RP58 Associates with Condensed Chromatin and Mediates a Sequence-specific Transcriptional Repression*

An approximately 120-amino acid domain present generally at the NH2 termini, termed the POZ domain, is highly conserved in various proteins with zinc finger DNA binding motifs. We have isolated a novel protein sharing homology with the POZ domain of a number of zinc finger proteins, including the human BCL-6 protein. By using a binding site selection technique (CAST), a high affinity binding site of the protein was determined to be (A/C)ACATCTG(G/T)(A/C), containing the E box core sequence motif. The protein was shown to repress transcription from a promoter linked to its target sequences and was hence named RP58 (RepressorProtein with a predicted molecular mass of 58kDa). Immunogold electron microscopic study revealed that almost all RP58 is localized in condensed chromatin regions. These observations demonstrate for the first time that a protein mediating a sequence-specific transcriptional repression associates with highly condensed chromatin. We suggest that RP58 may be involved in a molecular link between sequence-specific transcriptional repression and the organization of chromosomes in the nucleus.

In previous studies, we identified and cloned the gene coding for a DNA-binding protein, translin, exhibiting general binding activity to consensus sequences at breakpoint junctions of chromosomal translocations in many cases of lymphoid neoplasms (19 -22). Further molecular analysis revealed that the native form of translin is a ring-shaped octamer connected by the leucine zipper motifs of each dimer and that this structure is responsible for the binding to target sequences situated only at single-strand DNA ends (23). Subsequently, to investigate the functional significance of translin, we examined whether it might interact with other proteins using yeast two-hybrid interaction analysis and identified an associated 33-kDa protein partner, TRAX, with extensive amino acid homology (28). During the course of screening for translin-associated molecules in the yeast two-hybrid system, which was designed to avoid self-association of translin to form the ring-shaped structure, we finally selected one clone D15 whose product gave the strongest activation only with the translin bait lacking the leucine zipper motif. The protein encoded by D15, named RP58, contains a POZ domain in its amino-terminal region and Kruppel-type zinc finger motifs separated by a short conserved motif, TGEKP(Y/P)X (H/C link) in the carboxyl-terminal region.
The goal of this study was to elucidate the biological role of RP58 and, especially, its possible involvement in transcriptional regulation. In transient cotransfection experiments, RP58 was shown to repress transcription from a promoter linked to its target sequence containing the E box motif. Immunogold electron microscopic study revealed that RP58 is localized in condensed chromatin regions, suggesting a role for the sequence-specific transcriptional repression in the heterochromatin structure.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Cloning-DNA encoding the translin domain lacking a leucine zipper motif (amino acids 1-189) was cloned into the yeast GAL4 DNA binding domain vector, pGBT9 (CLONTECH). The resulting plasmid, GAL4bd-Tra (LZ delete), was used as bait to screen a human spleen cDNA library in two-hybrid interaction analysis which was performed following the Matchmaker Two-hybrid System Protocol (CLONTECH). Positive yeast clones were screened by activation of his and lacZ reporter genes. After transformation of yeast DNA into Escherichia coli, plasmids containing cDNA clones were identified by restriction mapping and further characterized by DNA sequencing.
Amplification of cDNA Ends-5Ј-Amplification of cDNA ends was carried out using a 5Ј-RACE 1 system (Life Technologies, Inc.). First-* This work was supported by a Human Science Research Fund (to M. K.). 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AJ001388.
Determination of the Consensus Binding Sites by CAST-GST-RP58 (⌬POZ) immobilized on GSH-Sepharose beads was equilibrated to CAST binding buffer containing 0.5% Nonidet P-40 and washed two times with the same buffer. The duplex degenerate oligonucleotide (10 g), CAST DDG (AGACGGATCCATTGCA(N) 20 CTGTAGGAATTCGG-A), the immobilized proteins (30 l), and 400 ng of poly(dI-dC) (Amersham Pharmacia Biotech) were mixed in 50 l of CAST binding buffer with 0.5% Nonidet P-40 and incubated for 20 min at room temperature. The complex was then washed five times in the same buffer and subjected to 10, 14, or 18 cycles of PCR amplification using the primers, CAST-16(ϩ) (AGACGGATCCATTGCA) and CAST-15(Ϫ) (TCCGAAT-TCCTACAG). Following six rounds of selection, the final 30 cycles of PCR were performed, and then the resulting products of approximately 50 bp were digested with BamHI and EcoRI and cloned into pBluescript II SK(ϩ) (Stratagene) for sequencing.
