Production, specificity, and functionality of monoclonal antibodies to specific peptide–major histocompatibility complex class II complexes formed by processing of exogenous protein

Several unanswered questions in T cell immunobiology relating to intracellular processing or in vivo antigen presentation could be approached if convenient, specific, and sensitive reagents were available for detecting the peptide–major histocompatibility complex (MHC) class I or class II ligands recognized by αβ T cell receptors. For this reason, we have developed a method using homogeneously loaded peptide–MHC class II complexes to generate and select specific mAb reactive with these structures using hen egg lysozyme (HEL) and I-Ak as a model system. mAbs specific for either HEL-(46–61)–Ak or HEL-(116–129)–Ak have been isolated. They cross-react with a small subset of I-Ak molecules loaded with self peptides but can nonetheless be used for flow cytometry, immunoprecipitation, Western blotting, and intracellular immunofluorescence to detect specific HEL peptide–MHC class II complexes formed by either peptide exposure or natural processing of native HEL. An example of the utility of these reagents is provided herein by using one of the anti-HEL-(46–61)–Ak specific mAbs to visualize intracellular compartments where I-Ak is loaded with HEL-derived peptides early after antigen administration. Other uses, especially for in vivo tracking of specific ligand-bearing antigen-presenting cells, are discussed.

Peripheral anti-Aβ antibody alters CNS and plasma Aβ clearance and decreases brain Aβ burden in a mouse model of Alzheimer's disease

Active immunization with the amyloid β (Aβ) peptide has been shown to decrease brain Aβ deposition in transgenic mouse models of Alzheimer's disease and certain peripherally administered anti-Aβ antibodies were shown to mimic this effect. In exploring factors that alter Aβ metabolism and clearance, we found that a monoclonal antibody (m266) directed against the central domain of Aβ was able to bind and completely sequester plasma Aβ. Peripheral administration of m266 to PDAPP transgenic mice, in which Aβ is generated specifically within the central nervous system (CNS), results in a rapid 1,000-fold increase in plasma Aβ, due, in part, to a change in Aβ equilibrium between the CNS and plasma. Although peripheral administration of m266 to PDAPP mice markedly reduces Aβ deposition, m266 did not bind to Aβ deposits in the brain. Thus, m266 appears to reduce brain Aβ burden by altering CNS and plasma Aβ clearance.

Anti-peptide aptamers recognize amino acid sequence and bind a protein epitope.

In vitro selection of nucleic acid binding species (aptamers) is superficially similar to the immune response. Both processes produce biopolymers that can recognize targets with high affinity and specificity. While antibodies are known to recognize the sequence and conformation of protein surface features (epitopes), very little is known about the precise interactions between aptamers and their epitopes. Therefore, aptamers that could recognize a particular epitope, a peptide fragment of human immunodeficiency virus type I Rev, were selected from a random sequence RNA pool. Several of the selected RNAs could bind the free peptide more tightly than a natural RNA ligand, the Rev-binding element. In accord with the hypothesis that protein and nucleic acid binding cusps are functionally similar, interactions between aptamers and the peptide target could be disrupted by sequence substitutions. Moreover, the aptamers appeared to be able to bind peptides with different solution conformations, implying an induced fit mechanism for binding. Just as anti-peptide antibodies can sometimes recognize the corresponding epitope when presented in a protein, the anti-peptide aptamers were found to specifically bind to Rev.

Monoclonal antibodies reveal receptor specificity among G-protein-coupled receptor kinases.

Guanine nucleotide-binding regulatory protein (G protein)-coupled receptor kinases (GRKs) constitute a family of serine/threonine kinases that play a major role in the agonist-induced phosphorylation and desensitization of G-protein-coupled receptors. Herein we describe the generation of monoclonal antibodies (mAbs) that specifically react with GRK2 and GRK3 or with GRK4, GRK5, and GRK6. They are used in several different receptor systems to identify the kinases that are responsible for receptor phosphorylation and desensitization. The ability of these reagents to inhibit GRK- mediated receptor phosphorylation is demonstrated in permeabilized 293 cells that overexpress individual GRKs and the type 1A angiotensin II receptor. We also use this approach to identify the endogenous GRKs that are responsible for the agonist-induced phosphorylation of epitope-tagged beta2- adrenergic receptors (beta2ARs) overexpressed in rabbit ventricular myocytes that are infected with a recombinant adenovirus. In these myocytes, anti-GRK2/3 mAbs inhibit isoproterenol-induced receptor phosphorylation by 77%, while GRK4-6-specific mAbs have no effect. Consistent with the operation of a betaAR kinase-mediated mechanism, GRK2 is identified by immunoblot analysis as well as in a functional assay as the predominant GRK expressed in these cells. Microinjection of GRK2/3-specific mAbs into chicken sensory neurons, which have been shown to express a GRK3-like protein, abolishes desensitization of the alpha2AR-mediated calcium current inhibition. The intracellular inhibition of endogenous GRKs by mAbs represents a novel approach to the study of receptor specificities among GRKs that should be widely applicable to many G-protein-coupled receptors.

