Abstract : The lectin of Helix pomatia (HPA), extracted from the albumin gland of the Roman snail and specific for the residue GalNAc, belongs to a new H type lectin family. It is used for twenty years as marker for metastatic adenocarcinoma (in particular breast, colon, lung) associated with poor life prognostic. Nevertheless, its use as routine tool in oncology is highly limited because of its incapability to produce it in a recombinant form. To avoid these difficulties, homologous proteins were searched in others invertebrates. Two H type lectins have been identified in the amiboe Dictyostelium discoideum (discoidins) and one in the coral Sinularia lochmodes (SLL-2). Discoidins are composed of two distinct domains, a C-terminal domain, specific for galactosylated residues and homologuous to HPA and an N-terminal domain, called discoidin domain, with unknown function. This thesis is focused, in a first time, on the continuation of structural characterization of discoidin 1 and on the production of the N-terminal domain of discoidin 2 to confirm the supposed lectin function. In a second time, confocal microscopy experiments showed that discoidins was not able to discriminate metastatic cancer cells to non metastatic ones, as HPA does. The construction, by mutagenesis, of a chimeric protein between discoidin 2, easily produced in E. coli, and HPA, began. The purpose was to give the same specificity as HPA. Last, SLL-2 was cloned and numerous expression assays, in a soluble form, and purification was tried to characterize the protein biochemistrycally and structurally. The aim was to test it as marker in histopathology.