We propose a self-consistent approach to analyze knowledge-based atom–atom potentials used to calculate protein–ligand binding energies. Ligands complexed to actual protein structures were first built using the SMoG growth procedure (DeWitte & Shakhnovich, 1996) with a chosen input potential. These model protein–ligand complexes were used to construct databases from which knowledge-based protein–ligand potentials were derived. We then tested several different modifications to such potentials and evaluated their performance on their ability to reconstruct the input potential using the statistical information available from a database composed of model complexes. Our data indicate that the most significant improvement resulted from properly accounting for the following key issues when estimating the reference state: (1) the presence of significant nonenergetic effects that influence the contact frequencies and (2) the presence of correlations in contact patterns due to chemical structure. The most successful procedure was applied to derive an atom–atom potential for real protein–ligand complexes. Despite the simplicity of the model (pairwise contact potential with a single interaction distance), the derived binding free energies showed a statistically significant correlation (∼0.65) with experimental binding scores for a diverse set of complexes.

CAP-23/NAP-22, a neuron-specific protein kinase C substrate, is Nα-myristoylated and interacts with calmodulin (CaM) in the presence of Ca2+ ions. Takasaki et al. (1999, J Biol Chem 274:11848–11853) have recently found that the myristoylated N-terminal nonapeptide of CAP-23/NAP-22 (mC/N9) binds to Ca2+-bound CaM (Ca2+/CaM). In the present study, small-angle X-ray scattering was used to investigate structural changes of Ca2+/CaM induced by its binding to mC/N9 in solution. The binding of one mC/N9 molecule induced an insignificant structural change in Ca2+/CaM. The 1:1 complex appeared to retain the extended conformation much like that of Ca2+/CaM in isolation. However, it could be seen that the binding of two mC/N9 molecules induced a drastic structural change in Ca2+/CaM, followed by a slight structural change by the binding of more than two but less than four mC/N9 molecules. Under the saturated condition (the molar ratio of 1:4), the radius of gyration (Rg) for the Ca2+/CaM-mC/N9 complex was 19.8 ± 0.3 Å. This value was significantly smaller than that of Ca2+/CaM (21.9 ± 0.3 Å), which adopted a dumbbell structure and was conversely 2–3 Å larger than those of the complexes of Ca2+/CaM with the nonmyristoylated target peptides of myosin light chain kinase or CaM kinase II, which adopted a compact globular structure. The pair distance distribution function had no shoulder peak at around 40 Å, which was mainly due to the dumbbell structure. These results suggest that Ca2+/CaM interacts with Nα-myristoylated CAP-23/NAP-22 differently than it does with other nonmyristoylated target proteins. The N-terminal amino acid sequence alignment of CAP-23/NAP-22 and other myristoylated proteins suggests that the protein myristoylation plays important roles not only in the binding of CAP-23/NAP-22 to Ca2+/CaM, but also in the protein–protein interactions related to other myristoylated proteins.

The EF-hand calcium-binding protein S100B has been shown to interact in vitro in a calcium-sensitive manner with many substrates. These potential S100B target proteins have been screened for the preservation of a previously identified consensus sequence across species. The results were compared to known structural and in vitro properties of the proteins to rationalize choices for potential binding partners. Our approach uncovered four oligomeric proteins tubulin (α and β), glial fibrillary acidic protein (GFAP), desmin, and vimentin that have conserved regions matching the consensus sequence. In the type III intermediate filament proteins (GFAP, vimentin, and desmin), this region corresponds to a portion of a coiled-coil (helix 2A), the structural element responsible for their assembly. In tubulin, the sequence matches correspond to regions of α and β tubulin found at the αβ tubulin interface. In both cases, these consensus sequence matches provide a logical explanation for in vitro observations that S100B is able to inhibit oligomerization of these proteins.

The pressure-induced unfolding of lysozyme was investigated in an aqueous guanidinium chloride solution by means of ultraviolet spectroscopy. Assuming a two-state transition model, volume changes were calculated from the slope of free energy vs. pressure plots over a temperature range of 10 to 60 °C. Between 25 and 60 °C, almost constant volume changes were observed in the transition region, which was reflected in almost identical slopes of the free energy change vs. pressure plots. On the other hand, the different slopes were observed in the pressure dependence of free energy change at temperatures lower than 25 °C. These data were interpreted as suggesting that a two-state model is not appropriate at low temperature, but instead one or more intermediates are present under these conditions. The volume changes for unfolding became less negative at temperatures higher than 25 °C.

