Interaction Between Methyl CpG-Binding Protein and Ran GTPase during Cell Division in Tobacco Cultured Cells

Abstract

• Background and Aims Methyl CpG-binding proteins are considered to play critical roles in epigenetic control of gene expression by recognizing and interacting with 5-methylcytosine (m⁵C) in eukaryotes. However, among 13 corresponding genes in Arabidopsis thaliana, designated as featuring a methyl-binding domain (MBD), only four have so far been shown actually to bind to m⁵C. One example, AtMBD5, was selected here to screen for interacting proteins.

• Methods Yeast two-hybrid assays were used for screening, and physical interaction was confirmed by pull-down and bimolecular fluorescence complementation (BiFC) assays. Cellular localization was analysed by fluorescence-tagged fusion proteins using tobacco (Nicotiana tabacum) cultured bright yellow 2 cells.

• Key Results A gene finally identified was found to encode AtRAN3, a protein that belongs to the Ran GTPase family, which plays a critical role in nucleocytoplasmic transport and spindle bipolarization during cell division. AtMBD5 and AtRAN3 were clearly shown to interact in the nucleus by BiFC. On co-expression of AtMBD5-cyan fluorescence protein and yellow fluorescence protein-AtRAN3 in tobacco cells, both localized to the nucleus in the resting stage, migrating to the cytoplasm, primarily around chromatin, during mitosis, particularly at metaphase.

• Conclusions These results suggest that AtMBD5 becomes localized to the vicinity of chromosomes with the aid of AtRAN3 during cell division, and may play an important role not only in maintenance of chromatin structures by binding to m⁵C, but also in progress through mitosis by detaching from m⁵C. The present findings also shed light on the physiological function of Ran GTPases, direct target proteins of which have not thus far been well defined, suggesting their key role in chromatin movements in plant cells.

INTRODUCTION

DNA of higher eukaryotes contains 5-methylcytosine (m⁵C) as a minor base, and its interaction with proteins possessing a methyl CpG-binding domain (MBD) is critical for epigenetic regulation of gene expression. In mammalian cells, at least five MBD proteins performing multiple functions, such as interaction with histone deacetylase (Nan et al., 1998) and induction of large-scale chromatin reorganization during terminal differentiation in mouse cells (Breno et al., 2005), are present. MBD3 was shown to be specifically localized at centrosomes in the early M-phase during cell division (Sakai et al., 2002). In plants, a database search revealed that proteins containing MBD form a small family, with 13 and 16 corresponding genes in Arabidopsis and rice, respectively (Springer and Kaeppler, 2005). Subsequent analyses using bacterially expressed Arabidopsis proteins indicated that only AtMBD5, AtMBD6, AtMBD7 and AtMBD11 actually bind to m⁵C, specifically recognizing methylated DNA (Springer and Kaeppler, 2005). AtMBD6 has been shown to interact with histone deacetylase (Zemach and Grafi, 2003), while knock-out of AtMBD11 resulted in abnormal development in Arabidopsis (Berg et al., 2003). AtMBD5 was found to be enriched in actively proliferating tissues (Ito et al., 2003) and to bind to DDM1, which might facilitate localization to specific nuclear domains (Zemach et al., 2005), including heterochromatin regions (Scebba et al., 2003). Taking all the available information together, it is conceivable that MBD proteins play a critical role in chromatin remodelling during development.

