Transcription Factor BORIS (Brother of the Regulator of Imprinted Sites) Directly Induces Expression of a Cancer-Testis Antigen, TSP50, through Regulated Binding of BORIS to the Promoter*

Cancer-testis antigens (CTAs) are normally expressed in testis but are aberrantly expressed in a variety of cancers with varying frequency. More than 100 proteins have been identified as CTA including testes-specific protease 50 (TSP50) and the testis-specific paralogue of CCCTC-binding factor, BORIS (brother of the regulator of imprinted sites). Because many CTAs are considered as excellent targets for tumor immunotherapy, understanding the regulatory mechanisms governing their expression is important. In this study we demonstrate that BORIS is directly responsible for the transcriptional activation of TSP50. We found two BORIS binding sites in the TSP50 promoter that are highly conserved between mouse and human. Mutations of the binding sites resulted in loss of BORIS binding and the ability of BORIS to activate the promoter. However, although expression of BORIS was essential, it was not sufficient for high expression of TSP50 in cancer cells. Further studies showed that binding of BORIS to the target sites was methylation-independent but was diminished by nucleosomal occupancy consistent with the findings that high expression of TSP50 was associated with increased DNase I sensitivity and high BORIS occupancy of the promoter. These findings indicate that BORIS-induced expression of TSP50 is governed by accessibility and binding of BORIS to the promoter. To our knowledge this is the first report of regulated expression of one CTA by another to be validated in a physiological context.

CTCF 5 is a versatile chromatin factor originally identified as a repressor of Myc transcription (1)(2)(3)(4). CTCF recognizes DNA target sequences through its 11 zinc finger (11ZF) domain that is highly conserved from Drosophila melanogaster to humans (5) and functions not only as a repressor but also as an activator to regulate expression of various genes including APP, RB1, TP53, TERT, and ARF (6 -11). In addition to functioning as a conventional transcription factor, CTCF is known to regulate gene expression through organization of chromatin structure (12,13). For instance, CTCF regulates expression of imprinted genes H19/Igf2 through binding to H19 ICR, which is methylated at the paternal allele and unmethylated at the maternal allele, in a CpG methylation-dependent manner (14,15). The binding of CTCF to the H19 ICR regulates promoter activity of H19/Igf2 through formation of intrachromosomal loops, which results in allele-specific expression (16 -18). Furthermore, recent reports identified additional unique functions of CTCF such as regulated stability of trinucleotide repeats (19) and latency of various viruses (20 -22).
In mammals, there is a single CTCF paralogous gene named BORIS (brother of the regulator of imprinted sites), also known as CTCFL and CT27 (cancer-testis antigen 27; see CTDatabase) (23,24). BORIS contains an 11ZF domain that is highly similar to the 11ZF domain of CTCF. We recently reported that BORIS plays important physiological functions in progression of spermatogenesis by regulating expression of a testis-specific splicing variant of cerebroside sulfotransferase (Cst), Cst form F TS (25).
CTAs such as BORIS are encoded by genes that show restricted expression in testis and tumors. For the restricted expression, CTAs are considered as candidate targets of cancer vaccines (26,27). Immunization against BORIS was shown to be effective in treating mice injected with BORIS-expressing cancer cells (28,29). Furthermore, it has been reported that BORIS regulates expression of genes encoding other CTAs including MAGE-A1, SPANX, and NY-ESO-1 (30,31). Recent publications suggested BORIS was involved in DNA hypomethylation in cancers that is important for derepression of CTAs expression (32,33). These studies suggest that BORIS is an important regulator of CTA expression in cancer cells.
PRSS50, also known as TSP50 (testes-specific protease 50) or CT20, encodes a protease originally found as a gene with a promoter that is hypomethylated in tumors (34). Like BORIS, TSP50 is known as a CTA. Although the physiologic function of TSP50 is not clarified, the characteristic of its expression makes it interesting as a potential target for cancer immunotherapy. Recent publications suggested that expression of TSP50 is regulated by CpG methylation of the promoter (35), similar to many other CTAs, and by various transcription factors including SP1, CCAAT/enhancer binding protein (C/EBP), and TP53 (36,37). However, none of these factors was shown to regulate expression of TSP50 in a physiological context, which make it unclear which factors are involved in vivo. Given the possible utility of TSP50 as a target for tumor immunotherapy, we felt that clarification of the regulatory mechanisms was an important issue. We recently reported that BORIS is involved in regulating Tsp50 expression in testis, although it was not clear whether reduced expression of Tsp50 in testis of Boris knockout (KO) mice was a primary or secondary effect of the loss of BORIS (25).
