Cell cycle behavior of human HP1 subtypes: distinct molecular domains of HP1 are required for their centromeric localization during interphase and metaphase

Eukaryotic chromosomes consist of cytologically defined domains of euchromatin and heterochromatin. Euchromatin contains transcriptionally potent genes which are subject to transcriptional activation or inactivation. In contrast, heterochromatin contains transcriptionally silent genes. Constitutive heterochromatin also has specialized chromosome structures such as centromeres and telomeres.

Although the molecular details that define euchromatin and heterochromatin are not fully understood, it is now clear that heterochromatin protein 1 (HP1)is a major constituent of heterochromatin and plays a key role in its formation and maintenance. HP1 is highly conserved from yeasts to humans(Eissenberg and Elgin, 2000). HP1 was first identified as a gene product of an allele of Su(var)2-5in Drosophila melanogaster (James and Elgin, 1986; Eissenberg et al., 1990). In Drosophila, HP1 is localized to pericentric and telomeric heterochromatin. When mutated HP1 was expressed in Drosophila, abnormal condensation and segregation of chromosomes was observed (Kellum and Alberts,1995). Involvement of HP1 in heterochromatin formation is now known to occur through its interaction with SUV39H1, the histone methyltransferase that methylates histone H3 at lysine 9. HP1 binds to histone H3, which has been methylated at lysine 9 by SUV39HI, and in turn it recruits SUV39H1 to the DNA, which further propagates methylation along the chromatin(Jenuwein, 2001). This relationship between HP1 and SUV39H1 is conserved in their Schizosaccharomyces pombe homologues Swi6 and Clr4(Bannister et al., 2001),suggesting evolutionary conservation of this mechanism of heterochromatin formation.

In humans, three subtypes of HP1 have been identified, HP1α,HP1β, and HP1γ (Singh et al.,1991; Saunders et al.,1993; Ye and Worman,1996). They share significant similarities in their amino acid sequences (see Fig. 7). HP1 consists of two highly conserved regions, the N-terminal chromo domain (CD)(Paro and Hogness, 1991) and the structurally related C-terminal chromo shadow domain (CSD)(Aasland and Stewart, 1995; Cowieson et al., 2000). These two domains are connected by a less well conserved hinge region (IVR) (see Fig. 7). The CSD is a unique motif found only in the HP1 family; the CD motif, however, is also found in Su(var)3-9 family proteins and Polycomb (Pc) family proteins(Jacobs et al., 2001).

HP1 seems to have pleiotropic functions in the genetic activities of chromosomes. It has been shown that mammalian HP1 homologues have a number of interacting partners such as the CAF-1 subunit p150(Murzina et al., 1999), SP100(Lehming et al., 1998; Seeler et al., 1998),TIF1-β (KAP-1) (Le Douarin et al.,1996; Ryan et al.,1999), Ku70 (Song et al.,2001), lamin B receptor (Ye and Worman, 1996; Ye et al.,1997), SUV39H1 (Aagaard et al.,1999), Ki-67 (Kametaka et al.,2002; Scholzen et al.,2002), histone H1-like protein(Nielsen et al., 2001),methylated histone H3 (Bannister et al.,2001; Lachner et al.,2001) and INCENP (Ainsztein et al., 1998). These proteins have a wide variety of functions involved in chromatin assembly, transcriptional regulation, telomere maintenance, and nuclear membrane formation. Many of the HP1 interacting partners listed above bind to the CSD of HP1, but interaction of lysine 9-methylated histone H3 with the CD region(Lachner et al., 2001) and INCENP with the IVR region (Ainsztein et al., 1998) have also been documented.

