Structural diversity in a family of homologous proteins.

An interesting example of a structurally diverse group of sequentially homologous proteins is analyzed at the level of molecular interactions. In this family, the EF-hand calcium-binding proteins, there are examples of at least three distinct mutual positions of the N and C-terminal domains, despite significant sequence homology between all members of this family. Why does a particular protein choose one arrangement over another? To answer this question, detailed models of all proteins in their native structures as well as all alternative sequence/structure combinations are built by comparative modeling. By studying and comparing interactions stabilizing native structures and destabilizing alternative conformations, it is possible to gain insight into how such conformational diversity is achieved. It is shown that some mechanisms used to achieve it are: correlated mutations on the surface of two units and the presence of additional domains/chain fragments stabilizing desired topologies. The implications of these findings, both for structure predictions for other members of this family as well as the general problem of quaternary structure formation, are discussed.

Conservative mutation Met8 --> Leu affects the folding process and structural stability of squash trypsin inhibitor CMTI-I.

Protein molecules can accommodate a large number of mutations without noticeable effects on their stability and folding kinetics. On the other hand, some mutations can have quite strong effects on protein conformational properties. Such mutations either destabilize secondary structures, e.g., alpha-helices, are incompatible with close packing of protein hydrophobic cores, or lead to disruption of some specific interactions such as disulfide cross links, salt bridges, hydrogen bonds, or aromatic-aromatic contacts. The Met8 --> Leu mutation in CMTI-I results in significant destabilization of the protein structure. This effect could hardly be expected since the mutation is highly conservative, and the side chain of residue 8 is situated on the protein surface. We show that the protein destabilization is caused by rearrangement of a hydrophobic cluster formed by side chains of residues 8, Ile6, and Leu17 that leads to partial breaking of a hydrogen bond formed by the amide group of Leu17 with water and to a reduction of a hydrophobic surface buried within the cluster. The mutation perturbs also the protein folding. In aerobic conditions the reduced wild-type protein folds effectively into its native structure, whereas more then 75% of the mutant molecules are trapped in various misfolded species. The main conclusion of this work is that conservative mutations of hydrophobic residues can destabilize a protein structure even if these residues are situated on the protein surface and partially accessible to water. Structural rearrangement of small hydrophobic clusters formed by such residues can lead to local changes in protein hydration, and consequently, can affect considerably protein stability and folding process.

Luminescence of peptide-bound terbium ions. Determination of binding constants.

Luminescence of Tb3+ ions bound to a calmodulin fragment has been studied. It is shown that during their lifetime excited ions dissociate from the peptide. If concentration of free peptide is high enough they can be coordinated again. As a consequence, observed terbium luminescence lifetime and intensity depends not only on binding equilibrium, but also on concentration of free peptide molecules. In such a system terbium binding constant cannot be correctly determined by simple steady-state measurements of luminescence intensities. Instead, terbium luminescence decay curves measured at various peptide concentrations must be analysed. Such an analysis has been made for a fragment of the IIIrd calcium binding domain of rat testis calmodulin. Rate constant of terbium association and the equilibrium binding constant corresponding to the best fit of theoretical functions to experimental points have been determined.

Synthesis, cloning and expression in Escherichia coli of a gene coding for the Met8-->Leu CMTI I--a representative of the squash inhibitors of serine proteinases.

A chemically synthesized gene coding for a Cucurbita maxima trypsin inhibitor modified at position P'3 (Met8-->Leu CMTI I), i.e. at the third position downstream of the reactive site bond (Arg5-Ile), was cloned into a derivative of the plasmid pAED4 that utilizes a T7 expression system. The gene was expressed in Escherichia coli as a fusion protein that accumulates in inclusion bodies. After reduction and CNBr cleavage of the fusion protein followed by oxidative refolding and reverse-phase HPLC, about 5 mg of pure protein was obtained per 1 of cell culture. Association constants of recombinant Leu-8-CMTI I with bovine beta-trypsin and human cathepsin G are the same, within experimental error, as for CMTI I isolated from a natural source.

