valuable figure of merit for assessing and improving protein design algorithms.

As an alternative method to study the heterotropic mechanism of Escherichia coli aspartate transcarbamoylase, a series of nucleotide analogs were used. These nucleotide analogs have the advantage over site-specific mutagenesis experiments in that interactions between the backbone of the protein and the nucleotide could be evaluated in terms of their importance for function. The ATP analogs purine 5′-triphosphate (PTP), 6-chloropurine 5′-triphosphate (Cl-PTP), 6-mercaptopurine 5′-triphosphate (SH-PTP), 6-methylpurine 5′-triphosphate (Me-PTP), and 1-methyladenosine 5′-triphosphate (Me-ATP) were partially synthesized from their corresponding nucleosides. Kinetic analysis was performed on the wild-type enzyme in the presence of these ATP analogs along with GTP, ITP, and XTP. PTP, Cl-PTP, and SH-PTP each activate the enzyme at subsaturating concentrations of l-aspartate and saturating concentrations of carbamoyl phosphate, but not to the same extent as does ATP. These experiments suggest that the interaction between N6-amino group of ATP and the backbone of the regulatory chain is important for orienting the nucleotide and inducing the displacements of the regulatory chain backbone necessary for initiation of the regulatory response. Me-PTP and Me-ATP also activate the enzyme, but in a more complex fashion, which suggests differential binding at the two sites within each regulatory dimer. The purine nucleotides GTP, ITP, and XTP each inhibit the enzyme but to a lesser extent than CTP. The influence of deoxy and dideoxynucleotides on the activity of the enzyme was also investigated. These experiments suggest that the 2′ and 3′ ribose hydroxyl groups are not of significant importance for binding and orientation of the nucleotide in the regulatory binding site. 2′-dCTP inhibits the enzyme to the same extent as CTP, indicating that the interactions of the enzyme to the O2-carbonyl of CTP are critical for CTP binding, inhibition, and the ability of the enzyme to discriminate between ATP and CTP. Examination of the electrostatic surface potential of the nucleotides and the regulatory chain suggest that the complimentary electrostatic interactions between the nucleotides and the regulatory chain are important for binding and orientation of the nucleotide necessary to induce the local conformational changes that propagate the heterotropic effect.

Several recent publications illustrated advantages of using sequence profiles in recognizing distant homologies between proteins. At the same time, the practical usefulness of distant homology recognition depends not only on the sensitivity of the algorithm, but also on the quality of the alignment between a prediction target and the template from the database of known proteins. Here, we study this question for several supersensitive protein algorithms that were previously compared in their recognition sensitivity (Rychlewski et al., 2000). A database of protein pairs with similar structures, but low sequence similarity is used to rate the alignments obtained with several different methods, which included sequence–sequence, sequence–profile, and profile–profile alignment methods. We show that incorporation of evolutionary information encoded in sequence profiles into alignment calculation methods significantly increases the alignment accuracy, bringing them closer to the alignments obtained from structure comparison.

In general, alignment quality is correlated with recognition and alignment score significance. For every alignment method, alignments with statistically significant scores correlate with both correct structural templates and good quality alignments. At the same time, average alignment lengths differ in various methods, making the comparison between them difficult. For instance, the alignments obtained by FFAS, the profile–profile alignment algorithm developed in our group are always longer that the alignments obtained with the PSI-BLAST algorithms. To address this problem, we develop methods to truncate or extend alignments to cover a specified percentage of protein lengths. In most cases, the elongation of the alignment by profile–profile methods is reasonable, adding fragments of similar structure. The examples of erroneous alignment are examined and it is shown that they can be identified based on the model quality.

The pressure-induced changes in 15N enriched HPr from Staphylococcus carnosus were investigated by two-dimensional (2D) heteronuclear NMR spectroscopy at pressures ranging from atmospheric pressure up to 200 MPa. The NMR experiments allowed the simultaneous observation of the backbone and side-chain amide protons and nitrogens. Most of the resonances shift downfield with increasing pressure indicating generalized pressure-induced conformational changes. The average pressure-induced shifts for amide protons and nitrogens are 0.285 ppm GPa−1 at 278 K and 2.20 ppm GPa−1, respectively. At 298 K the corresponding values are 0.275 and 2.41 ppm GPa−1. Proton and nitrogen pressure coefficients show a significant but rather small correlation (0.31) if determined for all amide resonances. When restricting the analysis to amide groups in the β-pleated sheet, the correlation between these coefficients is with 0.59 significantly higher. As already described for other proteins, the amide proton pressure coefficients are strongly correlated to the corresponding hydrogen bond distances, and thus are indicators for the pressure-induced changes of the hydrogen bond lengths. The nitrogen shift changes appear to sense other physical phenomena such as changes of the local backbone conformation as well. Interpretation of the pressure-induced shifts in terms of structural changes in the HPr protein suggests the following picture: the four-stranded β-pleated sheet of HPr protein is the least compressible part of the structure showing only small pressure effects. The two long helices a and c show intermediary effects that could be explained by a higher compressibility and a concomitant bending of the helices. The largest pressure coefficients are found in the active center region around His15 and in the regulatory helix b which includes the phosphorylation site Ser46 for the HPr kinase. This suggests that this part of the structure occurs in a number of different structural states whose equilibrium populations are shifted by pressure. In contrast to the surrounding residues of the active center loop that show large pressure effects, Ile14 has a very small proton and nitrogen pressure coefficient. It could represent some kind of anchoring point of the active center loop that holds it in the right place in space, whereas other parts of the loop adapt themselves to changing external conditions.

