During the development of vertebrate embryos, neural crest cells develop at the junction between the neural plate and the surface ectoderm. They migrate along spatially and temporally distinct pathways and differentiate into a variety of neuronal and nonneuronal cell types that are important in the formation of craniofacial structures, the cardiovascular system, and much of the peripheral nervous system (Le Douarin and Kalcheim, 1999). It has been shown that some cells in the cranial and trunk neural crests die by apoptosis during migration and that such cell death may play crucial roles in differentiation and migration of neural crest cells (Jeffs and Osmond, 1992; Jeffs et al., 1992; Hirata and Hall, 2000). Jeffs et al. (1992) observed cell death in specific populations of cranial neural crest (CNC) cells and postulated that it may contribute to the patterning of early CNC migration. In the trunk of embryos, Wakamatsu et al. (1998) demonstrated that some neural crest cells undergo apoptosis on the lateral migration pathway, and they argued that such cell death may help neural crest cells find the right migration pathway and remove “ectopic” cells by a “proofreading mechanism.” They suggested that some environmental cues regulate the survival and death of neural crest cells.
Developmental neuronal cell death has been reported to occur in both the central and peripheral nervous systems and is assumed to contribute to eliminating excessive and inappropriately located cells (Jacobson, 1991; Wakamatsu et al., 1998; Maynard et al., 2000). Various mechanisms have been proposed for developmental neuronal death, and it seems that the mechanism of cell death is not the same at different locations and at various stages in development. Since apoptotic death of neural crest cells often occurs after they leave the neural tube and become nonepithelial, we postulated that the survival of neural crest cells may depend on anchorage-dependent cell spreading and the maintenance of normal cytoarchitecture. Such anchorage-dependent survival has been noted in some epithelial cell types (Boudreau et al., 1995a, 1995b) but has not been demonstrated so far in neural crest cells.
To investigate whether anchorage-dependent cell spreading and maintenance of cytoarchitecture are required for survival of CNC cells, we undertook an in vitro study where their actin cytoskeleton and attachment to the substratum were disrupted by chemical agents. Cultured CNC cells emigrating out of the explanted mouse embryonic neural tube were treated with cytochalasin D and matrix metalloproteinase-2 (MMP-2) to disrupt their actin cytoskeleton and adhesion to the underlying substratum, respectively. Cytochalasin D blocks the formation of microfilament structures by abrogating actin polymerization and cytochalasin D-treated cells cannot maintain normal morphology (Korn, 1982). MMP-2 degrades extracellular matrix (ECM) components such as type IV and V collagens, laminin-5, elastin, fibronectin, and proteoglycans, thereby disrupting cell-ECM interactions (Okada et al., 1990; Giannelli et al., 1997; Zuo et al., 1998). It has been shown that MMP-2 has an important functional role in epithelial-mesenchymal transformation (EMT) and migration of neural crest cells and that perturbation of MMP activity may lead to neural crest-related congenital defects (Cai et al., 2000; Cai and Brauer, 2002; Duong and Erickson, 2004). We analyzed the effects of cytochlasin D and MMP-2 on the morphology and survival of cultured CNC cells by cytochemical staining for actin microfilaments and by in situ labeling of DNA fragmentation (TUNEL method). The disruption of actin fiber organization and adhesion to the substratum was followed by apoptotic death of cultured CNC cells, suggesting that properly organized cytoarchitecture and anchorage-dependent cell spreading are essential for survival of CNC cells.
MATERIALS AND METHODS
ICR strain mice (SLC Japan, Shizuoka, Japan) were maintained in a temperature- and humidity-controlled animal facility with a 12-hr light/12-hr dark cycle. Animals were given laboratory chow and tap water ad libitum. Virgin female mice were mated overnight with a male mouse, and the noon of the day on which a vaginal plug was found was taken as embryonic day 0.5 (E0.5). On E8.5, embryos were obtained by Caesarean section and transferred to sterile Tyrode buffer solution. The experimental protocol was approved by the Committee for Animal Experimentation of Kyoto University Graduate School of Medicine.