Plasmid Construction-To generate the luciferase reporter construct (BS10-pGL2C), BS10 (a DNA fragment containing 10 repeats of the binding core sequence, (A/C)ACATCTG or its antisense (Fig. 4B)) was digested with KpnI and SacI, and it was cloned into KpnI/SacI sites of pGL2C, a luciferase control reporter vector (Promega).
Transient Transfection and Reporter Gene Assays-Cells (Cos-7) maintained in culture medium (RPMI 1640 medium with 10% newborn calf serum) were grown to 50% confluence in 12-well plates and washed once with phosphate-buffered saline (PBS). For cotransfection experiments, the reporters and various amounts of effectors were mixed with the medium without serum (final volume of 75 l), followed by addition of 7.5 l of Superfect Transfection Reagent (Qiagen). p␤gal-control plasmids (CLONTECH) expressing bacterial ␤-galactosidase (0.5 g) were also cotransfected to serve as an internal control for transfection efficiency. After incubation for 10 min at room temperature, each mixture was added to wells with 400-l aliquots of culture medium and incubated for 3 h at 37°C. The cells were then washed once with PBS and incubated for 48 h in 1 ml of the culture medium. The assay for luciferase activity was performed as follows. Transfected cells were washed twice with PBS and lysed with 100 l of LC␤ lysis buffer (Nippongene). After centrifugation at 15,000 rpm for 3 min, 20 l of extract were mixed with 100 l of Luciferase Assay Reagent (Promega), and the produced light was measured over 3 s using a Lumat LB9501 luminometer (Berthold). Similarly, the ␤-galactosidase activity was measured using a Luminescent ␤-galactosidase Detection Kit II (CLONTECH).
Immunogold Electron Microscopy and Immunofluorescence-To investigate the localization of RP58 by immunogold electron microscopy, IMR32 cells (human neuroblastoma) were harvested by centrifugation at 1200 rpm for 5 min, washed with PBS, and then fixed with 4% paraformaldehyde and 0.25% glutaraldehyde. The samples were dehydrated in a graded ethanol series, embedded in Lowicryl K 4M resin, and then allowed to polymerize for 6 days under UV irradiation. Ultrathin sections were cut and mounted on 200-mesh nickel grids. To block nonspecific binding sites, they were incubated with 0.3% non-fat dried milk in PBS for 30 min. The sections were then incubated with an affinity purified rabbit antibody to the carboxyl-terminal peptide of RP58 (diluted 1:100) for 12 h at 4°C, rinsed in PBS, and incubated with a goat anti-rabbit IgG conjugated with 15-nm colloidal gold (British Biocell International) for 1 h at 37°C. They were rinsed in PBS followed by distilled water and stained with 4% uranyl acetate. Samples were observed under an electron microscope (Hitachi H-7000) operating at 75 kV.
For immunofluorescence studies, IMR32 cells were spread on poly-D-lysine-coated coverslips by centrifugation at 1200 rpm for 5 min in a cytocentrifuge and were treated with 0.1% Tween 20 for 5 min to permeabilize cell membrane. The antibody to the carboxyl-terminal peptide of RP58 was used at 1:500 dilution, followed by fluorescein isothiocyanate-conjugated goat anti-rabbit IgG (Cappel), at a dilution of 1:500. After washing with PBS, the cells were examined on a Zeiss microscope (Axioplan 2).