Transmembrane helices are the most readily predictable secondary structure components of proteins. They can be predicted to a high degree of accuracy in a variety of ways. Many of these methods compare new sequence data with the sequence characteristics of known transmembrane domains. However, the known transmembrane sequences are not necessarily representative of a particular organism. We attempt to demonstrate that parameters optimized for the known transmembrane domains are far from optimal when predicting transmembrane regions in a given genome. In particular, we have tested the effect of nucleotide bias upon the composition and hence the prediction characteristics of transmembrane helices. Our analysis shows that nucleotide bias of a genome has a strong and predictable influence upon the occurrences of several of the most important hydrophobic amino acids found within transmembrane helices. Thus, we show that nucleotide bias should be taken into account when determining putative transmembrane domains from sequence data.

Two high resolution crystal structures of Escherichia coli alkaline phosphatase (AP) in the presence of phosphonate inhibitors are reported. The phosphonate compounds, phosphonoacetic acid (PAA) and mercaptomethylphosphonic acid (MMP), bind competitively to AP with dissociation constants of 5.5 and 0.6 mM, respectively. The structures of the complexes of AP with PAA and MMP were refined at high resolution to crystallographic R-values of 19.0 and 17.5%, respectively. Refinement of the AP-inhibitor complexes was carried out using X-PLOR. The final round of refinement was done using SHELXL-97. Crystallographic analyses of the inhibitor complexes reveal different binding modes for the two phosphonate compounds. The significant difference in binding constants can be attributed to these alternative binding modes observed in the high resolution X-ray structures. The phosphinyl group of PAA coordinates to the active site zinc ions in a manner similar to the competitive inhibitor and product inorganic phosphate. In contrast, MMP binds with its phosphonate moiety directed toward solvent. Both enzyme-inhibitor complexes exhibit close contacts, one of which has the chemical and geometrical potential to be considered an unconventional hydrogen bond of the type C-H···X.

The distance between Ca2+-binding site III in the C-terminal domain and Cys35 in the N-terminal domain in cardiac muscle troponin C (cTnC) was determined with a single-tryptophan mutant using bound Tb3+ as the energy donor and iodoacetamidotetramethylrhodamine linked to the cysteine residue as energy acceptor. The luminescence of bound Tb3+ was generated through sensitization by the tryptophan located in the 12-residue binding loop of site III upon irradiation at 295 nm, and this sensitized luminescence was the donor signal transferred to the acceptor. In the absence of bound cation at site II, the mean interdomain distance was found to be 48–49 Å regardless of whether the cTnC was unbound or bound to cardiac troponin I, or reconstituted into cardiac troponin. These results suggest that cTnC retains its overall length in the presence of bound target proteins. The distribution of the distances was wide (half-width >9 Å) and suggests considerable interdomain flexibility in isolated cTnC, but the distributions became narrower for cTnC in the complexes with the other subunits. In the presence of bound cation at the regulatory site II, the interdomain distance was shortened by 6 Å for cTnC, but without an effect on the half-width. The decrease in the mean distance was much smaller or negligible when cTnC was complexed with cTnI or cTnI and cTnT under the same conditions. Although free cTnC has considerable interdomain flexibility, this dynamics is slightly reduced in troponin. These results indicate that the transition from the relaxed state to an activated state in cardiac muscle is not accompanied by a gross alteration of the cTnC conformation in cardiac troponin.

