Overexpression of the human MNB/DYRK1A gene induces formation of multinucleate cells through overduplication of the centrosome

Background Previously we cloned the human MNB/DYRK1A gene from the "Down syndrome critical region" on chromosome 21. This gene encodes a dual specificity protein kinase that catalyzes its autophosphorylation on serine/threonine and tyrosine residues. But, the functions of the MNB/DYRK1A gene in cellular processes are unknown. Results In this study, we examined HeLa cells transfected with cDNA encoding a green fluorescent protein (GFP)-MNB/DYRK1A fusion protein and found 2 patterns of expression: In one group of transfected cells, GFP-MNB/DYRK1A was localized as dots within the nucleus; and in the other group, it was overexpressed and had accumulated all over the nucleus. In the cells overexpressing GFP-MNB/DYRK1A, multinucleation was clearly observed; whereas in those with the nuclear dots, such aberrant nuclei were not found. Furthermore, in the latter cells, essential processes such as mitosis and cytokinesis occurred normally. Multinucleation was dependent on the kinase activity of MNB/DYRK1A, because it was not observed in cells overexpressing kinase activity-negative mutants, GFP-MNB/DYRK1A (K179R) and GFP-MNB/DYRK1A (Y310F/Y312F). Immunostaining of GFP-MNB/DYRK1A-overexpressing cells with specific antibodies against α- and γ-tubulin revealed that multiple copies of centrosomes and aberrant multipolar spindles were generated in these cells. Conclusions These results indicate that overexpression of MNB/DYRK1A induces multinucleation in HeLa cells through overduplication of the centrosome during interphase and production of aberrant spindles and missegregation of chromosomes during mitosis.


Background
Down syndrome (trisomy 21) is the most frequent birth defect and is a major cause of mental retardation and congenital heart disease [1]. Besides the characteristic set of facial and physical features of individuals afflicted with it, this syndrome is associated with defects of the immune and endocrine systems, an increased rate of leukemia, and early onset of Alzheimer disease [1]. Although little is known about the mechanism by which trisomy 21 interferes with normal development, the increased dosage of the chromosomal elements clearly implies altered levels of gene expression as a causative factor.
In most cases, patients with Down syndrome show trisomy of chromosome 21. Studies of cases with partial trisomy of chromosome 21 have suggested that the region around locus D21S55 is particularly important in the etiology of the syndrome [2][3][4]. This subchromosomal region is called the "Down syndrome critical region." Earlier we performed exon trapping experiments using a series of cosmid clones isolated from this chromosomal region, and identified the genomic structure and cDNA sequence of the human MNB/DYRK1A gene in this region [5][6][7].
The human MNB/DYRK1A gene is a human homolog of Drosophila minibrain and rat DYRK genes. Mutant flies with a reduced expression of minibrain have a reduced number of neurons in distinct areas of the adult brain; and this gene is therefore required for a distinct neuroblast proliferation during postembryonic neurogenesis [8]. A previous report on the rat DYRK protein showed that the DYRK gene encodes a dual specificity kinase that catalyzes its autophosphorylation on both serine/threonine and tyrosine residues [9][10][11]. The human MNB/DYRK1A gene has been suggested to be a strong candidate gene for learning defects associated with Down syndrome, because transgenic mice carrying a 180-kb YAC contig containing the human MNB/DYRK1A gene showed defects in learning and memory [12,13]. But, the roles of the MNB/ DYRK1A gene in cellular processes are far from established.
To clarify the physiological role of MNB/DYRK1A, in the present study we transfected HeLa cells with cDNA encoding a GFP-MNB/DYRK1A fusion protein. The subcellular localization of the transfected-MNB/DYRK1A depended on its expression level. In the case of a low level, the protein was concentrated at specific loci in the nucleus; but in the case of a high level, it was located all over the nucleus. This high-level expression of MNB/DYRK1A protein induced the overproduction of the centrosome and led to multinucleation in HeLa cells. These results indicate that MNB/DYRK1A may play a specific function in coordinat-ing nuclear division with other cell-cycle progression events.

