Corepressor SMRT binds the BTB/POZ repressing domain of the LAZ3/BCL6 oncoprotein

MATERIALS AND METHODS

RESULTS

We used the first 181 amino acids of LAZ3/BCL6, encompassing the BTB/POZ domain, as a bait in a yeast two-hybrid screen to isolate cofactors interacting with LAZ3/BCL6 (19). We isolated a cDNA encoding amino acids 194–657 [thereafter referred to as SMRT(cl2)] of the SMRT protein (Fig. 1 A and B).

Next, we set out to identify the region of LAZ3/BCL6 that is sufficient to interact with SMRT(cl2). When fused to GAL4act, a LAZ3/BCL6 derivative lacking the BTB/POZ domain [(GAL4act)LAZ(ΔBTB/POZ)] failed to interact with SMRT(cl2) in the yeast two-hybrid system (Fig. 1C), whereas the isolated BTB/POZ domain [(GALact)LAZ(1–140)] shows strong association (Fig. 1C). We conclude that the BTB/POZ domain is both necessary and sufficient for LAZ3/BCL6 to interact with SMRT(cl2). Interestingly, this region of SMRT includes the N-terminal repressing domains SRD-1 and SRD-2 (Fig. 1A), but not the C-terminal nuclear receptors interaction domains (22–26).

We next attempted to confirm the interaction between the BTB/POZ domain and SMRT in vitro (Fig. 2). In GST pulldown experiments, the BTB/POZ domain is able to retain the in vitro translated and ³⁵S-labeled full-length SMRT (Fig. 2, lane 4) about as efficiently as RARα does (Fig. 2, lane 3). As a control, GST alone is unable to pull down SMRT (Fig. 2 lane 2). We conclude that the LAZ3/BCL6 directly binds the SMRT protein in vitro through its BTB/POZ domain.

We next addressed in mammalian cells the interactions observed in yeast and in vitro between SMRT and the LAZ3/BCL6 BTB/POZ domain. To this end, we performed two-hybrid experiments in mouse C2 cells, using GAL4dbd-LAZ3/BCL6 and VP16 transactivation domain (VP16)-SMRT(cl2) chimeras. As expected, we observed a ≈300-fold increase in the GAL4 responsive reporter activity when a (VP16)SMRT(cl2) chimera is coexpressed with (GAL4dbd)LAZ(BTB/POZ) in these cells (33) (Fig. 3B). In contrast, the cotransfection of (VP16)SMRT(cl2) with either the isolated GAL4dbd or a chimera fusing the GAL4dbd to a LAZ3/BCL6 derivative lacking only the BTB/POZ domain [(GAL4dbd)LAZ(ΔBTB/POZ)] completely failed to activate the same reporter (Fig. 3B). Thus, in complete agreement with our results both in yeast and in vitro, the BTB/POZ domain of LAZ3/BCL6 interacts with SMRT in mammalian cells. We next addressed the possibility for DNA-bound LAZ3/BCL6 to recruit SMRT in mammalian cells (Fig. 3C). Concomitant expression of LAZ3/BCL6 and VP16-SMRT(cl2) leads to a ≈10-fold increase of the B6BStkluc reporter activity (18), which contains a LAZ3/BCL6 target sequence upstream of the minimal tk promoter (Fig. 3C). We conclude that DNA-bound LAZ3/BCL6 is indeed able to recruit SMRT in mammalian cells.

To further substantiate the interaction between SMRT and LAZ3/BCL6 in mammalian cells, we took advantage of the observation that overexpressed LAZ3/BCL6 forms readily detectable nuclear dots (9, 14). By immunofluorescence analyses, we show in Fig. 3D that LAZ3/BCL6 and SMRT completely colocalize in such nuclear dots when simultaneously overexpressed in C2 cells. These results again support the conclusion that LAZ3/BCL6 and SMRT form complexes in mammalian cell nuclei.