The DNA-binding domain of the yeast heat shock transcription factor (HSF) contains a strictly conserved proline that is at the center of a kink. To define the role of this conserved proline-centered kink, we replaced the proline with a number of other residues. These substitutions did not diminish the ability of the full-length protein to support growth of yeast or to activate transcription, suggesting that the proline at the center of the kink is not conserved for function. The stability of the isolated mutant DNA-binding domains was unaltered from the wild-type, so the proline is not conserved to maintain the stability of the protein. The crystal structures of two of the mutant DNA-binding domains revealed that the helices in the mutant proteins were still kinked after substitution of the proline, suggesting that the proline does not cause the α-helical kink. So why are prolines conserved in this and the majority of other kinked α-helices if not for structure, function, or stability? The mutant DNA-binding domains are less soluble than wild-type when overexpressed. In addition, the folding kinetics, as measured by stopped-flow fluorescence, is faster for the mutant proteins. These two results support the premise that the presence of the proline is critical for the folding pathway of HSF's DNA-binding domain. The finding may also be more general and explain why kinked helices maintain their prolines.

Little is known of the plant branched-chain α-ketoacid dehydrogenase complex. We have undertaken a detailed study of the structure of the dihydrolipoyl acyltransferase (BCE2) subunit that forms the core of the complex, to which two other enzymes attach. Mature Arabidopsis thaliana BCE2 was expressed in Escherichia coli. The soluble recombinant protein was purified using a Superose 6 size-exclusion column to >90% homogeneity and was catalytically active. The recombinant protein formed a stable complex with a native molecular mass of 0.95 MDa and an S coefficient of 19.4, consistent with formation of a 24-mer. Negative-staining transmission electron microscopy of the recombinant protein confirmed that BCE2 forms a core with octagonal symmetry. Despite divergence of mammalian and plant BCE2s, there is clearly conservation of structure that is independent of primary sequence.

The unrefined fold of Escherichia coli β-galactosidase based on a monoclinic crystal form with four independent tetramers has been reported previously. Here, we describe a new, orthorhombic form with one tetramer per asymmetric unit that has permitted refinement of the structure at 1.7 Å resolution. This high-resolution analysis has confirmed the original description of the structure and revealed new details. An essential magnesium ion, identified at the active site in the monoclinic crystals, is also seen in the orthorhombic form. Additional putative magnesium binding sites are also seen. Sodium ions are also known to affect catalysis, and five putative binding sites have been identified, one close to the active site. In a crevice on the protein surface, five linked five-membered solvent rings form a partial clathrate-like structure. Some other unusual aspects of the structure include seven apparent cis-peptide bonds, four of which are proline, and several internal salt-bridge networks. Deep solvent-filled channels and tunnels extend across the surface of the molecule and pass through the center of the tetramer. Because of these departures from a compact globular shape, the molecule is not well characterized by prior empirical relationships between the mass and surface area of proteins. The 50 or so residues at the amino terminus have a largely extended conformation and mostly lie across the surface of the protein. At the same time, however, segment 13–21 contributes to a subunit interface, and residues 29–33 pass through a “tunnel” formed by a domain interface. Taken together, the overall arrangement provides a structural basis for the phenomenon of α-complementation.

Within the tumor suppressor protein INK4 (inhibitor of cyclin-dependent kinase 4) family, p15INK4B is the smallest and the only one whose structure has not been determined previously, probably due to the protein's conformational flexibility and instability. In this work, multidimensional NMR studies were performed on this protein. The first tertiary structure was built by comparative modeling with p16INK4A as the template, followed by restrained energy minimization with NMR constraints (NOE and H-bonds). For this purpose, the solution structure of p16INK4A, whose quality was also limited by similar problems, was refined with additional NMR experiments conducted on an 800 MHz spectrometer and by structure-based iterative NOE assignments. The nonhelical regions showed major improvement with root-mean-square deviation (RMSD) improved from 1.23 to 0.68 Å for backbone heavy atoms. The completion of p15INK4B coupled with refinement of p16INK4A made it possible to compare the structures of the four INK4 members in depth, and to compare the structures of p16INK4A in the free form and in the p16INK4A-CDK6 complex. This is an important step toward a comprehensive understanding of the precise functional roles of each INK4 member.

Human infection with Toxoplasma gondii is an important cause of morbidity and mortality. Protozoan parasites such as T. gondii are incapable of de novo purine biosynthesis and must acquire purines from their host, so the purine salvage pathway offers a number of potential targets for antiparasitic chemotherapy. In T. gondii tachyzoites, adenosine is the predominantly salvaged purine nucleoside, and thus adenosine kinase is a key enzyme in the purine salvage pathway of this parasite. The structure of T. gondii adenosine kinase was solved using molecular replacement and refined by simulated annealing at 1.8 Å resolution to an R-factor of 0.214. The overall structure and the active site geometry are similar to human adenosine kinase, although there are significant differences. The T. gondii adenosine kinase has several unique features compared to the human sequence, including a five-residue deletion in one of the four linking segments between the two domains, which is probably responsible for a major change in the orientation of the two domains with respect to each other. These structural differences suggest the possibility of developing specific inhibitors of the parasitic enzyme.