HP1 subtypes have shown heterogeneous patterns of intracellular localization in accordance with the presence of a wide variety of interacting molecules. Localization of HP1 subtypes during the cell cycle have been extensively studied in human and mouse cells(Wreggett et al., 1994; Horsley et al., 1996; Furuta et al., 1997; Minc et al., 1999; Yamada et al., 1999; Minc et al., 2000; Sugimoto et al., 2001). In interphase nuclei, HP1 subtypes are localized in heterochromatin(Wreggett et al., 1994; Minc et al., 1999; Minc et al., 2000) in accordance with their interaction with SUV39H1(Aagaard et al., 1999), and localized at the PML nuclear bodies in interphase nuclei(Everett et al., 1999) in accordance with their interaction with SP100, a component of the PML nuclear body (Lehming et al., 1998; Seeler et al., 1998); HP1 subtypes are also localized at the periphery of nucleoli(Minc et al., 1999). In metaphase, HP1α is localized at the centromeric region(Minc et al., 1999; Yamada et al., 1999; Minc et al., 2001), as is its interacter INCENP (Ainsztein et al.,1998). Furthermore, live cell imaging of GFP-fused HP1αthroughout the cell cycle in human cells demonstrated that HP1α is colocalized with DsRed-fused CENP-B at the centromere during metaphase(Sugimoto et al., 2001). Compared with the reproducible results reported for HP1α, localization of HP1β and γ at the metaphase centromere has been debatable,depending on cell types and antibodies used(Wreggett et al., 1994; Minc et al., 1999; Saffery et al., 1999; Minc et al., 2000; Minc et al., 2001).

To determine molecular domains of HP1 subtypes required for their intracellular localization, we examined human HP1 subtypes and various truncation constructs in living HeLa cells, and determined the molecular domains responsible for their localization to centromeres and PML nuclear bodies. Here we report that HP1β associates with interphase centromeres but is replaced at the centromere by HP1α as the cell enters metaphase. The N-terminal region of HP1β including the CD, is responsible for localization to the interphase centromere and the C-terminal region of HP1α including the CSD and a portion of the IVR, is responsible for localization to the metaphase centromere. Simultaneous observations of the N-terminal fragment of HP1β and the C-terminal fragment of HP1α in living cells revealed that during late prophase the HP1β fragment dissociated from the centromere and as the cells entered metaphase the HP1α fragment accumulated at the centromere. Thus, localization of human HP1β and HP1α to centromeric heterochromatin during different phases of the cell cycle is mediated by distinct regions of these proteins. These results suggest specific roles for HP1β and HP1α at the interphase and metaphase centromere, respectively.

HeLa cells were obtained from the Riken Cell Bank. Cells were grown at 37°C in Dulbecco's Modified Eagle Medium (DMEM) (Invitrogen) containing 10% calf serum (Invitrogen). Human anti-centromere antisera was derived from a patient with scleroderma (Masumoto et al.,1989). Anti-PML monoclonal antibody was purchased from Santa Cruz Biotech (Santa Cruz, CA, USA). Species-specific, fluorescent secondary antibodies were purchased from Jackson Immunoresearch Laboratories (West Groove, PA, USA). Hoechst 33342 and 4,6-diamidino-2-phenylindole (DAPI) were purchased from Calbiochem (La Jolla, CA, USA). An anti-fluorescence fading reagent, 1,4-diazobicyclo-(2,2,2)-octane (DABCO), was purchased from Sigma-Aldrich (St. Louis, MO, USA).

The open reading frames (ORF) of human HP1α (GenBank accession number: NM_012117.1), HP1β (GenBank accession no.: NM_006807.2) and HP1γ (GenBank accession no.: AB030905.1) were cloned by RT-PCR using mRNA isolated from HeLa cells as the PCR template. Total RNA was isolated from HeLa cells using TORIZOL (Invitrogen) according to the manufacturer's instructions. Complementary DNAs were amplified with Pyrobest DNA polymerase (TaKaRa, Ohtsu, Japan) using specific primers for the ORF of each gene; the EcoRI linker was added to the 5′ primer, and a stop codon plus the BamHI linker was added to the 3′ primers. Synthesized cDNAs were digested with EcoRI and BamHI, then cloned into the pEYFP-C1, pECFP-C1 or pDsRed2-C1 vector (Clontech, Palo Alto,CA, USA). Truncated constructs of human HP1 were produced by PCR-based amplification using its ORF as the template. All clones were checked for their nucleotide sequence using ABI310 or ABI377 (Applied Biosystems Inc., Foster City, CA, USA).

