During morphogenesis in triploblastic animals, some cell types in the embryos perform a range of functions involving a variety of different cell behaviors. Identification of the range of functions performed by each cell type will be essential for a more complete understanding of how embryonic cells interact to achieve normal development (Wood and Jacinto, 2007). In starfish embryos, mesenchyme cells contribute to embryonic morphogenesis in at least two distinct ways. First, mesenchyme cells can control some components of the extracellular matrix (ECM). For example, they guide the basal lamina from the tip of the archenteron to the ectodermal stomodeum during mouth formation (Crawford and Abed, 1983; Abed and Crawford, 1986). Mesenchyme cells also exert mechanical tension against the fibrous component of the ECM to sustain embryonic and larval shape (Crawford, 1990; Crawford et al., 1997; Kaneko et al., 2005). Second, mesenchyme cells act as scavengers to clean out the blastocoel. The mesenchyme cells phagocytose cell debris dispersed in the blastocoel during reconstruction of gastrula embryos, thereby allowing morphogenesis to proceed (Tamura et al., 1998). The scavenger function of mesenchyme cells is also essential in normal bipinnaria larva for keeping the blastocoelic environment clean, in particular to deal with dying epithelial cells that fall into the blastocoel and with any unexpected foreign bodies that have invaded through the ectodermal wall (Furukawa et al., 2009). Thus, mesenchyme cells are multifunctional cells that support several aspects of starfish morphogenesis.
In some species of starfish, morphogenesis is achieved by a simple structural change in the epithelial monolayer that comprises the bulk of the embryonic and larval body (Kuraishi and Osanai, 1992). During the blastula stage, the epithelial layer forms a hollow sphere-like structure as a monolayer of cells. By the mid-gastrula stage, invagination of the vegetal portion of the epithelial monolayer occurs to initiate archenteron formation. In the mid-gastrula stage, the mesenchyme cells ingress into the blastocoel from the tip of the archenteron, and then disperse to the ectoderm and endoderm epithelial walls (Dan-Sohkawa et al., 1980). By the bipinnaria stage, the embryo has enlarged and successively formed several organs while preserving the epithelial monolayer throughout its body. With the exception of a transient period when a small number of muscle progenitor cells emerge from coelomic pouches (Crawford and Chia, 1978), mesenchyme cells are the only cells within the blastocoel. In 4-day-old bipinnaria larvae, mesenchyme cell numbers range from 150 to 190, approximately 1% of the total cell number in the larval body (Furukawa et al., 2009). Throughout this developmental period, the transparency of the embryonic and larval body of starfish allows visualization of the wide range of mesenchyme cell behaviors (Crawford and Chia, 1978; Dan-Sohkawa et al., 1980; Kaneko et al., 1990, 1995b; Kuraishi and Osanai, 1992).
One approach to identification of novel morphogenetic functions of mesenchyme cells is to exploit in vivo situations in which the embryonic body is comprised only of epithelial cells as the mesenchyme cells have yet to appear in the blastocoel. Comparison of the patterns of embryonic development in the presence or absence of mesenchyme cells will shed light upon possible novel functions of the mesenchyme cells. We previously developed an in vitro culture system in which a pure population of mesenchyme cells can be prepared for injection into the blastocoel of a blastula embryo (Kaneko et al., 1995a, b, 2005; Furukawa et al., 2009). Interestingly, in preliminary experiments using a variety of mesenchyme cell numbers, several of the blastula embryos grew larger than normal by the late gastrula and/or bipinnaria stages (see also Fig. 6). This finding indicates the possibility that mesenchyme cells induce epithelial cells to proliferate, resulting in the larger phenotype.
The purpose of this study was to determine whether mesenchyme cells have a function in inducing epithelial cells to proliferate and, to this end, we used our mesenchyme cell injection approach with blastula embryos of the starfish Asterina pectinifera. We also used a bacteria counter to assess total cell numbers in individual embryos and larvae. Our cytometric analyses and immunohistochemical observation of 5-bromo-2′-deoxyuridine (BrdU)-treated specimens showed that injection of mesenchyme cells could stimulate a significant increase in epithelial cell numbers. However, at the largest number of injected cells, we observed a lack of effect on epithelial cell proliferation and also morphogenetic abnormalities in the embryos. This response was associated with a severe condensation of the fibrous component of the extracellular matrix. We propose that epithelial cell proliferation in the starfish embryo requires the presence of an appropriately sized population of mesenchyme cells and also responds to the interaction of mesenchyme cells and the fibrous component of the extracellular matrix.
