Yin Yang 1 Physically Interacts with Hoxa11 and Represses Hoxa11-dependent Transcription*

Yin Yang 1 (YY1) plays an indispensable role in embryonic development. YY1 contains an evolutionarily conserved, 22-amino acid segment, the PHO homology region (PHR), which is located within its central domain (spacer) and has been shown previously to participate in the recruitment of Polycomb group of proteins and in YY1-mediated transcription. In this report, we show that the PHR physically interacts with several Abd-B-type Hox proteins. Although ectopic expression of Hoxa11 enhanced target promoter activity, overexpression of YY1 repressed this effect, which was abrogated by YY1 siRNA and the histone deacetylase inhibitor trichostatin A. We have further demonstrated that this suppression effect was the result of YY1-dependent recruitment of HDAC2 to the Hoxa11 target promoter. Taken together, our findings show that YY1 represses Hoxa11-dependent transcription via interactions with the Hox proteins and HDAC recruitment, providing a link between an Abd-type Hox protein and a Polycomb group protein at the level of direct protein-protein interactions. These findings not only provide a novel insight into YY1 function but also identify a new regulation of homeotic protein-mediated transcriptional regulation in general.

supported by recent findings that mammalian YY1 also functions as a PcG protein (8,9). YY1 is the only known PcG protein containing a zinc finger domain that has the sequence-specific DNA binding capability that is conserved in PHO. It has been demonstrated that PHO binds a 17-base pair (bp) PRE sequence in the Drosophila engrailed gene (5). These early studies in Drosophila suggest an important role of YY1 in regulating homeotic gene transcription.
Although YY1 is evolutionarily conserved from Drosophila to human, the sequence homology between YY1 and PHO is limited to the DNA-binding zinc fingers and a 22-aa segment located at the central region of the protein (aa 205-226, PHO homology region (PHR)) (5). The spacer region (aa 205-295) that encompasses PHR has been shown to play a role in YY1mediated transcriptional regulation (9). In addition, a recent study suggests that PHR may mediate physical interactions between PHO and other PcG proteins (10). A human YY1 with the spacer deleted has been shown to possess reduced transcriptional activity in vitro (11). Furthermore, removal of the spacer region in chicken DT40 cells abrogated the ability of cYY1 to support DT40 cell survival in vivo, indicating that the spacer region is required for YY1 biological functions (3). However molecular mechanisms that underlie the spacer and PHR are incompletely understood. To address the role of the YY1 spacer region, we performed a yeast two-hybrid assay using the spacer region as bait. We identified novel physical interactions between YY1 and the Abd-B-type Hox proteins and showed that the physical interactions are important for YY1 to negatively regulate Hox protein-dependent transcription in a manner that is dependent on the recruitment of histone deacetylases. Taken together, these findings suggest that YY1 not only functions as a PcG protein to directly repress Hox gene transcription but also participates in negatively regulating transcription mediated by the products of the Hox genes, i.e. via direct physical interactions of the homeodomain proteins.

MATERIALS AND METHODS
Yeast Two-hybrid Assay-The SalI-BglII cDNA fragment encoding amino acids 205-226 of human yy1 was cloned downstream of the GAL4 DNA-binding domain into pPC98 vector and used as bait. An 8.5-day mouse embryonic cDNA library was produced using pPC86 vector (Resgen, Invitrogen). The yeast strain MaV103 (MAT␣, gal4, gal80, leu2, trp1, his3, SPAL10::URA3) (12) was co-transformed with pPC98/hYY1-(205-226) and the cDNA library plasmids. Transformants were plated on synthetic complete media plates lacking histidine, leucine, and tryptophan in the presence of 12.5 mM 3-aminotriazole. It produced 5.8 ϫ 10 6 transformation efficiency. * This work was supported by Grant GM53874 from the National Institutes of Health (to Y. S.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 1  An X-gal filter assay was performed after the yeast two-hybrid assay for confirmation. Yeast colonies from minimal plates (ϪLeu, ϪTrp, ϪHis) were lifted onto nitrocellulose mem-branes, immersed in liquid nitrogen for 5 s, and then transferred onto Whatman filter papers saturated with 3 ml of Z-buffer (100 mM sodium phosphate, pH 7.0, 10 mM KCl, 1 mM MgSO 4 , 40 mM ␤-mercaptoethanol) containing 1 mg/ml X-gal. The color of the colonies was verified after putting the membranes at 30°C for 1 h.
