SEARCH

SEARCH BY CITATION

Keywords:

  • primitive streak;
  • midline;
  • cell death;
  • caspase inhibitors;
  • left–right asymmetry

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental Procedures
  6. Acknowledgements
  7. REFERENCES

During embryogenesis, left–right sidedness is established by asymmetric expression of laterality genes. A recent model predicts the presence of a functional midline that divides the left side of the embryonic disc from the right side, separating left- and right-inducing signals. We show evidence that this midline is formed from a distinct population of cells within the primitive streak. Cells in the dorsal midline of the chick primitive streak display unique expression of the gastrulation markers fgf-8 and brachyury. These midline cells are fated to die, and dead cells remain in the midline during gastrulation. Inhibition of midline cell death compromises the early expression of laterality genes, such as shh and nodal and randomizes the direction of heart looping. We suggest that cell death along the primitive streak midline is a novel mechanism involved in the regulation of left–right asymmetry during early embryogenesis. © 2002 Wiley-Liss, Inc.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental Procedures
  6. Acknowledgements
  7. REFERENCES

All bilateria, both deuterostomes and protostomes, develop the left and right sides of the body by bisecting one embryonic field into the two fields along the midline axis (Brown and Wolpert, 1990; Levin and Nascone, 1997). In higher organisms, bilateral embryonic fields become asymmetric. This mechanism ensures that laterality in organ formation, such as unidirectional looping of the heart and gut and unilateral budding of the liver from the gut is established (Levin and Nascone, 1997; Yost, 1998). The left–right sidedness in organ formation is regulated by asymmetric expression of laterality genes, such as shh and nodal (Levin et al., 1995; Levin and Mercola, 1999). Complications with midline and laterality identities lead to numerous congenital defects, such as cyclopia, situs inversus, and heterotaxia (Abernethy, 1793; Brown and Wolpert, 1990). However, little is known about the mechanisms to induce and/or maintain a functional midline that divides the left side of the embryo from the right side, separating left signals from right signals.

The future midline axis in nonamniotes is set up by the localization of maternal factors during oogenesis (Nusslein-Volhard, 1991; Grunert and St Johnston, 1996) or by the sperm entry site (Gerhart et al., 1981). In amniotes, however, sperm entry site has been linked only to the separation between the embryonic and nonembryonic cell lineages (Piotrowska and Zernicka-Goetz, 2001) and has not been implicated in the positioning of the midline. Division of the embryonic disc into left and right halves in nonrodent amniotes is not detectable until formation of the primitive streak, an organizing center of gastrulation, begins. The primitive streak originates from the posterior side of the blastodisc and grows anteriorly to extend across two thirds of the blastodisc (Hamburger and Hamilton, 1951). Although the streak bisects the blastodisc, it consists of many cells in width. Therefore, the exact position of the future midline is still uncertain at this stage. Once primitive streak formation is completed, a finer midline within the streak is visualized by formation of the primitive groove, where epiblast cells ingress to form the mesoderm and endoderm. However, nothing is known about the exact mechanism that defines the midline axis within the primitive streak.

Subsequently, asymmetric expression of laterality genes begins just after formation of Hensen's node at the anterior end of the streak and before development of the notochord. The notochord itself is known to be important for maintaining the position of the midline and in the asymmetric orientation of the organs during embryogenesis (Danos and Yost, 1996; Lohr et al., 1997; Melloy et al., 1998; Rebagliati et al., 1998; Sampath et al., 1998; Izraeli et al., 1999). Little is known about the mechanism that triggers the asymmetric expression of laterality genes. However, it is likely that the midline axis is set up before initiation of the asymmetric gene expression that determines the identity of the left and right embryonic fields.

The present study examines when the midline axis of the chick embryo is defined. Evidence is presented demonstrating that, before formation of Hensen's node and the primitive groove, a dorsal midline population (Lawson and Schoenwolf, 2001) within the primitive streak becomes distinguishable from other streak cells by their unique gene expression and cell migration. Histologic and electron microscopic analyses revealed that the morphology of these midline cells resembles that of a cell undergoing cell death. Higher levels of active Caspase 3 and propidium iodide staining were found in the midline population. Application of a Caspase family inhibitor, Z-VAD-FMK, resulted in reduced propidium iodide staining in the midline and randomization in the direction of heart looping. These data show that there is a unique population of cells within the primitive streak that delineates the midline axis before asymmetric gene expression. The midline cells may play an important role in establishing and/or maintaining the left–right axis during embryogenesis.

