RNA UK 2012

SAFB1- and SAFB2-mediated transcriptional repression: relevance to cancer

Elaine A. Hong, Hannah L. Gautrey, David J. Elliott, Alison J. Tyson-Capper


SAFB1 (scaffold attachment factor B1) and a second family member SAFB2, are multifunctional proteins implicated in a variety of cellular processes including cell growth, apoptosis and stress response. Their potential function as tumour suppressors has been proposed based on well-described roles in tran-scriptional repression. The present review summarizes the current knowledge of SAFB1 and SAFB2 proteins in transcriptional repression with relevance to cancer.

  • cancer
  • scaffold attachment factor
  • transcription repression
  • tumour suppressor


SAFB1 (scaffold attachment factor B1) was originally identified by two independent groups based on its ability to bind scaffold/matrix attachment regions [1] and to regulate transcription of the HSP27 (heat-shock protein 27) gene [2]. A subsequent study identified SAFB1 as an interacting partner of hnRNP (heterogeneous nuclear ribonucleoprotein) A1 [3]. These studies revealed multiple functions for SAFB1 in chromatin organization, transcriptional control and RNA processing. SAFB2 was eventually isolated and characterized as a second member of the SAFB family, and has more than 70% homology to SAFB1 at the protein and nucleotide level [4].

The SAFB1 gene maps to human chromosome 19p13.3 [5]. The SAFB2 gene is located adjacent to SAFB1 in a head-to-head arrangement, separated by a 490 bp bidirectional promoter [4]. Translation of the SAFB genes give rise to large multi-domain proteins that consist of a SAF-box (scaffold attachment factor-box) at the N-terminus; an RRM (RNA-recognition motif) and NLS (nuclear localization signal) in the central region; and glutamic acid/arginine and glycine-rich regions at the C-terminus (Figure 1). A potent transcriptional repression domain is located within the C-terminal regions of SAFB1 and SAFB2 which mediates the transcriptional repression activity of these proteins [6].

Figure 1 SAFB1 and SAFB2 domain structure

Functional domains of SAFB1 and SAFB2 containing the SAF-box, RRM, NLS, glutamic acid/arginine and glycine-rich regions. The C-terminal region that includes the repression domain and degree of homology between SAFB1 and SAFB2 is indicated [6].

SAFB1/SAFB2 recruitment to promoter regions

A major role for the SAFB proteins, particularly well-described for SAFB1, is in the transcriptional regulation of gene expression. This might potentially be mediated via direct promoter-binding or indirect transcription factor interactions. The first evidence of promoter-dependent transcriptional regulation was observed by SAFB1's ability to repress HSP27 promoter activity [2]. In this study, SAFB1 was shown to bind to the HSP27 promoter and significantly decrease HSP27 promoter activity in several breast cancer cell lines.

A later study further supported the promoter-dependent repression activity of SAFB1 protein on the expression of the XOR (xanthine oxidoreductase) gene [7]. Affinity purification of DNA probes containing E-box (enhancer box) and Ku86 sites yielded SAFB1 as one of the binding proteins. Further investigation revealed that SAFB1 negatively regulates XOR gene expression in part through its binding to the E-box of the XOR promoter and its interaction with Ku86 [7].

Interestingly, Omura et al. [8] reported that SAFB1 binding at the promoter of the SREBP1c (sterol-regulatory-element-binding protein 1c) gene could positively regulate the expression of this gene. However, this effect was only distinct when the RBMX (X-chromosome linked RNA binding motif) protein was equally present in the cell line tested. SAFB1 and SAFB2 proteins are known to interact with RBMX (also known as hnRNP G) [9].

Evidence from these studies suggest that recruitment of SAFB1 protein alone to promoter regions may not be sufficient to completely mediate transcriptional regulation of its target genes, but an interaction between SAFB1 and other factors is required to carry out this function. This is consistent with a recent report by Hammerich-Hille et al. [10] that showed lack of significant similarities between SAFB1/SAFB2 target genes identified using two different experimental methods, chromatin immunoprecipitation and genome-wide expression array. This study speculated that the majority of SAFB1/SAFB2 target genes are not regulated through their recruitment to promoter regions, but through other mechanisms.