The plasmids pGFP-CB1-160c and pCFP-CB1-160c, containing the N-terminal fragment of human CENP-B fused with GFP and CFP, respectively, were generated as follows: The N-terminal fragment of CENP-B (–2 to 481 bp) was excised from pETCBN-160 (Kitagawa et al.,1995) with NcoI (the NcoI digestion product was blunt-ended using Klenow) and BamHI. A BamHI-HindIII adaptor was prepared by annealing oligonucleotides 5′-GATCCATGGCTAGCATGACTGGTGGACAGCAAATGGGTCGGTAGTCGACA-3′ and 5′-AGCTTGTCGACTACCGACCCATTTGCTGTCCACCAGTCATGCTAGCCATG-3′. The CENP-B fragment with the adaptor was cloned into SalI (the SalI digestion product was blunt-ended using Klenow)-HindIII-cut pMTID13S(Stuart et al., 1984) to create pMTCB1-160c. Then, to create the plasmid pGFP-CB1-160c, the N-terminal fragment of CENP-B was excised from pMTCB1-160c with XbaI and HindIII (the HindIII digestion product was blunt-ended using Klenow) and introduced into XbaI-BclI-digested and blunt-ended pGFP-C1 (Clontech). To create the plasmid pCFP-CB1-160c, the N-terminal fragment of CENP-B was excised from the plasmid pGFP-CB1-160c with SalI and BamHI, and introduced into SalI-BamH-digested pECFP-C1 (Clonetech).

Transfection reactions were carried out with LipofectAmine PLUS(Invitrogen). HeLa cells cultured in a 35 mm glass-bottom culture dish(MatTek, Ashland, MA, USA) were transfected with 250 ng of plasmid DNA, and incubated at 37°C for 1.5 hours. For co-transfection with two or three kinds of plasmid DNA, 125 ng or 85 ng DNA of each plasmid were applied,respectively. Transfected cells were incubated in fresh DMEM medium containing 10% calf serum for 1 or 2 days before observation.

Indirect immunofluorescent staining was carried out essentially as described previously (Haraguchi et al.,2000). Cells cultured in a 35 mm glass-bottom dish were fixed with a mixture of 3.7% formaldehyde and 0.2% glutaraldehyde in DMEM culture medium for 20-30 minutes at room temperature. Fixed cells were then treated twice with 0.1% sodium borohydrate in phosphate-buffered saline (PBS) for 15 minutes at room temperature. The cells were then permeabilized with 0.1% Triton X-100 in PBS for 5 minutes, followed by incubation in 1% bovine serum albumin in PBS at room temperature for 1 hour. Anti-centromere human antisera and anti-PML mouse monoclonal antibody were then added at a dilution of 1:100 and 1:200,respectively, and incubated for 3 hours at room temperature. After the first antibody reaction, cells were washed five times with PBS for 10 minutes, and stained for 3 hours at room temperature with the relevant fluorescent secondary antibodies: Cy3-conjugated goat anti-mouse IgG or Cy5-conjugated goat anti-human IgG, both at a dilution of 1:200. The stained cells were washed five times with PBS for 10 minutes and then taken sequentially with 20%, 40%, 60% and 80% glycerol in PBS containing 0.5 μg/ml DAPI and 2.5%DABCO. Finally, cells were mounted in 90% glycerol for microscopic observation.

observation of live cells was carried out as previously described(Haraguchi et al., 1997; Haraguchi et al., 1999). Briefly, cells cultured in a 35 mm glass-bottom dish were stained with 100 ng/ml Hoechst 33342 for 30 minutes and then washed twice with DMEM containing 10% calf serum. The Hoechst 33342-stained cells were then cultured in phenol red-free DMEM supplemented with 10% fetal bovine serum (Invitrogen),antibiotics (50 unit/ml penicillin and 50 unit/ml streptomycin; purchased from Invitrogen), and 20 mM Hepes pH 7.4. These cells were incubated in a CO₂ incubator for at least 30 minutes and then placed on a microscopic stage in a temperature-controlled room. Mineral oil was layered onto the culture medium to avoid evaporation during microscopic observation. After 30 minutes of culture on the microscope stage, observation was carried out using a DeltaVision microscope system (Applied Precision, Issaquah, WA,USA) kept at 37°C in a temperature-controlled room(Haraguchi et al., 1999). The oil-immersion objective lens UApo40x/NA1.35 (Olympus Opitcal, Tokyo, Japan)was used for the observation.

