18th Sir Hans Krebs lecture. Knowledge-based protein modelling and design

Predicting protein secondary structure content. A tandem neural network approach

A priori knowledge of secondary structure content can be of great use in theoretical and experimental determination of protein structure. We present a method that uses two computer-simulated neural networks placed in "tandem" to predict the secondary structure content of water-soluble, globular proteins. The first of the two networks, NET1, predicts a protein's helix and strand content given information about the protein's amino acid composition, molecular weight and heme presence. Because NET1 contained more adjustable parameters (network weights) than learning examples, this network experienced problems with memorization, which is the inability to generalize onto new, never-seen-before examples. To overcome this problem, we designed a second network, NET2, which learned to determine when NET1 was in a state of generalization. Together, these two networks produce prediction errors as low as 5.0% and 5.6% for helix and strand content, respectively, on a set of protein crystal structures bearing little homology to those used in network training. A comparison between three other methods including a multiple linear regression analysis, a non-hidden-node network analysis and a secondary structure assignment analysis reveals that our tandem neural network scheme is, indeed, the best method for predicting secondary structure content. The results of our analysis suggest that the knowledge of sequence information is not necessary for highly accurate predictions of protein secondary structure content.

Structure-function relationships in the cysteine proteinases actinidin, papain and papaya proteinase omega. Three-dimensional structure of papaya proteinase omega deduced by knowledge-based modelling and active-centre characteristics determined by two-hydronic-state reactivity probe kinetics and kinetics of catalysis

1. A model of the three-dimensional structure of papaya proteinase omega, the most basic cysteine proteinase component of the latex of papaya (Carica papaya), was built from its amino acid sequence and the two currently known high-resolution crystal structures of the homologous enzymes papain (EC 3.4.22.2) and actinidin (EC 3.4.22.14). The method used a knowledge-based approach incorporated in the COMPOSER suite of programs and refinement by using the interactive graphics program FRODO on an Evans and Sutherland PS 390 and by energy minimization using the GROMOS program library. 2. Functional similarities and differences between the three cysteine proteinases revealed by analysis of pH-dependent kinetics of the acylation process of the catalytic act and of the reactions of the enzyme catalytic sites with substrate-derived 2-pyridyl disulphides as two-hydronic-state reactivity probes are reported and discussed in terms of the knowledge-based model. 3. To facilitate analysis of complex pH-dependent kinetic data, a multitasking application program (SKETCHER) for parameter estimation by interactive manipulation of calculated curves and a simple method of writing down pH-dependent kinetic equations for reactions involving any number of reactive hydronic states by using information matrices were developed. 4. Papaya proteinase omega differs from the other two enzymes in the ionization characteristics of the common (Cys)-SH/(His)-Im+H catalytic-site system and of the other acid/base groups that modulate thiol reactivity towards substrate-derived inhibitors and the acylation process of the catalytic act. The most marked difference in the Cys/His system is that the pKa for the loss of the ion-pair state to form -S-/-Im is 8.1-8.3 for papaya proteinase omega, whereas it is 9.5 for both actinidin and papain. Papaya proteinase omega is similar to actinidin in that it lacks the second catalytically influential group with pKa approx. 4 present in papain and possesses a catalytically influential group with pKa 5.5-6.0. 5. Papaya proteinase omega occupies an intermediate position between actinidin and papain in the sensitivity with which hydrophobic interaction in the S2 subsite is transmitted to produce changes in transition-state geometry in the catalytic site, a fact that may be linked with differences in specificity in P2-S2 interaction exhibited by the three enzymes.(ABSTRACT TRUNCATED AT 400 WORDS)

Identification of a functionally conserved surface region of rat cytochromes P450IA