Intensive studies with animal cells have revealed that Ran GTPase has two major roles in cell division: nucleocytoplasmic transport during interphase and spindle bipolarization during mitosis (Clarke and Zhang, 2001). On conformational changes due to GTP/GDP binding, Ran GTPase forms a steep gradient around the nuclear envelope, there regulating trafficking of macromolecules. It also interacts with chromosomes through formation of protein–protein complexes, including RCC1, and serves as a chromatin signal for spindle formation (Kahana and Cleveland, 1999; Hetzer et al., 2002). The available data strongly suggest that Ran GTPase plays a central role in nuclear division and transport between the nucleus and cytoplasm (Zheng, 2004) but few Ran GTPase targets have been identified, for example importin-β (Kunzler and Hurt, 2001). In plants, information on this protein is very limited. In Arabidopsis, four Ran GTPases have so far been identified, among which three (AtRAN1 to 3) appear to be almost identical (Haizel et al., 1997; Vernoud et al., 2003). They are expressed in meristematic tissues and embryos during development, suggesting roles in cell cycle progression, DNA replication and chromosome structure (Haizel et al., 1997).

Seven Arabidopsis MBDs were previously examined, and only AtMBD5 was found to bind specifically to m⁵C (Ito et al., 2003). Subsequent yeast two-hybrid screening identified Ran GTPase, which interacts with AtMBD5, as further characterized in the present study.

MATERIALS AND METHODS

Plant materials

Tobacco cultured cells, bright yellow 2 (BY2), were grown in Murashige and Skoog medium on a rotary shaker (115 r.p.m., 25 °C) in the dark. Transgenic BY2 cells were generated with Agrobacterium tumefaciens strain LBA4404 cells harbouring appropriate plasmids as described previously (Yamaguchi et al., 2003).

Gene isolation

Yeast two-hybrid screening was essentially performed as described by Yap et al. (2005). Briefly, a cDNA library was constructed with the HybriZAP-2·1^® two-hybrid pre-digested vector system (Stratagene, La Jolla, CA, USA), using RNA isolated from Arabidopsis seedlings. The bait plasmid was constructed by in-frame fusion of GBD-AtMBD5 containing a 1·2-kb AtMBD5 full-length cDNA (At3g46580) (Ito et al., 2003) to the GAL4 DNA binding domain in the pBD-GAL4 Cam vector (Stratagene). The bait plasmid and cDNA library clones were then transformed sequentially into the yeast strain Y190 (Clonetech, Palo Alto, CA, USA). A total of 10⁶ transformants were screened for complementation of growth on SD agar supplemented with 40 mm 3-amino-1,2,4-triazole and a mixture of appropriate amino acids, depleted of tryptophan, leucine and histidine. After secondary screening by x-gal filter lift assay for β-galactosidase activity, a 0·8-kb cDNA fragment was selected as a DNA sequence containing part of AtRAN3. Subsequently, a full-length cDNA was prepared by PCR based on the registered sequence (NCBI accession no. U73810), using the following set of primers: 5′GGATCCATGGCTCTACCTAACCAGC3′ (forward) and 5′CCGGGCGGGTAATCAGTGACAGATC3′ (reverse). The resulting fragment was cloned into the pGEM-T Easy vector and used for further experiments.

Fusion proteins

Plasmids for glutathione S-transferase (GST)-AtMBD5 and His-tagged AtRAN3 were constructed using pGEX4T-2 (Pharmacia Biotechnology, Piscataway, NJ, USA) (containing a GST tag of 25 kDa) and pET-32a (Novagen, Darmstadt, Germany) (containing thioredoxin, His and S tags of 21 kDa), respectively, as described (Yap et al., 2005) and transformed into Escherichia coli BL21. Expression of fusion proteins was then induced at 25 °C for 6 h with 1 mm isopropyl-β-d-thiogalactoside. GST-AtMBD5 protein was purified through a glutathione Sepharose column (Amersham Biosciences, Piscataway, NJ, USA), and His-tagged AtRAN3 with Ni-NTA agarose beads (Qiagen, Hilden, Germany) according to the manufacturers' instructions. Protein concentrations were estimated by the Bradford method.