HeLa cells cultured in a 60 mm culture dish were transfected with 750 ng DNA for 1.5 hours using LipofectAmine PLUS. The transfected cells were then cultured in DMEM supplemented with 10% calf serum at 37°C for 24 hours. Thymidine was added to the culture at a final concentration of 2.5 mM to synchronize the cell cycle. After incubation in the thymidine-containing medium for 18 hours, the cells were washed twice with DMEM and cultured in DMEM supplemented with 10% calf serum at 37°C for 8 hours to allow cells to proceed into mitosis. Mitotic cells were enriched by collecting cells that detached from the dish by gentle pipetting. Cells with no mitotic enrichment were collected using a rubber scraper. For hypotonic treatment, the collected cells were washed once with 75 mM KCl that had been pre-warmed to 37°C,and incubated for 20 minutes at 37°C. Mitotic cells were collected by pipetting, resuspended in 5 ml of 75 mM KCl and kept on ice. Then an equal volume of ice-cold buffer A (15 mM Hepes pH 7.4, 80 mM KCl, 20 mM NaCl, 0.5 mM EGTA, 2 mM EDTA, 0.5 mM spermidine, 0.2 mM spermine and 0.1%2-mercaptoethanol) (Hiraoka et al.,1990) containing 0.1% Triton X-100 (buffer A-T) was added to the cells while on ice. Cells were centrifuged, washed and then resuspended with buffer A-T at 4°C. The cells were placed on a glass slide, DAPI added for DNA counterstaining, and the cells were covered with a coverslip. Chromosome spreads were prepared by gently tapping the coverslip.

Fluorescence microscope images were obtained on a Peltier-cooled CCD using a DeltaVision microscope system. Details of the microscope system have been described previously (Haraguchi et al.,1997; Haraguchi et al.,1999). For high-resolution analysis of fixed specimens, the oil-immersion objective lens PlanApo60x/NA1.4 (Olympus Opitical) was used. Three-dimensional optical section images were taken at 0.5 μm focus intervals and computationally processed by an iterative deconvolution method(Agard et al., 1989).

We examined the localization of YFP-fusion constructs of human HP1 subtypes in HeLa cells. HeLa cells transiently expressing YFP-fused HP1 were fixed and observed under a fluorescence microscope. In interphase nuclei, YFP-fused HP1 showed two different staining patterns: one, a series of bright regularly shaped spots; and the other, less bright irregularly shaped areas with some diffuse staining of the nucleoplasm. HP1β was predominantly found in the irregularly shaped bodies associated with centromeric regions(Fig. 1A, arrowhead), and HP1α and γ were found also in the round bright spots of the PML nuclear bodies (Fig. 1A,arrow). During metaphase, HP1 subtypes diffused throughout the cytoplasm, and were excluded from the chromosomal region(Fig. 1B). HP1α, however,did show spot staining on the metaphase chromosomes(Fig. 1B,b); HP1β andγ did not (Fig. 1B,e,h). Immunofluorescence staining confirmed that the spot-like staining pattern of HP1α on metaphase chromosomes was at the centromere(Fig. 1C).

Localization of YFP-fused HP1 subtypes was also examined on spread preparations of metaphase chromosomes. When metaphase chromosome spreads were prepared from cell populations enriched with mitotic cells HP1α was clearly present at the centromere (Fig. 2a) whereas HP1β and γ showed faint or no localization at the centromere (Fig. 2c and e, respectively). In contrast, when chromosome spreads were prepared from whole cell populations with no enrichment, HP1β showed a somewhat increased tendency to be associated with the centromere(Fig. 2d) although its localization was much fainter than that of HP1α. Quantitative data is summarized in Table 1. The apparent centromere localization of HP1β was the result of interphase cells being present in the preparations whereas localization of HP1α andγ was not significantly affected by the presence of interface cells. This would suggest that relocalization of HP1β occurs during the cell cycle and may explain the discrepancies in the reported localization of HP1β at the metaphase centromere (see Discussion).

