Using linear and non-linear regression to fit biochemical data.

For biochemists or chemists the most common form of data analysis is likely to be regression analysis. This is a technique to find the 'best' values for various experimental parameters; defined as those values which, when used in an appropriate equation, result in the minimum deviation of the calculated results from the experimental data. Despite the widespread application of regression analysis, the basis of the technique and the underlying assumptions are often poorly understood or appreciated. This article describes the basics of linear and non-linear regression, the role of 'weighting' and the potential pitfalls of such analyses.

The Parasol Protocol for computational mutagenesis.

To aid in the discovery and development of peptides and proteins as therapeutic agents, a virtual screen can be used to predict trends and direct workflow. We have developed the Parasol Protocol, a dynamic method implemented using the AMBER MD package, for computational site-directed mutagenesis. This tool can mutate between any pair of amino acids in a computationally expedient, automated manner. To demonstrate the potential of this methodology, we have employed the protocol to investigate a test case involving stapled peptides, and have demonstrated good agreement with experiment.

A study of the specificity of barley chymotrypsin inhibitor 2 by cysteine engineering of the P1 residue.

A combination of oligonucleotide-directed mutagenesis and chemical modification was used to produce reactive site (P1) variants of chymotrypsin inhibitor II (CI2) in an attempt to create more potent inhibitors and examine inhibitory specificity. Mutagenesis to introduce a unique cysteine (CI2M59C) followed by modification to S-carboxamidocysteine with iodoacetamide produced a 7-fold more potent inhibitor of subtilisin BPN' than the wild type inhibitor. Modification with iodoacetic acid, which gives a negatively charged P1 residue (S-carboxymethylcysteine), generates a weaker inhibitor of subtilisin BPN' and chymotrypsin. Further chemical modification experiments of CI2M59C with a series of iodoalkanes of increasing chain lengths was used to determine the optimal P1 side chain length required for potent inhibition of porcine pancreatic elastase. A trend was observed which implies that for CI2 the original methionine residue or its isostere S-ethylcysteine are most effective residues at this position and not alanine as might have been expected from the substrate specificity. The mutant CI2M59C did not inhibit human neutrophil elastase. The iodoalkane modifications not only resulted in recovery of inhibitory activity but also proved to be substantially more potent inhibitors of human neutrophil elastase than wild-type CI2.

The role of the P2' position of Bowman-Birk proteinase inhibitor in the inhibition of trypsin. Studies on P2' variation in cyclic peptides encompassing the reactive site loop.

The role of the P2' residue in proteinase inhibitors of the Bowman-Birk family was investigated using synthetic cyclic peptides based on the reactive site loop of the inhibitor. A series of 21 variants having different P2' residues was tested for inhibition of trypsin, and the rate at which they were hydrolysed by this enzyme was also measured. Variation at P2' was found to result in marked differences in inhibitory potency, with the best sequence (Ile) having a Ki value of 9 nM. Peptides with P2' Gly, Pro or Glu failed to demonstrate any measurable inhibition (Ki>1 mM). The peptides also displayed significant differences in the rates at which they were hydrolysed, which varied by over three orders of magnitude between the difference sequences. There was found to be overall correlation between the Ki value and the rate of hydrolysis, with peptides that inhibited best also being hydrolysed more slowly. The results are discussed in light of the sequence information for Bowman-Birk inhibitor proteins.

Expression and kinetic characterization of barley chymotrypsin inhibitors 1a and 1b.

The genes for chymotrypsin inhibitors 1a and 1b (CI-1a and CI-1b) from barley have been expressed in E. coli, and the CI-1a and CI-1b proteins purified. These proteins, although highly homologous, differ in the active site region at P2, P1' and P3' (Schechter and Berger nomenclature), and so might be expected to have differing specificities. Despite this, analysis of the inhibition kinetics showed that each displayed very similar kinetic behaviour when tested against a range of proteinases. The specificity of the CI-1 proteins is different to that of the other main barley inhibitor, CI-2, and Ki values are found to follow the series subtilisin Carlsberg < neutrophil elastase approximately subtilisin BPN' << chymotrypsin. Only very weak inhibition is found of trypsin, and pancreatic elastase is not measurably inhibited. For the proteinases inhibited most strongly, characteristic slow-binding inhibition kinetics were observed, whereas classical inhibition applied to the weaker interactions. The results are consistent with the major determinant of specificity being the P1 residue of the inhibitor, which is the same in both CI-1a and CI-1b. Consistent with this, is the similar spectrum of specificity found for the homologous inhibitor eglin c from leech, which has the same P1 residue. Both the CI-1 proteins are found to be less stable than CI-2, with CI-1a being significantly less stable than CI-1b as measured by guanidinium hydrochloride unfolding experiments. Possible reasons for the reduced stability are discussed in view of the sequence differences between CI-1a, CI-1b and CI-2.