Blastula Embryos Show Proliferation of Epithelial Cells as They Develop to the Mid-gastrula Stage
The structures of blastula and mid-gastrula embryos were assessed with regard to the distributions of their constituent cell types (see the Experimental Procedures section). Blastula embryos were composed of a monolayer of epithelial cells and had a hollow sphere-like appearance (Fig. 1A,B). The epithelial cells retained this monolayer structure until the mid-gastrula stage (Fig. 1C,D). The blastula embryos formed an archenteron (endoderm) as they enlarged along the anteroposterior axis (Fig. 1C,D). Up to the mid-gastrula stage, no mesenchyme cells were present in the blastocoel. Cytometric analysis of 3 combined samples of 30 embryos yielded an average of 3,325.9 (SD ± 384.3) cells in each blastula embryo and of 4,744 (SD ± 384.1) cells in each mid-gastrula embryo. From the blastula to mid-gastrula stages, TUNEL (terminal deoxynucleotidyl transferase–mediated deoxyuridinetriphosphate nick end-labeling; Supp. Fig. S1, which is available online) and propidium iodide (PI) assays (data not shown) detected a small amount of apoptosis but no necrotic signals, respectively (Supp. Fig. S1).
Each Blastula Embryo Is Composed of 3,000–3,500 Cells
Blastula embryos from any given batch showed some variation in size, although they all appeared to have a similar, hollow sphere-like structure (Fig. 2A–F). We attempted to approximate the size of individual blastula embryos by calculating the volume occupied by the epithelial cells around the hollow sphere: the longest (R, r) and shortest axes (R′, r′) of both the outer and inner diameters of the sphere were measured (see Fig. 2G); (R+R′)/2 and (r+r′)/2 were defined as the outer diameter and inner diameters, respectively. Using the bacteria counter, the numbers of cells were separately determined for six blastula embryos in four independent experiments; total cell numbers were plotted against the estimated volume of each blastula embryo (Fig. 2H). Regardless of the significant differences in estimated volumes, the total cell numbers ranged only between 3,000 and 3,500. The plotted data in Figure 2H clearly demonstrate the absence of any relationship between total cell numbers and estimated volumes of the embryos; this is illustrated by comparison of the embryos labeled “a” and “c,” which have similar total cell numbers but differ widely in estimated volumes. The total cell numbers per embryo obtained using the bacteria counter are very similar to those estimated from the combined sample of 30 embryos (see also Supp. Table S1).
Injection of Mesenchyme Cells Into Blastula Embryos Increases Total Cell Numbers at the Mid-gastrula Stage
Because the bacteria counter produced a robust estimate of the total cell numbers in individual embryos, we used this method to address the question of whether mesenchyme cells have a morphogenetic function as an inducer of epithelial cell proliferation. Blastula embryos were injected with 40–50 cultured mesenchyme cells and allowed to develop until the mid-gastrula stage, at which time cytometric analysis was carried out (Figs. 3, 4). Control embryos were injected with similar numbers of ectodermal epithelial cells prepared from late-gastrula embryos; here, we term these cells “epithelial cells.” Immediately after injection, the mesenchyme or epithelial cells remained as an aggregate in the blastocoel (Fig. 3A,C). When blastula embryos injected with mesenchyme cells reached the mid-gastrula stage, the mesenchyme cells had separated and were dispersed beneath the epithelial cells of the ectodermal and endodermal wall (Fig. 3B). We found that the mesenchyme cell aggregates consisted of between 40 and 50 cells, based on embryos in which the cells had dispersed (Fig. 3B). In contrast, injected epithelial cells stayed in the aggregates in recipient embryos during the corresponding developmental period (Fig. 3D). These aggregates also appeared to contain between 40 and 50 cells based on analyses of paraformaldehyde (PFA)-fixed samples that had been stained with PI (Supp. Fig. S2). Sham-injected embryos, an additional control experiment, developed normally to the mid-gastrula stage (data not shown). There were no differences in apparent size and shape of the mid-gastrula embryos in the three experimental groups.