A quantitative test to estimate neutralizing antibodies to the hepatitis C virus: cytofluorimetric assessment of envelope glycoprotein 2 binding to target cells.

Hepatitis C virus (HCV) is a major cause of chronic hepatitis. The virus does not replicate efficiently in cell cultures, and it is therefore difficult to assess infection-neutralizing antibodies and to evaluate protective immunity in vitro. To study the binding of the HCV envelope to cell-surface receptors, we developed an assay to assess specific binding of recombinant envelope proteins to human cells and neutralization thereof. HCV recombinant envelope proteins expressed in various systems were incubated with human cells, and binding was assessed by flow cytometry using anti-envelope antibodies. Envelope glycoprotein 2 (E2) expressed in mammalian cells, but not in yeast or insect cells, binds human cells with high affinity (Kd approximately 10(-8) M). We then assessed antibodies able to neutralize E2 binding in the sera of both vaccinated and carrier chimpanzees, as well as in the sera of humans infected with various HCV genotypes. Vaccination with recombinant envelope proteins expressed in mammalian cells elicited high titers of neutralizing antibodies that correlated with protection from HCV challenge. HCV infection does not elicit neutralizing antibodies in most chimpanzees and humans, although low titers of neutralizing antibodies were detectable in a minority of infections. The ability to neutralize binding of E2 derived from the HCV-1 genotype was equally distributed among sera from patients infected with HCV genotypes 1, 2, and 3, demonstrating that binding of E2 is partly independent of E2 hypervariable regions. However, a mouse monoclonal antibody raised against the E2 hypervariable region 1 can partially neutralize binding of E2, indicating that at least two neutralizing epitopes, one of which is hypervariable, should exist on the E2 protein. The neutralization-of-binding assay described will be useful to study protective immunity to HCV infection and for vaccine development.

Interactions of protein antigens with antibodies.

There are now several crystal structures of antibody Fab fragments complexed to their protein antigens. These include Fab complexes with lysozyme, two Fab complexes with influenza virus neuraminidase, and three Fab complexes with their anti-idiotype Fabs. The pattern of binding that emerges is similar to that found with other protein-protein interactions, with good shape complementarity between the interacting surfaces and reasonable juxtapositions of polar residues so as to permit hydrogen-bond formation. Water molecules have been observed in cavities within the interface and on the periphery, where they often form bridging hydrogen bonds between antibody and antigen. For the most part the antigen is bound in the middle of the antibody combining site with most of the six complementarity-determining residues involved in binding. For the most studied antigen, lysozyme, the epitopes for four antibodies occupy approximately 45% of the accessible surface area. Some conformational changes have been observed to accompany binding in both the antibody and the antigen, although most of the information on conformational change in the latter comes from studies of complexes with small antigens.

Anti-nuclear antibody production and immune-complex glomerulonephritis in BALB/c mice treated with pristane.

