The novel lectin KM+ detects a specific subset of mannosyl-glycoconjugates in the rat cerebellum

KM+ is a D(+)mannose-specific lectin with a carbohydrate structure-affinity relationship different from those of most mannose-binding lectins. KM+ elicits carbohydrate-dependent biological effects in several mammalian cell types, but it has not yet been employed as a probe for the detection of its specific ligands. We show here for the first time the screening and partial identification of cerebellar mannosyl-glycoconjugates recognized by KM+, by means of lectin-histochemistry and lectin-blotting. Biotinylated KM+ stained most cellular structures in the adult rat cerebellum, particularly Purkinje cells bodies and the surface of granule cells, but not cellular processes. Capillaries in the choroid plexus were also strongly decorated, while blood vessels in the cerebellar parenchyma remained unstained. D(+)mannose, but not D(+)galactose, abolished the staining of all cerebellar structures. Higher inhibitory potencies were found for mannosyl-glycans such as mannotriose (man-α1,3-[man-α1,6]-man) and the biantennary heptasaccharide carried by the enzyme horseradish peroxidase. After separation of cerebellar proteins by SDS-PAGE, KM+ recognized three major unidentified mannosyl-glycoproteins of 132, 83 and 49 kDa. KM+ also detected high-Mw bands corresponding to the light and heavy chains of Type-I laminin, but not a 160-kDa cleavage product of laminin. We conclude that KM+ binds preferentially to a specific subset of mannose-containing glycoproteins in cerebellar tissue, thus being much more restricted than other mannose-specific lectins. KM+ can be used as a novel probe to screen the central nervous system for this specific subset of complex mannosyl-glycoconjugates. Published in 2004.     

Lectin KM+-induced neutrophil haptotaxis involves binding to laminin

The lectin KM+ from Artocarpus integrifolia, also known as artocarpin, induces neutrophil migration by haptotaxis. The interactions of KM+ with both the extracellular matrix (ECM) and neutrophils depend on the lectin ability to recognize mannose-containing glycans. Here, we report the binding of KM+ to laminin and demonstrate that this interaction potentiates the KM+-induced neutrophil migration. Labeling of lung tissue by KM+ located its ligands on the endothelial cells, in the basement membrane, in the alveolus, and in the interstitial connective tissue. Such labeling was inhibited by 400 mM D-mannose, 10 mM Manalpha1-3[Manalpha1-6]Man or 10 microM peroxidase (a glycoprotein-containing mannosyl heptasaccharide). Laminin is a tissue ligand for KM+, since both KM+ and anti-laminin antibodies not only reacted with the same high molecular mass components of a lung extract, but also determined colocalized labeling in basement membranes of the lung tissue. The relevance of the KM+-laminin interaction to the KM+ property of inducing neutrophil migration was evaluated. The inability of low concentrations of soluble KM+ to induce human neutrophil migration was reversed by coating the microchamber filter with laminin. So, the interaction of KM+ with laminin promotes the formation of a substrate-bound KM+ gradient that is able to induce neutrophil haptotaxis.     

Related mannose-specific lectins from different species of the family Amaryllidaceae

Bulbs from three species of the plant family Amaryllidaceae ( Narcissus pseudonurcissus L., Leucojum aestivum L. and Leucojum vernum L.) were found to contain mannose-specific lectins. These lectins were serologically identical to a previously reported Amaryllidaceae lectin from Galanthus nivalis L. bulbs, but had a different molecular structure. The lectins described in this paper are dimeric proteins composed of subunits of 13 kDa, which are not held together by disulphide bridges. In hapten-inhibition assays Amaryllidaceae lectins exhibited exclusive specificity towards mannose. Furthermore, they all had a high specific agglutination activity with trypsin-treated rabbit erythrocytes, whereas human red blood cells were not agglutinated.     