A region of rat cytochrome P450IA1 at residues 294-301 (Gln-Asp-Arg-Arg-Leu-Asp-Glu-Asn), equivalent to a proinhibitory region of cytochrome P450IA2, was identified by sequence alignment. Anti-peptide antibodies were successfully raised when the peptide was coupled through either its N- or its C-terminus to carrier protein, but no antibodies were produced against the so-called multiple peptide antigen, which consisted of eight copies of the peptide attached through its C-terminus to a synthetic base. Both of the anti-peptide antibodies bound specifically to cytochrome P450IA1 in the rat, as shown by e.l.i.s.a. and immunoblotting. They inhibited microsomal aryl hydrocarbon hydroxylase activity and the mutagenic activation of 2-acetylaminofluorene (these reactions are catalysed by cytochrome P450IA1), but not high-affinity phenacetin O-de-ethylation activity, which is catalysed by cytochrome P450IA2. However, there was differences in the properties of the two antisera in their binding to cytochromes P450IA1 in species other than the rat, their relative binding to the multiple peptide antigen, the yield of antibody following affinity purification using peptide coupled through its N-terminus to CNBr-activated Sepharose, and the binding of the purified preparations to N- and C-terminal-coupled peptide conjugates. These observations indicated that the antibodies were directed to the region of the peptide opposite to the end which was coupled to the carrier protein. Nevertheless, both of the antibody preparations bound equally well to the target cytochrome P450, thus indicating that, in the native protein, the whole of the peptide region is exposed on the surface of cytochrome P450IA1 and is available for binding by the antibodies. The role of this region appears to be the same in both cytochromes P450IA1 and P450IA2, despite the difference in its primary structure in the two cytochromes P450.

Model for the structure of formaldehyde dehydrogenase based on alcohol dehydrogenase

The three-dimensional structure of rat liver formaldehyde dehydrogenase (FALDH), previously known as class III alcohol dehydrogenase, was constructed using computer graphics and computer programs developed for model building. The construction is based on horse liver alcohol dehydrogenase (EE-ADH), whose structure has been elucidated by X-ray crystallography. The high sequence homology between the two enzymes makes knowledge-based modelling feasible in this case. The model shows a remarkable similarity to horse liver alcohol dehydrogenase especially in the NAD-binding domain. Certain mutations, and the one insertion in FALDH compared to EE-ADH in particular, have cause important changes in the substrate binding site, and thus aliphatic alcohols have been replaced by hemi-thioacetals as favourable substrates.

A new method for building protein conformations from sequence alignments with homologues of known structure

We describe a largely automatic procedure for building protein structures from sequence alignments with homologues of known structure. This procedure uses simple rules by which multiple sequence alignments can be translated into distance and chirality constraints, which are then used as input for distance geometry calculations. By this means one obtains an ensemble of conformations for the unknown structure that are compatible with the rules employed, and the differences among these conformations provide an indication of the reliability of the structure prediction. The overall approach is demonstrated here by applying it to several Kazal-type trypsin inhibitors, for which experimentally determined structures are available. On the basis of our experience with these test problems, we have further predicted the conformation of the human pancreatic secretory trypsin inhibitor, for which no experimentally determined structure is presently available.

Molecular modeling and dynamics of biologically active peptides: application to neuropeptide Y

A methodology is presented to allow results from molecular dynamics simulations to be combined with pharmacological binding and activity measurements in order to study the structure-function relationships of neuropeptide Y. This approach is a general one and should also be applicable to other peptides for which stable structures are known to exist in solution. The basis of the method is the calculation of energetically stable structures via a simulate and test approach. This approach uses molecular dynamics simulations to search conformational space in order to find low potential energy structures. The energetic stability of the structures is then tested via additional simulations. Once energetic stability has been achieved, perturbations of the structure may be performed via molecular modeling. The simulate and test approach is then used to obtain energetically stable structures for the perturbed compound. Comparison between the energetically stable starting and perturbed structures can then be made concerning both structural and dynamic changes. By using an energetically stable structure prior to the perturbation, the assumption can be made that the calculated differences are primarily due to the perturbation rather than to one or both of the structures reaching a more energetically favorable state. It should be emphasized that the calculations are being performed employing a limited physical model such that the influence of that model on the observed results must always be taken into account.

Identification of native protein folds amongst a large number of incorrect models. The calculation of low energy conformations from potentials of mean force

We present an approach that is able to detect native folds amongst a large number of non-native conformations. The method is based on the compilation of potentials of mean force of the interactions of the C beta atoms of all amino acid pairs from a database of known three-dimensional protein structures. These potentials are used to calculate the conformational energy of amino acid sequences in a number of different folds. For a substantial number of proteins we find that the conformational energy of the native state is lowest amongst the alternatives. Exceptions are proteins containing large prosthetic groups, Fe-S clusters or polypeptide chains that do not adopt globular folds. We discuss briefly potential applications in various fields of protein structural research.

The 3-D structure of HIV-1 proteinase and the design of antiviral agents for the treatment of AIDS

A proteinase is essential for replication of HIV. Cloning and chemical synthesis have provided a sufficient supply of HIV-1 proteinase for the determination of its three-dimensional structure. Analogies between the structures of HIV-1 proteinase and the mammalian enzyme renin, which is involved in the control of blood pressure, have given important clues concerning the design of specific inhibitors that have antiviral activity.

