Nuclear receptor NHR-25 is required for cell-shape dynamics during epidermal differentiation in Caenorhabditis elegans

Developmental processes involve shape changes of epithelial cells. These cellular movements depend upon cytoskeletal actin and cell-adhesion molecules, and are regulated by conserved pathways. In the nematode Caenorhabditis elegans, the best-studied process of epithelial cell motility is the embryonic ventral closure (Chin-Sang and Chisholm, 2000; Simske and Hardin, 2001). Considerably less is known about the regulation of cellular movements that shape C. elegans epidermis during the postembryonic development.

The postembryonic epidermis of C. elegans offers an excellent model to study dynamic modulation of cell morphology, cell contacts and cell fusion. The worm epidermis consists of the hyp7 syncytium, hyp cells of the head and tail, the ventral P cells and the lateral seam cells. Ten seam cells initially present on each side of a freshly hatched worm serve as stem cells, dividing during the four larval stages (L1-L4) to generate epidermal cells that join the hyp7 syncytium, neuroblasts and additional seam cells (Sulston and Horvitz, 1977) (Fig. 1). During the L1 stage, seam cells V1-V6 are connected via adherens junctions to the next seam cell in the row and to six pairs of the P cells. The initial V-cell division produces anterior daughters that first separate the pairs of P cells and then fuse with the hyp7 syncytium, thus creating gaps between seam cells. The seam cells elongate in the anteroposterior direction to re-establish adherens junctions with their neighbors (Austin and Kenyon, 1994; Podbilewicz and White, 1994). The cycle of loss and reconstitution of contacts between adjacent seam cells reiterates once per larval stage, whenever newborn cells fuse with the hyp7 syncytium.

Regaining contacts between seam cells is important for at least four reasons. First, reaching the contact is thought to be a signal for the seam cells to stop elongating (Austin and Kenyon, 1994). Second, cell-ablation experiments have shown that a unique seam cell, V5.p, requires its neighbors to produce a neuroblast (Sulston and White, 1980; Austin and Kenyon, 1994). Third, the integrity of the seam is crucial for the proper anchoring of vulval and uterine cells, and for the attachment of vulval muscles (i.e. conditions for the normal functioning of the egg-laying system) (for a review, see Michaux et al., 2001). Finally, after the connected seam cells cease dividing by the L4 stage, they fuse into a syncytium, which makes adult-specific cuticular ridges called alae. Understandably, gaps in the syncytium cause gaps in the alae (Singh and Sulston, 1978).

Regulation of the active process of seam-cell reconnection is poorly understood. We find that a transcription factor, the nuclear receptor NHR-25, is responsible for the renewal of seam-cell contacts following the asymmetric cell division. RNA interference (RNAi) silencing of nhr-25 blocks the elongation of seam cells that consequently remain isolated within the hyp7 syncytium. The lack of cell contacts caused by the loss of nhr-25 function has profound consequences for the ensuing differentiation of the larval and adult epidermis. Although some of these consequences resemble the effects of seam-cell ablation (Sulston and White, 1980), the V5.p cell devoid of contacts with its neighbors surprisingly still gives rise to the anterior neuroblast.

Of the enormous family of 284 C. elegans nuclear receptors, NHR-25 is one of the few that have been ascribed a biological role, and is also one that is evolutionarily conserved (Asahina et al., 2000; Gissendanner and Sluder, 2000; Maglich et al., 2001). The mammalian NHR-25 counterpart SF-1/FTZ-F1 regulates steroid-hormone synthesis (Parker and Schimmer, 1997) and is essential for adrenal and male gonad development in mice (Luo et al., 1994; Sadovsky et al., 1995). In Drosophila, two FTZ-F1 protein isoforms play distinct developmental roles. Interaction of maternal αFTZ-F1 with a Hox protein [Fushi tarazu (FTZ)] is necessary for embryonic segmentation (Guichet et al., 1997; Yu et al., 1997; Suzuki et al., 2001; Yussa et al., 2001). The other isoform, βFTZ-F1, is a key regulator of the steroid-dependent replacement of the larval cuticle (molting) and of metamorphosis that transforms the larva into the pupa and the adult (Broadus et al., 1999; Yamada et al., 2000; Thummel, 2001).