Analysis of molecular recognition using optical biosensors.

Optical biosensors have made the analysis of molecular interactions rapid and convenient. The field is still young, with commercial instrumentation having been available for less than ten years. For this reason instrument development is still an important factor and the past year has seen continuing advances in instrumentation and particularly in novel sensor surfaces.

Structure of immunoglobulin G by scanning tunnelling microscopy.

Scanning tunnelling microscopy (STM) has been used to examine the shape of individual immunoglobulin G (IgG) molecules deposited onto a graphite surface. IgG was chosen for this study as it has a well-characterized and distinctive three-dimensional structure. The micrographs clearly reveal the IgG molecule as trilobed, corresponding with the known structural organization of IgG. Comparison of these images with the structure of IgG determined by X-ray crystallography shows that the STM images are consistent with the crystal structure. This illustrates that STM is a valuable technique for examining protein structure, allowing rapid determination of the overall molecular shape that is consistent with more established techniques.

The role of threonine in the P2 position of Bowman-Birk proteinase inhibitors: studies on P2 variation in cyclic peptides encompassing the reactive site loop.

Previously, we have described a template-assisted combinatorial peptide library based on the anti-tryptic reactive site loop of a Bowman-Birk inhibitor (BBI). Sequences that displayed inhibitory activity re-directed towards chymotrypsin were found to have a consensus binding motif, with their most striking feature being that exclusively threonine was found at the P2 position. The present study investigates the reason for this surprising specificity by maintaining the binding motif but systematically varying the P2 residue. From analysis of 26 variants, it is found that the requirements for inhibitory activity at P2 are finely tuned, and in agreement with the library work, threonine at P2 provides optimal inhibition. In addition, peptides with threonine at P2 are significantly less susceptible to hydrolysis. Examination of all available BBI sequences shows that threonine is very highly conserved at P2, which implies that the functional requirement extends to the full-length BBI protein. Our results are consistent with a dual requirement for hydrophobic recognition within the S2 pocket and maintenance of an inhibitory conformation via hydrogen bonding within the reactive-site loop. As the isolated peptide loop reproduces the active region of full-length BBI, these results explain why threonine is well conserved at P2 in this class of inhibitor. Furthermore, they illustrate that proteinase inhibitor specificity can have characteristics that are not easily predicted from information on the substrate preferences of a proteinase.

Selection of chymotrypsin inhibitors from a conformationally-constrained combinatorial peptide library.

A synthetic library of cyclic peptides was constructed utilizing the anti-tryptic loop region of the Bowman-Birk inhibitor, D4 from Macrotyloma axillare, as a template. The loop region of this proteinase inhibitor was reproduced by an 11 residue sequence, conformationally constrained by the presence of a disulfide bridge, to act as a mimetic of the functional reactive site region of this protein. This sequence, plus a pentaglycine spacer arm, was used to create a "one bead, one peptide" combinatorial library after on-resin deprotection and cyclization. Randomization at three positions considered to be important for proteinase specificity (P2, P1 and P'2) with the genetically coded amino acids (minus cysteine) plus norleucine generated 8000 permutations. Screening this library with biotinylated alpha-chymotrypsin under appropriate conditions revealed a small number (<0.05%) of beads that selectively bound the labeled proteinase. The sequences present on these active beads were determined, and found to have a well-defined consensus. Analysis of chymotrypsin inhibition in solution using re-synthesized peptides reveals that the sequences identified are potent inhibitors with Ki values in the nanomolar range. These results show that directed randomization of the canonical loop is a powerful way of generating proteinase inhibitors with targeted specificities. Incorporation of selective random changes within a defined structural framework is found to be an effective means of generating variation in large synthetic systems. The functional basis for inhibition by the identified sequences is discussed.

The Bowman-Birk inhibitor reactive site loop sequence represents an independent structural beta-hairpin motif.