Cell Culture, Transfection, GST Pulldown, Immunoblot, and Immunoprecipitation-F9 cells and HeLa cells were cultured in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. F9 cells were induced to differentiate by retinoic acid for 4 days. HeLa cells were transfected using Lipofectamine 2000 (Invitrogen), and immunoprecipitation was achieved as described previously (15). For GST binding assays, in vitro translated Hoxa11 and Hoxa11 lacking the homeodomain mutant products were allowed to interact with either GST or GST-YY1. The procedure for GST pulldown assays was described previously (16). Anti-Hoxa11 antibodies were generously provided by Daniela Bomgardner (17), and anti-HA (H.10) and anti-FLAG (F3165) antibodies were purchased from Babco and Sigma. Anti-YY1 (H414), anti-HDAC1 (H-51), and anti-HDAC2 (H-54) antibodies were from Santa Cruz Biotechnology.
Electrophoretic Mobility Shift Assay-Probes containing Hox and non-Hox consensus binding sites were used in this study  (18). The sense strands of complementary oligonucleotides for Hox binding sequences was 5Ј-TCGAGCCAGATCTGACAG-TTTTACGACAGATCTCCC-3Ј and for non-Hox binding site sequences was 5Ј-TCGAGCCAGATCTGACAGTTTGATG-ACAGATCTCCC-3Ј. Both were end-labeled and annealed overnight with their corresponding antisense oligonucleotides to generate double-stranded DNA probes. His-YY1 and His-YY1(⌬PHR) fusion proteins were prepared and purified as instructed by Qiagen. GST and GST/Hoxa11 proteins were purified using Glutathione-Agarose (G-4510) from Sigma. Labeled probes were incubated individually with various fusion proteins in the presence of 25 mM Hepes (pH 7.91), 65 mM KCl, 1 mM dithiothreitol, 3 mM MgCl 2 , 12% glycerol, 2 g of bovine serum albumin, and 0.5 g of dI-dC as nonspecific competitor. Reaction mixtures were incubated at 4°C for 1 h and applied to a 10% polyacrylamide gel. Gel electrophoresis was performed using 0.5ϫ Tris borate-EDTA buffer.
For histone deacetylases (HDAC) inhibitor analysis, 400 ng of Hoxa11 was co-transfected with 100 ng of HBS-Luc reporter in HeLa cells, treated with 100 ng of TSA in complete Dulbecco's modified Eagle's medium, and incubated for 18 h. In the YY1 knockdown experiment, 400 ng Hoxa11 was transfected with 300 ng of U6 vector-YY1 siRNA and 100 ng of HBS-Luc reporter in HeLa cells. Cell lysates were analyzed 48 h after transfection. Results were all normalized by the corresponding ␤-gal activity.

YY1 Interacts with Abd-B-type
Hox Proteins-Human YY1 contains a 22-amino acid fragment (aa 205-226) in the spacer region (aa 205-295) that is Ͼ80% conserved in Drosophila PHO (5). To better characterize the function of the PHR in embryogenesis, we performed a yeast two-hybrid screen of an 8.5 day post coitum mouse embryo library. Amino acids 205-226 (PHR) of YY1 was used as bait (Fig.  1A). We selected color changes via X-gal filter assays. Three independent cDNAs were identified, and they encoded portions of the homeodomain proteins with a high degree of homology, i.e. Hoxa9, Hoxd9, and Hoxa11 (19,20). These proteins are closely related to the products of the abdominal-B (Abd-B) HOM-C genes in Drosophila, which are situated at the 5Ј-end of the mammalian Hox clusters. Overlapping expression of these genes was found in developing limbs, indicating their specific regional identity (21,22). YY1 PHR Binds Hoxa11 Homeobox Domain in Vitro and in Vivo-To confirm the yeast two-hybrid results and to characterize the Hoxa11 interaction domain of YY1, we carried out a GST binding assay using in vitro translated products, fulllength Hoxa11 and Hoxa11 homeodomain (⌬HD) deletion mutant (Fig. 1B). Comparable amounts of GST and GST-YY1 were used in the assays (data not shown). As shown in Fig. 1C, only full-length Hoxa11 directly interacted with YY1 (lane 3) whereas Hoxa11 ⌬HD was unable to bind (lane 6), indicating that the DNA binding domain of Hoxa11 is essential for the direct interaction with YY1.
