CBFA2T3-ZNF652 Corepressor Complex Regulates Transcription of the E-box Gene HEB*

Transcriptional repression plays a critical role in development and homeostasis. The ETO family represents a group of highly conserved and ubiquitously expressed transcriptional regulatory proteins that are components of a diverse range of multiprotein repressor complexes. ETO proteins function as transcriptional repressors by interacting with a number of transcription factors that bind to their cognate consensus DNA binding sequences within the promoters of target genes. We previously reported that the classical C2H2 zinc finger DNA-binding protein, ZNF652, specifically and functionally interacts with the ETO protein CBFA2T3 and has a role in the suppression of breast oncogenesis. Here we report the identification and validation of the ZNF652 consensus DNA binding sequence. Our results show that the E-box gene HEB is a direct target of CBFA2T3-ZNF652-mediated transcriptional repression. The CBFA2T3-ZNF652 complex regulates HEB expression by binding to a single ZNF652 response element located within the promoter sequence of HEB. This study also shows that the NHR3 and NHR4 domains of CBFA2T3 interact with a conserved proline-rich region located within the C terminus of ZNF652. Our results, together with previous reports, indicate that HEB has a complex relationship with CBFA2T3; CBFA2T3 interacts with ZNF652 to repress HEB expression, and in addition CBFA2T3 interacts with the HEB protein to inhibit its activator function. These findings suggest that CBFA2T3-ZNF652-mediated HEB regulation may play an important role in hematopoiesis and myogenesis.

Gene regulation occurs through the balanced activity of transcriptional activators and repressors and is critical for the ordered development and maintenance of homeostasis. Inherited or acquired defects in transcription factor structure and function can result in irreversible alterations in this balance. The ensuing aberrant expression of target genes can lead to developmental abnormalities or to the initiation and promo-tion of cancer (1). Spatial and temporal gene regulation by transcriptional activation has been intensely investigated, whereas the critical role of transcriptional repression in development and disease has only been recently recognized (2). CBFA2T1 (RUNX1T1, MTG8) together with CBFA2T2 (MTGR1) and CBFA2T3 (MTG16) form a group of ubiquitously expressed transcriptional regulatory proteins called the "ETO" family. The ETO proteins have modular structures and are characterized by the presence of four regions, NHR1 to NHR4, so named because of their homology to the Drosophila nervy protein (3). ETO proteins are primarily nuclear-localized and are capable of oligomerization and interaction with other proteins through their NHR 2 domains (4 -6). The NHR1 (also called eTAFH) domain of CBFA2T1 has been shown to bind the repression domain I of nuclear receptor co-repressor N-CoR in vitro and repress transcription in a reporter assay (7). NHR1 also mediates ETO interactions with a conserved activation domain of the E-box proteins HEB and E2A and inhibits their transactivation functions (8). Therefore, the NHR1 motif can interchange negative and positive coregulatory proteins to control transcription (9). The second domain, NHR2, is described as a hydrophobic heptad repeat domain and is required for homo-and heterotetramerization of the ETO proteins (6). The tetramerization of the ETO proteins is not essential for interaction with the corepressors N-CoR, SMRT, mSin3A, and HDAC1-3. However, NHR2-mediated tetramerization of the AML1-CBFA2T1 fusion protein contributes to the development of leukemia in t(8:21) translocation carriers (6). The third domain in CBFA2T1, NHR3, has been shown to interact with the regulatory subunit of type II cyclic AMP-dependent protein kinase (PKA RII␣) in lymphocytes (10). The last of these domains, NHR4, also called the MYND (Myeloid-Nervy-DEAF-1) domain, contains two nonclassical zinc fingers and, in conjunction with NHR1 and NHR3, interacts with the core-pressor N-CoR (7). Additionally, the NHR4 domain of CBFA2T1 also directly interacts with SMRT (5,11), although the significance of this interaction in AML1-CBFA2T1-induced leukemogenic activity has not been completely resolved (12,13). Consistent with the role of the ETO proteins in transcriptional repression, NHR2, NHR3, and NHR4 also interact with various members of the HDAC family.
CBFA2T1 is the most studied of the ETO proteins because of its involvement in the non-random t (8;21) translocations that are frequently associated with acute myeloid leukemia (AML). The t (8;21) in AML generates an in-frame fusion of the coding sequences of the N-terminal zinc finger DNA-binding domain of AML1 (also called RUNX1) with almost the entire CBFA2T1 encoding region, leading to a chimeric AML1-CBFA2T1 protein (14). AML1 normally functions as an activator of transcription, but when fused with CBFA2T1, it acts as a repressor. The resulting changes in gene expression are thought to be the basis for the development of leukemia. The ETO protein CBFA2T3 is also involved in a recurrent t (16;21) translocation with AML1 and is associated with therapy-related myeloid malignancy (15).