NHR-25 has previously been implicated in several aspects of epidermal development in C. elegans, including embryogenesis, molting, dauer formation and the differentiation of the vulva (Asahina et al., 2000; Gissendanner and Sluder, 2000; Chen et al., 2004; Gissendanner et al., 2004; Hwang and Sternberg, 2004). Here, we show that NHR-25 plays an essential role in postembryonic epidermal-cell elongation and junction that ensure proper differentiation of seam cells and their epidermal descendants. Loss of nhr-25 function also affects epidermal-cell morphology to the point that secretion of a new cuticle and consequently molting are compromised.

C. elegans strains were reared at 20°C according to standard protocols (Brenner, 1974). The N2 variety Bristol was used as the wild type. The following transgenic strains were used.

• GS2806 – dpy-20(e1362); arIs51[dpy-20(+); cdh-3::gfp] (Pettitt et al., 1996) (gift from I. Greenwald, Columbia University, NY).

• RG242 – veIs13[col-19::gfp; rol-6(su1006)] (Abrahante et al., 1998) (gift from A. E. Rougvie, University of Minnesota, Minneapolis, MN).

• JR667 – unc-119(e2498::Tc1); wIs51[unc-119(+); scm::gfp] (Terns et al., 1997) (from the Caenorhabditis Genetics Center).

• SU93 – jcIs1[ajm-1::gfp; unc-29(+); rol-6(su1006)] (Mohler et al., 1998) (from the Caenorhabditis Genetics Center).

• CF237 – muIs[mab-5::lacZ, unc-31(+)] unc-31(e169) (Salser and Kenyon, 1992) (from the Caenorhabditis Genetics Center).

• NL2099 – rrf-3(pk1426) (Simmer et al., 2002) (from the Caenorhabditis Genetics Center).

An nhr-25::gfp construct encoding the first six amino acid residues of NHR-25 fused to green fluorescent protein (GFP) was prepared by fusing a region from a PstI site 10,635 bp upstream of the nhr-25 start codon (Asahina et al., 2000) with gfp in the pPD95.79 vector without the nuclear localization signal. Transformation was performed as described previously (Mello and Fire, 1995).

A 900-bp XhoI-EcoRI fragment of an nhr-25 cDNA clone yk342d8 (from Y. Kohara, National Institute of Genetics, Mishima, Japan) that excluded a sequence encoding the conserved DNA-binding domain was cloned into pBluescript KS^– for double-stranded RNA (dsRNA) synthesis in vitro or the pPD129.36 vector (a gift from A. Fire, Carnegie Institution of Washington, Baltimore, MD) for dsRNA expression in Escherichia coli HT115 (a gift from D. Court, National Cancer Institute, Frederick, MD). For nhr-23 RNAi, dsRNA was synthesized from a linearized vector 4666 (Kostrouchova et al., 2001). RNAi silencing by dsRNA injection into the adult gonad was done as described previously (Asahina et al., 2000); RNAi by feeding was performed essentially as described (Timmons et al., 2001). Expression of dsRNA in E. coli HT115, grown to an optical density at 600 nm (OD₆₀₀) of 0.4 in LB medium (Sambrook et al., 1989) containing 50 μ g ml^–1 ampicillin or carbenicillin and 12.5 μg ml^–1 tetracycline, was induced for 4 hours with 0.4 mM isopropyl-β-D-thiogalactoside (IPTG). Bacteria were then concentrated by centrifugation and seeded on nematode growth medium (NGM; Brenner, 1974)/agarose plates containing both antibiotics and 0.4 mM IPTG. Worms were placed on the seeded plates and kept at 20°C. Bacteria carrying pPD129.36 without an insert were used for control.