Between five and eight mid-gastrula embryos, similar to those depicted in Figure 3, were prepared for cytometric analyses from each experimental group (Fig. 4). We found that the embryos injected with mesenchyme cells had a significant increase in the total number of cells (approximately 1.3-fold) compared with those in two control experiments (Fig. 4, Supp. Table S1). The total cell numbers of mid-gastrula embryos in the two control experiments were similar to those obtained in our earlier analysis of the combined sample of thirty mid-gastrula embryos allowed to develop normally (see Supp. Table S1).
Injected Mesenchyme Cells Do Not Show Mitotic Activity in the Blastocoel of Recipient Embryos
To determine whether proliferation of the injected mesenchyme cells contributed to the increase in total cell number, we cultured some blastula embryos in artificial seawater containing BrdU. These embryos were then examined under the laser confocal microscope (Fig. 5). BrdU-positive signals were detected in almost all epithelial cells of the ectodermal and endodermal wall but not in mesenchyme cells (Fig. 5A,B). Optical sections of mesenchyme cells at a higher magnification confirmed that there was no indication of BrdU staining in these cells (Fig. 5C).
Effect of Injection With an Excess Number of Mesenchyme Cells on Later Morphogenesis
To examine whether varying the numbers of injected mesenchyme cells influenced later morphogenesis in recipient embryos, we compared development in embryos injected at the blastula stage with either the equivalent number of mesenchyme cells as in the experiments described above or with approximately three to four times that number. The injected embryos were allowed to develop to the 3-day-old bipinnaria stage. Examination of embryos with dispersed rhodamine B isothiocyanate-celite (RITC)-labeled mesenchyme cells after two days of development indicated that the two embryo treatment groups received approximately 40–50 or 150–200 cells, respectively (Fig. 6A,B,D,E). The injected mesenchyme cells were dispersed as single cells (Fig. 6C,F) and coexisted with the innate mesenchyme cells (see also Supp. Fig. S3). In the description below, we refer to these different quantities of injected mesenchyme cells as moderate or excess, respectively. A sham injection group was used as the control. Blastula embryos injected with the moderate number of mesenchyme cells showed normal morphogenetic development, except for size, compared with embryos in the sham control group at the late-gastrula and bipinnaria stages (Fig. 6B,C,H,I). In contrast, blastula embryos injected with the excess number of mesenchyme cells showed significantly altered development with some morphological abnormalities, e.g., the tip of the archenteron and the anterior portion of the ectodermal wall failed to bulge and constrict, respectively, 2 days after injection (Fig. 6E). Subsequently, further skewed morphogenesis resulted in 3-day-old bipinnaria larvae with abnormal morphologies, although the mouth, esophagus, stomach, intestine, coelomic pouches and ciliary bands did form (Fig. 6F). These larvae appeared smaller than those in the sham control group (Fig. 6C,F,I). In all three experimental groups, innate mesenchyme cells showed normal ingression into the blastocoels of recipient embryos.
Total cell numbers were counted in 3-day-old bipinnaria larvae of all three experimental groups using the bacteria counter (Fig. 7). A remarkable difference in total cell numbers was found in the larvae derived from the injection experiment with the moderate number of mesenchyme cells compared with those from the other two experimental groups. The 3-day-old bipinnaria larvae produced by blastula embryos that had been injected with the moderate number of mesenchyme cells had approximately a 1.3-fold larger total cell number than those from the sham control group. The total cell numbers in the 3-day-old bipinnaria larvae produced by blastula embryos that had been injected with the excess number of mesenchyme cells were not significantly different from those of the sham control group (means ± SD; n = 3).
Injection of an Excess Number of Mesenchyme Cells Causes a Severe Condensation of a Fibrous Component the ECM
We previously reported that anomalous epithelial shape formation is closely related to an abnormal distribution of a fibrous component (4H11 fibers) of the ECM (Kaneko et al., 2005). These fibers are formed by a 370-kDa proteinaceous molecule that is recognized by the monoclonal antibody 4H11 Mab (Kaneko et al., 2005). Here, we investigated the distribution patterns of 4H11 fibers in the 3-day-old bipinnaria larvae of the three experimental groups described in Figure 6. No significant differences in the distribution of 4H11 fibers were present between the 3-day-old bipinnaria larvae injected with the moderate number of mesenchyme cells and those in the sham experiment (Fig. 8A,C). By contrast, 3-day-old bipinnaria larvae that had received an excess number of mesenchyme cells showed an unusual distribution of 4H11 fibers. The blastocoelic space of these larvae lacked 4H11 fibers in the anterior portion of the larval body (Fig. 8B, asterisk). Moreover, the injected mesenchyme cells were detected at every location that had 4H11 fibers but were never seen in regions lacking 4H11 fibers (Fig. 8B). This morphology was common to all the larvae in the excess mesenchyme cell experimental group.