Results
MNB/DYRK1A protein is expressed at high levels in various organs including the central nervous system during the embryonic period, but its level gradually decreases with postnatal growth and is extremely low in the adult [7]. This finding may indicate a significant role of this protein in the proliferation and differentiation of cells during development. To examine the physiological role of MNB/ DYRK1A, we transiently transfected HeLa cells with cDNA encoding a GFP-MNB/DYRK1A fusion protein and determined its subcellular localization. The fluorescent fusion protein was found inside the nucleus of HeLa cells and showed a speckled pattern, as reported previously (Fig.  1A, panel C) [11]. We also found other cells with GFP-MNB/DYRK1A in a diffuse pattern all over the nucleus, in which cells multiple nuclei were often observed (Fig. 1A, panel D). Such a speckled pattern or multinucleation was not observed if cDNA encoding GFP or GFP-NLS, a fusion protein of GFP and bipartite nuclear localization sequence corresponding to amino acids 104-133 in MNB/DYRK1A, was used for the transfection. GFP was mainly detected in the cytoplasm, and GFP-NLS was localized in the nucleus, where it accumulated in nucleoli (Fig.  1A, panels A and B). These subcellular localization patterns were observed independently of the expression level of GFP and GFP-NLS. We next expressed MNB/DYRK1A protein without the GFP tag in HeLa cells and found cells with nuclear dots (Fig. 1B, panel C) and multiple nuclei (Fig. 1B, panel D). Thus, it appears that the speckled distribution of GFP-MNB/DYRK1A in the nucleus of HeLa cells and the emergence of multinucleate cells are not artifacts caused by high-level expression of the GFP fusion protein.
Next, we performed detailed analysis of the multinucleate cells by confocal laser scanning microscopy to exclude the possibility that expression of the GFP-MNB/DYRK1A fusion protein produced nuclei surrounded by a continuous and polymorphous nuclear membrane. Serial optical section image data for the cells strongly expressing MNB/ DYRK1A were obtained at 0.4-µm intervals (Fig. 2). Observation of all of the focal planes of cells with 3 nuclei showed each nucleus to be independent, i.e., to have its own nuclear membrane, indicating that MNB/DYRK1A indeed induced multinucleation if its expression level was high.
We previously prepared a polyclonal antibody (anti-MNB antibody) immunospecific for the MNB/DYRK1A protein, one that recognized 2 major bands and 1 minor band with apparent molecular masses of about 90 kDa when samples from various rat tissues were analyzed [7]. These 3 MNB/DYRK1A proteins were also detected in nontransfected HeLa cells by immunoblot analysis using this anti-MNB antibody (data not shown), but nontransfected HeLa cells were stained only weakly with this antibody (Fig. 3, panel A). To determine the expression level of GFP-MNB/DYRK1A protein in multinucleate cells, we stained the cells with the anti-MNB antibody after transfection with GFP-MNB/DYRK1A cDNA. When cells with the dotted distribution of GFP-MNB/DYRK1A were immunostained with this antibody, a similar dotted pattern was observed (Fig. 3, panel B). On the other hand, cells with the diffuse distribution of the GFP-MNB/ DYRK1A fusion protein all over the nuclei were stained strongly with the anti-MNB antibody (Fig. 3, panels C and D). This strong staining was not observed when control immunostaining was performed without the addition of anti-MNB antibody as the primary antibody (Fig. 3, panel E). These results show that the GFP-MNB/DYRK1A protein level was much higher in cells with the diffuse distribution of this protein than in the nontransfected control cells or transfected cells having the dotted distribution of GFP-MNB/DYRK1A.
