YY1/BCCIP Coordinately Regulates P53-Responsive Element (p53RE)-Mediated Transactivation of p21Waf1/Cip1

Transactivation of p21 (cyclin-dependent kinase inhibitor 1A, CDKN1A) is closely related to the recruitment of transcription cofactors at the p53 responsive elements (p53REs) in its promoter region. Human chromatin remodeling enzyme INO80 can be recruited to the p53REs of p21 promoter and negatively regulates p21. As one of the key subunits of the INO80 complex, YY1 has also been confirmed to bind to the p53RE sites of p21 promoter. Importantly, YY1 was recently reported to be bound and stabilized by BCCIP (BRCA2 and CDKN1A-interacting protein). Therefore, we hypothesized that the YY1/BCCIP complex plays an important role in regulating the transactivation of p21. Here we present evidence that the YY1/BCCIP complex coordinatively regulates p53RE-mediated p21 transactivation. We first confirmed the cross-interaction between YY1, BCCIP, and p53, suggesting an intrinsic link between three proteins in the regulation of p21 transcription. In dual luciferase assays, YY1 inhibited p53RE-mediated luciferase activity, whereas BCCIP revealed the opposite effect. More interestingly, the region 146–270 amino acids of YY1, which bound to BCCIP, increased p53-mediated luciferase activity, indicating the complexity of the YY1/BCCIP complex in co-regulating p21 transcription. Further in-depth research confirmed the co-occupancy of YY1/BCCIP with p53 at the p53RE-proximal region of p21. Lentiviral-mediated knockdown of BCCIP inhibited the recruitment of p53 and YY1 at the p53RE proximal region of p21; however, this phenomenon was reversed by expressing exogenous YY1, suggesting the collaborative regulation of YY1/BCCIP complex in p53RE-mediated p21 transcription. These data provide new insights into the transcriptional regulation of p21 by the YY1/BCCIP complex.


Introduction
The TP53 gene encodes the p53 protein that plays critical role in tumor prevention by taking control of a wide variety of cellular responses and the expression of multiple genes that regulates stress signal pathways [1]. In cancer cells, p53 is usually degraded and therefore becomes inactive [2]. However, p53 is activated upon cellular stresses such as DNA damage and transcriptionally activates sets of genes to

Stabilization of BCCIP on p53 Was Regulated by YY1 in HCT116 (p53+/+) Cells
In our preliminary experiments, we noticed that p53 protein expression levels were regulated by BCCIP. Thus, in order to confirm this phenomenon, we first examined the endogenous p53 protein level in Myc-BCCIP-expressing cells. The result, as shown in Figure 1A, endogenous p53 protein was dose-dependently increased in Myc-BCCIP-expressed cells at 72 h after transfection without affecting p53 mRNA levels ( Figure 1B). To further explore the role of BCCIP in stabilizing p53, endogenous p53 protein was examined in pLVX-shNT (non-targeting shRNA) and pLVX-shBCCIP transfected cells. As expected, the endogenous p53 protein level in shBCCIP-expressing cells clearly decreased compared to the shNT group ( Figure 1C, lane 2), suggesting the role of BCCIP in stabilizing endogenous p53 protein. Similarly, decreased endogenous p53 and its downstream target gene p21 proteins were observed by fluorescent staining of shBCCIP-expressing HCT116 cells ( Figure 1D). To further confirm the function of BCCIP in stabilizing p53, the effect of transient transfection of BCCIP on the level of exogenous p53 protein was detected in the presence of MG132 and HA-ubiquitin. Similar results were then observed in cells transfected with Flag-p53. Compared to the pcDNA3.1 vector (empty vector) group ( Figure 1E, lane 1), higher p53 protein levels were observed at 48 h in the Myc-BCCIPα-transfected cells. Importantly, a dose-dependent decrease of ubiquitination was exhibited in BCCIP-transfected groups after Flag immunoprecipitation (IP), suggesting that BCCIP may regulate p53 stability through the ubiquitin-proteasome-mediated degradation pathway.