Gene knock-out studies in mice have been utilized to determine the normal cellular role of the ETO family of proteins, although the observed phenotypes may be influenced by functional complementation between the ETO family members. Studies of knock-out mice show that Cbfa2t1 plays a critical role in gut development (16), whereas Cbfa2t2 is required for maintenance of the secretory cell lineage in the small intestine (17).
Further clues regarding the normal function of the ETO proteins have been obtained through their interacting proteins. CBFA2T1 regulates early adipogenesis by inhibiting transcriptional activity of C/EBP␤, a member of the C/EBP (CCAAT/ enhancer-binding protein) family of leucine zipper transcription factors (18). Both CBFA2T1 and CBFA2T3 have been shown to interact with the corepressor ATN1 (19). An expansion of a polyglutamine repeat within the ATN1 protein is associated with the neurodegenerative disease dentatorubral-pallidoluysian atrophy (DRPLA) (19). As a component of multiprotein complexes, CBFA2T3 has been shown to coordinate cellular proliferation and differentiation during erythropoiesis (20,21).
CBFA2T3 is located in the 16q24.3 loss-of-heterozygosity region in breast cancer, and our functional studies are consistent with a role of CBFA2T3 as a breast cancer tumor suppressor (22,23). We have previously shown that loss, or reduced levels, of CBFA2T3 expression is associated with breast oncogenesis (22). CBFA2T3 has also been shown to interact with the soluble intracellular domain, termed s80, of ERBB4 and to have a role in ERBB4-dependent differentiation (24). Reporter assays have shown ERBB4 also regulates CBFA2T3-mediated transcriptional repression. ERBB4 and CBFA2T3 are both expressed in normal breast tissue, and changes in their levels are associated with breast cancer. Interaction between CBFA2T3 and ERBB4 may therefore have an important role in the normal development of the mammary gland, and their deregulation may lead to breast cancer (24). On the whole, although the original description of the ETO family was based on their involvement in leu-kemia, these proteins have now been shown to have a range of other functions (14).
ETO proteins do not bind DNA. Gene-specific repression by ETO-containing multiprotein complexes is mediated through their interaction with transcription factors that can bind directly to the promoters of target genes. These DNA-binding transcription factors are predominantly classical C 2 H 2 zinc finger proteins. It is remarkable that all of the identified ETOinteracting DNA-binding zinc finger proteins are involved in cancer. For example, BCL6 (B-cell lymphoma 6 protein) is frequently translocated and mutated in diffuse large cell lymphoma (26), and PLZF (promyelocytic leukemia zinc finger) is disrupted by a t(11;17) rearrangement in acute promyelocytic leukemia (27). The zinc finger protein Gfi-1 is required for the development of neuroendocrine cells, sensory neurons, and both T-and B-lymphocytes (28). Germ-line mutations in the Gfi-1 coding sequence are associated with severe congenital neutropenia or non-immune chronic idiopathic neutropenia in adults (29). Recent findings also suggest that Gfi-1 can act as either an oncogene or tumor suppressor gene (28,30). Therefore, the ETO family of proteins generates a diverse constellation of transcriptional repressor complexes through interaction with various transcription factors and corepressors.
We have recently shown that the classical C 2 H 2 zinc finger DNA-binding protein ZNF652 specifically and functionally interacts with the ETO protein CBFA2T3 to repress transcription (31). ZNF652 has seven classical C 2 H 2 zinc finger motifs conforming to the consensus CX 2 CX 12 HX 3 H sequence, with three of these joined by part or all of a consensus TGEKP linker sequence, suggesting a role as a DNA-binding protein (32,33). The CBFA2T3-ZNF652 complex was proposed to repress transcription of genes that have roles in the oncogenesis of breast (31). As an approach to discovering ZNF652 target genes, we have identified the ZNF652 consensus DNA binding sequence and subsequently have determined that the promoter of the HEB gene contains a functional ZNF652 DNA binding motif. We present data that HEB expression is regulated by the CBFA2T3-ZNF652 repressor complex. In addition, we also have determined the minimal interaction domains for ZNF652 and CBFA2T3.
