ceh-16/engrailed patterns the embryonic epidermis of Caenorhabditis elegans

The conserved homeodomain transcription factor engrailed was first identified in 1926 as a spontaneous recessive mutation in Drosophila(Eker, 1929). engrailed turned out to be a key gene involved in the development of Drosophila appendages and segments, where it specifies the posterior compartment. engrailed has therefore been named a `selector' gene(Garcia-Bellido and Santamaria,1972; Lawrence and Morata,1976; Nusslein-Volhard and Wieschaus, 1980). In vertebrates, En1 is recruited during the development of vertebrate limbs. It is essential for the maintenance of the ventral compartment specification(Loomis et al., 1996) and has a crucial role in patterning the mid-hindbrain boundary(Danielian and McMahon,1996).

One striking fact is the similarity between the involvement of engrailed in the specification of the posterior compartment in the wing imaginal disc of Drosophila and in the specification of the ventral ectoderm in the developing limb in mouse. It has been shown that during embryonic development the anterior ectodermal ridge (AER) of the developing limb remarkably resembles the anterior/posterior (A/P) compartment boundary in the fly. Both the A/P boundary and the AER express decapentaplegic (dpp)/BMP2 homologous proteins of the TGF-β family, which are repressed by en/EN1 in adjacent cells. Additionally, in both cases the expression of dpp/BMP2 is induced by hedgehog (hh)/Shh from neighboring cells. These neighboring cells in Drosophila are the posterior cells, where hhdepends on engrailed. In the mouse limb, Shh is expressed posteriorly adjacent to, and maintained by, the AER, depending indirectly on EN-1 function(reviewed by Hidalgo, 1998). In addition to the regulation of organizers, engrailed is required to preserve compartment boundaries in Drosophila. When the selector gene engrailed is removed, in vivo, from a posterior clone of cells in the wing, those cells gain anterior affinity. They sort out from posterior cells and, if in contact with anterior cells, sort into and mix with them(Lawrence and Struhl, 1982; Morata and Kerridge, 1982). This segregation mechanism might be controlled in part by regulation of cell adhesion molecules (Dahmann and Basler,1999). It has been suggested that the ancestral function of engrailed may be neuronal targeting, because it regulates the connectivity through the transcriptional regulation of cell adhesion molecules in the central nervous system in arthropods and vertebrates(Gibert, 2002; Vincent, 1998). Moreover, engrailed has been proposed to play a general role in segmentation of protostomes (Prud'homme et al.,2003).

In Caenorhabditis elegans, the functions of engrailedhave not yet been described. Instead, the function of GATA and other homeobox-containing transcription factors have been studied in the patterning of the epidermis in roundworms. Briefly, the GATA factor ELT-1 specifies general epidermal identity (Page et al.,1997). The epidermis is subsequently patterned in three morphologically distinguishable major areas during embryogenesis: (1) dorsal cells that fuse to form the syncytia hyp6 and hyp7 during embryonic elongation(Podbilewicz and White, 1994),(see Movie 1 in the supplementary material); (2) two single left and right rows of lateral seam cells; and (3) the ventral P cells whose descendants either fuse postembryonically to hyp6 and hyp7 or develop vulval structures and the ventral nerve cord (Podbilewicz and White, 1994; Sulston et al., 1983). LIN-39/HoxD4/Dfd and CEH-20/Exd play a crucial role in repressing the cell fusion of some posterior descendants of the P cells(Clark et al., 1993; Maloof and Kenyon, 1998; Shemer and Podbilewicz, 2002; Wang et al., 1993). In addition, the operon encoding the two GATA factors ELT-5(=EGL-18) and ELT-6 is important for differentiation/fusion-repression in the lateral seam cells(Koh and Rothman, 2001) and for cell fusion-repression in the vulval precursor cells (VPCs), where it is controlled by LIN-39/HoxD4/Dfd (Koh et al., 2002). A general effector for cell fusion in the epidermis of the worm is the transmembrane protein EFF-1(Mohler et al., 2002). Moreover, EFF-1 is both necessary and sufficient for epithelial and myoepithelial cell fusion in C. elegans(Shemer et al., 2004). It has also been shown that LIN-39 represses the expression of eff-1 in the VPCs (Mohler et al., 2002; Shemer and Podbilewicz,2002).