Here we report that BORIS is directly involved in activation of TSP50 expression by binding to the promoter region. We identified two BORIS target sites in the TSP50 promoter that are highly conserved among species. We found that BORISinduced expression of TSP50 was dependent on regulated binding of BORIS to the promoter. The promoter of TSP50 was DNase I-hypersensitive in the cell line Delta47, which expresses TSP50 at high levels, but not in other cell lines with little or no expression. This suggested that accessibility to the promoter is the limiting factor for binding of BORIS and downstream expression of TSP50. To our knowledge, this is the first report demonstrating direct regulation of a CTA by BORIS in a physiological context.
Mice-Boris knock-out mice were generated as described previously (25). All animal experiments were conducted in compliance with the Animal Care and Use Committee of the NIAID, National Institutes of Health.
Antibodies-The antibody against mouse BORIS used for chromatin immunoprecipitation (ChIP) analysis was described previously (25). Mouse monoclonal antibodies specific for human BORIS were produced using recombinant human BORIS protein prepared from larvae of Trichoplusia ni infected with baculovirus expressing full-length human BORIS. 6 Nine antibody-producing clones that produced antibodies active in mobility super-shift assay were chosen, and ChIP assays were carried out using a mixture of these antibodies.
Quantitative PCR (qPCR)-Total RNA was isolated from tissues and cells using RNeasy Mini kit (Qiagen, Valencia, CA). cDNA was prepared with oligo-dT primers using SuperScript III First-strand Synthesis System (Invitrogen) according to the manufacturer's protocol. qPCR was performed using Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) and the 7900HT sequence detection system (Applied Biosystems). The expression of TSP50 and BORIS in primary melanomas was analyzed using TissueScan Melanoma Tissue qPCR Panel I (Origene, Rockville, MD) and Mx3000p (Agilent Technologies, Santa Clara, CA). Fisher's exact test was conducted to analyze statistical significance of correlation between expression of BORIS and TSP50 in melanomas. Primers for qPCR were designed using Primer Express software (Applied Biosystems). Primer sequences are listed in supplemental Table  1 or described in a previous publication (25) Expression levels were normalized against the housekeeping gene GAPDH except for the melanoma panel, which was pre-normalized against ACTB. Relative expression levels were determined by standard curve method or comparative Ct method according to the manufacturer's protocol. Each sample was analyzed in triplicate. Student's t test was performed to evaluate statistical significance. Data shown are the means Ϯ S.D.
Purification of Spermatocytes and Round Spermatids-Spermatocytes and round spermatids were purified as previously described (25). Briefly, dissociated testicular cells were separated by centrifugal elutriation. Partially purified spermatocyte and round spermatid fractions were incubated with 10 M Vybrant DyeCycle Green (Invitrogen) for 30 min at 32°C followed by staining with 4,6-diamidino-2-phenylindole (DAPI). Spermatocytes and round spermatids were sorted according to DNA content to improve purity using FACSAria (BD Biosciences). DAPI-positive dead cells were eliminated.
Electrophoretic Mobility Shift Assay (EMSA) and Methylation Interference Assay-EMSA and methylation interference assays were performed as described previously (25). Mouse and human TSP50 promoters were analyzed with BORIS protein of mouse and human origin, respectively. Because the amino acid sequences of human and mouse CTCF are highly conserved, we used mouse CTCF for analyses of both the mouse and human TSP50 promoters. Partially purified recombinant human BORIS protein prepared from larva of T. ni infected with baculovirus expressing full-length human BORIS was used for EMSA. 6 Briefly, larvae were homogenized in buffer containing 40 mM HEPES (pH 7.6), 100 mM NaCl, 2 mM MgSO 4 , 10 M ZnSO 4 , and 4ϫ Complete protease inhibitor mixture without EDTA (Roche Applied Science). Homogenate was centrifuged at 10,000 ϫ g for 10 min. The bulk of the BORIS protein was detected in the pellet using antibodies against the V5 tag. The pellet was washed sequentially with homogenization buffer containing 1, 2, and 4 M urea. After final washing, BORIS was extracted using homogenization buffer containing 6 M urea and 500 mM NaCl. The DNA binding ability and specificity of the recombinant protein was confirmed in EMSA tests using a variety of known BORIS binding DNA sequences as well as negative controls. The binding specificity was further confirmed by 6 A. A. Vostrov, manuscript in preparation. mobility supershift assay using antibodies against human BORIS (Abcam, Cambridge, MA). Partially purified recombinant mouse BORIS and CTCF proteins were prepared from Escherichia coli transformed with pET16b-mouse Boris and mouse Ctcf, respectively, following the procedure described above. Bacteria were disrupted by sonication. Samples prepared from E. coli transformed with pET16b empty vector were used as a negative control. For EMSA with nucleosomal DNA fragments, nucleosomes were reconstituted using a Chromatin Assembly kit (Active Motif, Carlsbad, CA) according to the manufacturer's protocol. Luciferase and control lysate from larva of T. ni were used as negative controls in EMSA for in vitro translated protein and recombinant protein prepared with baculovirus, respectively. Primers to prepare DNA fragments of human TSP50 promoter for EMSA are listed in supplemental Table 1.