On the other hand, according to our previous reports, BCCIP can also stabilize intracellular YY1 protein [17]. So, to know whether the stabilization effect of BCCIP on p53 is impacted by YY1, the effect of YY1 on p53 was first examined. In order to better visualize endogenous p53 by cell staining, fluorescent staining experiments were performed after treating cells with 5-Fluorouracil (5FU). Compared with the shNT group, it was clear that YY1 knockdown up-regulated p53 and its target gene p21 expression, and this effect was more pronounced in 5FU-treated cells (Figure 2A). In contrast, compared to the empty vector group, endogenous p53 protein levels in BCCIP transfected cells were inhibited by transfecting Flag-tagged full-length YY1 and deletion mutant YY1/146-270 amino acids (YY1/∆146-270), which were the binding regions for BCCIP ( Figure 2B, lanes 4-9/IB:p53). After confirming the negative regulation of YY1 on p53, an experiment with a design shown in Figure 2C was carried out. Consistent with previous results, endogenous p53 protein was increased by overexpressing BCCIP (lanes 2 and 6), and this increased level was further raised by transient knockdown of YY1 (lanes 4 and 8). However, in the stably expressing shYY1 cells, the dose-dependent increase of endogenous p53 by BCCIP was significantly suppressed ( Figure 2D, lanes 5-6 in comparison with lanes 2-3), indicating the complexity of YY1/BCCIP in regulating endogenous p53 protein levels.

Cross-Interaction between BCCIP, YY1, and p53 Was Verified in HCT116 (p53+/+) Cells.
The stabilizing effect of BCCIP on p53 combined with previous reports suggested that a cross-interaction between BCCIP, p53, and YY1 may exist. To address this issue, co-transfection/Co-IP experiments were performed. At first, the interaction between YY1 and BCCIP was tested in p53 wild type (p53+/+) and p53-null (p53−/−) HCT116 cells. As shown in Figure  3A, endogenous p53 could only be detected in HCT116 (p53+/+) cells (Input, IB:p53). After confirming the expression levels of Flag-YY1 and Myc-BCCIP (Input), Co-IP experiments were done using anti-Flag M2 and anti-c-Myc affinity agarose. It was obvious that BCCIP (both BCCIPα and BCCIPβ) and YY1 were bound each other (lanes 2-3, Flag IP and Myc IP), but in the case where p53 was absent, the binding ability of the two proteins was significantly reduced (lanes 8-9, Flag IP/IB:Myc and Myc IP/IB:Flag). Next, the interaction between p53 and BCCIP was determined in stably expressing pLVX-shNT and pLVX-shYY1 cells ( Figure 3B). The YY1 knockdown efficiency is shown in the input panel (IB:YY1). Similarly, in cells lacking YY1, the binding ability of p53 and BCCIP was much weaker than in cells with normal expression levels of YY1 (lanes 8-9, Flag IP/IB:Myc and Myc IP/IB:Flag). It is worth noting that the binding activity between p53 and BCCIPβ

Cross-Interaction between BCCIP, YY1, and p53 Was Verified in HCT116 (p53+/+) Cells
The stabilizing effect of BCCIP on p53 combined with previous reports suggested that a cross-interaction between BCCIP, p53, and YY1 may exist. To address this issue, co-transfection/Co-IP experiments were performed. At first, the interaction between YY1 and BCCIP was tested in p53 wild type (p53+/+) and p53-null (p53−/−) HCT116 cells. As shown in Figure 3A, endogenous p53 could only be detected in HCT116 (p53+/+) cells (Input, IB:p53). After confirming the expression levels of Flag-YY1 and Myc-BCCIP (Input), Co-IP experiments were done using anti-Flag M2 and anti-c-Myc affinity agarose. It was obvious that BCCIP (both BCCIPα and BCCIPβ) and YY1 were bound each other (lanes 2-3, Flag IP and Myc IP), but in the case where p53 was absent, the binding ability of the two proteins was significantly reduced (lanes 8-9, Flag IP/IB:Myc and Myc IP/IB:Flag). Next, the interaction between p53 and BCCIP was determined in stably expressing pLVX-shNT and pLVX-shYY1 cells ( Figure 3B). The YY1 knockdown efficiency is shown in the input panel (IB:YY1). Similarly, in cells lacking YY1, the binding ability of p53 and BCCIP was much weaker than in cells with normal expression levels of YY1 (lanes 8-9, Flag IP/IB:Myc and Myc IP/IB:Flag). It is worth noting that the binding activity between p53 and BCCIPβ (lane 9) was extremely low compared to that between p53 and BCCIPα (lane 8). The same co-transfection/Co-IP experiments were carried out in cells stably expressing pLVX-shNT and pLVX-shBCCIP ( Figure 3C). The BCCIP knockdown efficiency is shown in the input panel (IB:BCCIP). Compared to the shNT group, the interaction between p53 and YY1 was decreased in cells lacking BCCIP (lane 6 compared to lane 2). Taken together, the protein-protein interaction between YY1, BCCIP, and p53 suggested the existence of an intrinsic link in biological function between three proteins. (lane 9) was extremely low compared to that between p53 and BCCIPα (lane 8). The same co-transfection/Co-IP experiments were carried out in cells stably expressing pLVX-shNT and pLVX-shBCCIP ( Figure 3C). The BCCIP knockdown efficiency is shown in the input panel (IB:BCCIP). Compared to the shNT group, the interaction between p53 and YY1 was decreased in cells lacking BCCIP (lane 6 compared to lane 2). Taken together, the protein-protein interaction between YY1, BCCIP, and p53 suggested the existence of an intrinsic link in biological function between three proteins.

