Expression of the Celsr/flamingo homologue, c-fmi1, in the early avian embryo indicates a conserved role in neural tube closure and additional roles in asymmetry and somitogenesis


  • Caroline J. Formstone,

    Corresponding author
    1. MRC Centre for Developmental Neurobiology, New Hunts House, Kings College London, London, United Kingdom
    • MRC Centre for Developmental Neurobiology, New Hunts House, Kings College London, London SE1 1UL, UK
    Search for more papers by this author
  • Ivor Mason

    Corresponding author
    1. MRC Centre for Developmental Neurobiology, New Hunts House, Kings College London, London, United Kingdom
    • MRC Centre for Developmental Neurobiology, New Hunts House, Kings College London, London SE1 1UL, UK
    Search for more papers by this author


Flamingo is one of a core group of proteins that regulate planar cell polarity of epithelial structures within the Drosophila embryo while their vertebrate counterparts have been implicated in the coordination of convergent extension movements during gastrulation and in neural tube closure, suggesting that planar polarity mechanisms also function in these processes. Failure of neural tube closure is one of the most common human birth defects, and a murine flamingo (fmi) homologue, Celsr1/fmi-1, was identified as the defective gene in two mouse mutants exhibiting failure of closure 1 of the neural tube. This failure resulted in craniorachischisis in which the neural tube is open from the midbrain posteriorly. The avian embryo provides a tractable system to study neural tube closure. We have identified a chick Celsr1/fmi-1 orthologue, c-fmi1 and provide the first study of expression of an avian flamingo gene. We show that expression is highly dynamic in the early embryo and that c-fmi1 transcripts become enriched within the avian neural epithelium at the initiation of neural tube closure, suggesting a conserved function for Flamingo proteins in this process. Our data also suggest a role for c-fmi1 in myotome development. Developmental Dynamics 232:408–413, 2005. © 2004 Wiley-Liss, Inc.


The seven-pass transmembrane protocadherin Flamingo (also known as Starry Night; Stan) regulates several developmental processes during Drosophila embryogenesis, including orientation of wing prehairs within the plane of the epithelium (planar cell polarity, PCP; Usui et al., 1999; Chae et al., 1999) and formation of dendritic fields within the peripheral nervous system (Gao et al., 2000).

Recently, it has been shown that two homozygous mouse mutants carrying single amino acid substitutions within the coding region of a murine Flamingo homologue, Celsr1, exhibit severe neural tube defects as a result of failure to initiate neural tube closure (Curtin et al., 2003). Neural tube closure in the mouse begins at the hindbrain/cervical boundary around embryonic day 8.5 (closure 1) and spreads rostrally and caudally from this site. Failure of closure 1 results in a condition known as craniorachischisis, for which almost the entire neural tube from the midbrain to the lower spine remains open (for a recent review, see Copp et al., 2003). Consistent with an involvement in closure 1, Celsr1 is intensely expressed within the neural plate before and during neurulation (Hadjantonakis et al., 1998). Defective genes in several mouse mutants exhibiting craniorachischisis have been identified recently (for recent review, see Copp et al., 2003). It is noteworthy that, like Flamingo, their Drosophila homologues regulate cellular polarity within the plane of the wing and ommatidial epithelia.

PCP has also been implicated in the control of convergent extension movements within gastrulating Xenopus and zebrafish embryos during which cells accumulate at the dorsal side of the gastrula and concomitantly extend its anterior–posterior (A-P) axis (for recent review, see Heisenberg and Tada, 2002).

Since identifying three Flamingo homologues in mouse (Celsr1, Celsr2, and Celsr3; Hadjantonakis et al., 1998, Formstone and Little, 2001), we subsequently have sought flamingo-related genes in other vertebrate species to exploit each, as appropriate, as a model system. Here, we describe the pattern of expression of the Celsr1 orthologue in chick, c-flamingo1 (c-fmi1). We find that expression of c-fmi1 within the neural epithelium is consistent with, and suggestive of, a function in initial closure of the neural tube in the chicken. At later stages, c-fmi1 transcripts become increasingly polarized along the apical–basal axis of the neuroepithelium, suggestive of subsequent secondary functions. Notably, asymmetric distribution of c-fmi1 to the apical border of the closed neural tube is mirrored in other embryonic tubular structures such as the notochord and the pronephros. In addition, we identify a potential role for vertebrate Flamingo proteins in somitogenesis, a function not yet ascribed to any mammalian flamingo-related gene.