Neural Crest Cell Explant Culture
Primary culture of mouse CNC cells was carried out according to the method originally described by Ito and Takeuchi (1984) and Murphy et al. (1991). Each mouse embryo (E8.5) was treated with 0.25% trypsin/PBS for 15 min and the hindbrain neural folds were isolated. The dorsal ridge of the neural folds was surgically removed free from the surrounding tissues and cultured on a 35 mm culture dish coated with fibronectin (Becton Dickinson). The culture medium contained 90% α-MEM medium (Gibco-BRL), 10% fetal bovine serum (Gibco-BRL), 4% chick embryo extract (Gibco-BRL), 50 U/ml penicillin, 50 μg/ml streptomycin (Gibco-BRL), and 1% insulin-transferrin-selenium-X (Gibco-BRL). The cultures were maintained for 48 hr at 37°C in a CO2 incubator under a humidified atmosphere composed of 95% air and 5% CO2. At 48 hr after culture, the neural folds in the colonies were scraped away with syringe needles, and the remaining CNC cells, which had migrated out of the neuroepithelium, were used for the following experiments.
Cranial neural crest cells were identified by immunohistochemical staining with a rat antimouse monoclonal antibody 4E9R (Kubota et al., 1996), which was kindly provided by Dr. Kazuo Ito (Osaka University Graduate School of Science, Toyonaka, Japan). The 4E9R antibody has been shown to identify migratory mouse neural crest cells (Kubota et al., 1996).
Treatment With Cytochalasin D and MMP-2
After preparation of cultured CNC cells as described above, the culture medium was replaced by a serum-free medium to eliminate the influence of substances contained in fetal bovine serum. Then, either cytochalasin D (1 μg/ml; Sigma, St. Louis, MO) or MMP-2 (2.0 unit/ml; Cosmo Bio, Tokyo, Japan) was added to the medium for experimental groups. For stock solutions, cytochalasin D and MMP-2 were dissolved in DMSO and PBS, respectively. The final concentration of cytochalasin-D (1 μg/ml) was determined according to previous studies in which microtubule assembly and cell shape of cultured cells were disrupted (Nishi et al., 2002; Nemeth et al., 2004). The concentration of MMP-2 (2.0 unit/ml) was used because the concentration 1.0 unit/ml or below had no observable effects on cultured CNC cells. In control groups, an equivalent volume of the vehicle was added to the culture medium. After adding the chemicals, each culture dish was further incubated in a CO2 incubator and was observed and photographed at every hour using a Nikon Diaphoto phase-contrast microscope (Nikon, Tokyo, Japan). At 12 hr after treatment, cells in culture dishes were fixed with 4% paraformaldehyde for further stainings and the TUNEL reaction.
Dual Staining for Actin Fibers and Nuclear DNA
The fixed CNC cells were washed three times with PBS, immersed in 0.1% Triton X-100/PBS, then incubated for 10 min at room temperature with 33 μM phalloidin- rhodamin (Sigma) and 0.025% Hoechst33342 (Sigma) dissolved in PBS. After incubation, the cells in the dishes were washed three times with PBS and coverslipped. The cells were observed using an Axioplan 2 fluorescence microscope (Carl Zeiss, Germany) equipped with fluorescein filters and were photographed using a CoolSNAP CCD camera (Photometrics).
In Situ Labeling of DNA Fragmentation by TUNEL Method
The in situ visualization technique for DNA fragmentation was carried out according to the TUNEL method (Gavrieli et al., 1992; Mori et al., 1995). Fixed CNC cells were washed three times with double-distilled water, air-dried, and treated with 5 μg/ml proteinase K (Wako Pure Chemical, Tokyo, Japan). After proteinase K treatment, the cells were incubated in TdT buffer containing 12.5 μM biotinylated-dUTP (Boehringer Mannheim) and 0.15 unit/μl TdT (Takara, Kyoto, Japan) at 37°C for 70 min. Then, they were reacted with rhodamine-avidin (Sigma) for visualization of fragmented DNA. After being rinsed several times with PBS, the samples were coverslipped and observed. The samples were photographed using a Zeiss Axioplan 2 fluorescence microscope and a CoolSNAP CCD camera (Photometrics). For each sample, the ratio of TUNEL-positive cells to the total cells was calculated. Cell counts were executed on the monitor of a Power Macintosh computer (Apple) using a counting tool of the National Institutes of Health Image program (http://rsb.info.nih.gov/nih-image).