Evidence for a sugar receptor (lectin) in the peritrophic membrane of the blowfly larva, Calliphora erythrocephala Mg. (Diptera)

For the first time a sugar receptor (lectin) has been localized by electron microscopy in an invertebrate. The peritrophic membrane of the blowfly larva, Calliphora erythrocephala, is shown here to express lectins with high specificity for mannose. The lectin is restricted to the lumen side of the peritrophic membrane. The surface of the midgut epithelium is devoid of mannose-specific lectins. It is suggested that the midgut epithelium has lost these lectins during the course of evolution in favour of the peritrophic membrane which is secreted by specialized cells only at the beginning of the midgut.Peritrophic membranes and the midgut epithelium lack lectins specific for galactose. The lumen side of the peritrophic membrane of the larvae has mannose and/or glucose residues, and it is densely packed with two species of bacteria, Proteus vulgaris and P. morganii. These also have mannose-specific lectins as well as mannose residues on their pili. The existence of mannose-specific receptors and mannose residues on both, peritrophic membranes and bacteria, leads to the assumption of mutual adherence between the two surfaces.     

In vitro and in vivo binding of snowdrop (Galanthus nivalis agglutinin; GNA) and jackbean (Canavalia ensiformis; Con A) lectins within tomato moth (Lacanobia oleracea) larvae; mechanisms of insecticidal action

When fed in semi-artificial diet the lectins from snowdrop (Galanthus nivalis: GNA: mannose-specific) and jackbean (Canavalia ensiformis: Con A: specific for glucose and mannose) were shown to accumulate in vivo in the guts, malpighian tubules and haemolymph of Lacanobia oleracea (tomato moth) larvae. Con A, but not GNA, also accumulated in the fat bodies of lectin-fed larvae. The presence of glycoproteins which bind to both lectins in vitro was confirmed using labelled lectins to probe blots of polypeptides extracted from larval tissues. Immunolocalisation studies revealed a similar pattern of GNA and Con A binding along the digestive tract with binding concentrated in midgut sections. Binding of lectins to microvilli appeared to lead to transport of the proteins into cells of the gut and malpighian tubules. These results suggested that both lectins are able to exert systemic effects via transport from the gut contents to the haemolymph across the gut epithelium. The delivery of GNA and Con A to the haemolymph was shown to be dependent on their functional integrity by feeding larvae diets containing denatured lectins. Con A, but not GNA, was shown to persist in gut and fat body tissue of lectin-fed larvae chased with control diet for three days. Con A also shows more extensive binding to larval tissues in vitro than GNA, and these two factors are suggested to contribute to the higher levels of toxicity shown by Con A, relative to GNA, in previous long term bioassays.     

Carbohydrate and glycoprotein specificity of two endogenous cerebellar lectins

Two endogenous cerebellar mannose binding lectins have been isolated in an active form by immunoaffinity chromatography employing their respective immobilized antibodies. One of them, termed cerebellar soluble lectin (CSL), was extracted in the absence of detergents, whereas the other, called Receptor 1 (R1), was soluble only in the presence of detergents. Tests of inhibition of agglutination of erythrocytes were performed with mono-, oligo and polysaccharides, as well as glycoconjugates of known structures. On the basis of agglutinating activities these 2 lectins are different from the previously reported lectins in brain, since they were not inhibited by galactosides and lactosides and were only marginally inhibited by glycosaminoglycans. CSL and R1 were better inhibited by mannose-rich glycopeptides as compared to the corresponding oligosaccharides. The different inhibition patterns obtained with glycans of known structures indicated that these lectins are very discriminative. Although CSL and R1 have similar specificities, they differed in their binding properties towards glycopeptides of ovalbumin. Both lectins showed considerable affinity for endogenous cerebellar glycopeptides, also rich in mannose. These glycopeptides belong to a few endogenous Con A-binding cerebellar glycoprotein subunits and are not present on other endogenous Con A-binding glycoproteins. In the forebrain, where CSL and R1 were also present, at least some of the glycoproteins interacting with the lectins were different from that observed in the cerebellum. Our data overall suggest that specific cell recognition in the nervous system could be invoked via the interactions between widely distributed lectins and cell-specific glycoproteins.     

Specificity of a target cell-derived stop signal for afferent axonal growth.