Results and Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental Procedures
  6. Acknowledgements
  7. REFERENCES

Identification of Midline Cells

In the amniote, asymmetric expression of laterality genes begins only after Hensen's node, an organizing center of gastrulation, develops at the rostral end of the primitive streak (Levin et al., 1995). Before this molecular event, it is hypothesized that a midline must be established to set up this asymmetric patterning to separate the left side from the right side of the developing embryo. To test this hypothesis, we examined the expression of fgf-8 (Levin et al., 1995) and brachyury (Wilson et al., 1995), genes essential for gastrulation in the chick embryo (Fig. 1). At prestreak stages, these gastrulation genes were expressed broadly in posterior regions of the blastodisc (Fig. 1a,c), with no indication of differential midline expression. However, as soon as primitive streak formation began (Hamburger and Hamilton, 1995), a midline cell population became molecularly distinct from the lateral streak cells (Fig. 1b,d–h). This previously unidentified midline population was first visualized as a straight line of cells with a higher level of fgf-8 expression than surrounding cells in the developing primitive streak (Fig. 1b,e). As the primitive streak developed further, these midline cells down-regulated the two marker genes (Fig. 1d,f–h), whereas expression of both genes was up-regulated in the lateral streak cells. This unique expression pattern of gastrulation genes was restricted to cells of the dorsal midline (Lawson and Schoenwolf, 2001) within the primitive streak (Fig. 1i–l). These results demonstrate that molecular specification of the dorsal midline is already initiated in early streak formation, before the development of Hensen's node and primitive groove.

thumbnail image

Figure 1. Midline cells are molecularly specified before Hensen's node formation. Expression of gastrulation genes fgf-8 (a,b,e,f,i,j) and brachyury (c,d,g,h,k,l) was examined with RNA in situ hybridization at the prestreak Hamburger and Hamilton (HH) stage 1(9) (a,c), the mid-streak HH stage 3 (b,d,e,g), and the definitive streak HH stage 4 (f,h). Note that only higher expression was selectively detected with submaximal color development. i–l: Transverse sections of e–h, respectively; Purple staining, the area positive for transcripts; arrows, the midline population exhibiting a unique expression pattern of gastrulation genes in the primitive streak midline; ps, primitive streak; hn, Hensen's node; L and R, left and right sides of embryonic discs, respectively.

Download figure to PowerPoint

To examine how the identity of this midline is maintained during gastrulation, the fate of the primitive streak cells was traced. By using a replication-defective retrovirus with a lacZ reporter gene (Wei and Mikawa, 2000a, b), the early primitive streak was labeled and the embryos allowed to gastrulate. The majority of tagged cells ingressed and migrated from the primitive streak (Fig. 2a), following their known migration pathways of gastrulation (Lawson and Schoenwolf, 2001). Importantly, there was a distinct cell population that did not ingress, but remained in the midline of the primitive streak throughout gastrulation (Fig. 2a–c). This noningressing midline cell population may be responsible for maintaining midline identity during gastrulation. However, an array of accumulated cell fragments (Bancroft and Bellairs, 1974) was detected along the primitive groove (Fig. 2d–f). In addition, the midline cells were highly positive for ubiquitinated proteins (Fig. 2g), a marker for epithelial-mesenchymal transformation and cell migration (Wunsch and Hass, 1995). Alternatively, the conjugated form of ubiquitin is a known component of cell death pathways in development (Schwartz et al., 1990). The positive staining in the midline may also indicate regulated cell death, giving rise to the cell fragments detected along the primitive groove.

thumbnail image

Figure 2. Midline cells undergo cell death during primitive streak formation. a–c: Retroviral cell labeling and tracing of primitive streak cells. LacZ-containing viral particles were injected into the primitive streak at Hamburger and Hamilton (HH) stage 3 before mesoderm formation, and the fate of labeled cells was examined at HH stages 5 (a) and 8 (b). c: High-power view of the primitive streak in b. d–f: Scanning electron photomicrographs of HH stage 5 primitive streak (d), and high-power views of Hensen's node (e) and the primitive groove (f) of HH stage 8 streak. g: Antibody staining for the conjugated form of ubiquitin. h,i: Detection of the activated form of Caspase-3 in midline cells of the developing primitive streak at HH stages 2+ (h) and 3 (i). j–l: Propidium iodide staining of control (j), z-VAD-FMK treated (k), and z-WEHD-FMK treated (l) HH stage 4 embryos. Arrows, midline cells; hn, Hensen's node; hf, head fold; ps, primitive streak; blue line, outline of the developing streak.