CASTing (Cyclic Amplification and Selection of Targets) protocol-To identify ZNF652 DNA binding sequence by CASTing (34), the bacterially expressed MBP-ZNF652 fusion protein was affinity-purified using amylose-coated resin under nondenaturing conditions. A library of double-stranded DNA (dsDNA) molecules containing 18 degenerate nucleotides flanked by known sequences was generated according to the published method (35). Briefly, random sequence oligonucleotide 5Ј-CAGGTCAGTTCAGCGGATCCTGTCG(N) 18 GAG-GCGAATTCAGTGCAACTGCAGC-3Ј was annealed to 5Ј-GCTGCAGTTGCACTGAATTCGCCTC-3Ј and filled in by DNA polymerase I, large fragment (Klenow) to generate a pool of dsDNA for the binding assay. This pool of dsDNA (250 ng) was incubated with 1 g of affinity-purified MBP-ZNF652 in 1ϫ DNA-binding buffer (20 mM Hepes-KOH, pH 7.9, 100 mM KCl, 2 mM MgCl 2 , 10 M ZnSO 4 , 0.5 mM DTT, 0.2 mM EDTA, 0.1% Nonidet P-40, 10% glycerol, 250 g/ml BSA) in a 50-l final volume at 4°C for 1 h. MBP-ZNF652-DNA complexes were captured on amylose resin, eluted, and then PCR-amplified using the primers 5Ј-GCTGCAGTTG-CACTGAATTCGCCTC-3Ј and 5Ј-CAGGTCAGTTCAGC-GGATCCTGTCG-3Ј with Platinum Taq DNA polymerase High Fidelity (Invitrogen), initially for 2 min at 94°C and then 15 cycles of 94°C for 10 s, 58°C for 10 s, and 68°C for 10 s. The amplified DNA from each round was used for the subsequent round of binding and ZNF652-sequence specific enrichment. Five rounds of selection were performed. DNA amplified from the fifth round of selection was digested with EcoRI and BamHI and cloned at the same restriction enzyme sites in BlueScript vector pKS(ϩ) (Stratagene), and a total of 27 clones were sequenced.
Nuclear Extracts-For nuclear extracts, nuclei from ZR75-1 breast cancer cells (4 ϫ 10 7 ) known to express endogenous ZNF652 (31) were isolated according to the published method (36). Nuclei were resuspended in cold buffer containing 20 mM Hepes-KOH, pH 7.9, 420 mM NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT, and 25% glycerol (37) and sonicated on a Vibra-Cell VCX 130 (Sonics) with three 10-s pulses at 30% amplitude. Nuclear lysates were then centrifuged, and the supernatants were assayed for protein content using a BCA protein assay kit (Pierce). Aliquots were stored at Ϫ80°C until required. Nuclear extracts were also prepared from Jurkat and HEK293T cells transiently expressing HA-ZNF652.
Western Blotting-Western blots carrying input and immunoprecipitated samples were first incubated with primary antibody (rat anti-HA or mouse anti-Myc) followed by the appropriate HRP-conjugated antibodies. Proteins were visualized using the enhanced chemiluminescence kit (ECL, Amersham Biosciences).
Dual-Luciferase Reporter Assays-Approximately 1.75 ϫ 10 5 CHO cells were transiently transfected using Lipofectamine 2000 with 200 ng of pGL2-4ϫZNF652-TK-Luc, constructs expressing varying amounts of ZNF652 and/or Myc-CBFA2T3, and 25 ng of pRL-TK plasmid (Promega) as an internal transfection control. The cells were harvested after 16 h, lysed, and assayed using the Dual-Luciferase reporter assay system (Promega). Luciferase values were normalized to Renilla luciferase activity and expressed as relative light units. For reporter assays on the HEB promoter, HeLa cells (1.5 ϫ 10 5 ) were transfected with 200 ng of luciferase reporter together with varying amounts of constructs expressing Myc-ZNF652 and/or Myc-CBFA2T3 and 50 ng of pRL-TK plasmid (Promega). Cells were harvested and analyzed as described above. All luciferase reporter assays were performed in triplicate and repeated at least three times with the data presented as mean Ϯ S.E. ZNF652 and CBFA2T3 expression was confirmed by Western blot analysis using the appropriate primary and secondary antibodies.
Reverse Transcription Real-time PCR-Total RNA was extracted using an RNeasy mini kit (Qiagen) along with On-Column RNase-free DNase digestion. cDNAs were generated by reverse transcribing the total RNA using oligo(dT) 24 and Moloney murine leukemia virus reverse transcriptase (H Ϫ ) (Promega). Real-time PCR was performed on these cDNA samples to measure the levels of HEB transcript using forward (5Ј-GTCACTACTTCAAGCACAGACCTGA-3Ј) and reverse (5Ј-GCGTTCTCTGGCATTGTTAGCCAT-3Ј) primers and of the housekeeping gene cyclophilin A using forward (5Ј-GGC-AAATGCTGGACCCAACACAAA-3Ј) and reverse (5Ј-CTA-GGCATGGGAGGGAACAAGGAA-3Ј) primers. Real-time PCRs were carried out using iQ SYBR Green Supermix (Bio-Rad) on an iQ real-time PCR detection system under the following conditions: 95°C for 3 min; 40 cycles of 94°C for 15 s, 60°C for 15 s, and 72°C for 30 s. HEB expression was normalized against the levels of cyclophilin A.
