We present direct evidence for a change in protein structural specificity due to hydrophobic core packing. High resolution structural analysis of a designed core variant of ubiquitin reveals that the protein is in slow exchange between two conformations. Examination of side-chain rotamers indicates that this dynamic response and the lower stability of the protein are coupled to greater strain and mobility in the core. The results suggest that manipulating the level of side-chain strain may be one way of fine tuning the stability and specificity of proteins.

In the past few years, a new generation of fold recognition methods has been developed, in which the classical sequence information is combined with information obtained from secondary structure and, sometimes, accessibility predictions. The results are promising, indicating that this approach may compete with potential-based methods (Rost B et al., 1997, J Mol Biol 270:471–480). Here we present a systematic study of the different factors contributing to the performance of these methods, in particular when applied to the problem of fold recognition of remote homologues.

Our results indicate that secondary structure and accessibility prediction methods have reached an accuracy level where they are not the major factor limiting the accuracy of fold recognition. The pattern degeneracy problem is confirmed as the major source of error of these methods. On the basis of these results, we study three different options to overcome these limitations: normalization schemes, mapping of the coil state into the different zones of the Ramachandran plot, and post-threading graphical analysis.

Inspection of high resolution three-dimensional (3D) structures from the protein database reveals an increasing number of cis-Xaa-Pro and cis-Xaa-Yaa peptide bonds. However, we are still far from being able to predict whether these bonds will remain cis upon single-site substitution of Pro or Yaa and/or cleavage of a peptide bond close to it in the sequence. We have chosen oxidized Escherichia coli thioredoxin (Trx), a member of the Trx superfamily with a single α/β domain and cis P76 to determine the effect of single-site substitution and/or cleavage on this isomer. Standard two-dimensional (2D) NMR analysis were performed on cleaved Trx (1–73/74–108) and its P76A variant. Analysis of the NOE connectivities indicates remarkable similarity between the secondary and supersecondary structure of the noncovalent complexes and Trx. Analysis of the 2D version of the HCCH-TOCSY and HMQC-NOESY-HMQC and 13C-filtered HMQC-NOESY spectra of cleaved Trx with uniformly 13C-labeled I75 and P76 shows surprising conservation of both cis P76 and packing of I75 against W31. A similar NMR analysis of its P76A variant provides no evidence for cis A76 and shows only subtle local changes in both the packing of I75 and the interstrand connectivities between its most protected hydrophobic strands (β2 and β4). Indeed, a molecular simulation model for the trans P76A variant of Trx shows only subtle local changes around the substitution site. In conclusion, cleavage of R73 is insufficient to provoke cis/trans isomerization of P76, but cleavage and single-site substitution (P76A) favors the trans isomer.

Hirudin is an anticoagulant polypeptide isolated from a medicinal leech that inhibits thrombin with extraordinary potency (Kd = 0.2–1.0 pM) and selectivity. Hirudin is composed of a compact N-terminal region (residues 1–47, cross-linked by three disulfide bridges) that binds to the active site of thrombin, and a flexible C-terminal tail (residues 48–64) that interacts with the exosite I of the enzyme. To minimize the sequence of hirudin able to bind thrombin and also to improve its therapeutic profile, several N-terminal fragments have been prepared as potential anticoagulants. However, the practical use of these fragments has been impaired by their relatively poor affinity for the enzyme, as given by the increased value of the dissociation constant (Kd) of the corresponding thrombin complexes (Kd = 30–400 nM). The aim of the present study is to obtain a derivative of the N-terminal domain 1–47 of hirudin displaying enhanced inhibitory potency for thrombin compared to the natural product. In this view, we have synthesized an analogue of fragment 1–47 of hirudin HM2 in which Val1 has been replaced by tert-butylglycine, Ser2 by Arg, and Tyr3 by β-naphthylalanine, to give the BugArgNal analogue. The results of chemical and conformational characterization indicate that the synthetic peptide is able to fold efficiently with the correct disulfide topology (Cys6–Cys14, Cys16–Cys28, Cys22–Cys37), while retaining the conformational properties of the natural fragment. Thrombin inhibition data indicate that the effects of amino acid replacements are perfectly additive if compared to the singly substituted analogues (De Filippis V, Quarzago D, Vindigni A, Di Cera E, Fontana A, 1998, Biochemistry 37:13507–13515), yielding a molecule that inhibits the fast or slow form of thrombin by 2,670- and 6,818-fold more effectively than the natural fragment, and that binds exclusively at the active site of the enzyme with an affinity (Kd,fast = 15.4 pM, Kd,slow = 220 pM) comparable to that of full-length hirudin (Kd,fast = 0.2 pM, Kd,slow = 5.5 pM). Moreover, BugArgNal displays absolute selectivity for thrombin over the other physiologically important serine proteases trypsin, plasmin, factor Xa, and tissue plasminogen activator, up to the highest concentration of inhibitor tested (10 μM).

