Planarians have regenerative ability made possible by pluripotent stem cells referred to as neoblasts. Classical ultrastructural studies have indicated that stem cells can be distinguished by a unique cytoplasmic structure known as the chromatoid body and their undifferentiated features, and they are specifically eliminated by X-ray irradiation. Recently, by using fluorescence activated cell sorting (FACS), planarian cells were separated into two X-ray-sensitive fractions (X1 and X2) and an X-ray-insensitive fraction (XIS) according to DNA content and cytoplasmic size. Here we analyzed the fractionated cells by transmission electron microscopy (TEM). First, we found that both undifferentiated cells (stem cells) and regenerative cells (differentiating cells) were concentrated in the X1 fraction containing the S/G2/M phase cells. The regenerative cells were considered to be committed stem cells or progenitor cells, suggesting that some stem cells may maintain proliferative ability even after cell fate-commitment. Second, we succeeded in identifying a new type of stem cells, which were small in size with few chromatoid bodies and a heterochromatin-rich nucleus. Interestingly, they were concentrated in the X2 fraction, containing G0/G1 phase cells. These results suggest that planarian stem cells are not homogeneous, but may consist of heterogeneous populations, like mammalian stem cells.
Planarians have the ability to regenerate their whole body after being cut into many segments (Agata et al. 2007). How can planarians regenerate their entire body from small fragments? It has been shown that planarians have many pluripotent stem cells throughout the mesenchymal space, referred to classically as neoblasts. It was believed that in planarians, neoblasts were the only cells with the capacity to proliferate, giving rise to all cell types during regeneration, and that once committed or differentiated, cells could no longer proliferate. However, the detailed properties of planarian neoblasts are still unclear. It has been suggested that a large G2-arrested cell population does not exist, based on BrdU-labeling and staining with antiphospho-histone H3 (Newmark & Sánchez Alvarado 2000). Although the cell cycle parameters of neoblasts have begun to be revealed, it has still not been elucidated whether planarians have stem cells of a single type or multiple types (Agata & Watanabe 1999; Shibata et al. 1999; Salvetti et al. 2000; Agata 2003; Reddien & Sánchez Alvarado 2004).
Morphological examination of neoblasts shows that they are small in size, and possess a high nucleus/cell size ratio. Additionally, classical ultrastructural studies have shown that planarian neoblasts have typical undifferentiated features, with minimal cytoplasm containing a few mitochondria, many free ribosomes, and little or no endoplasmic reticulum (Hay & Coward 1975; Hori 1982, 1992a, 1997), and a unique cytoplasmic structure known as the chromatoid body (Morita et al. 1969; Hay & Coward 1975; Hori 1982). In general, chromatoid bodies are composed of amorphous materials of moderate electron density and are present near the nuclear pores (Morita et al. 1969; Hori 1982; Hori & Kishida 2003).
Molecular biological studies have shown that chromatoid bodies contain RNA (Hori 1982; Auladell et al. 1993). In 1999, our group focused on the morphological similarity of the chromatoid body to the germ granule in other animals, and then screened planarian genes for planarian homologs of major components of germ granules, and finally succeeded in isolating the first marker gene for the planarian neoblasts, DjvlgA (Dugesia japonica vasa-like gene A; Shibata et al. 1999). After this success, many different groups isolated homologs of germplasm components from planarians, and many homologs that are specifically expressed in neoblasts or germline stem cells were identified, such as pumilio homolog Djpum, piwi homologs Smedwi and Djpiwi, smed-bruno-like gene bruli, and the nanos-related gene Djnos (Reddien et al. 2005; Salvetti et al. 2005; Guo et al. 2006; Rossi et al. 2006; Sato et al. 2006).
During the process of differentiation, the size and number of chromatoid bodies are progressively reduced (Hori 1982; Auladell et al. 1993), and the neoblast-like cells start to develop Golgi apparatus and rough endoplasmic reticulum (Hori 1992a, 1997). Thus, ‘regenerative cells’, which have features intermediate between those of neoblasts and differentiated cells, appear. Such cells can be observed extensively during the process of regeneration in the stump region of the blastema (Hori 1983, 1992b, 1997).