Escherichia coli CspA is a member of the cold shock protein family. All cold shock proteins studied to date fold rapidly by an apparent two-state mechanism. CspA contains an unusual cluster of aromatic amino acids on its surface that is necessary for nucleic acid binding and also provides stability to CspA (Hillier et al., 1998). To elucidate the role this aromatic cluster plays in the determining the folding rate and pathway of CspA, we have studied the folding kinetics of mutants containing either leucine or serine substituted for Phe18, Phe20, and/or Phe31. The leucine substitutions are found to accelerate folding and the serine substitutions to decelerate folding. Because these residues exert effects on the free energy of the folding transition state, they may be necessary for nucleating folding. They are not responsible, however, for the very compact, native-like transition state ensemble seen in the cold shock proteins, as the refolding rates of the mutants all show a similar, weak dependence of unfolding rate on denaturant concentration. Using mutant cycle analysis, we show that there is energetic coupling among the three residues between the unfolded and transition states, suggesting that the cooperative nature of these interactions helps to determine the unfolding rate. Overall, our results suggest that separate evolutionary pressures can act simultaneously on the same group of residues to maintain function, stability, and folding rate.

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.

Calerythrin is a 20 kDa calcium-binding protein isolated from gram-positive bacterium Saccharopolyspora erythraea. Based on amino acid sequence homology, it has been suggested that calerythrin belongs to the family of invertebrate sarcoplasmic EF-hand calcium-binding proteins (SCPs), and therefore it is expected to function as a calcium buffer. NMR spectroscopy was used to obtain structural information on the protein in solution. Backbone and side chain 1H, 13C, and 15N assignments were obtained from triple resonance experiments HNCACB, HN(CO)CACB, HNCO, CC(CO)NH, and [15N]-edited TOCSY, and HCCH-TOCSY. Secondary structure was determined by using secondary chemical shifts and characteristic NOEs. In addition, backbone N-H residual dipolar couplings were measured from a spin-state selective [1H, 15N] correlation spectrum acquired from a sample dissolved in a dilute liquid crystal. Four EF-hand motifs with characteristic helix-loop-helix patterns were observed. Three of these are typical calcium-binding EF-hands, whereas site 2 is an atypical nonbinding site. The global fold of calerythrin was assessed by dipolar couplings. Measured dipolar couplings were compared with values calculated from four crystal structures of proteins with sequence homology to calerythrin. These data allowed us to recognize an overall similarity between the folds of calerythrin and sarcoplasmic calcium-binding proteins from the sandworm Nereis diversicolor and the amphioxus Branchiostoma lanceolatum.

We have recently constructed hyperactive human deoxyribonuclease I (DNase I) variants that digest double-stranded DNA more efficiently under physiological saline conditions by introducing positively charged amino acids at eight positions that can interact favorably with the negatively charged DNA phosphates. In this study, we present data from supercoiled DNA nicking, linear DNA digestion, and hyperchromicity assays that distinguish two classes of DNase I hyperactive variants based upon their activity dependence on Ca2+. Class A variants are highly dependent upon Ca2+, having up to 300-fold lower activity in the presence of Mg2+ alone compared to that in the presence of Mg2+ and Ca2+, and include Q9R, H44K, and T205K, in addition to wild-type DNase I. In contrast, the catalytic activity of Class B variants, which comprise the E13R, T14K, N74K, S75K, and N110R hyperactive variants, is relatively Ca2+ independent. A significant proportion of this difference in Ca2+-dependent activity can be attributed to one of the two structural calcium binding sites in DNase I. Compared to wild-type, the removal of Ca2+ binding site 2 by alanine replacements at Asp99, Asp107, and Glu112 decreased activity up to 26-fold in the presence of Mg2+ and Ca2+, but had no effect in the presence of Mg2+ alone. We propose that the rate-enhancing effect of Ca2+ binding at site 2 can be replaced by favorable electrostatic interactions created by proximal positively charged amino acid substitutions such as those found in the Class B variants, thus reducing the dependence on Ca2+.

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 oxidative refolding of ribonuclease A has been investigated in several experimental conditions using a variety of redox systems. All these studies agree that the formation of disulfide bonds during the process occurs through a nonrandom mechanism with a preferential coupling of certain cysteine residues. We have previously demonstrated that in the presence of glutathione the refolding process occurs through the reiteration of two sequential reactions: a mixed disulfide with glutathione is produced first which evolves to form an intramolecular S-S bond. In the same experimental conditions, protein disulfide isomerase (PDI) was shown to catalyze formation and reduction of mixed disulfides with glutathione as well as formation of intramolecular S-S bonds.