In this study we show how ceh-16/engrailed controls the differentiation of the seam cells, thereby patterning the embryonic epidermis of C. elegans. ceh-16/engrailed represses the fusion of the seam cells with the neighboring epidermal cells by repressing the expression of the fusion effector eff-1. ceh-16/engrailed also triggers the expression of elt-5 and other seam cell markers and is indispensable for alae formation (a hallmark of seam cell differentiation). We also show that in the ceh-16/engrailed mutant the seam cells lose their lateral position and migrate either dorsally or ventrally, intermingling with these cells. Therefore, seam cells in the wild-type context seem to act by preventing cell migration and maintaining embryonic compartment.

The ceh-16 cDNA was cloned by RT-PCR using SMART™ RACE cDNA Amplification Kit protocol (Clontech). As the 5′ UTR contains stop codons in all three reading frames, the methionine in the second exon constitutes the bona fide start methionine. GenBank sequence identifier:AY647457.

ceh-16(lg16)III, ceh-16(lg17)III were obtained by screening ethylmethanesulfonate (EMS) mutagenized worm libraries via PCR according to Anderson (Anderson,1995); breakpoints of the deletions were sequenced twice independently using standard procedures. The deletion in ceh-16(lg16)spans 2218 bp and encompasses 560 bp of the promoter region, the transcriptional start and the first four of the five exons. Exon 5 encodes the C-terminal 51 amino acids. By molecular means this allele is predicted to be a null allele. In ceh-16(lg17), the deletion spans 867 bp encompassing 471 bp of the promoter region, the transcriptional start and most of the second exon including the start methionine and the first 56 amino acids. Since the transcription start in ceh-16(lg17) is severely compromised and the phenotype is identical in both ceh-16(lg16)(Table 1) and ceh-16(lg17), both alleles probably constitute a null allele. Breakpoint for ceh-16(lg16): GATCGAAAAAGTAGTG/CAGTTGTTTTGGCATGA. Breakpoint for ceh-16(lg17): TAATTCCCATGTTATATT/GCACAAGATATTCCGATC. The sequences of the primers used for screening are available upon request. Both mutants were out-crossed ten times prior to analysis.

Nematodes were maintained as previously described(Wood, 1988). Strains were kept at 20°C unless otherwise noted. The wild-type strain N2 was used unless otherwise stated. Strains with the following mutant alleles,chromosomal aberrations or transgenic arrays were used in this work: eff-1(hy21)II (Mohler et al.,2002), jcIs1 IV (integrated ajm-1::gfp strain), ced-1(e1735)I; unc-119(ed4) III; wIs78 [contains pDP#MM016B (unc-119+), pJS191 (ajm-1::GFP); pMF1 (SCM::GFP— nuclear seam cell marker), and cosmid F58E10], dpy-20(e2017)IV, wIs66 [elt-5::gfp containing pKK7(Koh and Rothman, 2001)], Ex[eff-1p::gfp] (Mohler et al.,2002).

Transgenic strains were obtained using standard procedures(Mello et al., 1991), adapted as in Cassata et al. (Cassata et al.,2000). ceh-16::gfp translational fusions were injected at a concentration of 30 μg/μl along with 50 μg/μl pRF4(rol-6 dm) plasmid. Roller lines were crossed into heterozygous ceh-16(lg16) and tested for their rescue ability by selecting homozygous transgenics (genotypization by PCR). Thereafter, the rescued strain was crossed into jcIs1 for microscopical analyses. Transgenic Ex[wrt-5::gfp] and Ex [wrt-2::gfp] strains were a generous gift from T. Burglin (Aspock et al.,1999). The wrt-5::gfp and wrt-2::gfpextrachromosomal arrays were integrated as follows: 50 transgenic L4 were irradiated with UV using a Stratalinker (Model 1800) from Stratagene at 30,000μJ/cm². After irradiation the animals were singled. After starvation the plates were chunked to let the worms crawl out of the agar; 250 were singled and analyzed for their ability to produce 100% transgenics in the offspring. One integrated line of each transgenic was isolated in this way. Both were out-crossed twice. Transgenic lines of nhr-73::gfp and nhr-74::gfp were obtained as described in Miyabayashi et al.(Miyabayashi et al.,1999).