Promoter Precipitation-Wild type and mutant HEB promoter fragments were PCR-amplified from the pGL3-HEB-E-Luc and pGL3-HEB-mut-E-Luc constructs, respectively, using HEB-F (5Ј-TGCTTCTCGAGGTGCCTGCGAGCGC-CTCAT-3Ј) and the biotinylated GLprimer2-R (5Ј-CTTTA-TGTTTTTGGCGTCTTCCA-3Ј) primer, which anneals within the vector sequences. PCR products were gel-purified and immobilized on streptavidin-coated M-280 Dynabeads. HuT78 total cell extracts were used for the endogenous ZNF652 and CBFA2T3 binding assay. For the second assay, total cell extracts from HEK293T cells ectopically expressing the FLAG-ZNF652 and Myc-CBFA2T3 were used. Uncharged beads or beads conjugated with the HEB promoter fragment carrying the wild type or mutant ZNF652 DNA binding JULY 4, 2008 • VOLUME 283 • NUMBER 27 sequence were incubated with the appropriate cell extracts in binding buffer (20 mM Tris-HCl, pH 8.0, 10% glycerol, 6 mM MgCl 2 , 5 mM DTT, 0.1 mM EDTA, 0.01% Nonidet P-40) (20) for 2 h at room temperature and then washed three times with the same buffer. The proteins bound to the immobilized DNA were recovered by incubating the beads with 10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 1 M NaCl buffer, resolved on SDS-PAGE, and analyzed by Western blotting using the appropriate primary and secondary antibodies.
siRNA-mediated ZNF652 Knockdown-MCF7 breast cancer cells were transfected with either ZNF652-specific (target sequences, 5Ј-GUAGAGAAAGUCAGCGUUA-3Ј or 5Ј-GAG-AAGCACAUGAACGUUA-3Ј (Dharmacon RNA Technologies)) or scrambled control (Qiagen) siRNA at 100 nM using lipitoid transfection reagent according to the published method (40). Cells were collected 48 h after transfection and analyzed for levels of ZNF652 and HEB protein using Western blotting. Protein loading was confirmed by probing with anti-␤-actin antibody.

Identification and Validation of ZNF652 Consensus DNA
Binding Sequence-The binding sequence for ZNF652 was identified using a CASTing protocol (34). Bacterially expressed MBP-ZNF652 fusion protein was used in five sequential rounds of binding and elution cycles to enrich for ZNF652 binding sequences from a random pool of dsDNA. This random pool was generated by annealing two complementary oligonucleotides with 18 degenerate bases (see "Experimental Procedures"). The enrichment of the ZNF652 DNA binding sequence was monitored by performing EMSA on equal amounts of the pools of DNA amplified from rounds 3 to 5 of selection (   pools sequentially increased with the three successive cycles of selection (Fig. 1A, compare lanes 6, 9, and 12). No such binding was detected with the MBP protein alone (Fig. 1A, lanes 13-15). PCR products from the fifth round of selection were cloned and sequenced. Sequence alignment of 27 individual clones revealed the consensus binding sequence CGAAAGGGT-TAAT (Fig. 1B).

C G A A A G G G T T A A T Consensus
To confirm that MBP-ZNF652 specifically binds to this consensus sequence, complementary oligonucleotides with wild type or mutant ZNF652 DNA binding sequences were designed. These oligonucleotides were annealed to generate short DNA fragments and used in EMSA. Whereas MBP-ZNF652 bound to the short dsDNA carrying the wild type sequence, no such binding was detected with the mutated sequence ( Fig. 2A). In addition, binding to the wild type probe decreased in correspondence to the presence of increasing amounts of unlabeled wild type probe DNA ( Fig. 2A, lanes 4 -6). However, no change in the binding to the wild type probe was observed in the presence of increasing amounts of unlabeled mutant probe DNA ( Fig. 2A,  lanes 9 -11). MBP protein alone did not bind to the wild type probe (Fig.  2A, lane 2). In a separate EMSA, both the recombinant MBP-ZNF652 and endogenous-ZNF652 proteins (from ZR75-1 breast cancer cells) were observed to bind to the wild type probe. The specificity of this binding was evident as both the MBP-ZNF652-DNA and endogenous ZNF652-DNA complexes supershifted in the presence of an anti-ZNF652 antibody (Fig. 2B). However, such a supershift was not observed in the presence of a nonspecific antibody. These results show that the ZNF652 consensus DNA binding sequence identified from the CASTing protocol binds the ZNF652 protein.