Luciferase Assay-Fragments containing the entire 5Ј-untranslated region (5Ј-UTR) sequences of Tsp50 and its promoter sequences starting from position Ϫ150, Ϫ300, or Ϫ600 were cloned into pGL3-basic vector. NIH3T3 cells were cotransfected using FuGENE 6 (Roche Applied Science) with luciferase constructs together with pRL-CMV vector as an internal control and expression vectors for BORIS or CTCF in pCIneo according to the manufacture's protocol. Cells were cultured for 48 h at 37°C with 5% CO 2 . Luciferase assays were carried out with the Dual-Luciferase Reporter Assay System (Promega, Madison, WI) according to the manufacturer's protocol. The luciferase activities were normalized to the empty vector-transfected samples for each luciferase construct. Student's t test was performed to evaluate statistical significance. Data shown are the means Ϯ S.D.
ChIP assay-ChIP assays were carried out as previously described (25) with slight modifications. Briefly, fixed cells or tissue lysates corresponding to 1 ϫ 10 6 cells or 20 mg of tissue in 100 l of SDS lysis buffer (20 mM Tris-HCl (pH 8), 0.1% SDS, 2 mM EDTA, 150 mM NaCl, 1% Triton X-100) were incubated with Dynabeads M-280 (Invitrogen) preconjugated with antibodies for 1.5 h at 4°C with rotation. DNA fragments bound to beads were purified and subjected for qPCR. Primers used for qPCR are listed in supplemental Table 1. Student's t test was performed to evaluate statistical significance. Data shown are the means Ϯ S.D.
Bisulfite Sequencing-Bisulfite modification of genomic DNA was carried out using the Imprint DNA Modification kit (Sigma) according to the manufacturer's protocol. The bisulfite-modified TSP50 promoter region from Ϫ198 to Ϫ32 was amplified using 5Ј-GGGTTAAGGAGGGAAGTTAGT-3Ј and 5Ј-CTAATAAACTAAATAAAAATACTCCTAATA-3Ј as primers with PCR Platinum Taq Polymerase (Invitrogen) under the following conditions: 94°C for 2 min; 35 cycles of 94°C for 30 s, 52°C for 30 s, 72°C for 30 s; 72°C for 5 min. A CpG island from Ϫ65 to ϩ159 was amplified using 5Ј-GGG-GTATTAGGAGTATTTTTATTTAGTTTATT-3Ј and 5Ј-ACCTCAACAACAAAAACAACAACAA-3Ј as primers under the following conditions: 94°C for 2 min; 35 cycles of 94°C for 30 s, 55°C for 30 s, 72°C for 30 s; 72°C for 5 min. Amplicons were purified and cloned into pGEM-T vector (Promega). After transformation, plasmids from individual bacterial colonies were extracted and subjected to sequencing (Genomics Research Facility, Rocky Mountain Laboratories, NIAID, National Institutes of Health).
Methylation-specific PCR-Methylation status was analyzed by PCR using bisulfite-treated DNA as templates with primers specific for methylated or unmethylated sequences with designs based on the promoter region or the CpG island inside the transcribed region of TSP50. Primers used for these experiments are shown in supplemental Table 1. PCR was performed under the following conditions: 94°C for 2 min; 35 cycles of 94°C for 30 s, annealing temperature of 57°C (unmethylated promoter)/57°C (methylated promoter)/53°C (unmethylated transcribed region)/60°C (methylated transcribed region) for 30 s, 72°C for 30 s; 72°C for 5 min. PCR products were electrophoresed on 3% agarose gels and visualized by ethidium bromide staining.
DNase I Hypersensitive Assay-DNase I hypersensitive assays were carried out as previously described with slight modifications (38). Briefly, 2 ϫ 10 7 cells homogenized with a Dounce homogenizer were split into four samples and treated with 0, 80, 160, or 320 units/ml of DNase I for 3 min at 25°C. Samples were then treated with RNase A followed by treatment with proteinase K. 70 g of purified DNA digested with NheI and KpnI was separated on 0.8% agarose gels. The probe for hybridization was generated by PCR using 5Ј-CAG-TCTTTTCAGTGTCCCACCG-3Ј and 5Ј-CCCTTGGAGTC-TGTATGTTGGC-3Ј as primers.