The pathogenesis of systemic lupus erythematosus is thought to be primarily under genetic control, with environmental factors playing a secondary role. However, it has been shown recently that intraperitoneal injection of pristane (2,6,10,14-tetramethylpentadecane) induces autoantibodies typical of lupus in BALB/c mice, a strain not usually considered to be genetically susceptible to the disease. In this study, the induction of autoimmune disease by pristane was investigated. BALB/c mice receiving pristane were tested for autoantibody production and histopathological evidence of glomerulonephritis. Six of 11 mice developed IgM anti-single-stranded DNA antibodies shortly after receiving pristane and 4 developed IgM anti-histone antibodies, but anti-double-stranded DNA antibodies were absent. IgG anti-DNA and anti-histone antibodies were absent. In contrast, the lupus-associated anti-nuclear ribonucleoprotein/Sm and anti-Su autoantibodies produced by these mice were predominantly IgG. In addition to autoantibodies, most of the mice developed significant proteinuria. Light microscopy of the kidney showed segmental or diffuse proliferative glomerulonephritis. Electron microscopy showed subepithelial and mesangial immune-complex deposits and epithelial foot process effacement. Immunofluorescence revealed striking glomerular deposition of IgM, IgG, and C3 with a mesangial or mesangiocapillary distribution. Thus, pristane induces immune-complex glomerulonephritis in association with autoantibodies typical of lupus in BALB/c mice. These data support the idea that lupus is produced by an interplay of genetic and environmental factors and that unlike the MRL or (NZB x W)F1 mouse models, in which genetic susceptibility factors are of primary importance, environmental factors are of considerable importance in the autoimmune disease of pristane-treated BALB/c mice.

Anti-C5 monoclonal antibody therapy prevents collagen-induced arthritis and ameliorates established disease.

Activated components of the complement system are potent mediators of inflammation that may play an important role in numerous disease states. For example, they have been implicated in the pathogenesis of inflammatory joint diseases including rheumatoid arthritis (RA). To target complement activation in immune-mediated joint inflammation, we have utilized monoclonal antibodies (mAbs) that inhibit the complement cascade at C5, blocking the generation of the major chemotactic and proinflammatory factors C5a and C5b-9. In this study, we demonstrate the efficacy of a mAb specific for murine C5 in the treatment of collagen-induced arthritis, an animal model for RA. We show that systemic administration of the anti-C5 mAb effectively inhibits terminal complement activation in vivo and prevents the onset of arthritis in immunized animals. Most important, anti-C5 mAb treatment is also highly effective in ameliorating established disease. These results demonstrate a critical role for activated terminal complement components not only in the induction but also in the progression of collagen-induced arthritis and suggest that C5 may be an attractive therapeutic target in RA.

Restoration of surface IgM-mediated apoptosis in an anti-IgM-resistant variant of WEHI-231 lymphoma cells by HS1, a protein-tyrosine kinase substrate.

The HS1 protein is one of the major substrates of non-receptor-type protein-tyrosine kinases and is phosphorylated immediately after crosslinking of the surface IgM on B cells. The mouse B-lymphoma cell line WEHI-231 is known to undergo apoptosis upon crosslinking of surface IgM by anti-IgM antibodies. Variants of WEHI-231 that were resistant to anti-IgM-induced apoptosis expressed dramatically reduced levels of HS1 protein. Expression of the human HS1 protein from an expression vector introduced into one of the variant cell lines restored the sensitivity of the cells to apoptosis induced by surface IgM crosslinking. These results suggest that HS1 protein plays a crucial role in the B-cell antigen receptor-mediated signal transduction pathway that leads to apoptosis.

Phenotypic knockout of the high-affinity human interleukin 2 receptor by intracellular single-chain antibodies against the alpha subunit of the receptor.

The experimental manipulation of peptide growth hormones and their cellular receptors is central to understanding the pathways governing cellular signaling and growth control. Previous work has shown that intracellular antibodies targeted to the endoplasmic reticulum (ER) can be used to capture specific proteins as they enter the ER, preventing their transport to the cell surface. Here we have used this technology to inhibit the cell surface expression of the alpha subunit of the high-affinity interleukin 2 receptor (IL-2R alpha). A single-chain variable-region fragment of the anti-Tac monoclonal antibody was constructed with a signal peptide and a C-terminal ER retention signal. Intracellular expression of the single-chain antibody was found to completely abrogate cell surface expression of IL-2R alpha in stimulated Jurkat T cells. IL-2R alpha was detectable within the Jurkat cells as an immature 40-kDa form that was sensitive to endoglycosidase H, consistent with its retention in a pre- or early Golgi compartment. A single-chain antibody lacking the ER retention signal was also able to inhibit cell surface expression of IL-2R alpha although the mechanism appeared to involve rapid degradation of the receptor chain within the ER. These intracellular antibodies will provide a valuable tool for examining the role of IL-2R alpha in T-cell activation, IL-2 signal transduction, and the deregulated growth of leukemic cells which overexpress IL-2R alpha.

