The identification and functional validation of tumor suppressor gene have been a research hotspot. Here, we have established three groups of tissue-specific mouse models of PTEN and (or) INK4a/ARF gene based on MB1-Cre-LoxP model. Tumor incidence in PTEN and INK4a/ARF double knockout mouse was 100%, including tumor in the pleural cavity, abdomen, and tumor linked with intestine. Flow cytometry analysis demonstrated that cells of lymphoid tissues were mainly Mac-1+, and tumor cells were Mac-1 and B220 double positive. The characteristics of this hematopoietic neoplasm still need to be studied. However, mice deficient for either PTEN or INK4a/ARF did not show the occurrence of similar tumors. B cell development was blocked in the pre-pro-B stage or earlier in either PTEN or compound PTEN & INK4a/ARF deficient mice, while there were no such changes in INK4a/ARF null mice. Thus, PTEN and INK4a/ARF show synergy in constraining tumorigenesis. INK4a/ARF deficiency likely facilitates tumor development in PTEN null mice, although the mechanism of which remains undetermined. Additional aspect of this thesis focused on identifying potential tumor suppressors that are methylated in acute myeloid leukemia. Using the method of COBRA and MSP, we have identified a few potential tumor suppressors such as ALOX12. The methylation status around the ALOX12 CpG-island inversely correlated with the expression levels of mRNA. On the other hand, its expression was restored by demethylation reagent such as 5-aza treatment. Further functional studies are undertaken.