SAFB1/SAFB2 interaction with transcription factors

The most evident mechanism of SAFB1/SAFB2 transcriptional regulation is through interactions with various transcription factors, mainly facilitated by its C-terminal repression domain. Previous identification of SAFB1 binding at the promoter region of the oestrogen-regulated HSP27 gene prompted Oesterreich et al. [11] to investigate the association between SAFB1 protein and ERα (oestrogen receptor α) action. Their study revealed a ligand-independent SAFB1 interaction with ERα protein via multiple domains, with a notably strong interaction at the DNA-binding/hinge regions of ERα. SAFB1 protein represses ERα transcriptional activity through its interaction with the ERα DNA-binding domain; however, this did not prevent ERα from binding DNA [11]. Similar observations were also reported for SAFB2 hence, SAFB1 and SAFB2 are both ERα co-repressors [4]. Interestingly, our recent findings show that SAFB1 and SAFB2 expression itself changes in response to increasing concentrations of β-oestradiol in different breast cancer cell lines [12].

SAFB1 protein was identified in another yeast two-hybrid screen, in an attempt to identify PPARγ (peroxisome-proliferator-activated receptor γ) interacting proteins that could influence its nuclear receptor activity [13]. In this study, Debril et al. also revealed the promiscuous nature of SAFB1, when interactions were detected with several other nuclear receptors, mainly in a ligand-independent manner, including FXRα (farnesoid X receptor a), RORα1 (RAR-related orphan receptor α1), VDR (vitamin D receptor), LRH-1 (liver receptor homologue 1) and the transcriptional factor c-Jun.

Interactions between SAFB1/SAFB2 and the multifunctional tumour suppressor protein p53 has also been described in an extended study that identified a common mediator, SRPK1a (serine/arginine protein kinase 1a) [14]. The p53 protein physically interacts with the C-terminal region of SAFB1/SAFB2, and protein co-localization was visible under treatment with a genotoxic agent, 5-FU (5-fluorouracil). This study showed that SAFB1 repressed p53-dependent transcriptional activity [14].

Regulation of SAFB1/SAFB2 transcriptional repression

Even though the role of SAFB1/SAFB2 has been well-studied, knowledge about the mechanism of SAFB1/SAFB2 repressive activity is still limited. Mapping of the repression domain within SAFB1/SAFB2 revealed an intrinsic repression domain located in the C-terminal region (between amino acids 599 and 915) that is essential for the ERα–co-repressor activity [6]. Although SAFB1 interacts with ERα at its central domain containing the RRM, a SAFB1-mutant (ΔRRM) still functions as an ERα co-repressor, emphasizing the importance of the C-terminus repression domain to mediate its co-repressor action [6].

Repression through this SAFB C-terminal domain could partially be relieved by HDAC (histone deacetylase) inhibitors, suggesting that HDAC activity may be required for SAFB1/SAFB2-mediated repression [6,15]. A yeast two-hybrid assay, however, did not reveal any direct interaction between the SAFB1/SAFB2 repression domain and any known HDAC proteins or members of HDAC complexes [6]. An indirect interaction with HDAC complexes was therefore speculated for SAFB1/SAFB2. Further investigation revealed N-CoR (nuclear receptor co-repressor) as an interacting mediator between SAFB1 and HDAC3 [16]. N-CoR is a well-characterized co-repressor that modulates nuclear receptor repression by recruiting HDACs or HDAC complexes to target promoters to remodel local chromatin, resulting in transcriptional repression (reviewed in [17]). Jiang et al. [16] reported that N-CoR interacts directly with the C-terminus repression domain of SAFB1 and partial co-localization was observed through confocal microscopy. The repressive effect of SAFB1 was diminished in the absence of N-CoR, suggesting that its repressive ability is partly a HDAC-dependent action mediated through N-CoR [16].

Figure 2 SAFB1 and SAFB2 function in transcriptional regulation

Regulation of SAFB1 and SAFB2 function mediated through (a) direct binding to the promoter region, (b) association to transcriptional factors as co-regulators, (c) interaction with C-terminal domain of RNA polymerase II, and (d) formation of mRNA splicing complexes [26].

A recent study by Garee et al. [18] presented evidence for an alternative mechanism that could regulate SAFB1/SAFB2 action through the post-translational modification by SUMO (small ubiquitin-related modifier) known as SUMOylation. Similar to ubiquitination, SUMOylation is an enzymatic process that occurs at a unique four amino acid motif that includes a lysine residue that is modified (reviewed in [19]). SUMOylation of transcription factors and co-regulators has been commonly shown to negatively regulate transcriptional activity (reviewed in [20]). Potential lysine modification sites were identified at amino acids 231 and 294 within SAFB1, and these sites were 100% conserved in SAFB2 [18]. SUMOylation of SAFB1 occurs predominantly at amino acid 294 and was found to be a common effect that is not cell line specific. Interestingly, SAFB1 transcriptional repression ability was abolished when both SUMOylation sites were mutated, consequently linking SAFB1 SUMOylation and its repressive activity. SUMOylation of SAFB1 did not show a change in subcellular localization, protein t1/2 or alter interaction with N-CoR. However, SAFB1 SUMOylation significantly decreased its interaction with HDAC3 [18].