The conformational changes during the photocycle of the photoactive yellow protein have been the subject of many recent studies. Spectroscopic measurements have shown that the photocycle also occurs in a crystalline environment, and this has been the basis for subsequent Laue diffraction and cryocrystallographic studies. These studies have shown that conformational changes during the photocycle are limited to the chromophore and its immediate environment. However, spectroscopic studies suggest the presence of large conformational changes in the protein. Here, we address this apparent discrepancy in two ways. First, we obtain a description of large concerted motions in the ground state of the yellow protein from NMR data and theoretical calculations. Second, we describe the high-resolution structure of the yellow protein crystallized in a different space group. The structure of the yellow protein differs significantly between the two crystal forms. We show that these differences can be used to obtain a description of the flexibility of the protein that is consistent with the motions observed in solution.

The native form of some proteins such as strained plasma serpins (serine protease inhibitors) and the spring-loaded viral membrane fusion proteins are in a metastable state. The metastable native form is thought to be a folding intermediate in which conversion into the most stable state is blocked by a very high kinetic barrier. In an effort to understand how the spontaneous conversion of the metastable native form into the most stable state is prevented, we designed mutations of α1-antitrypsin, a prototype serpin, which can bypass the folding barrier. Extending the reactive center loop of α1-antitrypsin converts the molecule into a more stable state. Remarkably, a 30-residue loop extension allows conversion into an extremely stable state, which is comparable to the relaxed cleaved form. Biochemical data strongly suggest that the strain release is due to the insertion of the reactive center loop into the major β-sheet, A sheet, as in the known stable conformations of serpins. Our results clearly show that extending the reactive center loop is sufficient to bypass the folding barrier of α1-antitrypsin and suggest that the constrain held by polypeptide connection prevents the conversion of the native form into the lowest energy state.

Proteins often require cofactors to perform their biological functions and must fold in the presence of their cognate ligands. Using circular dichroism spectroscopy, we investigated the effects of divalent metal binding upon the folding pathway of Escherichia coli RNase HI. This enzyme binds divalent metal in its active site, which is proximal to the folding core of RNase HI as defined by hydrogen/deuterium exchange studies. Metal binding increases the apparent stability of native RNase HI chiefly by reducing the unfolding rate. As with the apo-form of the protein, refolding from high denaturant concentrations in the presence of Mg2+ follows three-state kinetics: formation of a rapid burst phase followed by measurable single exponential kinetics. Therefore, the overall folding pathway of RNase HI is minimally perturbed by the presence of metal ions. Our results indicate that the metal cofactor enters the active site pocket only after the enzyme reaches its native fold, and therefore, divalent metal binding stabilizes the protein by decreasing its unfolding rate. Furthermore, the binding of the cofactor is dependent upon a carboxylate critical for activity (Asp10). A mutation in this residue (D10A) alters the folding kinetics in the absence of metal ions such that they are similar to those observed for the unaltered enzyme in the presence of metal.

Orthologs typically retain the same function in the course of evolution. Using β-decarboxylating dehydrogenase family as a model, we demonstrate that orthologs can be confidently identified. The strategy is based on our recent findings that substitutions of only a few amino acid residues in these enzymes are sufficient to exchange substrate and coenzyme specificities. Hence, the few major specificity determinants can serve as reliable markers for determining orthologous or paralogous relationships. The power of this approach has been demonstrated by correcting similarity-based functional misassignment and discovering new genes and related pathways, and should be broadly applicable to other enzyme families.