It is also well known that planarian stem cells are sensitive to X-ray irradiation (Wolff & Dubois 1948). By using FACS (fluorescence activated cell sorting) our laboratory previously succeeded in fractioning planarian cells into X1 (X-ray sensitive fraction 1), X2 (X-ray sensitive fraction 2) and XIS (X-ray insensitive fraction) cell fractions. These fractions were defined according to their DNA and cytoplasmic content and comparison of the sorting patterns of intact and X-ray-irradiated planarian cells. The relative proportions of the three cell populations X1 : X2 : XIS were about 1:2:6 (Hayashi et al. 2006). These three fractions were further characterized by quantitative real-time polymerase chain reaction (PCR) analysis using molecular markers, which revealed that the X1 fraction predominantly contained proliferating cells that were X-ray sensitive, and the X2 fraction contained a limited number of non-proliferating cells that were X-ray sensitive. Consequently, it was suggested that the X1 fraction cells are stem cells in the S/G2/M phase, while X2 cells include stem cells in the G1/G0 phase. The XIS fraction contains a variety of differentiated cells, including neurons and muscle cells (Hayashi et al. 2006).
The aim of the current study was to characterize the cells in the above three fractions by ultrastructural analysis using criteria that allow one to judge the differentiation stage. Initially, we ascertained whether or not the X-ray-sensitive cells are morphologically stem cells. Next, we examined the X1 and X2 fraction stem cells in more detail, and further investigated the relationship between the ultrastructural features and the characteristic state of X1 and X2 stem cells, such as cell cycle phase and developmental stage. Furthermore, we hoped to elucidate with respect to their morphological features whether planarian stem cells consist of a single cell type or multiple cell types.
Materials and methods
A clonal strain, SSP (diploid, 2n = 16), of the planarian Dugesia japonica was used. This strain was derived from a single progeny that hatched from an egg laid by a sexualized worm of strain GI (mixoploid, 2n + 3n; Orii et al. 1999). The worms were asexually maintained in autoclaved tap water at 22–24°C and fed chicken liver twice a week (Ito et al. 2001). Animals 8 mm in length were starved for more than 1 week prior to the experiment, and used for cell sorting.
Preparation for FACS
Planarians were cut into small pieces on ice with a scalpel. The small pieces were treated with 0.1% trypsin (DIFCO, Franklin Lakes, NJ, USA) in 5/8 Holtfreter's solution for 1 h at 20°C (Asami et al. 2002; Ogawa et al. 2002; Hayashi et al. 2006). The dissociated cells were stained using the following fluorescent dyes: Hoechst 33342 (Sigma, St Louis, MO, USA) for DNA staining and Calcein AM (Sigma) for cytoplasm staining, both for 2 h at 20°C. Subsequent staining with propidium iodide (PI; Dojindo, Kumamoto, Japan) to determine cell viability was carried out for 10 min on ice, immediately before cell sorting (Hayashi et al. 2006).
Electron microscopy of dissociated cells
The cells isolated by FACS were centrifuged at 1800 g for 10 min using a swinging-bucket rotor (Tomy, Tokyo, Japan, TMS-21). The samples were fixed in 1.2% glutaraldehyde in 0.1 m sodium cacodylate buffer (pH 7.4) for 1 h at 4°C. The samples were then washed for 15 min in 0.1 m sodium cacodylate buffer and postfixed in 2% osmium tetroxide in the same buffer for 1 h at 4°C, and then encompassed with 0.1% agarose gel in the same buffer to prevent the cells from being dispersed. The samples were dehydrated at 4°C by passage through a series of increasing concentrations of ethanol (70–80–90–95–100%) and finally through acetone for 15 min. After infiltration with QY-1 (Nisshin EM Corporation, Tokyo, Japan) for 30 min, cells were embedded in Epon 812 (Nisshin EM Corporation). Ultra-thin sections were cut with a diamond knife on an ultramicrotome. The sections were stained with 4% uranyl acetate and lead citrate, and observed by transmission electron microscopy.
Electron microscopy of intact animals
The cells isolated by FACS were fixed in 2.5% glutaraldehyde in 0.1 m sodium cacodylate buffer (pH 7.4) for 1 h at 4°C. The samples were then washed for 15 min in 0.1 m sodium cacodylate buffer and postfixed in 2% osmium tetroxide in the same buffer for 1 h at 4°C. After rinsing for 15 min in the buffer, the samples were dehydrated at 4°C by passage through a series of increasing concentrations of ethanol (70–80–90–95–100%) and finally through propylene oxide for 15 min. These samples were embedded in Epon 812. The subsequent methods were carried out as described in the previous section (Hori 1997).