Next, to investigate the subcellular localization of MNB/ DYRK1A protein during the cell cycle, especially during mitosis, we transiently expressed GFP-MNB/DYRK1A in HeLa cells and chased interphase cells with a speckled fluorescent pattern in a time-lapse manner (Fig. 5). At 240 minutes when the nuclear envelope disappeared, GFP-MNB/DYRK1A fluorescence gave a diffuse pattern and was located in the cytoplasm with relative exclusion from the condensed chromatin. After reformation of the nuclear envelope at telophase (see 360 minutes), GFP-MNB/DYRK1A again exhibited a speckled pattern. Thus, MNB/DYRK1A was concentrated at specific loci in the nucleus during interphase, whereas during mitosis it was located all over the cells and not condensed with either chromosome, mitotic spindle, or spindle pole. Fig. 5 also shows that the interphase cells with a speckled pattern in their nucleus progressed into M phase and produced 2 daughter cells. Further, both the condensation and subsequent segregation of chromosomes occurred in these cells as properly as in non-transfected control cells. On the other hand, chromosome missegregation was observed in HeLa cells with overexpressed MNB/DYRK1A, as will be described later (Fig. 6A, panel d). MNB/DYRK1A protein is thought to contribute to the characteristic features of Down syndrome. Although multinucleation can be produced via various mechanisms, defects in the mitotic machinery may play a major role in this nuclear abnormality. The mitotic machinery accurately separates and distributes chromosomes into each daughter cell. This accurate segregation is achieved by mitotic spindles composed of microtubules. So, we determined the effect of over-expressed GFP-MNB/ DYRK1A on microtubule structures by immunostaining cells with an antibody specific for α-tubulin, a main constituent of microtubules (Fig. 6). When interphase cells were observed after the transfection with GFP, α-tubulin was detected in the cytoplasm, especially in the nuclear periphery (Fig. 6A, panel a). In a multinucleate cell, αtubulin showed a localization similar to that in the GFPtransfected cell, but it was localized around a mass of nuclei rather than around each nucleus (Fig. 6A, panel c).
At the M phase of the cell cycle, typical mitosis with a bipolar mitotic spindle was observed in the GFP-transfected cells (Fig. 6A, panel b), whereas aberrant mitotic spindles were noted in cells expressing a high level of GFP-MNB/DYRK1A (Fig. 6A, panel d). Further analysis of mitotic cells overexpressing GFP-MNB/DYRK1A revealed that tripolar spindles appeared in prometaphase cells (Fig.  6B, panel a) and that multiple mitotic spindle poles and abnormal chromosome condensation appeared in metaphase cells and late-anaphase cells (Fig. 6B, panels b and  c).
The centrosome is the major microtubule-organizing center in eukaryotic cells and features prominently in mitosis, where it is required for the establishment of spindle bipolarity, spindle microtubule assembly, and balanced segregation of chromosomes. So, the aberrant mitotic spindles seen in MNB/DYRK1A-overexpressing cells may have resulted from the generation of multiple centrosomes. So next we visualized these structures by immunostaining cells with an antibody specific for γtubulin, a well-characterized component of centrosomes in all phases of the cell cycle (Fig. 7). HeLa cells transfected with the GFP construct, which was used as a control, contained 1 or 2 centrosomes juxtaposed to the nucleus at interphase (Fig. 7, panel A). At M phase, they contained 2 centrosomes, between which the chromosomes were aligned (Fig. 7, panel B). In contrast, MNB/ DYRK1A-overexpressing cells contained more than 2 centrosomes at interphase (Fig. 7, panels C and D) and also at M phase (Fig. 7, panel E), thus indicating centrosome overduplication. Since overduplication of the centrosome was not detected in the cells in which MNB/DYRK1A was localized as nuclear dots within the nucleus (data not shown), such overduplication must have resulted from MNB/DYRK1A overexpression. Further, the presence of cells with centrosome amplification but not with multi-nucleation indicates that a high level of MNB/DYRK1A induces the overduplication of the centrosome in interphase, subsequently producing multiple spindle poles leading to the multinucleation.
We next quantified the number of cells with an abnormal number of centrosomes at 24 hours and 48 hours after the transfection with the cDNA encoding the GFP-MNB/ DYRK1A fusion protein. Cells with 3 or more centrosomes were observed predominantly in the case of GFP-MNB/DYRK1A transfection (Fig. 8). The percentage of cells with 2 centrosomes was also higher in this cell population, than in that transfected with GFP or GFP-NLS. Since centrosome amplification was not observed even in cells that expressed GFP or GFP-NLS at a high level, the overproduction of centrosomes is concluded to have occurred through the overexpression of MNB/ DYRK1A protein.