Candidate young adult heterozygous ceh-16(lg16) or ceh-16(lg17) P0 animals were singled and grown on agar plates for 4-5 hours (this allowed each worm to lay 15-20 eggs). Thereafter they were removed and the genotype was determined via PCR. The sum of all the eggs (of the positive plates) was counted. The next day, the number of dead eggs was determined. A similar procedure was adopted with [ceh-16(lg16)/+]X[ceh-16(lg17)/+] crosses: males derived from a ceh-16(lg17) cross with N2 males were crossed with ceh-16(lg16) hermaphrodites. These hermaphrodites were used as P0 for the lethality test. The presence of both alleles in the offspring was tested via PCR. Dead eggs were counted as above.

The full-length ceh-16 cDNA was cloned into pBluescript II SK-using HindIII and BamHI sites. In vitro transcribed ssRNA from linearized vectors was produced using commercially available T3 and T7 RNA polymerase systems (Promega). Annealed dsRNA was injected into young adult hermaphrodites at a concentration of 1 μg/μl. The offspring was analyzed(by microscopy or lethality test as described above). elt-5(RNAi)experiments were performed similarly using the F55A8.1 RNAi clone from the MRC C. elegans RNAi bank.

The PCR amplified full-length ceh-16 cDNA was inserted in frame into the KpnI site of the heatshock promoter (hsp16-2)contained in construct pPD49.78 (Stringham et al., 1992) and sequenced.

To perform heatshock experiments, two independent lines of transgenic worms were constructed by injecting 5 μg/μl of heatshock promoter (with or without ceh-16 for control purposes) along with pMH86(dpy-20+) (Han and Sternberg,1991) into a dpy-20(e2017)IV; wIs66[elt-5::gfp] strain. Plates containing transgenic animals were heatshocked three times for 1 hour at 30°C (recover period between heatshocks: 1 hour at 15°C). Transgenic offspring L1 was analyzed the next day for ectopic expression of ELT-5::GFP.

Antibody staining was performed according to Waddle et al.(Waddle et al., 1994). Light microscopy was performed using a Zeiss Axioplan2 imaging microscope, Zeiss Axiocam HRc camera and Axiovision software. Confocal microscopy was performed as in Shemer and Podbilewicz (Shemer and Podbilewicz, 2002), and confocal time-lapse movies were recorded taking Z series projections each 5-10 minutes for 1.5 hours at 20°C(Rabin and Podbilewicz,2000).

The structure and some functions of engrailed are conserved from arthropods to mammals. To study for the first time engrailed in an unsegmented animal, the nematode C. elegans, we first cloned the cDNA by RT-PCR (see Materials and methods). The encoded protein sequence contains all the critical domains of an engrailed-like factor, unlike any other protein encoded in the worm genome(Fig. 1B; and see Fig. 1 in the supplementary material).

The expression of ceh-16 in C. elegans was analyzed using transgenic animals bearing full-length translational gfp fusions. This fusion construct rescued the mutant (see below), suggesting that all required promoter elements are represented. Expression is most robust throughout embryonic development (Fig. 1C,D) from 250 minutes after the first cleavage until the early 3-fold stage. Expression was observed in the nuclei of hyp5, H0-H2, V1-V6 and T. These cells constitute a bilateral row of cells called seam cells(Fig. 1G,H). Additional ceh-16::gfp expression was also observed in cells of the AB lineage at stages prior to the one described above (28-56 cell stage). These cells were determined in 100 minute embryos as being ABprap, ABarpp, ABpraa, ABplap,ABplaa, ABarpa and ABaraa (not shown). In later embryonic stages we observed expression in anterior neurons (not determined, see Fig. 2 in the supplementary material) and in the DA1 and DD1 motoneurons after hatching.