BCCIP and YY1 Coordinate the p53 Responsive Element (RE) Mediated Luciferase Activity in HCT116 (p53+/+) Cells.
To further explore the coordinative roles of BCCIP and YY1 in the regulation of the p53RE-mediated transcription of downstream target genes, multiple p53 responsive elements containing pp53-TA-Luc plasmid (p53RE-Luc) ( Figure 4A) were used as a model to reflect p53 transcriptional activity. P53RE-Luc activity was determined by dual luciferase assay. Then, the impact of YY1 on p53 transactivation was estimated by co-transfection of p53RE-Luc with Flag-YY1 plasmids or YY1-specific siRNA. In contrast to basal-level luciferase activity, co-transfection of p53RE-Luc and YY1 dose-dependently decreased p53RE luciferase activity ( Figure 4B, columns 3-4), and this effect, in turn, was reversed by simultaneous transfection of siYY1 ( Figure 4C, columns 4-5). Exogenous Flag-YY1 proteins are shown in the upper right corner of Figure 4B. Next, we measured the effect of BCCIP on p53RE luciferase activity. In contrast to the effect of YY1, co-transfection of p53RE-Luc and BCCIP (both BCCIPα and BCCIPβ) dose-dependently increased p53RE luciferase activity compared the basal level ( Figure 4D). However, the basal level of p53RE luciferase activity was inhibited by transfecting pLVX-shBCCIP ( Figure 4E). Exogenous BCCIP proteins and BCCIP knockdown efficiency are shown in the upper right corners of Figure 4D and Figure 4E, respectively.

BCCIP and YY1 Coordinate the p53 Responsive Element (RE) Mediated Luciferase Activity in HCT116 (p53+/+) Cells
To further explore the coordinative roles of BCCIP and YY1 in the regulation of the p53RE-mediated transcription of downstream target genes, multiple p53 responsive elements containing pp53-TA-Luc plasmid (p53RE-Luc) ( Figure 4A) were used as a model to reflect p53 transcriptional activity. P53RE-Luc activity was determined by dual luciferase assay. Then, the impact of YY1 on p53 transactivation was estimated by co-transfection of p53RE-Luc with Flag-YY1 plasmids or YY1-specific siRNA. In contrast to basal-level luciferase activity, co-transfection of p53RE-Luc and YY1 dose-dependently decreased p53RE luciferase activity ( Figure 4B, columns [3][4], and this effect, in turn, was reversed by simultaneous transfection of siYY1 ( Figure 4C, columns 4-5). Exogenous Flag-YY1 proteins are shown in the upper right corner of Figure 4B. Next, we measured the effect of BCCIP on p53RE luciferase activity. In contrast to the effect of YY1, co-transfection of p53RE-Luc and BCCIP (both BCCIPα and BCCIPβ) dose-dependently increased p53RE luciferase activity compared the basal level ( Figure 4D). However, the basal level of p53RE luciferase activity was inhibited by transfecting pLVX-shBCCIP ( Figure 4E). Exogenous BCCIP proteins and BCCIP knockdown efficiency are shown in the upper right corners of Figures 4D and 4E, respectively. activity (columns 2-5) was similar to Figure 5D. There was a slight increase for the BCCIP-transfected group (column 7) compared to the full-length YY1-transfection group. In addition, an additive effect between BCCIP and the YY1/146-270 region on p53RE luciferase activity was observed (column 9 in comparison with column 4). However, BCCIP did not change the effect of deletion mutant YY1 (YY1/∆146-270) on p53RE luciferase activity (columns 10 and 11 in comparison with column 5).  BCCIP and YY1 coregulation of p53RE luciferase activity was determined by co-transfection of p53RE-Luc with BCCIP and YY1. As shown in Figure 5A, decreased p53RE luciferase activity by YY1 was dose-dependently enhanced by transfecting both BCCIPα and BCCIPβ ( ## p < 0.01, columns 4-7 in comparison with column 3). In contrast, the increase in p53RE luciferase activity due to exogenous BCCIPα and BCCIPβ was superimposed by knocking down YY1 ( Figure 5B). Next, p53RE luciferase activity was measured by co-transfection of Flag-YY1 (0.3 and 0.9 µg) in the stably expressing pLVX-shBCCIP cells. shBCCIP-311 and shBCCIP-730 represent two different targeting sequences for BCCIP. The results were shown in Figure 5C. No matter which sequence region was targeted, compared to shNT groups, knocking down BCCIP