Download figure to PowerPoint

Midline Cells Undergo Cell Death

To test the possibility that the cells in the midline were undergoing cell death, tests specific for nonnecrotic cell death were carried out. The activated form of Caspase 3, a proteinase specifically activated in multiple pathways of programmed cell death (Vaux and Korsmeyer, 1999), was also detected in the midline cell population in the developing primitive streak by using immunohistochemical analysis (Fig. 2h,i). Furthermore, nuclei of the midline cells were positively stained with propidium iodide (Fig. 2j), an impermeable DNA-binding dye used to detect cell death (Telford et al., 1992). Importantly, introduction of the Caspase family inhibitor benzyloxycarbonyl-val-ala-asp (OMe) fluoromethylketone (z-VAD-FMK) (Ekert et al., 1999) in ovo during primitive streak formation resulted in a significant loss of propidium iodide-positive cells in the midline (Fig. 2k). The effect of z-VAD-FMK was dose-dependent, saturating in the range of 10 micromolar of the antagonist added. No significant loss of propidium iodide-positive nuclei resulted from treatment with an inhibitor for Caspase-5, z-WEHD-FMK, which is not a main component in embryonic programmed cell death (Jacobson et al., 1997; Meier et al., 2000) (Fig. 2l). These results suggest that the midline cell population is fated to undergo cell death.

If cell death along the primitive streak midline plays a role in maintaining midline identity, dead cells or cell debris must persist in the midline during gastrulation. However, cell debris resulting from apoptotic or necrotic cell death is known to be quickly scavenged by neighboring cells (Meier et al., 2000; Vaux and Korsmeyer, 1999). Therefore, we hypothesized that the cell death process along the midline may be distinct from ordinary apoptosis or necrosis, as is the case of other examples of cell death during embryonic development (Schwartz et al., 1993; Achlze-Osthoff et al., 1994; Jacobson et al., 1997; Vaux and Korsmeyer, 1999). Indeed, propidium iodide- and lacZ-positive cells were found to remain in the midline and exhibited no significant segregation from the dorsolateral primitive cells (Fig. 3a–c). Although midline cells were clearly distinct from other streak cells by their lower electron density, higher content of subcellular granules, and accumulated cell fragments above them (Fig. 3d,e), it was evident that they remained incorporated in the dorsal layer of the primitive streak during gastrulation.

thumbnail image

Figure 3. Midline cells maintain their position in the primitive streak during gastrulation. a,b: Fluorescein-conjugated phalloidin staining of adherence belts (green), outlining the propidium iodide-positive midline cells (red) (a) and the lateral streak cells (b). c: Transverse section of the late primitive streak labeled with lacZ-virus at Hamburger and Hamilton stage 3 as Figure 1. A lacZ-positive cell (arrow) remained in the dorsal midline. d,e: Transmission electron photomicrographs of the dorsal midline of the primitive streak. Note that midline cells (asterisks) remain incorporated in the dorsal layer of the streak, just underneath vesicular fragments (arrows) accumulated in the groove. e: High-power view of the midline cells in c. f: Highly fenestrated (arrowheads) cell membrane of dorsal midline cells. g: Loss of membrane fenestration in midline cells after z-VAD-FMK treatment.

Download figure to PowerPoint

Another distinct feature of the midline cells was the highly fenestrated cell membrane (Fig. 3f), perhaps resulting from production of cell fragments seen along the midline. This unique fenestration of the cell membrane, together with unrestricted accessibility of propidium iodide to nuclei, suggested that the midline cell membrane may be highly permeable or leaky. This is consistent with the model that establishment of left–right asymmetry requires disruption of intercellular communications between the left and right sides of the embryonic disc (Levin and Nascone, 1997; Levin and Mercola, 1999), because increased membrane leakiness is a known trigger to close intercellular connexon channels (Goodenough et al., 1996). Importantly, the fenestration of the cell membrane was significantly lost in z-VAD-FMK–treated embryos (Fig. 3g) in which midline cell nuclei were no longer stained with propidium iodide (Fig. 2k).