Suppression and promotion of tumor growth by monoclonal antibodies to ErbB-2 differentially correlate with cellular uptake.

Amplification and overexpression of the erbB-2/neu protooncogene are frequently associated with aggressive clinical course of certain human adenocarcinomas, and therefore the encoded surface glycoprotein is considered a candidate target for immunotherapy. We previously generated a series of anti-ErbB-2 monoclonal antibodies (mAbs) that either accelerate or inhibit the tumorigenic growth of erbB-2-transformed murine fibroblasts. The present study extended this observation to a human tumor cell line grown as xenografts in athymic mice and addressed the biochemical differences between the two classes of mAbs. We show that the inhibitory effect is dominant in an antibody mixture, and it depends on antibody bivalency. By using radiolabeled mAbs we found that all of three tumor-inhibitory mAbs became rapidly inaccessible to acid treatment when incubated with tumor cells. However, a tumor-stimulatory mAb remained accessible to extracellular treatments, indicating that it did not undergo endocytosis. In addition, intracellular fragments of the inhibitory mAbs, but not of the stimulatory mAb, were observed. Electron microscopy of colloidal gold-antibody conjugates confirmed the absence of endocytosis of the stimulatory mAb but detected endocytic vesicles containing an inhibitory mAb. We conclude that acceleration of cell growth by ErbB-2 correlates with cell surface localization, whereas inhibition of tumor growth is associated with an intrinsic ability of anti-ErbB-2 mAbs to induce endocytosis. These conclusions are relevant to the selection of optimal mAbs for immunotherapy and may have implications for the mechanism of cellular transformation by an overexpressed erbB-2 gene.

Influencia del bloqueo del factor de necrosis tumoral-alfa sobre los niveles del factor reumatoide y anticuerpos anti-péptidos cíclicos citrulinados en pacientes con artritis reumatoide

La respuesta clínica a los anti-TNF-α en los pacientes con Artritis Reumatoide (AR) es muy variable, en muchos casos se logra la remisión de la enfermedad, sin embargo una cantidad sustancial de pacientes no responden y persisten con actividad de la enfermedad o presentan recaídas a pesar del tratamiento. Los marcadores clínicos, radiológicos, serológicos y genéticos disponibles para el diagnóstico y seguimiento de la enfermedad tienen un valor limitado a la hora de predecir de manera precisa la respuesta al tratamiento y las recaídas. El factor reumatoide (FR) y los anticuerpos anti-péptidos cíclicos citrulinados (anti-PCC) son los principales marcadores biológicos en la AR y forman parte de los criterios de clasificación de la enfermedad, sin embargo, la relación entre los cambios en la concentración de estos autoanticuerpos y la respuesta a los anti-TNF- es variable en los diferentes estudios y no se aceptan como factores de predicción y de seguimiento de la respuesta a estos fármacos. Las concentraciones séricas del fármaco y los anticuerpos anti-fármaco (ADAb anti-drug antibodies) han sido estudiados como marcadores séricos relacionados con la actividad del enfermedad. Diferentes investigadores han demostrado que la pérdida de eficacia a los anti-TNF-α está asociada con el desarrollo de ADAb, que a su vez se correlaciona con la ausencia de concentraciones séricas adecuadas del fármaco...

Co-encapsulation of enzymes and antibodies for chemical deactivation of pathogens on paper