This paper reports the structural characterization of the one-disulfide intermediate population during the oxidative refolding of Ribonuclease A under the presence of PDI and glutathione with the aim of defining the role of the enzyme at the early stages of the reaction. The one-disulfide intermediate population occurring at the early stages of both the uncatalyzed and the PDI-catalyzed refolding was purified and structurally characterized by proteolytic digestion followed by MALDI-MS and LC/ESIMS analyses. In the uncatalyzed refolding, a total of 12 disulfide bonds out of the 28 theoretical possible cysteine couplings was observed, confirming a nonrandom distribution of native and nonnative disulfide bonds. Under the presence of PDI, only two additional nonnative disulfides were detected. Semiquantitative LC/ESIMS analysis of the distribution of the S-S bridged peptides showed that the most abundant species were equally populated in both the uncatalyzed and the catalyzed process.

This paper shows the first structural characterization of the one-disulfide intermediate population formed transiently during the refolding of ribonuclease A in quasi-physiological conditions that mimic those present in the ER lumen. At the early stages of the process, three of the four native disulfides are detected, whereas the Cys26–Cys84 pairing is absent. Most of the nonnative disulfide bonds identified are formed by nearest-neighboring cysteines. The presence of PDI does not significantly alter the distribution of S-S bonds, suggesting that the ensemble of single-disulfide species is formed under thermodynamic control.

The protease domain of coagulation factor VIIa (FVIIa) is homologous to trypsin with a similar active site architecture. The catalytic function of FVIIa is regulated by allosteric modulations induced by binding of divalent metal ions and the cofactor tissue factor (TF). To further elucidate the mechanisms behind these transformations, the effects of Zn2+ binding to FVIIa in the free form and in complex with TF were investigated. Equilibrium dialysis suggested that two Zn2+ bind with high affinity to FVIIa outside the N-terminal γ-carboxyglutamic acid (Gla) domain. Binding of Zn2+ to FVIIa, which was influenced by the presence of Ca2+, resulted in decreased amidolytic activity and slightly reduced affinity for TF. After binding to TF, FVIIa was less susceptible to zinc inhibition. Alanine substitutions for either of two histidine residues unique for FVIIa, His216, and His257, produced FVIIa variants with decreased sensitivity to Zn2+ inhibition. A search for putative Zn2+ binding sites in the crystal structure of the FVIIa protease domain was performed by Grid calculations. We identified a pair of Zn2+ binding sites in the Glu210–Glu220 Ca2+ binding loop adjacent to the so-called activation domain canonical to serine proteases. Based on our results, we propose a model that describes the conformational changes underlying the Zn2+-mediated allosteric down-regulation of FVIIa's activity.

The refolding of barstar, the intracellular inhibitor of barnase, is dominated by the slow formation of a cis peptidyl prolyl bond in the native protein. The triple mutant C40/82A P27A in which two cysteine residues and one trans proline were replaced by alanine was used as model system to investigate the kinetics and structural consequences of the trans/cis interconversion of Pro48. One- and two-dimensional real-time NMR spectroscopy was used to follow the trans/cis interconversion after folding was initiated by rapid dilution of the urea denatured protein. Series of 1H, 15N HSQC spectra acquired with and without the addition of peptidyl prolyl isomerase unambiguously revealed the accumulation of a transient trans-Pro48 intermediate within the dead time of the experiment. Subtle chemical shift differences between the native state and the intermediate spectra indicate that the intermediate is predominantly native-like with a local rearrangement in the Pro48 loop and in the β-sheet region including residues Tyr47, Ala82, Thr85, and Val50.