Inhibition of Midline Cell Death Results in Laterality Defects

If the unique features of cell death in the midline play a role in establishing the laterality of the embryo, inhibition of midline cell death should result in a compromised left–right asymmetry. To test this possibility, z-VAD-FMK–treated embryos were allowed to develop further. They underwent normal gastrulation compared with control, sham-treated embryos, although cell death along the streak midline was greatly diminished (Fig. 4a–d). However, the z-VAD-FMK–treated embryos exhibited an altered expression of laterality genes, such as sonic hedgehog (shh) and nodal, which are involved in specification of left–right asymmetry (Levin et al., 1995). In control embryos, the expression of shh and nodal was localized primarily to the left side of the Hensen's node (Fig. 4e–g). In contrast, in z-VAD-FMK–treated embryos, the expression of shh was significantly down-regulated and expanded diffusely into the right side of the node (Fig. 4h,i). Down-regulation of nodal was also evident (Fig. 4j) as seen in some laterality mutants (Bisgrove et al., 2000), although the expression was still asymmetric. No significant changes were detected in expression of a posterior-lateral marker cdx (Marom et al., 1997; Ehrman and Yutzey, 2001) and an anterior marker rax (Ohuchi et al., 1999) (not shown).

thumbnail image

Figure 4. Inhibition of midline cell death randomizes the laterality of heart looping. a: Control stage 5 embryo with fully developed primitive streak (ps) and Hensen's node (hn). b: Propidium iodide signals in the streak midline of a. c: Normal formation of ps and hn in z-VAD-FMK–treated embryo. d: Propidium iodide signals are absent from the midline of c. e–j: RNA in situ hybridization analysis of the expression of shh (e,f,h,i) at stage 5 and nodal (g,j) at stage 7 in control (e–g) and z-VAD-FMK–treated embryos (h–j). Note the down-regulated, diffuse expression of laterality genes (arrows) in z-VAD-FMK–treated embryos. Six to 10 embryos were examined for each group. hf, head fold. k: Ventral views of control stage 12 embryos exhibiting correct right-sided (R) heart looping (asterisk). l–o: z-VAD-FMK–treated embryos displaying a reversed, left-sided (L) looping (l,m), no or ambiguous looping (n), and right-sided heart looping (o). p: Frequency of heart looping after treatment with Caspase antagonists. n, the number of embryos; amb, ambiguous.

Download figure to PowerPoint

After this subtle change in expression of the laterality genes during early gastrulation stages, embryos that lacked midline cell death exhibited an altered pattern of heart looping at neurula stages (Fig. 4k–p). Sham-treated embryos predominantly underwent normal, right-sided looping of the heart tube as seen in untreated embryos. In contrast, a large population of z-VAD-FMK–treated embryos (Fig. 4l–p) exhibited either a left-sided looping (43.6%) or ambiguous looping (15.4%) of the heart, whereas 41% of treated embryos underwent right-sided looping. Except for this apparent randomization of heart looping, no other morphogenetic differences were evident in these early embryos. It cannot be ruled out that the heart looping defect results from a nonspecific side effect of the Caspase family inhibitor. Importantly, however, the correct, right-sided heart looping (Fig. 4p) was preserved in a group treated with the Caspase-5 inhibitor, z-WEHD-FMK, which failed to inhibit midline cell death (Fig. 2l). Thus the alteration in the direction of heart looping was induced only when cell death along the midline was lost. The above results suggest that midline cell death is necessary for proper expression of laterality genes and normal heart looping.

The importance of the midline in human development is highlighted in the numerous types of laterality disorders (Abernethy, 1793, Brown and Wolpert, 1990; Aylsworth, 2001). The conditions of heterotaxia, situs inversus, and isomerism all show defects in positioning of the organs around the midline axis of the body. Signaling networks that are localized asymmetrically are the source for patterning the left–right axis and are responsible for laterality and positioning (Capdevila et al., 2000). The model relies on the presence of a functional midline that separates left signals from right signals (Levin and Nascone, 1997; Levin and Mercola, 1999). In nonamniote vertebrates, the future midline of the organism is specified during the first cleavage that divides a fertilized egg into the left and right halves along the plane determined by a sperm entry site (Gerhart et al., 1981; Yost, 1998). In amniotes, however, the embryonic midline is a hypothetical line until Hensen's node and the primitive groove form (Brown and Wolpert, 1990; Levin and Nascone, 1997; Piotrowska and Zernicka-Goetz, 2001).