Le papier bioactif est obtenu par la modification de substrat du papier avec des biomolécules et des réactifs. Ce type de papier est utilisé dans le développement de nouveaux biocapteurs qui sont portables, jetables et économiques visant à capturer, détecter et dans certains cas, désactiver les agents pathogènes. Généralement les papiers bioactifs sont fabriqués par l’incorporation de biomolécules telles que les enzymes et les anticorps sur la surface du papier. L’immobilisation de ces biomolécules sur les surfaces solides est largement utilisée pour différentes applications de diagnostic comme dans immunocapteurs et immunoessais mais en raison de la nature sensible des enzymes, leur intégration au papier à grande échelle a rencontré plusieurs difficultés surtout dans les conditions industrielles. Pendant ce temps, les microcapsules sont une plate-forme intéressante pour l’immobilisation des enzymes et aussi assez efficace pour permettre à la fonctionnalisation du papier à grande échelle car le papier peut être facilement recouvert avec une couche de telles microcapsules. Dans cette étude, nous avons développé une plate-forme générique utilisant des microcapsules à base d’alginate qui peuvent être appliquées aux procédés usuels de production de papier bioactif et antibactérien avec la capacité de capturer des pathogènes à sa surface et de les désactiver grâce à la production d’un réactif anti-pathogène. La conception de cette plate-forme antibactérienne est basée sur la production constante de peroxyde d’hydrogène en tant qu’agent antibactérien à l’intérieur des microcapsules d’alginate. Cette production de peroxyde d’hydrogène est obtenue par oxydation du glucose catalysée par la glucose oxydase encapsulée à l’intérieur des billes d’alginate. Les différentes étapes de cette étude comprennent le piégeage de la glucose oxydase à l’intérieur des microcapsules d’alginate, l’activation et le renforcement de la surface des microcapsules par ajout d’une couche supplémentaire de chitosan, la vérification de la possibilité d’immobilisation des anticorps (immunoglobulines G humaine comme une modèle d’anticorps) sur la surface des microcapsules et enfin, l’évaluation des propriétés antibactériennes de cette plate-forme vis-à-vis l’Escherichia coli K-12 (E. coli K-12) en tant qu’un représentant des agents pathogènes. Après avoir effectué chaque étape, certaines mesures et observations ont été faites en utilisant diverses méthodes et techniques analytiques telles que la méthode de Bradford pour dosage des protéines, l’électroanalyse d’oxygène, la microscopie optique et confocale à balayage laser (CLSM), la spectrométrie de masse avec désorption laser assistée par matrice- temps de vol (MALDI-TOF-MS), etc. Les essais appropriés ont été effectués pour valider la réussite de modification des microcapsules et pour confirmer à ce fait que la glucose oxydase est toujours active après chaque étape de modification. L’activité enzymatique spécifique de la glucose oxydase après l’encapsulation a été évaluée à 120±30 U/g. Aussi, des efforts ont été faits pour immobiliser la glucose oxydase sur des nanoparticules d’or avec deux tailles différentes de diamètre (10,9 nm et 50 nm) afin d’améliorer l’activité enzymatique et augmenter l’efficacité d’encapsulation. Les résultats obtenus lors de cette étude démontrent les modifications réussies sur les microcapsules d’alginate et aussi une réponse favorable de cette plate-forme antibactérienne concernant la désactivation de E. coli K-12. La concentration efficace de l’activité enzymatique afin de désactivation de cet agent pathogénique modèle a été déterminée à 1.3×10-2 U/ml pour une concentration de 6.7×108 cellules/ml de bactéries. D’autres études sont nécessaires pour évaluer l’efficacité de l’anticorps immobilisé dans la désactivation des agents pathogènes et également intégrer la plate-forme sur le papier et valider l’efficacité du système une fois qu’il est déposé sur papier.

Co-encapsulation of enzymes and antibodies for chemical deactivation of pathogens on paper