Taken together, these results show that YFP-fused HP1 subtypes change their intracellular localization from interphase to metaphase. All HP1 subtypes associate with interphase centromeres, but only HP1α shows strong association with metaphase centromeres. These intracellular localization patterns of YFP-fused HP1 are consistent with the previous observations obtained using antibodies or GFP-fusion constructs(Minc et al., 1999; Sugimoto et al., 2001) (see Introduction). Thus, we determined molecular domains required for the intracellular localization using truncation constructs of the YFP-fusion HP1 subtypes in HeLa cells.

As HP1α was uniquely localized to the metaphase centromere, we determined the necessary molecular domains of HP1α required for this localization. To this end, we constructed a series of truncated HP1αpeptides tagged with YFP (Fig. 3A). First, we divided HP1α into three domains, namely the N-terminal region, consisting of residues 1-75, which contains the CD; the C-terminal region consisting of residues 121-191, which contains the CSD; and the IVR region, consisting of residues 76-120. However, none of these domains was found to be localized to the metaphase centromere (data not shown).

We then examined residues 1-170 containing the CD and the IVR and residues 76-191 containing the IVR and the CSD. We found that the latter peptide was localized to the metaphase centromere (Fig. 3A; image data not shown). Further experiments using truncations of HP1α determined that the peptide consisting of residues 101-180 was localized to the metaphase centromere, whereas the peptide consisting of residues 111-180 was not (Fig. 3B), indicating that residues 101-180 of HP1α is the minimum region necessary for localization of HP1α to the metaphase centromere. Interestingly, residues 98-176 of HP1β and residues 83-170 of HP1γ,which correspond to residues 101-180 of HP1α, were also localized to the metaphase centromere (Fig. 4B, Fig. 5B) in spite of the fact that neither full-length HP1β nor HP1γ was localized to the metaphase centromere (Fig. 1B),suggesting that this region of HP1 has a fundamental role in the localization to the metaphase centromere.

HP1 peptides that were localized to the centromere at metaphase were not localized to the centromere in interphase (HP1α 101-180, compare Fig. 3C with Fig. 3B; HP1β 98-176,compare Fig. 4C with B;HP1γ 83-170, compare Fig. 5C with B). Instead they localized exclusively to PML nuclear bodies (data not shown). The shorter fragments containing only the CSD rgion were also present (HP1α 121-191 in Fig. 3D; HP1β 117–185 in Fig. 4D; HP1γ 111-173 in Fig. 5D). Thus, the CSD portion of HP1 is sufficient for localization to PML nuclear bodies in interphase,whereas the additional IVR residues are required for localization to the metaphase centromere.

As the residues of HP1 that were localized to the metaphase centromere were not localized to the interphase centromere, we examined the molecular domains of HP1 necessary for localization to the centromere within the interphase nucleus. Our results showed that in HP1α and HP1β the N-terminal fragments containing the CD were localized to the interphase centromere, but this fragment in HP1γ was poorly localized (HP1α 1-75 in Fig. 3C; HP1β 1-76 in Fig. 4C; HP1γ 1-75 in Fig. 5C). These results indicate that the domains of HP1 required for centromeric localization during interphase are different from those required for centromeric localization during metaphase.