We have determined the NMR structure in aqueous solution of a disulphide-cyclised 11-residue peptide that forms a stable beta-hairpin, incorporating a type VIb beta-turn. The structure is found to be extremely well ordered for a short peptide, with the 30 lowest energy simulated annealing structures having an average pairwise r.m.s. deviation of only 0.36 A over the backbone. All but three side-chains adopt distinct conformations, allowing a detailed analysis of their involvement in cross-strand interactions. The peptide sequence analysed originates from a previously reported study, which identified potent inhibitors of human leukocyte elastase from screening a combinatorial peptide library based on the short protein beta-sheet segment that forms the reactive site loop of Bowman-Birk inhibitors. A detailed comparison of the peptide's solution structure with the corresponding region in the whole protein structure reveals a very good correspondence not only for the backbone (r.m.s. deviation approximately 0.7 A) but also for the side-chains. This isolated beta-hairpin retains the biologically active "canonical conformation" typical of small serine proteinase inhibitor proteins, which explains why it retains inhibitory activity. Since the structural integrity is sequence-inherent and does not depend upon the presence of the remaining protein, this beta-hairpin represents an independent structural motif and so provides a useful model of this type of protein architecture and its relation to biological function. The relationship between the conformation of this beta-hairpin and its biological activity is discussed.

Binding of complement subcomponent C1q to mouse IgG1, IgG2a and IgG2b: a novel C1q binding assay.

A C1q binding assay is presented which is suitable for use in comparison of the binding ability of different antibodies, and which allows the quantitative determination of their binding constants. The assay system uses IgG bound to a hapten-derivatized Affigel support. No non-specific binding is observed to a DNP-derivatized support, allowing the use of anti-DNP antibodies. With mouse anti-DNP hybridoma IgGs it was found that C1q binding followed the series IgG2a greater than IgG2b much greater than IgG1, in accordance with the complement fixing ability of these subclasses. Since it is relatively simple to couple any antigen to Affigel , this assay system should be generally applicable to any antibody-antigen system.

Effector functions of a monoclonal aglycosylated mouse IgG2a: binding and activation of complement component C1 and interaction with human monocyte Fc receptor.

Aglycosylated monoclonal anti-DNP mouse IgG2a produced in the presence of tunicamycin was compared with the native monoclonal IgG2a with respect to its ability to interact with the first component of complement, C1, and to compete with human IgG for binding to human monocyte Fc receptors. The aglycosylated IgG2a was found to bind subcomponent C1q with an equivalent capacity to the native IgG2a, but the dissociation constant was found to be increased three-fold. When activation of C1 by the glycosylated and aglycosylated IgG2a was compared, the rate of C1 activation by the aglycosylated IgG2a was reduced approximately three-fold. In contrast aglycosylation was accompanied by a large decrease (greater than or equal to 50-fold) in the apparent binding constant of monomeric IgG2a to human monocytes. The data suggest that the aglycosylated IgG2a has a structure which differs in the CH2 domain from the native IgG2a, and that the heterogeneous N-linked oligosaccharides of this monoclonal IgG2a which occur at a conserved position in the CH2 domain play a role in maintaining the integrity of its monocyte-binding site. This lack of monocyte binding may result either from a localized conformational change occurring in a single CH2 domain or from an alteration in the CH2-CH2 cross-domain architecture which is normally structured by a pair of opposing and interacting oligosaccharides. The minimal changes in C1q binding and C1 activation suggest that the oligosaccharides are, at most, indirectly involved in these events.

The effect of aglycosylation on the binding of mouse IgG to staphylococcal protein A.

Aglycosylated IgG produced by hybridoma cells cultured in the presence of tunicamycin was compared with normal IgG for its ability to bind to staphylococcal protein A. No differences were found in binding or elution profiles. It is concluded that aglycosylation does not produce major structural alterations at the CH2-CH3 interface of the Fc region of IgG.

Synthetic peptide mimics of the Bowman-Birk inhibitor protein.

Proteins of the Bowman-Birk inhibitor family of serine proteinase inhibitors interact with the enzymes they inhibit via an exposed surface loop that adopts the canonical proteinase inhibitory conformation. The resulting non-covalent complex renders the proteinase inactive. This inhibition mechanism is common for the majority of serine proteinase inhibitor proteins and many analogous examples are known. A particular feature of the Bowman-Birk inhibitor protein, however, is that the interacting loop is a particularly well-defined disulfide-linked short beta-sheet region. Moreover, synthetic peptides based on this region keep the same structure as the corresponding part of the full sized protein and also retain inhibitory activity. This review describes the background to inhibition by Bowman-Birk inhibitor proteins (and derived peptides) and shows how peptides based on the reactive site can be manipulated in order to generate potent proteinase inhibitors with redirected specificities.

Analysis of kinetic data of antibody-antigen interaction from an optical biosensor by exponential curve fitting.

