Chmadrin: A novel Ki-67 antigen-related perichromosomal protein possibly implicated in higher order chromatin structure

Chromosome condensation involves successive orders of DNA folding. The DNA fiber is first compacted by winding around the histone core, generating a nucleosome (Kornberg, 1974). The nucleosome fiber is then shortened and thickened to give a 30 nm fiber, probably with the aid of histone H1 (Horowitz et al., 1994). A considerable body of evidence suggests that the 30 nm fiber is then gathered into loops in both the interphase nuclei and mitotic chromosomes (reviewed by Gasser et al., 1989). Higher order chromatin structure is generally believed to be organized by holding these loops together in a cell-cycle dependent manner. A biochemical fraction, which is referred to as the chromosome scaffold, has been postulated to act as a loop fastener, since chromosome scaffolds isolated from mitotic chromosomes retain their characteristic chromosomal morphology with paired sister chromatids and condensed centromeres (Adolph et al., 1977; Earnshaw and Laemmli, 1983). One of the major components of the chromosome scaffold fraction has been shown to be topoisomerase II (Earnshaw et al., 1985; Gasser et al., 1986), which was demonstrated to function in both mitotic chromosome condensation in vitro (Adachi et al., 1991; Hirano and Mitchson, 1993) and in vivo (Uemura et al., 1987). To date, topoisomerase II represents the only molecule which is functionally implicated in mitotic chromosome condensation prior to the recent discovery of the SMC (structural maintenance of chromosomes) protein family.

The SMC protein family plays a global role in higher order chromosome dynamics (reviewed by Hirano et al., 1995; Koshland and Strunnikov, 1996). It has been shown that a subset of SMC proteins are essential for chromosome condensation in both in vivo and in vitro conditions (Strunnikov et al., 1993, 1995; Chuang et al., 1994; Hirano and Mitchson, 1994; Saitoh et al., 1994; Saka et al., 1994). In addition, some of those proteins have been shown to associate with non-SMC proteins to form higher order complexes (Chuang et al., 1996; Guacci et al., 1997; Hirano et al., 1997; Michaelis et al., 1997; Losada et al., 1998). In Xenopus, for example, XCAP-C and XCAP-E, both of which are members of the SMC family of proteins, form a multiprotein complex (termed 13S condensin) containing three additional subunits XCAP-D2, XCAP-G and XCAP-H (Hirano et al., 1997).

XCAP-H may functionally link SMC proteins and topoisomerase II, since the Barren protein, a Drosophila homologue of XCAP-H, interacts with both SMC and topoisomerase II (Bhat et al., 1996). The functional role of SMC complexes in mitotic chromosome condensation has been partly elucidated as the result of the recent identification of their biochemical activities, i.e. DNA renaturation activity (Sutani and Yanagida, 1997) and DNA supercoiling activity (Kimura and Hirano, 1997).

In addition, a class of proteins which localize around mitotic chromosomes (perichromosomal proteins) have also been identified. These proteins have been classified into several distinct classes, based on their subcellular localization during interphase by Hernandez-Verdun and Gautier (1994). Of these proteins, the most prominent may be the Ki-67 antigen (pKi-67). pKi-67 is a nucleolar protein which was originally identified by Gerdes et al. (1983) using a monoclonal antibody against a nuclear antigen from a Hodgkin’s lymphoma-derived cell line and was recently cloned in both human and mouse (Schlüter et al., 1993; Starborg et al., 1996). pKi-67 has been widely used as a proliferation marker, since its expression is observed only in growing cells (Gerdes, 1990). Because of the apparent contact of perichromosomal proteins with mitotic chromosomes, their significance in mitotic chromosome organization has been an ongoing assumption. However, this assumption has not yet been experimentally verified.