Of the three HP1 subtypes HP1β was localized to the interphase centromere most strongly, whereas HP1α was the only subtype localized to the metaphase centromere. In order to examine the switching of HP1α and HP1β during the cell cycle, we attempted to observe HP1α and HP1β simultaneously in living cells. To avoid possible interaction between HP1α and HP1β through their CSD(Le Douarin et al., 1996; Ye et al., 1997), we used the N-terminal fragment of HP1β (residues 1-76) and the near C-terminal fragment of HP1α (residues 101-180); these fragments do not interact with each other. Fig. 6 shows an example of HeLa cells expressing the residues 1-76 of HP1β fused with YFP, the residues 101-180 of HP1α fused with DsRed2, and the N-terminal fragment of CENP-B fused with CFP. Between interphase and prophase, the HP1α fragment was not localized to the centromere, but instead localized to PML nuclear bodies (Fig. 6,frames at 0:00-1:44). However, the HP1β fragment did localize to the centromere, stained with CENP-B-CFP, during this period(Fig. 6, frames at 0:00-1:02). In late prophase, the HP1β fragment dissociated from the centromere(Fig. 6, frames at 1:44-2:01),and, in the same cell, the HP1α fragment accumulated at the centromere just prior to metaphase (Fig. 6, frame at 2:29). During anaphase, the HP1α fragment dissociated from the centromere (Fig. 6, frame at 2:34) and the HP1β fragment began to accumulate first on the chromosome (Fig. 6, frame at 2:45), and then on the centromere(Fig. 6, frame at 2:51). Accumulation of the HP1α fragment at the midbody was also observed during cytokinesis (Fig. 6,frames at 2:45 and 2:51). Full-length HP1α also showed accumulation at the midbody (data not shown). By the next interphase, the HP1α fragment reformed foci with PML nuclear bodies and the HP1β fragment relocalized to centromeric regions (an example for later interphase is shown in Fig. 6, frame at 16:09). These results demonstrate that as a result of the interaction of different domains of human HP1 proteins with interphase and metaphase centromeric regions,HP1β is localized to the centromere during interphase and is replaced by HP1α during metaphase (summarized in Fig. 7; see Discussion).

In this study, we determined the molecular domains of human HP1 that are required for the intracellular localization of its subtypes. While localization of full-length HP1 is heterogeneous over several different intracellular structures, truncated domains of HP1 showed distinct localization to specific structures (summarized in Fig. 7). Within the interphase nucleus, the CD region of HP1 localizes to the centromere and other heterochromatin, but does not associate with PML nuclear bodies, whilst the CSD region localizes exclusively to PML nuclear bodies. Interestingly, in metaphase, centromere localization requires the CSD but not the CD. Heterogeneous localization of HP1 reflects the presence of its various interacting molecules, and the distinct localization of different domains reflects domain-specific binding partners. We speculate that centromeric localization of HP1 probably involves distinct binding partners during interphase and metaphase.

In interphase, all three human HP1 subtypes localize to centromeric regions and other heterochromatin regions to varying extents. Of the subtypes,HP1β most preferentially localized to heterochromatin. We determined that the N-terminal fragment containing the CD was the molecular domain of HP1 required for localization to interphase centromeric heterochromatin. Localization of the CD region of HP1 to heterochromatin probably reflects the fact that the CD of HP1 binds to lysine 9 methylated histone H3(Bannister et al., 2001; Lachner et al., 2001) while a CSD interacting partner, SUV39H1, methylates lysine 9 of histone H3(Rea et al., 2000). It is now hypothesized that the CD of HP1 recognizes modified histones, the `histone code', and recruits a variety of proteins through interaction with the CSD to establish chromatin structure (Jenuwein and Allis, 2001).

Our results also showed that all three human HP1 subtypes were localized to the PML nuclear bodies, as previously reported(Everett et al., 1999). HP1α and γ were preferentially localized to PML nuclear bodies. The PML nuclear body is an intranuclear structure that contains a number of proteins, such as PML, SP100, ISG20, PIC1/SUMO1, LYSP100, PLZF, INT6, CBP,RB1, RFP and ribosomal protein P (Lamond and Earnshaw, 1998; Seeler and Dejean, 1999). It has been shown that HP1 interacts with SP100,one of the major components of the PML nuclear body(Lehming et al., 1998; Seeler et al., 1998; Seeler and Dejean, 1999). In this study, we determined that the CSD is the molecular domain of HP1 required for localization to PML nuclear bodies. This is consistent with the fact that the CSD portion of HP1 interacts with SP100(Lehming et al., 1998; Seeler et al., 1998).

In metaphase, HP1α localized to the centromere whereas HP1β and HP1γ did not. This result is consistent with previous observations obtained from antibodies or GFP fusion constructs(Sugimoto et al., 2001). However, it has also been reported that HP1β localizes to the centromere of some, but not all, metaphase chromosomes(Saffery et al., 1999) and that HP1γ is localized to the metaphase centromere(Minc et al., 2000). Furuta et al. (Furuta et al., 1997)demonstrated that centromere localization of HP1β is observed at metaphase, but is most prominent at anaphase, suggesting cell cycle regulations. We also occasionally observed some faint centromere staining for HP1β or HP1γ in metaphase chromosome spreads that also contained some interphase cells (Fig. 2; Table 1), suggesting that their localization may be affected by unknown factors from interphase cells. Since our experiments of HP1 truncations showed that all HP1 subtypes are potentially capable of binding to the metaphase centromere, centromere localization of HP1 is probably regulated during mitosis by protein modification or interaction with other proteins. Also their localization may depend on heterochromatin structures, which vary in different cell types.