Methods of peptide conformation studies.

In solution most of the peptides assume multiple flexible conformations. Determination of the dominant conformers and evaluation of their populations is the aim of peptide conformation studies, in which theoretical and experimental methods play complementary roles. Molecular dynamics or Monte Carlo methods are quite effective in searching the conformational space accessible to a peptide but they are not able to estimate, precisely enough, the populations of various conformations. Therefore, they must be supplemented by experimental data. In this paper, a short review of the experimental methods, most widely used in peptide conformational studies, is presented. Among them NMR plays the leading role. Valuable information is also obtained from hydrogen exchange, fluorescence resonance energy transfer, and circular dichroism measurements. The advantages and shortcomings of these methods are discussed.

Helix-coil transition theories. Are they correct?

Principles of contemporary theoretical description of alpha-helix formation by polypeptide chains in water solution are shortly presented and critically discussed. The theory treats the unfolded state of a peptide as "random coil"--an ideal conformation quite distant from reality. We suggest that for this reason the helix propagation parameters of amino-acid residues, determined using series of model peptides with different sequential patterns, are not the same. Interpretation of the so called "nucleation parameter" is erroneous. In fact, it is not determined by the helix nucleation process but rather by a specific situation of residues at the helix N- and C-termini, and it strongly depends on solvation of their NH and CO groups, respectively. Consequently, helical segments with terminal sequences dominated by residues with strongly hydrophobic, bulky side chains can be very unstable. We postulate that an unexpectedly high stability of very short, pre-nucleated helices studied by us arises from a "helix end separation effect": separated helix termini are better solvated than when they overlap each other. Because of this effect, helix initiation may be much more difficult than predicted by the theoretical "helix nucleation parameters".

A comparative CD and fluorescence study of a series of model calcium-binding peptides.

Lanthanide-saturated peptides analogous to calcium-binding loops of EF-hand proteins can be used to stabilize the alpha-helical structure of peptide or protein segments attached to their C-termini. To study conformational properties of such loop-containing hybrids it is necessary to produce them in bacteria. In peptides obtained in this way the helix will be destabilized by the negatively charged C-terminal alpha-carboxyl groups. We propose to block them by the homoserine lactone. The results presented in this paper indicate that the presence of the lactone even at the C-terminus of the loop does not have any negative effect on the loop helix-nucleation ability. On the other hand, the presence of the alpha-NH3+ at the loop N-terminus leads to a drop of metal-binding constant and loss of the rigid structure of the alpha-helical segment of the loop. The alpha-amino group separated by one glycine residue from the loop N-terminus should also be avoided because it perturbs the conformation of the N-terminal part of the loop and may reduce the loop affinity to lanthanide ions.

Molecular cloning and expression in Escherichia coli of a gene coding for bovine S100A1 protein and its Glu32-->Gln and Glu73-->Gln mutants.

Calcium binding S100A1 protein consists of two S100 alpha subunits. On the basis of sequence homology to other S100 proteins it is believed that the binding loops are formed by amino-acid residues 19-32 and 62-73 of S100 alpha polypeptide chain. In the oxidized form of the protein the subunits are linked covalently with each other by a disulphide bond between their Cys85 residues. A synthetic gene coding for bovine S100 alpha subunit was constructed and cloned into a derivative of pAED4 plasmid. The gene was expressed in Escherichia coli utilizing the T7 expression system. The expression products were purified and identified using mass spectrometry and by sequencing of their N- and C-termini. Three different forms (a, b, and c) of S100 alpha were produced: with the native sequence, with the initiator methionine at the N-terminus, and with an additional alanine at the C-terminus as well as with the initiator methionine. The material was partly oxidized. Interestingly, only the homodimers of a, b, and c species were formed. The total yield of the protein was about 50 mg/l of culture. Genes coding for Glu32-->Gln and Glu73-->Gln mutants of S100 alpha were obtained by site-directed mutagenesis and expressed in the same system. In both cases similar mixtures of oxidized and reduced a, b, and c species have been obtained. The total yield of E73Q mutant is similar to that of the native protein and that of E32Q lower by about a half. As expected, the mutants of S100 alpha subunits bind only one calcium ion.

