CBFA2T3–ZNF651, like CBFA2T3–ZNF652, functions as a transcriptional corepressor complex

1. Introduction 

Transcriptional activation and repression is critical in development and differentiation. Although about 10% of proteins encoded by the human genome are transcription factors, only a small number of these have been functionally characterised [1]. Transcription factors directly bind their cognate DNA binding sequences generally located within the promoter regions of their target genes. A number of these DNA binding transcription factors are evolutionarily conserved, particularly within regions of the protein that directly bind the DNA sequences. For example, Gfi-1 and Gfi-1b are highly conserved within their DNA binding zinc finger regions and both recognise the same DNA binding motif. However, their differing functions are attributed to dissimilarity between the two proteins outside their zinc finger regions [2].

CBFA2T1, CBFA2T2 and CBFA2T3 (MTG8, R1 and 16, respectively) constitute a group of ubiquitously expressed transcriptional regulatory proteins sometimes referred to as the “ETO” family. ETO proteins do not directly bind DNA, but exhibit their repressor activity by interacting with transcription factors (for example, BCL6, PLZF, Gfi-1, ZNF652) that bind directly to their cognate DNA binding sequences located within the promoters of target genes. ETOs are scaffold proteins that recruit a range of corepressor proteins such as N-CoR, SMRT, Sin3A and ATN1 and HDACs to generate complexes that function to repress gene transcription [3].

We have previously shown that the classical C2H2 zinc finger DNA binding protein ZNF652 specifically and functionally interacts with the ETO protein CBFA2T3 to repress transcription [4]. The CBFA2T3–ZNF652 complex was proposed to repress transcription of genes that have roles in breast oncogenesis [4]. Subsequently, we identified the ZNF652 consensus DNA binding sequence, and showed that CBFA2T3–ZNF652 represses HEB expression by binding to a single ZNF652 DNA binding motif located within the HEB promoter [5]. We have now identified ZNF651 (also called ZBTB47) as a ZNF652 paralogue. The deduced ZNF652 and ZNF651 amino acid sequences are highly conserved within the zinc finger region, and we show that ZNF651 can also bind to the consensus ZNF652 DNA binding sequence. We present data showing that CBFA2T3–ZNF651 functions as a repressor complex in a manner similar to CBFA2T3–ZNF652. We also determined that both ZNF651 and ZNF652 share a region of homology through which they interact with CBFA2T3. We predict that ZNF651 and ZNF652 perform functionally similar roles in a tissue-specific manner.

2. Materials and methods 

2.4. Dual luciferase reporter assays 

Dual luciferase assays were performed using HeLa cells as previously reported [5]. Briefly, HeLa cells were co-transfected with the firefly luciferase expressing pGL2-IRF2BP1-TK-Luc and Renilla luciferase expression plasmid pRL-TK and the indicated expression constructs (see Fig. 4). The firefly luciferase activity was expressed relative to the Renilla luciferase activity. All reporter assays were performed in triplicate and repeated at least three times with the data presented as mean±S.E.

View Large Image Download to PowerPoint

Fig. 4. CBFA2T3 enhances ZNF651- and ZNF652-mediated transcriptional repression. Increasing levels of ZNF651 and ZNF652 result in a dose dependent decrease in pGL2-IRF2BP1-TK-Luc reporter gene activity and this is further reduced in the presence of CBFA2T3. Dual-luciferase reporter assays were performed to determine the effect of ZNF651 and ZNF652 on transcriptional activity of the pGL2-IRF2BP1-TK-Luc reporter construct (schematic shown below the graph). ZNF651/ZNF652 DNA binding site located at −928 to −941 within the IRF2BP1 promoter is shown. The core ZNF652 DNA binding sequence is underlined. HeLa cells were transfected with pGL2-IRF2BP1-TK-Luc together with or without different expression constructs as shown. pRL-TK-Renilla luciferase vector was used as a transfection control. Promoter activity was calculated from the ratio of firefly to Renilla luciferase activities. Repressive activity was enhanced in the presence of ZNF651 or ZNF652 alone or in combination with CBFA2T3. Data shown is representative of three independent experiments and is presented as mean±S.E. (n=3).