Pull-down binding assay

Approximately 5 μg of GST fusion protein or GST alone as a control was bound to a glutathione Sepharose™ 4B column (MicroSpin GST Purification Module, Amersham Pharmacia Biotech) and incubated with 25 μg of His-tagged AtRAN3 at 4 °C for 16 h. To obtain GTP- and GDP-bound forms, 25 μg AtRAN3 was incubated with 1 mm GTP[γS] or GDP[βS] at 30 °C for 40 min prior to the incubation. After three washes with washing buffer (300 mm NaCl, 50 mm Na₂HPO₄, pH 7·3, 50 mm imidasol), GST-fusion proteins bound to the columns were eluted with a buffer containing 10 mm reduced glutathione in 50 mm Tris-HCl at pH 8. Eluted proteins were fractionated by 7·5 % polyacrylamide-SDS gel electrophoresis, and subjected to immunoblot assay using rabbit anti-His-tag antibodies (Santa Cruz Biotechnology, Santa Cruz, CA, USA) and anti-rabbit horseradish peroxidase conjugate antibodies (BioRad, Hercules, CA, USA).

Vector construction

Expression vectors for epifluorescence experiments were prepared as follows: the AtMBD5-green fluorescence protein (GFP) vector was constructed by introducing the full-length AtMBD5 cDNA into the cauliflower mosaic virus (CaMV)35S-sGFP(S65T)-NOS vector at SalI and NcoI sites as previously described (Ito et al., 2003). The AtMBD5-CFP vector was reconstituted from the AtMBD5-GFP vector by substituting GFP and the terminator by cyan-fluorescence protein (CFP) and its terminator fragment. The yellow fluorescence protein (YFP)-AtRAN3 vector was produced step-wise by first inserting AtRAN3 into the YFP vector (provided by Dr A. von Arnim, University of Tennessee, Knoxville) at BglII and XbaI sites, and then its CaMV35S-YFP-AtRAN3 fragment into pBI101 at SalI and BamHI sites. For the double-expression vector, the CaMV5S-YFP-AtRAN3 plasmid was digested with BamHI, and the BamHI-cleaved CaMV35S-AtMBD5-CFP fragment was inserted to yield YFP-AtRan3-AtMBD5-CFP. AtRCC1 was prepared by PCR amplification from an Arabidopsis cDNA library and fused to CaMV35S-GFP vector as described (Yap et al., 2005). All the plasmids obtained were transfected into Agrobacterium tumefaciens LBA4404 strains, which were then used to transform tobacco cultured BY2 cells (Yap et al., 2005).

Bimolecular fluorescence complementation analysis

AtMBD5 and AtRAN3 cDNA fragments were subcloned into the SalI/NotI sites of pSY728 and pSY736 containing the N-terminal fragment of the YFP (YN), and pSY738 and pSY735 vectors containing the C-terminal fragment of the YFP (YC) (Bracha-Drori et al., 2004). The resulting eight plasmids, pSY728-AtMBD5 (AtMBD5-YN), pSY728-AtRAN3 (AtRAN3-YN), pSY738-AtMBD5 (AtMBD5-YC), pSY738-AtRAN3 (AtRAN3-YC), pSY736-AtMBD5 (YN-AtMBD5), pSY736-AtRAN3 (YN-AtRAN3), pSY735-AtMBD5 (YC-AtMBD5) and pSY735-AtRAN3 (YC-AtRAN3), were co-bombarded under various combinations (Table 1) with gold particles (Bio-Rad) into onion epidermal cell layers as described (Yap et al., 2005). CaMV35S-YFP plasmid was used as a control. After incubation at 25 °C for 48 h in darkness, the epidermal cell layers were viewed under a microscope (Leica MZ FLIII) equipped with a fluorescence module.

Table 1.

Vector combination and YFP reconstitution in BiFC analysis

Combination of constructs	YFP fluorescence
YN-MBD5/YC-RAN3	–
YN-MBD5/RAN3-YC	–
YN-RAN3/YC-MBD5	+
YN-RAN3/MBD5-YC	–
MBD5-YN/YC-RAN3	+
MBD5-YN/RAN3-YC	–
RAN3-YN/YC-MBD5	–
RAN3-YN/MBD5-YC	–

A set of the indicated constructs was bombarded into onion epidermal cell layers, incubated at 25 °C in the dark for 48 h and observed with a microscope. The combination which showed positive YFP fluorescence is indicated with a plus (+), and that with no fluorescence is indicated with a minus (–).