With a novel model culture system in which afferents are co-cultured with purified populations of target neurons, we have demonstrated that a target cell within the central nervous system (CNS), the cerebellar granule neuron, poses a "stop-growing signal" for its appropriate afferents, the mossy fibers. To ask whether this stop signal is afferent specific, we co-cultured granule neurons with another cerebellar afferent system, the climbing fibers from the inferior olivary nuclei, which normally contact Purkinje neurons, and with retinal ganglion cell afferents, which never enter the cerebellum. Granule neurons do not pose a stop signal to either of these afferents. In contrast to pontine mossy afferents that grow well on laminin and showed reduced outgrowth on granule neurons, both olivary and retinal fibers displayed similar growth on laminin alone or on granule neurons. In addition, each afferent showed different degrees of fasciculation and growth cone morphology on laminin. Thus, the growth arrest signal sent by granule neurons is specifically recognized by their appropriate afferents. Moreover, these three types of afferents exhibit varying growth patterns on the same noncellular and cellular substrates, implicating distinct molecular characteristics of growth regulation for different classes of neurons that would contribute to specificity of synapse formation.     

Uptake by macrophages of a biotinylated oligo-alpha-deoxythymidylate by using mannosylated streptavidin.

Streptavidin substituted with mannose residues increased by 20-fold the intracellular concentration of a biotinylated dodecakis(alpha-deoxythymidylate) in macrophages by comparison with the uptake of free oligodeoxynucleotide. Streptavidin, the bacterial homologue of the very basic avidin, which does not contain any carbohydrate moieties and is a neutral protein, was substituted with 12 mannose residues in order to be recognized and internalized by mannose-specific lectins on the surface of macrophages. A 3'-biotinylated and 5'-fluoresceinylated dodecakis (alpha-deoxythymidylate) was synthesized and bound onto mannosylated streptavidin. The conjugate was isolated, and by using flow cytometry, it was shown that the uptake of fluoresceinylated oligodeoxynucleotides bound to mannosylated streptavidin by macrophages is 20-fold higher than that of free oligodeoxynucleotides and that the uptake was competively inhibited by mannosylated serum albumin. Glycosylated streptavidin conjugates recognizing specific membrane lectins on different cells provide the possibility to target biotinylated antisense oligodeoxynucleotides and to increase the biological effect of these chemotherapeutic agents.     

Profile of resistance of human immunodeficiency virus to mannose-specific plant lectins

The mannose-specific plant lectins from the Amaryllidaceae family (e.g., Hippeastrum sp. hybrid and Galanthus nivalis) inhibit human immunodeficiency virus (HIV) infection of human lymphocytic cells in the higher nanogram per milliliter range and suppress syncytium formation between persistently HIV type 1 (HIV-1)-infected cells and uninfected CD4(+) T cells. These lectins inhibit virus entry. When exposed to escalating concentrations of G. nivalis and Hippeastrum sp. hybrid agglutinin, a variety of HIV-1(III(B)) strains were isolated after 20 to 40 subcultivations which showed a decreased sensitivity to the plant lectins. Several amino acid changes in the envelope glycoprotein gp120, but not in gp41, of the mutant virus isolates were observed. The vast majority of the amino acid changes occurred at the N glycosylation sites and at the S or T residues that are part of the N glycosylation motif. The degree of resistance to the plant lectins was invariably correlated with an increasing number of mutated glycosylation sites in gp120. The nature of these mutations was entirely different from that of mutations that are known to appear in HIV-1 gp120 under the pressure of other viral entry inhibitors such as dextran sulfate, bicyclams (i.e., AMD3100), and chicoric acid, which also explains the lack of cross-resistance of plant lectin-resistant viruses to any other HIV inhibitor including T-20 and the blue-green algae (cyanobacteria)-derived mannose-specific cyanovirin. The plant lectins represent a well-defined class of anti-HIV (microbicidal) drugs with a novel HIV drug resistance profile different from those of other existing anti-HIV drugs.     

Preparation of a monolithic capillary column with immobilized alpha-mannose for affinity chromatography of lectins

A simple method for the preparation of an affinity monolithic (also called continuous bed) capillary column for alpha-mannose-specific lectins is described. 2-Hydroxyethyl methacrylate in combination with (+)-N,N -diallyltartardiamide (DATD) and piperazine diacrylamide (PDA, 1,4-bisacryloyl-piperazine) as crosslinkers, were used as monomers for the monolith. After oxidation of DATD with periodate, alpha-mannose with spacer was bound to the aldehyde groups of the polymeric skeleton via reductive amination to form an affinity column for the separation, enrichment or binding studies of mannose-specific lectins. The permeability of the column was excellent. The porosity of the monolith was investigated by scanning electron microscope (SEM) and inverse size exclusion chromatography (ISEC). The affinity of the monolith was evaluated by frontal analysis (FA) and fluorescence microscopy (FM) using fluorescently labeled concanavalin (Con A). Frontal affinity chromatography showed a specific interaction of two different lectins with the alpha-mannose-modified monolith. According to FM the affinity sites were evenly distributed over the monolithic bed.     