Morphologic Changes of Cultured CNC Cells Treated With Cytochalasin D and MMP-2
In control cultures, numerous cells began to emigrate out of the explanted neural tube within 4 hr after culture and their emigration and mitotic divisions continued for over 72 hr. Migrating cells at the peripheral part of the colonies showed the bipolar or multipolar morphology, which is typical for neural crest cells, as has been described by Ito and Takeuchi (1984) (Fig. 1A). Immunohistochemical staining with a monoclonal antibody against neural crest cells (4E9R) (Kubota et al., 1996) revealed that over 90% of the cells emigrating out of the explanted neural tube were 4E9R-positive, indicating that they were CNC cells (Fig. 1B). By 60 hr in culture, many of those emigrated cells began to extend axon-like protrusions and appeared to undergo morphologic differentiation (data not shown). Therefore, the cells cultured for 48 hr were used for the following experiments.
When CNC cells were cultured with cytochalasin D (1 μg/ml), obvious morphologic changes were observed within 1 hr. Many cells showed a blebbing appearance, which is characteristic to apoptotic cells. By 12 hr after culture, the cells retracted their cell protrusions (filopodia) and became shrunken and rounded (Fig. 1C). When MMP-2 (2.0 units/ml) was added to the culture medium, cultured cells showed morphologic changes such as membrane blebbing and retraction of filopodia, and the cell area was reduced as compared with controls (Fig. 1D). Some cells became rounded.
Morphologic Changes of Nucleus and Actin Fiber Organization of Cultured CNC Cells
When the cells in control cultures were subjected to dual staining for actin fibers and nuclear DNA, the morphology of cell nuclei and the arrangement of actin stress fibers in the cytoplasm appeared normal (Fig. 2A and B). Actin fibers were well organized and arranged orderly in their cytoplasm. The cells were spread well and had numerous filopodia.
When the cultured CNC cells were treated with cytochalasin D, they became remarkably shrunken within 1 hr and their actin fibers became disorganized and condensed around nuclei (Fig. 2C). Most of the cells showed nuclear shrinkage and chromatin condensation (Fig. 2D), which are often observed in apoptotic cells. When cultured cells were treated with MMP-2, many of the cells appeared to retract filopodia and the cell area was reduced by 12 hr after treatment, indicating that the cell-matrix attachment was disrupted to some extent (Fig. 2E). Some cells with cytoplasmic shrinkage were associated with actin fiber condensation around the nuclei, and condensed nuclei were intensely stained with Hoechst 33342, indicative of possible apoptotic bodies (Fig. 2F).
Apototic Death of Cultured CNC Cells
In each culture, apoptotic cells were identified by the TUNEL method, which detects DNA fragmentation, and the number of TUNEL-positive cells was counted. TUNEL-positive cells often had condensed nuclei and small apoptotic bodies (Fig. 3A–C). The ratio of TUNEL-positive cells to the total number of cells was calculated by counting at least 200 cells in each culture and the data were compared between the groups (n = 5 for each group). The frequencies of apoptotic CNC cells in cultures treated with cytochalasin D and MMP-2 were 13.4% ± 2.0% (mean ± SE) and 20.1% ± 2.0%, respectively, which were significantly higher than the control value (5.4%% ± 0.5%; Fig. 3C). The frequency of apoptotic cells was higher in the MMP-2-treated group than in the cytochalasin D-treated group, although the morphologic change was less severe in the former group (Figs. 1 and 2).