Epifluorescence analysis

Transgenic BY2 cells were examined under a fluorescence microscope (Olympus Provis AX70) using the U-MGFPHQ cube for GFP, the U-MYFPHQ cube for YFP and the U-MCFPHQ cube for CFP fluorescence (Olympus, Tokyo, Japan) (Yap et al., 2005). DNA staining was performed with an aliquot of culture medium containing 1 mg mL⁻¹ 4′,6-diamido-2-phenylindole (DAPI).

RESULTS

Identification of AtRAN3

Among 10⁶ yeast transformants derived from the Arbidopsis cDNA library with full-length AtMBD5 as the bait in the yeast two-hybrid system, six clones were initially positive in both histidine and β-galactosidase complementation assays. The plasmids recovered from these transformants were subjected to sequence determination and one was found to encode a Ran-related protein. To confirm the specificity of the interaction, the recovered plasmid was co-transformed with either GBD-AtMBD5 (bait plasmid) or the pBD-GAL4 Cam vector into yeast Y190. Only the transformants containing plasmids encoding both AtMBD5 and its interaction partner were positive in the histidine and β-galactosidase reporter assays (Fig. 1A), indicating that the interaction between them was indeed specific. Its putative peptide sequence indicated the isolated cDNA clone to be a 5′-truncated product, encoding a polypeptide of 91 amino acids at the C-terminus (Fig. 1B). Consequently, the full-length cDNA was isolated from the Arabidopsis cDNA library by PCR, and sequence analysis indicated that it encoded AtRAN3 (gene ID: 835612) (Fig. 1B). The observed interaction of AtMBD5 and AtRAN3 in the yeast two-hybrid system was confirmed in an in vitro pull-down assay (Fig. 1C). His-tagged AtRAN3 was applied to a glutathione-Sepharose column containing AtMBD5-GST or GST proteins, eluted with a buffer containing reduced GST, separated by SDS-PAGE and subjected to immunoblot assays with anti-His-tag antibodies. The results clearly showed that AtRAN3 was detectable upon incubation with AtMBD5-GST but not with GST protein alone, indicating specific binding of AtRAN3 to AtMBD5 (Fig. 1C). Ran GTPase is known to change its location and targeting of interacting proteins depending on whether it is in the GTP- or GDP-bound forms (Zheng, 2004). In order to determine which preferentially binds to AtMBD5, pull-down assays were performed in the presence of GTP and GDP analogues (Fig. 1C). GTP[γS] binds to AtRAN3, but is not hydrolysed, resulting in formation of a constitutively active complex. By contrast, GDP[βS] cannot be converted into GTP, leaving the complex inactive (Terryn et al., 1995). Binding tests clearly showed that AtMBD5 preferentially interacted with the active form of AtRAN3 (Fig. 1C). A densitometric estimation indicated that the active form had five-fold higher binding efficiency than the inactive form (data not shown).

Fig. 1.