To further investigate the epidermal expression, an in situ staining with a monoclonal antibody (4D9) that recognizes engrailed proteins in many species was performed (Patel et al.,1989). Localized immunoreactivity in the seam cells was detected during the same stages of embryonic development(Fig. 1E,F). This additionally confirms the expression of ceh-16 in the nuclei of these cells.

In summary, ceh-16 encodes an engrailed homolog in C. elegans. ceh-16 is expressed in the AB lineage during early embryogenesis, in the lateral seam cells and neurons during morphogenesis, and in two motoneurons during postembryonic stages.

To study the functions of ceh-16 in vivo, two deletion alleles[ceh-16(lg16) and ceh-16(lg17)] were isolated from an EMS deletion library. As both mutations delete the transcriptional start and a large part of the coding region, they are predicted to be null alleles(Fig. 1A; see Materials and methods for details). Both mutant alleles are recessive embryonic lethal, do not complement each other and show a very similar phenotype to worms subjected to ceh-16 RNA interference (RNAi)(Table 1; and see Movie 3 in the supplementary material). As the penetrance of the phenotype of both deletion alleles is identical, we used ceh-16(lg16) and ceh-16(RNAi) for further experiments. The full-length translational gfp construct used for expression studies(Fig. 1A) was sufficient to fully rescue the mutant phenotype (see below), whereas a shorter translational gfp construct (Fig. 1A) that contained only the transcriptional start and the first exon did not (data not shown), confirming the specificity of the phenotype.

Microscopic analysis of the ceh-16(—) embryos revealed that the epidermal cells were disorganized, causing severe morphological defects and lack of elongation (Fig. 2C,D). To examine what led to this terminal phenotype the rescuing transgene was used as a marker for ceh-16(+) cells in a ceh-16 mutant background. As, in transgenic C. elegans,extrachromosomal arrays are frequently lost, embryos that expressed the array in a subset of the seam cells were analyzed (mosaic analysis). We found that,in mosaic animals, seam cells lacking ceh-16 in their nuclei showed a dorsal and/or ventral displacement with no obvious directional preference. In addition to the positioning defects, the ceh-16(—) cells fused to the dorsal or ventral epidermis (Fig. 2G,H; arrows). In some mosaic animals (8/63), mutant seam cells projected ventrally in such a way that may have destabilized ventral closure(George et al., 1998),implying that embryonic lethality may be a result of leakage of internal cells. In Fig. 2I,J we show an example of such an embryo with free undetermined cells near the ventral closure that may have leaked out of the embryo at the ventral side. In all the mosaic animals, the seam cells expressing CEH-16::GFP [ceh-16(+)]were rescued and those lacking CEH-16::GFP [ceh-16(—)] showed the phenotype described above (displacement and/or ectopic fusion; Fig. 2E,F). Partial rescued animals manage to hatch showing an attenuated phenotype(Fig. 2K,L). As AB precursors that give rise to ventral and dorsal epidermis express ceh-16 at earlier stages, we cannot rule out expression of ceh-16 below detection level in dorsal and ventral epidermis, but we did not detect any defects in the hypodermis of these areas (e.g. the vulvae were perfectly formed in mosaic semi-rescued animals; not shown). We conclude from this analysis that ceh-16 is required for embryonic seam cell development.