Discussion
The transfection experiments with cDNA encoding GFP-MNB/DYRK1A showed that the subcellular distribution pattern of GFP-MNB/DYRK1A protein depended on its intracellular expression level. When the expression level of transfected MNB/DYRK1A cDNA was low, MNB/ DYRK1A protein was localized with a speckled pattern in the nucleus. A similar subnuclear localization of MNB/ DYRK1A has been observed in both COS-7 cells and HEK293 cells transfected with the GFP fusion protein of MNB/DYRK1A [10,11]. Furthermore, we found that MNB/DYKR1A without a GFP tag became localized with a speckled pattern in the nucleus. These findings indicate that the speckled distribution of GFP-MNB/DYRK1A in the nucleus is not an artifact. Thus, MNB/DYRK1A probably localizes to subnuclear domains in the nucleus; and its association with this distinct subnuclear structure may be critical for some specific function of MNB/ DYRK1A, although we couldn't detect this subnuclear localization of endogenous MNB/DYRK1A by using our specific antibody against MNB/DYRK1A. Similar speckled patterns of subnuclear localization have been shown for other proteins [14][15][16][17][18]. One of them is the transcription factor forkhead in rhabdomysarcoma (FKHR), and this protein is known to co-localize and interact with MNB/DYRK1A [19]. We found that GFP-MNB/DYRK1A co-immunoprecipitated with FKHR by immunoprecipitation using anti-FKHR antibody (manuscript in preparation), indicating that FKHR binds to GFP-MNB/DYRK1A in HeLa cells. FKHR transcription factors mediate cell-cycle regulation of a variety of cell lines, dependent on the cell-cycle inhibitor p27 kip 1 . They also play a role in the control of gene expression by insulin, as well as in the regulation of apoptosis mediated by survival factors [20][21][22]. These signals trigger the phosphorylation Overproduction of centrosomes in GFP-MNB/DYRK1A-overexpressing cells of FKHR at 3 residues (Thr24, Ser253 and Ser319) catalyzed by protein kinase B through a phosphoinositide-3kinase-dependent pathway. MNB/DYRK1A phosphorylates Ser329 residue on FKHR in vivo [19], but upstream kinases that can phosphorylate and activate MNB/ DYRK1A are still unknown. The phosphorylation of these residues including Ser329 on FKHR has been reported to reduce the proportion of FKHR present within the nuclei and to decrease the ability of FKHR to stimulate gene transactivation [19][20][21][22]. MNB/DYRK1A in a discrete subnuclear structure may, therefore, play a role in the control of cell-cycle progression or apoptosis by regulating the nuclear level of FKHR.
HeLa cells with a speckled pattern in their nucleus progressed into M phase and produced 2 daughter cells, and both the condensation and subsequent segregation of chromosomes occurred in these cells as properly as in non-transfected control cells. On the other hand, in HeLa cells expressing a high level of MNB/DYRK1A, multinucleation was observed. Immunostaining with antibody specifically recognizing γ-tubulin revealed that the multinucleation had resulted from overduplication of the centrosome. Balczon and co-workers [23] reported that CHO cells arrested at the G1/S boundary of the cell cycle by treatment with hydroxyurea underwent multiple rounds of centrosome replication in the complete absence of DNA synthesis and cell division. Thus, one possible explanation for the overduplication of the centrosome, which was seen in this study, is that the overexpression of MNB/DYRK1A influences cell-cycle progression, possibly by affecting the nuclear level of FKHR. For cell division to occur properly, the centrosome must be duplicated once during each cell cycle; and thus in normal cells, the centrosome duplication cycle is tightly regulated. Failure of the normal cycle of this duplication would result either in cell-cycle arrest before the onset of mitosis or in the formation of an aberrant monopolar or multipolar spindle [24][25][26][27]. It is possible that the overexpression of MNB/ DYRK1A induces the overduplication of the centrosome prior to the next mitosis, subsequently producing multiple spindle poles leading to the multinucleation.
MNB/DYRK1A is a dual specificity protein kinase whose activity depends on the phosphorylation of tyrosines in its activation loop [28,29]. Outside this catalytic domain, the sequence comprises a bipartite nuclear translocation signal (amino acids 105-139). A previous report showed that the nuclear translocation of MNB/DYRK1A is mediated by this signal sequence but that its characteristic subnuclear localization depends on additional N-terminal elements [11]. Thus, the protein kinase activity of MNB/DYRK1A is not required for its subnuclear localization. We have also obtained data leading to the same conclusion: when cDNA encoding kinase activity-negative mutants, GFP-MNB/DYRK1A (K179R) and GFP-MNB/ DYRK1A (Y310F/Y312F), were used to transfect HeLa cells, speckled signals were detected in the nucleus (data not shown). On the other hand, cells with multinuclei were not observed if the kinase activity-negative mutants were overexpressed. It thus appears that the kinase activity of MNB/DYRK1A is essential for the formation of multinucleation, but not for its speckled subnuclear formation with FKHR.
Centrosome overduplication and multi-polar mitotic spindles were also observed in some tumor cells after γirradiation [30,31]. Since mitotic cell death was predominantly observed in these irradiated cells, cells expressing a higher level of MNB/DYRK1A than that in normal cells may die in a similar fashion. A possibility that mitotic cell death or apoptosis may occur in MNB/DYRK1A-overexpressing cells is worth studying, because brain development is markedly affected in Down syndrome patients and the number of neurons is reduced in their brain [32,33].