In this report, we describe the identification of a novel perichromosomal protein, chmadrin, from rat kangaroo (potoroo) PtK₂ cells. The predicted amino acid sequences of chmadrin show significant similarities to pKi-67 in limited regions and the structural motifs that are common to chmadrin and pKi-67 are present. The central region of the chmadrin protein is a characteristic repetitive sequence domain (called the ‘chmadrin repeat domain’) composed of tandemly repeated highly conserved 20 or 21-residue elements (‘chmadrin motif’), which correspond to the motif found in the central region of pKi-67. The C-terminal region of chmadrin also has a repetitive structure consisting of eight relatively conserved residues with irregular intervals. Careful comparison also revealed a weak but significant similarity in the C-terminal regions of chmadrin and pKi-67, which are characterized by multiple appearances of LR (leucine and arginine) pairs with irregular spacing (referred to as ‘LR domains’). Through a series of transient transfections, the regions which are likely to be responsible for the characteristic cellular localization of chmadrin, i.e. predominantly nucleolar during interphase and chromosomal in the mitotic phase, were revealed. The nucleolar localization during interphase is likely to be conferred by residues 494-778, a region well conserved between chmadrin and pKi-67. Since the LR domain targeted the body of mitotic chromosomes when expressed transiently, the LR domain is likely to be essential, but not sufficient, for the perichromosomal localization of chmadrin. We also found that the LR domain, which is expressed during interphase induces the formation of a heterochromatin-like structure as a structural constituent. These observations point to the possible involvement of chmadrin in the organization of higher order chromatin structures such as mitotic chromosomes and interphasic heterochromatin.

PtK₂ (rat kangaroo kidney epithelium) cells were maintained in modified Eagle’s medium supplemented with 10% fetal bovine serum and non-essential amino acids. For immunofluorescence, cells were grown directly on sterile acid-washed 12 mm diameter coverslips in 35 mm dishes.

2A11 was prepared using recombinant human Hsc70 (70 kDa heat shock protein cognate) protein as the immunogen. Preparation of the immunogen was carried out as previously described (Imamoto et al., 1992). 2A11 was purified from ascitic fluid of mice with implanted 2A11 hybridomas by hydroxylapatite (HTP-grade, Bio-Rad) chromatography with buffers containing 0.5 M NaCl and step-gradient concentration of sodium phosphate (pH 6.7). 2A11 was eluted with a buffer containing 0.5 M NaCl and 150 mM sodium phosphate. The class of 2A11 was determined to be IgM, as evidenced by the use of a mouse monoclonal antibody isotyping kit (Amersham).

A λZAPII cDNA library of PtK₂ cell mRNA, having a complexity of 9×10⁵ independent clones, was constructed according to the protocol recommended by the manufacturer (Stratagene) and was screened with 2A11. Plaque purified phages were isolated with three or four rounds of screening, and in vivo excision was carried out using Exassist phage and SOLR recipient cells (Stratagene) to obtain plasmids. The 5′-end of the obtained cDNA was recovered by the 5′ RACE method (Frohman et al., 1988) using the 5′ RACE System, Version 2.0 (Life Technologies, Inc.) as follows. GSP1 (5′-TTTCC-TTCATTTTGAGAGCC-3′) was annealed to 0.5 μg of PtK₂ total RNA and the first strand of cDNA was synthesized with SUPER SCRIPT™ II (Life Technologies, Inc.) at 50°C. After the addition of dCTPs to the 5′ end of the first strand by terminal deoxynucleotidyl transferase, two rounds of PCR were performed. Initially, the first strand was amplified with anchor primer (5′-CUACUACUACUAG-GCCACGCGTCGACTAGTACGGGIIGGGIIGGGIIG-3′) and GSP2 (5′-TTCGGCTGCGGAGTAAAAATCTGCC-3′). The initial PCR product was then amplified with AUAP primer (5′-GGCCACGC-GTCGACTAGTAC-3′) and GSP3 (5′-ATATTAGGATCCTCACCT-CCTGTCCTGGCTGCCTCGT-3′). The second PCR product was directly cloned into pGEM-T vector (Promega). The resulting plasmids were sequenced by the dideoxy chain termination method using an automated DNA sequencer (LiCor Model 4000L) and Thermosequenase (Amersham).