Tertiary structural constraints on protein evolutionary diversity: templates, key residues and structure prediction

The pattern of residue substitution in divergently evolving families of globular proteins is highly variable. At each position in a fold there are constraints on the identities of amino acids from both the three-dimensional structure and the function of the protein. To characterize and quantify the structural constraints, we have made a comparative analysis of families of homologous globular proteins. Residues are classified according to amino acid type, secondary structure, accessibility of the sidechain, and existence of hydrogen bonds from sidechain to other sidechains or peptide carbonyl or amide functions. There are distinct patterns of substitution especially where residues are both solvent inaccessible and hydrogen bonded through their sidechains. The patterns of residue substitution can be used to construct templates or to identify 'key' residues if one or more structures are known. Conversely, analysis of conversation and substitution across a large family of aligned sequences in terms of substitution profiles can allow prediction of tertiary environment or indicate a functional role. Similar analyses can be used to test the validity of putative structures if several homologous sequences are available.

Calculation of conformational ensembles from potentials of mean force. An approach to the knowledge-based prediction of local structures in globular proteins

We present a prototype of a new approach to the folding problem of polypeptide chains. This approach is based on the analysis of known protein structures. It derives the energy potentials for the atomic interactions of all amino acid residue pairs as a function of the distance between the involved atoms. These potentials are then used to calculate the energies of all conformations that exist in the data base with respect to a given sequence. Then, by using only the most stable conformations, clusters of the most probable conformations for the given sequence are obtained. To discuss the results properly we introduce a new classification of segments based on their conformational stability. Special care is taken to allow for sparse data sets. The use of the method is demonstrated in the discussion of the identical oligopeptide sequences found in different conformations in unrelated proteins. VNTFV, for example, adopts a beta-strand in ribonuclease but it is found in an alpha-helical conformation in erythrocruorin. In the case of VNTFV the ensemble obtained consists of a single cluster of beta-strand conformations, indicating that this may be the preferred conformation for the pentapeptide. When the flanking residues are included in the calculation the hepapeptide P-VNTFV-H (ribonuclease) again yields an ensemble of beta-strands. However, in the ensemble of D-VNTFV-A (erythrocruorin) the major cluster is of alpha-helical type. In the present study we concentrate on the local aspects of protein conformations. However, the theory presented is quite general and not restricted to oligopeptides. We indicate extensions of the approach to the calculation of global conformations of proteins as well as conceivable applications to a number of molecular systems.

Structural features of neutral protease from Bacillus subtilis deduced from model-building and limited proteolysis experiments

The overall folding of neutral protease from Bacillus subtilis has been predicted by computer-aided modelling, taking as a basis the known three-dimensional structure of thermolysin. As expected from the 50% similarity of sequence between the two proteins, the structure of B. subtilis protease is similar to that of thermolysin, including the two-domain topology and location of elements of regular secondary structure (helices and strands), whereas specific differences were predicted in loop regions. A protruding and loose loop predicted in B. subtilis has been detected also experimentally by a limited proteolysis approach. Incubation of B. subtilis protease at pH 9.0 for 24 h at room temperature with trypsin at 20:1 ratio (by mass) leads to a specific and almost quantitative fission of the Arg214-Asn215 peptide bond located in a highly exposed, and thus probably flexible, loop of the protease. On the other hand, thermolysin was completely resistant to tryptic hydrolysis when reacted under identical conditions. The 'nicked' B. subtilis protease can be isolated by gel filtration chromatography at neutral pH, whereas the two constituting fragments 1-214 and 215-300 are separated under protein-denaturing conditions. Overall, these results indicate that the limited proteolysis approach can pinpoint a peculiar difference in surface structure between the two similar protein molecules of B. subtilis neutral protease and thermolysin and emphasize the potential use of proteolytic enzymes as structural probes of globular proteins.

Definition of general topological equivalence in protein structures. A procedure involving comparison of properties and relationships through simulated annealing and dynamic programming

A protein is defined as an indexed string of elements at each level in the hierarchy of protein structure: sequence, secondary structure, super-secondary structure, etc. The elements, for example, residues or secondary structure segments such as helices or beta-strands, are associated with a series of properties and can be involved in a number of relationships with other elements. Element-by-element dissimilarity matrices are then computed and used in the alignment procedure based on the sequence alignment algorithm of Needleman & Wunsch, expanded by the simulated annealing technique to take into account relationships as well as properties. The utility of this method for exploring the variability of various aspects of protein structure and for comparing distantly related proteins is demonstrated by multiple alignment of serine proteinases, aspartic proteinase lobes and globins.