Nucleosome Phasing-Nucleosome positioning was analyzed as described previously with modifications (39, 40). 1.5 ϫ 10 7 cells were harvested and washed with cold PBS. Cells were suspended in solution A (10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl 2 , 0.4% Nonidet P-40, protease inhibitor mixture) and incubated for 10 min on ice followed by treatment with solution B (10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl 2 , 0.4% Nonidet P-40, 1 mM CaCl 2 , protease inhibitor mixture). Samples were treated with 120 units/ml micrococcal nuclease for 10 min at 37°C, and the reaction was terminated by adding solution C (100 mM EDTA, 4% SDS). After treatment with 0.3 mg/ml RNase A for 30 min at 37°C, samples were treated overnight at 50°C with 20 mg/ml proteinase K. Purified DNA was separated on 1.5% agarose gels, and the band corresponding to mononucleosomes was excised from the gel followed by purification with QIAquick Gel Extraction kit (Qiagen). Nucleosome positioning was identified by tiling qPCR using mononucleosomal DNA as a template. Primers were designed by Primer3 from the Whitehead Institute to produce overlapping amplicons covering the promoter region of TSP50. Primers used for qPCR are shown in supplemental Table 1. Data were normalized to the signal of the highest peak.

Characterization of Tsp50 Expression in Wild Type (WT) and
Boris KO Mice-As a first step in evaluating the role played by BORIS in expression of Tsp50, we quantified levels of Tsp50 transcripts in WT and Boris KO mice by qPCR (Fig. 1A). As reported previously (34), Tsp50 transcripts were present at high levels only in testis and were significantly lower in testis of Boris KO mice (25). To determine if Tsp50 was differentially ex-pressed among subsets of spermatogenic cells, we compared the levels of transcripts in purified spermatocytes and round spermatids. We found that transcript levels in round spermatids were 16-fold higher than in SC (Fig. 1B, left panel). The differential distribution of Tsp50 expression between these subsets of spermatogenic cells was quite similar to that described previously for Cst form F, a reported target of BORIS, and Boris as illustrated in Fig. 1B (middle and right panels, respectively) (25).
The Mouse Tsp50 Promoter Contains Two BORIS Binding Sites-The above findings prompted us to determine if BORIS was directly involved in governing Tsp50 expression by binding to the promoter region. To examine this issue, we generated five DNA fragments (Fr), designated Fr1 to Fr5, that span the promoter region ( Fig. 2A) and tested them for BORIS binding by EMSA. BORIS bound uniquely and specifically to Fr2 (Fig.  2B), which includes sequences around transcription start site (TSS). We next examined the functional relevance of these results by generating a series of promoter truncations ( Fig. 2A) for use in luciferase reporter assays. NIH3T3 cells were cotransfected with each of the promoter constructs together with expression vectors for BORIS or CTCF. The results of these studies demonstrated that all three reporters were significantly activated by BORIS but not by CTCF (Fig. 2C). These results indicated that the activation of Tsp50 promoter was directly dependent on binding of BORIS to sequences downstream of Ϫ150 of the Tsp50 promoter. To more precisely define the tar-get sites for BORIS on the Tsp50 promoter, we conducted methylation interference assays using the Fr2 sequence as a probe (Figs. 3, A-C). Because CTCF is known to share target sequences with BORIS in vitro for the highly conserved DNA recognition domain, we first probed the Fr2 sequence using the 11ZF domain of CTCF that efficiently binds to target sequences to identify contact nucleotides (Fig. 3B). We identified two binding sites in the TSP50 promoter that we designated as B1 and B2. To validate BORIS would also recognize these residues, we carried out methylation interference assays using mouse BORIS protein and found that BORIS bound exactly the same guanine residues as 11ZF domain of CTCF (Fig. 3C). This result clearly shows that BORIS and CTCF recognize the same target sequences in the Tsp50 promoter. Furthermore, we conducted EMSA using either WT Fr2 (Fr2-wt) or Fr2 carrying mutations at contact nucleotides within B1 and/or B2 (Fig. 3, A and D). BORIS was bound to Fr2-wt as well as Fr2 with mutations  within either B1 (Fr2-B1mut) or B2 (Fr2-B2mut). In contrast, almost no band-shift was seen with Fr2-carrying mutations in both B1 and B2 (Fr2-B1/B2mut). We then tested the Fr2-wt and mutated Fr2 sequences in EMSA for binding of either fulllength CTCF or the 11ZF domain (Fig. 3E). The results of these studies showed that, like BORIS, both CTCF and the 11ZF domain bound to Fr2-wt, B1mut, and B2mut but that binding to Fr2-B1/B2mut was markedly reduced or undetectable. It is noteworthy that EMSA with BORIS, CTCF, and 11ZF domain showed the slower migrating shifted bands due to double occu-pancy of Fr2-wt (Fig. 3D, arrow 1; Fig. 3E, arrow 1; Fig. 3E, arrow 3, respectively) and the more rapidly migrating shifted band due to single occupancy of the single mutants, Fr2-B1mut and Fr2-B2mut (Fig. 3D, arrow 2; Fig. 3E, arrow 2; Fig. 3E, arrow 4, respectively).