Androcam is a testis-specific protein of Drosophila melanogaster, with 67% sequence identity to calmodulin and four potential EF-hand calcium-binding sites. Spectroscopic monitoring of the thermal unfolding of recombinant calcium-free androcam shows a biphasic process characteristic of a two-domain protein, with the apo-N-domain less stable than the apo-C-domain. The two EF hands of the C-domain of androcam bind calcium cooperatively with 40-fold higher average affinity than the corresponding calmodulin sites. Magnesium competes with calcium binding [Ka(Mg) ∼3 × 103 M−1]. Weak calcium binding is also detected at one or more N-domain sites. Compared to apo-calmodulin, apo-androcam has a smaller conformational response to calcium and a lower α-helical content over a range of experimental conditions. Unlike calmodulin, a tryptic cleavage site in the N-domain of apo-androcam remains trypsin sensitive in the presence of calcium, suggesting an altered calcium-dependent conformational change in this domain. The affinity of model target peptides for androcam is 103–105 times lower than for calmodulin, and interaction of the N-domain of androcam with these peptides is significantly reduced. Thus, androcam shows calcium-induced conformational responses typical of a calcium sensor, but its properties indicate calcium sensitivity and target interactions significantly different from those of calmodulin. From the sequence differences and the altered calcium-binding properties it is likely that androcam differs from calmodulin in the conformation of residues in the second calcium-binding loop. Molecular modeling supports the deduction that there are significant conformational differences in the N-domain of androcam compared to calmodulin, and that these could affect the surface, conferring a different specificity on androcam in target interactions related to testis-specific calcium signaling functions.

Turkey ovomucoid third domain (OMTKY3) is a canonical inhibitor of serine proteinases. Upon complex formation, the inhibitors fully exposed P1 residue becomes fully buried in the preformed cavity of the enzyme. All 20 P1 variants of OMTKY3 have been obtained by recombinant DNA technology and their equilibrium association constants have been measured with six serine proteinases. To rationalize the trends observed in this data set, high resolution crystal structures have been determined for OMTKY3 P1 variants in complex with the bacterial serine proteinase, Streptomyces griseus proteinase B (SGPB). Four high resolution complex structures are being reported in this paper; the three β-branched variants, Ile18I, Val18I, and Thr18I, determined to 2.1, 1.6, and 1.7 Å resolution, respectively, and the structure of the Ser18I variant complex, determined to 1.9 Å resolution. Models of the Cys18I, Hse18I, and Ape18I variant complexes are also discussed. The β-branched side chains are not complementary to the shape of the S1 binding pocket in SGPB, in contrast to that of the wild-type γ-branched P1 residue for OMTKY3, Leu18I. χ1 angles of approximately 40° are imposed on the side chains of Ile18I, Val18I, and Thr18I within the S1 pocket. Dihedral angles of +60°, −60°, or 180° are more commonly observed but 40° is not unfavorable for the β-branched side chains. Thr18I Oγ1 also forms a hydrogen bond with Ser195 Oγ in this orientation. The Ser18I side chain adopts two alternate conformations within the S1 pocket of SGPB, suggesting that the side chain is not stable in either conformation.