Eggs laid by control or RNAi-treated adults 14 hours after dsRNA injection were collected at 12-hour intervals. Larvae 0-24 hours after hatching were then subjected to immunostaining as described by Bossinger et al. (Bossinger et al., 2001). Worms were washed in distilled water, transferred onto slides coated with 0.1% poly-L-lysine, 0.2% gelatin, 0.4 mM potassium chromium (III) sulfate, squashed with a cover slip and frozen in liquid nitrogen. Samples were then fixed and rehydrated by the following steps: 10 minutes in 100% methanol, 20 minutes in 100% acetone, 10 minutes in 90% ethanol, 10 minutes in 60% ethanol (all incubations at –20°C) and, finally, 10 minutes in 30% ethanol at room temperature. Slides were washed three times for 10 minutes each in TBT (20 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.1% Tween-20). Incubations were done in TBT containing 1% bovine serum albumin and 1% skimmed milk at 4°C overnight for primary antibodies and 2-3 hours at room temperature for Alexa-488- or Cy3-conjugated secondary antibodies. A rabbit antiserum against a portion of the ligand-binding domain [amino acids 294-475 (Asahina et al., 2000)] of bacterially expressed NHR-25 protein was used at a dilution of 1:1000. The anti-DLG-1 antibody (kindly provided by O. Bossinger, Heinrich-Heine Universität, Düsseldorf, Germany) was diluted 1:400 and the MH27 antibody (Francis and Waterston, 1985) recognizing AJM-1 was diluted 1:500. Stained worms were observed using either the Zeiss LSM 410 laser-scanning microscope or with Zeiss Axioplan 2 equipped for epifluorescence.

Worms were washed with distilled water, frozen in liquid nitrogen and vacuum dried for 45 minutes. Following incubation in acetone for 3 minutes at –20°C, worms were vacuum dried for 10 minutes. β-Galactosidase activity was detected by incubation at room temperature in 200 mM sodium phosphate (pH 7.5), 1 mM MgCl₂, 0.004% sodium dodecyl sulfate (SDS), 10 mM potassium ferricyanide/ferrocyanide, 75 μg ml^–1 kanamycin, 0.04% X-gal. The reaction was terminated by washing with S-basal (100 mM NaCl, 50 mM potassium phosphate, 5 mg/l cholesterol) medium and samples were mounted in 90% glycerol. This procedure has been adopted from http://cobweb.dartmouth.edu/~ambros/worms/17.html.

Control and nhr-25(RNAi) adults were washed in S-basal medium and anesthetized in 8% ethanol. Worms were fixed in 0.2 M cacodylate buffer (pH 7.2) with 2.5% glutaraldehyde at 4°C for 4 hours followed by post-fixation in 2% OsO₄ at 4°C for 2 hours. For transmission electron microscopy (TEM), samples were dehydrated by an acetone series and embedded in Epon resin. Ultrathin sections were double stained with uranyl acetate and lead citrate, and viewed with a JEOL 1010 transmission electron microscope. For scanning electron microscopy (SEM), samples were subjected to critical-point drying, coated with gold and viewed with a JEOL 6300 scanning electron microscope.

The lateral seam cells of the C. elegans epidermis are arranged in a row, with each seam cell being connected via adherens junctions to its anterior and posterior neighbor, and to the surrounding hyp7 syncytium. During each larval stage, the seam-cell contacts are interrupted by an asymmetric division of the V1-V6 seam cells, whose newborn anterior daughters fuse with the hyp7 syncytium, leaving temporary gaps (Austin and Kenyon, 1994) (Fig. 1). Staining of adherens junctions with antibodies against DLG-1 or AJM-1 (data not shown) shows that the V1.p-V6.p posterior daughters actively bridge the gaps to restore their contacts in normally developing L1 animals (Fig. 2A-D).