Poly(A)⁺ RNAs of PtK₂ cells were purified by guanidium thiocyanate extraction and binding to Oligotex-dT30<super> resin (Roche). For northern blots, 5 μg of poly(A)⁺ RNA was run on a formaldehyde-0.8% agarose gel, and then transferred to Hybond-N⁺ (Amersham). For each lane of genomic southern blots, 10 μg of PtK₂ chromosomal DNA digested with a restriction enzyme (BamHI or EcoRI) was separated on 1% agarose gel and transferred to Hybond-N⁺. Both blots were probed with a cDNA fragment (bases 2463-4492) labeled with [α-³²P]dCTP (NEN Research Products) using Ready-To-Go DNA labeling kit ([α-³²P]dCTP) (Pharmacia).

A fragment of chmadrin cDNA corresponding to the amino acid residues 2263-2603 was ligated into the pGEX-5X-2 vector (Pharmacia) in the correct frame. The GST-fusion protein was purified by affinity chromatography on glutathione Sepharose-4B resin (Pharmacia) and used to immunize two rabbits (kbs:JW) purchased from Kitayama Labes Co., Ltd (Japan). To purify anti-chmadrin antibodies, the antisera of these rabbits were first adsorbed against GST coupled to CNBr-activated Sepharose beads (Pharmacia), and then were incubated with GST-chmadrin (amino acids 2263-2603) coupled to CNBr-beads. Bound antibodies were then eluted with 0.1 M glycine-HCl (pH 2.2), neutralized immediately with 1 M Tris (pH 8.0) and dialyzed against PBS (137 mM NaCl, 3 mM KCl, 8 mM Na₂HPO₄ and 1 mM KH₂PO₄). The antibodies were further incubated with GST-Sepharose to adsorb residual traces of anti-GST antibodies.

Cells were rinsed in cold PBS in the presence of protease inhibitors (1 μg/ml of pepstatin, leupeptin and aprotinin and 1 mM PMSF), and lysed in boiling SDS sample buffer (Laemmli, 1970). Extracts were separated on a 2-15% gradient polyacrylamide gel (multigel 2/15, Daiichi Pure Chemicals Co., Ltd) and proteins were transferred to nitrocellulose membranes (Schleicher & Schuell, Keene, NH). The filters were blocked with blocking buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl and 3% skimmed milk) for 1 hour at room temperature, then incubated with affinity purified anti-chmadrin antibodies at 5 μg/ml in blocking buffer for 3 hours at room temperature. Bound antibodies were detected with alkaline phosphatase-conjugated goat antibodies to rabbit IgG (Bio-Rad) using standard methodology.

PtK₂ cells grown on coverslips were fixed with 4% formaldehyde in PBS for 10 minutes at room temperature, rinsed with PBS and permeabilized with 0.5% Triton X-100 in PBS for 5 minutes. The cells were placed in blocking solution (PBS containing 5 mg/ml BSA, 2% goat control serum and 50 mM glycine) for 1 hour at room temperature, followed by the same solution with antibodies for more than ten hours at 4°C (preimmune antibody, 2A11 and affinity purified anti-chmadrin antibody at 5 μg/ml, anti-HA monoclonal antibody 12CA5 (Boehringer-Mannheim) at 10 μg/ml and anti-nucleolin monoclonal antibody 4E2 (MBL) at 5 μg/ml). After washing with PBS, rabbit and mouse antibodies were detected with fluorescein isothiocyanate (FITC)-conjugated goat anti-rabbit IgG (1:100, TAGO) and RITC-conjugated goat anti-mouse IgG (1:100, TAGO), respectively. The coverslips were then washed again with PBS, counterstained with Hoechst 33342 (1 μg/ml in PBS) for 5 minutes and mounted in Mowiol containing triethylenediamine. The samples were examined using a DeltaVision deconvolution microscopy system (Applied Precision, Inc.) or an Axiophot microscope (Carl Zeiss, Inc.).