Microcalorimetry has been used to measure the stabilities of mutational variants of yeast iso-1 cytochrome c in which F82 and L85 have been replaced by other hydrophobic amino acids. Specifically, F82 has been replaced by Y and L85 by A. The double mutant F82Y, L85A iso-1 has also been studied, and the mutational perturbations are compared to those for the two single mutants, F82Y iso-1 and L85A iso-1. Results are interpreted in terms of known crystallographic structures. The data show that (1) the destabilization of the mutant proteins is similar in magnitude to that which is theoretically predicted by the more obvious mutation-induced structural effects; (2) the free energy of destabilization of the double mutant, F82Y, L85A iso-1, is less than the sum of those of the two single mutants, almost certainly because, in the double mutant, the -OH group of Y82 is able to protrude into the cavity formed by the L85A substitution. The more favorable structural accommodation of the new -OH group in the double mutant leads to additional stability through (1) further decreases in the volumes of internal cavities and (2) formation of an extra protein–protein hydrogen bond.

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.

The reaction mechanism of phosphoryl transfer catalyzed by UMP/CMP-kinase from Dictyostelium discoideum was investigated by semiempirical AM1 molecular orbital computations of an active site model system derived from crystal structures that contain a transition state analog or a bisubstrate inhibitor. The computational results suggest that the nucleoside monophosphate must be protonated for the forward reaction while it is unprotonated in the presence of aluminium fluoride, a popular transition state analog for phosphoryl transfer reactions. Furthermore, a compactification of the active site model system during the reaction and for the corresponding complex containing AlF3 was observed. For the active site residues that are part of the LID domain, conformational flexibility during the reaction proved to be crucial. On the basis of the calculations, a concerted phosphoryl transfer mechanism is suggested that involves the synchronous shift of a proton from the monophosphate to the transferred PO3-group. The proposed mechanism is thus analogous to the phosphoryl transfer mechanism in cAMP-dependent protein kinase that phosphorylates the hydroxyl groups of serine residues.

Escherichia coli thioredoxin is a 108 amino acid oxidoreductase and contains a single Met residue at position 37. The protein contains a long α-helical stretch between residues 32 and 49. The central residue of this helix, Pro40, has been replaced by Ser. The stabilities of the oxidized states of two proteins, the single mutant M37L and the double mutant M37L,P40S, have been characterized by differential scanning calorimetry (DSC) and also by a series of isothermal guanidine hydrochloride (GuHCl) melts in the temperature range of 277 to 333 K. The P40S mutation was found to stabilize the protein at all temperatures upto 340 K though both proteins had similar Tm values of about 356 K. At 298 K, the M37L,P40S mutant was found to be more stable than M37L by 1.5 kcal/mol. A combined analysis of GuHCl and calorimetric data was carried out to determine the enthalpy, entropy, and heat capacity change upon unfolding. At 298 K there was a large, stabilizing enthalpic effect in P40S though significant enthalpy-entropy compensation was observed and the two proteins had similar values of ΔCp. Thus, replacement of a Pro in the interior of an α helix can have substantial effects on protein stability.

The crystal structure of a covalently cross-linked Lactobacillus casei thymidylate synthase has been determined at 2.8 Å resolution. The sites for mutation to achieve the bis-disulfide linked dimer were identified using the disulfide modeling program MODIP. The mutant so obtained was found to be remarkably thermostable. This increase in stability has been reasoned to be entirely a consequence of the covalent gluing between the two subunits.

Alan Fersht has revised his classic text Structure and Mechanism (Freeman, 1985) and extended it to include applications of the protein engineering method to characterizing transition states in protein folding and enzyme catalysis. His earlier textbook was so good it scared off possible competitors, and its notable features are preserved in the new text. It provides enough information about techniques to make close connections between discussions of mechanism and actual data. It packs an amazing amount of information and viewpoint into a short space without losing readability. It does not just derive equations, it explains their physical meanings and it aims to develop the reader's physical insight. The assumptions behind an approach are laid out clearly and, when a topic is controversial, the author gives his own viewpoint at the end. The pace of research today is such that few leading scientists find time to write textbooks, and the scientific community is surely grateful to Alan Fersht for revising his. However, as he points out, a main section of the book covering classical enzymology has not been revised.

