Mitotic phosphorylation of CCCTC‐binding factor (CTCF) reduces its DNA binding activity

During mitosis, higher order chromatin structures are disrupted and chromosomes are condensed to achieve accurate chromosome segregation. CCCTC‐binding factor (CTCF) is a highly conserved and ubiquitously expressed C2H2‐type zinc finger protein which is considered to be involved in epigenetic memory through regulation of higher order chromatin architecture. However, the regulatory mechanism of CTCF in mitosis is still unclear. Here we found that the DNA‐binding activity of CTCF is regulated in a phosphorylation‐dependent manner during mitosis. The linker domains of the CTCF zinc finger domain were found to be phosphorylated during mitosis. The phosphorylation of linker domains impaired the DNA‐binding activity in vitro. Mutation analyses showed that amino acid residues (Thr289, Thr317, Thr346, Thr374, Ser402, Ser461, and Thr518) located in the linker domains were phosphorylated during mitosis. Based on these results, we propose that the mitotic phosphorylation of the linker domains of CTCF is important for the dissociation of CTCF from mitotic chromatin.

During mitosis, higher order chromatin structures are disrupted and chromosomes are condensed to achieve accurate chromosome segregation. CCCTC-binding factor (CTCF) is a highly conserved and ubiquitously expressed C2H2-type zinc finger protein which is considered to be involved in epigenetic memory through regulation of higher order chromatin architecture. However, the regulatory mechanism of CTCF in mitosis is still unclear. Here we found that the DNA-binding activity of CTCF is regulated in a phosphorylation-dependent manner during mitosis. The linker domains of the CTCF zinc finger domain were found to be phosphorylated during mitosis. The phosphorylation of linker domains impaired the DNAbinding activity in vitro. Mutation analyses showed that amino acid residues (Thr289, Thr317, Thr346, Thr374, Ser402, Ser461, and Thr518) located in the linker domains were phosphorylated during mitosis. Based on these results, we propose that the mitotic phosphorylation of the linker domains of CTCF is important for the dissociation of CTCF from mitotic chromatin.
Cis-acting regulatory DNA elements such as insulators and enhancers are involved in the temporal and cell type-specific control of gene expression through the formation of higher order chromatin structure. Higher order chromatin structure is tightly regulated throughout the cell cycle to achieve proper and dynamic chromosome processes. CCCTC-binding factor (CTCF) is a transcription factor containing 11 highly conserved zinc finger motifs (ZF1-ZF11; Fig. 1) responsible for its DNA-binding activity [1,2]. CTCF also mediates higher order chromatin structure formation by modulation of chromatin loops that define the boundary between active and inactive chromatin [3,4].
More than 3% of the total number of human genes belongs to the zinc finger protein family. Most zinc finger proteins, including CTCF, have tandemly repeated multiple zinc finger motifs separated by a highly conserved short linker peptide sequence, TGEKP. This linker domain is important for the structural stabilization of the zinc finger motif by forming an alpha-cap structure [5], and the DNAbinding activity of zinc finger motifs is regulated by phosphorylation of threonine residues in the linker domain during mitosis [6][7][8]. CTCF has 10 linker domains, but only linker domain 9 located between ZF9 and ZF10 contains the exact TGEKP sequence. It has been reported that CTCF is highly phosphorylated at multiple amino acid residues including T518 at the linker domain 9 [9][10][11]. However, the effect of mitotic phosphorylation on CTCF DNA-binding activity with regard to each phosphorylation site is not well understood.

Materials and methods
Cell culture, synchronization, and transfection HeLa S3 cells and MCF-7 cells were maintained at 37°C in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% fetal bovine serum. Cells were synchronized at mitotic phase by two cycles of "excess thymidine blockage followed by nocodazole arrest". Briefly, at 12 h post treatment of 2.5 mM thymidine (Sigma-Aldrich Co. LLC., St. Louis, MO, USA), cells were released into a fresh culture medium without an excess amount of thymidine for 10 h, and then synchronized again at G1/S boundary in growth medium with 2.5 mM thymidine for 14 h. At 6 h post release from thymidine treatment, cells were treated with 165 nM nocodazole (Sigma-Aldrich Co. LLC.) for 6 h. Mitotic cells were collected by gentle shaking of cell culture dishes. Transient DNA transfection assays were performed using Gene Pulser Xcell System (Bio-Rad Laboratories Inc., Hercules, CA, USA) according to the manufacturer's protocol.