appeared to result in no big changes in p53RE luciferase activity reduction due to YY1. Based on our previous report [17], the 146-270 amino acids of YY1 are the binding region for BCCIP. We therefore designed experiments as shown in Figure 5D. The upper panel presents the deletion mutants of YY1. As a result, the truncated YY1 (YY1/146-270 amino acids) showed an activity opposite to that of the full-length YY1 (column 4 compared to column 3). Interestingly, the p53RE luciferase activity of YY1-lacking residues 146-270 (YY1/∆146-270) remained the same as that of the full-length YY1 (columns 3 and 5), indicating that the region 146-270 amino acids of YY1 may be related to the transcriptional activation of YY1. Similar experiments were then performed with or without BCCIP transfection ( Figure 5E). In the absence of BCCIP, the impact of both BCCIPα and BCCIPβ on p53RE luciferase activity (columns 2-5) was similar to Figure 5D. There was a slight increase for the BCCIP-transfected group (column 7) compared to the full-length YY1-transfection group. In addition, an additive effect between BCCIP and the YY1/146-270 region on p53RE luciferase activity was observed (column 9 in comparison with column 4). However, BCCIP did not change the effect of deletion mutant YY1 (YY1/∆146-270) on p53RE luciferase activity (columns 10 and 11 in comparison with column 5). A classic chemotherapeutic drug, 5-Fluorouracil (5FU), has been shown to induce p53 [21]. In our experiment conditions, p53/p53T18P (threonine 18 site phosphorylated p53) and its target gene p21, but not GADD45/Bax/Bcl2, were significantly induced by 250 μM 5FU ( Figure 6A). However,

P53 Downstream Target Gene p21 Was Coordinately Regulated by the YY1/BCCIP Complex in HCT116 (p53+/+) Cells
A classic chemotherapeutic drug, 5-Fluorouracil (5FU), has been shown to induce p53 [21]. In our experiment conditions, p53/p53T18P (threonine 18 site phosphorylated p53) and its target gene p21, but not GADD45/Bax/Bcl2, were significantly induced by 250 µM 5FU ( Figure 6A). However, this induction was blocked in the pLVX-shBCCIP-transfected cells ( Figure 6B, lane 4). In addition, consistent with previous reports [17], YY1 protein levels significantly declined in BCCIP knockdown cells with or without 5FU treatment ( Figure 6B, lane 2/IB:YY1). Next, the coordinative effects of BCCIP/YY1 on p53 and p21 protein levels were examined by western blot in 250 µM 5FU-exposed HCT116 cells. The results, as shown in Figure 6C, indicate a dose-dependent increase of p53T18P and p21 in BCCIP-transfected cells in the presence of 5FU (lanes 2-3 and lanes 5-6 compared to lane 1 and 4, respectively). However, this phenomenon disappeared by co-transfecting BCCIP with YY1 (lanes 7-10). Protein levels were also visualized by cell staining after co-transfection of BCCIP and YY1 in the presence or absence of 5FU cells ( Figure 6D). Similar to western blot results, induction of p21 was observed in 5FU-exposed cells. This induction was enhanced by BCCIP transfection. However, it was suppressed by co-transfecting YY1, suggesting the collaborative work of BCCIP/YY1 on p53 and its target gene p21.  Figure 6B, lane 4). In addition, consistent with previous reports [17], YY1 protein levels significantly declined in BCCIP knockdown cells with or without 5FU treatment ( Figure 6B, lane 2/IB:YY1). Next, the coordinative effects of BCCIP/YY1 on p53 and p21 protein levels were examined by western blot in 250 μM 5FU-exposed HCT116 cells. The results, as shown in Figure 6C, indicate a dose-dependent increase of p53T18P and p21 in BCCIP-transfected cells in the presence of 5FU (lanes 2-3 and lanes 5-6 compared to lane 1 and 4, respectively). However, this phenomenon disappeared by co-transfecting BCCIP with YY1 (lanes 7-10). Protein levels were also visualized by cell staining after co-transfection of BCCIP and YY1 in the presence or absence of 5FU cells ( Figure 6D). Similar to western blot results, induction of p21 was observed in 5FU-exposed cells. This induction was enhanced by BCCIP transfection. However, it was suppressed by co-transfecting YY1, suggesting the collaborative work of BCCIP/YY1 on p53 and its target gene p21.