In co-immunoprecipitation (co-IP) experiments, FLAGtagged Hoxa11 was co-transfected with HA-tagged YY1 protein into HeLa cells. Cell lysates were immunoprecipitated with FIGURE 2. YY1 enhances Hoxa11 DNA binding activity. An electrophoretic mobility shift assay was preformed with fusion proteins and an oligonucleotide containing a consensus core (TTAC) HBS. Binding of Hoxa11 to HBS was specific, as unlabeled HBS probed specifically abrogated its binding (lanes 5 and 7). A supershifted band was observed following the addition of Hoxa11 antibodies (lane 6). Increasing amounts of YY1 or YY1 (⌬PHR) protein were incubated with Hoxa11. A representative gel of three experiments is shown.
The results indicate that YY1 acts as a binding partner of Hoxa11; increasing the amount of YY1 strongly enhanced Hoxa11 binding to HBS, whereas mutant YY1 (⌬PHR) only enhanced Hoxa11 binding slightly (lanes 8 -11).
an anti-FLAG antibody and Western blotted with an anti-HA antibody.
As shown in Fig. 1D, YY1 PHR physically interacts with Hoxa11 (Fig. 1D, lane 4). This binding is specific as no interaction was found when Hoxa11, YY1 or YY1 (⌬PHR) were expressed alone (Fig. 1D, lanes 1-3). Importantly, Hoxa11 cotransfected with the mutant YY1 deleted of PHR (⌬PHR) also failed to support physical interaction, indicating that PHR is necessary for YY1 to interact with Hoxa11 (Fig. 1D, lanes 5). Similarly, a reciprocal co-IP assay was performed using an anti-HA antibody for the YY1 protein, and Western blotted with an anti-FLAG antibody for the detection of Hoxa11 protein, and essentially the same result was obtained (data not shown).
The interaction between Hoxa11 and YY1 was also detected in the F9 embryonic carcinoma cell line where Hoxa11 was found to have the highest level of protein expression 4 days post retinoic acid induction (Fig. 1E). In contrast, no Hoxa11 protein was detected if F9 cells remain undifferentiated. Both differentiated and undifferentiated F9 cell lysates were immunoprecipitated with an anti-Hoxa11 antibody and Western blotted with an anti-YY1 antibody. As shown in Fig. 1E, endogenous YY1 was pull down in F9 differentiated but not undifferentiated cells, confirming that endogenous YY1 physically interacts with Hoxa11 in vivo.
YY1 Enhances Hoxa11 DNA Binding-The ability of YY1 to interact with Hoxa11 protein suggests that YY1 may influence Hox protein activity via protein-protein interactions. Because cofactors are generally required for Hox proteins to bind their downstream targeting genes (23), we next asked whether YY1 interaction with Hoxa11 affects Hoxa11 DNA binding activity. All Abd-like Hox proteins preferentially bind DNA sequences containing a TTAC core recognition sequence (18). We performed an electrophoretic mobility shift assay based on this Hox binding sequence. As shown in Fig. 2, binding of Hoxa11 to HBS is specific as the unlabeled HBS probed specifically abrogated its binding (lanes 5 and  7). Furthermore, a supershifted band was observed with the addition of the Hoxa11 antibodies (lane 6). No binding was detected for the GST or YY1 protein (lanes 2 and  12). Interestingly, compared with Hoxa11 alone, increasing amount of YY1 strongly enhanced Hoxa11 binding to HBS, whereas the mutant YY1 (⌬PHR) only enhanced Hoxa11 binding slightly (Fig. 2, lanes 5, 8, 9, 10 and 11). These results suggest that YY1 enhances Hoxa11 binding to its DNA sequences via physical interactions mediated by PHR.
YY1 Interacts with Hoxa11 and Represses Transcriptional Activity-YY1 is capable of activating or repressing transcription by direct association with other transcription factors without necessarily binding to DNA (24). To understand the functional consequence of the YY1 and Hoxa11 protein interaction, we next performed reporter assays using a triple TTAC Hox binding sequence in a luciferase (Luc) reporter (14). As shown in Fig. 3, Hoxa11 activated the HBS-Luc reporter activity (3-fold); however, co-transfection with increasing amounts of YY1 suppressed promoter activities (lanes 5 and 6). Importantly, no repression was observed when mutant YY1 (⌬PHR) was used to replace wild type YY1 (Fig. 3, lanes 7 and 8), suggesting that the YY1 physical interaction with Hoxa11 is likely to be essential for regulating Hoxa11 activity.