The monoclonal antibody 1696, directed against the HIV-1 protease, displays strong inhibitory effects toward the catalytic activity of the enzyme of both the HIV-1 and HIV-2 isolates. This antibody cross-reacts with peptides that include the N-terminus of the enzyme, a region that is well conserved in sequence among different viral strains and which, furthermore, is crucial for homodimerization to the active enzymatic form. This observation, as well as antigen-binding studies in the presence of an active site inhibitor, suggest that 1696 inhibits the HIV protease by destabilizing its active homodimeric form. To characterize further how the antibody 1696 inhibits the HIV-1 and HIV-2 proteases, we have solved the crystal structure of its Fab fragment by molecular replacement and refined it at 3.0 Å resolution. The antigen binding site has a deep cavity at its center, which is lined mainly by acidic and hydrophobic residues, and is large enough to accommodate several antigen residues. The structure of the Fab 1696 could form a starting basis for the design of alternative HIV protease-inhibiting molecules of broad specificity.

Short chain l-3-hydroxyacyl CoA dehydrogenase (SCHAD) is a soluble dimeric enzyme critical for oxidative metabolism of fatty acids. Its primary sequence has been reported to be conserved across numerous tissues and species with the notable exception of the pig heart homologue. Preliminary efforts to solve the crystal structure of the dimeric pig heart SCHAD suggested the unprecedented occurrence of three enzyme subunits within the asymmetric unit, a phenomenon that was thought to have hampered refinement of the initial chain tracing. The recently solved crystal coordinates of human heart SCHAD facilitated a molecular replacement solution to the pig heart SCHAD data. Refinement of the model, in conjunction with the nucleotide sequence for pig heart SCHAD determined in this paper, has demonstrated that the previously published pig heart SCHAD sequence was incorrect. Presented here are the corrected amino acid sequence and the high resolution crystal structure determined for pig heart SCHAD complexed with its NAD+ cofactor (2.8 Å; Rcryst = 22.4%, Rfree = 28.8%). In addition, the peculiar phenomenon of a dimeric enzyme crystallizing with three subunits contained in the asymmetric unit is described.

Unfolded apocytochrome c acquires an α-helical conformation upon interaction with lipid. Folding kinetic results below and above the lipid's CMC, together with energy transfer measurements of lipid bound states, and salt-induced compact states in solution, show that the folding transition of apocytochrome c from the unfolded state in solution to a lipid-inserted helical conformation proceeds via a collapsed intermediate state (IC). This initial compact state is driven by a hydrophobic collapse of the polypeptide chain in the absence of the heme group and may represent a heme-free analogue of an early compact intermediate detected on the folding pathway of cytochrome c in solution. Insertion into the lipid phase occurs via an unfolding step of IC through a more extended state associated with the membrane surface (IS). While IC appears to be as compact as salt-induced compact states in solution with substantial α-helix content, the final lipid-inserted state (Hmic) is as compact as the unfolded state in solution at pH 5 and has an α-helix content which resembles that of native cytochrome c.

Decorsin is an antagonist of integrin αIIb β3 and a potent platelet aggregation inhibitor. A synthetic gene encoding decorsin, originally isolated from the leech Macrobdella decora, was designed, constructed, and expressed in Escherichia coli. The synthetic gene was fused to the stII signal sequence and expressed under the transcriptional control of the E. coli alkaline phosphatase promoter. The protein was purified by size-exclusion filtration of the periplasmic contents followed by reversed-phase high-performance liquid chromatography. Purified recombinant decorsin was found to be indistinguishable from leech-derived decorsin based on amino acid composition, mass spectral analysis, and biological activity assays. Complete sequential assignments of 1H and proton bound 13C resonances were established. Stereospecific assignments of 21 of 25 nondegenerate β-methylene groups were determined. The RGD adhesion site recognized by integrin receptors was found at the apex of a most exposed hairpin loop. The dynamic behavior of decorsin was analyzed using several independent NMR parameters. Although the loop containing the RGD sequence is the most flexible one in decorsin, the conformation of the RGD site itself is more restricted than in other proteins with similar activities.