By using in vitro culture of migrating CNC cells, we have demonstrated that the survival of CNC cells is dependent on their cytoarchitecture maintained by actin stress fibers as well as on anchorage-dependent cell spreading. When the intercellular actin fiber organization of cultured CNC cells was disrupted by cytochalasin D, they rapidly became rounded and underwent apoptotic morphologic changes. It is interesting to note that the dying cells showed such morphologic alterations before they became TUNEL-positive. Since cytochalasin D abrogates actin polymerizaion and disrupts actin filaments (Walling et al., 1988), it is likely that malfunction of actin fibers and/or disruption of normal cytoarchitecture can induce cell death without directly affecting nuclear DNA. Although DNA fragmentation is a major molecular landmark of apoptotic cell death, it has been shown that apoptosis is a process involving some cytoplasmic alterations and does not necessarily require nuclear DNA fragmentation or other changes in the nucleus at its initial step (Cohen et al., 1992; Falcieri et al., 1993; Oberhammer et al., 1993). It has been well accepted that the maintenance of cell morphology and differentiation is dependent on the actin cytoskeleton (Hay, 1993). In addition, recent studies suggest that the cytoskeleton and cortical actin network play critical roles in various intercellular signaling (Aplin and Juliano, 1999; Zoubiane et al., 2004). Therefore, disruption of the actin cytoskeleton by cytochalasin D may not only have affected the cytoarchitecture but also have precluded effective signaling via the cytoskeleton, resulting in cell death.
MMP-2 degrades various ECM proteins including fibronectin, type IV and V collagens, laminin, and elastin (Vassalli and Pepper, 1994; Werb and Chin, 1998). Recently, Duong and Erickson (2004) demonstrated that MMP-2 is expressed as neural crest cells detach from the neural epithelium during EMT but is rapidly extinguished as they disperse. They also showed that MMP inhibitors and knockdown of MMP-2 expression perturb EMT that generates neural crest cells but do not affect migration of neural crest cells, suggesting that MMP-2 plays a crucial role at some steps in EMT of neural crest cells but is not required for the later migration process. Cai et al. (2000) observed the expression of MMP-2 mRNA in the craniofacial region of avian embryos and found that early migrating CNC cells do not synthesize MMP-2 mRNA but can interact with extracellular MMP-2 protein synthesized by the mesoderm. Furthermore, Cai and Brauer (2002) showed that neural crest migration is decreased when the MMP activity is perturbed. Thus, it seems that proteolytic degradation of ECM by MMP-2 is required at some stages of neural crest development but its expression may be finely regulated spatially and temporally so that the differentiation and migration of neural crest cells can take place properly. Patch mutant mice, which have a deficit in MMP-2 and membrane-type MMP expression and decreased migratory capacity of craniofacial mesenchyme (Robbins et al., 1999), exhibit neural crest-related craniofacial and cardiac defects (Morrison-Graham et al., 1992; Schatteman et al., 1995).
We showed in the present study that treatment of cultured CNC cells with MMP-2 disrupts cell-matrix interaction and induces apoptotic death. When treated continuously with MMP-2, cultured CNC cells retracted their protrusions (filopodia) and appeared to detach from the substratum, increasing TUNEL-positive cells by 12 hr. Thus, adhesion of CNC cells to ECM seems essential not only for maintaining their normal cell shape but also for their survival. This finding is consistent with the previous finding that some epithelial cells undergo apoptosis when they are separated from the basement membrane (Schmidt et al., 1993; Frisch and Francis, 1994). We have confirmed that degradation of ECM components by MMP-1 (type I collagenase) also resulted in apoptotic death of cultured mouse CNC cells (data not shown). In the present study, it was noted that the proportion of TUNEL-positive cells, which is indicative of DNA fragmentation, was higher in MMP-2-treated cells than in cytochalasin D-treated cells, although the morphologic effect appeared more severe in the latter (Figs. 2 and 3). When the CNC cells were treated with 1 μg/ml cytochalasin D, many cells became rounded and their nuclei appeared condensed. However, their nuclei became rodlike (Fig. 2D), which were different from typically pycnotic nuclei, as were seen in MMP-2-treated cells (Fig. 2F). Thus, actin fiber disorganization induced by cytochalasin D may not instantly result in apoptotic death of CNC cells.