We defined the domain of HP1α that is required for its localization to the metaphase centromere. This domain consists of a portion of the IVR and the CSD region, as opposed to the CD which is required for localization to the interphase centromere (Fig. 7). Replacement of an HP1β peptide containing the CD region with an HP1α peptide containing the IVR/CSD region at the centromere during mitosis was clearly demonstrated by observation of live cells (see Fig. 6). Interestingly, the corresponding IVR/CSD domain of HP1β and HP1γ was also localized to the metaphase centromere whereas full-length HP1β or HP1γ was not. These results show that HP1β and HP1γ potentially have the capability of localizing to the metaphase centromere, but the neighboring domains somehow interfere with metaphase centromere localization. These results also suggest that their localization at metaphase centromere is regulated by small variations of amino acid sequences or modifications such as phosphorylation. Putative phosphorylation sites regulated during the cell cycle have been reported (Minc et al.,1999) although there is no direct evidence that phosphorylation is involved in HP1 localization.

Centromeres are likely to be reorganized during progression through mitosis. The human centromere/kinetochore consists of many structural units(Clarke et al., 1999): the sister chromatids cohesion element (INCENP and Rad21), centromeric chromatin element (CENP-B), inner kinetochore element (CENP-A, CENP-C), outer kinetochore element (CENP-F), and fibrous corona (CENP-E, dynein, dynactin). Additionally, the spindle checkpoint apparatus, including Mad, Bub and other proteins, is localized on the outer kinetochore element. It should be noted that a portion of the IVR, in addition to the CSD, of HP1 is required for its localization to the metaphase centromere. INCENP is known to interact with HP1 through the IVR (residues 64-111)(Ainsztein et al., 1998). Our results show that neither the IVR nor the CSD alone is sufficient for localization to the metaphase centromere. Thus, INCENP is not sufficient for localizing HP1 to the metaphase centromere, and other protein molecules interacting with the CSD appear to be required, in combination with INCENP or other unknown IVR interactive molecules, for this effect.

In accordance with the specific localization of HP1α to the metaphase centromere, some cellular events are known to be specifically associated with HP1α. In highly invasive, metastatic breast carcinoma cell lines,HP1α expression is decreased at both mRNA and protein levels, whereas expression of HP1β or HP1γ is unaffected(Kirschmann et al., 2000). Conversely, the protein level of HP1α increases under anti-IgM induced apoptotic conditions in a human Burkitt lymphoma cell line(Brockstedt et al., 1998). HP1α interacts with INCENP (Ainsztein et al., 1998), and INCENP interacts with Aurora B kinase(Adams et al., 2001). Deletion of Scc1/Rad21 in DT40 cells eliminates the localization of INCENP to the inner centromeric domain (Sonoda et al.,2001). Furthermore, deletion of Swi6 (S. pombe HP1)causes loss of sister centromeric cohesion(Bernard et al., 2001). These results would suggest that the HP1α localized to the mitotic centromeric heterochromatin may contribute to the stability of sister chromatid cohesion or activation of the kinetochore checkpoint. Reduced HP1α at the metaphase centromere may also be a cause of chromosome instability in cancer cells.

HP1 subtypes interact with a variety of molecules. This generates a heterogeneous pattern of intracellular localization and dynamic relocalization during the cell cycle. Molecular dissection of their domains provides a simpler view of their localization through interaction with specific binding partners.

We thank David Alexander for critical reading of the manuscript. This work was supported by grants from the Japan Science and Technology Corporation(CREST; to H.M. and Y.H.), Grant-in-Aid for Scientific Research on Priority Areas B (to T.H., H.M. and Y.H.), and Grant-in-Aid for Basic Research 21 for Breakthroughs in Info-Communications (to H.M.).