Using a combination of theoretical sequence structure recognition predictions and experimental disulfide bond assignments, a three-dimensional (3D) model of human interleukin-7 (hIL-7) was constructed that predicts atypical surface chemistry in helix D that is important for receptor activation. A 3D model of hIL-7 was built using the X-ray crystal structure of interleukin-4 (IL-4) as a template (Walter MR et al., 1992, J Mol Biol. 224:1075–1085; Walter MR et al., 1992, J Biol Chem 267:20371–20376). Core secondary structures were constructed from sequences of hIL-7 predicted to form helices. The model was constructed by superimposing IL-7 helices onto the IL-4 template and connecting them together in an up–up down–down topology. The model was finished by incorporating the disulfide bond assignments (Cys3, Cys142), (Cys35, Cys130), and (Cys48, Cys93), which were determined by MALDI mass spectroscopy and site-directed mutagenesis (Cosenza L, Sweeney E, Murphy JR, 1997, J Biol Chem 272:32995–33000). Quality analysis of the hIL-7 model identified poor structural features in the carboxyl terminus that, when further studied using hydrophobic moment analysis, detected an atypical structural property in helix D, which contains Cys130 and Cys142. This analysis demonstrated that helix D had a hydrophobic surface exposed to bulk solvent that accounted for the poor quality of the model, but was suggestive of a region in IL-7 that maybe important for protein interactions. Alanine (Ala) substitution scanning mutagenesis was performed to test if the predicted atypical surface chemistry of helix D in the hIL-7 model is important for receptor activation. This analysis resulted in the construction, purification, and characterization of four hIL-7 variants, hIL-7(K121A), hIL-7(L136A), hIL-7(K140A), and hIL-7(W143A), that displayed reduced or abrogated ability to stimulate a murine IL-7 dependent pre-B cell proliferation. The mutant hIL-7(W143A), which is biologically inactive and displaces [125I]-hIL-7, is the first reported IL-7R system antagonist.

NMR spectroscopy was used to search for mechanistically significant differences in the local mobility of the main-chain amides of Bacillus circulans xylanase (BCX) in its native and catalytically competent covalent glycosyl-enzyme intermediate states. 15N T1, T2, and 15N{1H} NOE values were measured for ~120 out of 178 peptide groups in both the apo form of the protein and in BCX covalently modified at position Glu78 with a mechanism-based 2-deoxy-2-fluoro-β-xylobioside inactivator. Employing the model-free formalism of Lipari and Szabo, the measured relaxation parameters were used to calculate a global correlation time (τm) for the protein in each form (9.2 ± 0.2 ns for apo-BCX; 9.8 ± 0.3 ns for the modified protein), as well as individual order parameters for the main-chain NH bond vectors. Average values of the order parameters for the protein in the apo and complexed forms were S2 = 0.86 ± 0.04 and S2 = 0.91 ± 0.04, respectively. No correlation is observed between these order parameters and the secondary structure, solvent accessibility, or hydrogen bonding patterns of amides in either form of the protein. These results demonstrate that the backbone of BCX is well ordered in both states and that formation of the glycosyl-enzyme intermediate leads to little change, in any, in the dynamic properties of BCX on the time scales sampled by 15N-NMR relaxation measurements.

The activation domain of human procarboxypeptidase A2 (ADA2h) aggregates following thermal or chemical denaturation at acidic pH. The aggregated material contains well-defined ordered structures with all the characteristics of the fibrils associated with amyloidotic diseases. Variants of ADA2h containing a series of mutations designed to increase the local stability of each of the two helical regions of the protein have been found to have a substantially reduced propensity to form fibrils. This arises from a reduced tendency of the denatured species to aggregate rather than from a change in the overall stability of the native state. The reduction in aggregation propensity may result from an increase in the stability of local relative to longer range interactions within the polypeptide chain. These findings show that the intrinsic ability of a protein to form amyloid can be altered substantially by protein engineering methods without perturbing significantly its overall stability or activity. This suggests new strategies for combating diseases associated with the formation of aggregated proteins and for the design of novel protein or peptide therapeutics.

Native state hydrogen exchange of cold shock protein A (CspA) has been characterized as a function of the denaturant urea and of the stabilizing agent trimethylamine N-oxide (TMAO). The structure of CspA has five strands of β-sheet. Strands β1–β4 have strongly protected amide protons that, based on experiments as a function of urea, exchange through a simple all-or-none global unfolding mechanism. By contrast, the protection of amide protons from strand β5 is too weak to measure in water. Strand β5 is hydrogen bonded to strands β3 and β4, both of which afford strong protection from solvent exchange. Gaussian network model (GNM) simulations, which assume that the degree of protection depends on tertiary contact density in the native structure, accurately predict the strong protection observed in strands β1–β4 but fail to account for the weak protection in strand β5. The most conspicuous feature of strand β5 is its low sequence hydrophobicity. In the presence of TMAO, there is an increase in the protection of strands β1–β4, and protection extends to amide protons in more hydrophilic segments of the protein, including strand β5 and the loops connecting the β-strands. TMAO stabilizes proteins by raising the free energy of the denatured state, due to highly unfavorable interactions between TMAO and the exposed peptide backbone. As such, the stabilizing effects of TMAO are expected to be relatively independent of sequence hydrophobicity. The present results suggest that the magnitude of solvent exchange protection depends more on solvent accessibility in the ensemble of exchange susceptible conformations than on the strength of hydrogen-bonding interactions in the native structure.