To further confirm that endogenous mammalian nuclear ZNF652 binds to the identified consensus sequence, competitive EMSA experiments were performed using ZR75-1 nuclear extracts (Fig. 2C). These results were similar to those obtained using the bacterially expressed MBP-ZNF652 fusion protein ( Fig. 2A). Finally, the specificity of nuclear ZNF652 protein binding to the consensus ZNF652 DNA sequence was confirmed by supershift EMSA using nuclear extracts of ZR75-1 and HEK293T cells transiently expressing HA-ZNF652 (Fig.  2D). ZNF652-DNA complexes underwent a supershift in the presence of an anti-ZNF652 antibody but not in the presence of a nonspecific antibody (Fig. 2D, compare lane 3 with 5 and 9 A. with 11). Nuclear extracts of the breast cancer cell line MCF7 and leukemic cell lines KU-812 and K-562 were also seen to bind specifically to the wild type ZNF652 probe (data not shown). Taken together, these results show that the identified ZNF652 DNA binding sequence (hereafter called ZNF652 response element (RE)) is biologically authentic. Therefore, we hypothesized that genes containing the ZNF652 RE in their promoters would be transcriptionally repressed by the CBFA2T3-ZNF652 complex. The CBFA2T3-ZNF652 Complex Mediates Transcriptional Repression through a ZNF652 RE-Dual-Luciferase assays were used to determine whether the CBFA2T3-ZNF652 interaction complex was able to mediate transcriptional repression through the identified ZNF652 RE. For these assays, the pGL2-4ϫZNF652-TK-Luc construct carrying four tandem copies of the ZNF652 RE was used. Luciferase activity of the pGL2-4ϫZNF652-TK-Luc reporter was considered as the basal transcriptional activity. Reporter assays showed a dose-dependent decrease in transcription of the luciferase gene in response to increasing levels of ZNF652; this repression was enhanced in the presence of increasing levels of CBFA2T3 (Fig. 3). However, expression of CBFA2T3 alone had no significant effect on luciferase activity from the pGL2-4ϫZNF652-TK-Luc reporter. These results show that ZNF652 and CBFA2T3 act as functional corepressors of transcription.

HEB (TCF12) Is a Target of CBFA2T3-ZNF652-mediated Transcriptional Regulation-The
Eukaryotic Promoter Database (EPD release 84) was used to search for gene promoters that contain the consensus ZNF652 RE and thus are potentially regulated by ZNF652. The human HEB promoter contains a putative ZNF652 RE (AAGGGTTAA) located at position Ϫ306 upstream of the translational start site that is conserved in the mouse HEB promoter (located at position Ϫ332).
Reporter assays using a construct in which this HEB promoter sequence drives the luciferase gene (pGL3-HEB-E-Luc) showed a dose-dependent transcriptional repression in the presence of ectopically expressed ZNF652. Ectopic expression of CBFA2T3 further increased this repression (Fig. 4). ZNF652mediated repression was abrogated in the construct pGL3-HEB-mut-E-Luc carrying the mutated ZNF652 RE (Fig. 4). These results suggest that ZNF652 protein binds the ZNF652 RE within the HEB promoter sequence to induce transcriptional repression, which is enhanced in the presence of the CBFA2T3 corepressor.
The CBFA2T3-ZNF652 Complex Specifically Binds to the ZNF652 RE within the HEB Promoter Region-A polyclonal rabbit anti-CBFA2T3 antibody was raised and affinity-purified. This antibody was seen to specifically detect the CBFA2T3 protein from the HEK293T cells ectopically expressing Myc-CBFA2T1, Myc-CBFA2T2, or Myc-CBFA2T3 protein (  5A). Anti-CBFA2T3 antibody could also specifically detect the endogenous CBFA2T3 protein from MCF7 and Jurkat nuclear extracts (Fig. 5B).
In vitro promoter binding assays were used to show the binding of CBFA2T3-ZNF652 to the single ZNF652 RE within the HEB promoter. Firstly, promoter binding assays were performed using protein lysates from HEK293T cells exogenously expressing FLAG-ZNF652 and Myc-CBFA2T3. Both ZNF652 and CBFA2T3 bound to the HEB promoter carrying the wild type ZNF652 RE but did not bind to the HEB promoter when  ). B, affinity-purified rabbit anti-CBFA2T3 polyclonal antibody specifically detects the endogenous CBFA2T3 protein. Nuclear extracts of MCF7 (9 g) and Jurkat (6 g) cells were Western blotted with anti-CBFA2T3 antibody. Lysate from HEK293T cells ectopically expressing Myc-CBFA2T3 was used as a positive control. C, exogenously expressed ZNF652 and CBFA2T3 proteins bind to the ZNF652 RE located within the HEB promoter. WT and MT HEB promoter sequences were PCR-amplified from the appropriate reporter constructs using a pair of primers (one of which was biotinylated at the 5Ј-end) and immobilized to streptavidin-coated magnetic beads. Magnetic beads charged with promoter sequences (lanes 2 and 3)

CBFA2T3-ZNF652 Mediates Transcriptional Repression of HEB