DNA-binding assay
Asynchronously growing MCF-7 cells were incubated on ice for 5 min in the hypotonic buffer. Supernatant and pellet fractions were separated by centrifugation at 2300 g for 5 min, and the pellet fractions were incubated on ice for 5 min in a buffer (10 mM Tris/HCl, pH 7.9, 275 mM NaCl, 1 mM MgCl 2 , 0.1% Triton X-100, 1 mM Na 3 VO 4 , 1 mM NaF, 5 mM glycerol 2-phosphate). Nocodazole-arrested MCF-7 cells were incubated on ice for 5 min in the hypotonic buffer. CTCF was immunoprecipitated from the lysates using the anti-CTCF antibody. After washing with a buffer containing 10 mM Tris/HCl, pH 7.9, 500 mM NaCl, 1 mM MgCl 2 , 0.1% Triton X-100, 1 mM Na 3 VO 4 , 1 mM NaF, 5 mM glycerol 2-phosphate, immunoprecipitated proteins were incubated with or without lambda protein phosphatase (P7053, New England Biolabs Inc., Ipswich, MA, USA) at 30°C for 2 h, and then incubated in a DNA-binding buffer (20 mM Tris/HCl, pH 7.4, 150 mM NaCl, 2 mM MgCl 2 , 0.01% NP-40, 6.25% glycerol, 1 mM Na 3 VO 4 , 1 mM NaF, 5 mM glycerol 2-phosphate) with 0.15 ng of the rRNA gene upstream region fragment and 50 ng of poly (dIdC). The rRNA gene upstream region fragment used here is amplified from genomic DNA and purified from agarose gel. The exact sequence of the fragment corresponding to the region between nucleotide positions À961 and À851, where +1 is set to be the transcription start site [14] is as follows: 5 0 -GGTCCACGGGCCGCCCTGCCAGCCGGATCTGTCT CGCTGACGTCCGCGGCGGTTGTCGGGCTCCATCT GGCGGCCGCTTTGAGATCGTGCTCTCGGCTTCCG GAGCTGCG-3 0 , where the CTCF-binding site is underlined. The amounts of coimmunoprecipitated DNA were analyzed by qPCR and normalized by the protein amount of CTCF measured by IMAGEJ software (developed at the National Institutes of Health, Bethesda, MD, USA) from western blotting results.

CTCF is dissociated from mitotic chromatin
It is reported that most of the C2H2 zinc finger family proteins are excluded from mitotic chromosomes [11].
To determine the amount of CTCF bound to mitotic chromatin, subcellular fractionation was performed using a hypotonic buffer. HeLa S3 cells were synchronized at mitosis as described above. The amount of CTCF in the supernatant fraction obtained from mitotic cells was increased compared to asynchronous cells ( Fig. 2A). It has been reported that CTCF binds upstream of the rRNA gene promoter [14], so that next we examined the amount of CTCF bound to the rRNA gene locus during mitosis by chromatin immunoprecipitation (ChIP) assays. The amount of CTCF on the rRNA gene locus in mitotic cells was 80% less than that in asynchronous cells (Fig. 2B). After release from the mitotic block, the amount of CTCF on the rRNA gene was restored (Fig. 2B). These results suggest that CTCF is dissociated from chromatin during mitosis and reassociated upon G1 entry.

Phosphorylation of CTCF in mitosis
It has been reported that a variety of DNA-binding factors including transcription factors are phosphorylated and released from highly condensed chromosomes during mitosis [15]. To examine whether CTCF is also phosphorylated in the mitotic phase, we carried out phos-tag SDS/PAGE [16] which relies on the fact that phosphorylated proteins migrate slower than unphosphorylated proteins in the phos-tag gel. CTCF in mitotic cell lysates migrated slower than that in asynchronous cell lysates on phos-tag SDS/PAGE (Fig. 3A), suggesting that mitotic CTCF is phosphorylated and is therefore slower to migrate. To confirm the phosphorylation of CTCF during mitosis, the immunoprecipitated CTCF was treated with lambda protein phosphatase and subjected to phos-tag electrophoresis. The phosphatase treatment resulted in the increased migration rate of mitotic CTCF (Fig. 3B,  lanes 3, 4). These results suggest that CTCF is phosphorylated during mitosis. Both mitotic and asynchronous CTCF showed a slight increase in migration rate after lambda protein phosphatase treatment (Fig. 3B, lanes 1, 2). This result suggests phosphorylation of CTCF in interphase. It has been reported that  the amino acid residues (Ser604, Ser609, Ser610, Ser612) of the C-terminal domain of CTCF are phosphorylated by CK2 [17]. Phosphorylation of these amino acid residues, especially Ser612, is involved in functional switching of CTCF from transcriptional repressor to activator in c-myc [18]. Thus, CTCF is phosphorylated in not only mitosis but also interphase.
The linker domains of CTCF zinc finger domain are phosphorylated during mitosis It has been reported that serine residues at amino acid positions 604, 609, 610, and 612 of CTCF are phosphorylated by protein kinase CK2 [17]. Thr518, which is located in the linker domain 9, is phosphorylated during mitosis [11]. Thus, we constructed alanine-substituted mutants of these respective amino acid residues ( Fig. 4A; designated 4A and T518A). Similar to the results of 3 9 FLAG-CTCF WT (Fig. 4B, lanes 3 and 4), ectopically expressed 3 9 FLAG-CTCF 4A and T518A were still phosphorylated during mitosis (Fig. 4B, lanes 5-8), suggesting that these residues are not critical as a mitotic phosphorylation signal. To determine the location of the phosphorylated amino acid residues of CTCF during mitosis, we substituted serine and threonine residues at amino acid positions 289, 317, 346, 374, 402, 431, 461, and 518 of the linker domains (8A mutant, Figs 1 and 4A). 3 9 FLAG-CTCF 8A in mitotic cells migrated similarly to asynchronous cells (Fig. 4B lanes 9 and 10), suggesting that some of these amino acid residues are phosphorylated during mitosis. Furthermore, serine residues at amino acid position 604, 609, 610, and 612 are not phosphorylated in mitosis. Next, to identify the phosphorylated amino acid residues of linker domains in mitosis, we introduced respective alanine substitutions to seven of eight amino acid residues. Among them, only the 3 9 FLAG-CTCF 7A + T431 mutant shows the same migration rate as 3 9 FLAG-CTCF 8A mutant (Fig. 4C lanes 2 and 8). These results strongly suggest that each candidate amino acid residue of the linker domains except for T431 is phosphorylated during mitosis.