As seam cells that were ceh-16(—) and were not rescued in mosaics abnormally fuse to the syncytial hypodermis during embryonic elongation, we hypothesized that ceh-16 normally inhibits cell fusion and that this inhibition could be executed through the repression of eff-1, a gene that is probably necessary for all cell fusion events in C. elegans (Mohler et al.,2002; Shemer and Podbilewicz,2002). In order to test this hypothesis, we constructed animals that were mosaic for ceh-16 and homozygous for eff-1(hy21). We found that in these animals the cell fusion phenotype of ceh-16(—) cells was suppressed(Fig. 3A,B; Table 2). Moreover, in double mutants completely lacking the ceh-16 rescuing fragment, cell fusion still did not take place as a consequence of eff-1 absence(Table 2). In addition, eff-1p::gfp was ectopically expressed in ceh-16(—)seam cells, suggesting that ceh-16 may be a transcriptional repressor of eff-1 that is active in the seam cells(Fig. 3C,D; Table 2; see Movie 3 in the supplementary material). When eff-1 was de-repressed by ceh-16(RNAi) in a wild-type background, the seam cells that ectopically express eff-1 started to fuse (n=7),(Fig. 3E-G). Therefore, these experiments show that eff-1 is epistatic to ceh-16. eff-1may be a target of ceh-16, and ceh-16 acts in the seam cells to repress eff-1 expression. ceh-16 activity is therefore required to block seam cell fusion with the dorsal and ventral epidermal cells in embryos.

To study whether ceh-16 is controlling for seam cell fate differentiation, various integrated and non-integrated gfp strains expressing early or late seam cell markers were subjected to ceh-16(RNAi) (Fire et al.,1998) and their progeny was analyzed(Fig. 4; see Materials and methods). The early markers tested were elt-5 (=egl-18), nhr-73 and nhr-74 (Koh and Rothman,2001; Miyabayashi et al.,1999). The late markers tested were scm-1, wrt-2 and wrt-5 (not shown) (Aspock et al.,1999; Koh and Rothman,2001). We found that the progeny of transgenic animals subjected to ceh-16(RNAi) showed the typical epidermal defects described in Fig. 2, with the same penetrance for ceh-16(RNAi) (Table 1). These defects were accompanied in all cases by diminished or abolished seam cell marker expression, including that of the GATA factor ELT-5(=EGL-18) (integrated strain, n>20; Fig. 4A-B,G-H), known to be necessary for seam cell specification (Koh and Rothman, 2001), and the nuclear hormone receptors NHR-73(extrachromosomal array, downregulation in 8/8 early embryos [control 2/12]; Fig. 4C-D,I-J) and NHR-74(extrachromosomal array, downregulation in 9/10 early embryos [control 3/16]; Fig. 4C-D,I-J)(Miyabayashi et al., 1999). The fact that ceh-16 regulates nhr-73/74 is interesting,because elt-5 does not (Koh and Rothman, 2001). These results suggest that ceh-16 is a candidate for an early seam cell `determinant' required before elt-5expression. To strengthen this hypothesis, transgenic animals in which ceh-16 was expressed ubiquitously under the control of a heatshock promoter were studied (Stringham et al.,1992). L1 larvae expressing ceh-16 ectopically showed elt-5::gfp expression in the dorsal epidermis, where elt-5is normally never seen [48% (58/122)]. By contrast, control animals transgenic for the heatshock vector alone showed no ectopic gfp staining (0/138)(Fig. 4M-P). Thus, ceh-16/engrailed acts via elt-5. This is further confirmed by the fact that in elt-5(RNAi) animals seam cells fuse(Koh and Rothman, 2001), and this fusion phenotype is also due to de-repression of eff-1(de-repression of eff-1::gfp was observed in 4/20 elt-5(RNAi) embryos; see Fig. 3 in the supplementary material).

Moreover, analysis of semi-rescued mosaic L1 larvae (>50) revealed that in animals that developed to adult stage individual seam cells that had not expressed ceh-16 during embryogenesis failed to differentiate into larval seam cells, indicated by the absence of alae formation [specific cuticular structures secreted by the seam cells in L1 stage(Fig. 5A,B)]. Together, these experiments suggest that ceh-16 is required for early seam cell differentiation/specification.