Molecular Cloning of RP58 in Two-hybrid Interaction
Analysis-To understand better the functional significance of translin and its relation to other proteins, we examined whether it might be a member of a multicomponent complex using a protein-protein interaction analysis. To avoid self-association of translin to form the ring-shaped structure, DNA encoding its domain lacking the leucine zipper motif (amino acids 1-189) was cloned into the yeast GAL4 DNA binding domain vector and used as bait to screen a human spleen cDNA library in two-hybrid interaction analyses. From screening 6.2 ϫ 10 6 individual colonies for activation of the his reporter gene, 20 clones were obtained. Of these 20, nine were selected by activation of the lacZ reporter gene, but eight turned out to be the gene coding for TRAX showing extensive homology with translin (24). We finally selected one clone D15 (2024-nt cDNA) whose product gave the strongest activation with the translin bait. In contrast to the TRAX, which was selected by both the full-length and deleted translin, the D15 clone was obtained only when translin lacked its leucine zipper motif. In order to generate a full-length cDNA, we used 5Ј-RACE PCR to extend the D15 in the 5Ј direction. Analysis of seven resultant clones resulted in the identification of an additional 207-nt sequences in all cases, preceding the 5Ј end of the original clone, suggesting that this position might be a major transcription start site. Further screening of a human spleen cDNA library with the D15 probe resulted in the identification of several overlapping clones extending 2015-nt sequences in the 3Ј direction. As a result, we could construct the composite cDNA (4246 nt) including the AATAAA polyadenylation signal and poly(A) tract (Fig. 1A, GenBank TM accession number AJ001388). Nucleotide sequence analysis of the cDNA revealed an open reading frame (ORF) encoding a protein of 522 amino acids with a molecular mass of 58 kDa, RP58. Since no other favorable ORF was identified, we assumed that the ATG codon (nt 524 -526) immediately 3Ј of the in-frame termination codon (TGA) was the FIG. 1. A, the nucleotide sequence of the RP58 cDNA and its deduced amino acid sequence. The isolated clone (D15; nt 208 -2231), its 5Ј-RACE product (nt 1-355), and several overlapping clones were combined to construct the composite cDNA. The horizontal arrows indicate the sequences targeted by the specific primers (R1, R2, and R3) used for the 5Ј-RACE system. The 5Ј and 3Ј ends of the D15 clone and the 5Ј-RACE products are indicated by the vertical arrows. The 522-amino acid sequence deduced from the determined nucleotide sequence of the RP58 cDNA is shown. The in-frame translation stop codons (TGA; nt 488 -490, TAA; 2090 -2092) and polyadenylation signal are underlined. The four C 2 -H 2 zinc finger motifs translational initiation codon. The RP58 protein contains the putative nuclear localization signal in the amino-terminal region (KKKLKEK, amino acids 114 -120). The carboxyl-terminal region of the RP58 protein contains 4 sets of Kruppel-type zinc finger motifs separated by a short conserved motif, TGEKP(Y/P)X (H/C link). Data base searches using BLAST and FASTA programs revealed that the amino-terminal region (amino acids 1-115), termed the POZ domain here, is highly conserved in a number of C 2 H 2 zinc finger proteins (30 -40% identity), including the human BCL-6 protein, the human HIC-1 protein, the human PLZF protein, the human KUP protein, clone 18, mouse ZF-5, and human ZID (Fig. 1B). On the other hand, the central region between the POZ domain and zinc finger motifs was not found to have any significant similarity with previously described proteins.
Tissue-specific Expression of RP58 -To determine the size and expression pattern of RP58 mRNA, multi-tissue Northern blots were probed with a 1.5-kb RP58 ORF. A major transcript of approximately 4.3 kb was detected in lymphoid tissues, testis, heart, brain, skeletal muscle, and pancreas and at much lower levels in other tissues, along with a minor band at 9 kb (Fig. 1C). The size of the major transcript was consistent with that of RP58 cDNA, suggesting that the RP58 cDNA contains a full-length cDNA. Interestingly, a faint band at 7 kb appeared to be expressed only in brain. Our preliminary studies suggested that this 7-kb RNA may be derived from a particular portion of the brain. The expression of several different transcripts is suggestive of alternative splicing, as is frequently observed for mRNAs encoding zinc finger proteins (25,26). Fig. 1D illustrates expression of RP58 protein in various human tissues, studied by immunoblotting with an anti-RP58 antibody. Specific binding was found for a 60-kDa band which corresponds to the full length of RP58 protein. In addition, a 48-kDa band, thought to be the truncated form, 2 was detected. The results clearly indicated that there is no direct correlation between the mRNA level and the amount of protein in the spleen, brain, and skeletal muscle (compare Fig. 1, C and D). Comparison of protein samples revealed that the 60-kDa protein level in brain is several hundredfold higher than in any other tissues (Fig. 1E), suggesting post-transcriptional control.