Laminin binding and internalization by human and murine mammary gland cell lines in vitro.

We have studied the binding and internalization of Engelbreth-Holm-Swarm mouse sarcoma laminin labeled with colloidal gold (LN-G40) by human and murine mammary gland cell lines. Interactions between the LN-G40 probe and the cells spread on a glass coverslip were monitored with video-enhanced contrast microscopy (Nanovid). Transmission electron microscopy allowed the quantitation of the LN-G40 probe at various cellular locations. During the first 15 min, a homogeneous binding of LN-G40 probe to the cell surface was observed with all cell lines. This binding did not occur with gold particles that were not conjugated to laminin. Then, the LN-G40 probe began to cluster on the cell surface and was, during the following 20 h, internalized by pits that were not coated. In the cells, the LN-G40 probe sometimes showed saltatory movements along linear tracks. The LN-G40 probe was intracellularly found in vesicles, multivesicular bodies, cisternal structures, and lysosomes, suggesting the degradation of the internalized laminin. However, not all cell surface-bound LN-G40 probe was internalized after 20 h. Differences between the cell lines were quantitative, but no clear correlation could be made between migration of cells on laminin and internalization of laminin.     

Neutrophil haptotaxis induced by the lectin KM+

KM+ is a D(+)mannose binding lectin from Artocarpus integrifolia that induces neutrophil migration in vitro and in vivo.This attractant activity was shown to be caused by haptotaxis rather than chemotaxis. The inhibition by D(+)mannose of the neutrophil attraction exerted by KM+, both in vitro and in vivo, supports the idea that haptotaxis is triggered in vivo by the sugar binding sites interacting with glycoconjugates located on the neutrophil surface and in the extracellular matrix. In the present study an in vivo haptotaxis assay was performed by intradermally (i.d.) injecting 125I-KM+ (200 ng), which led to a selective staining of loose connective tissue and vascular endothelium. The radiolabelled area exhibited a maximum increase (five-fold) in neutrophil infiltration 3 h after injection, relative to i.d. 200 ng 125I-BSA. We characterized the ex vivo binding of KM+ to tissue elements by immunohistochemistry, using paraformaldehyde-fixed, paraffin-embedded, untreated rat skin. Bound KM+ was detected with an affinity-purified rabbit IgG anti-KM+ and visualized with an alkaline phosphatase based system. KM+ binding to connective tissue and vascular endothelium was inhibited by preincubating KM+ with 0.4 m MD(+)mannose and was potentiated by heparan sulfate (100 g ml–1). An in vitro assay carried out in a Boyden microchamber showed that heparan sulfate potentiated the attractant effect of 10 g KM+ by 34%. The present data suggest that KM+ induces neutrophil migration in vivo by haptotaxis and that the haptotactic gradient could be provided by the interaction of the KM+ carbohydrate recognition site(s) with mannose-containing glycoconjugate(s) in vascular endothelium and connective tissue. Heparan sulfate would act as an accessory molecule, enhancing the KM+ tissue binding and potentiating the induced neutrophil haptotaxis.     

Preparation of a monolithic capillary column with immobilized α-mannose for affinity chromatography of lectins

A simple method for the preparation of an affinity monolithic (also called continuous bed) capillary column for α-mannose-specific lectins is described. 2-Hydroxyethyl methacrylate in combination with (+)-N,N´-diallyltartardiamide (DATD) and piperazine diacrylamide (PDA, 1,4-bisacryloyl-piperazine) as crosslinkers, were used as monomers for the monolith. After oxidation of DATD with periodate, α-mannose with spacer was bound to the aldehyde groups of the polymeric skeleton via reductive amination to form an affinity column for the separation, enrichment or binding studies of mannose-specific lectins. The permeability of the column was excellent. The porosity of the monolith was investigated by scanning electron microscope (SEM) and inverse size exclusion chromatography (ISEC). The affinity of the monolith was evaluated by frontal analysis (FA) and fluorescence microscopy (FM) using fluorescently labeled concanavalin (Con A). Frontal affinity chromatography showed a specific interaction of two different lectins with the α-mannose-modified monolith. According to FM the affinity sites were evenly distributed over the monolithic bed.     