As stated above, we have observed regulation of early and late seam cell markers, as well as the loss of the alae morphological structures indicative of the seam cell differentiation. One could argue that ceh-16(—) seam cells during embryogenesis ectopically fuse with the hypodermal syncytium, and therefore, as part of the syncytium, lose their ability to express late markers or to form alae. The loss of differentiation might, therefore, be due to indirect effects and not due to the lack of the specific onset of differentiating genes regulated by ceh-16. To exclude this possibility, we tested ceh-16 mutants in which fusion did not take place. This was accomplished in one of two ways: First, ceh-16(—) seam cells failed to secrete alae also when they`occasionally' escaped fusion (Fig. 5C,D). Second, in the fusion-incompetent background of eff-1(hy21), 25/25 animals showed alae-gaps in eff-1(—);ceh-16(—) mosaic L1 larvae (not shown). This experiment strongly suggests that ceh-16 is required for the determination of seam cell fate and not only to repress cell fusion.

We had observed that ceh-16(—) seam cells migrate(Fig. 2). We decided to analyze this phenotype in more detail. Of the mosaic embryos (n=20), 70%displayed abnormal cell positioning phenotype (bean to comma stage), with no directional preference (see Fig. 2E,F; arrows). Do ceh-16(—) cells in mosaic animals leave their position (migration), or is the loss of collinear arrangement due to earlier events, as ceh-16 is expressed earlier(Fig. 1C,E)?

To answer this question, ceh-16 mosaic embryos in an eff-1(—) background were analyzed. In these animals ectopic fusions were not present (Fig. 3B; Table 2). As in a ceh-16(—);eff-1(+) background, ceh-16(—);eff-1(—) cells were not in their normal position. Of 40 embryos tested from bean to 1.5-fold stage of elongation, 33 embryos showed a seam cell-defective phenotype (83%). Out of 62 aberrant seam cells, 48 (77%) had either a minor or strong projection intercalating with the ventral P cells. The minor projection often preceded a more pronounced migration of the entire cell, visible when the embryo was re-analyzed at a later stage (1.8-fold). The remaining 23% of the ceh-16(—) seam cells displayed migration toward the dorsal side(when animals were analyzed at a later stage). Unlike in eff-1(—) and wild-type animals, in eff-1(—);ceh-16(—) double mutants the shape of the seam cells was not wild type and the cell belt margins were no longer straight (see Movie 2 in the supplementary material), but were dented or intercalated(Fig. 6; see Movie 3 in the supplementary material). Moreover, migration can be observed live in a time-lapse experiment, where a ceh-16(—) seam cell, marked by the ectopic expression of eff-1p::gfp, migrated dorsally (see Movie 3 in the supplementary material: in the depicted experiment, not all the seam cells expressed GFP, probably due to mosaic expression of the extrachromosomal array). We conclude from these experiments that ceh-16 is required for the maintenance of correct boundaries between the lateral rows of seam cells and the ventral and dorsal row of epidermal cells during embryonic development.

In this work we have cloned and characterized the cDNA of ceh-16,the only engrailed-like gene in C. elegans. In Drosophila and vertebrates engrailed-like genes have been duplicated during evolution (Gibert,2002), whereas in the nematode Caenorhabditis elegansonly one engrailed-like gene exists (this work). Phenotypic analysis of mutants and mosaic animals revealed that ceh-16/engrailed is required embryonically for the differentiation of the seam cells. The animals lacking ceh-16/engrailed activity died during embryogenesis, most probably due to defective elongation. Moreover, seam cells that failed to express ceh-16/engrailed fused to surrounding ceh-16/engrailed(—) cells in an eff-1-dependent manner. We also showed that eff-1 is epistatic to ceh-16/engrailed and that ceh-16/engrailed acts as a transcriptional repressor for eff-1 in the seam cells. In addition, the function of ceh-16/engrailed is required to maintain the embryonic seam cell fate and the linear lateral position of the seam cells.