Conformational properties of Ca2+-binding segments of proteins from the troponin C superfamily.

The troponin C superfamily consists of about 100 Ca2+-binding proteins. Sequence variations observed in these proteins have been analyzed and lead to the following conclusions. (1) There are some strict rules defining the set of calcium ligands necessary for effective Ca2+ binding. (2) If they are fulfilled, the Ca2+ binding constant depends on tertiary interactions within a protein, as well as the free energy of secondary structures of its polypeptide chain. The former provide a constant contribution to the free energy of protein folding and the Ca2+-binding process. (3) The observed variety in Ca2+-binding constants of these proteins results from the various abilities of segments of these proteins to assume the correct secondary structure.

Investigations on purine and pyrimidine bases stacking associations in aqueous solutions by the fluorescence quenching method. I. Autoassociation of 2-aminopurine.

A general equation was derived, describing fluorescence quantum yield and lifetime of an autoassociating compound in liquid solutions. The autoassociation of 2-aminopurine in aqueous solution was examined within the range from 0 to 90 degrees C. The compound seemed to associate cooperatively. The thermodynamic parameters of polymerization change with temperature, so that its free enthalypy deltaG = 0.0797 T2 + 45.4 T - 7893. The dimerization enthalpy and entropy are approximately temperature-independent (deltaH2 = -4.17 kcal/mol deltas2 = -10.9 e.u.), although the function: delta g2 = -0.0308T2 +30.3T - 7213 fits experimental points better. The observed dependences can be explained by the increasing role of the hydrophobic effect with temperature and size of the aggregates. The association rate constants were determined, and a two-step reaction mechanism was demonstrated. The first step is diffusion-controlled. The second is characterized by an activation energy of approximately 2 kcal/mol and an encounter distance of approximately 8.3 angstroms.

Investigations on purine and pyrimidine bases stacking associations in aqueous solutions by the fluorescence quenching method. II. Heteroassociation between 2-aminopurine and thymidine.

Heteroassociation between A and B compounds in liquid solution was considered. Provided that concentration of A molecules is low, a general equation describing fluorescence quantum yield and lifetime of compound A as a function of B molecules concentration was derived. The heteroassociation between 2-aminopurine and thymidine in aqueous solutions was examined within the range of temperatures 0 to 90 degrees C. The equilibrium constants of the first step of association, namely heterodimer formation, were determined and its thermodynamic parameters (deltaH equals - 2.76 kcal/mol, deltaS equals - 5.9 e.u.) were calculated. The observed changes of the stacking rate constants with temperature confirm the two-step mechanism of the reaction. The activation energy (approximately 10.7 angstroms) are only slightly larger than in the case of 2-aminopurine autoassociation, most probably because of a stronger solvation of thymidine molecules.

Conformational study of two synthetic peptides with sequence analogies to the N-terminal fragment of RNase A.

The conformational properties of two synthetic model peptides, AEAAHAAEAAHMG (PA) and AEAAHAFEAAHMG (PF), have been studied using CD and 1H-NMR methods. In both peptides, glutamate and histidine residues are situated in such a way that two salt bridges between Glu- (i) and His+ (i + 3) can be formed. A salt bridge of this type (Glu- 9-His+ 12) was postulated previously to stabilize, to a great extent, the alpha-helical conformation of isolated N-terminal fragments of RNase A: C-peptide and S-peptide (A. Bierzynski, P.S. Kim and R.L. Baldwin, Proc. Natl. Acad. Sci. U.S.A. 79 (1982) 2470). Although in both PA and PF salt bridges between glutamates and histidines are formed, as demonstrated by the pH-titration curves of the glutamate gamma-proton signals, no traces of helical conformation have been detected. Evidently, the Glu- (i)-His+ (i + 3) salt bridges do not stabilize the alpha-helical conformation. A comparative analysis of PA and PF NMR spectra provides strong evidence that the phenylalanyl ring in PF interacts not only with the hydrophobic methyl groups of almost all alanine residues but also with the histidine rings and the glutamate side chains in their protonated as well as deprotonated forms. Similar interactions, involving Phe 8, can be expected in the N-terminal fragments of RNase and should be taken into account as an important factor determining the conformational properties of C- and S-peptides.