3. Results 

3.1. ZNF651 and ZNF652 are paralogues that are differentially expressed in human tissues 

ZNF651 is located on chromosome 3 and encodes a 371 amino acid protein that is shorter than the 606 amino acid ZNF652 protein encoded by the ZNF652 gene located on chromosome 17. Alignment of the deduced amino acid sequences of ZNF651 and ZNF652 shows 77% overall similarity. The two proteins are highly conserved (95%) within their zinc finger regions. Besides this zinc finger region, a short carboxy-terminal proline-rich sequence of ZNF651 and ZNF652 is the only other region with significant similarity (85%) (Fig. 1). Whereas both ZNF651 and ZNF652 are differentially expressed in different tissues, only ZNF652 is expressed in blood and mammary tissue (Fig. 2).

View Large Image Download to PowerPoint

Fig. 1. ZNF651 and ZNF652 are paralogues. Alignment of the deduced amino acid sequences of ZNF651 and ZNF652. Conserved regions are shown in yellow. The zinc finger domains are marked with lines above the sequence and amino acids conserved within the carboxy-terminal proline-rich regions are underlined.

View Large Image Download to PowerPoint

Fig. 2. ZNF651 and ZNF652 are differentially expressed in different human tissues. The graph shows levels of ZNF651 and ZNF652 expression in selected human tissues. Data for the transcripts per million in the pool for each tissue was accessed from http://www.ncbi.nlm.nih.gov/UniGene/ESTProfileViewer.cgi?uglist=Hs.409561 (ZNF651) and http://www.ncbi.nlm.nih.gov/UniGene/ESTProfileViewer.cgi?uglist=Hs.463375 (ZNF652).

3.2. ZNF651 and ZNF652 bind to the identical consensus DNA binding sequence 

Using a CASTing protocol, we previously identified NBNAVGGGTTAAN (where N=AGCT, B=GCT and V=AGC) as the ZNF652 DNA binding sequence [5]. As ZNF651 and ZNF652 shared 95% similarity within their zinc finger regions, we premised whether ZNF651 binds to the ZNF652 consensus DNA binding sequence. This was determined by an EMSA using nuclear extracts from HEK293T cells ectopically expressing HA-tagged ZNF651 or ZNF652 proteins. Ectopically expressed HA-ZNF651 and HA-ZNF652 proteins were used due to unavailability of an EMSA compatible anti-ZNF651 antibody. Binding reactions were performed by mixing the nuclear extracts and radio-labelled short double stranded DNA probes carrying either wild type ZNF652 or mutant sequences and resolved on a non-denaturing acrylamide gel. Whereas both the ZNF652 and ZNF651 proteins bound to the wild type probes (Fig. 3, lanes 4 and 10) no such binding was seen with the mutant probes (lanes 5 and 11). To confirm the specificity of protein–DNA complexes, rabbit anti-ZNF652 and rat anti-HA antibodies were used in a supershift EMSA. Whereas the ZNF652– and ZNF651–wild type DNA probe complexes supershifted in the presence of anti-ZNF652 (lane 6) and anti-HA (lane 12) antibodies, respectively, no further mobility shift was detected in the presence of appropriate non-specific antibodies (lanes 8 and 14). Taken together, these results suggest that both ZNF652 and ZNF651 can specifically bind to the same DNA binding sequence that was previously identified as the consensus ZNF652 DNA binding sequence. Although both proteins bind to the same DNA binding sequence, subtle differences in their strength of binding may be present.

View Large Image Download to PowerPoint

Fig. 3. ZNF651 and ZNF652 proteins bind to the same DNA binding sequence. The nuclear extracts from HEK293T cells ectopically expressing either HA-tagged ZNF652 (lanes 4–9) or ZNF651 (lanes 10–15) were incubated with 32P-labelled wild type (WT) probe consisting of consensus ZNF652 binding sequence, or a non-specific (NS) probe containing a randomised sequence of the same length, and analysed by EMSA. 18- and 35-mer oligonucleotide probes were used for ZNF651 and ZNF652 binding reactions, respectively. Binding reactions containing WT (lane 1, 18-mer), NS (lane 2, 18-mer) or WT (lane 3, 35-mer) oligonucleotide probes and HEK293T nuclear extracts were used as controls. The anti-ZNF652 (lanes 6–7), anti-HA (lanes 12–13) or appropriate non-specific (lanes 8–9 and 14–15) antibody was added to the reactions for supershift analysis (lanes 6–9 and 12–15, respectively). Protein–DNA complexes and protein–DNA-supershift (SS) complexes that supershifted only in the presence of anti-ZNF652 or anti-HA antibody are shown with arrows.