Cells isolated by FACS were counted from the electron micrographs. The cell types: stem cells, differentiating cells and differentiated cells, were classified based on criteria from classical ultrastructural studies of regeneration blastemas (Hay & Coward 1975; Hori 1982, 1997).
Measurement of stem cells
Stem cells were measured to determine the length, area of the nucleus, area of the cytoplasm, and the ratio of the area of the nucleus to the cell size. Measurements were carried out on a Macintosh computer using the public domain NIH Image program (US National Institutes of Health, http://rsb.info.nih.gov/nih-image/).
Dissociated planarian cells can be classified into three types
The cells in intact planarians have been classified into three types according to ultrastructural criteria indicating the differentiation stage: neoblasts, differentiated cells and intermediate cells between neoblasts and differentiated cells, also called regenerative cells (Hori 1992a, 1997; see Introduction). We classified the cells of each FACS fraction, X1, X2 and XIS, obtained from whole planarians using similar criteria (Fig. 1). For this, we randomly picked up about 100 cells from each FACS fraction and categorized them into three cell types, namely, stem cells, differentiating cells and differentiated cells. Three typical examples of FACS-sorted cells are shown in Figure 2. The first has typical undifferentiated cytoplasm with chromatoid bodies (Fig. 2A,A′). This cell was categorized as a ‘stem cell’, which may be comparable to the ‘undifferentiated cells’ in intact planarians. The second cell has slightly developed multilayered rough endoplasmic reticulum (RER), but otherwise has cytoplasmic features resembling those of stem cells (Fig. 2B,B′). Thus, it was categorized as a ‘differentiating cell’, which may be comparable to the ‘regenerative cells’. The third, a differentiated cell, has developed RER and other distinguishing characteristics such as organelles specific to that cell type. The cell of Figure 2C could be judged a rhabdite-forming cell by the specific organelles (Fig. 2C,C′).
However, images of some cells could not be classified into the three categories. For example, some cells showed cytoplasmic characteristics similar to stem cells and differentiating cells, but chromatoid bodies were not observed in the section. It is possible that chromatoid bodies from those cells might have been present in other sections. However, in the current study, we could not test that possibility, so these cells were categorized as unclassified cells (Fig. 2D,D′). The four types of cells, stem cells, differentiating cells, differentiated cells and unclassified cells, are summarized in Figure 2E.
Stem cells are highly enriched in the X1 fraction
The cells of the X1, X2 and XIS fractions were classified into three categories: ‘stem cells’, ‘differentiating cells’ and ‘differentiated cells’ according to their ultrastructural characteristics. The percentages of cells in each category are summarized in Figure 3. The XI fraction was previously defined as the fraction containing predominantly X-ray sensitive cells, with strong Hoechst staining, which specifically stains nuclei, and weak Calcein AM staining, which specifically stains cytoplasm. It was also suggested that the cells in the X1 fraction consist of stem cells in the proliferative phase, because they had high Hoechst intensity, indicating that they had high DNA content (Hayashi et al. 2006). Sixty-seven percent of the 108 cells counted in the X1 fraction were identified as stem cells, i.e. they had both chromatoid bodies and minimal cytoplasm (Fig. 3, red column), indicating that the X1 fraction consisted predominantly of stem cells.
Seventeen percent of the cells in the X1 fraction were classified as differentiating cells since they had slightly developed RER. However, these cells retained other cytoplasmic features similar to those of the stem cells, including chromatoid bodies (Fig. 3, yellow column), and thus may have corresponded to the regenerative cells in intact animals (Hori 1997). It has been thought that differentiating cells are not dividing. However, unexpectedly, these cells were detected in the X1 fraction, which contains cells in S/G2/M phase.
Only 4% of the X1 fraction cells showed differentiated morphology (Fig. 3, green column). Additionally, 12% of the cells had cytoplasmic characteristics similar to those of stem cells and differentiating cells, but chromatoid bodies were not observed in the section. This population of cells is represented by the white column (Fig. 3). These results suggest that the majority of proliferating cells consisted of stem cells and differentiating cells.