Coregulation of the YY1/BCCIP Complex on p53 Target Gene p21 Was Clarified in HCT116 (p53+/+) Cells.
In order to further confirm the coordinative effect of BCCIP/YY1 on p53 and its target gene p21, six primer sets including p53 binding sites in the p21 locus were used to amplify ChIP DNA ( Figure  7A). Prepared chromatin lysate from 0 or 250 μM 5FU-treated cells were applied for p53, YY1, and BCCIP ChIP assays. P53 and p21 proteins induced by 5FU were analyzed by western blot with

Coregulation of the YY1/BCCIP Complex on p53 Target Gene p21 Was Clarified in HCT116 (p53+/+) Cells
In order to further confirm the coordinative effect of BCCIP/YY1 on p53 and its target gene p21, six primer sets including p53 binding sites in the p21 locus were used to amplify ChIP DNA ( Figure 7A). Prepared chromatin lysate from 0 or 250 µM 5FU-treated cells were applied for p53, YY1, and BCCIP ChIP assays. P53 and p21 proteins induced by 5FU were analyzed by western blot with specific antibodies ( Figure 7B). It was clear that p53 ( Figure 7C), YY1 ( Figure 7D), and BCCIP ( Figure 7E) were co-occupied at −2.3 kb upstream of the p21 transcriptional start site (TSS). The enrichment of three proteins at −2.3 kb upstream of the p21 locus in 5FU-treated cells was more obvious. Although not very strong, we can see the recruitment of p53 at −1.8 kb and p53/YY1/BCCIP at −0.9 kb upstream of the p21 TSS. From the above experimental results, the existence of cross-talk between p53, YY1, and BCCIP in the transcriptional regulation of p21 can be speculated. To address this speculation, ChIP assays were further carried out in stably expressing shBCCIP cells with or without 5FU treatment. Select primer sets at the p21 locus including -2.3 kb and -1.5 kb were used to amplify ChIP DNA. As expected, silencing BCCIP prevented the recruitment of both p53 and YY1 at −2.3 kb upstream of the p21 TSS compared to the shNT group ( Figure 8A,B). Interestingly, transient transfection of YY1 restored the recruitment of p53 and YY1 at −2.3 kb upstream of the p21 TSS ( Figure 8C,D). However, no binding peaks of p53 and YY1 were seen at the non-p53 binding site (−1.5 kb). Protein levels, including all of the experimental conditions in Figure 8, are shown in E. specific antibodies ( Figure 7B). It was clear that p53 ( Figure 7C), YY1 ( Figure 7D), and BCCIP ( Figure 7E) were co-occupied at −2.3 kb upstream of the p21 transcriptional start site (TSS). The enrichment of three proteins at −2.3 kb upstream of the p21 locus in 5FU-treated cells was more obvious. Although not very strong, we can see the recruitment of p53 at −1.8 kb and p53/YY1/BCCIP at −0.9 kb upstream of the p21 TSS. From the above experimental results, the existence of cross-talk between p53, YY1, and BCCIP in the transcriptional regulation of p21 can be speculated. To address this speculation, ChIP assays were further carried out in stably expressing shBCCIP cells with or without 5FU treatment. Select primer sets at the p21 locus including -2.3 kb and -1.5 kb were used to amplify ChIP DNA. As expected, silencing BCCIP prevented the recruitment of both p53 and YY1 at −2.3 kb upstream of the p21 TSS compared to the shNT group ( Figure 8A,B). Interestingly, transient transfection of YY1 restored the recruitment of p53 and YY1 at −2.3 kb upstream of the p21 TSS ( Figure 8C,D). However, no binding peaks of p53 and YY1 were seen at the non-p53 binding site (−1.5 kb). Protein levels, including all of the experimental conditions in Figure 8, are shown in E. Chromatin lysates were prepared from cells with or without 5FU treatment. 5FU-induced proteins were analyzed by western blot (B). ChIP experiments were performed using anti-p53 (C), anti-YY1 (D), and anti-BCCIP (E) antibodies. ChIP DNA amplified with qPCR. The y-axis shows the ratio of ChIP DNA signals to IgG, and all signals were normalized to input (n = 3). * p < 0.05, ** p < 0.01, t-test, comparison to 5FU untreated group.

