The transforming proteins of acute promyelocytic leukaemias (APL) are fusions of the promyelocytic leukaemia (PML) and the promyelocytic leukaemia zinc-finger (PLZF) proteins with retinoic acid receptor-alpha (RARalpha). These proteins retain the RARalpha DNA- and retinoic acid (RA)-binding domains, and their ability to block haematopoietic differentiation depends on the RARalpha DNA-binding domain. Thus RA-target genes are downstream effectors. However, treatment with RA induces differentiation of leukaemic blast cells and disease remission in PML-RARalpha APLs, whereas PLZF-RARa APLs are resistant to RA. Transcriptional regulation by RARs involves modifications of chromatin by histone deacetylases, which are recruited to RA-target genes by nuclear co-repressors. Here we show that both PML-RARalpha and PLZF-RARalpha fusion proteins recruit the nuclear co-repressor (N-CoR)-histone deacetylase complex through the RARalpha CoR box. PLZF-RARalpha contains a second, RA-resistant binding site in the PLZF amino-terminal region. High doses of RA release histone deacetylase activity from PML-RARalpha, but not from PLZF-RARalpha. Mutation of the N-CoR binding site abolishes the ability of PML-RARalpha to block differentiation, whereas inhibition of histone deacetylase activity switches the transcriptional and biological effects of PLZF-RARalpha from being an inhibitor to an activator of the RA signalling pathway. Therefore, recruitment of histone deacetylase is crucial to the transforming potential of APL fusion proteins, and the different effects of RA on the stability of the PML-RARalpha and PLZF-RARalpha co-repressor complexes determines the differential response of APLs to RA.

We have defined two principles of corepressor function that account for differences in transcriptional repression by nuclear hormone receptors (NHRs). First, we have determined that receptor stoichiometry is a crucial determinant of transcriptional repression mediated by the corepressors N-CoR and SMRT. This provides a molecular explanation for the observation that NHRs repress transcription as dimers but not monomers. Second, corepressor function is restricted by steric effects related to DNA binding in a receptor-specific manner. Thus, although N-CoR and SMRT are capable of binding to several NHRs in solution, they are highly selective about receptor binding on DNA, a context that reflects their in vivo function more accurately. These stoichiometric and steric principles govern specific interactions between corepressors and NHRs, thus providing evidence that N-CoR and SMRT do not serve redundant functions but rather contribute to receptor-specific transcriptional repression.[Key Words: Corepressor function; transcriptional repression,-NHRs; stoichiometry; steric effects; nuclear hormone receptor; orphan receptor; PPAR; thyroid hormone receptor] Received November 4, 1996; revised version accepted March 4, 1997.Nuclear hormone receptors (NHRs) regulate cellular growth and differentiation and organ development by modulating gene transcription. In addition to ligand-dependent gene activation, selected NHRs including thy roid hormone receptor (TR) and retinoic acid receptor (RAR) repress basal transcription in the absence of ligand (Brent et al. 1989;Graupner et al. 1989;Baniahmad et al. 1992;Fondell et al. 1993;Casanova et al. 1994). The overall level of transcription of a specific gene is deter mined by the integration of positive and negative effects exerted by transcription factors on the basal transcrip tion apparatus. In this way, transcription of a gene may depend on the net influence of multiple ligands and di verse signal transduction pathways that act both directly and via intervening proteins termed coactivators or co repressors.Nuclear receptor corepressor (N-CoR) and SMRT (silencing mediator for retinoid and thyroid hormone receptor) (Chen and Evans 1995) are considered corepressor proteins because they interact with unliganded NHRs and function as adaptors to con vey a repressive signal to the transcription apparatus. Ligand binding to the NHR leads to a conformational change in the receptor that results in dissociation of the corepressor. These events permit the NHR to bind dis-*Corresponding author. E-MAIL lazar@mail.med.upenn.edu; FAX (215) 898-5408. tinct adaptor proteins known as coactivators, which are involved in transactivation, such as SRC-1 (Onate et al.