X2 fraction is a mixture of stem cells and differentiated cells
The X2 fraction was defined as the fraction of cells with low Hoechst and Calcein AM intensity, and approximately half of the cells were X-ray sensitive (Hayashi et al. 2006). It was technically much more difficult to obtain a sufficient number of cells from this fraction for ultrastructural observation using our experimental conditions, as compared to X1 and XIS fraction cells. This might have been due to problems resulting from this being a population of small cells. Finally, we were able to observe 115 cells in the X2 fraction. Nineteen percent of the cells counted in the X2 fraction were identified as typical stem cells with minimal cytoplasm and chromatoid bodies (Fig. 3). The remaining cells consisted of differentiating cells (2%), differentiated cells (56%) and unclassified cells (23%). Various differentiated cells were identified as gland cells, rhabdite-forming cells and muscle cells according to their cytoplasmic morphology, but most of the cells in this fraction seemed to be fragmented. It is notable that many unclassified cells (23%) were also observed in this fraction. If these unclassified cells were stem cells that possessed chromatoid bodies not detectable in the particular section examined, then the total fraction of stem cells detected by ultrastructural analysis in X2 (19% plus 23%) would closely correspond to the fraction of X-ray-sensitive cells in X2 (about half).
Absence of stem cells in the XIS fraction
The XIS fraction was defined as the fraction containing X-ray-insensitive cells with low Hoechst and high Calcein AM staining intensity (Hayashi et al. 2006). Almost all (97%) of the 117 cells counted in the XIS fraction were differentiated cells (Fig. 3). These differentiated cells included gland cells, rhabdite-forming cells, muscle cells, flame cells and some unclassifiable differentiated cells. Chromatoid bodies were found in only six of these differentiated cells. The fact that stem cells were not detected ultrastructurally in this fraction of X-ray-insensitive cells confirmed that stem cells are sensitive to X-ray irradiation. The absence of stem cells in the XIS fraction indicates that the X1 and X2 fractions successfully captured much of the whole spectrum of planarian stem cells.
Interestingly, the XIS fraction did not contain differentiating cells, except for two cells (2%). Contrary to our expectation, differentiating cells (which are X-ray sensitive and contain a large quantity of DNA) may be able to divide, because such cells were not present in the XIS fraction and were concentrated in the X1 fraction instead.
Stem cells can be classified into two types by their morphology
In the course of comparing the compositions of the cells among the fractions, we found that the stem cells could be further classified into two types according to whether their nucleus appeared to contain a large amount of euchromatin (A type) or heterochromatin (B type) (Fig. 4). Among the stem cells in the combined X1 and X2 fractions, 75 cells were judged to be A type and 11 cells B type, according to their nuclear features, and eight cells could not be clearly categorized as A or B, and were not further analyzed. Ultrastructural comparison of stem cells classified as A type (with a euchromatin-rich nucleus) and B type (with a heterochromatin-rich nucleus) revealed that the A type had a large cell size with numerous chromatoid bodies in the cytoplasm (Fig. 4A,A′,A″), while the B type was smaller in size, with few chromatoid bodies (Fig. 4B,B′,B″).
The length, areas of the nucleus and cytoplasm, and ratio of the nucleus to whole cell size were measured using NIH Image software (see Materials and Methods; Fig. 5A–C). For this analysis, the same FACS-sorted fractions of cells as used in the experiment shown in Figure 3 were used. The average of the longest length in A type cells was 6.5 ± 1.3 µm, and that in the B type cells was 4.7 ± 0.8 µm (P < 0.005). The cell length of the A type was thus 1.8 µm (27.7%) longer than that of the B type (Fig. 5A). In addition, the area of the nuclei of the A type was 14.8 ± 5.8 µm2, while that of the B type was 10.1 ± 2.8 µm2 (P < 0.005). The area of the nuclei of the A type was larger than that of the B type by about 4.7 µm2 (31.6%) (Fig. 5B). The nucleus/cell size (N/C) ratio of the A type was 52.3 ± 10.8%, while that of the B type was 63.7 ± 7.2% (P < 0.005). The N/C ratio of the B type was larger than that of the A type by 11.4% (Fig. 5C). The number of chromatoid bodies was also counted. About half of the A type cells contained four or more chromatoid bodies (data not shown), whereas there were no B type cells containing more than three chromatoid bodies, and about half of the B type cells contained only one chromatoid body (data not shown). The average number of chromatoid bodies in A type cells was 3.3 ± 2.0, and that in B type cells was 1.6 ± 0.7 (Fig. 5D) (P < 0.005). Thus, the A type cells possessed more chromatoid bodies than the B type cells (Fig. 5D). These results indicate that the A and B types of dissociated stem cells showed distinct ultrastructural and morphometric features.