To assess the functional relevance of these findings, we performed luciferase assays in NIH3T3 cells using Tsp50 promoter constructs starting from Ϫ600 that were WT or mutated at the B1, B2, or both the B1 and B2 sites. BORIS-induced activation of the promoter was reduced by mutations in either site and lost FIGURE 3. Identification of BORIS binding sites on mouse Tsp50 promoter. A, shown is the DNA sequence of mouse Tsp50 promoter. The numbering is relative to the TSS of Tsp50, which is marked by an arrow. The first ATG shown in bold is boxed. The solid line denotes Fr2. Contacting guanine residues of B1 and B2 BORIS binding sites are shown with asterisks. Residues marked with squares were converted to adenine in mutant constructs. B, methylation interference assay of the Fr2 fragment using 11ZF protein is shown. Partial sequences of the fragment around B1 and B2 sites are shown. Only the top strand is shown, as the bottom strand did not show any differences. Left lane, unbound fragments (Free); right lane, bound fragments (Bound). Asterisks denote contacting guanine residues. C, methylation interference assay of the Fr2 fragment using BORIS is shown. Partial sequences of the fragment around B1 and B2 sites are shown. Only the top strand is shown, as the bottom strand did not show any differences. Left lane, unbound fragments (Free); right lane, bound fragments (Bound). Asterisks denote contacting guanine residues. D, binding of BORIS to the mutated fragments was analyzed by EMSA. N, negative control; B, BORIS; Wt, Fr2-wt; B1, Fr2-B1mut; B2, Fr2-B2mut; B1/B2, Fr2-B1/B2mut. Arrow 1, shifted band for simultaneous binding of BORIS to both B1 and B2 site; arrow 2, shifted band for binding of BORIS to one site. E, binding of CTCF and 11ZF to the mutated fragments was analyzed by EMSA as in D. N, negative control; ZF, 11ZF; C, CTCF. Arrow 1, shifted band for simultaneous binding of CTCF to both B1 and B2 site; arrow 2, shifted band for binding of CTCF to one site; arrow 3, shifted band for simultaneous binding of 11ZF to both B1 and B2 site; arrow 4, shifted band for binding of 11ZF to one site. F, luciferase assays were performed using reporter constructs of the mutated TSP50 promoter (n ϭ 3). Empty vector, BORIS, or CTCF expression vectors were cotransfected with each construct. Luciferase activities were normalized against the empty vector-transfected sample. An asterisk denotes statistical significance (p Ͻ 0.005). G, binding of BORIS on 1.9 kb upstream from TSS (5Ј) and mouse Tsp50 promoter region (Pr) in testes of WT and Boris KO mice was examined by ChIP-qPCR using anti-mouse BORIS antibodies (n ϭ 3). An asterisk denotes statistical significance (p Ͻ 0.005).
almost completely with a construct bearing mutations in both sites (Fig. 3F). Taken together, these observations indicated that recognition of the Tsp50 promoter by BORIS, involving binding to both the B1 and B2 sites, was responsible for the promoter activation.
These findings prompted us to determine if BORIS bound to Tsp50 promoter sequences in vivo. To do this we performed ChIP-qPCR analyses using BORIS antibodies with testis tissue from WT and Boris KO mice (Fig. 3G). The results showed marked binding of BORIS to sequences in the Tsp50 promoter region in WT but not in Boris KO testis, whereas enrichment was not found in the 5Ј negative control region of either sample. From this, we conclude that BORIS binds directly to specific sequences within the Tsp50 promoter and is responsible for transcriptional activation of the gene in testis.
BORIS Binding Sites in the TSP50 Promoter Are Conserved between Mice and Humans-Because some CTAs are recognized to be important targets for immunotherapy of cancer and others, such as BORIS, to have potential for preventative vaccines (28, 41-43), we felt it is important to characterize mech-anisms governing expression of TSP50 in human cells. A comparison of the TSP50 promoter regions of various mammalian species revealed areas of strong conservation around the B1 and B2 BORIS target sites defined for mice (Fig. 4A), suggesting that similar regulatory mechanisms might be operative across species lines. This suggestion was strengthened by the results of methylation interference assays of the human TSP50 promoter region (Fig. 4B). These studies identified two BORIS binding regions with contact guanines lying in the regions of homology to the mouse B1 and B2 binding sites. We validated the importance of these two sites for binding of BORIS by EMSA using probes with WT sequences and sequences mutated at contact residues in B1, B2, or both sites (Fig. 4C). Studies performed with a WT probe revealed two shifted bands with the slower migrating band (Fig. 4C, arrowhead), corresponding to the occupancy of both target sites and the more rapidly migrating band (Fig. 4C, arrow), corresponding to the occupancy of single sites as seen with probes mutated at only B1 or B2. No band shifts were observed with a probe mutated at both B1 and B2, paralleling similar studies of the mouse TSP50 promoter. Addi- Gray boxes indicate conserved residues among species. The TSS is marked by an arrow. Asterisks denote contacting guanine residues of human or mouse. Residues marked with squares were converted to adenine in mutant constructs. B, shown is a methylation interference assay using DNA fragment containing B1 and B2 sites of human TSP50 promoter. Partial sequences of the fragment around B1 and B2 sites are shown. Only the top strand is shown, as the bottom strand did not show any differences. Left lane, unbound fragments (Free); right lane, bound fragments (Bound). Asterisks denote contacting guanine residues. C, binding of BORIS to the mutated fragments of human TSP50 promoter was analyzed by EMSA as Fig. 3D. Arrowhead, shifted band for simultaneous binding of BORIS to both B1 and B2 site; arrow, shifted band for binding of BORIS to one site. An asterisk indicates a shifted band most likely for binding of degradated BORIS protein. D, binding of CTCF and 11ZF to the mutated fragments of human TSP50 promoter was analyzed by EMSA as Fig. 3E. Arrow 1, shifted band for simultaneous binding of CTCF to both B1 and B2 site; arrow 2, shifted band for binding of CTCF to one site; arrow 3, shifted band for simultaneous binding of 11ZF to both B1 and B2 site; arrow 4, shifted band for binding of 11ZF to one site. N, negative control; ZF, 11ZF; C, CTCF.