The crystal structures of four active site-directed thrombin inhibitors, 1–4, in a complex with human α-thrombin have been determined and refined at up to 2.0 Å resolution using X-ray crystallography. These compounds belong to a structurally novel family of inhibitors based on a 2,3-disubstituted benzo[b]thiophene structure. Compared to traditional active-site directed inhibitors, the X-ray crystal structures of these complexes reveal a novel binding mode. Unexpectedly, the lipophilic benzo[b]thiophene nucleus of the inhibitor appears to bind in the S1 specificity pocket. At the same time, the basic amine of the C-3 side chain of the inhibitor interacts with the mostly hydrophobic proximal, S2, and distal, S3, binding sites. The second, basic amine side chain at C-2 was found to point away from the active site, occupying a location between the S1 and S′1 sites. Together, the aromatic rings of the C-2 and C-3 side chains sandwich the indole ring of Trp60D contained in the thrombin S2 insertion loop defined by the sequence “Tyr-Pro-Pro-Trp.” [The thrombin residue numbering used in this study is equivalent to that reported for chymotrypsinogen (Hartley BS, Shotton DM, 1971, The enzymes, vol. 3. New York: Academic Press. pp 323–373).] In contrast to the binding mode of more classical thrombin inhibitors (d-Phe-Pro-Arg-H, NAPAP, Argatroban), this novel class of benzo[b]thiophene derivatives does not engage in hydrogen bond formation with Gly216 of the thrombin active site. A detailed analysis of the three-dimensional structures not only provides a clearer understanding of the interaction of these agents with thrombin, but forms a foundation for rational structure-based drug design. The use of the data from this study has led to the design of derivatives that are up to 2,900-fold more potent than the screening hit 1.

Calbindin D28k is a highly conserved Ca2+-binding protein abundant in brain and sensory neurons. The 261-residue protein contains six EF-hands packed into one globular domain. In this study, we have reconstituted calbindin D28k from two fragments containing three EF-hands each (residues 1–132 and 133–261, respectively), and from other combinations of small and large fragments. Complex formation is studied by ion-exchange and size-exclusion chromatography, electrophoresis, surface plasmon resonance, as well as circular dichroism (CD), fluorescence, and NMR spectroscopy. Similar chromatographic behavior to the native protein is observed for reconstituted complexes formed by mixing different sets of complementary fragments, produced by introducing a cut between EF-hands 1, 2, 3, or 4. The C-terminal half (residues 133–261) appears to have a lower intrinsic stability compared to the N-terminal half (residues 1–132). In the presence of Ca2+, NMR spectroscopy reveals a high degree of structural similarity between the intact protein and the protein reconstituted from the 1–132 and 133–261 fragments. The affinity between these two fragments is 2 × 107 M−1, with association and dissociation rate constants of 2.7 × 104 M−1 s−1 and 1.4 × 10−3 s−1, respectively. The complex formed in the presence of Ca2+ is remarkably stable towards unfolding by urea and heat. Both the complex and intact protein display cold and heat denaturation, although residual α-helical structure is seen in the urea denatured state at high temperature. In the absence of Ca2+, the fragments do not recombine to yield a complex resembling the intact apo protein. Thus, calbindin D28k is an example of a protein that can only be reconstituted in the presence of bound ligand. The α-helical CD signal is increased by 26% after addition of Ca2+ to each half of the protein. This suggests that Ca2+-induced folding of the fragments is important for successful reconstitution of calbindin D28k.

We propose a new approach that permits correlation of specific domains defined by their primary sequence with their location in the structure of complex macromolecular aggregates. It is based on the combination of well-established structural analysis methods that incorporate the use of overlapping peptides on cellulose membranes for the isolation and purification of specific antibodies from a polyclonal antiserum. Monospecific antibodies to the connector protein of bacteriophage φ29 were isolated from polyclonal antisera using a new development of the spotscan method. These antibodies can be purified in quantities that allow antigenicity testing in enzyme-linked immunosorbent assays, Western blotting and immunoprecipitations, demonstrating the specificity of this isolation procedure. This approach has allowed us to generate direct antibody probes for immunoelectron microscopy mapping of different connector protein domains in a low resolution three-dimensional epitope map.

The human C3a anaphylatoxin receptor (C3aR) is a G protein-coupled receptor (GPCR) composed of seven transmembrane α-helices connected by hydrophilic loops. Previous studies of chimeric C3aR/C5aR and loop deletions in C3aR demonstrated that the large extracellular loop2 plays an important role in noneffector ligand binding; however, the effector binding site for C3a has not been identified. In this study, selected charged residues in the transmembrane regions of C3aR were replaced by Ala using site-directed mutagenesis, and mutant receptors were stably expressed in the RBL-2H3 cell line. Ligand binding studies demonstrated that R161A (helix IV), R340A (helix V), and D417A (helix VII) showed no binding activity, although full expression of these receptors was established by flow cytometric analysis. C3a induced very weak intracellular calcium flux in cells expressing these three mutant receptors. H81A (helix II) and K96A (helix III) showed decreased ligand binding activity. The calcium flux induced by C3a in H81A and K96A cells was also consistently reduced. These findings suggest that the charged transmembrane residues Arg161, Arg340, and Asp417 in C3aR are essential for ligand effector binding and/or signal coupling, and that residues His81 and Lys96 may contribute less directly to the overall free energy of ligand binding. These transmembrane residues in C3aR identify specific molecular contacts for ligand interactions that account for C3a-induced receptor activation.