A Lectin from Leaves of Neoregelia flandria Recognizes D-Glucose, D-Mannose and N-Acetyl D-glucosamine, Differing from the Mannose-Specific Lectins of Other Monocotyledonous Plants

A mannose-specific lectin was isolated from leaves of Neoregeliaflandria, an ornamental plant that belongs to Bromeliaceae,a family of monocotyledons. The amino acid composition and molecularmass of the lectin were similar to those of mannose-specificlectins from other monocotyledons. However, in a test to examinethe inhibition of hemagglutination, it became apparent thatthe isolated lectin recognized D-glucose and N-acetyl D-glucosaminein addition to D-mannose, unlike mannose-specific lectins fromthe monocotyledons that have been reported to date. (Received May 17, 1996; Accepted August 19, 1996)     

The galactose-binding and mannose-binding jacalin-related lectins are located in different sub-cellular compartments

A galactose-specific and a mannose-specific lectin of the family of the jacalin-related lectins have been localized by immunofluorescence microscopy. The present localization studies provide for the first time unambiguous evidence for the cytoplasmic location of the mannose-specific jacalin-related lectin from rhizomes of Calystegia sepium, which definitely differs from the vacuolar location of the galactose-specific jacalin from Artocarpus integrifolia. These observations support the hypothesis that the galactose-specific jacalin-related lectins evolved from their mannose-specific homologues through the acquisition of vacuolar targeting sequences.     

A biological role of the carbohydrate moieties of laminin

The ways in which the carbohydrate moieties of laminin affect its cellular interactions have been examined by two different experimental approaches. In one approach, we used lectins in order to block specific carbohydrates on laminin which previously had been dried onto a plastic surface. We found that wheat germ agglutinin and Griffonia simplicifolia agglutinin I blocked the binding of the neuron-like rat pheochromocytoma cell line PC12. However, when concanavalin A was used cell binding was unaffected but neurite outgrowth was prevented, compared to controls, over a 24-h period. In the second approach we used unglycosylated laminin as a substratum on the plastic surface. We have developed a method for the purification of unglycosylated laminin from tunicamycin treated cultures of a mouse embryonal carcinoma derived cell line, M1536 B3, and have partially characterized the purified material. A mixture of unglycosylated and glycosylated laminin was selectively purified from the M1536 B3 cell lysate by an anti-EHS laminin monoclonal antibody immunoaffinity column. The unglycosylated laminin was separated from glycosylated laminin using G. simplicifolia lectin affinity chromatography. The lectins, wheat germ agglutinin, G. simplicifolia agglutinin I, and concanavalin A, did not bind to any of the subunits of unglycosylated laminin in Western blots. The unglycosylated laminin migrated as a single band in agarose-gel electrophoresis under nonreducing conditions indicating that it is a fully assembled and disulfide bonded molecule. Circular dichroism studies showed no differences between glycosylated and unglycosylated laminin, indicating similar molecular conformations. Western blots using antibodies specific for the A, B1, and B2 chains of laminin showed that unglycosylated laminin contained each of these subunits. We then performed cell binding and spreading or neurite outgrowth assays using unglycosylated laminin. A mouse melanoma cell line, B16 F1, bound to this laminin in the same numbers as to the control glycosylated laminin, but cell spreading was minimal. When this unglycosylated laminin was used as a substrate for PC12 cells neurite outgrowth was impaired; no effect was noted on the number of cells bound, compared to glycosylated laminin. We conclude from these results that once cells become bound to laminin the carbohydrate residues of that glycoprotein must be available to enable the cells to spread or to extend neurite processes.     

