Oncogene (2007) 26, 5439–5449 & 2007 Nature Publishing Group All rights reserved 0950-9232/07 $30.00 www.nature.com/onc REVIEW HDAC3: taking the SMRT-N-CoRrect road to repression P Karagianni and J Wong1 Department of Molecular and Cellular Biology, Baylor College of Medicine, One Baylor Plaza, Houston, TX, USA Known histone deacetylases (HDACs) are divided into repression when targeted to promoters, as well as different classes, and HDAC3 belongs to Class I. Through physical association with the DNA-binding factor forming multiprotein complexes with the corepressors YY1 (Yang et al., 1997; Dangond et al., 1998; Emiliani SMRT and N-CoR, HDAC3 regulates the transcription et al., 1998). Together, these findings suggested that this of a plethora of genes. A growing list of nonhistone ubiquitously expressed protein could be involved in the substrates extends the role of HDAC3 beyond transcrip- regulation of mammalian gene expression. On the other tional repression. Here, we review data on the composi- hand, HDAC3 contains an intriguingly variable C tion, regulation and mechanism of action of the SMRT/ terminus, with no apparent similarity with other N-CoR-HDAC3 complexes and provide several examples HDACs. This observation led to the hypothesis that of nontranscriptional functions, to illustrate the wide HDAC3 may have some unique properties and may variety of physiological processes affected by this deacety- not be completely redundant with the other HDACs lase. Furthermore, we discuss the implication of HDAC3 (Yang et al., 1997). This is also suggested by the in cancer, focusing on leukemia. We conclude with some differential localization of HDAC3, which, unlike the thoughts about the potential therapeutic efficacies of predominantly nuclear HDACs 1 and 2, can be found in HDAC3 activity modulation. the nucleus, the cytoplasm and at the plasma membrane Oncogene (2007) 26, 5439–5449; doi:10.1038/sj.onc.1210612 (Takami and Nakayama, 2000; Longworth and Laimins, 2006). Detailed domain analysis has revealed Keywords: HDAC3; SMRT; N-CoR; corepressor complex; that the protein contains both nuclear import and histone deacetylation; repression and cancer export signals, which account for this distinct localiza- tion pattern (Yang et al., 2002). The subcellular distribution of HDAC3 agrees with its nuclear function and opens up a whole new world of potential Introduction cytoplasmic substrates and regulators. Initially cloned based on sequence similarity with the previously identified histone deacetylases (HDACs) 1 HDAC3 complexes and 2, HDAC3 was the third mammalian deacetylase to be identified (Yang et al., 1997; Dangond et al., 1998; Composition Emiliani et al., 1998). Its open reading frame predicted a A breakthrough in the understanding of the biological 428 aa protein, with an estimated molecular mass of role of HDAC3 came with the realization that the 49 kDa. Human HDAC3 is 53% identical with human enzyme forms a stable complex with nuclear receptor HDAC1 and 52% identical with human HDAC2 at the corepressor (N-CoR) and silencing mediator of retinoic amino acid level (Yang et al., 1997; Dangond et al., and thyroid receptors (SMRT). Biochemical purifica- 1998). This similarity led to the classification of the tions of both N-CoR/SMRT as well as HDAC3- above three proteins into one family. HDAC8 would be associated proteins, mostly based on immunoaffinity later added to this class, which includes mammalian precipitation followed by mass spectrometry, converged enzymes related to yeast Rpd3 (Buggy et al., 2000). to this conclusion (Guenther et al., 2000; Wen et al., Similar to HDACs 1 and 2, HDAC3 is also ubiquitously 2000; Li et al., 2000; Zhang et al., 2002; Yoon et al., expressed. As expected by the sequence identity, early 2003). At least in HeLa cells, the majority of cellular functional analysis of the HDAC3 protein revealed HDAC3 is found to associate with SMRT and N-CoR common features with HDACs 1 and 2, namely complexes. Both N-CoR and SMRT had been discov- deacetylation of histone substrates, transcriptional ered as interacting partners of unliganded TR and RAR and mediators of their repressive functions (Chen and Correspondence: Dr J Wong, Department of Molecular and Cellular Evans, 1995; Horlein et al., 1995). The two corepressors Biology, Baylor College of Medicine, One Baylor Plaza, Houston, TX are high-molecular-weight proteins (B270 kDa), very 77030, USA. similar at the amino acid level, containing nuclear E-mail: [email protected] 1Current address: Institute of Biomedical Sciences, College of Life receptor-interacting domains as well as multiple repres- Science, East China Normal University, Dongchuan Road 500, sor domains (Chen and Evans, 1995; Horlein et al., Shanghai 200241, China. 