CEH-16 shares the archetypal structure of engrailed class proteins from arthropods, annelids, chordates and vertebrates (see Fig. 1 in the supplementary material). Besides the homeodomain referred to as Engrailed Homology domain 4 (EH4), CEH-16 possesses at least EH1, EH2 and EH3 [of the five known EH domains (Gibert,2002)]. EH1 is constituted by the engrailed repressor domain, which binds to Groucho in Drosophila. Groucho can act as a co-repressor of transcription (Jimenez et al., 1997), is conserved in C. elegans and binds to the EH1 domain, which is also present in unc-4(Winnier et al., 1999). Interestingly, we have shown that ceh-16/engrailed represses the transcription of eff-1, but we do not know whether this repression is direct or indirect, nor if it is dependent on Groucho-like genes. EH2 and EH3 are involved in the binding of Hox homeodomain co-factors such as Pbx/Exd in Drosophila. These bind again to other homeodomain co-factors of the HtH/Prep/Meis gene family; these interactions are important for nuclear import(Abu-Shaar et al., 1999; Berthelsen et al., 1999; Kurant et al., 1998; Rieckhof et al., 1997). Exd,HtH and Engrailed can form a functional triple repressor complex in Drosophila (Kobayashi et al.,2003). In C. elegans, like ceh-16/engrailed, the orthologs of Exd (=ceh-20/40) and Hth (=unc-62) are also involved in embryonic epidermal development(Van Auken et al., 2002). As the EH2/3 domains in CEH-16 are present, CEH-20/40 and UNC-62 might also be co-factors of CEH-16/Engrailed in C. elegans.

Cell fusion has been shown to control cell fates. When cell fusion is blocked during postembryonic development in eff-1 mutants, unfused VPCs can also respond to neighboring signals and adopt vulval fates (reviewed by Shemer and Podbilewicz, 2003). The result is an ectopic and non-functional vulva (Mohler et al.,2002).

We have shown that the regulation of epidermal cell fusion is also crucial during embryonic development. Seam cells act as a non-fusing `inter-zone'between the dorsal and the ventral areas of the embryo. Although fusion repression prevents the seam cells from becoming a part of the dorsal syncytium, we have shown by bypassing ectopic fusions in an eff-1mutant background that ceh-16/engrailed is necessary for the differentiation of the seam cells, also in a fusion negative genetic background. In the seam cells that lack ceh-16/engrailed, eff-1 is de-repressed. The de-repression of eff-1 occurs simultaneously with expression of eff-1 in the other epidermal cells (e.g. in the forming hyp7 syncytium). ceh-16/engrailed therefore installs an additional program in a subset of cells otherwise committed to behaving like the surrounding epidermal cells. In summary, ceh-16/engrailed, during embryonic development, primes a transcriptional cascade necessary for seam cell differentiation. To allow this separate seam cell differentiation, and to maintain the lateral epidermal cell fate, ceh-16/engrailed also represses the fusion of the seam cells.

ceh-16/engrailed commits the lateral epidermis to seam cell fate in part by regulating the expression of elt-5 (=egl-18). As mutations of elt-5, like ceh-16, have also been shown to prevent cell fusions and, to a minor extent, inappropriate cell migration of the seam cells (Koh and Rothman,2001), we suggest a regulatory cascade in which ceh-16controls elt-5, which may repress eff-1 expression and participates in anti-migratory mechanisms(Fig. 7C,D). Although we are able to ectopically express elt-5 by misexpression of ceh-16we do not know if elt-5 is a direct target of ceh-16. Moreover, supporting an indirect regulation, we found no putative ceh-16/engrailed binding site in the elt-5 locus by screening in silico, using the reported Drosophila engrailedbinding sequence (Solano et al.,2003). Are all the functions of ceh-16/engrailed mediated by elt-5? Although the phenotype of elt-5 larvae is very similar to the one seen in ceh-16 mutants, we think that this is not the case. Koh et al. (Koh et al.,2002) showed that elt-5 controls many markers of the seam cells. But nhr-73 and nhr-74 are not regulated by elt-5, so the authors speculated that there must be an additional factor X, which might act in parallel to elt-5. We have shown that ceh-16 regulates nhr-73/74 and elt-5. So ceh-16 may be the factor X, which is placed upstream of elt-5 (Fig. 7C). Interestingly, in the ventral region of the epidermis, cell fusions are controlled by the expression of another homeobox repressor, lin-39/HOXD4/Dfd and ceh-20/Exd(Shemer and Podbilewicz,2002), which also act through elt-5(Koh et al., 2002). In this region, elt-5, controlled by lin-39, is essential for vulva formation. Therefore, it seems that a concerted spatial-temporal(lateral-ventral) expression of different homeodomain proteins controls differentiation of respective epidermal structures by recruiting in part the same factors (such as elt-5 and eff-1; Fig. 7D). The occurrence of elt-5 in lateral and ventral domains of the epidermal cells might be required for the regulation in both areas of eff-1(Fig. 7D).