To identify flamingo-related genes in chick, we screened a stage 10 (HH10; Hamburger and Hamilton, 1951) chick library (a gift of D. Wilkinson, NIMR, UK) at reduced stringency with partial cDNAs comprising the seven-transmembrane and cytoplasmic domains of all three murine flamingo genes (Formstone and Little, 2001) and identified several overlapping cDNA clones. BLAST analysis revealed each of these cDNAs to be most closely related to human and mouse Celsr1/flamingo-1 (Fig. 1). Sequence has been deposited with GenBank (accession no. AY426608), denoted c-flamingo-1 (c-fmi1), and comprises a cDNA encompassing the HormR domain through to the 3′ untranslated region.

Figure 1.

Sequence alignment of c-fmi1 and Celsr1, Celsr2, and Celsr3. Predicted amino acid sequence of cysteine-rich and seven-pass transmembrane domains (hydrophobic membrane domains highlighted in red) from c-Fmi-1 and the murine flamingo-related proteins CELSR1, CELSR2, and CELSR3. Sequence comparison reveals that c-Fmi1 is the orthologue of CELSR1.

c-fmi1 Is Expressed Within the Gastrulating Chick Embryo

During late streak stages (HH4), c-fmi-1 expression was polarized along the A-P axis of the embryo with expression restricted to the node and emergent head process (arrowhead in Fig. 2A) and caudal-most regions of the primitive streak (arrow in Fig. 2A). By HH5, c-fmi-1 transcripts were enriched within the vicinity of the regressing node (Fig. 2B) and detected more faintly within the primitive streak. At HH6, c-fmi-1 expression was present within the neural plate (Fig. 2C) but absent more caudally.

Figure 2.

c-fmi1 expression is dynamic within the early avian embryo and correlates with initial closure of the avian neural tube. A–C, E–H, and M–O are dorsal views of whole-mount embryos. Embryonic stages according to Hamburger Hamilton (HH) are indicated. All sections are transverse 50-μm Vibratome sections taken from embryo C (section D), embryo H (sections I, K), and embryo M (section L). The level at which each section was taken is indicated on panels C, H, and M. A: c-fmi1 is expressed at the rostral-most tip of the embryo within the node (arrowhead) and also within the most caudal regions of the primitive streak (arrow) at HH4. B: c-fmi1 transcripts are enriched within the vicinity of the regressing node (arrow) at HH5. C: At HH6, c-fmi1 expression appears in the neural plate and is highly enriched within the developing mesencephalon. There is an absence of c-fmi1 transcripts within more caudal regions of the embryo. D: c-fmi1 transcripts are highly enriched within the lateral neural plate but absent from the midline floor plate (arrow). The underlying notochord is positive for c-fmi1 expression (arrowhead). E: c-fmi1 expression is enriched within a medial stripe within the apposing walls of the neural furrow (arrow) but remains absent from the midline floor plate. There is lower expression of c-fmi1 elsewhere in the neural plate. F: By HH7+, c-fmi1 expression is transiently up-regulated rostrally as the anterior neuropore forms. More caudally and marking neural furrow formation, medial stripes of elevated c-fmi1 expression are observed (arrow) flanking the floor plate, just rostral to the somites, the boundaries of which are also marked by c-fmi1 expression (arrowhead). G: By HH8−, the medial stripes of c-fmi1 transcripts (arrow) within the neural walls shift caudally with the neural furrow in the wake of the regressing node. c-fmi1 expression has reduced significantly within the presumptive forebrain (arrowhead). H: Expression of c-fmi1 has been lost from the presumptive forebrain (arrowhead) at HH8, allowing a clear view of c-fmi1 expression in the underlying prechordal plate (black arrow) and in the notochord lying caudal to it. c-fmi1 expression is highly enriched in the walls of the neural plate before closure. Caudally, unilateral expression of c-fmi1 is apparent within the lateral plate mesoderm, where it is restricted to the right hand side (white arrow). I: c-fmi1 expression is elevated within the closing lateral walls of the mesencephalon and is expressed within the notochord (arrow). J: The walls of the closing neural plate strongly express c-fmi1. K: Reduced levels of c-fmi1 are apparent in caudal regions of the neural plate and within associated midline tissue. c-fmi1 transcripts are restricted to the right-hand lateral plate mesoderm (arrow). L: By HH8+, as the neural folds converge toward the midline within the future brain, c-fmi1 expression is lost from the dorsal-most regions of the neural epithelium. M: A sharp transient boundary of c-fmi1 expression is observed within the presumptive forebrain (arrow), and c-fmi1 transcripts mark the medial aspect of somites 2–5 (and inset showing a magnified dorsal view of somites 3–6. N: At HH9, c-fmi1expression is appreciably down-regulated within the developing brain but is still elevated within the caudal neural tube and expression within the lateral plate mesoderm has become bilateral (arrows). O: At HH10, the level of c-fmi1 transcripts is elevated within the presumptive posterior diencephalon (arrow). Expression of c-fmi1 remains restricted to the medial aspect of the developing somites.