Most vectors used in this study were prepared from pME-HA-CaMKIV (Okajima et al., 1996), in which the CaMKIV cDNA is inserted as a SalI-NotI segment. Chmadrin cDNAs encoding the residues 1-493, 494-778, 1276-1462 and 1783-2264 were amplified by PCR with Cloned Pfu polymerase (Stratagene) using appropriate primer sets. Restriction sites (SalI-NotI) were incorporated in every primer set. The PCR products were digested with SalI and NotI and replaced with the CaMKIV cDNA in pME-HA-CaMKIV to yield pHA-(1-493), pHA-(494-778), pHA-CR and pHA-LR. An oligonucleotide cassette encoding the NLS of SV40 large T antigen, which was made by annealing ΔSN-NLS(+) (5′-TCGAGCCCAA-GAAGAAGCGCAAGGTGGTCGACCTGCAGGC-3′) and ΔSN-NLS(−) (5′-GGCCGCCTGCAGGTCGACCACCTTGCGCTTCTTC-TTGGGC-3′), was inserted into the SalI and NotI sites of pME-HA-CaMKIV to yield pHA-NLS. Note that the original SalI site in pME-HA-CaMKIV was disrupted and a new SalI site was created instead just after the nucleotides encoding NLS. Starting with pHA-NLS, pHA-NLS-LR was constructed using the same procedure described above for pHA-LR. pGFP-NLS-LR was originated from pEGFP-C1 (Clontech), which encodes enhanced green fluorescent protein. First, an oligonucleotide cassette, ΔXE-NLS, which was made by annealing ΔXE-NLS(+) (5′-TCGACCCAAGAAGAAG-CGCAAGGTGGAG-3′) and ΔXE-NLS(−) (5′-AATTCTCCACCT-

TGCGCTTCTTCTTGGG-3′), was inserted into the XhoI and EcoRI sites of pEGFP-C1 to yield pGFP-NLS. Then, the cDNA of chmadrin encoding the LR domain (residues 1783-2259) flanked by EcoRI and SalI sites was inserted into the EcoRI and SalI sites of pGFP-NLS to yield pGFP-NLS-LR. The cDNA sequences in all constructs were confirmed by DNA sequence analysis.

PtK₂ cells were plated onto 12 mm coverslips in 35 mm dishes 1-2 days prior to use. Plasmids were introduced using cationic-lipid mediated transfection using Lipofectamine (Gibco BRL) and transfection conditions as recommended by the supplier for 28-30 hours.

The chmadrin sequence data has been submitted to the DDBJ/EMBL/GenBank database under accession number AB013085.

A monoclonal antibody, termed 2A11, was originally prepared to recombinant human 70 kDa heat shock cognate protein (Hsc70). Unexpectedly, 2A11 decorated the surface region of mitotic chromosomes of potoroo PtK₂ cells, as evidenced by immunofluorescence (Fig. 1A). Immunoblotting analysis with a PtK₂ cell lysate showed that 2A11 recognized three proteins with apparent molecular masses of approximately 40, 70 and 300 kDa (Fig. 1E, lane 1). The 70 kDa protein appeared to be a potoroo Hsc70. Therefore, it was assumed that one or both of the strongly cross-reacted 40 and 300 kDa proteins were localized around the mitotic chromosomes. Since the characteristic perichromosomal staining was obtained only in the case of PtK₂ cells, and not in human or mouse cells, we attempted to identify the perichromosomal protein(s) by screening a PtK₂ cDNA expression library with 2A11.

Of the 9×10⁵ plaques screened, 13 overlapping clones were isolated and spanned about 8.3 kb of contiguous cDNA containing a putative ORF of about 7.8 kb. The authenticity of the translational start site was verified by 5′RACE using a pair of nested oligonucleotide primers which were near the first ATG obtained. Two 5′RACE products which differed in length by 40 bp at their 5′ ends were isolated as described in Materials and Methods, but in both cases, no other ATG was present in the sequences obtained. Moreover, the position of the first ATG was relevant to that of human pKi-67, which is similar to the cloned molecule especially in the N-terminal region (62% identity in the first 100 residues, details below). From this collective data, we concluded that the complete coding sequence was obtained.