Conclusions
In this study, we found that overexpression of MNB/ DYRK1A induced overduplication of the centrosome during interphase, resulting in aberrant spindles and missegregation of chromosomes during mitosis and subsequent multinucleation. A major goal of Down syndrome research is to correlate dosage imbalance of specific genes from human chromosome 21 with various clinical aspects of the syndrome. Our experimental system using overexpression of MNB/DYRK1A is a very useful model for studying the effect of gene dosage in Down syndrome in vitro. Further studies on the molecular mechanism underlying centrosome dysregulation by overexpression of MNB/DYRK1A should provide more important insights into the role of this protein kinase in Down syndrome.

Materials
The following materials were purchased from the sources indicated: Mouse antibodies against α-tubulin, γ-tubulin, and FLAG epitope from Sigma Chemical Co.

Cell Culture
HeLa cells were maintained at 37°C in a 5% CO 2 atmosphere in Dulbecco's modified Eagle's (DME) medium supplemented with 10% fetal calf serum and 10 µg/ml kanamycin.

Immunostaining of cells for confocal laser scanning microscopic observation
Transient transfection with constructs encoding GFP-MNB/DYRK1A and FLAG epitope-tagged MNB/DYRK1A were carried out by using LipofectAMINE reagent as recommended by the manufacturer. Briefly, HeLa cells were seeded onto cover slips in 35-mm dishes at least more than 24 hours before transfection. The plasmid DNA was mixed with LipofectAMINE reagent in serum-free Opti-MEM, incubated at room temperature for 30 minutes, and then added to the seeded cells. The total amount of transfected plasmid DNA was 1.0 µg per dish. At 3 hours after the addition of the plasmid DNA, the transfection mixture was replaced with DME medium supplemented with 10% fetal calf serum. The cells were then incubated in a CO 2 incubator for 24-48 hours, washed twice with ice-cold phosphate-buffered saline (PBS), and then fixed with methanol for 5 minutes at -20°C. The fixed cells were washed 3 times with PBS and were subsequently incubated for 2 hours with anti-α-tubulin antibody (1:200), anti-γ-tubulin antibody (1:200), anti-FLAG antibody (1:200), or anti-MNB antibody (1:50) [7]. They were then incubated for 1 hour with 1 µM TOTO-3 and goat Cy3conjugated anti-mouse IgG antibody (1:100), donkey Cy3-conjugated anti-rabbit IgG antibody (1:100), or goat fluorescein isothiocyanate (FITC)-conjugated anti-mouse IgG antibody (1:100) at room temperature. The stained cells were observed with a confocal laser scanning microscope (MRC1024, Bio-Rad). A total of 200 cells were examined for centrosome number and for percentage of multinucleate cells in each of 3 independent experiments.

Microscopic observation of MNB/DYRK1A in living cells during the cell cycle
The procedure for preparation of fluorescently-stained living cells for microscopic observation was described previously [34,35]. Briefly, HeLa cells were plated on a 35-mm glass-bottom culture dish (MatTek Corp., Ashland, MA) and cultured for 1 day in a CO 2 incubator in 2 ml of DME medium supplemented with 10% fetal calf serum. The cells were then transiently transfected with the plasmid DNA encoding the GFP-MNB/DYRK1A fusion gene by using LipofectAMINE reagent. At 24 hours after the transfection, the cells were stained with 100 ng/ml of Hoechst33342 for 5-30 minutes and then washed 3 times with DME medium supplemented with 10% calf serum. The Hoechst33342-stained cells were cultured in a phenol red-free DME medium supplemented with 10% fetal bovine serum in a CO 2 incubator for at least 30 minutes and used for microscopic observation. This medium also contained 80 µg/ml of kanamycin sulfate and Hepes buffer (pH 7.4) at a final 20 mM concentration.

Author's contribution
EF carried out the construction of the expression vectors and immunocytochemical studies, and drafted the manuscript. TH and YH participated in the observation of dynamics of chromosomes and MNB/DYRK1A in living cells during the cell cycle. JK and TH participated in the design of the study. FI and NS conceived the study, raised the funding for the project, and participated in its design and coordination.
All authors read and approved the final manuscript.
Publish with Bio Med Central and every scientist can read your work free of charge