Evolution of a protein superfamily: relationships between vertebrate lens crystallins and microorganism dormancy proteins

A search of sequence databases shows that spherulin 3a, an encystment-specific protein of Physarum polycephalum, is probably structurally related to the beta- and gamma-crystallins, vertebrate ocular lens proteins, and to Protein S, a sporulation-specific protein of Myxococcus xanthus. The beta- and gamma-crystallins have two similar domains thought to have arisen by two successive gene duplication and fusion events. Molecular modeling confirms that spherulin 3a has all the characteristics required to adopt the tertiary structure of a single gamma-crystallin domain. The structure of spherulin 3a thus illustrates an earlier stage in the evolution of this protein superfamily. The relationship of beta- and gamma-crystallins to spherulin 3a and Protein S suggests that the lens proteins were derived from an ancestor with a role in stress-response, perhaps a response to osmotic stress.

Molecular anatomy: phyletic relationships derived from three-dimensional structures of proteins

A distance measure that reflects the dissimilarity among structures has been developed on the basis of the three-dimensional structures of similar proteins, this being totally independent of sequence in the sense that only the relative spatial positions of mainchain alpha-carbon atoms need be known. This procedure leads to phyletic relationships that are in general correlated with the sequence phylogenies based on residue type. Such relationships among known protein three-dimensional structures are also a useful aid to their classification and selection in knowledge-based modeling using homologous structures. We have applied this approach to six homologous sets of proteins: immunoglobulin fragments, globins, cytochromes c, serine proteinases, eye-lens gamma crystallins, and dinucleotide-binding domains.

X-ray analysis of HIV-1 proteinase at 2.7 A resolution confirms structural homology among retroviral enzymes

Knowledge of the tertiary structure of the proteinase from human immunodeficiency virus HIV-1 is important to the design of inhibitors that might possess antiviral activity and thus be useful in the treatment of AIDS. The conserved Asp-Thr/Ser-Gly sequence in retroviral proteinases suggests that they exist as dimers similar to the ancestor proposed for the pepsins. Although this has been confirmed by X-ray analyses of Rous sarcoma virus and HIV-1 proteinases, these structures have overall folds that are similar to each other only where they are also similar to the pepsins. We now report a further X-ray analysis of a recombinant HIV-1 proteinase at 2.7 A resolution. The polypeptide chain adopts a fold in which the N- and C-terminal strands are organized together in a four-stranded beta-sheet. A helix precedes the single C-terminal strand, as in the Rous sarcoma virus proteinase and also in a synthetic HIV-1 proteinase, in which the cysteines have been replaced by alpha-aminobuytric acid. The structure reported here provides an explanation for the amino acid invariance amongst retroviral proteinases, but differs from that reported earlier in some residues that are candidates for substrate interactions at P3, and in the mode of intramolecular cleavage during processing of the polyprotein.

Protein motifs and data-base searching

Protein structure and sequence motifs are now recognized for many different protein families and topologies. To aid identification and use of these motifs in modelling and prediction, it has become necessary to establish consistent data bases of protein structure, including not only coordinates, but also derived data such as secondary structure location and solvent accessibilities. This article describes first attempts to construct such data bases and explains how they can be used.

Conformation of beta-hairpins in protein structures. A systematic classification with applications to modelling by homology, electron density fitting and protein engineering

A systematic classification of beta-hairpin structures which takes into account the polypeptide chain length and hydrogen bonding between the two antiparallel beta-strands is described. We have used this classification of beta-hairpin structures and their specific sequence pattern to derive rules which demonstrate its usefulness in assisting modelling beta-hairpins. These rules can be applied to comparative model building, modelling into electron density and in the prediction of conformation of beta-hairpins to aid protein engineering.

Rebuilding flavodoxin from C alpha coordinates: a test study

The tertiary structure of flavodoxin has been model built from only the X-ray crystallographic alpha-carbon coordinates. Main-chain atoms were generated from a dictionary of backbone structures. Side-chain conformations were initially set according to observed statistical distributions, clashes were resolved with reference to other knowledge-based parameters, and finally, energy minimization was applied. The RMSD of the model was 1.7 A across all atoms to the native structure. Regular secondary structural elements were modeled more accurately than other regions. About 40% of the chi 1 torsional angles were modeled correctly. Packing of side chains in the core was energetically stable but diverged significantly from the native structure in some regions. The modeling of protein structures is increasing in popularity but relatively few checks have been applied to determine the accuracy of the approach. In this work a variety of parameters have been examined. It was found that close contacts, and hydrogen-bonding patterns could identify poorly packed residues. These tests, however, did not indicate which residues had a conformation different from the native structure or how to move such residues to bring them into agreement. To assist in the modeling of interacting side chains a database of known interactions has been prepared.