To verify whether this cDNA encodes a perichromosomal protein, a rabbit polyclonal antibody (pAb-C) was raised against a bacterial recombinant protein corresponding to the putative C-terminal region of the polypeptide. Affinity purified pAb-C stained the perichromosomal region of mitotic chromosomes in a manner indistinguishable from 2A11 (Fig. 1B). The staining was abolished when pAb-C was preincubated with an excess amount of the recombinant protein used for immunization (data not shown). By immunoblotting analysis, pAb-C mainly recognized a ∼300 kDa protein (Fig. 1E, lane 2), which is, most likely, the same protein recognized by 2A11. A ∼210 kDa protein, commonly recognized by both antibodies, was likely to be a degradation product of chmadrin, since the intensity of the band increased when protease inhibitors were omitted during the protein extraction (not shown). From these observations, we concluded that the isolated cDNA clone actually encodes the ∼300 kDa perichromosomal protein. Because the characteristic staining pattern surrounding the mitotic chromosomes is reminiscent of ‘KUMADORI’, a red or blue cosmetic make-up around the eyes of Kabuki (Japanese classical play) actors, we designate the protein identified here ‘chmadrin’.

A full-length ORF of chmadrin reported here was assembled from three different overlapping cDNA clones and the 5′RACE product. Northern blotting of PtK₂ poly(A)⁺ RNA probed with the central portion of the cDNA (nucleotide 2463-4492) showed a hybridization signal around 8.4 kb long (Fig. 2A). When PtK₂ genomic DNAs, digested with BamHI or EcoRI, were probed with the same cDNA, only a single band was detected in each case (Fig. 2B), indicating that chmadrin is encoded by a single gene.

A putative ORF of chmadrin encodes a predicted polypeptide of 2,603 amino acids (Fig. 3A) with a calculated molecular mass of 291,864 Da and a pI of 9.7 (∼19% lysine + arginine + histidine). A database search revealed certain similarities to the Ki-67 antigen (pKi-67) in limited regions:

(1) the most N-terminal 100 residues (62% identity in amino acids), (2) residues 498-548 in chmadrin (67% identity), and

(3) residues 633-778 in chmadrin (54% identity) (Fig. 3B).

One of the most characteristic features of the chmadrin polypeptide is that it contains two repetitive sequence domains, which are clearly evident in the homology plot analysis with itself (Fig. 3C, left). One is a highly conserved repetitive sequence domain located in the central portion of the polypeptide, named the ‘chmadrin repeat domain’, which contains tandemly repeated 20 or 21-residue elements with an insertion of 19 residues at residue 1265-1283 (Fig. 3D). The consensus sequence of the chmadrin repeat (the bottom line of Fig. 3D) contains a possible phosphorylation site sequence TP (threonine and proline) for Cdc2 kinase (Nigg, 1995). Each element of the chmadrin repeats is highly similar to the ‘Ki-67 motif’ which is a highly conserved 22-residue element found in ‘Ki-67 repeats’ (repeats of 122-residue elements) of human pKi-67 (Fig. 3E).

The other is a loosely conserved repetitive sequence domain located in the C-terminal portion (residue approximately 1,700-2,600) of the polypeptide, referred to as the ‘LR domain’, since the functional significance of the LR (leucine and arginine) pairs found in the domain are suggested by a structural comparison between chmadrin and pKi-67 (see below). Each repeating unit of the LR domain of chmadrin contains eight relatively conserved residues (shown in Fig. 3F) flanked by 18-90 less-conserved amino acids. A careful comparison reveals that the motif having irregularly spaced LR pairs also exists in the C-terminal region of human pKi-67 (Fig. 3G). The region which contains this motif extends for about 900 residues in chmadrin, but only for about 220 residues in pKi-67. The similarity in this domain is shown in the boxed region of a homology plot (Fig. 3C, right) and the portion which can be best aligned is illustrated in Fig. 3H. The overall structural similarity between chmadrin and pKi-67 is well documented by the homology plot shown in Fig. 3C (right).

Additional motifs are found in the chmadrin polypeptide, including putative phosphorylation sites for various kinases (Hanks and Quinn, 1991), a basic amino-acid tract-like nuclear localization signal (NLS) (Dingwall and Lasky, 1991) beginning at residues 358, 505, 1618 and 1897, the type A motif for NTP binding (Saraste et al., 1990) at residues 104-111, an FHA (Forkhead-associated) domain which is implicated in protein-protein interaction (Hofmann and Bucher, 1995) at residues 27-77 and four strong PEST sites which are implicated in proteolysis (Rogers et al., 1986) at residues 784-806, 1241-1251, 1532-1542 and 1796-1806.