this RE was mutated (Fig. 5C). Secondly, promoter binding assays were performed using protein lysates from the human T-cell line that expresses both endogenous ZNF652 and CBFA2T3. Endogenous CBFA2T3 and ZNF652 also specifically bound to the HEB promoter carrying the wild type ZNF652 RE. However, no such binding was observed with the HEB promoter containing a mutated ZNF652 RE (Fig. 5D). Furthermore, the stoichiometric level of ZNF652 binding was higher than that of CBFA2T3. Finally, ChIP assay was used to confirm in vivo binding of ZNF652 to its cognate RE located within the HEB promoter region (Fig. 5E). PCR data showed that the HEB promoter sequence was enriched with an anti-ZNF652 antibody but not with rabbit IgG. Negative control ␤-globin sequences were also not enriched by either anti-ZNF652 antibody or control IgG (Fig. 5E). These results suggested that ZNF652 binds to the HEB promoter in vivo. As CBFA2T3 does not directly bind DNA, these assays confirmed that ZNF652 binds to the single wild type ZNF652 RE within the HEB promoter region and is associated with CBFA2T3 (31). HEB Is a Direct Transcriptional Target of ZNF652-To further confirm that ZNF652 mediates transcriptional repression of HEB in vivo, HEK293T cells ectopically expressing FLAG-ZNF652 alone or with Myc-CBFA2T3 were used. The relative levels of HEB transcription were assayed using real-time reverse transcription-PCR. The HEK293T cells expressing exogenous ZNF652 had a reduced level of HEB expression compared with the empty vector-transfected HEK293T cells (Fig. 6A). ZNF652mediated repression of HEB was enhanced in the presence of CBFA2T3. Exogenous expression of CBFA2T3 alone in the HEK293T cells also resulted in a minimal repression of HEB, an observation that can be attributed to the presence of endogenous ZNF652 (Fig.  6A, lane 4). Similar to the transcript levels, HEB protein levels were also repressed in HEK293T cells ectopically expressing either ZNF652 or CBFA2T3 or both (Fig. 6A). These results show that ZNF652 mediates transcriptional repression of HEB in vivo and that this repression is enhanced in the presence of CBFA2T3.
To further demonstrate that ZNF652 represses HEB transcription, the effect of knockdown of ZNF652 expression on levels of HEB protein was investigated. MCF7 cells were transfected with either scrambled or two different ZNF652-specific siRNAs, and HEB and ZNF652 protein levels were determined subsequently by Western blot analysis. As predicted, higher levels of HEB protein were detected in cells with ablated ZNF652 expression, but no such increase was detected in the presence of scrambled siRNA (Fig. 6B).
NHR3 and NHR4 Motifs of CBFA2T3 Are Both Required for Interaction of CBFA2T3 with ZNF652-We have shown previously that CBFA2T3 interacts with ZNF652 to repress transcription (31). To more precisely define the regions of CBFA2T3 that interacted with ZNF652, various fragments of CBFA2T3 were used in co-immunoprecipitation assays (Fig.  7A). Full-length CBFA2T3 and the fragment CBFA2T3-3, containing both the NHR3 and NHR4 domains, interacted with ZNF652 (Fig. 7B, lanes 9 and 11), whereas interaction was not observed with the fragments CBFA2T3-1 (containing NHR2 alone), CBFA2T3-2 (containing NHR3 alone) and CBFA2T3-4 (containing NHR4 alone) (Fig. 7B, lanes 7, 8, and 10). These results suggest that both the NHR3 and NHR4 domains of CBFA2T3 are required for interaction with ZNF652.
ZNF652 Interacts with the CBFA2T3 via Its C-terminal Region-Immunoprecipitation experiments were undertaken to define the region of ZNF652 that interacts with the CBFA2T3-3. Immunoprecipitations on HEK293T cells expressing CBFA2T3-3 and either full-length or two different C-terminal fragments of ZNF652 were performed (Fig. 8A). ZNF652-2 weakly interacted with CBFA2T3-3, but no such interaction was detected with ZNF652-4 (Fig. 8B), suggesting  that CBFA2T3-ZNF652 interaction is stabilized in the presence of the zinc finger region (31). These results further confirm that ZNF652 interacts with CBFA2T3 via the NHR3 and NHR4 domains. ZNF652 Interacts with CBFA2T3 via a Proline-rich Stretch within its C-terminal Region-Recently it was demonstrated that a proline-rich motif of N-CoR/SMRT has high affinity for the MYND domain located within the NHR4 motif of CBFA2T1. This finding was consistent with the data on the solution structure of the NHR4-SMRT/N-CoR peptide complex (12) and an earlier report showing that NHR3-4 domains of CBFA2T1 interact with N-CoR repression domain III, a region of N-CoR that contains a conserved PPLXP binding motif for corepressor proteins (7). We aligned the amino acid sequences of ZNF652, N-CoR, and SMRT (Fig. 9A). We reasoned that CBFA2T3 interacts with ZNF652 through a similar proline-rich region (amino acids 565-574) within the C terminus of ZNF652. To test this possibility, we determined the interaction of ZNF652 with either PPVP mutated to AAAA (amino acids 565-568; HA-ZNF652-mut1) or PPPP mutated to AAAA (amino acids 571-574; HA-ZNF652-mut2) and CBFA2T3 using co-immunoprecipitation assays. Consistent with this premise, neither of the ZNF652 mutants interacted with the CBFA2T3 protein, suggesting that a proline-rich region located within the C terminus of ZNF652 (amino acids 565-574) is the minimal region of interaction with CBFA2T3 (Fig. 9B). To confirm that the two ZNF652 mutant proteins retained their three-dimensional structure, we determined their ability to bind the ZNF652 RE within the HEB promoter using promoter precipitation assays. Both the wild type and mutant ZNF652 proteins were able to bind the ZNF652 RE, indicating that alanine substitutions do not impair the structure or DNA binding ability of the ZNF652 mutants (Fig. 9C). Taken together, these results suggest that a C-terminally located proline-rich region of ZNF652 interacts with the NHR3-NHR4 domain of CBFA2T3.