Because nhr-25 is essential in the epidermis (Asahina et al., 2000), we tested its role in the dynamic changes of seam-cell shape. Worms were subjected to RNAi by injection of adult hermaphrodites with nhr-25 dsRNA. L1 progeny stained shortly after hatching showed wild-type pattern of adherens junctions (Fig. 2E). Seam cells then divided and their daughters fused normally with the hyp7 syncytium, leaving the seam cells separated by gaps (Fig. 2F,G). However, the seam cells then failed to extend processes and remained isolated in nearly all of the nhr-25(RNAi) worms (Fig. 2H). Some animals displayed gaps between all seam cells whose parents produced a hyp daughter during the L1 division, others only between some of these cells, probably reflecting an uneven RNAi penetrance. Gaps never occurred between the head seam cells H0 and H1.a (Fig. 3C), because H1 produces its hyp daughter posteriorly (Fig. 1). Consequently, the H2.p seam cell was isolated, whereas H0 and H1.a stayed in contact throughout the L1 and L2 stages.

To see whether the observed cellular defects were specific to NHR-25, adherens junctions were stained in L1 worms following RNAi silencing of a functionally related nuclear receptor, NHR-23 (Kostrouchova et al., 1998; Kostrouchova et al., 2001). Although the expected nhr-23(RNAi) anomalies such as molting defects and dumpy (Dpy) resulted (not shown), silencing of nhr-23 did not perturb the ability of seam cells to re-establish contacts after the asymmetric division and hyp cell fusion (Fig. 2I). Therefore, NHR-25 has a unique role in epidermal development.

Next, we asked whether the ability of seam cells to restore their contacts depends on a postembryonic effect of NHR-25 or whether nhr-25 RNAi perturbs some earlier embryonic function. Seam cells normally regain their contacts within 1 hour after their anterior daughters have fused with hyp7 (Austin and Kenyon, 1994) (i.e. by 8 hours after hatching). RNAi applied by feeding freshly hatched larvae of the RNAi-sensitive rrf-3(pk1426) strain on E. coli expressing nhr-25 dsRNA produced seam-cell contact phenotypes identical to those obtained by injection (e.g. Fig. 3G), indicating that NHR-25 is required after hatching.

At the L2 stage, seam cells V1.p-V4.p and V6.p duplicate, whereas V5.p generates an anterior neuroblast (V5.pa) that in turn gives rise to a sensory organ composed of four cells, which will hereafter be referred to as the postdeirid (Fig. 1). Based on V-cell laser ablation experiments, postdeirid formation requires a physical contact of V5.p with both V6.p and V4.p or another anterior seam cell that can extend far enough to reach V5.p (Sulston and White, 1980; Austin and Kenyon, 1994). The killing of V6 leads to ectopic expression of the homeotic gene mab-5 in V5.p and its descendants, which prevents postdeirid differentiation (Hunter et al., 1999). The absence of V5.p neighbors does not preclude postdeirid formation in mab-5 mutant background (Austin and Kenyon, 1994).

Staining of the adherens junctions confirmed that during and immediately after the L2 doubling division all the seam and future postdeirid cells maintained contacts in wild-type worms (Fig. 3A,B). By contrast, in nhr-25(RNAi) animals, the Vn.p cells remained isolated during and after their duplication (Fig. 3C). Interestingly, the V5.pa cells in these animals divided in a postdeirid neuroblast manner regardless of the absence of V5.p contacts with V4.p or V6.p (Fig. 3D,E). By the time that the V cells underwent their next asymmetric division in early L3 control animals, the four postdeirid cells have lost adherens junctions (Fig. 3F). The same pattern of postdeirid formation was observed in nhr-25(RNAi) worms even when V5 descendants lacked contacts with both their anterior and their posterior neighbors (Fig. 3G). Thus, nhr-25 silencing had a similar effect to mab-5 mutation, in that it allowed the postdeirid to form in the absence of the cell-cell contacts. However, this effects was not due to the loss of mab-5 expression, because a mab-5::lacZ construct was ectopically active in seam cells positioned anterior to V6.p in nhr-25(RNAi) worms (Fig. 4).