The metallo-β-lactamases require divalent cations such as zinc or cadmium for hydrolyzing the amide bond of β-lactam antibiotics. The crystal structure of the Zn2+-bound enzyme from Bacteroides fragilis contains a binuclear zinc center in the active site. A hydroxide, coordinated to both zinc atoms, is proposed as the moiety that mounts the nucleophilic attack on the carbonyl carbon atom of the β-lactam bond of the substrate. It was previously reported that the replacement of the active site Cys181 by a serine residue severely impaired catalysis while atomic absorption measurements indicated that binding of the two zinc ions remained intact. Contradicting data emerge from recent mass spectrometry results, which show that only a single zinc ion binds to the C181S metallo-β-lactamase. In the current study, the C181S mutant enzyme was examined at the atomic level by determining the crystal structure at 2.6 Å resolution. The overall structure of the mutant enzyme is the same as that of the wild-type enzyme. At the mutation site, the side chain of Ser181 occupies the same position as that of the side chain of Cys181 in the wild-type protein. One zinc ion, Zn1, is present in the crystal structure; however, the site of the second zinc ion, Zn2 is unoccupied. A water molecule is associated with Zn1, reminiscent of the hydroxide seen in the structure of the wild-type enzyme but farther from the metal. The position of the water molecule is off the plane of the carboxylate group of Asp103; therefore, the water molecule may be less nucleophilic than a water molecule which is coplanar with the carboxylate group.

We present an analysis of the assumptions behind some of the most commonly used methods for evaluating the goodness of the fit between a sequence and a structure. Our studies on a lattice model show that methods based on statistical considerations are easy to use and can capture some of the features of protein-like sequences and their corresponding native states, but unfortunately are incapable of recognizing, with certainty, the native-like conformation of a sequence among a set of decoys. Meanwhile, an optimization method, entailing the determination of the parameters of an effective free energy of interaction, is much more reliable in recognizing the native state of a sequence. However, the statistical method is shown to perform quite well in tests of protein design.

The mechanism of the disulfide-bond forming enzyme DsbA depends on the very low pKa of a cysteine residue in its active-site and on the relative instability of the oxidized enzyme compared to the reduced one. A thermodynamic cycle has been used to correlate its redox properties to the difference in the free energies of folding (ΔΔGred/ox) of the oxidized and reduced forms. However, the relation was proved unsatisfied for a number of DsbA variants. In this study, we investigate the thermodynamic and redox properties of a highly destabilized variant DsbAP151A (substitution of cis-Pro151 by an alanine) by the means of intrinsic tryptophan fluorescence and by high-sensitivity differential scanning calorimetry (HS-DSC). When the value of ΔΔGred/ox obtained fluorimetrically for DsbAP151A does not correlate with the value expected from its redox potential, the value of ΔΔGred/ox provided by HS-DSC are in perfect agreement with the predicted thermodynamic cycle for both wild-type and variant. HS-DSC data indicate that oxidized wild-type enzyme and the reduced forms of both wild-type and variant unfold according to a two-state mechanism. Oxidized DsbAP151A shows a deviation from two-state behavior that implies the loss of interdomain cooperativity in DsbA caused by Pro151 substitution. The presence of chaotrope in fluorimetric measurements could facilitate domain uncoupling so that the fluorescence probe (Trp76) does not reflect the whole unfolding process of DsbAP151A anymore. Thus, theoretical thermodynamic cycle is respected when an appropriate method is applied to DsbA unfolding under conditions in which protein domains still conserve their cooperativity.

Coiled coils consist of two or more amphipathic α-helices wrapped around each other to form a superhelical structure stabilized at the interhelical interface by hydrophobic residues spaced in a repeating 3–4 sequence pattern. Dimeric coiled coils have been shown to often form in a single step reaction in which association and folding of peptide chains are tightly coupled. Here, we ask whether such a simple folding mechanism may also apply to the formation of a three-stranded coiled coil. The designed 29-residue peptide LZ16A was shown previously to be in a concentration-dependent equilibrium between unfolded monomer (M), folded dimer (D), and folded trimer (T ). We show by time-resolved fluorescence change experiments that folding of LZ16A to D and T can be described by [equation] and [equation]. The following rate constants were determined (25 °C, pH 7): k1 = 7.8 × 104 M−1 s−1, k−1 = 0.015 s−1, k2 = 6.5 × 105 M−1 s−1, and k−2 = 1.1 s−1. In a separate experiment, equilibrium binding constants were determined from the change with concentration of the far-ultraviolet circular dichroism spectrum of LZ16A and were in good agreement with the kinetic rate constants according to KD = k1/2k−1 and KT = k2/k−2. Furthermore, pulsed hydrogen-exchange experiments indicated that only unfolded M and folded D and T were significantly populated during folding. The results are compatible with a two-step reaction in which a subpopulation of association competent (e.g., partly helical) monomers associate to dimeric and trimeric coiled coils.