Le papier bioactif est obtenu par la modification de substrat du papier avec des biomolécules et des réactifs. Ce type de papier est utilisé dans le développement de nouveaux biocapteurs qui sont portables, jetables et économiques visant à capturer, détecter et dans certains cas, désactiver les agents pathogènes. Généralement les papiers bioactifs sont fabriqués par l’incorporation de biomolécules telles que les enzymes et les anticorps sur la surface du papier. L’immobilisation de ces biomolécules sur les surfaces solides est largement utilisée pour différentes applications de diagnostic comme dans immunocapteurs et immunoessais mais en raison de la nature sensible des enzymes, leur intégration au papier à grande échelle a rencontré plusieurs difficultés surtout dans les conditions industrielles. Pendant ce temps, les microcapsules sont une plate-forme intéressante pour l’immobilisation des enzymes et aussi assez efficace pour permettre à la fonctionnalisation du papier à grande échelle car le papier peut être facilement recouvert avec une couche de telles microcapsules. Dans cette étude, nous avons développé une plate-forme générique utilisant des microcapsules à base d’alginate qui peuvent être appliquées aux procédés usuels de production de papier bioactif et antibactérien avec la capacité de capturer des pathogènes à sa surface et de les désactiver grâce à la production d’un réactif anti-pathogène. La conception de cette plate-forme antibactérienne est basée sur la production constante de peroxyde d’hydrogène en tant qu’agent antibactérien à l’intérieur des microcapsules d’alginate. Cette production de peroxyde d’hydrogène est obtenue par oxydation du glucose catalysée par la glucose oxydase encapsulée à l’intérieur des billes d’alginate. Les différentes étapes de cette étude comprennent le piégeage de la glucose oxydase à l’intérieur des microcapsules d’alginate, l’activation et le renforcement de la surface des microcapsules par ajout d’une couche supplémentaire de chitosan, la vérification de la possibilité d’immobilisation des anticorps (immunoglobulines G humaine comme une modèle d’anticorps) sur la surface des microcapsules et enfin, l’évaluation des propriétés antibactériennes de cette plate-forme vis-à-vis l’Escherichia coli K-12 (E. coli K-12) en tant qu’un représentant des agents pathogènes. Après avoir effectué chaque étape, certaines mesures et observations ont été faites en utilisant diverses méthodes et techniques analytiques telles que la méthode de Bradford pour dosage des protéines, l’électroanalyse d’oxygène, la microscopie optique et confocale à balayage laser (CLSM), la spectrométrie de masse avec désorption laser assistée par matrice- temps de vol (MALDI-TOF-MS), etc. Les essais appropriés ont été effectués pour valider la réussite de modification des microcapsules et pour confirmer à ce fait que la glucose oxydase est toujours active après chaque étape de modification. L’activité enzymatique spécifique de la glucose oxydase après l’encapsulation a été évaluée à 120±30 U/g. Aussi, des efforts ont été faits pour immobiliser la glucose oxydase sur des nanoparticules d’or avec deux tailles différentes de diamètre (10,9 nm et 50 nm) afin d’améliorer l’activité enzymatique et augmenter l’efficacité d’encapsulation. Les résultats obtenus lors de cette étude démontrent les modifications réussies sur les microcapsules d’alginate et aussi une réponse favorable de cette plate-forme antibactérienne concernant la désactivation de E. coli K-12. La concentration efficace de l’activité enzymatique afin de désactivation de cet agent pathogénique modèle a été déterminée à 1.3×10-2 U/ml pour une concentration de 6.7×108 cellules/ml de bactéries. D’autres études sont nécessaires pour évaluer l’efficacité de l’anticorps immobilisé dans la désactivation des agents pathogènes et également intégrer la plate-forme sur le papier et valider l’efficacité du système une fois qu’il est déposé sur papier.

Conjugation of an antibody to cross-linked fibrin for targeted delivery of anti-restenotic drugs

There is an urgent need to treat restenosis, a major complication of the treatment of arteries blocked by atherosclerotic plaque, using local delivery techniques. We observed that cross-linked fibrin (XLF) is deposited at the site of surgical injury of arteries. An antibody to XLF, conjugated to anti-restenotic agents, should deliver the drugs directly and only to the site of injury. An anti-XLF antibody (H93.7C.1D2/48; 1D2) was conjugated to heparin (using N-succinimidyl 3-(2-pyridyldithio)-propionate), low molecular weight heparin (LMWH) (adipic acid dihydrazide) and rapamycin (1-ethyl-3-(3-dimethylaminopropyl)carbodiimide/N-hydroxysuccinimide), and the conjugates purified and tested for activity before use in vivo. Rabbits had their right carotid arteries de-endothelialised and then given a bolus of 1D2-heparin, 1D2-LMWH or 1D2-rapamycin conjugate or controls of saline, heparin, LMWH, rapamycin or 1D2 (+/-heparin bolus) and sacrificed after 2 or 4 weeks (12 groups, n=6/group). Rabbits given any of the conjugates had minimal neointimal development in injured arteries, with up to 59% fewer neointimal cells than those given control drugs. Rabbits given 1D2-heparin or 1D2-LMWH had an increased or insignificant reduction in luminal area, with positive remodelling, while the medial and total arterial areas of rabbits given 1D2-rapamycin were not affected by injury. Arteries exposed to 1D2-heparin or 1D2-rapamycin had more endothelial cells than rabbits given control drugs. Thus, XLF-antibodies can site-deliver anti-restenotic agents to injured areas of the artery wall, where the conjugates can influence remodelling, re-endothelialisation and neointimal cell density, with reduced neointimal formation. (C) 2004 Elsevier B.V. All rights reserved.