Inhibition of Histone Deacetylase Activity and Depletion of YY1 Abrogated the Repression of YY1 to Hoxa11-mediated Transcription-It has been shown that YY1 recruits histone modifiers to mediate transcriptional regulation, and numerous studies have revealed a functional significance of YY1 associa-FIGURE 3. YY1 represses Hoxa11 reporter activity. Increasing amounts of YY1 or YY1 (⌬PHR) were transiently transfected into HeLa cells with Hoxa11 and HBS-Luc reporter. Extracts were prepared 36 h after transfection for luciferase activity assay. Results, normalized with the corresponding ␤-gal activity, are shown as bar graphs with mean Ϯ S.D. from three independent transfections. Control experiments using YY1 or YY1 (⌬PHR) transfected with HBS-Luc reporter showed no effect on reporter activity (lanes 2-3). Hoxa11 co-transfected with increasing amounts of YY1 suppressed promoter activities (lanes 5-6), whereas no repression was observed when mutant YY1 (⌬PHR) was co-transfected with Hoxa11 (lanes 7-8).
tion with HDACs (25)(26)(27). Therefore, we hypothesized that the repression effect observed above might be because of YY1 recruitment of HDACs. To investigate this hypothesis, the histone deacetylase inhibitor TSA was added to HeLa cells cotransfected with Hoxa11 and the HBS-Luc reporter. Significantly, compared with the controls, Hoxa11 activated promoter activity in the presence of TSA by 10-fold (Fig. 4A, lane 5), indicating that inhibition of HDAC activities reversed the suppressive effect of the endogenous YY1. Consistent with the observation that YY1 mediated Hoxa11-dependent transcription, knockdown of YY1 by siRNA in HeLa cells also enhanced Hoxa11 reporter activity (Fig. 4A, lane 4). In contrast, control experiments using the U6 siRNA vector showed no effect on reporter activity (Fig. 4A, lane 3). Thus, these findings suggest that YY1 represses Hoxa11-mediated transcription by recruiting HDACs.
YY1 Is Required for Hoxa11 to Recruit HDAC2-The TSA experiments described above suggest that HDACs might be associated physically with the YY1-Hoxa11 complex. To address this possibility, we transfected FLAG-Hoxa11 into HeLa cells and immunoprecipitated Hoxa11 by the FLAG antibodies followed by Western blotting to detect the presence of HDACs. As shown in Fig. 5, the FLAG antibodies coimmunoprecipitated HDAC2 from cells expressing FLAG-Hoxa11 but not from HeLa cells without FLAG-Hoxa11 expression (Fig. 5, compare lane 2 with lane 1). Strikingly, this interaction is dependent on YY1, as the Hoxa11-HDAC2 interaction was specifically disrupted by YY1 depletion (Fig. 5, lane 4). Unlike HDAC2, we found that Hoxa11 interacted with HDAC1 in the absence of YY1 (Fig. 5, lane 4). Taken together, these findings suggest that YY1 may serve as a bridge for the recruitment of HDAC2 for repression of Hoxa11-dependent transcription.

DISCUSSION
YY1 is essential for cell proliferation and differentiation and has been shown to play an indispensable role in embryonic development (1)(2)(3). YY1 is a PcG protein (8), and the PHO homology region (PHR, aa 205-226) has been shown to interact with other PcG proteins such as E(z) (10). A main function of the PcG proteins is to mediate repression of Hox gene transcription during development (28,29). Consistently, YY1 has been shown to repress homeotic genes during development (8,9). In this report, we show that PHR also mediates YY1 interactions with the products of the Abd-B-type Hox genes. This interaction results in the enhancement of Hoxa11 DNA binding activity and repression of Hoxa11-mediated transcription. Importantly, YY1 represses Hoxa11-mediated downstream target gene transcription in a manner that is dependent on its physical interactions with Hoxa11 and the recruitment of HDACs. Our findings suggest that YY1 as a PcG protein not only directly represses Hox gene transcription but also inhibits the transcriptional activities of the Hox gene products such as Hoxa11. This reveals a new layer of regulation that involves PcG and the homeodomain proteins. Because the physical interaction between YY1 and Hoxa11 appears to be mediated by the conserved PHR, our findings further suggest that the negative regulation conferred by YY1 through protein-protein interaction is also evolutionarily conserved from Drosophila to human.