DISCUSSION
DNA-binding activator or repressor proteins play a critical role in regulation of gene expression. Dysregulation of their functional activity can lead to a number of malignancies, including breast cancer (1). The identification of transcription factor binding sequences is necessary to identify target genes and determine their precise biological function.
ETO proteins do not directly bind DNA but perform their repressive role through interaction with transcription factors that are capable of binding their cognate response elements located within promoter sequences of the target genes. The ETO family of proteins has been shown to interact with the zinc finger proteins Gfi-1, BCL6, PLZF, and GATA-1, as well as with transcription factors such as HEB and SCL/TAL-1 (stem cell leukemia/T-cell acute lymphocytic leukemia-1), to repress gene expression. In addition, we have shown that ZNF652 is a novel ETO-interacting protein that functions as a DNA-binding transcription factor and provides further complexity to the ETO family of repressor complexes (31). The major established function of these ETO-based complexes is in hematopoiesis, with dysregulation leading to leukemia (41). However, the expression of the ETO proteins and their interacting DNA-binding transcription factors are not limited to the hematopoietic lineages, as more diverse roles are currently emerging. For example, the CBFA2T3-ZNF652 complex has a role in suppressing breast oncogenesis (31), and the BCL6 oncoprotein is implicated in the pathogenesis of B-cell lymphomas (42). In addition, BCL6 is expressed in a higher proportion of clinically aggressive breast tumors and has been shown to prevent mammary epithelial differentiation (43).
The novel classical zinc finger protein ZNF652 specifically interacts with the breast tumor suppressor CBFA2T3 (31). This is a transcriptional repressor, because a GAL4-ZNF652 fusion protein, when tethered to DNA via a GAL4 DNA binding sequence, represses transcription in reporter assays, and this repression is enhanced in the presence of the CBFA2T3 protein (31). To assist the identification of gene targets of the CBFA2T3-ZNF652 complex, CGAAAGGGTTAAT was identified as the ZNF652 consensus DNA binding sequence (Fig.  1B). The authenticity of the identified ZNF652 DNA binding sequence is supported by multiple assays including normal, competitive and supershift EMSAs using purified recombinant MBP-ZNF652 and endogenous and ectopically expressed mammalian ZNF652 proteins (Fig. 2). Furthermore, CBFA2T3 and ZNF652 form a transcriptional repressor complex on this ZNF652 RE (Fig. 3). Our in vivo assays show that ZNF652 represses transcription of HEB by binding to a single ZNF652 RE within the HEB promoter sequence, and this repression is enhanced in the presence of CBFA2T3 (Fig. 4). This result was further supported by our in vivo and in vitro promoter binding assays showing that the CBFA2T3-ZNF652 complex is recruited to the HEB promoter (Fig. 5). Furthermore, ZNF652 ablation resulted in an increased level of HEB expression (Fig. 6). Based on these findings, it is proposed that CBFA2T3 and ZNF652 form a complex on the HEB promoter, subsequently recruiting corepressors and HDACs and resulting in repression of transcription.
Our results show that HEB is a direct target of CBFA2T3-ZNF652-mediated transcriptional repression. It is of interest that all of the ETO family proteins can interact directly with HEB to generate transcription complexes and modulate the transcriptional activation function of HEB (8). The three-dimensional structure of the CBFA2T1 NHR1 domain interacting with a 17-amino acid peptide derived from the HEB activation domain-1 (AD1) was reported recently (44). As the NHR1 domains of the ETO proteins are highly conserved, it is predicted that both CBFA2T2 and CBFA2T3 would also interact directly with HEB through their NHR1 domain. NHR1-mediated CBFA2T1 interaction with HEB activation domain-1 inhibits its transcriptional activation function by blocking the p300/ CBP (CREB-binding protein) occupancy of HEB and promoting the recruitment of HDAC containing complexes (8).