Discussion
In general, tumor suppressor gene p53 appears inactivated in many cancers. However, p53 can respond to various stimuli as a cellular stress sensor that triggers cell cycle arrest, apoptosis, DNA damage caused by various chemotherapeutic drugs, hypoxia, oxidative stress, and nutrient starvation periods [5,22]. Based on specific cellular responses, p53 transactivates and induces particular p53-mediated target genes such as p21 [23,24]. Like other transcription factors, p53 contains specific domains that are responsible for sequence-specific DNA binding and transcriptional activation [25]. It has been reported that activated p53 can recognize and bind directly to two highly conserved p53REs in the p21 promoter and induce transcriptional activation of the p21 gene [8]. Using Re-ChIP experiments, we previously provided evidence that the INO80 chromatin remodeling enzyme in conjunction with p53 co-occupies at both -2.2kb and -1.0kb upstream of the p21 TSS and negatively regulates p21 expression in a p53-mediated mechanism [9]. It is worth noting that as an essential co-activator and a subunit of the INO80 complex, YY1 can recruit the INO80 complex to some target genes such as CDC6 and GRP78 [26]. Our data, combined with previously reported data, have confirmed that YY1 can be recruited at the p53 binding site in Stably expressing shNT or shBCCIP cells were transiently transfected with Flag-YY1 for 48 h. Then, prepared chromatin lysate was applied to ChIP assays. The y-axis shows the ratio of ChIP DNA signals to IgG, and all signals were normalized to input (n = 3). * p < 0.05, ** p < 0.01, t-test, comparison to shNT group. (E) Indicated protein levels in A, B, C and D.

Discussion
In general, tumor suppressor gene p53 appears inactivated in many cancers. However, p53 can respond to various stimuli as a cellular stress sensor that triggers cell cycle arrest, apoptosis, DNA damage caused by various chemotherapeutic drugs, hypoxia, oxidative stress, and nutrient starvation periods [5,22]. Based on specific cellular responses, p53 transactivates and induces particular p53-mediated target genes such as p21 [23,24]. Like other transcription factors, p53 contains specific domains that are responsible for sequence-specific DNA binding and transcriptional activation [25]. It has been reported that activated p53 can recognize and bind directly to two highly conserved p53REs in the p21 promoter and induce transcriptional activation of the p21 gene [8]. Using Re-ChIP experiments, we previously provided evidence that the INO80 chromatin remodeling enzyme in conjunction with p53 co-occupies at both -2.2kb and -1.0kb upstream of the p21 TSS and negatively regulates p21 expression in a p53-mediated mechanism [9]. It is worth noting that as an essential co-activator and a subunit of the INO80 complex, YY1 can recruit the INO80 complex to some target genes such as CDC6 and GRP78 [26]. Our data, combined with previously reported data, have confirmed that YY1 can be recruited at the p53 binding site in the p21 promoter and be involved in the regulation of p21 transcription [9,13], suggesting that the recruitment of INO80 complex to the p53 binding site of the p21 TSS may be achieved by YY1. Based on the most recent report, the recruitment of INO80 complex to the p53 binding site in the p21 promoter may function in two different ways according to the intracellular condition: i) the INO80 complex represses p21 expression by regulating the promoter proximal nucleosome arrangement under normal cellular conditions; ii) in Doxo-treated cells, the INO80 complex rapidly removes accumulated H2A.Z and relieves its inhibition of p21, thereby inducing the transcriptional expression of p21 [27].