Mitotic phosphorylation of CTCF decreases the DNA-binding activity
To determine whether the mitotic phosphorylation of linker domains affects the DNA-binding activity, CTCF was purified using anti-CTCF antibody from either asynchronous or mitotic MCF-7 cell extracts and subjected to in vitro DNA-binding assays. The purified CTCF, bound to protein A agarose beads, was treated with or without lambda protein phosphatase, and incubated with a fragment of the rRNA gene that contains the CTCF-binding site. Quantification of fragments that interacted with CTCF was accomplished by qPCR and normalized against the protein amount of CTCF that was eluted from protein A agarose beads and quantified by western blotting. The DNA-binding activity of mitotic CTCF was 80% lower than asynchronous CTCF (Fig. 5A). However, phosphatase treatment restored the DNA-binding activity of mitotic CTCF (Fig. 5A). These results indicate that CTCF decreases its DNA-binding activity in a phosphorylation-dependent manner during mitosis. Next, we examined the DNA-binding activity of a phosphomimetic mutant, termed CTCF-8D, in which eight putative mitotic phosphorylation sites were changed to aspartic acid. 3 9 FLAG-tagged WT and 8D mutants of CTCF gave very similar expression levels in HeLa S3 cells (Fig. 5B). To analyze the DNA-binding activity of the phosphomimetic mutant of CTCF, we carried out ChIP assays of exogenously expressed CTCF. The amount of the CTCF-binding site in the rRNA gene locus bound to 3 9 FLAG-CTCF 8D was reduced by 30% compared to wild-type (Fig. 5C). These results suggest that the mitotic phosphorylation of CTCF decreases its DNA-binding activity.

Discussion
It has been reported that the phosphorylation of linker domains reduces the DNA-binding activity of C2H2 zinc finger proteins [6][7][8]. Structural analyses indicate that the threonine residues of linker domains are required for the stabilization of the interaction between DNA and zinc finger proteins by providing the capping structure for the alpha-helix of the zinc finger motif [5]. Our results suggest that the DNAbinding activity of CTCF is similarly regulated by the phosphorylation of linker domains during mitosis. It has been reported that serine/threonine kinases TOPK/PBK [19] and Cdk1 [20] phosphorylate the threonine residues of the conserved linker domains. Compared with other phosphorylated residues, the phosphorylation level of Ser461 in the linker domain 7 was weak, but a significant amount of band shift was observed during mitosis (Fig. 4). The peptide sequence of the linker domain 7 is different from the consensus motif (Fig. 1). Thus, it is possible that the DNA-binding activity of CTCF is regulated by an unknown protein kinase in addition to TOPK/PBK and Cdk1. It has been proposed that mitotic chromosomes exist in a highly condensed state to ensure stable and integral chromosome segregation [21]. Because transcription-related factors are excluded from mitotic chromatin, mitotic chromosomes are transcriptionally inactive [15]. In contrast, some markers for epigenetic gene control are retained during mitosis [22]. It is possible that transcription and the formation of higher order chromatin structures mediated by CTCF might result in a disadvantage to achieve accurate chromosome segregation; however, we found that a part of CTCF still interacts with mitotic chromatin as reported previously ( Fig. 2A) [23,24]. It is known that CTCF recognizes a wide variety of target gene loci using various combinations of the zinc finger domains [25,26]. We found that mitotic CTCF is detected on phos-tag SDS/PAGE as a smeared migration pattern (Figs 3 and 4), suggesting that CTCF is phosphorylated heterogeneously in mitosis. Therefore, it is possible that the partial phosphorylation of CTCF allows for a partial interaction with gene loci even during mitosis, possibly for the probable bookmarking of chromatin domains to achieve proper gene expression at the subsequent G1 phase.

Conclusions
Here we have shown that CTCF dissociates from mitotic chromosomes. Mutation analyses indicated that CTCF is phosphorylated in mitosis at Thr289, Thr317, Thr346, Thr374, Ser402, Ser461, and Thr518, all of which are located in linker domains. The mitotic phosphorylation of CTCF resulted in the reduction of the DNA-binding activity to a CTCF-binding site on rRNA upstream regions. These results suggest that the DNA-binding activity of CTCF could be regulated through the phosphorylation of linker domains during mitosis.