Interaction between AtMBD5 and AtRAN3. (A) Yeast two-hybrid assay. Yeast AH109 cells were co-transformed with GBD-AtMBD5 and AD-AtRAN3 constructs (left panel). The transformants were planted on SD agar supplemented with an amino acid mixture depleted of histidine and tryptophan (SD/–WL) (lane 1), or histidine, tryptophan and leucine in the absence (SD/–HWL) (lane 2) or presence of 40 mm 3-amino-1,2,4-triazole (lane 3). Colonies cultured on the SD/–WL plate were assayed for β-galactosidase by the filter lift method (lane 4). Transformants from the SD/–HWL agar plate were cultured in SD/–WL broth and β-galactosidase activity was estimated using O-nitriphenyl-β-d-galactopyranoside as the substrate and expressed in Miller units (right panel). The value shown is the average of triplicate measurements with standard deviation. (B) Amino acid sequence of AtRAN3 (gene ID: 835612) and its tobacco orthologue, NtRan-A1 (accession no. P41918). Alignment was performed with the Clustal W program. Identical residues are shaded, and a fragment initially identified by the yeast two-hybrid screening is underlined. Specific motifs conserved among Ran GTPases are indicated; G1–G3 and P-loop are for GTP binding, and Switch 1 and Switch 2 are for interaction with RanBD1 and karyopherin-β2, respectively (Macara, 1999). (C) Pull-down assay. Interactions between AtMBD5 and AtRAN3 in the absence or presence of GTP[γS] or GDP[βS] were examined. Approximately 5 μg GST-AtMBD5 (47 kDa) proteins or GST proteins as the control were bound to a glutathione column and 25 μg His-tagged AtRAN3 (45 kDa) was added to each column. After elution with reduced glutathione, proteins were separated on SDS–PAGE (CBB staining, left panel). Loaded samples were molecular markers (lane 1), input His-tagged AtRAN3 (lane 2) (indicated by an open arrowhead), GST protein (lane 3) and GST-AtMBD5 (indicated by a closed arrowhead) (lanes 4–6). His-tagged AtRAN3 was added prior to elution in the absence (lanes 3 and 4) or presence of GTP[γS] (lane 5) or GDP[βS] (lane 6). Proteins were then blotted onto nylon membrane, and AtRAN3 was detected with anti-His-tag antibodies (indicated by an open arrowhead) (right panel).

Intracellular localization

The cellular localization of AtMBD5 and AtRAN3 proteins was examined with GFP- and YFP-fused constructs, respectively. Plasmids containing CaMV 35S promoter AtMBD5-GFP and AtRAN3-YFP genes were constructed and introduced into tobacco cultured BY2 cells via Agrobacterium transformation. On epifluorescence analysis, both AtMBD5-GFP and YFP-AtRAN3 constructs gave fluorescent signals in nuclei during the resting stage, whereas the control GFP fluorescence was observed throughout the cell (Fig. 2A). Upon merging with DAPI-stained images, both proteins, particularly AtMBD5, were apparently localized to regions where DNA is condensed. This pattern was consistent with the previous observation that AtMBD5 is located within heterochromatin (Scebba et al., 2003). Interactions between AtRAN3 and AtMBD5 in planta were directly examined by bimolecular fluorescence complementation (BiFC) analysis, in which active YFP is reconstituted only when non-fluorescent N-terminal (YN) and C-terminal (YC) YFP fragments are brought together by protein–protein (AtRAN3–AtMBD5) interactions (Hu et al., 2002; Bracha-Drori et al., 2004). To determine the optimal condition, various combinations of vectors and proteins were examined by transient co-expression in onion epidermal cell layers via particle bombardment. Eight initial combinations were used, for example AtMBD5 fused to the C-terminus of the YN fragment (YN-AtMBD5), AtRAN3 fused to the C-terminus of the YC fragment (YC-AtRAN3) and vice versa (Table 1). Among these eight combinations, cells co-expressing AtMBD5-YN and YC-AtRAN3 showed clear YFP fluorescence in the nucleus (Fig. 2B, upper panel). A combination of YN-AtRAN3 and YC-AtMBD5 also exhibited a positive signal, but all other combinations did not emit fluorescence, suggesting that the position and orientation of each YFP fragment are critical to recover activity (Table 1). These results indicated that the observed fluorescence was the result of reconstitution of YFP through a specific interaction between AtMBD5 and AtRAN3. Control fluorescence of YFP alone was observed throughout the cell (Fig. 2B, lower panel). Note that negative controls for BiFC assays such as non-fluorescent transformants are technically not available, because such cells are not visible via microscopy, and not detectable owing to the transient expression system of the particle bombardment. Hence, the positive signal itself is considered to be good evidence for a specific interaction between the two proteins (see, for example, Kapoor et al., 2005). Upon enlargement, bright speckled signals were observed in the nucleus (Fig. 2B, right panel). Cellular localization of AtMBD5 and its close relative, AtMBD6, has been reported to be at regions of heterochromatin (Scebba et al., 2003), and at perinucleolar chromocentres (Zemach et al., 2005). The BiFC fluorescence points to the location of interaction of two proteins, but not necessarily to the location of each protein. Yet the speckled pattern was consistent with the former case and also with the localizing pattern of AtMBD5 (Fig. 2A). It was thus conceivable that, at resting stage, the AtMBD5/AtRAN3 complex is localized to heterochromatin regions.