1995; Ordentlich et al., 1999; Park et al., 1999). Role of HDAC3 in cancer therapy P Karagianni and J Wong 5440 Although N-CoR and SMRT share many functions, context-specific fashion, the consensus favors a stable they are likely not completely redundant, as suggested core N-CoR/SMRT complex comprising HDAC3, N- by the lethality of N-CoR deficient mice (Jepsen et al., CoR/SMRT, TBL1/TBLR1 and GPS-2. It is note- 2000). Identification of HDAC3 as the catalytic worthy that the core complex of SMRT and N-CoR component of the N-CoR/SMRT complexes provided containing HDAC3, TBL1, TBLR1 and GPS2 is very a mechanistic link between transcriptional repression stable, resistant to high concentration of salt and and histone deacetylation. Via their interactions with a detergents (Zhang et al., 2002). Furthermore, the number of different transcription factors, the two association of HDAC3 with N-CoR/SMRT is evolutio- corepressors recruit HDAC3 to specific promoters, narily conserved, as the yeast SET3 complex is believed where the enzyme deacetylates histones and mediates to be orthologous to the mammalian N-CoR/SMRT silencing of the corresponding genes. complexes (Pijnappel et al., 2001). The purified N-CoR/SMRT complexes contain addi- tional components including transducin b-like 1 (TBL1) Substrate specificity (Guenther et al., 2000; Li et al., 2000), the TBL1-related The existence of different HDACs and HDAC complexes protein (TBLR1) (Zhang et al., 2002; Yoon et al., 2003), raises the question of potential specificity in their and G protein pathway suppressor 2 (GPS-2) (Zhang enzymatic activities and general functions. Early studies et al., 2002; Yoon et al., 2003). Both TBL1 and TBLR1 showed that, like HDAC1 and HDAC2, HDAC3 isolated interact directly with SMRT and N-CoR but not with from mammalian cells can deacetylate both H3 and H4 in HDAC3 (Guenther et al., 2000; Zhang et al., 2002; Yoon free histones or nucleosome substrates (Yang et al., 1997; et al., 2003). In contrast to SMRT and N-CoR, TBL1 Dangond et al., 1998; Emiliani et al., 1998). In a more and TBLR1 are not required for HDAC3 enzymatic detailed study, Johnson et al. (2002) immunoprecipitated activity (Guenther et al., 2000; Zhang et al., 2002). HDAC1, -2, -3 and -6 from cell extracts and compared Although their functions may be at least partially their kinetics of cleavage of the acetyl group on different redundant, TBL1 and TBLR1 are essential for transcrip- lysine residues of hyperacetylated free histones. Although tional repression mediated by TR and other transcription this approach does not provide pure enzymatic prepara- factors (Guenther et al., 2000; Yoon et al., 2003). Work tions, some differences in the preferred lysine residues of by Perissi et al. (2004) further indicates that TBL1 and the individual complexes were observed. The in vitro TBLR1 are critical for transcriptional activation by a results suggested that HDAC3 could completely deacety- number of different nuclear receptors examined. Accord- late H2A, H4K5Ac and H4K12Ac, but only partially ing to the proposed model, in addition to their repressive deacetylate H3, H2B, H4K8Ac and H4K16Ac (Johnson function, TBL1 and TBLR1 also mediate proteasome- et al., 2002). Compared with HDAC1, HDAC3 deacety- dependent degradation of SMRT/N-CoR complexes lated H4K8Ac, H4K16Ac and H2B at the same rate, but from promoters, to allow for de-repression and subse- it deacetylated H4K5Ac, H4K12Ac and H2AK5Ac much quent recruitment of coactivators. Thus, the two proteins more rapidly. become essential for gene expression. However, Yoon Looking at an endogenous retinoic acid-regulated et al. (2005) failed to observe any effect of TBL1/TBLR1 gene, Hartman et al. (2005) observed a similar specificity in transcriptional activation, leaving the exact roles of of the N-CoR/SMRT-HDAC3 complex. Upon ligand TBL1/TBLR1 controversial. removal, recruitment of the corepressor complex re-estab- Human GPS2 was initially identified via its ability to lishes the repressed state of the gene and this involves suppress lethal G protein subunit-activating mutations progressive deacetylation of lysines in H4 tails. Deacety- in the yeast pheromone response pathway (Spain et al., lation of the H4 lysines occurs in a nonrandom pattern, 1996). Overexpression of GPS2 in mammalian cells has starting with K5 and followed by K8, K12 and K16, been shown to potently suppress JNK1 activation by and this ordered process is dependent on the activity of serum factors or TNFa (Spain et al., 1996; Jin et al., HDAC3 (Hartman et al., 2005). 1997) and promote the transactivation activities of Using an in vitro reconstituted chromatin template, bovine papillomavirus E2 and tumor suppressor p53 Vermeulen et al. (2004) tested histone deacetylation proteins (Breiding et al., 1997; Peng et al., 2001). Within specificity of SMRT/N-CoR complexes. When targeted the N-CoR complex, GPS-2 interacts directly with to preacetylated nucleosomal templates, Sin3/HDAC N-CoR and TBL1 and may aid the assembly and was found to deacetylate both H3 and H4, whereas the stabilization of the complex (Zhang et al., 2002). N-CoR/SMRT-HDAC3 complex displayed preferential Importantly, the association with N-CoR/HDAC3 is activity toward H3 (Vermeulen et al., 2004). This result required for GPS-2 to inhibit the JNK pathway (Zhang is somewhat surprising, considering the data discussed et al., 2002).