Ligand-independent transcriptional repression is an important function of nuclear hormone receptors. An interaction screen with the repression domain of the orphan receptor RevErb identified N-CoR, the corepressor for thyroid hormone receptor (TR) and retinoic acid receptor (RAR). N-CoR is likely to be a bona fide transcriptional corepressor for RevErb because (i) RevErb interacts with endogenous N-CoR, (ii) ectopic N-CoR potentiates RevErb-mediated repression, and (iii) transcriptional repression by RevErb correlates with its ability to bind N-CoR. Remarkably, a region homologous to the CoR box which is necessary for TR and RAR to interact with N-CoR is not required for RevErb. Rather, two short regions of RevErb separated by ϳ200 amino acids are required for interaction with N-CoR. The primary amino acid sequence of the N-terminal region of RevErb essential for N-CoR interaction is not homologous to that of TR or RAR, whereas similarities exist among the C-terminal domains of the receptors. N-CoR contains two adjacent but distinct interaction domains, one of which binds tightly to both RevErb and TR whereas the other binds more weakly and differentially interacts with the nuclear receptors. These results indicate that multiple nuclear receptors, utilizing different primary amino acid sequences, repress transcription by interacting with N-CoR.Regulation of gene expression is essential for cellular differentiation, development, and maintenance of homeostasis. These processes are regulated at the transcriptional level by sequence-specific transcriptional activators and repressors which communicate with the basal transcription apparatus (for reviews, see references 51 and 59). Many activation domains have been shown to interact directly with components of the basal transcription apparatus, while others interact indirectly by contacting intermediate proteins termed adapters or coactivators. Activation domains have been classified by the amino acid residues that predominate: acidic (29, 38), glutamine rich (12), or proline-rich (41). It has been hypothesized that activation domains with diverse structures may recruit different components of the basal apparatus or different classes of adapter molecules, thereby effecting transcription through distinct mechanisms. Indeed, the acidic activation domain of VP16 interacts with TAF II 40 to mediate activated transcription (23), while the Gln-rich activation domain of Sp1 interacts with TAF II 110 (10).Several sequence-specific transcriptional repressors have recently been identified, but their mechanisms of action are less well understood. Repression domains have shown little or no sequence similarity to each other, indicating that they may have different downstream targets (28,39,47,52,53). As with the activators, some repression domains contact members of the basal machinery and/or interact with adapter proteins with the characteristics of a corepressor (48,54). A protein is considered to function as a corepressor for a specific transcription factor when it fulfills the ...

Nuclear hormone receptors are potent repressors of transcription in the unliganded state. We describe here the cloning of a nuclear receptor corepressor that we call SUN-CoR (Small Unique Nuclear receptor CoRepressor), which shows no homology to previously described nuclear hormone receptor corepressors, N-CoR, or SMRT. SUN-CoR is a highly basic, 16-kDa nuclear protein that is expressed at high levels in adult tissues and is induced during adipocyte and myogenic differentiation. Gene expression is regulated at the transcriptional level by sequence-specific transcriptional activators and repressors. The mechanism by which transcriptional repressors modulate gene expression is beginning to be elucidated. One mechanism involves protein-protein interaction with the basal transcription apparatus preventing recruitment of general transcription factors or maintaining the complex in a conformation that is inactive for transcriptional initiation. Many activation and repression domains have been shown to interact directly with components of the basal transcription apparatus, whereas others interact indirectly by first contacting an intermediate protein termed a coactivator or corepressor (reviewed in ref.

. An expression vector was prepared containing a cDNA coding for a truncated version of the intermediate filament (IF) protein desmin. The encoded truncated desmin protein lacks a portion of the highly conserved a-helical rod region as well as the entire nonhelical carboxy-terminal domain . When transiently expressed in primary fibroblasts, or in differentiating postmitotic myoblasts and multinucleated myotubes, the truncated protein induces the complete dismantling of the preexisting vimentin or desmin/vimentin IF networks, respectively. Instead, in both cell types vimentin and desmin are packaged into hybrid spheroid bodies scattered throughout the cytoplasm .Despite the complete lack of intact IFs, myoblasts S NCE their description in maturing myotubes and chondroblasts over twenty years ago (38), intermediate filaments (IF)' have been the subject of considerable study. Much has been learned concerning the requirements for particular protein domains, as well as sites of phosphorylation, for IF assembly in cell-free systems (22, 36, 37), for incorporation of IF proteins into preexisting filament networks in vivo (1,2,24,46,57, 66,69), as well as for de novo IF assembly within living cells (57) . The sequence of several IF genes have been delineated, and the flanking regulatory regions of some have been determined as well (52,56,70; and see reference 61 for review) . In contrast to this steady accumulation of molecular and biochemical data, and of information regarding the temporal and spatial expression of members of this multigene family during development (17,31,50,51), virtually nothing is known about the biological function(s) of any class of IFs .Several experimental approaches have been taken to attempt to uncover functional roles of IFs. Acrylamide has been shown to induce alterations in intermediate filament