An optical biosensor system employing a resonant mirror (RM), with a stirred cuvette has been used to follow the interaction of a recombinant antibody fragment with its antigen, hen egg lysozyme. The data generated by the biosensor were analysed in order to determine the kinetic constants for the interaction using a linear transform (derivative analysis). For comparison the data were also analysed using an exponential curve fitting routine. It was demonstrated that the exponential curve fitting method produced results which were in agreement with the existing linear transform method. It was also shown that early fitting of the association phase response, using the exponential curve fitting routine between 0 and 70 s after sample addition, yielded sufficient information to provide a prediction of Kon. The potential use of the optical biosensor for the rapid monitoring of protein production and purification is discussed.

Imaging of proteins by scanning tunnelling microscopy.

Scanning tunnelling microscopy has been used to examine the structure of proteins deposited on a graphite surface. Three molecules have been studied; immunoglobulin G (IgG), Complement component 1q (C1q) and ATP-citrate lyase (ACL). The images show IgG as a tri-lobed molecule, consistent with the known 3D structure as determined by X-ray crystallography. The C1q images differ from the well known "tulip bunch" model derived by electron microscopy, but are consistent with the model if it is assumed that the six globular heads have aggregated. Molecules of ACL are visible as discrete units, with some hints of substructure. These results highlight the potential of STM in studying protein structures, but also illustrate the difficulties of interpreting micrographs of proteins whose structure is currently unknown.

Potent and specific inhibition of the biological activity of the type-II transmembrane serine protease matriptase by the cyclic microprotein MCoTI-II.

Matriptase is a type-II transmembrane serine protease involved in epithelial homeostasis in both health and disease, and is implicated in the development and progression of a variety of cancers. Matriptase mediates its biological effects both via as yet undefined substrates and pathways, and also by proteolytic cleavage of a variety of well-defined protein substrates, several of which it shares with the closely-related protease hepsin. Development of targeted therapeutic strategies will require discrimination between these proteases. Here we have investigated cyclic microproteins of the squash Momordica cochinchinensis trypsin-inhibitor family (generated by total chemical synthesis) and found MCoTI-II to be a high-affinity (Ki 9 nM) and highly selective (> 1,000-fold) inhibitor of matriptase. MCoTI-II efficiently inhibited the proteolytic activation of pro-hepatocyte growth factor (HGF) by matriptase but not by hepsin, in both purified and cell-based systems, and inhibited HGF-dependent cell scattering. MCoTI-II also selectively inhibited the invasion of matriptase-expressing prostate cancer cells. Using a model of epithelial cell tight junction assembly, we also found that MCoTI-II could effectively inhibit the re-establishment of tight junctions and epithelial barrier function in MDCK-I cells after disruption, consistent with the role of matriptase in regulating epithelial integrity. Surprisingly, MCoTI-II was unable to inhibit matriptase-dependent proteolytic activation of prostasin, a GPI-anchored serine protease also implicated in epithelial homeostasis. These observations suggest that the unusually high selectivity afforded by MCoTI-II and its biological effectiveness might represent a useful starting point for the development of therapeutic inhibitors, and further highlight the role of matriptase in epithelial maintenance.

Structural analysis of foot-and-mouth disease virus 3C protease: a viable target for antiviral drugs?

Foot-and-mouth disease virus causes a major global agricultural problem that is difficult to control with existing vaccines. Structural analyses of the viral 3C protease not only have provided fresh insights into the catalytic mechanism of an unusual class of chymotrypsin-like cysteine proteases, but also are generating valuable information to drive the quest for effective antiviral therapies.

Use of nonlinear regression to analyze enzyme kinetic data: application to situations of substrate contamination and background subtraction.

In a recent publication, A. Lundin, P. Arner, and J. Hellmér [Anal. Biochem. 177, 125-131 (1989)] describe a method whereby kinetic substrate assays can be performed when the assay mixture includes a significant contaminating levels of substrate. Their method requires various rearrangements of the data, and involves three separate linear regression calculations. We show how the same data may be analyzed directly, and far more simply, by nonlinear regression. Unlike the linear regression method, nonlinear regression allows direct calculation of the actual values for Km, Vmax, and the concentration of contaminating substrate (as well as estimates of their standard errors); the former method gives only apparent values. The nonlinear regression technique is also statistically a more valid means of analysis, as the rearrangements required to give linearized equations will considerably distort the error distribution and render simple unweighted linear regression inappropriate. The ease of incorporating extra parameters into standard equations when nonlinear regression is used is further illustrated by fitting enzyme reaction data which describe a first-order process when a significant nonspecific background is present. For this equation no simple rearranged linear plot is possible, but nonlinear regression is easily applied to determine the kinetic parameters.