Recent works demonstrated that some transcriptional repressors recruit histone deacetylases (HDACs) either through direct interaction, or as a member of a multisubunit repressing complex containing other components referred to as corepressors. For instance, the bHLH-Zip transcriptional repressors MAD/MXI recruit HDACs together with the mSIN3 corepressors, whereas unliganded nuclear receptors contact another corepressor, SMRT (or its relative N-CoR), which, in turn, associates with both mSIN3 and HDACs to form the repressor complex. Recently, we reported that SMRT also directly associates with LAZ3(BCL-6), a POZ/Zn finger transcriptional repressor involvedin the pathogenesis of non-Hodgkin lymphomas. However, whether LAZ3 recruits the HDACs-containing repression complex is currently unknown. We report here that LAZ3 associates with corepressor mSIN3A both in vivo and in vitro , and found that a central region, which harbours autonomous repression activity, is mainly responsible for this interaction. Conversely, the N-terminal half of mSIN3A is both necessary and sufficient to bind LAZ3. Moreover, we show that LAZ3 also interacts with an HDAC (HDAC-1) through its POZ domain in vitro while the immunoprecipitation of LAZ3 results in the coretention of an endogenous HDAC activity in vivo . Finally, inhibitors of HDACs significantly reduce the LAZ3-mediated repression. Taken together, we conclude that LAZ3 recruits a repressing complex containing SMRT, mSIN3A and a HDAC, and that its full repressing potential on transcription requires HDACs activity. Our results identify HDACs as molecular targets of LAZ3 oncogene and further strengthen the connection between aberrant chromatin acetylation and human cancers.

One of the most frequent genetic abnormalities associated with non Hodgkin lymphoma is the structural alteration of the 5' non coding/regulatory region of the BCL6 (LAZ3) protooncogene. BCL6 encodes a POZ/ Zn ®nger protein, a structure similar to that of many Drosophila developmental regulators and to another protein involved in a human hematopoietic malignancy, PLZF. BCL6 is a sequence speci®c transcriptional repressor controlling germinal center formation and T cell dependent immune response. Although the expression of BCL6 negatively correlates with cellular proliferation in di erent cell types, the in¯uence of BCL6 on cell growth and survival is currently unknown so that the way its deregulation may contribute to cancer remains elusive. To directly address this issue, we used a tetracycline-regulated system in human U2OS osteosarcoma cells and thus found that BCL6 mediates growth suppression associated with impaired S phase progression and apoptosis. Interestingly, overexpressed BCL6 can colocalize with sites of ongoing DNA synthesis, suggesting that it may directly interfere with S phase initiation and/or progression. In contrast, the isolated Zn ®nger region of BCL6, which binds BCL6 target sequence but lacks transcriptional repression activity, slows, but does not suppress, U2OS cell growth, is less e cient at delaying S phase progression, and does not trigger apoptosis. Thus, for a large part, the e ects of BCL6 overexpression on cell growth and survival depend on its ability to engage protein/protein interactions with itself and/or its transcriptional corepressors. That BCL6 restricts cell growth suggests that its deregulation upon structural alterations may alleviate negative controls on the cell cycle and cell survival.

We recently isolated the RhoH/TTF gene by its fusion to the LAZ3/BCL6 gene, in a non-Hodgkin's lymphoma (NHL) cell line, which bore a t(3;4)(q27;p11 ± 13) translocation. This gene encodes a novel Rho GTP-binding protein and is speci®cally expressed in hematopoietic tissues. We made its precise mapping at band 4p13, and described its partial genomic structure. Using¯uorescence in situ hybridization and molecular analyses, we report here on the rearrangement of the RhoH/TTF gene, at band 4p13, in four cases of NHL with t(3;4)(q27;p13) translocation and its fusion to the LAZ3/BCL6 gene at band 3q27, in three of these cases. RT ± PCR analysis of two cases allowed the detection of variable fusion transcripts emerging from the rearranged alleles, and in one case, a deregulated expression of both RhoH/TTF and LAZ3/BCL6 genes, by promoter substitution, was observed. We also show here another rearrangement of the RhoH/TTF gene in a patient with multiple myeloma and t(4;14)(p13;q32) translocation, with breakage within the IGH gene. It is the ®rst report which describes the recurrent chromosomal alteration of a GTP-binding protein encoding gene, in patients with hematopoietic malignancies.

Most acute promyelocytic leukemia (APL) cases are associated with recurrent translocations between the gene of retinoic receptor alpha and that of PML (t(15;17)) or PLZF (t(11;17)). PML localizes onto discrete intranuclear domains, the PML-nuclear bodies, and displays antioncogenic and pro-apoptotic properties. PLZF encodes a transcription factor belonging to the POZ/domain and KruÈ ppel zinc ®nger (POK) family which interacts directly with PML. PLZF is related to another POK protein, LAZ3(BCL6), which is structurally altered, and presumably misexpressed, in many non-Hodgkin lymphoma (NHL) cases. PLZF and LAZ3 share many functional properties: both inhibit cell growth, concentrate into punctated nuclear subdomains and are sequence-speci®c transcriptional repressors recruiting a histone deacetylaserepressing complex. Given these similarities, we tested whether both proteins could be targeted by each other. Here, LAZ3 and PLZF are shown to colocalize onto nuclear dots. Moreover, truncated derivatives of one protein, which display a diuse nuclear localization, are recruited onto nuclear dots by the full-length other. The colocalization and the reciprocal`rescue' is the result of a direct interaction between LAZ3 and PLZF, as indicated by yeast two hybrid assays, in vitro immunoprecipitations, and GST pull down experiments. In contrast to LAZ3 homomerization, LAZ3/PLZF heteromerization in yeast does not solely depend on POZ/POZ contacts but rather also relies on interactions between the two zinc ®nger regions and`cross' contacts between the zinc ®nger region and the POZ domain of each partner. Likewise, LAZ3 shows some colocalization with the PLZF partner PML upon stable overexpression of both proteins in CHO cells and interacts with PML in yeast. Finally, endogenous LAZ3 and PLZF are co-induced and partially colocalized in myeloid MDS cells. These data indicate that a physical interaction between LAZ3 and PLZF underlies their simultaneous recruitment onto multiproteic nuclear complexes, presumably involved in transcriptional silencing and whose integrity (for APL) and/or function (for APL and NHL) may be altered in oncogenesis. Oncogene (2000) 19, 6240 ± 6250.