The XI fraction was highly enriched in A type stem cells: about 89% of the stem cells in the X1 fraction were categorized as the A type. In contrast, B type stem cells were concentrated in the X2 fraction (32%) (Fig. 5E).
Identification of type A and B stem cells in intact planarian
In previously reported ultrastructural studies, type B-like stem cells have never been described. To investigate whether these cells can be identified only when they are dissociated, or can be detected in the intact planarians, we observed the cells in intact planarians using the same fixation conditions and staining methods. We carefully checked the mesenchymal space, and succeeded in the identification of candidate cells. In Figure 6A the gross location of the type B-like cells is indicated. They are located in the mesenchymal space, but close to the body muscle layer underneath the ciliary epithelial cells. A higher-magnification view of Figure 6A is shown in Figure 6B. Both type A-like and type B-like cells could be observed and easily distinguished according to the differences of their nuclear features (Fig. 6B). The A-like neoblasts with a euchromatin-rich nucleus (Fig. 6B, lower cell) had a large cell size and many chromatoid bodies in the cytoplasm (Fig. 6C). The B-like neoblasts with a heterochromatin-rich nucleus (Fig. 6B, upper cell) were smaller in size and had fewer chromatoid bodies (Fig. 6D).
In total, 15 A-like neoblasts and 12 B-like neoblasts were examined in intact planarians. The length, areas of the nucleus and cytoplasm, and ratio of the nucleus to whole cell size were measured using NIH Image software (see Materials and methods; Fig. 6E–G). The average of the longest length in A-like neoblasts was 9.6 ± 2.9 µm, and that in the B-like neoblasts was 6.2 ± 1.4 µm (P < 0.005). The cell average length of the type A-like neoblasts was 3.4 µm (35%) longer than that of the type B-like neoblasts (Fig. 6E). In addition, the area of the nuclei of A-like neoblasts was 17.8 ± 6.3 µm2, while that of the B type was 10.2 ± 3.3 µm2 (P < 0.005). The area of the nuclei of A-like neoblasts was larger than that of the B type by about 7.6 µm2 (42.7%) (Fig. 6F). The nucleus/cell size ratio of the type B-like neoblasts was larger than that of the type A-like neoblasts by 13.5% (Fig. 6G) (P < 0.005). The average number of chromatoid bodies in A-like neoblasts was 4.4 ± 2.1, and that in B type cells was 2.1 ± 0.9 (Fig. 6H) (P < 0.005). Thus, the A-like neoblasts possessed more chromatoid bodies than the B-like neoblasts (Fig. 6H).
We conclude that type B cells were not artificially formed in the process of dissociation, but rather that they really existed in the mesenchymal space in vivo. They were concentrated in the X2 fraction, and by using FACS, we were able to identify the type B cells as a newly detected subpopulation of stem cells.
Detection of a differentiating metaphase cell in intact planarian
When we observed stem cells in intact planarians by transmission electron microscopy (TEM), we found a metaphase cell possessing the characteristic cytoplasmic features of differentiating cells with chromatoid bodies and slightly developed RER cisternae. A typical cell with these features is shown in Figure 7. Condensed chromosomes with high electron density were observed in the central region of the cell, and the cell did not have a nuclear membrane. According to these features, this cell was judged to be in metaphase. Interestingly, the cell possessed slightly developed RER cisternae, as well as a chromatoid body. These features of the cell indicated that this cell was not a stem cell (undifferentiated cell) but a differentiating cell (regenerative cell) of the metaphase stage. In the analysis of FACS-sorted planarian cells, differentiating cells were most abundant in the X1 fraction with strong Hoechst staining (Fig. 3). According to our observation of the cell shown in Figure 7 in an intact planarian, we conclude that differentiating cells may still have the ability to divide.
Ultrastructural characterization of three FACS fractions of planarian cells
Ultrastructural analysis of the three FACS fractions of cells from whole planarians revealed that those fractions had distinct profiles with respect to their state of differentiation. The XI fraction contained predominantly stem cells and a smaller proportion of differentiating cells (Fig. 3). Half of the X2 fraction cells were differentiated cells, while at least one-fifth of the cells in this fraction were stem cells. The XIS fraction contained many types of differentiated cells, such as gland cells, rhabdite-forming cells, muscle cells, flame cells and other unclassifiable differentiated cells, but no stem cells. The present ultrastructural analysis and the previously reported quantitative real-time PCR data of proliferating cell markers and X-ray sensitivity of the cells in the FACS fractions (Hayashi et al. 2006) together provide strong evidence that highly purified populations of planarian stem cells were isolated by FACS.