Supracrystallographic resolution of interactions contributing to enzyme catalysis by use of natural structural variants and reactivity-probe kinetics

1. The influence on the reactivities of the catalytic sites of papain (EC 3.4.22.2) and actinidin (3.4.22.14) of providing for interactions involving the S1-S2 intersubsite regions of the enzymes was evaluated by using as a series of thiol-specific two-hydronic-state reactivity probes: n-propyl 2-pyridyl disulphide (I) (a 'featureless' probe), 2-(acetamido)ethyl 2'-pyridyl disulphide (II) (containing a P1-P2 amide bond), 2-(acetoxy)ethyl 2'-pyridyl disulphide (III) [the ester analogue of probe (II)] and 2-carboxyethyl 2'-pyridyl disulphide N-methylamide (IV) [the retroamide analogue of probe (II)]. Syntheses of compounds (I), (III) and (IV) are reported. 2. The reactivities of the two enzymes towards the four reactivity probes (I)-(IV) and also that of papain towards 2-(N'-acetyl-L-phenylalanylamino)ethyl 2'-pyridyl disulphide (VII) (containing both a P1-P2 amide bond and an L-phenylalanyl side chain as an occupant for the S2 subsite), in up to four hydronic (previously called protonic) states, were evaluated by analysis of pH-dependent stopped-flow kinetic data (for the release of pyridine-2-thione) by using an eight-parameter rate equation [described in the Appendix: Brocklehurst & Brocklehurst (1988) Biochem. J. 256, 556-558] to provide pH-independent rate constants and macroscopic pKa values. The analysis reveals the various ways in which the two enzymes respond very differently to the binding of ligands in the S1-S2 intersubsite regions despite the virtually superimposable crystal structures in these regions of the molecules. 3. Particularly striking differences between the behaviour of papain and that of actinidin are that (a) only papain responds to the presence of a P1-P2 amide bond in the probe such that a rate maximum at pH 6-7 is produced in the pH-k profile in place of the rate minimum, (b) only in the papain reactions does the pKa value of the alkaline limb of the pH-k profile change from 9.5 to approx. 8.2 [the value characteristic of a pH-(kcat./Km) profile] when the probe contains a P1-P2 amide bond, (c) only papain reactivity is affected by two positively co-operative hydronic dissociations with pKI congruent to pKII congruent to 4 and (d) modulation of the reactivity of the common -S(-)-ImH+ catalytic-site ion-pair (Cys-25/His-159 in papain and Cys-25/His-162 in actinidin) by hydronic dissociation with pKa approx. 5 is more marked and occurs more generally in reactions of actinidin than is the case for papain reactions.(ABSTRACT TRUNCATED AT 400 WORDS)

The use of folding patterns in the search of protein structural similarities; a three-dimensional model of phosphoribosyl transferases

A new way to predict the topologies of proteins of unknown three-dimensional structure is derived from the comparison of the distribution of the strongest predicted secondary structures with equivalent distributions recorded for proteins of known X-ray structures. As an illustration the tentative three-dimensional model of phosphoribosyl transferases which was proposed by Argos et al. is rediscussed.

Magnifying endoscopic observation of the gastric mucosa, particularly in patients with atrophic gastritis

The gastric mucosal surface was observed using the magnifying fibergastroscope (FGS-ML), and the fine gastric mucosal patterns, which were even smaller than one unit of gastric area, were examined at a magnification of about 30. For simplicification, we classified these patterns by magnifying endoscopy in the following ways; FP, FIP, FSP, SP and MP, modifying Yoshii's classification under the dissecting microscope. The FIP, which was found to have round and long elliptical gastric pits, is a new addition to our endoscopic classification. The relationship between the FIP and the intermediate zone was evaluated by superficial and histological studies of surgical and biopsy specimens. The width of the band of FIP seems to be related to the severity of atrophic gastritis. Also, the transformation of FP to FIP was assessed by comparing specimens taken from the resected and residual parts of the stomach, respectively. Moreover, it appears that severe gastritis occurs in the gastric mucosa which shows a FIP. Therefore, we consider that the FIP indicates the position of the atrophic border.