It has been reported that various types of epithelial cells need anchorage-dependent cell spreading for survival and that they undergo apoptotic death when their cell-matrix attachment is disrupted (Boudreau et al., 1995a, 1995b; Roberts et al., 2002). Frisch and Francis (1994) coined the term “anoikis” for such a kind of apoptotic cell death. It is likely that adhesion-dependent regulation of cell survival is mediated by integrin signaling and that efficient cell-ECM adhesion is required for forming focal adhesion plaques by clustering ECM-receptor integrins. It has been shown that some signaling pathways of tyrosine phosphorylation mediated by focal adhesion kinase (FAK) are required for organization and maintenance of actin stress fibers to assemble cytoskeleton-associated proteins, such as paxillin, vinculin, talin, tenascin, and α-actinin, at the cytoplasmic domains of integrins (Clark and Brugge, 1995; Miyamoto et al., 1995). Frisch et al. (1996) showed that the interaction of integrins with ECM proteins can activate FAK and thereby suppress apoptosis in endothelial and other epithelial cells.
In the case of endothelial cells, not only cell-ECM adhesion but also cell spreading are important for preventing them from entering the process of apoptosis. Re et al. (1994) demonstrated that endothelial cells cultured under low concentrations of fibronectin or vitronectin became rounded and underwent rapid cell death, while the cells became flattened and remained viable under high substrate concentrations. They concluded that cell-matrix attachment is not sufficient for sustaining cell viability but cells need to achieve some shape changes to survive. Chen et al. (1997) directly examined the effects of cell spreading on growth and viability and showed that growth and survival of cultured endothelial cells increased as the extent of cell spreading increased but were independent of the total area of cell-ECM contact. Furthermore, it was shown that DNA synthesis is tightly coupled with the cell shape (Folkman and Moscona, 1978) and that stretch stimuli activate DNA synthesis and cell proliferation (Lansman et al., 1987; Olesen et al., 1988). These results support the hypothesis that unfavorable alterations of cell architecture and cell spreading can affect the survival and differentiation of some cells and trigger their apoptosis. During the migration of neural crest cells, their interaction with ECM may be crucial. It is likely that some proteinases including MMPs synthesized by mesenchymal cells help the EMT and migration of neural crest cells but their expression needs to be temporally and spatially regulated not only to facilitate their differentiation but also to keep them viable.
Recent molecular studies have demonstrated that a cystein protease family caspase, which is an ICE/CED-3 gene product, is involved in the onset of apoptosis (Miura et al., 1993; Nicholson et al., 1995; Kuida et al., 1996; Mashima et al., 1997; Miller, 1997). Caspase-1 (ICE) and caspase-3 (CPP-32), which cleave the existing actin fibers as a substrate, are activated following the disruption of cell-ECM adhesion (Boudreau et al., 1995b; Mashima et al., 1997). Actin is an inhibitor of deoxyribonuclease I (DNase I), which is a candidate endonuclease responsible for apoptotic DNA fragmentation (Peitsch et al., 1993), and the actin cleaved by ICE/CPP-32 loses the ability to inhibit DNase I (Kayalar et al., 1996). These data are consistent with the assumption that the disruption of actin fiber organization is a critical event for commencement of apoptosis. It is possible that the perturbation of actin microfilaments and the disruption of cell adhesion to ECM or adjacent cells can induce cell shape changes and perturb some signaling pathways in the cell.
As for the roles of caspases in developmental apoptosis, Umpierre et al. (2001) demonstrated that activated caspase-3 and DNA fragmentation were colocalized in the mesenchymal cells of branchial arches and in neuroepithelial cells of day 9 mouse embryos. The cells expressing caspase-3 were found to be abundant in the mesenchyme of the first and second branchial arches, which coincided with the CNC cell migratory areas in E8.5 mouse embryos (Kubota et al., 1996). It was also shown that caspase-3-deficient mice exhibit embryonic or early postnatal lethality and their central nervous system development was severely impaired (Kuida et al., 1996). Further, caspase-3 deficiency in mice resulted in decreased embryonic neuroblast apoptosis and neoplastic growth of the brain (Pompeiano et al., 2000). Thus, caspase-3 seems to play a critical role in the induction of morphogenetic apoptosis in embryos, especially in the nervous system. These reports warrant further investigation for elucidating the mechanisms of differentiation and migration of CNC cells as well as the significance of their apoptotic death observed in craniofacial morphogenesis.
The authors are grateful to Dr. Kazuo Ito, Osaka University Graduate School of Science, for providing the 4E9R antibody and Dr. Shigehito Yamada for his technical assistance.