Structure of porphobilinogen synthase from Pseudomonas aeruginosa in complex with 5-fluorolevulinic acid suggests a double Schiff base mechanism

The seeds of jack fruit (Artocarpus integrifolia) contain two tetrameric lectins, jacalin and artocarpin. Jacalin was the first lectin found to exhibit the beta-prism I fold, which is characteristic of the Moraceae plant lectin family. Jacalin contains two polypeptide chains produced by a post-translational proteolysis which has been shown to be crucial for generating its specificity for galactose. Artocarpin is a single chain protein with considerable sequence similarity with jacalin. It, however, exhibits many properties different from those of jacalin. In particular, it is specific to mannose. The structures of two crystal forms, form I and form II, of the native lectin have been determined at 2.4 and 2.5 A resolution, respectively. The structure of the lectin complexed with methyl-alpha-mannose, has also been determined at 2.9 A resolution. The structure is similar to jacalin, although differences exist in details. The crystal structures and detailed modelling studies indicate that the following differences between the carbohydrate binding sites of artocarpin and jacalin are responsible for the difference in the specificities of the two lectins. Firstly, artocarpin does not contain, unlike jacalin, an N terminus generated by post-translational proteolysis. Secondly, there is no aromatic residue in the binding site of artocarpin whereas there are four in that of jacalin. A comparison with similar lectins of known structures or sequences, suggests that, in general, stacking interactions with aromatic residues are important for the binding of galactose while such interactions are usually absent in the carbohydrate binding sites of mannose-specific lectins with the beta-prism I fold.     

Mannose-binding plant lectins: different structural scaffolds for a common sugar-recognition process

Mannose-specific lectins are widely distributed in higher plants and are believed to play a role in recognition of high-mannose type glycans of foreign micro-organisms or plant predators. Structural studies have demonstrated that the mannose-binding specificity of lectins is mediated by distinct structural scaffolds. The mannose/glucose-specific legume (e.g., Con A, pea lectin) exhibit the canonical twelve-stranded beta-sandwich structure. In contrast to legume lectins that interact with both mannose and glucose, the monocot mannose-binding lectins (e.g., the Galanthus nivalis agglutinin or GNA from bulbs) react exclusively with mannose and mannose-containing N-glycans. These lectins possess a beta-prism structure. More recently, an increasing number of mannose-specific lectins structurally related to jacalin (e.g., the lectins from the Jerusalem artichoke, banana or rice), which also exhibit a beta-prism organization, were characterized. Jacalin itself was re-defined as a polyspecific lectin which, in addition to galactose, also interacts with mannose and mannose-containing glycans. Finally the B-chain of the type II RIP of iris, which has the same beta-prism structure as all other members of the ricin-B family, interacts specifically with mannose and galactose. This structural diversity associated with the specific recognition of high-mannose type glycans highlights the importance of mannose-specific lectins as recognition molecules in higher plants.     

Accessibility of the high-mannose glycans of glycoprotein gp120 from human immunodeficiency virus type 1 probed by in vitro interaction with mannose-binding lectins

The direct interaction of mannose-specific plant lectins with gp120 of HIV-1 was studied by surface plasmon resonance. Inhibition experiments indicated that exposed high mannose type glycans play a key role in the interaction. Most of the lectins specifically accommodate outer alpha1,2-, alpha1,3-, or alpha1,6-linked di- or trimannosides, and especially legume lectins, also interact with the trimannoside core of the complex type glycans. The unexpected affinity of some lectins towards gp120 presumably results from conformational differences in their binding sites. These results demonstrate that mannose-specific plant lectins are powerful tools to study the accessibility and elucidate the function of the gp120 glycans in the recognition and infection of the host cells by HIV-1.     

Specificity of a target cell-derived stop signal for afferent axonal growth

With a novel model culture system in which afferents are co-cultured with purified populations of target neurons, we have demonstrated that a target cell within the central nervous system (CNS), the cerebellar granule neuron, poses a ?stop-growing signal”? for its appropriate afferents, the mossy fibers. To ask whether this stop signal is afferent specific, we co-cultured granule neurons with an other cerebellar afferent system, the climbing fibers from the inferior olivary nuclei, which normally contact Purkinje neurons, and with retinal ganglion cell afferents, which never enter the cerebellum. Granule neurons do not pose a stop signal to either of these afferents. In contrast to pontine mossy afferents that grow well on laminin and showed reduced outgrowth on granule neurons, both olivary and retinal fibers displayed similar growth on laminin alone or on granule neurons. In addition, each afferent showed different degrees of fasciculation and growth cone morphology on laminin. Thus, the growth arrest signal sent by granule neurons is specifically recognized by their appropriate afferents. Moreover, these three types of afferents exhibit varying growth patterns on the same noncellular and cellular substrates, implicating distinct molecular characteristics of growth regulation for different classes of neurons that would contribute to specificity of synapse formation. © 1992 John Wiley & Sons, Inc.