A detailed analysis was carried out to compare the morphologies of 4H11 fibers in the region around the mouth of the blastocoelic spaces of larvae of the moderate and excess mesenchyme cell groups. In the moderate cell number group, this area was characterized by a V-shaped adoral ciliary band and the buccal cavity was located in the upper half of the mouth (Fig. 8D). The 4H11 fibers were present as a fine meshwork emanating from underneath the ectodermal wall in the vicinity of the esophagus (Fig. 8E). In contrast, in the excess cell number group, the adoral ciliary band had an oblong appearance and no buccal cavity was formed (Fig. 8F). Moreover, the esophagus connecting the mouth and stomach was present as a flattened duct (data not shown). At the same focal depth as the image shown in Figure 8E, 4H11 fibers were severely condensed and many of the injected mesenchyme cells were located there (Fig. 8G). Severe condensation of the 4H11 fibers was observed in all 3-day-old bipinnaria larvae of the excess mesenchyme cell group.
To elucidate the multiple functions performed by mesenchyme cells, we investigated the effects on epithelial cell proliferation of injection of mesenchyme cells into blastula embryos of the starfish Asterina pectinifera. Three principal findings were obtained in this study: (1) Embryos injected with 40 to 50 mesenchyme cells at the blastula stage had an approximately 1.3-fold larger total cell number at the mid-gastrula and 3-day-old bipinnaria stages than control embryos (Figs. 4, 7). (2) In two control groups that were either injected with epithelial cells or had a sham injection, the mid-gastrula embryos and 3-day-old bipinnaria larvae were composed of approximately 4,600 to 5,000 cells and 15,000 to 17,000 cells, respectively (Figs. 4, 7). (3) Injected mesenchyme cells did not undergo mitosis in the recipient embryos (Fig. 5). These data demonstrate that mesenchyme cells have the ability to induce epithelial cell proliferation, at least under our experimental conditions. This function might be opposed by other functions of the mesenchyme cells during starfish development, such as control of the ECM or clearance of the blastocoel (see Introduction).
The present study deals with an aspect of the epithelial–mesenchymal interaction. Our experimental approach had two particular features: first, the use of blastula embryos for injection, and, second, the use of a bacteria counter to determine cell numbers. The blastula embryo is a very simple body structure, characterized by a hollow sphere-like epithelial monolayer (Fig. 1). Up to the mid-gastrula stage, the embryo increases in total cell numbers with little apoptosis (Supp. Fig. S1). In addition, it is also sufficiently early in development to precede the appearance of innate mesenchyme cells at the mid-gastrula stage (Fig. 1). This last property is advantageous as it allowed us to obtain clear results without any “noise” due to innate mesenchyme cells. In addition, use of a bacteria counter allowed us to carry out a cytometric analysis on total cell numbers in individual embryos and larvae. In a preliminary experiment, a conventional cell counter was found to give barely acceptable estimates. The data we obtained from individual mid-gastrula embryos of a sham control experimental group (Fig. 4, see also Supp. Table S1) are comparable to those reported by Dan-Sohkawa and Satoh (1978) who reported an average total cell number of 4,908 using a squash preparation method.
Several studies have examined the influence of mesenchyme cells on epithelial cell proliferation. Most of these previous studies have been concerned with vertebrate organogenesis, such as of the pancreas (Bhushan et al., 2001), liver (Matsumoto et al., 2008), skin (Ming Kwan et al., 2004), salivary gland (Yamamoto et al., 2008), or tooth (Fukumoto et al., 2006). To the best of our knowledge, the present study is the first to describe the proliferative profile of epithelial cells during in vivo morphogenesis of a deuterostome invertebrate. Because the starfish is phylogenetically close to the origin of the chordate line, the function of mesenchyme cells described in this study might also be present in other invertebrate embryos of the chordate line. Our observations raise the question of how this function has been preserved during evolution.