The Consensus Binding Sequence for RP58 -We have employed a CAST protocol (27) to determine the consensus binding sequence for RP58. Based on previous reports (1, 2) showing inhibition of its DNA binding ability by the amino-terminal domains of ZID and ZF5, a truncated form of RP58 (a GST fusion protein lacking the POZ domain), GST-RP58 (⌬POZ), was immobilized on GSH-Sepharose beads and incubated with a pool of synthetic oligonucleotides. The bound complexes were isolated and subjected to 10, 14, or 18 cycles of PCR amplification. Following several rounds of amplification, the resulting products of approximately 50 bp were cloned and sequenced.  2. A, the RP58-binding sites selected by CAST. After cloning by CAST, the 56 fragments obtained were sequenced and classified into 22 independent clones. The approximately 20-bp sequences of the central portions between CAST-16(ϩ) and CAST-15(Ϫ) primers are shown. The perfect match for the binding core sequence ((A/C)ACATCTG) and its antisense are indicated with a black background and gray shading, respectively. One base mismatched sequences are underlined. The lowercase letters indicate the bases derived from primer sequences used for PCR. B, the consensus sequence for the RP58-binding sites. The percentage compositions of bases around the binding core sequences were calculated for the 44 sequences in the 22 independent clones. The nucleotides derived from the primer sequences were excluded from the calculation. (Fig. 2B). Interestingly, the most conserved sequence of the core was found to be the consensus E box sequence, CANNTG (28). To determine whether the consensus binding sites determined by CAST really bind to the RP58, an EMSA was performed with the recombinant RP58 using the various sequences listed in Fig. 2A. In this experiment, to avoid possible steric hindrance by GST, full-length RP58 containing a sixhistidine tag was used instead of the GST fusion protein. Of six fragments studied, CAST4 or CAST2-2, which have a perfect match for the core sequence, bound to the RP58 with the highest affinity. The binding affinity decreased with the increase of mis-match sequences and the pool of unselected sequences, N 20 , did not form the complex, suggesting RP58 binding specificity (Fig. 3A). In competition assays using cold CAST4 or N 20 fragments, the binding specificity of RP58 was again confirmed (Fig. 3B). Furthermore, DNA-protein complex formation was efficiently inhibited by the addition of anti-RP58 antibody, but not by normal rabbit IgG, suggesting the involvement of RP58 protein in the complex.

Sequence-specific Transcriptional Repression Mediated by
RP58 -Some of the POZ/zinc finger proteins including BCL-6, PLZF, and ZF5 are known to function as sequence-specific transcriptional repressors (2,(32)(33)(34). Based on the consensus binding sequence for RP58, we asked whether RP58 can also repress the transcription of reporter genes from promoters linked to the DNA target site, BS10, containing 10 copies of the binding core sequence for RP58 (Fig. 4B). The RP58 expression vector, pcDNA3.1-RP58, and the BS10-pGL2C reporter vector containing the luciferase gene driven by the SV40 early promoter (SV) linked to the DNA target site, BS10, were cotransfected into COS-7 cells (Fig. 4A). In this experiment, no significant difference was detectable between the basal levels of luciferase expressions of BS10-pGL2C and pGL2C (data not  3. A, identification of the RP58-DNA complex by EMSA. EMSA was performed as described earlier (21). Briefly, recombinant RP58 (0.4 g) was incubated at room temperature for 20 min with 10,000 cpm of 32 P-labeled various oligonucleotides selected by CAST and 1 g of poly(dI-dC) in CAST binding buffer. DNA-protein complexes were then separated on a 5% polyacrylamide gel, dried, and autoradiographed. The names of the clones used as probes are shown. B, competition assay. A competition assay was performed using the CAST4 fragment as a probe. Increasing amounts of cold CAST4 or N 20 fragments (10, 50, and 250 ng) and anti-RP58 antibody (rabbit Ig G), anti-RP58, or normal rabbit IgG, NR-IgG (0.1 and 0.5 g), were added as competitors.
shown). The results indicate that the pcDNA3.1-RP58 led to a strong and dose-dependent repression of the reporter gene expression, whereas the plasmid expressing the TRAX protein, pcDNA3.1-TRAX, did not affect the transcriptional activity (Fig. 4C). The transrepression activity of RP58 proved to be specific and dependent upon the presence of the RP58 domain in the effector vector and the BS10 sequence in the reporter vector.
Association of RP58 with Condensed Chromatin-The properties intrinsic to RP58, sequence-specific DNA-binding protein, transcriptional repression, and the presence of nuclear localization signal in the amino-terminal region, raise the possibility that it may localize to a particular region of the nucleus. Therefore, we examined localization of RP58 in the nucleus of IMR32 cells (human neuroblastoma) by immunofluorescence experiments. Indirect immunofluorescence of RP58 proteins revealed a punctate nuclear staining pattern, especially in regions adjacent to the nuclear membrane (Fig. 5). To investigate further its distribution at the ultrastructural level, we performed immunogold cytochemistry. When sections of IMR32 cells were labeled with anti-RP58 antibody, followed by the second antibody conjugated with 15-nm gold particles, over 90% of the gold particles were found to be distributed in the electron-dense chromatin (Fig. 6).