Following the round of proliferative division at the L2 stage, the isolated couples of seam cells in nhr-25(RNAi) animals divided further to produce cells that should normally assume the hyp7 fate (Fig. 1). However, instead of fusing with the syncytium, the newborn cells remained attached to the Vn.p stem cells via abnormal anteroposterior and dorsoventral adherens junctions (Fig. 3C,H,I). The presumed hyp7 cells within these clusters continued to express a seam-cell-specific nuclear marker scm::gfp in the JR667strain (Fig. 3H) and could not be distinguished from the original seam cells. Some of the clusters contained more than the four cells normally generated by a single Vn.p during the L2 stage (Fig. 1), implying additional cell divisions of the incorrectly specified daughters (Fig. 3H,I).

Seam cells produce hyp7-fusing anterior daughters once per larval stage and thus need to elongate and reconnect periodically (Fig. 1). It was of interest to know whether the effect of nhr-25 RNAi on cell-contact renewal was limited to L1 or whether it also affected later stages. By transferring 14-hour-old larvae from control bacteria to bacteria expressing nhr-25 dsRNA, we applied RNAi about 6 hours after seam cells have regained their contacts (Austin and Kenyon, 1994). Loss of NHR-25 at this time thus could not affect seam-cell elongation following the first L1 division, yet it resulted in gaps between seam cells visible at the L4 stage (Fig. 5C). When worms were subjected to RNAi at 20 hours after hatching, gaps were observed in the adult seam (data not shown). These results suggested that nhr-25 is required throughout development for the periodic restoration of mutual seam-cell contacts.

By the late L4 stage, seam cells had fused into a continuous syncytium in control animals (Fig. 5A). By contrast, gaps between seam cells prevented the formation of such an uninterrupted syncytium in nhr-25(RNAi) worms (Fig. 5B,C,E). Interestingly, the expression of a reporter derived from the C. elegans cdh-3 cadherin gene (cdh-3::gfp) was reduced or undetectable in the isolated seam cells, in contrast to those of control animals (Fig. 5D,E). When some seam cells in nhr-25(RNAi) worms eventually fused, their incorrect alignment along the anteroposterior axis caused bifurcations of the syncytium (Fig. 5F). In addition, nhr-25(RNAi) adults contained on each lateral side up to 26 (20±3 on average, n=50) nuclei positive for the seam-cell-specific marker (Fig. 5H) as opposed to the 16±0.3 (n=50) nuclei in controls (Fig. 1). The superfluous nuclei were smaller than normal and were close to each other (Fig. 5H), suggesting that they arose by extra cell divisions of the incorrectly specified seam-cell daughters (Fig. 3H).

The effects of nhr-25 could be autonomous to seam cells if nhr-25 is expressed there. To test this possibility, we used an antibody against the NHR-25 protein. This antibody stained the same nuclei lit by the scm::gfp seam-cell-specific marker at L1 (Fig. 6A) and later stages (data not shown). A weaker signal was observed in the nuclei of hyp7 cells produced by the first asymmetric division of the V cells. A transgenic nhr-25::gfp construct confirmed that nhr-25 was expressed in the seam cells from L1 (not shown) until adulthood (Fig. 6B,C); consistent with the antibody staining, this construct was less active in the hyp7 daughters.

The seam syncytium makes a cuticular structure called the alae, which consists of three continuous ridges on each side of the worm (Fig. 7A,C). Compared with controls, alae of nhr-25(RNAi) adults were shallow (Fig. 7B) and the middle ridge was interrupted (Fig. 7D). Gaps (Fig. 7E) and minor branching (data not shown) in nhr-25(RNAi) alae corresponded with gaps and bifurcations in the underlying seam syncytium, respectively (Fig. 5). Branching of the alae has been known to occur in mutants for certain genes, such as rol-6, which encodes a cuticular collagen (Cox et al., 1980; Thein et al., 2003). The rol-6(su1006) mutation is routinely used as a dominant visible marker for C. elegans transformation (Mello and Fire, 1995) and by itself caused moderate alae branching in about 47% of cases in two independent strains (Fig. 7F, Table 1). nhr-25 RNAi in the rol-6(su1006) background dramatically enhanced the severity of branching (Fig. 7G) and increased its incidence to over 83% (Table 1).