Investigations on purine and pyrimidine bases stacking associations in aqueous solutions by the fluorescence quenching method. III. Intramolecular association of 9,9'-[1,3-propylene]-bis-2-aminopurine.

General equations relating fluorescence quantum yield and lifetime of a compound with its intramolecular stacking equilibrium and kinetics were derived. Intramolecular stacking association of 9,9'-[1,3-propylene]-bis-2-aminopurine in aqueous solution was examined within the range of temperatures from 0 to 90 degrees C. A two-state thermodynamic model of the association was verified. The stacking enthalpy and entropy can be taken, with a good approximation, as temperature-independent (deltaH equals - 2.0 kcal/mol, deltaS=-3.25 e.u.) although the function deltaG=-0.00886T2 + 8.847 T - 2876 describes more precisely the observed changes of stacking free enthalpy with temperature. The association rate constants were detemined. Activation energy of the reaction (2 kcal/mol) is the same as in the case of association between free 2-aminopurine molecules. It confirms a two-step mechanism of the process. The advantages and shortcomings of the fluorescence quenching method are discussed.

Conformational role of His-12 in C-peptide of ribonuclease A.

Possible interactions of the His-12 ring with other side chain and backbone groups of C-peptide lactone (CPL) are discussed. The works published so far are critically reviewed and compared with the latest results obtained by the authors. The main new conclusion is that in the helical conformation of CPL, the Phe-8 and His-12 rings are clustered together. Studies of Phe-8----Ala analogs of CPL and calculations of ring current effects satisfactorily explain the observed environmental shifts of Phe-8 and His-12 protons in NMR spectra of CPL. Interaction between both rings is favorable for alpha-helix formation, but cannot explain an increase in helix stability related with protonation of His-12. This effect arises from favorable interactions of the charged His+-12 ring with the helix backbone.

Fluorescence quenching and spin label electron spin resonance studies of stacking self-association in aqueous solutions of 2-aminopurine riboside and its 5'-mono- and -diphosphate.

The autoassociation of 2-aminopurine riboside (rn2Pur) and its 5'-mono- (P-rn2Pur) and 5'-diphosphate (PP-rn2Pur) in neutral aqueous solutions was investigated using fluorescence quenching and ESR spin-label methods within the range 276-358 K. Respective equilibrium constants and thermodynamic functions were derived therefrom assuming two models of infinite autoassociation: (i) an isodesmic one (K2 = K3 = ... Kp), and (ii) one in which K2 no equal to K2 = K4 ... Kp. Comparative analysis of these data and that of the parent 2-aminopurine, obtained previously, allowed us to formulate the following conclusions: (1) the mechanism of autoassociation of rn2Pur varies with temperature in such a way that a T = 318 K the isodesmic model is fulfilled (K2 = Kp); at high temperatures Kp/K2 greater than 1, i.e. the process is cooperative, while at lower temperatures it becomes anticooperative (Kp/K2 greater less than 1); (2) at 298 K the tendency to autoassociation decreases in the order; rn2Pur greater than P-rn2Pur greater than PP-rn2Pur; (3) rn2Pur forms highly packed complexes with the bases stacked and the ribofuranose residues interacting via hydrogen bonds or water bridges; (4) autoassociation of P-rn2Pur and PP-rn2Pur is mainly governed by stacking of the bases, while the ribose phosphate residues attain a trans configuration corresponding to the lowest electrostatic repulsion between charged phosphate groups; even at high ionic strength (I = 0.8), a positive electrostatic contribution to the free enthalpy of autoassociation is observed; (5) the two methods employed gave similar results for P-rn2Pur, but somewhat different ones for rn2Pur because the presence of the spin label (nitroxide stable radical) at the 2'(3')-OH group of the ribose residues prevents its interaction via hydrogen bonding with an unlabeled one of an adjacent nucleoside.