These results suggest that the transcriptional repression activity of RP58 might be important in controlling heterochromatin-mediated gene inactivation processes. DISCUSSION We have isolated a novel zinc finger protein, RP58, sharing homology with a number of transcription regulators at the amino terminus and determined the RP58 consensus binding site using the CAST technique. Interestingly, the preferred binding sequence consensus for RP58, (A/C)ACATCTG(G/T)(A/ C), contains the E box sequence of CANNTG, with most clones containing a CATCTG core. E boxes were first identified as in vivo protein-binding sites in the immunoglobulin heavy chain (IgH) enhancer (29 -31). Subsequently, E box binding activity was found for members of the helix-loop-helix class of proteins (28)  We have demonstrated that in transient transfection assays the RP58 protein can mediate a sequence-specific transcriptional repression. Transcriptional repression activities have been proposed for other POZ/zinc finger proteins including BCL-6 (32, 33), ZF5 (2) and PLZF (34). Although a number of the POZ/zinc finger proteins studied so far have displayed a consistent transcriptional repression activity, some POZ/zinc finger proteins display transcriptional activation (4) or the ability to modify nucleosomal structure (35), suggesting multifunctional aspects of this protein family. Other lines of evidence have indicated that the POZ domain acts in specific protein-protein interactions serving to organize higher order macromolecular complexes and thereby affect transcriptional activity (1,6). These results suggest that transcriptional repression is not an intrinsic property of the POZ domain. Therefore, the repressor activity of RP58 protein might be regulated by controlling the affinity to its target sequence through interaction with the POZ domains of other partner proteins.
A search of the eukaryotic promoter data base using the BLAST program revealed that the binding core sequence of FIG. 5. Localization of RP58 by indirect immunofluorescence. IMR32 cells were stained with the antibody to the carboxyl-terminal peptide of RP58, followed by secondary staining with fluorescein isothiocyanate-conjugated goat anti-rabbit IgG.
FIG. 6. Immunogold electron microscopic detection of RP58. IMR32 cells were processed for immunogold labeling by incubation with an affinity purified rabbit antibody to the carboxyl-terminal peptide of RP58, followed by a goat anti-rabbit IgG conjugated with 15-nm colloidal gold. RP58 is present in the putative promoter regions of various genes including the rat and bovine preprotachykinin genes (GenBank TM accession numbers E30009 and E11122), the human, mouse, and rat anion exchanger (AE1 or Band3) genes (GenBank TM accession numbers X77739, E25042, and L02942), the human Na/H exchanger (NHE-1) gene (Gen-Bank TM accession number E40005), and the human protein C gene (GenBank TM accession number E320001) ( Table I). The binding core sequences for RP58 were found 46 -252 bp upstream from the transcription start sites, suggesting that these genes are possible targets for the RP58 protein. Since levels of RP58 protein vary in human tissues, it is likely that transcription of target genes is regulated by the differential expression. Considering the extraordinary amount of RP58 protein expressed in brain, further studies focused on brain might provide a clue as to its functional significance.
A number of heterochromatin-associated proteins contain a conserved domain, designated the "chromodomain" that is necessary for chromosome binding and sufficient to define binding specificity (36). The chromodomain-containing proteins are not thought to bind to DNA directly but probably associate with heterochromatin through protein-protein interactions. Since RP58 does not contain the chromodomain, it is of particular interest to determine whether it associates with condensed chromatin regions through interaction with other domains of partner proteins or binds to DNA even when the DNA is highly compacted in the mitotic chromosome. Our observations also support the previous studies indicating that the Drosophila GAGA transcription factor, which belongs to the zinc finger protein family with POZ domain, can associate with specific DNA regions of heterochromatin throughout the cell cycle (11). In contrast to the GAGA factor functioning as a transcriptional activator, RP58 is a sequence-specific DNA-binding protein with a transcriptional repression activity, the first to be reported to associate with condensed chromatin. RP58 was originally found by screening for translin-associated molecules in yeast two-hybrid interaction analyses, the protein demonstrating interaction with translin lacking its leucine zipper motif, but not with full-length translin. Although it seems likely that the interaction occurs in the processing of multimeric translin structure, final conclusions can only be drawn when the interaction of these molecules is proven in vivo.
Transcriptional repression is an important component of gene regulation, but its mechanism is not yet as fully understood as that for transcriptional activation. The experiments reported here provide the molecular basis for addressing many intriguing questions such as the relation between gene-specific repression and the organization of chromosomes in the nucleus (37,38), and its implications for human diseases associated with the silencing mechanisms involved in genetic imprinting.
Further investigation of RP58 functions is certainly warranted to acquire a better understanding of these problems.