Another cuticular structure affected by nhr-25 RNAi were the annuli, transverse rings around the whole body that are interrupted only by the lateral alae. Compared with wild-type controls, the annuli were shallower and sparser in all nhr-25(RNAi) adults, particularly on the lateral sides, where a thick layer of hyp7 syncytium underlies the cuticle (Fig. 7D). The subcellular morphology of the lateral part of the adult syncytium displayed striking anomalies (Fig. 7B). On the dorsolateral and ventrolateral sides, the hyp7 syncytium was abnormally enlarged instead of forming a thin layer between the cuticle and muscles (Fig. 8A,B). Epidermal attachments (fibrous organelles that normally fasten muscles to the cuticle) could not be found in nhr-25(RNAi) adults, and the structure of the epidermal basement membrane and the muscle morphology were altered as well. In addition, no struts that would link the two layers of the adult cuticle were visible. Instead, unusual vesicles that might represent faulty cuticle secretion were trapped between the cuticle layers (Fig. 8B). These anomalies of the epidermal ultrastructure are probably the cause of defective L4-to-adult molt, observed in about 30% of wild-type and in 100% of rrf-3(pk1426) worms that had been fed on nhr-25 dsRNA-expressing bacteria from hatching. Unshed L4 cuticle remained attached to their heads or tails and caused body constrictions in these nhr-25(RNAi) adults (Fig. 8C,D).

By preventing seam cells from restoring mutual contacts after asymmetric divisions, nhr-25 RNAi emulated the effect of V-cell ablation. Anterior daughters of the isolated seam cells assumed the wrong fate: they failed to fuse with the hyp7 syncytium, continued expressing a seam-specific transgene and occasionally divided, thus resembling their parent stem cells. These results were consistent with the early observations of Sulston and White (Sulston and White, 1980) that `a gap in the seam can cause the remaining cells to proliferate more than usual', this being `due to some of the anterior daughters, which normally fuse with the hypodermal syncytium, persisting as seam cells and subsequently dividing again'. It is not clear yet why the absence of the cell contact conserves the seam fate in the anterior daughters. However, the extra seam cells were not produced by additional heterochronic divisions, such as those occurring in mutants for another nuclear receptor DAF-12 (Antebi et al., 2000).

Consistently with its transcriptional function, we have detected NHR-25 in the nuclei of seam and, at a lower level, in hyp7 epidermal cells. This finding supports the possibility that the role of NHR-25 in the renewal of seam-cell contacts might be autonomous to the seam cells, although we still cannot exclude the possibility that NHR-25 is required in the surrounding hyp7 syncytium. Moderate antibody staining and activity of nhr-25::gfp constructs (Gissendanner and Sluder, 2000) (this study) in hyp7 suggest that NHR-25 also plays a role in this part of the epidermis. Such a role is apparent from the effect of nhr-25 RNAi on the hyp7 subcellular structure.

The failure of seam cells to restore contacts might be caused by their inability to emit a mutually attractive signal, incompetence to sense such a signal, disrupted cytoskeleton or impaired recognition and adhesion to their neighbors. However, we have seen normal expression and localization of DLG-1, which is required for proper cytoskeleton organization (Bossinger et al., 2001; Firestein and Rongo, 2001; McMahon et al., 2001) and of AJM-1, an essential component of adherens junctions (Koppen et al., 2001), in seam cells affected by nhr-25 RNAi. Also, E-cadherin (HMR-1), α-catenin (HMP-1) and β-catenin (HMP-2), which connect cytoskeleton with the extracellular matrix (Costa et al., 1998), have appeared undisturbed by a loss of nhr-25 function (Chen et al., 2004). We have found that cdh-3::gfp, which encodes a promoter fusion of a Fat-like cadherin, is underexpressed in seam cells that have lost mutual adherens junctions as a result of nhr-25 RNAi. CDH-3 is required for differentiation of the hyp10 syncytium and has also been suggested to play a role in cell-cell recognition, adhesion and the maintenance of the seam integrity (Pettitt et al., 1996). Reduced cdh-3 expression could therefore offer a plausible explanation for the loss of cell-cell contacts in nhr-25(RNAi) animals. However, although cdh-3 is expressed in seam cells throughout development, seam-cell morphology appears normal in cdh-3 mutants, possibly because of gene(s) compensating for CDH-3 function (Pettitt et al., 1996). Thus, cdh-3 might not be the sole gene downstream of NHR-25 action.