3.4. The CBFA2T3–ZNF651 complex specifically binds to the ZNF652 DNA binding sequence 

An in vitro DNA binding assay was performed to determine whether CBFA2T3 associates with ZNF651 when bound to the ZNF652 DNA binding sequence. Nuclear extracts from HEK293T cells ectopically expressing Myc-ZNF651 and Myc-CBFA2T3 proteins were incubated with DNA fragments carrying either wild type or mutant ZNF652 DNA binding sequences. Both ZNF651 and CBFA2T3 proteins bound to the wild type ZNF652 DNA binding sequence but no such binding to the mutant sequences was observed (Fig. 5). A low level of CBFA2T3 binding to the magnetic beads charged with mutant DNA sequence appeared to be non-specific as a comparable level of CBFA2T3 binding was also observed in the uncharged beads (Fig. 5A; lanes 3 and 4).

View Large Image Download to PowerPoint

Fig. 5. CBFA2T3 interacts to the ZNF651 bound to the ZNF652 DNA binding sequence and ZNF651 interacts with CBFA2T3 through its conserved proline-rich carboxy-terminal sequence. (A) The 5′-end biotinylated DNA carrying wild type (WT) or mutant (MT) DNA binding sequences were immobilised to streptavidin-coated magnetic beads. Magnetic beads charged with DNA sequences (lanes 2 and 3) or uncharged beads (lane 4) were incubated with protein lysates from HEK293T cells ectopically expressing Myc-tagged ZNF651 and CBFA2T3 proteins and then washed and eluted. Input and eluted proteins were analysed by Western blotting with anti-Myc antibody. Low level of non-specific binding of the CBFA2T3 to the magnetic beads was detected. (B) HEK293T cells were transfected with constructs expressing HA-tagged ZNF651 proteins alone (lanes 1 and 4) or with Myc-CBFA2T3 (lanes 2 and 5) or Myc-CBFA2T3-3 (lanes 3 and 6) as shown. Cell lysates were co-immunoprecipitated with anti-Myc antibody. Inputs (lanes 1–3) and immunoprecipitates (lanes 4–6) were Western blotted with anti-HA or anti-Myc antibodies.

4. Discussion 

Transcriptional regulation is an extremely complex and tightly controlled process and is the result of finely balanced activities of activator and repressor proteins. These controls are critical to a multitude of developmental and differentiation processes while deregulation of transcriptional regulation can lead to disease or cancer. The observation that a significant proportion of the human genome encodes for transcription factors, whose key functional motifs are generally highly conserved among diverse organisms, signifies the critical role they play in essential developmental processes.

ETOs are modular proteins that do not directly bind DNA but interact with transcription factors bound to their cognate DNA binding sequences located within the promoter regions of target genes and recruit a range of corepressors to facilitate transcriptional repression. The ETO proteins interact with the zinc finger proteins, Gfi-1, BCL6, PLZF, GATA-1 and transcription factors HEB and TAL-1/SCL to repress gene expression [5]. We showed that ETOs can also exhibit their transcriptional repressor activity by interacting with a novel zinc finger protein ZNF652 [4]. In this report we have now shown that CBFA2T3 can also exhibit its repressive activity through a ZNF652 paralogue ZNF651 adding to the existing range of ETOs-transcriptional complex-mediated gene repression. We have presented in vitro and in vivo data showing that ZNF651 and ZNF652 can bind to the same DNA binding consensus sequence and that CBFA2T3 interacts with both of these proteins. We have shown that both CBFA2T3–ZNF651 and CBFA2T3–ZNF652 function as corepressors on the ZNF652 DNA binding site located within the IRF2BP1 promoter. As CBFA2T3–ZNF651 did not show transcriptional repression on the HEB promoter ([5] and data not shown) and as ZNF651 and ZNF652 proteins are differentially expressed among different tissues, we premise that the two complexes may exhibit both tissue and gene specific roles.

Although dysregulation of most ETO-based complexes is associated with leukaemia [3], down regulation of CBFA2T3 in breast tumours and functional studies are consistent with a role of CBFA2T3 as a breast tumor suppressor [6]. We have reported that CBFA2T3 suppresses breast oncogenesis through its interaction with ZNF652 [4]. A recent report showed that CBFA2T3 interacts with the soluble intracellular domain, termed s80, of ERBB4 and has a role in ERBB4-dependent differentiation [7]. Whereas CBFA2T3 is ubiquitously expressed, ZNF651 and ZNF652 show differential expression among a range of human tissues. Therefore, we suggest that CBFA2T3–ZNF651 and CBFA2T3–ZNF652 complexes exhibit their transcriptional repressor function in a tissue-specific manner.