Fig. 2.

Intracellular localization. (A) AtMBD5-GFP (top row) or YFP-AtRAN3 (middle row) was stably expressed in tobacco BY2 cells, and cells at resting stage were observed under interference contrast (Interference contrast), and by staining with DAPI (DAPI) and by epifluorescence of GFP for AtMBD5 or YFP for AtRAN3 (Fluorescence). DAPI and epifluorescence images are merged (Merge). As a control, BY2 cells expressing CaMV35S-GFP were used (bottom row). (B) Bimolecular fluorescence complementation analysis. Onion epidermal cells were bombarded with expression vectors containing YC-AtRAN3 or AtMBD5-YN and transient epifluorescence was observed (top row, left). In the nucleus, enlarged and speckled spots are clearly shown (top row, right). As a positive control, onion epidermal cells were bombarded with an expression vector containing YFP alone and transient epifluorescence was observed (bottom image).

Co-localization of AtRAN3 and AtMBD5 during cell division

As Ran GTPases have been recognized to play a critical role in cell division, the above observation suggested that AtMBD5 might also be involved in cell division together with AtRAN3. To test this hypothesis, the cellular localization of AtMBD5 was first examined by GFP-fusion protein (Fig. 3). In cells at prophase, AtMBD5-GFP was only localized in the nucleus (Fig. 3A, top panel). At metaphase, however, AtMBD5-GFP was found mostly outwith chromosomes (Fig. 3A, second panel), particularly at the position of spindle microtubules (Fig. 3B). At anaphase, the protein began to align on chromosomes (Fig. 3A, third panel), and at telophase, AtMBD5 was again localized only in the nucleus (Fig. 3A, fourth panel). To examine whether this pattern was associated with AtRAN3, both proteins were simultaneously expressed in a single cell, and the behaviour of each was assessed by differential fluorescence for YFP (AtRAN3) and CFP (AtMBD5), together with DAPI staining for chromosomes (Fig. 4A). In cells at interphase, AtMBD5 was localized exclusively in the nucleus, whereas AtRAN3 was localized in both the nucleus and cytoplasm (Fig. 4A, top panel). At metaphase, they were mostly localized outwith chromosomes, particularly at the position of spindle microtubules (Fig. 4A, second panel). At anaphase, both began to align on chromosomes, but AtRAN3 was also localized between the poles (near the cell plate; Fig. 4A, third panel). At telophase, AtMBD5 was almost exclusively localized on chromosomes, whereas AtRAN3 was localized on chromosomes and in regions of cell plate growth (Fig. 4A, bottom panel). These results indicated that the cellular localization of AtMBD5 essentially coincides with that of AtRAN3. This was particularly distinct at metaphase, indicating that AtMBD5 moved in harmony with AtRAN3, although during later phases the former appeared to act partly independently of the latter.

Fig. 3.