Two types of stem cells
The stem cells were classified into two types, named A and B, according to the difference of their euchromatin/heterochromatin content at the electron microscopic observation level (Fig. 4). A type stem cells (defined as having a euchromatin-rich nucleus) were large in size, with many chromatoid bodies. B type stem cells (defined as having a heterochromatin-rich nucleus) were small in size, with few chromatoid bodies (Fig. 5). The A type stem cells were enriched in the X1 fraction, and X1 fraction cells had high Hoechst staining intensity, suggesting that the A type stem cells were in proliferative phase. In contrast, B type stem cells were scarcely detected in the X1 fraction. In previous ultrastructural studies, type B-like stem cells have never been described. Consequently, we checked the mesenchymal space in intact planarians to find the neoblasts, and we thereby succeeded in identification of type B-like stem cells (Fig. 6). This result suggested that the type B-like stem cells exist in the intact animal, and are not artificially formed during the process of dissociation, FACS sorting and fixation for TEM.
There are two possibilities regarding the relationship of the two types of stem cells. The first possibility is that the difference of A and B type cells may simply be due to their being in different phases of the cell cycle. However, it has not been reported that the G1 cell possesses a heterochromatin-rich nucleus. The second possibility is that the B type cell may be a stem cell of a new class, which is in the G0 state. The difference of the number of chromatoid bodies suggests that the difference of cell state is not due to a simple difference of the phase of the cell cycle. To clarify this point we will need further analysis at the molecular level.
Do intermediate cells maintain proliferative capacity?
We found that differentiating cells, which were characterized ultrastructurally by their possession of slightly developed RER cisternae, and thus appeared to correspond to the regenerative cells in intact planarians, were most abundant in the X1 fraction (Fig. 3). Interestingly, the X1 fraction was defined as the fraction with high intensity of Hoechst staining, and with predominantly X-ray sensitive cells, suggesting that the differentiating cells of the X1 fraction may maintain proliferative activity. When we observed stem cells in intact planarians, we found a metaphase cell possessing the characteristic cytoplasmic features of the differentiating cells with chromatoid bodies and slightly developed RER cisternae (Fig. 7). The structures indicated by arrows in Figure 7B were judged to be slightly developed RER rather than nuclear membrane for three reasons. First, the structures closely resembled the slightly developed RER shown in regenerative cells in Hori (1992a). Second, although Morita & Best (1984), described dispersed pieces of nuclear membranes seen in the periphery of metaphase cells, we considered that the structures we observed were too long to be such dispersed pieces. Third, the structures were located separately from the condensed chromosomes, not surrounding them. Together, these present observations lead us to suggest that the stem cells after commitment or progenitor cells may maintain proliferative ability.
We have summarized the findings of this study in Figure 8. In the classical view, planarian stem cells were considered to be homogeneous, and called neoblasts (Fig. 8A). However, our studies show a more complex view of the planarian stem cell system (Fig. 8B). The stem cells of planarians can be divided into at least two types: A and B types, and the stem cells after commitment may maintain proliferative activity, like mammalian stem cells. Recently, our colleagues clearly demonstrated that some neoblasts have become committed in germ line stem cells expressing the germ line-specific gene ‘Djnos’ and start to proliferate after sexualization, even though they show the same morphology as the A type stem cells (Sato et al. 2006). Understanding the heterogeneity of stem cells may be a key to further elucidating the mechanisms of regeneration and reproduction in planarians.
This work was supported by Special Coordination Funds for Promoting Science and Technology (KA), and a Grant-in-Aid for Creative Research (KA) and Scientific Research on Priority Areas (KA). We thank Dr T. Inoue for teaching us how to use NIH Image. We are grateful to Dr E. Nakajima and Mr J. Pulvers for correction of the English and the discussion in this report. We also thank other members of our laboratory for their encouragement. We are grateful to Dr S. Yonemura, Ms. M. Uwo, Ms. N. Inoue, Dr M. Hirata (RIKEN CDB) and Dr M. Kashikawa for technical advice about TEM.