Another candidate NHR-25 target is acn-1, which encodes a peptidase of the angiotensin-converting enzyme family. Expression of an acn-1::gfp reporter was reduced in seam cells upon nhr-25 RNAi, and some phenotypes caused by silencing of acn-1, including abnormal alae and molting defects, resembled those induced by nhr-25 RNAi (Brooks et al., 2003). However, although we reproduced these acn-1 RNAi phenotypes, we found that Vn.p cells re-established their contacts normally, therefore excluding ACN-1 as an effector of NHR-25 in this aspect of development. Gaps in the seam were observed in mutants for a transmembrane protein of the plexin family PLX-1 (Fujii et al., 2002). Plexins are receptors for semaphorins, secreted or transmembrane cues for axon guidance (Raper, 2000). Of two C. elegans plexin and three semaphorin genes, plx-1 and mab-20 [encoding semaphorin 2a (Roy et al., 2000)] are active in the larval seam cells.

The disrupted seam cell contacts are probably unrelated to molting defects seen in nhr-25(RNAi) worms, because gaps between seam cells occur even in nhr-25(RNAi) animals that had molted normally and molting defects have not been reported for plx-1 mutants that display the seam-cell phenotype.

Laser ablations (Sulston and White, 1980) suggested that the physical contact of V5.p with V6.p and V4.p was a signal for V5.p to produce an anterior neuroblast V5.pa and hence the postdeirid. Austin and Kenyon (Austin and Kenyon, 1994) made a strong case for cell-cell contact, as opposed to a diffusible signal, by showing that a delay in the V5.p commitment was equal to the time needed for a distant seam cell to reach V5.p. Without the contact, no postdeirid was formed. We were therefore surprised to see that, in nhr-25(RNAi) L2 worms, the V5.pa cells divided in the neuroblast fashion even when V5.p was disconnected from its neighbors.

For V5.pa to assume the neuroblast fate, a homeotic gene (mab-5) acting downstream in the Wnt signaling must be turned off in V5.p. The downregulation of mab-5 requires V6.p (Hunter et al., 1999) and a negative regulator of the Wnt pathway PRY-1 (Maloof et al., 1999; Korswagen et al., 2002). Ectopic mab-5 expression that occurs in the absence of V6.p or in a pry-1(–) background causes V5.p to divide symmetrically as other Vn.p cells do at the L2 stage. Misexpression of mab-5 and the resulting inhibition of the postdeirid program requires the normal function of bar-1, encoding a β-catenin, and some other components of the Wnt pathway (Hunter et al., 1999; Maloof et al., 1999; Korswagen et al., 2002).

How can a V5.p cell, devoid of contacts with V6.p in nhr-25(RNAi) larvae, give rise to a postdeirid-like structure? We hypothesize that blocking of the postdeirid program requires NHR-25 in a similar way to that in which it requires mab-5 or bar-1 function. Because mab-5::lacZ is ectopically expressed in nhr-25(RNAi) background, NHR-25 might act in the Wnt pathway, at the same level as MAB-5 or downstream of it. We propose that NHR-25 might exert its effect by directly interacting with the MAB-5 protein. This possibility is supported by a cooperation between a Drosophila NHR-25 ortholog αFTZ-F1 and a MAB-5-related homeodomain protein FTZ during embryonic segmentation (Guichet et al., 1997; Yu et al., 1997; Schwartz et al., 2001; Suzuki et al., 2001; Yussa et al., 2001). Moreover, NHR-25 has been recently shown to interact with the C. elegans homeodomain proteins LIN-39 and NOB-1 during vulval and embryonic development, respectively (Chen et al., 2004). Our preliminary data suggest that MAB-5 modulates NHR-25-dependent transcriptional activation in a yeast-based assay (M.S., M. van Gilst, K. Yamamoto, M.J. and M.A., unpublished).