The role of Hox proteins as activators or repressors is contextdependent and may involve differential recruitment of coregulators (30). Previous published studies show that some Hox proteins mediate transcriptional repression via direct interactions with HDACs (30 -33). Our findings have extended these earlier studies and identified YY1 as a possible bridging protein between Hox proteins and HDACs. YY1 transcriptional repression activity has been shown to be mediated by its association with HDACs via aa residues 170 -200 (GK-rich) and the C-terminal zinc finger domain (34). Indeed, when the zinc finger region-deleted mutant of YY1 was co-transfected with Hoxa11, it failed to suppress HBS-Luc reporter expression, consistent with the idea that recruitment of HDACs is important for repression (data not shown). RNA interference (RNAi) knockdown of YY1 did not result in the same level of derepression of the target gene as it did in the TSA treatment. It is possible that this may be due to an incomplete knockdown of YY1 or an involvement of additional histone deacetylases such as HDAC1 in the regulation.
The Abd-B-type Hox genes are situated at the 5Ј-end of the Hox clusters. Only one Abd-B gene is found in Drosophila, but a total of 16 Abd-B-type genes have been discovered in mammals (35,36). The 5Ј-end of Hox genes is generally expressed in the hind limb bud (37). These genes exhibit overlapping domains of expression during limbs development, consistent with their key roles in regional identity specification (21). Hoxa11 has been shown to be expressed in the developing limbs and caudal body (13). Interestingly, elevated YY1 mRNA expression has been found in the limb bud and tail tip region as well (6). Given the findings reported in our current study, it is possible that YY1 and Hox proteins coordinate to regulate gene expression during embryogenesis.
The Hox-YY1 interaction may also be involved in regulating cell survival and/or apoptosis. Several lines of evidence indicate that Hox function is linked to apoptosis. For instance, HoxA5 has been shown to regulate expression of p53, a tumor suppressor gene, in breast cancer tissue (38). Miguel-Aliaga and Thor (39) have shown that neuronal segmental specificity is mediated by differential expression of the Abd-B Hox genes, which protect pioneer neurons from apoptosis via repression of the two RHG motif cell death activators, rpr and grim. Significantly, mammalian YY1 has also been shown to be a negative regulator of p53 recently (3,40,41). Removal of the PHR in chicken YY1 abrogated the ability of cYY1 to rescue cells lacking endogenous cYY1, 3 suggesting that YY1 PHR is necessary for YY1 to support cell survival. Taken together, these findings suggest that YY1 may regulate important developmental processes as well as cell survival, in part by regulating Hox protein activities via protein-protein interactions.
Previous studies have shown that YY1 is required for Ezh2 binding to the silenced state of muscle specific genes (42) and that YY1 interacts with the EED protein in the EED-EZH human homolog PcG complex (43). These findings implicate YY1 in epigenetic regulation, specifically in mediating Hox gene repression. Importantly, our study has identified a new relationship between PcG (YY1) and Hox proteins at the level of direct protein-protein interaction. We speculate that other Hox proteins such as Hoxa9 and Hoxd9 may also be regulated by YY1 via a similar mechanism. Altogether, our findings suggest that YY1 may participate in the regulation of embryonic development in part by modulating the activity of the Hox proteins. These findings provide a new insight into YY1 function and offer a potential mechanism for the critical role of YY1 in mouse embryonic development.   1-4, respectively). Two days after transfection, cell lysates were immunoprecipitated (IP) with an anti-FLAG antibody and Western blotted with an anti-HDAC2 antibody. The direct Western blots for FLAG-Hoxa11, HDAC1, HDAC2, YY1, and ␤-actin are shown in the bottom panel. The results show that overexpression of Hoxa11 pulls down HDAC2 in the presence of endogenous YY1 (lane 1). Knockdown of YY1 using siRNA can disrupt this interaction (lane 4). Unlike HDAC2, we found that Hoxa11 interacted with HDAC1 in the absence of YY1 (lane 4).