Our present study has defined a previously unidentified functional midline as a molecularly specified cell population that is fated to die and maintains its position in the primitive streak. It is currently unknown how this unique cell population remains in the midline, escaping from phagocytic clearance that is the ordinary fate of dead cells. A “midline barrier model” has been proposed, where the midline blocks the spread of signals from the left to the right sides of the embryo (Levin and Nascone, 1997; Levin and Mercola, 1999). The persistent cell death in the streak midline may be a cellular mechanism that builds this physical barrier, dividing the embryonic disc into the left and right sides during gastrulation, even before Hensen's node formation. Randomized heart looping in embryos that have lost midline cell death suggests a critical role for this newly identified midline cell population in regulating left–right asymmetry at the earliest stages of embryonic development.

Experimental Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental Procedures
  6. Acknowledgements
  7. REFERENCES

Detection of Markers for Gastrulation, Laterality, and Cell Death

Fertilized chicken eggs were incubated at 38°C. Whole embryos were isolated, staged (Hamburger and Hamilton, 1951), and processed for antibody staining or RNA in situ hybridization as described (Henrique et al., 1995) with three times 30-min washes in RIPA buffer in the place of proteinase K treatment. Antibody staining for the conjugated form of ubiquitin was performed according to Wunsch and Haas (1995). The activated form of Caspase-3 was detected by using anti-active Caspase 3 antibody (1:150 dilution, BD Pharmingen). Embryos were fixed for 20 min in 4% paraformaldehyde and then washed in PBS three times. The embryos were incubated in PBS containing 1% BSA and 0.1% Triton-X for 30 min and then incubated with the anti-Caspase 3 antibody in PBS containing a 1:200 dilution of goat serum overnight at 40°C. Embryos were then washed three times 10 min in PBS and then incubated with a 1:150 dilution of the Alexa 488 anti-goat antibody (Molecular Probes) for 30 min at room temperature. Embryos were then washed three times and visualized under fluorescence. Signals from three embryos were superimposed, as described (Wei and Mikawa, 2000a). Propidium iodide staining for cell death detection was performed according to the manufacturer's instructions by using a 10-mM concentration dissolved in 1× Tyrode's solution (Molecular Probes). Staining of the adherence belt with fluorescein-conjugated phalloidin was carried out as described (Wei and Mikawa, 2000a).

In Ovo Injection of lacZ Retrovirus

A replication-defective spleen necrosis virus encoding lacZ (Wei and Mikawa, 2000a, b) was harvested and concentrated to titers of ∼107–108 virions/ml just before injection. A small window of ∼20 mm diameter was opened in the shell at the blunt end of the incubated egg and the underlying shell membrane removed. Embryos were illuminated by cool light through a blue filter and monitored through a color video camera connected to a trinocular stereomicroscope. Viral particles were microinjected into the developing primitive streak and embryos returned to incubator to develop further, as described (Wei and Mikawa, 2000a, b). Infected cells and their progeny were detected with X-gal staining for β-galactosidase, as described (Wei and Mikawa, 2000a).

Inhibition of Cell Death

Caspase inhibitors, z-VAD-FMK and z-WEHD-FMK (BioVision, Inc.), of 2–10 mM in DMSO were diluted to desired concentration with a buffer containing DMEM and 7% FCS just before injection. By using a small needle attached to a Hamilton syringe with tubing, 1–4 μl of inhibitor or buffer containing 0.1–0.5% DMSO was injected in ovo into the perivitelline space above the blastodisc at 4 hr and 10 hr of incubation. Injected eggs were returned to incubator until reaching desired developmental stages, then removed and tested with propidium iodide for the presence of cell death in the midline. After recording propidium iodide signals, embryos were fixed and processed for subsequent analyses.

Imaging

All images were captured by a digital photo-camera (SONY DKC-5000), adjusted for brightness and contrast, and then cropped by using Adobe Photoshop 6.0 (Adobe Systems, Inc.).

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental Procedures
  6. Acknowledgements
  7. REFERENCES

We thank M. Mercola, G. Schoenwolf, H.J. Yost, D. Pennisi, D. Reese, and V. Ballard for their invaluable comments on the manuscripts; L. Cohen-Gould, L. Miroff, and A. Dizon for technical assistance; H. Ohuchi, K. Yutzey, R. Rannyan, and C. Tabin for marker genes; A. L. Haas for antibodies to the conjugated form of ubiquitin.

REFERENCES

  1. Top of page
  2. Abstract
  3. Introduction
  4. Results and Discussion
  5. Experimental Procedures
  6. Acknowledgements
  7. REFERENCES