Although loss of nhr-25 causes disruption of seam-cell contacts and compromises the proper differentiation of the seam and the cuticular alae, these effects are unlikely to be directly connected with overall molting defects seen in nhr-25(RNAi) worms. Molting is a complex process that can fail at any of several steps, such as expression of a cuticle component, secretion of cuticle by the epidermis, separation of a new cuticle from the old one and behavior leading to the old cuticle removal. Although the effect of nhr-25 RNAi on shedding cuticle has been well established (Asahina et al., 2000; Gissendanner and Sluder, 2000; Chen et al., 2004; Gissendanner et al., 2004), the studies have not discriminated between the possible causes. Based on the disturbed epidermal ultrastructure and the presence of abnormal vesicles between cuticle layers, we propose that NHR-25 acts generally upon cuticle secretion. Missing cuticular struts and aberrant alae and annuli are examples of cuticular structures that are not properly formed in nhr-25(RNAi) adults. Interestingly, silencing of nhr-25 enhances the effect of a semidominant mutation in the collagen gene rol-6 on alae branching. Although we could speculate that rol-6 or other cuticle gene(s) might be regulated by NHR-25, the observed interaction with rol-6(su1006) probably results from the effect of nhr-25 RNAi on increasing the incidence of aberrant seam-cell contacts.

Molting defects, manifested as unshed cuticle, also result from RNAi silencing of the nuclear receptor NHR-23 (Kostrouchova et al., 1998). Reduced mRNA levels from the collagen gene dpy-7 are observed in nhr-23(RNAi) animals and suggest a role of NHR-23 in cuticle synthesis (Kostrouchova et al., 2001). We have previously raised the point that molting relies on the same genetic regulation in insects and nematodes, with NHR-23 and NHR-25 being its evolutionarily conserved core (Asahina et al., 2000). This idea considered the role of the NHR-23 and NHR-25 orthologs DHR3 and βFTZ-F1, respectively, in Drosophila molting and metamorphosis, triggered by the steroid hormone ecdysone (Thummel, 2001). In the ecdysone pathway, expression of βFTZ-F1 requires DHR3 (Kageyama et al., 1997; Lam et al., 1997; White et al., 1997). Although a similar epistasis has not been shown for NHR-23 and NHR-25 (Kostrouchova et al., 2001), both factors were recently suggested to activate a common target gene, acn-1, whose silencing itself disrupts molting (Brooks et al., 2003). We find that nhr-23 RNAi does not interfere with the ability of seam cells to restore adherens-junction contacts upon asymmetric divisions. Thus, although NHR-25 and NHR-23 might execute the molting process through a partially common genetic pathway, NHR-25 has a unique role in the epidermis independent of NHR-23. The dual role of NHR-25 (epidermal-cell differentiation and molting) is intriguingly parallel to the two distinct functions of the Drosophila FTZ-F1 isoforms, αFTZ-F1 in segmentation and βFTZ-F1 in steroid-dependent metamorphosis and molting.

We thank the Electron Microscopy Lab at the Institute of Parasitology for technical help, O. Bossinger for providing the anti-DLG-1 antibody and for useful advice, S. Hirose for supporting the preparation of the NHR-25 antibody and M. Labouesse for comments on the manuscript. We appreciate receiving plasmid vectors from A. Fire and C. elegans strains from I. Greenwald, A. Rougvie and the Caenorhabditis Genetics Center, supported by the National Institutes of Health. The nhr-25 clone yk342d8 was provided by Y. Kohara and the nhr-23 plasmid 4666 by M. Kostrouchova. The MH27 antibody was obtained from the Developmental Studies Hybridoma Bank at the University of Iowa. This work was supported by grant KJB5022303 and project Z60220518 from the Academy of Sciences of the Czech Republic. M.S. was supported by a graduate studies program 524/03/H133 from the Grant Agency of the Czech Republic.