Deprotonation of Glu9 destabilizes the alpha-helix in C-peptide of RNase A.

pK values of Glu2, Glu9, and His12 in the lactone and carboxylate forms of C-peptide at 4.5 and 21 degrees, 0.1 M NACl, have been determined from pH-tritration shifts of n.m.r. signals of the glutamate gamma and histidine ring protons. These have been used in an analysis of the well known pH-induced helix-coil transition in the C-peptide lactone. It has been shown that Glu12 deprotonation results in a twofold increase of the helix content, whereas deprotonation of Glu9 leads to a 28% drop. This latter effect is probably related to Glu-9-His+12 salt bridge formation. The transition within the pH range 5.5-7.5 follows exactly the deprotonation curve of His+12 residue.

Theoretical estimation of the calcium-binding constants for proteins from the troponin C superfamily based on a secondary structure prediction method. I. Estimation procedure.

Proteins belonging to the TNC superfamily are known to be built of two, three, four, or six domains of closely similar amino acid sequences. Each domain binds no more than one calcium ion and shows a characteristic helix-loop-helix structure when in the calcium-bound state. Conformational properties of all the domains known so far have been analysed by us using a secondary structure prediction method (Garnier, J., Osguthorpe, D.J. & Robson, B. (1978). J. molec. Biol. 120, 97). Significant differences in distribution of residues predicted as being in the helical, beta-turn, and coil conformations have been found between the strongly, weakly, and non-binding domains. We could determine the ideal prediction pattern characteristic for the domains with the highest affinity for calcium. On the basis of our analysis and observations made by other authors we worked out a few simple rules which made it possible to compare conformational properties of a given domain with the ideal reference pattern and estimate, in this way, the Ca2+-binding constant of the domain. In native proteins the domains are known to be organized in pairs. The Ca2+-binding constant for a two-domain region could be evaluated from the sum of the estimation points attributed to each of its components. Using our method it is possible to predict the binding constants of typical domains and two-domain regins with a precision of one order of magnitude. Data on amino acid sequences and calcium-binding constants of all known proteins, believed to be the members of the TNC superfamily, have been reviewed. References to virtually all papers published on this subject before the end of 1987 are given.

Theoretical estimation of the calcium-binding constants for proteins from the troponin C superfamily based on a secondary structure prediction method. II. Applications.

Ca2+-binding properties of the following proteins, classified as members of the troponin C (TNC) superfamily have been discussed: TNCs, calmodulins (CaMs), vitamin D-dependent calcium-binding proteins (CaBPs), myosin light chains (LCs), S-100 chains, parvalbumins (PVs), oncomodulin (OCM), sarcoplasmic calcium binding proteins (SCPs), calcineurin B (CB) and calcium vector protein (CaVP). Assuming the most probable domain pairing, the Ca2+-binding constants of these proteins have been predicted from their sequences using the method presented in the preceding paper. The results are critically compared with the available experimental data. For some proteins (TNCs, CaMs, CaBPs, LCs, CB and CaVP) our predictions are consistent with the experimental results. For the others, substantial discrepancies between the predicted and measured KCa values are observed. They result from some structural peculiarities of those proteins: a unique, three-domain organization in the case of PVs and OCM, unusual sequences of binding loops in the case of S-100 and a lack of a standard helix-loop-helix organization of Ca2+-binding domains in the case of SCPs.