tional studies of the WT and mutant probes were performed in EMSA using CTCF and 11ZF domain (Fig. 4D). In keeping with the observations made with the WT probe and BORIS, we observed two shifted bands in the presence of CTCF. Consistently, the patterns of band shifts seen with 11ZF protein resembled those seen with CTCF. Taken together, these results indicated that the mechanisms governing regulation of TSP50 expression by BORIS are conserved between mice and humans.
Expression of BORIS Is Essential but Not Sufficient for Expression of TSP50 in Human Cancer Cell Lines-To analyze the regulation of TSP50 expression by BORIS in human cells, we quantified expression of both BORIS and TSP50 in five BORISpositive cell lines, five BORIS-negative cell lines, and NHDF, which are supposed to be negative for both BORIS and TSP50 (Fig. 5A). The results showed that TSP50 transcripts were expressed by three BORIS-positive cell lines of different origins (Delta47, a multiple myeloma cell line; EKVX, a non-small cell lung carcinoma line; OVCAR8, an ovarian tumor cell line) but were almost completely negative in two other BORIS-positive lines (K562, a myelogenous leukemia cell line; MM-S1, a multiple myeloma cell line). It is noteworthy that the levels of TSP50 transcripts in Delta47 cells were more than 17-fold higher than in the other expressing lines and that there was no direct correlation between the levels of transcripts for TSP50 and BORIS in the cell lines expressing both genes. None of the BORIS-negative cancer cell lines including NHDF expressed TSP50. To investigate the relationship between the expression of BORIS and TSP50 in primary cancers, we analyzed their expression in melanomas, which are known to express various CTAs (Fig. 5B). Among 40 melanoma samples analyzed, nine showed higher expression of TSP50 than the background detected in normal skin samples (Fig. 5B, 1-9). Of these nine cases, five melanomas, including samples showing high expression of TSP50 (Fig. 5B, 1-3 and 5) expressed BORIS. There was a statistically significant correlation between the expression of BORIS and TSP50 (p ϭ 0.016), consistent with the idea that BORIS has a crucial function in activation of the TSP50 expression. On the other hand, BORIS expression was not sufficient to induce expression of TSP50 in melanomas as four cases were BORIS-positive but TSP50-negative (Fig. 5B, 13, 17, 36, and 37), similar to the observations made with cancer cell lines. These results suggest that the mechanisms involved in regulation of TSP50 expression by BORIS in cancer cell lines as well as primary melanomas are not completely congruent with those operative in normal testis.
Our studies about mouse Tsp50 promoter indicated that BORIS-induced expression of Tsp50 was controlled by binding of BORIS to the promoter. To determine if this mechanism was also operative in cancer cells, we performed ChIP-qPCR assays of the TSP50 promoter and an irrelevant site 3Ј to the promoter with antibodies against human BORIS for each of the cancer cell lines (Fig. 5C). We identified high enrichment of BORIS at the TSP50 promoter in Delta47 cells but markedly less enrichment for the other BORIS-positive cell lines. Taken together, these results indicate that expression of TSP50 in cancer cells may be controlled by regulating access of BORIS protein to the promoter rather than simply by the levels of BORIS expression as determined by transcript levels.