Song et al. [21] also recently revealed another post-translational modification for SAFB2 protein in an attempt to identify BRCA1 (breast-cancer susceptibility gene 1) protein ubiquitination substrates. BRCA1 protein interacts with BARD1 (BRCA1-associated RING domain protein 1) to form a stable heterodimer complex that displays ubiquitin E3 ligase activity (reviewed in [22]). Unlike most ubiquitin ligases, the BRCA1/BARD1 complex catalyses unconventional formation of Lys6-linked chains which do not result in protein degradation [23]. BRCA1 has been shown to induce ubiquitination of SAFB2 and result in an increased SAFB2 protein expression [21]. The overexpression of SAFB2 then significantly reduced levels of BARD1 via its C-terminal repression domain, but did not affect BRCA1 expression. Taken together, these results showing up-regulation of SAFB2 through BRCA1/BARD1-mediated ubiquitination and consequent down-regulation of BARD1 by SAFB2 overexpression, suggest a possible feedback loop that regulates SAFB2 and BARD1 protein levels [21]. Song et al. [21] also proposed a possible role for SAFB1/SAFB2 as the missing link between BRCA1 and its known function in repressing ERα transcriptional activity [24,25].

SAFB1/SAFB2 in cancer

The strong evidence of interactions between SAFB1/SAFB2 proteins with key players in tumorigenesis indicates an important role for SAFB1/SAFB2 in cancer. Findings from functional and clinical studies provide further evidence to implicate the relevance of SAFB1/SAFB2 in cancer, particularly in breast tumorigenesis. SAFB1/SAFB2 have now been described as candidate tumour suppressor proteins when high frequency of loss of heterozygosity near the SAFB1/SAFB2 locus on chromosome 19p13 was observed in breast cancer patients [27]. However, linkage studies in Swedish families indicated that the SAFB1/SAFB2 is not involved in hereditary breast cancer, but may be important in sporadic breast cancer [28]. Although the expression of SAFB1 in breast cancer cell lines and breast tumour tissues displayed high variability, overexpression of SAFB1 caused cell growth inhibition [29], while low levels of SAFB1/SAFB2 in invasive breast tumour samples correlates with poor patients survival [30]. Moreover, mouse embryonic fibroblasts with a genetic deletion of SAFB1 exhibit important characteristics of carcinogenesis, including cellular immortalization, increased cell transformation, ability to proliferate in growth-restricting conditions and increased anchorage-independent growth [31]. A more recent study to identify SAFB1/SAFB2 endogenous target genes in breast cancer cells has confirmed their role as transcriptional repressors, regulating genes involved in apoptosis and the immune system [10].


Over a decade of research has revealed the characteristics of SAFB1/SAFB2, and a myriad of evidence has implicated their contribution in cancer progression. SAFB1/SAFB2 may have more complex roles in cellular functions involving RNA processing and metabolism that could potentially affect a wide array of signalling pathways in cancer in addition to those discussed here (Figure 2). Additional studies are necessary to clearly define the mechanism of SAFB1/SAFB2 action in the context of tumorigenesis to facilitate the improvement of cancer treatment strategies.


E.A.H. was sponsored by the Dr William Harker Foundation and the Overseas Research Students Awards Scheme.


  • RNA UK 2012: An Independent Meeting held at The Burnside Hotel, Bowness-on-Windermere, Cumbria, U.K., 20–22 January 2012. Organized and Edited by Raymond O'Keefe and Mark Ashe (Manchester, U.K.).

Abbreviations: BRCA1, breast-cancer susceptibility gene 1; BARD1, BRCA1-associated RING domain protein 1; E-box, enhancer box; ERα, oestrogen receptor α; HDAC, histone deacetylase; hnRNP, heterogeneous nuclear ribonucleoprotein; HSP27, heat-shock protein 27; N-CoR, nuclear receptor co-repressor; NLS, nuclear localization signal; RBMX, X-chromosome linked RNA-binding motif; RRM, RNA-recognition motif; SAFB1, scaffold attachment factor B1; SAF-box, scaffold attachment factor-box; XOR, xanthine oxidoreductase


View Abstract