Localization of AtMBD5 during cell division. (A) BY2 cells expressing AtMBD5-GFP were observed by epifluorescence at the indicated phase during cell division (GFP). After fixation on glass slides, transgenic cells in the indicated division phase were stained with DAPI (DAPI), and differentially observed. Two fluorescence images were then merged (Merge). (B) At metaphase, enlarged and microtubule-like structures are visible (arrowhead).

Fig. 4.

Co-localization during cell division. (A) Subcellular localization of AtMBD5 and AtRan3 during mitosis. BY2 cells were stably transformed with the double-expression vector for AtMBD5 and AtRAN3 (top row), and after fixation on slide glasses, transgenic cells in the indicated division phase were stained with DAPI, and differentially observed using specified filters for each fluorescence. After taking fluorescence photographs for DAPI (DAPI), CFP (AtMBD5) and YFP (AtRAN3), images were merged (Merge). (B) Subcellular localization of control proteins. Transgenic BY2 cells stably expressing 35S-GFP (top row), AtRCC1-CFP (middle row) or tobacco α-tubulin-GFP (bottom row) were observed at metaphase. Images from DAPI staining (DAPI) and from epifluorescence (Fluorescence) are merged (Merge). Note that NtRCC1 remains within chromosomes, whereas α-tubulin relocates to spindle regions.

At metaphase, chromosomes align at the centres of cells, and spindle microtubules (polar microtubule) are formed (Clarke and Zhang, 2001). Subsequently, the localization patterns of AtMBD5 and AtRAN3 were compared with those of control proteins at metaphase (Fig. 4B). Control GFP was observed in cytoplasm, particularly near dividing areas, where cytoplasmic density was high (Fig. 4B, top panel). By contrast, AtRCC1 was primarily located in chromatin (Fig. 4B, second panel), consistent with reports that RCC1 stays with chromatin throughout cell division (Kahana and Cleveland, 1999). Alpha-tubulin, which is the major component of microtubules, became concentrated in cell centres, forming mitotic spindles (Fig. 4B, bottom panel). The localization of AtMBD5 and AtRAN3 was consistent with this observation, strongly suggesting both to be associated with spindle microtubules.

DISCUSSION

The present study demonstrated that AtMBD5 interacts with a Ran GTPase, AtRAN3, and that they dynamically move around the chromatin during cell division. Although these observations with fluorescence markers clearly showed co-migration of the two proteins, the significance must be carefully considered within the experimental framework. For example, it can be argued that their behaviour might not necessarily equate to that of their native counterparts, because the two proteins were over-expressed under control of the CaMV 35S promoter, and the experimental system was heterogeneous, introducing Arabidopsis genes into tobacco cultured cells. The first is a general point with the use of fluorescence protein markers, and currently no clear rebuttals have been published. However, as the CaMV 35S promoter-driven RCC1 proteins were correctly observed to be associated with chromatin throughout cell division under the present experimental conditions, as observed in mammalian cells (Moore et al., 2002), our observations are considered to reflect the native location of the two proteins. The second point raises the question as to whether the interaction between MBD and Ran GTPase also takes place in Arabidopsis cells. A homology search identified only five out of 221 amino acids to differ between AtRAN3 and tobacco Ran GTPase, NtRan-A1 (accession no. P41918) (97·8 % similarity) (see Fig. 1B), strongly suggesting them to be orthologues. Although a tobacco MBD5 orthologue(s) has yet to be identified, it is highly conceivable that a set of two protein species is common between Arabidopsis and tobacco plants. Taking these conditions into account, the results are interpreted as follows.