The distribution of chmadrin protein in PtK₂ cells was examined through their cell cycle by indirect immunofluorescence using affinity purified anti-chmadrin antibodies (pAb-C). The staining pattern was very similar to that of pKi-67 (Gerdes et al., 1984; Kill, 1996; Bridger et al., 1998). During interphase, two staining patterns were obtained. In a large population of nuclei, chmadrin was localized mainly in the nucleolus, but also, to some extent, in the nucleoplasmic foci (Fig. 4A). In certain nuclei, nucleoplasmic staining was more evident and extended even to the nuclear envelope (Fig. 4B). In both types of nuclei, the nucleoplasmic staining was overlapped with regions which were densely stained with Hoechst 33342, probably heterochromatin. This type of localization in both nucleoli and heterochromatic foci has also been described for pKi-67 (Bridger et al., 1998), although no specific function(s) at each region has been described. At the onset of mitosis, when the chromosomes became compacted, chmadrin was relocated to the periphery of the chromosomes and all chromosomes were coated with chmadrin (Fig. 4C). Chmadrin remained bound to the periphery of the chromosomes through mitosis (Fig. 4D-F). In the nuclei of daughter cells just after mitosis, chmadrin was detected in both the nucleoplasm and the presumptive new nucleoli. Nucleoplasmic chmadrin was detected in the form of discrete dotted structures (Fig. 4G).

To begin the functional characterization of chmadrin, we were initially interested in identifying the regions of chmadrin which are involved in cell-cycle dependent subcellular localization. Since chmadrin behaves in a manner similar to pKi-67 throughout the cell cycle, the essential regions for the localization would be expected to reside in the regions which are conserved between chmadrin and pKi-67. Following this consideration, we constructed four expression plasmids encoding the HA epitope-tagged truncated forms of chmadrin depicted in Fig. 5A, each of which shows certain structural similarity to pKi-67 as described above. These plasmids were transiently introduced into PtK₂ cells, and then the subcellular localization of expressed proteins were examined by using anti-HA monoclonal antibody. Of four proteins, HA-(494-778) was localized preferentially to the interphase nucleoli (Fig. 5G′), but not to heterochromatic foci as was observed for endogenous chmadrin (Fig. 4A,B). It should be noted that a small population of HA-(494-778) appeared to approach the mitotic chromosomes (Fig. 5C′). HA-LR, which contains eleven repetitive structural units, targeted precisely to the body of mitotic chromosomes but not perichromosomal region (Fig. 5E′), suggesting that the LR domain is necessary, but not sufficient, for the perichromosomal targeting of chmadrin. Both HA-(1-493) and HA-CR were localized in nuclei in interphase (Fig. 5F′,H′) and diffusely localized in mitotic phase (Fig. 5B′,D′). The mechanism by which chmadrin is well organized in the perichromosomal layer could not be addressed in this series of experiments.