The ba3-type cytochrome c oxidase from Thermus thermophilus is known as a two subunit enzyme. Deduced from the crystal structure of this enzyme, we discovered the presence of an additional transmembrane helix “subunit IIa” spanning the membrane. The hydrophobic N-terminally blocked protein was isolated in high yield using high-performance liquid chromatography. Its complete amino acid sequence was determined by a combination of automated Edman degradation of both the deformylated and the cyanogen bromide cleaved protein and automated C-terminal sequencing of the native protein.

The molecular mass of 3,794 Da as determined by MALDI-MS and by ESI requires the N-terminal methionine to be formylated and is in good agreement with the value calculated from the formylmethionine containing sequence (3,766.5 Da + 28 Da = 3,794.5 Da). This subunit consits of 34 residues forming one helix across the membrane (Lys5–Ala34), which corresponds in space to the first transmembrane helix of subunit II of the cytochrome c oxidases from Paracoccus denitrificans and bovine heart, however, with opposite polarity. It is 35% identical to subunit IV of the ba3-cytochrome oxidase from Natronobacterium pharaonis.

The open reading frame encoding this new subunit IIa (cbaD) is located upstream of cbaB in the same operon as the genes for subunit I (cbaA) and subunit II (cbaB).

Unlike most protein crystals, form IX of bovine pancreatic ribonuclease A diffracts well when severely dehydrated. Crystal structures have been solved after 2.5 and 4 days of desiccation with CaSO4, at 1.9 and 2.0 Å resolution, respectively. The two desiccated structures are very similar. An RMS displacement of 1.6 Å is observed for main-chain atoms in each structure when compared to the hydrated crystal structure with some large rearrangements observed in loop regions. The structural changes are the result of intermolecular contacts formed by strong electrostatic interactions in the absence of a high dielectric medium. The electron density is very diffuse for some surface loops, consistent with a very disordered structure. This disorder is related to the conformational changes. These results help explain conformational changes during the lyophilization of protein and the associated phenomena of denaturation and molecular memory.

Backbone dynamics of the basic/helix-loop-helix domain of Pho4 from Saccharomyces cerevisae have been probed by NMR techniques, in the absence of DNA, nonspecifically bound to DNA and bound to cognate DNA. Alpha proton chemical shift indices and nuclear Overhauser effect patterns were used to elucidate the secondary structure in these states. These secondary structures are compared to the co-crystal complex of Pho4 bound to a cognate DNA sequence (Shimizu T, Toumoto A, Ihara K, Shimizu M, Kyogou Y, Ogawa N, Oshima Y, Hakoshima T, 1997, EMBO J 15: 4689–4697). The dynamic information provides insight into the nature of this DNA binding domain as it progresses from free in solution to a specifically bound DNA complex. Relative to the unbound form, we show that formation of either the nonspecific and cognate DNA bound complexes involves a large change in conformation and backbone dynamics of the basic region. The nonspecific and cognate complexes, however, have nearly identical secondary structure and backbone dynamics. We also present evidence for conformational flexibility at a highly conserved glutamate basic region residue. These results are discussed in relation to the mechanism of sequence specific recognition and binding.

Conformational changes occurring within the NS3 protease domain from the hepatitis C virus Bk strain (NS31–180) under different physico-chemical conditions either in the absence or in the presence of its cofactor Pep4A were investigated by limited proteolysis experiments. Because the surface accessibility of the protein is affected by conformational changes, when comparative experiments were carried out on NS31–180 either at different glycerol concentrations or in the presence of Pep4A, differential peptide maps were obtained from which protein regions involved in the structural changes could be inferred. The surface topology of isolated NS31–180 in solution was essentially consistent with the crystal structure of the protein with the N-terminal segment showing a high conformational flexibility. At higher glycerol concentration, the protease assumed a more compact structure showing a decrease in the accessibility of the N-terminal segment that either was forced to interact with the protein or originate intermolecular interactions with neighboring molecules. Binding of the cofactor Pep4A caused the displacement of the N-terminal arm from the protein moiety, leading this segment to again adopt an open and flexible conformation, thus suggesting that the N-terminus of the protease contributes only marginally to the stability of the complex. The observed conformational changes might be directly correlated with the activation mechanism of the protease by either the cosolvent or the cofactor peptide because they lead to tighter packing of the substrate binding site.