In the resting stage, AtRAN3 was localized in nuclei. Upon onset of cell division, it shifted to the vicinity of chromosomes at metaphase, consistent with reports for mammalian cells (Kahana and Cleveland, 1999). The region involved was found to be microtubule spindles, as mutant Ran yielded spindle abnormalities in vivo (Di Fiore et al., 2004). At anaphase, AtRAN3 was observed in and between chromosomes, and at telophase, it was localized markedly in chromosomes and in the growing cell plate. This pattern is essentially the same as that observed in mammalian cells (Zhang et al., 1999), suggesting that plant Ran GTPases share a common role during cell division. AtMBD5 also changed its localization to the vicinity of chromosomes. As BiFC analysis clearly indicated their physical interaction in vivo, it was conceivable that both AtRAN3 and AtMBD5 proteins co-operatively function during cell division. This was particularly distinct during metaphase, in which both proteins proved to be present in or around spindles, as confirmed by similar localization patterns of α-tubulin, a major component of microtubules. At anaphase, AtMBD5 was still co-localized with AtRAN3 in and between chromosomes, but at telophase, it was condensed into chromosomes, whereas AtRAN3 was in both chromosomes and the growing cell plate. Ran GTPases are considered to be critical for microtubule formation and nuclear envelope formation (Kunzler and Hurt, 2001). Association of AtMBD5 with AtRAN3 during metaphase, and partial dissociation at anaphase and telophase suggested that AtMBD5 might actively participate in spindle formation but not necessarily in cell plate formation. To date, the targets of Ran GTPase have not been completely clarified. Known Ran-binding proteins include RanGAP, RCC1, karyopherins, RanBPs and importin-β (Kunzler and Hurt, 2001), the data all being derived from mammalian cells. The present findings showing that MBD proteins are also bound to the active form of Ran GTPase would be of help for further understanding of the mechanism of cell division, particularly in plants.

The biological significance of the interaction between the two proteins is not clear, but two alternative explanations are conceivable. First, AtMBD5 may be positively involved in microtubule formation together with AtRAN3. If this is the case, AtMBD5 might possess a specific function(s) as a protein in addition to its m⁵C binding ability. Second, by detaching from DNA, AtMBD5 could allow m⁵C to become exposed to other interacting factors, thereby helping associations between microtubules and chromatin. In this case, AtMBD5 would serve as a repressor of m⁵C under non-dividing quiescent conditions. Although there is no clear answer as to which scenario is more probable, the present findings indicate that AtMBD5 mobilizes around chromatin with the aid of AtRAN3, which functions as a transporter in nuclear trafficking, and therefore that MBDs may play an important role not only in maintenance of chromatin structures but also in progress through mitosis in plant cells (Fig. 5). To substantiate this idea, it was attempted to analyse mutant Arabidopsis for both proteins. To this end, three T-DNA inserted lines at the AtMBD5 locus (SALK117619, SALK11193 and ET8226) were first assayed, with the expectation of finding some dominant-negative phenotypes. However, their growth and stature did not differ from those of the wild-type plant, perhaps due to gene redundancy as previously suggested (Zemach et al., 2005). Further examination using double or triple mutants together with AtRAN3 mutants will provide clear-cut results as to their biological roles.

Fig. 5.

Model for mitotic activity involving Ran and MBD proteins. At prophase, AtMBD5 is tightly bound to DNA anchoring on m⁵C. AtRAN3 cycles between GTP-bound active and GDP-bound inactive forms (Kahana and Cleveland, 1999). The active form is translocated into the nucleus, and interacts with AtMBD5 (A). At metaphase, AtRAN3-GTP forms a steep gradient around the chromosome, and promotes spindle assembly by interacting with several specific proteins such as importins (Clarke and Zhang, 2001). At this time, AtMBD5 probably detaches from DNA with the aid of AtRAN3, and migrates to the spindle fibre (B). At anaphase and telophase, the cycling of AtRAN3 induces nuclear envelope reassembly, partially releasing AtMBD5, which again binds to m⁵C in DNA (C). Note that DNA is illustrated rather than chromatin to simplify the interaction between MBD and m⁵C.

Overall, the findings presented here may imply additional and profound roles of DNA methylation, at least in plant cells. It is tempting to speculate that m⁵C serves as a landmark for proteins that bind to DNA and mobilize chromatin during cell division.