c-fmi1 Is Expressed Within the Neuroepithelium in a Manner Consistent With a Role in the Initial Closure of the Chick Neural Tube

Within the emerging neuroepithelium, before neural tube closure, c-fmi-1 mRNA was enriched within a sharply defined area of the head, the presumptive mesencephalon (Fig. 2C). The presumptive mesencephalon is the site at which neural tube closure initiates within the chick embryo (Colas and Schoenwolf, 2001). As the neural plate is sculpted, narrowing mediolaterally but elongating rostrocaudally in the wake of the regressing node, mesencephalic neural tissue flanking the floor plate begins to appose at an increasingly greater angle and a neural furrow forms. Generally, it is accepted that the force required for initial furrowing derives from the median hinge point (MHP), created by cell wedging within the midline floor plate. Within the mesencephalon, c-fmi1 was strongly expressed within the walls of the neural plate (Fig. 2C,D) with reduced levels of c-fmi1 expression detected rostral and caudal to this region. c-fmi1 transcripts were absent from the midline floor plate (Fig. 2C–E), suggesting that c-fmi1 does not directly affect cell behavior within the MHP. Underlying the floor plate, the notochord was c-fmi1-positive (arrow in Fig. 2D) and expression of c-fmi1 within the notochord persisted through later stages of development (Figs. 2H,I, 3A,C,E,G). Instead, c-fmi1 transcripts were enriched within two medial stripes flanking the c-fmi1-negative floor plate (Fig. 2E). By HH7+, the stripes of enhanced c-fmi1 expression were restricted to a small region just anterior to the first somites (arrow in Fig. 2F), whereas more rostrally within the neural plate, where the neural walls had elevated further, strong c-fmi1 expression was more uniform across the neural epithelium. By HH8 (Fig. 2H–J), c-fmi1 transcripts were highly enriched within the walls of the closing neural tube but, as the neural folds converged toward the dorsal midline before their fusion, c-fmi1 transcripts were lost from the dorsal-most regions of the neural epithelium within the neural folds (arrow in Fig. 2L).

Figure 3.

c-fmi1 transcripts localize asymmetrically within tubular structures of the avian embryo and are associated with myotome formation. A,C,E,G: Transverse 50-μm Vibratome sections showing different anteroposterior (A-P) levels from the same Hamburger and Hamilton stage (HH) 18 embryo. B,D,F: Dorsal views of the same HH14 embryo at different A-P levels. Embryonic stages are indicated on each panel. A: Section taken at the level of the most recently cleaved somite of an HH18 embryo. Expression of c-fmi1 within the closed neural tube is enriched within the apical domain most notably within the roof plate region (arrow-a). The floor plate, however, has homogeneous c-fmi1 expression. c-fmi1 transcripts are also localized to the apical surface of the notochord (arrowhead) and the pronephros (arrow-b). B: Somites 17–20 of an HH14 embryo. c-fmi1 transcripts are observed along the entire rostrocaudal length of the medial aspect of each epithelial somite and appear to be localized apically within these cells (arrow). C: Section at somitic level 27–30 of an HH18 embryo. c-fmi1 expression remains distributed apically within the neural tube (excluding the floor plate), notochord (arrowhead) and pronephros (arrow-a). Within the epithelial somite, c-fmi1 transcripts also appear to be restricted to the apical surface of c-fmi1-expressing cells located within the medial part of the somite (arrow-b). D: Somitic level 9–14 of an HH14 embryo where somites are undergoing decondensing. Expression of c-fmi1 is tightly associated with, and extends along the entire rostrocaudal length of, the medial aspect of each somite (arrow). E: Somitic level 19–22 of an HH18 embryo. c-fmi1 expression remains apically restricted within the neural tube (excluding the floor plate), notochord (arrowhead) and pronephros (arrow-a). A tight knot of c-fmi1 expression exists within the dorsomedial region of the dissociating somite (arrow-b). F: Somitic level 3–7 of an HH14 embryo. Arrows indicate expansion of c-fmi1 expression in a ventrolateral direction within the maturing somites, initiating at the rostral edge (shown in the more posterior somite indicated by arrow), thus giving the impression of a transient triangle-like shape similar to that observed with myoD (Kalcheim et al., 1999). c-fmi1 expression subsequently becomes homogeneous within the dorsal part of the somite (shown in more anterior somite indicated by arrow). G: Somitic level 11–15 of an HH18 embryo. c-fmi1 expression remains apically restricted within the neural tube (excluding the floor plate) and notochord (arrowhead). c-fmi1 is also expressed within the myotome of the maturing somite as it extends ventrolaterally (arrow). nt, neural tube.