We showed that CBFA2T3 interacts with ZNF652 through a proline-rich region located within the carboxy-terminal region of ZNF652. We also showed that both NHR3 and NHR4 motifs of CBFA2T3 are required for its interaction with ZNF652 [5]. Here we have shown that CBFA2T3-3 carrying the NHR3–NHR4 motifs interacts with ZNF651 and that this interaction most likely also occurs through ZNF651 carboxy-terminal proline-rich region. This is because, besides zinc finger regions, carboxy-terminal proline-rich sequence is the only other region with significant similarity between ZNF651 and ZNF652. Previous studies on the ETO member CBFA2T1 showed that both the NHR3 and NHR4 domains were also required for its interaction with the corepressor N-CoR [8], [9] and this interaction occurs through a conserved proline-rich PPLXP motif within N-CoR [8]. A number of other proteins have been reported to interact with MYND domains (NHR4) through proline-rich domains and a PXLXP peptide motif has been proposed (Fig. 6). ZNF651 and ZNF652 carboxy-terminal proline-rich region conforms to this motif. MYND domains are defined by a C4–C2HC consensus and are frequently implicated in transcriptional repression [10], [11]. Interaction of ZNF651 and ZNF652 with CBFA2T3 through their carboxy-terminal proline-rich conserved sequence further emphasises the functional significance of proline-rich regions in protein–protein interaction and cell signaling [12].

View Large Image Download to PowerPoint

Fig. 6. Conserved proline-rich motifs facilitate functional protein–protein interactions. The diagram shows the structure of ZNF651 and ZNF652 proteins. An alignment of the proline-rich regions of ZNF652, ZNF651, N-CoR, and SMRT proteins is also shown. Amino acids (aa) identical among the aligned sequences are shaded.

Similarly to other examples of paralogous transcription factors, we find that ZNF651 and ZNF652 share a number of similarities with the Gfi-1 and Gfi-1b transcriptional repressor proteins. Both ZNF651 and ZNF652 are highly conserved paralogous transcriptional factors that are located on different chromosomes (3 and 17, respectively) and may have resulted from gene duplication. Both proteins bind to the same DNA binding sequences located within promoter regions of their target genes and interact with the corepressor CBFA2T3 through a proline-rich region located within their respective carboxyl-terminal regions. ZNF651 and ZNF652 are conserved within their zinc finger and proline-rich regions that are critical for DNA binding and CBFA2T3 corepressor interaction, respectively. Likewise, Gfi-1 and Gfi-1b are highly conserved pair of paralogous transcriptional repressors located on different chromosomes (chromosome 1 and 9, respectively) that may also have resulted from gene duplication, and bind to the same DNA binding sequence and interact with CBFA2T3. However, unlike ZNF651 and ZNF652, Gfi proteins interact with CBFA2T3 through their evolutionarily conserved six carboxy-terminal zinc-finger motifs.

Typically, ETO proteins form complexes with the DNA binding proteins and recruit corepressors such as N-CoR, Sin3A, SMRT and HDACs. It is not known whether CBFA2T3–ZNF652 recruits unique co-factors although ZNF651- and ZNF652-specific protein motifs may provide interfaces for interaction with as yet unidentified co-factors thereby providing functional specificity to the two transcriptional repressors. These findings further define the complexity and diverse nature of the ETO-based repressor complexes. ZNF651 and ZNF652 are differentially expressed in different tissues (Fig. 2). Expression of ZNF651 is absent while ZNF652 is expressed in blood and mammary tissue. The presence of relatively high ZNF652 expression in mammary tissue is consistent with the original identification of ZNF652 as a CBFA2T3 interacting protein in a yeast two-hybrid screen of a breast cDNA library [4]. It is suggested that although ZNF651 and ZNF652 transcription factors interact with the same consensus DNA binding sequence, functional specificity is provided by their tissue distribution. Future work will determine tissue-specific functional differences between the CBFA2T3–ZNF651 and CBFA2T3–ZNF652 transcriptional repressor complexes.