Our quantitative data provide two insights into the morphogenetic function of mesenchyme cells as an inducer of epithelial cell proliferation. First, they provide some indication of the potential of a single mesenchyme cell to induce proliferation of epithelial cells. Blastula embryos injected with approximately 50 mesenchyme cells showed an increase in total cell number of approximately 1,500 cells at the mid-gastrula stage (Fig. 4, Supp. Table S1). Thus, each mesenchyme cell potentially induces proliferation of an average of 30 epithelial cells from the blastula to mid-gastrula stage. This potential induction is estimated as 90 epithelial cells at the 3-day-old bipinnaria stage (Fig. 7, Supp. Table S2). Based on this analysis, a single mesenchyme cell can be inferred to govern proliferation of dozens of epithelial cells during at least a few days of development. One obvious caveat regarding this interpretation, however, is that our results were obtained from experiments involving injection of mesenchyme cells prepared in culture and not from transplantation experiments with innate mesenchyme cells.
Second, we need to consider the number of injected mesenchyme cells. Blastula embryos injected with the excess number of mesenchyme cells failed to show any increase in epithelial cell numbers (Fig. 7). This failure occurred without any indication of a disturbance to the appearance of the innate mesenchyme cells (Supp. Fig. S3). These observations suggest that the ratio of the number of mesenchyme cells to that of epithelial cells has a limiting influence on the ability of the epithelial cells to proliferate normally in interaction with mesenchyme cells during morphogenesis. It is of particular interest that failure of epithelial cell proliferation occurred concomitantly with abnormal morphogenesis in the larvae (Fig. 6). In particular, the blastocoelic space of the abnormal larvae lacked 4H11 fibers. Absence of 4H11 fibers from the blastocoelic spaces of larvae was also seen after treating starfish blastula embryos with the 4H11 Mab (see Fig. 6B of Kaneko et al., 2005). We also showed previously that some 4H11 fibers play a role as a flexible substrate that facilitates epithelial movements beneath ectodermal and endodermal walls (Kaneko et al., 2005). In this study, the altered distribution patterns of the 4H11 fibers was a direct consequence of an excess number of mesenchyme cells, which resulted in a change from the normal fine meshwork of 4H11 fibers to a severely condensed appearance (Fig. 8E,G). Moreover, many of the injected mesenchyme cells were observed to penetrate into the structures formed by the severely condensed 4H11 fibers (Fig. 8G). It is likely that the severe condensation of the 4H11 fibers renders them incapable of acting as a flexible substrate for epithelial movement and thus is an underlying cause of the failure of epithelial morphogenesis and epithelial cell proliferation. Our analyses suggest that epithelial cell proliferation may be regulated by a close interaction of the proper number of mesenchyme cells with 4H11 fibers.
Our observations here also suggest that mesenchyme cells play a role as supportive cells for epithelial cells as they proliferate during morphogenesis. It is unclear, at present, to what extent innate mesenchyme cells interact with epithelial cells to induce proliferation, and, if so, for how long and at which times during development. Two important questions were identified from this study. The first concerns normal starfish developmental processes: do innate mesenchyme cells have a ubiquitous role in epithelial cell proliferation to grow the embryonic body or do they participate in specialized organogenesis? The second question that needs to be addressed is the underlying molecular mechanism(s) that determine how mesenchyme cells function as a promoter of epithelial cell proliferation during normal embryonic development.
Mature eggs were obtained from the ovaries of starfish (Asterina pectinifera) treated with 1-methyladenine (Kanatani, 1969), and were fertilized with diluted sperm from the testes. Fertilized eggs were allowed to develop until the blastula, mid-gastrula of 3 days-bipinnaria stages at 20°C in artificial seawater (ASW, Marineart SF-1, Tomita Pharmaceutical). For the purposes of this study, we defined the blastula stage as newly-hatched embryos, and the mid-gastrula stage as those with invagination of the archenteron halfway across the blastocoel and just before the appearance of mesenchyme cells. These developmental stages were reached at 16 hr and 24 hr after fertilization, respectively.