Concomitant with caudal extension of neural closure is apposition of the neural walls rostrally to form the anterior neuropore. c-fmi1 expression was initially elevated at the rostral tip of the prospective forebrain from HH6 to 7+ (Fig. 2C,E,F), but by HH8−, before closure of the anterior neuropore (Schoenwolf, 1979), c-fmi1 expression had declined in this region (arrowhead in Fig. 2G) and was barely detectable by HH8 (Fig. 2H). However, transcripts were detected in the underlying prechordal plate (Fig. 2H). By HH8+, a sharp transient boundary of c-fmi1 expression had formed within the developing forebrain (Fig. 2M) possibly marking the boundary between telencephalon and diencephalon. As development proceeded, c-fmi1 expression remained strong throughout the developing spinal cord (Fig. 2M,N) but was reduced at future brain levels after neural closure. By HH10, however, c-fmi1 expression was elevated within the presumptive posterior diencephalon (Fig. 2O).

c-fmi1 Transcripts Exhibit Tissue Specific Asymmetric Localization Within Multiple Tissues

Subsequent to the early localization of elevated c-fmi1 expression within medial stripes flanking the midline (HH7–8; Fig. 2E–G) and after neural closure, c-fmi1 transcripts became concentrated within the apical ventricular domain of the pseudostratified neuroepithelium, including within the roof plate region along the entire rostrocaudal neural axis (Fig. 3A,C,E,G). We saw a similar asymmetric distribution of c-fmi1 transcripts to the apical domain of two other “cord-like” structures, the notochord and the developing pronephric duct (Fig. 3A,C,E,G) at HH18. Finally, analysis of both whole-mount (Fig. 3B) and sectioned embryos (Fig. 3C) suggested that c-fmi1 transcripts were also restricted to the apical surface of c-fmi1-expressing cells within the medial aspect of the epithelial somite.

Asymmetric expression of RNA transcripts also promotes the generation of the left–right (LR) axis (Levin, 1997). After the initial break of bilateral symmetry within the embryo, LR axis positional information is transferred first to the node and subsequently to lateral plate mesoderm where side-specific domains of gene expression are established (Pagan-Westphal and Tabin, 1998). Later, LR axial information is translated into organ-specific asymmetric morphogenesis. We observed a transient expression of c-fmi1 in the posterior right hand lateral plate mesoderm at HH8 (Fig. 2H,K). One stage later, however, c-fmi1 expression appeared bilateral within the lateral plate mesoderm (Fig. 2N) but was lost from this area by HH10 (data not shown).

c-fmi1 Is Expressed During Somitogenesis and Is Associated With the Site of Origin of the First Wave of Muscle Pioneers Which Contribute to the Formation of the Myotome

An unexpected feature of c-fmi1 gene expression was its association with somitogenesis. c-fmi1 was first observed faintly at somite boundaries around HH8- (three-somite stage; Fig. 2F) before strong expression of c-fmi1 in the overlying neural plate and neural tube. Later, c-fmi1 transcripts were apparent at low levels within the medial aspect of each somite, excluding the first formed somite (Fig. 2M and inset, Fig. 3B, C), abutting the neural tube and homogeneously distributed along the entire rostrocaudal length of the medial aspect of each epithelial somite. As development proceeds, epithelial somites dissociate into dermomyotome and sclerotome, and medial expression of c-fmi1 within newly formed somites resolved to a tight dorsomedial zone of expression (Fig. 3E). However, expression of c-fmi1 remained homogeneous along the rostrocaudal length of the somite (Fig. 3D). More rostrally, c-fmi1 expression was expanded ventrolaterally within the maturing somites (Fig. 3F,G). Initial medial expression of c-fmi1 correlates with the site of origin of muscle pioneers (Kalcheim et al., 1999) thought to represent the first wave of muscle progenitor cells that lead to formation of the myotome. Subsequently, during somite dissociation, this epithelial sheet of muscle precursors localize dorsomedially (Kahane et al., 2002) and then translocate laterally down underneath the forming dermomyotome. The dorsolateral expansion of c-fmi1 expression in more mature somites is consistent with the presence of c-fmi1 in these muscle pioneers.


An HH10 chick library was screened by reduced stringency hybridization by using partial cDNAs for all three murine Celsr genes as described previously (Formstone and Little, 2001). In situ hybridization was performed as described previously (Formstone and Little, 2001) using a riboprobe for c-fmi1 (corresponding to nucleotides 1971 to 2836 of AY426608). For sectioning, embryos were embedded in gelatin–albumin (0.45% gelatin, 25% albumin, 20% sucrose) and fixed with 2.5% glutaraldehyde, and 50 microns sections were cut on a Vibratome.


We thank Imelda McGonnell for help with Vibratome sectioning.