HEB is involved in both hematopoiesis and myogenesis (45,46) and is associated with various malignancies including gliomas (47). HEB, along with E2-2, E12, and E47, belongs to a family of evolutionarily conserved basic helix-loop-helix proteins called E-box transcription factors. HEB and E2-2 proteins are encoded by unique genes, whereas E12 and E47 result from differential splicing of the E2A gene transcript. HEB (along with E2A) plays an essential role by positively regulating a number of target genes critical in B-and T-cell lineage differentiation and development (48). Systematic gene replacement experiments show that signals leading to transcriptional activation of HEB versus E2A are crucial for B-and T-cell lineage commitment and differentiation (49). Our results show that the CBFA2T3-ZNF652 complex represses HEB expression. Therefore, we predict that CBFA2T3-ZNF652-mediated HEB repression may determine the relative levels of HEB and thus the ratio of E2A-E2A homodimers to E2A-HEB heterodimers, critical in B-and T-cell development.
CBFA2T3 performs its transcriptional repressor function in erythropoiesis as a component of a multiprotein complex. Within this complex, CBFA2T3 associates with the hematopoietic-specific E-box SCL protein and a number of other transcription factors (for example, HEB, GATA-1, E2A, E12/47, E2.2, Lbd-1, LMO2, and SSDP2) (20,21). The stoichiometry of CBFA2T3 relative to the level of SCL, E2A, or HEB activators within this multiprotein complex varies during erythroid differentiation, and levels of CBFA2T3 determine both the ability to activate or repress transcription of essential SCL target genes and the timing of their expression (20). Therefore, it is evident that HEB has a complex relationship with CBFA2T3, as the CBFA2T3-ZNF652 transcriptional complex represses HEB A.  1-6), and immunoprecipitates (lanes 7-12) were Western blotted (WB) with anti-HA (upper panel) or anti-Myc (lower panel) antibodies. C, wild type and mutated ZNF652 proteins retain the ability to bind the ZNF652 RE within the HEB promoter. Streptavidin beads conjugated to the biotinylated PCR product carrying the wild type HEB promoter sequence were prepared as described for Fig. 5C. Magnetic beads charged with the HEB promoter sequence (lanes 4 -6) or uncharged beads (lane 7) were incubated with the total cell lysates from HEK393T cells transfected with constructs expressing HA-tagged wild type ZNF652 (lanes 4 and 7), ZNF652-mut1 (lane 5), or ZNF652-mut2 (lane 6) and washed, and bound proteins were eluted with 1 M NaCl. Inputs (lanes 1-3) and promoter precipitates (lanes 4 -6) were analyzed by Western blotting with anti-HA antibody.
expression, and the CBFA2T3 protein directly interacts with the HEB protein to inhibit its activator function. Although CBFA2T3 does not interact with the ZNF652 zinc finger region (amino acids 243-491), this region is required in order to stabilize the interaction of the C-terminal region of ZNF652 (amino acids 498 -606) with CBFA2T3 (31). Findings presented in this study define the NHR3 and NHR4 domains of CBFA2T3 as the regions that interact with ZNF652 (Fig. 7). It has been shown for CBFA2T1 that both the NHR3 and NHR4 domains were also required for its interaction with the corepressor N-CoR (7,50). Additional studies have shown that CBFA2T1 interacts through a conserved proline-rich PPLXP motif within N-CoR, and reporter assays confirmed that this interaction is functionally important (7). A number of proteins have been reported to interact with MYND domains through their PXLXP peptide motif. MYND domains are defined by a C 4 -C 2 HC consensus and are frequently implicated in transcriptional repression (12,51). Therefore, we predict that the MYND domain of CBFA2T3 interacts with ZNF652 through a prolinerich PXLXP motif (amino acids 565-574; PPVPHLPPPP) located within the C-terminal region of ZNF652 (Fig. 9). In summary, both the CBFA2T3-ZNF652 and CBFA2T1-N-CoR interactions are mediated through the NHR3 and NHR4 regions. This interaction occurs through the MYND-type domain located within the NHR4 and respective proline-rich PXLXP motifs of either ZNF652 or N-CoR. It is of importance to note that SMRT lacks a PXLXP motif but interacts with MYND through a PPPLI sequence (12). This suggests that variable proline-rich sequences can interact with MYND domains; further work will be required to define precisely the functional residues within these proline-rich motifs.
ETOs have a modular structure due to the presence of conserved NHR domains that facilitate multiple complex protein interactions (41). ETOs perform their transcriptional repressor role by interacting with various DNA-binding transcription factors. Each ETO-interacting DNA-binding transcription factor interacts through a specific region within the ETO proteins (Fig. 10). The interaction of CBFA2T3 with ZNF652 occurs through its NHR3-NHR4 domains. This may reflect a unique functional relationship between the CBFA2T3 and ZNF652 proteins, as other known ETO-interacting zinc finger tran-scription factors interact through the NHR1-NHR2 domains (8,26,27,30,31) (Fig. 10).
We are currently working toward genome-wide identification of ZNF652 gene targets, particularly ZNF652-responsive genes directly associated with breast oncogenesis, using a ChIP-ChIP approach. Such investigations are critical in determining how ZNF652 regulates normal gene expression in cell fate, development, and differentiation and how ZNF652-mediated transcriptional alterations of target genes can lead to cell proliferation and/or apoptosis.