Methylation of the Body of TSP50 Gene Rather Than of the Promoter Region Correlates with Expression-It is known that binding of CTCF and BORIS to their target sequences is affected by DNA methylation. Moreover, other analyses demonstrated that the TSP50 promoter was methylated in a TSP50negative testicular embryonic carcinoma cell line and in tissue negative for TSP50, whereas the promoter was hypomethylated in testis and spermatogenic cells that expressed TSP50 (35). These data prompted us to analyze the cell lines described above for the methylation status of the TSP50 promoter, first by methylation-specific PCR (Fig. 6A, left column). Although the TSP50 promoter of Delta47 cells was unmethylated, the promoter of several TSP50-negative cell lines, including K562 and MM-S1 cells that were BORIS-positive, was also unmethylated. These results suggested that demethylation of the promoter was not sufficient to recruit BORIS to the promoter and induce expression of TSP50. Because the CpG island of the TSP50 promoter extends into the body of the gene, we also analyzed the methylation status of CpGs downstream of the TSS. Interestingly, CpGs inside the transcribed region of TSP50 were completely unmethylated in Delta47 cells, whereas they were fully methylated in other cell lines (Fig. 6A, right column). Parallel studies of the same CpGs in NHDF also showed relatively high methylation of the sequences. To extend our understandings of the relations between the methylation status of CpG islands in the TSP50 promoter region and the body of the gene to TSP50 expression, we analyzed their methylation status in selected cell lines as well as normal liver and NHDF by bisulfite sequencing (Fig. 6B). The TSP50 promoter region in TSP50-negative K562 cells, 8226 cells, NHDF, and liver was hypomethylated, validating that methylation of the promoter does not contribute in a major way to suppression of TSP50 expression in cancer and normal cells. Notably, CpGs flanking the B1 and B2 sites in the promoter region of the K562 cell line were not methylated, indicating that demethylation is not sufficient to induce binding of BORIS to the promoter. Moreover, EMSA using in vitro methylated fragments demonstrated that binding of BORIS to these sequences is methylation-independent (Figs. 6, C and D). Taken together, these results indicate that BORIS binding to the TSP50 promoter is not regulated by methylation of DNA.
Binding of BORIS to Nucleosome-free Target Sites-Areas of chromatin associated with active gene transcription frequently exhibit nucleosome-free status that results in the generation of DNase I hypersensitive site. Because the removal of nucleosome enables access of transcription factors to the target sequence, we performed DNase I hypersensitive assays of the TSP50 promoter region to analyze the status of chromatin in relation to binding of BORIS using cells that expressed both TSP50 and BORIS (Delta47), cells that are BORIS-positive but TSP50-negative (K562), and cells that were negative for the expression of both genes (8226) (Figs. 7, A and B). The results of these studies showed that Delta47 alone had a DNase I hypersensitive site in the TSP50 promoter region. To precisely determine the DNase I hypersensitive site in the promoter, we stud- ied nucleosome phasing of the TSP50 promoter region in the Delta47 and K562 cells (Fig. 7C). Strikingly, the promoter region from approximately Ϫ200 to the TSS that encompasses the BORIS binding sites was nucleosome-free in Delta47 cells but nucleosome-occupied in K562 cells. These data suggested that nucleosome occupancy might govern the ability of BORIS to bind to target sites in the TSP50 promoter. To analyze this possibility, TSP50 promoter sequences containing the two BORIS binding sites were used as free or mononucleosomal DNA in EMSA with BORIS (Fig. 7D). Nucleosome modification resulted in a distinct shifted band in the absence of added BORIS (Fig. 7D, arrow 3). The addition of increasing amounts of BORIS resulted in progressive reductions in the amount of unshifted free probe (Fig. 7D, arrow 4), whereas the intensity of the nucleosomal band was unchanged (Fig. 7D, arrow 3). These results suggested that BORIS is not able to bind the TSP50 promoter region when it is occupied by nucleosomes.
Histone Modifications around BORIS Binding Sites in the TSP50 Promoter-Recent studies showed that CTCF binding sites provide an anchor point for the positioning of flanking nucleosomes that are highly enriched for the specific histone marks such as H2A.Z, whereas there is no enrichment for other modifications including H3K27me3 (44). To explore in a preliminary way if BORIS target sites in the TSP50 promoter might be similarly structured, we performed ChIP-qPCR analyses of H2A.Z and H3K27me3 occupancy in the cell lines examined for binding of BORIS to TSP50 promoter (Fig. 8). The results showed that there was significant enrichment for H2A.Z only in Delta47 cells, whereas the levels of H3K27me3 occupancy were the lowest in the cells. These data implied that BORIS binding might, like CTCF, be associated with occupancy of H2A.Z and canonical histone modifications, although genome-wide studies of BORIS binding sites are clearly required to determine if this model can be generalized.

DISCUSSION
Previous studies from our laboratory demonstrated that Tsp50, a gene encoding CTA, was differentially expressed in testis from WT and Boris KO mice, suggesting that BORIS may be directly or indirectly involved in governing its expression (25). The results of the current study provide definitive evidence that TSP50 is a direct transcriptional target of BORIS in both mice and humans. It was recently reported that TSP50 is expressed at higher levels in conjunction with BORIS in spermatocytic seminomas than in seminoma/dysgerminomas (45). Our findings suggest that the high expression of TSP50 in spermatocytic seminoma can be ascribed to expression of BORIS.