Injection of Mesenchyme Cells or Epithelial Cells
A pure population of mesenchyme cells for injection was prepared from cell cultures as described previously (Kaneko et al., 1995a). In brief, the cells were washed with ASW, then labeled with rhodamine B isothiocyanate-celite (RITC) in culture at room temperature (RT) for 2 hr (Ettensohn and McClay, 1988). The cells were washed with ASW and then dislodged from the culture dish using a rubber policeman. The epithelial cells for injection were obtained from late-gastrula embryos. The embryos were first labeled with RITC for 6 to 10 hr, by which time they had reached the late gastrula stage. They were washed with ASW and the epithelial cells were removed from the ectoderm as described previously (Kaneko et al., 1990). There was no evidence of fluorescence decay of the RITC label in the mesenchyme or epithelial cells during the next 3 days. Mesenchyme or epithelial cells were aspirated into a microneedle, and injected into the blastocoel of each blastula embryo as described previously (Kaneko et al., 2005). The numbers of injected mesenchyme or epithelial cells were counted as they were ejected from the microneedle. In the sham experiment, a microneedle was pushed into the blastocoel and then removed without injecting any solution. Each injected embryo was separately placed in 2 ml of ASW in a 24-well multi-plate (Sumitomo Bakelite Co., Ltd.), and allowed to develop at 20°C to the desired developmental stages. They were analyzed using a laser scanning confocal microscope (Olympus FluoView FV300) equipped with a helium-argon laser.
Two methods were used for cell counting. First, 30 embryos were collected and briefly pretreated in dissociation medium containing 0.1% trypsin (DMET: 1.2 M glycine; 2 mM EDTA; 0.1% trypsin). This medium was then replaced with 100 μl of DMET, and the embryos were incubated at RT for 5 min. They were dissociated into single cells by aspiration with a micropipette, and cell numbers were counted using a Thoma hemocytometer (#A106, Sunlead Glass Corporation; SLGC, Japan). An estimate of the total cell number of each embryo was then made. In the second approach, total cell numbers were determined separately for each embryo or larva. The embryos were treated as above except that only 2 μl of DMET was used, and a Sigmacote (Sigma)-coated mouth pipette and bacteria counter (#A161, SLGC, Japan) were used. The bacteria counter is designed to allow determination of cell numbers even in a small volume of a live cell suspension.
Immediately after injection with mesenchyme cells, some blastula embryos were cultured in 5-bromo-2′-deoxyuridine (5 μM; BrdU, Sigma) in ASW until the mid-gastrula stage when they were fixed.
Fixation and Staining
In some experiments, embryos were fixed with 4% paraformaldehyde (PFA) in ASW at RT for 20 min or at 4°C over night. After washing with phosphate-buffered saline containing 0.01% Triton X-100 (PBST), the PFA-fixed embryos were post-fixed with ice-cold acetone at −20°C for 15 min. In the experiment to assess the structure of the embryos, the fixed normal blastula and mid-gastrula embryos were incubated in RNase A (200 μg/ml; Boehringer Mannheim, Germany) at 37°C for 20 min to remove the cytoplasmic RNA that quenches nuclear fluorescence signals at these developmental stages. The embryos were then incubated in PBST containing 0.165 μM of fluorescein isothiocyanate (FITC)-labeled phalloidin (Sigma) and 1 μg/ml propidium iodide (PI; Sigma) at RT for 30 min. In the BrdU treatment experiment, the fixed mid-gastrula embryos were incubated with 2N HCl at RT for 2 hr. They were stained with FITC-conjugated rat anti-BrdU (1:20 dilution with PBST, Oxford Biotechnology Ltd.) at RT for 30 min. Analyses of the distribution pattern of 4H11 fibers and of the structure of the fibrous component of the extracellular matrix were performed using samples that were immunofluorescently stained with the 4H11 Mab as described previously (Kaneko et al., 2005). In each experiment, samples were washed three to five times with PBST at RT after staining. The embryos were mounted on tape spacer, and viewed under an Olympus laser scanning confocal microscope as described above.
We thank members of the Asamushi Marine Biological Station of Tohoku University for supplying the starfish. We also thank Drs. Ritsu Kuraishi and Ryohei Furukawa, Keio University, for critical discussion, Dr. Katsuhiro Ohta (Department of Mathematics, Keio University) for critical suggestions regarding quantitative analyses in this study. G.H. was supported by a KLL Research Grant for Ph.D. Program (Keio University) and H.K. received a Keio Gijuku Fukuzawa Memorial Fund for the Advancement of Education and Research.