In today's society, cancer is a group of diseases that represents one of the major health problems. Physicians often face cancers that are resistant to conventional treatments. In addition, these treatments are often toxic for patients and at the origin of significant side effects. This is why researchers constantly search for new therapies and new drugs that can enhance the efficiency of current chemotherapy. A large number of interesting molecules for the development of new treatments come from the plant world. Pectin is one of these molecules of interest. It is a polysaccharide found in all higher plant cell wall. Once modified by pH treatment (MCP) or by heat treatment, pectin can induce tumor cell death. However, the type of cell death, the mechanisms involved and the nature of the active fragments are not well known. In this thesis, we aimed at characterizing the type of cell death induced by pectin having undergone a heat treatment. To do so, we used two tumor cell lines, HepG2 and A549 cells. We also tried to identify fragments that yield the cytotoxicity. We have shown that citrus pectin modified by heat treatment (HFCP) induces death of HepG2 and A549 cells. This cell death is different from classical apoptosis since we did not observe caspase-3 activity in HepG2 cells and we did not observe DNA cleavage in both cell types. Inhibition of caspases by Z-VAD-fmk did not affect HFCP-induced cell death in HepG2 cells, indicating that heat modified citrus pectin could induce caspase-independent cell death. We also showed that heat fragmented citrus pectin induced ubiquitination of intracellular proteins in HepG2 cells and an early induction of autophagy. The study of putative pectic fragment receptors did not allow us to identify the receptor capable of binding the active fragments. However, it allowed us to demonstrate that MCP fragments and HFCP have a different nature since the invalidation of galectin-3, which is able to recognize the MCP fragments, does not influence the death induced by HFCP. Analysis of active fragments allowed us to highlight that the cytotoxic activity induced by heat-treated pectin comes from small sized fragments having undergone a chemical modification of their structure which is probably due to Maillard reactions. We have identified at least one of the active compounds of the HFCP, it is called 4,5-dihydroxy-cyclopenten-1-one. This compound has been synthesized in the laboratory and actually displayed cytotoxic activity for cancer cells. To our knowledge, it is the first time that a change of structure is demonstrated during a heat treatment of pectin. This is also the first time that one of the active fragments is identified. Moreover, induction of autophagy and of ubiquitination after cell incubation with heat fragmented pectin was never shown before. This thesis confirms the great potential of modified pectin to improve cancer treatments.