The expression of lin-39 in VPCs is regulated via wnt/rassignaling pathways (Eisenmann et al.,1998). But how is the expression of ceh-16/engrailedregulated? The exclusive epidermal expression of ceh-16/engrailed in the seam cells cannot be explained by lineage relationships as it is for engrailed-positive cells within a compartment in Drosophila(reviewed by Dahmann and Basler,1999). One possible way to explain such a simultaneous expression of the same gene in the same topological area is the existence of an extracellular signal directing the expression of ceh-16/engrailedfrom an organizing zone. There is so far only one extracellular signal known to control embryonic epidermal development in C. elegans: the ephrin signaling pathway (Chin-Sang et al.,1999; Chin-Sang et al.,2002; George et al.,1998). The ephrins act in the neuroblasts, regulating the ventral closure of the overlying epidermal cells. This function is indirect and the signaling pathway mediating this function is unknown. We consider it unlikely that ceh-16/engrailed expression is dependent on ephrin function, as the phenotype we have observed occurs in animals that have already performed ventral closure. It is, however, conceivable that ephrin signaling attracts ceh-16(—) seam cells to the ventral side at later stages (see below).

The mosaic analyses and the time-lapse recordings show that the seam cells form a straight cell line that acts as a migration barrier(Fig. 7A,B). Differential cell adhesion properties may account for cells segregating from this line, as in mosaic animals ceh-16(—) cells invade the neighboring tissues. At this stage we cannot say whether this phenomenon is cell-autonomous or not. Since ceh-16/engrailed is expressed earlier in precursors of ventral and dorsal epidermis, it may be necessary for correct migration events there as well but not detectable by our experimental means. That ceh-16/engrailed may elicit such a phenotype might be due to the de-regulation of homophilic cell-surface molecules regulating adhesion and/or cell motility. Analogously, in Drosophila such mechanisms have been hypothesized to be involved in the formation/maintenance of compartment boundaries, where the additional paracrine function of hh is required for the boundary to be held in place(Dahmann and Basler, 1999). Studies based on rRNA have demonstrated that nematodes had previously been misplaced, and their true position is in a sister group of the arthropods(Adoutte et al., 2000; Aguinaldo et al., 1997). Therefore their evolutionary relationship to arthropods is closer than expected. The question arises whether engrailed in nematodes is controlling cell migration mechanisms by regulating the same molecules as in arthropods (Drosophila). To answer this question the identification in both species of these cell-adhesion molecules (controlled by engrailed) is required.

In conclusion, this work shows how engrailed patterns the embryonic epidermis of C. elegans. For this purpose, ceh-16/engrailed acts as a differentiation factor, as a cell migration inhibitor, and we describe for the first time how an engrailed-like gene controls animal developmental processes also by the regulation of cell fusion.

We would like to thank T. R. Burglin, N. Patel, J. Rothman, A. Chisholm, A. Fire and P. Sengupta for reagents and helpful discussions, and to Anna Poli and members of our labs for comments on the manuscript. Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). G.C. was a recipient of the EMBO long-term fellowship program. Research in the B.P. lab was supported by the Israel Science Foundation — The Charles H. Revson Foundation (grant no. 203/00-2), US-Israel Binational Science Foundation and Human Frontier Science Program. Research in the lab of R.B. was funded by the European Community, the Friedrich-Baur Stiftung, and the Fonds der Chemischen Industrie.