On the other hand, we recently found that BCCIP not only stabilizes YY1 protein, but also directly binds to YY1 in HCT116 cells. The 146-270 amino acid of YY1 seemed to be the key region for maintaining the interaction between YY1 and BCCIP [17]. However, the interaction between the two proteins was extremely weakened in p53-null HCT116 cells ( Figure 3A), demonstrating that the intracellular function of the YY1/BCCIP complex required the presence of p53 in cells. What is more, in our experimental conditions, p53 can bind to YY1 and BCCIP, respectively, but once YY1 or BCCIP was knocked down, the binding activity of p53 with the other protein was decreased, indicating the cross-interaction between p53, YY1, and BCCIP in cells. It is well known that the stability of the p53 protein directly affects the transcriptional activation of downstream target genes [28]. Given that the stabilization of p53 was regulated by both BCCIP and YY1, the coordinative function between p53, BCCIP, and YY1 on p53RE-mediated downstream target gene transcription, including p21, was surmised. Consistent with this, the luciferase activity of p53RE-Luc, which contains multiple p53REs, was regulated by both YY1 and BCCIP. As previously reported [14,19], YY1 inhibited p53RE-mediated transcriptional activation, whereas BCCIP revealed the opposite effect. Therefore, we can imagine that under normal circumstances, p53/YY1/BCCIP may maintain a dynamic balance by restricting each other. Any increase or decrease in either side can break this balance and affect the transcriptional activity of p53.
Given that YY1 contains both transcriptional activation and repression domains, it can activate or repress gene transcription by recruiting different transcriptional cofactors to its activation or repression domain [10,29]. Our data supports this view. Contrary to the effect of full-length YY1, 146-270 amino acids of YY1 increased p53RE-mediated transcriptional activation. However, the deletion mutants YY1/∆146-270 maintained the same p53RE luciferase activities as the full-length YY1, suggesting that this region of YY1 may be involved in p53RE-mediated transcriptional activation. Once this region is blocked by interacting proteins, such as BCCIP, YY1 functions as a transcription inhibitor. We mentioned earlier that the binding ability of YY1 and BCCIP was significantly attenuated in p53-null HCT116 cells. In contrast, YY1/BCCIP complex affected p53 stability in HCT116 (p53+/+) cells. Interestingly, the effect of YY1/BCCIP complex on p53 was strongly related to the knockdown level of YY1. In the case of transiently knocking down YY1 and simultaneously transfecting BCCIP, the p53 protein level was increased ( Figure 2C). However, when stably knocking down YY1 and then transfecting BCCIP, the p53 protein level in cells declined ( Figure 2D). In another case, exogenous BCCIP dose-dependently increased p53RE-Luc-mediated p21 and p53T18P expression levels, suggesting that p53 phosphorylation increases the transcriptional activity of p53 by enhancing p53 stability [28]. More in-depth research confirmed that p53 co-localized with YY1 and BCCIP at p53RE sites in the p21 promoter, suggesting the coordinative roles of YY1/BCCIP/p53 in regulating p53RE-mediated p21 transcription (Figure 9). For example, the binding of p53 and YY1 to the p53RE sites in p21 was reduced by knocking down BCCIP, but this phenomenon was rescued by over-expressing YY1. Figure 9. Coordinated regulation of p53RE-dependent gene transcription by p53/BCCIP/YY1. BCCIP promotes p53RE-mediated transactivation by binding to and stabilizing p53 and further activates p21 transcription. This pathway is regulated by intracellular YY1 protein levels. Increased YY1 inhibits p53RE-mediated transactivation, and the opposite effect is seen in transient knockdown of YY1. However, stable knockdown of YY1 attenuates BCCIP transcription, further affecting p53 stability, which suggests that there may be feedback inhibition. Black solid arrows, published data [17]; black T dotted arrow, indirect inhibition; blue solid arrows, possible pathway of p53RE-dependent gene transcription; blue T dotted arrow, stably expressing shYY1 affect the INO80/YY1-BCCIP pathway.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript; or in the decision to publish the results.