Through the above experiments, we noticed that the expression of HA-LR, but not the other constructs, brought about slight changes in chromatin structure (Fig. 5I). The higher the level of expressed protein, the stronger the effect tended to be (Fig. 6A-C). HA-LR, having a calculated molecular mass of 56 kDa, could migrate into nuclei, probably via the association with a limited number of other nuclear proteins, since HA-LR seemed to migrate into nuclei in a saturable manner (a great excess of the protein apparently resides in the cytoplasm, Fig. 6C). Supposing that HA-LR protein translocated into the nuclei has an impact on chromatin structure, the impact would increase by compelling the protein to migrate into nuclei with the aid of an NLS (nuclear localization signal). To test the idea, we examined the effect of HA-NLS-LR, in which the NLS from the SV40 large T antigen was inserted between the HA tag and the LR domain. As expected, the HA-NLS-LR migrated into nuclei efficiently and drastically induced the characteristic alteration of chromatin structure (Fig. 6D,E), which seemed to be a ‘heterochromatin-like structure’ as judged by its strong staining with Hoechst 33342. Even in the cells that contained nuclei similar to those shown in Fig. 6D, microinjected NLS-containing proteins were normally transported into the nuclei (data not shown). When the microinjection technique was employed for the nuclear introduction of the expression plasmid, all cells showed an aberrant compaction of chromatin by 14 hours. In spite of its strong impact on the chromatin structure, the overproduced HA-NLS-LR was not detected at the induced heterochromatin-like structure by the anti-HA antibody (Fig. 6F). Since there is a possibility that immunodetection is hindered by the difficulty of antibody access to the antigens, we also used GFP (green fluorescent protein) as a probe for the localization. GFP-NLS-LR, in which GFP was fused with NLS and the LR domain in this order, also induced the characteristic compaction of chromatin, and was detected precisely on the induced structure both before (data not shown) and after fixation (Fig. 6G-I). The localization of the expressed LR domain which had been probed with GFP might be more realistic than that probed with anti-HA antibody, since probing with GFP avoids the necessity of fixation, permeabilization and the immunodetection steps, which potentially obscure the actual localization. It is not certain, however, whether the apparent targeting of GFP-NLS-LR to nucleoli (Fig. 6I) reflected the intrinsic nature of the LR domain, since a small but significant population of GFP-NLS was also observed in nucleoli (Fig. 6M).

In control cells expressing GFP-NLS, no alteration of chromatin structure was observed (Fig. 6L), confirming that the effect observed after the overproduction of GFP-NLS-LR was produced by the LR domain itself. Immunofluorescence using an antibody against nucleolin, a marker protein of the nucleolus, showed that the structure induced by the overproduction of the LR domain was not fragmented nucleoli (Fig. 6J). We also noted that the localization of nucleolin shifted to the periphery of nucleoli and the nucleoplasmic space after the overproduction of the LR domain (compare Fig. 6J and 6N), suggesting that the LR domain could have some impact on the nucleolar structure, either directly or indirectly.

We report herein the molecular identification and characterization of a novel perichromosomal protein, chmadrin, from PtK₂ cells along with the identification of functional domains required for directing the protein to interphase nucleoli and mitotic chromosomes. It is also noteworthy that a structural domain, which is shown to be essential for targeting to mitotic chromosomes (called LR domain), caused the formation of a heterochromatin-like structure, when overproduced in interphase nuclei. Collectively, these data suggest that the chmadrin protein may play a role in higher order chromatin organization.

Chmadrin was identified via the use of a monoclonal antibody 2A11, originally raised against human Hsc70. Since cDNA clones obtained by immunoscreening with 2A11 had sequences which overlapped with the central portion of the chmadrin protein, it is likely that the 2A11 epitope resides in the overlapping region. The authenticity of the cDNA was confirmed as follows. The calculated size of the protein encoded by the cDNA is in close agreement with the size of the protein recognized by 2A11 and pAb-C (a newly generated antibody against the putative C-terminal portion of the cDNA) in western blotting (Fig. 1E). The action of pAb-C in recognizing the periphery of mitotic chromosomes was indistinguishable from 2A11, and was abolished by preincubation with the antigenic recombinant protein. The 5′-end of the cDNA was also confirmed by 5′-RACE analysis.

It should also be noted that the cDNA sequence described here contains 15-times chmadrin repeat units. We also obtained a clone having 19 chmadrin repeat units whose sequence is identical to that of the genomic DNA (data not shown). These isoforms appear to be derived by alternative splicing from a single gene. The question of whether each isoform plays a distinct role and the issue of how its expression is alternatively regulated remain to be elucidated.

An analysis of the primary structure of the chmadrin protein clearly shows its similarity to pKi-67. The similarity is found only in limited regions including the N-terminal region, the chmadrin repeat domain and the LR domain. The NTP binding motif is commonly found in both proteins, although in different positions; namely the N-terminal region of the chmadrin protein and the C-terminal region of pKi-67. In addition to their primary structures, both proteins show similar subcellular localization patterns throughout the cell cycle, which strongly suggests that they may have some analogous functions.