We next explored the functional consequences of the physical interaction between LAZ3/BCL6 and SMRT. We and others have shown that the LAZ3/BCL6 protein acts as a transcriptional repressor on its cognate DNA-binding sequence (15–18). SMRT interacts with unliganded RARs and thyroid hormone receptors (TRs) and represses transcription when tethered to DNA, thereby fulfilling the criteria of a corepressor. If SMRT is also a limiting corepressor of LAZ3/BCL6, then its overexpression should potentiate the efficiency of LAZ3/BCL6-mediated repression. Results shown in Fig. 4 fully substantiate this hypothesis. Overexpression of SMRT enhances the silencing effect mediated by LAZ3/BCL6 on its own binding site (Fig. 4). Furthermore, the effect of SMRT on LAZ3/BCL6-mediated repression is dose-dependent (data not shown). Thus, SMRT both interacts with LAZ3/BCL6 and potentiates its repressing activity.

DISCUSSION

The present study defines SMRT as a corepressor of the LAZ3/BCL6 oncoprotein. This conclusion arises from the following data: (i) yeast and mammalian two-hybrid systems show that SMRT interacts with the repressing BTB/POZ domain of LAZ3/BCL6; (ii) GST pulldown assay further demonstrates that SMRT and BTB/POZ directly associate in vitro; (iii) DNA-bound LAZ3/BCL6 is able to recruit SMRT in vivo; (iv) LAZ3/BCL6 and SMRT fully colocalize in nuclear dots; and (v) SMRT potentiates the LAZ3/BCL6-mediated transcriptional repression. In addition, our results point out that SMRT function is not restricted to nuclear receptor-mediated repressing activity. Rather, SMRT is a shared corepressor recruited to DNA by unliganded nuclear receptors, in particular RAR and TR, the BTB/POZ domain/zinc finger transcription factors, and Mad/Max heterodimeric complex (26).

The understanding of the LAZ3/BCL6-mediated transcriptional repression could shed light on the mechanisms underlying its implication in the appearance of diffuse large cell lymphomas. Indeed, LAZ3/BCL6 is a specific regulator of germinal center formation (10, 11) and is structurally altered in most cases of diffuse large cell lymphomas, a germinal center-derived tumor. Importantly, these structural alterations seem to be restricted to the 5′ noncoding region, and hence are thought to alter LAZ3/BCL6 expression while leaving intact its product (4–8). Thus, upon LAZ3/BCL6 structural alteration and deregulation, the ectopic repression of normal LAZ3/BCL6 target genes in late stages of B cell differentiation, especially after the exit of germinal center (9, 11), may contribute to diffuse large-cell lymphoma pathogenesis. In addition, as SMRT is a shared corepressor, misregulated LAZ3/BCL6 exerts an indirect effect by interfering with other SMRT-dependent regulatory pathways. In this regard, it should be noted that a shift from Myc/Max to Mad/Max complexes occur during cellular differentiation, and some RAR ligands appear to play an important role in the control of B cell proliferation (35, 36).

Several Drosophila and vertebrate BTB/POZ domain/zinc fingers proteins can, like LAZ3/BCL6, repress transcription through sequence-specific DNA binding (37–43). Interestingly, the BTB/POZ of promyelocytic leukemia zinc fingers (PLZF), a relative of LAZ3/BCL6 (13, 42), both autonomously represses transcription (43) and interacts with SMRT (R.J.L. and R.M.E., unpublished work), suggesting that SMRT may be a common corepressor for BTB/POZ-containing transcriptional repressors. PLZF originally was cloned because of its fusion to RARα in a subset of acute promyelocytic leukemia (42), and its BTB/POZ domain is chiefly responsible for the dominant negative properties of the PLZF-RARα protein on the ligand-dependent RARα transcriptional activity (44, 45). Together with our data, these findings suggest that the interaction between SMRT and the BTB/POZ domain of PLZF may interfere with the transcriptional activity of the RARα moiety within the PLZF-RARα chimera and thus directly contribute to its oncogenic activity.

It has been postulated that several nuclear BTB/POZ proteins could act by remodeling chromatin structures (13). In this respect, it is worth noting that SMRT recently was shown to form a ternary complex with mSin3A, a corepressor of the Mad and Mxi bHLH-ZIP proteins, and HDAC-1, a histone deacetylase homologous to the yeast rpd-3 gene product (26, 46–51). Interestingly, LAZ3/BCL6 interacts with the region of SMRT that has strong repression activity (Fig. 1A), raising the possibility that it also may associate with mSin3A/HDAC-1 and form a multimeric repression complex involving histone deacetylation activity.