Enzymes that are involved in the synthesis and processing of oligosaccharides, such as glycosidases, are important catalysts for the specific assembly of oligosaccharide structures on proteins. Design and preparation of selective inhibitors of these enzymes are of high interest as these molecules can be used to modulate cellular functions. Moreover, they may provide potential drugs in new therapeutic strategies. In particular, swainsonine (Swa), a natural inhibitor of Golgi α-mannosidase II, reduces certain tumors and hematological dysfunctions. Nevertheless, some side effects resulted in limitations for the development of this compound in medicinal treatments. Recently, our laboratory has developed a new combinatorial methodology for the disclosure of glycosidase inhibitors based on the dihydroxypyrrolidine scaffold found in Swa. This method allowed the preparation of a new family of selective and competitive a-mannosidase inhibitors with inhibitory activities (Ki values) ranging from 3 µM to 135 nM for the best congeners toward α-mannosidase from jack bean, a reliable model enzyme for mammalian Golgi α-mannosidase II. The aim of the present work was to design and synthesise optimized inhibitors of α-mannosidases II based on a 2-aminomethyl-3,4-dihydroxypyrrolidine scaffold and to assess their potential as new anti cancer leads. In a first time, a structural study was carried out through the analysis of x-ray structures of the best inhibitors previously developed in our laboratory complexed to Golgi α-mannosidase II from Drosophila melanogaster (dGMII), which presents a high sequence identity with the human enzyme. The binding mode of these inhibitors could be thus established. This study combined with docking experiments allowed us to design inhibitors of α-mannosidase II, with improved efficacy. The best congener (3R,4R,5R)-3,4-dihydroxy-5-({[(1R)-2-hydroxy-1-phenylethyl]amino}methyl)-1-methylpyrrolidin-2-one displayed an enhanced inhibitory potency against dGMII (IC50 = 0.5 µM vs 80 µM for the first generation inhibitor). Introduction of a polar substituent on the C(5) position of the pyrrolidine ring accounted for this improved activity as revealed by the x-ray structure of this new derivative complexed to dGMII which displayed additional interactions with active site residues. In a second time, these new α-mannosidase inhibitors have been evaluated as anti cancer agents and their effects were determined on human cancer cells from glioblastoma and melanoma, two tumors associated with a high proliferative and invasive potential, multiple resistance toward conventional chemotherapeutic agents, and poor prognosis. As these compounds were not able to inhibit cellular growth, probably due to a poor cellular uptake, derivatization of our inhibitors into more lipophilic esters, ethers and amides derivatives was envisaged. In particular, introduction of a 4-bromobenzoate moiety resulted in an improvement of cellular uptake, leading to a complete inhibition of cellular growth at 300 µM with inhibitors (2R)-2-({[(2R,3R,4S)-3,4-dihydroxypyrrolidin-2-yl]methyl}amino)-2-phenylethyl 4-bromobenzoate and (2R)-2-({[(2R,3R,4R)-3,4-dihydroxy-1-methyl-5-oxopyrrolidin-2-yl]methyl}amino)-2-phenylethyl 4-bromobenzoate. Under the same conditions, Swa didn't provide any significant growth inhibition. Moreover, exposure of human fibroblasts, as model of healthy cells, to our lead derivatives revealed some selectivity toward tumoral cells. It was hypothesized that cellular esterases were able to hydrolyse the ester moiety and thus release the more hydrophilic parent inhibitors within the tumoral cells. Finally, in order to get more insights into the mode of action of our derivatives, their inhibitory activity on α-mannosidases from human glioblastoma extracts was determined and compared to the activity of Swa. The first generation inhibitor presented a moderate inhibitory potency on α-mannosidases whereas the optimized derivative displayed potent inhibitory activity on human α-mannosidases (IC50 = 50 µM decreased to 0.5 µM). Nevertheless, Swa presented higher potency than our new lead (IC50 = 10-50 nM). It was also determined by gel filtration that the cellular α-mannosidases targeted by our pyrrolidine based inhibitors displayed apparent molecular weight of about 120 kD.