In fact, the ability to induce aberrant chromatin compaction by the LR domain of chmadrin observed in this study was also detected by the C-terminal region of human pKi-67 (unpublished observation). As shown in the structural comparison in Fig. 3F-H, however, the similarity between the C-terminal regions of both proteins is quite low, which is rather advantageous in predicting precise sequence requirements for the compaction of chromatin.

The structural basis for the specific cellular localization of the chmadrin protein was uncovered, in part, via a series of transient expression analyses, in which a structural comparison with pKi-67 was helpful. Nucleolar localization activity during interphase can be roughly assigned to residues 494-778 of the protein, which contains basic amino-acid tracts similar to the NoLS (nucleolar localization signal) found within other nucleolar proteins such as p120, B23 and several viral proteins (Valdez et al., 1994; Mears et al., 1995; Li et al., 1996). Chromosomal targeting activity was found to reside in the LR domain. The LR domain has no similarity to any known motifs, suggesting a role in chromatin targeting. In addition to the LR domain, some other regions might be required for the perichromosomal localization of chmadrin. The region harboring the nucleolar targeting activity (residues 494-778) is a candidate region, since this region was found to come close to mitotic chromosomes when overexpressed (Fig. 5C′).

The dynamic relocation from nucleoli to the perichromosomal region at the onset of mitosis is one of the most characteristic behaviors common to chmadrin and pKi-67. It is noteworthy that both the chmadrin repeat domain and the Ki-67 motif contain a consensus phosphorylation site (T/S)P for cell cycle-dependent kinases (Nigg, 1995). Phosphorylation at these sites may explain the mechanism for the temporal regulation of their transition between the nucleoli and the perichromosomal region. The balance of the localization between the nucleoli and heterochromatic foci within interphase nuclei may also be regulated by phosphorylation.

Our transient expression analysis showed that the overexpressed LR domain in PtK₂ cells caused the formation of an aberrant heterochromatin-like structure in interphase nuclei, which was enhanced when the LR domain was compelled to migrate into nuclei by fusing with the NLS. Since the impact on the chromatin structure was observed independently of the type of probes used for the localization (both when using HA and GFP), the impact was actually brought about by the LR domain. We propose that the LR domain directly induced the formation of a heterochromatin-like structure as a structural constituent, since the GFP-NLS-LR was localized precisely to this structure. The failure to detect HA-NLS-LR at the induced heterochromatin by anti-HA antibody might be explained by the protein topology within the highly compacted chromatin structure and the resulting inaccessibility of the antibody to antigens, although the view should be supported in future studies. In contrast, HA-LR targeting to mitotic chromosomes could be easily detected by the anti-HA antibody, which may reflect differences in the mode of chromatin targeting of the LR domain in interphase and mitotic phase. Although further studies will be required to absolutely verify that the structure induced by the overproduction of the LR domain is physiologically relevant to ‘heterochromatin’ and that the activity of the LR domain described here reflects the intrinsic activity of chmadrin, chmadrin might play a role in the organization of higher order chromatin structures through the interaction of the LR domain with chromatin. The assumed role of chmadrin would not be limited to mitotic chromosomes but might be expanded to interphasic heterochromatin, since a population of chmadrin does exist at the heterochromatic foci in interphase nuclei.

For a long period, the notion that perichromosomal proteins might participate in higher order chromatin organization has been reiterated, in the absence of experimental data. In this study, we empirically support this suggestion for the first time by demonstrating that a portion of chmadrin, referred to as the LR domain, has a potent activity for inducing the compaction of chromatin. Further studies on the chmadrin protein will shed new light on our understanding of the fashion of the organization of higher order chromatin structure such as mitotic chromosomes and the interphasic heterochromatin.

We are grateful to Mrs J. N. Takagi for her secretarial work. This work was supported by a Grant-in-Aid for Scientific Research on Priority Areas (No. 07282103), Grant-in-Aid for Scientific Research (B) (No. 08458229) and Grant-in-Aid for COE Research (No. 07CE2006) from the Japanese Ministry of Education, Science, Sports and Culture and the Nissan Science Foundation. M.T. was a research fellow of the Japanese Society for the promotion of Science.