Transcriptional activation of TSP50 by BORIS was shown to be dependent on occupancy of two targets sites in the promoter region that were conserved between mice and humans. Both binding sites share sequence similarity with the BORIS binding sequences we found in the promoter of Cst form F TS (25) as well FIGURE 7. Regulation of BORIS binding to TSP50 promoter. A, shown is a restriction map of human TSP50 promoter region. The probe used for DNase I hypersensitive assay is denoted. B, genomic DNA prepared from each of the cell lines was treated with DNase I followed by restriction digestion of NheI and KpnI. Samples were hybridized using the probe shown in A. An arrow indicates a hypersensitive site. An arrowhead denotes intact fragments generated by NheI and KpnI digestion. An asterisk shows nonspecific band. C, nucleosomal positioning around the TSS of human TSP50 promoter in Delta47 cells and K562 cells was analyzed by tilling qPCR using micrococcal nuclease-treated genomic DNA as templates. D, BORIS binding to the free (Free) or nucleosomal (Nuc) DNA fragment containing both the B1 and B2 sites of human TSP50 promoter was analyzed by EMSA. Arrow 1, shifted band for simultaneous binding of BORIS to both B1 and B2 site; arrow 2, shifted band for binding of BORIS to one site; arrow 3, shifted band for nucleosome; arrow 4, free fragment. N, negative control. as CTCF recognition sequences (46). EMSA analyses indicated that both binding sites of mouse and human TSP50 promoter could be occupied simultaneously by BORIS or by CTCF. Introduction of mutations to either B1 or B2 sites resulted in compromised activation of the mouse Tsp50 promoter by BORIS (Fig. 3F), indicating a requirement for both binding sites to induce full activation of Tsp50 promoter.
Analyses of human cancer cell lines failed to show a direct correlation between expression of BORIS and TSP50 transcripts. This indicated that mechanisms other than those responsible for expression of BORIS were required to drive high level expression of TSP50. ChIP analyses showed that BORIS was highly enriched at the TSP50 promoter in the Delta47 cells that expressed both BORIS and TSP50, but not in other BORISpositive cell lines, including K562 cells that express Ͼ2-fold higher levels of BORIS transcripts than Delta47 cells but are negative for TSP50 expression. Although it is well established that binding of CTCF to its target sequences is regulated by CpG methylation (14,47), we found that the differences in BORIS binding to the promoter could not be attributed to differences in methylation as the promoter was shown to be unmethylated even in BORIS-positive but TSP50-negative cells including K562 cells and binding of BORIS to the TSP50 promoter was methylation-independent. Although previous studies by others suggested that regulation of TSP50 expression was governed by promoter methylation (35), our results indicated that this is not the primary mechanism of regulation even though the methylation status may affect the total extent of activation. Our data suggest instead that it is the methylation status of the CpG island inside the transcribed region rather than the promoter that governs high expression of TSP50 as the CpG island inside the gene body was highly methylated in TSP50-low or -negative cells, whereas both the promoter and gene body were demethylated in Delta47 cells. Consistently, a recent study showed that methylation downstream of TSSs is tightly linked to gene silencing (48). On the other hand, although TSP50 was not expressed in NHDF and liver, their CpG island inside the transcribed region was not fully methylated like others, indicating partial demethylation is not enough to induce expression of TSP50 without BORIS. Further analyses will be required to clarify the effect of CpG methylation in TSP50 expression.
We found that the BORIS target sites in the TSP50 promoter were nucleosome-free in Delta47 but nucleosome-occupied in K562 (Figs. 7, B and C). In vitro studies strongly suggested that binding of BORIS to the TSP50 promoter was dependent on nucleosome-free sequences (Fig. 7D). Analysis of nucleosome positioning also revealed that nucleosomes on the TSP50 promoter were well phased in Delta47 but not in K562 cells (Fig.  7C). In view of the reported activity of CTCF in phasing nucleosomes around its target sites (44), it seems quite possible that BORIS could have similar functional attributes. Another potentially important parallel between BORIS and CTCF biding sites are the features of flanking nucleosomes that appear to be similarly marked by enrichment for H2A.Z and relative lack of H3K27me3 (Fig. 8). Genome-wide analysis of BORIS binding sites and histone marks are needed to clarify this possibility.
Finally, it was reported that down-regulation of BORIS (49) or TSP50 (50) expression results in apoptosis of cancer cell lines, suggesting that induction of TSP50 expression by BORIS may contribute to the survival of cancer cells. Although the potential clinical relevance of these observations is at present only conjectural, there is increasing evidence that BORIS and other CTAs may be important targets for immunotherapy (28,41). Further studies are clearly required to explore these possibilities.