C. elegans dystroglycan DGN-1 functions in epithelia and neurons, but not muscle, and independently of dystrophin

Dystroglycan (DG) is a crucial receptor for the basement membrane (BM) in vertebrate muscle. It is composed of an extracellular α subunit and a transmembrane β subunit produced by proteolytic processing of a single precursor (Ibraghimov-Beskrovnaya et al.,1992). Laminin G-like domains in extracellular matrix (ECM)proteins, such as laminin, agrin and perlecan, and the neural cell surface neurexins bind O-linked glycans on α-DG(Hohenester et al., 1999; Michele et al., 2002).β-DG binds to α-DG (Sciandra et al., 2001) and mediates its transmembrane linkage to intracellular cytoskeletal and signaling proteins(Winder, 2001). DG is a component of the skeletal muscle dystrophin-glycoprotein complex (DGC) that includes the transmembrane sarcoglycan complex and the cytoplasmic dystrophin-associated protein complex (DAPC), which contains the actin-binding protein dystrophin, dystrobrevin and syntrophin(Cohn and Campbell, 2000; Winder, 2001). Loss-of-function mutations in DG and other DGC components result in sarcolemmal damage and muscular dystrophy(Cote et al., 1999; Cohn and Campbell, 2000; Cohn et al., 2002; Michele and Campbell, 2003; Parsons et al., 2002).

DG also functions as a BM receptor in non-muscle tissues. Unconditional knockout of DG in mice results in early embryonic lethality due to failure of extraembryonic BM formation (Williamson et al., 1997). DG functions as an important laminin receptor in developing kidney, lung and salivary epithelia(Durbeej et al., 1995; Durbeej et al., 2001). The loss of DG in epithelial-derived breast tumor cells leads to a failure of ECM-induced cell polarization and enhanced invasiveness(Muschler et al., 2002). Brain-specific knockout of DG results in pial BM discontinuities and cortical neuron migration defects (Michele and Campbell, 2003; Moore et al.,2002). Knockout of DG in Schwann cells produces defects in myelination and nodal architecture (Saito et al., 2003). The importance of DGC components to DG function outside of muscle is unclear, as mice lacking both dystrophin and utrophin, or the sarcoglycan complex, do not display the nervous system defects or embryonic lethality seen in the brain-specific and unconditional DG knockouts,respectively (Rafael et al.,1999; Imamura et al.,2000).

Homologs of DG and other DGC components have been identified in Drosophila melanogaster and Caenorhabditis elegans(Dekkers et al., 2004; Deng et al., 2003; Grisoni et al., 2002). Drosophila DG is expressed in follicle and imaginal disc epithelia and the oocyte, and its loss disrupts epithelial and oocyte polarity(Deng et al., 2003). Drosophila DG and other DGC components are also expressed in the nervous system and some muscle (Dekkers et al., 2004). In C. elegans, DAPC complex homologs function in muscle with the acetylcholine transporter SNF-6 to regulate cholinergic stimulation (Bessou et al.,1998; Gieseler et al.,2001; Grisoni et al.,2003; Kim et al.,2004), but the roles of other DGC components have not been extensively characterized.

We report the characterization of DGN-1, the C. elegans ortholog of vertebrate DG. DGN-1 is expressed in epithelia in the gonad and other tissues, and in neurons. DGN-1 is not found in muscle and does not function with the conserved DAPC complex in C. elegans muscle. DGN-1 plays an important role in gonad epithelial development, where it is likely to mediate the function of laminin. DGN-1 also affects guidance and the extension of cell processes along BM surfaces. These findings suggest a conserved role for DG in mediating epithelial and neural cell responses to the ECM.

Culture and manipulation of C. elegans were performed according to standard methods (Brenner,1974) at 20°C. The following strains were used: wild-type N2 var. Bristol; NJ52 epi-1(rh27); NJ244 epi-1(rh92); NJ590 epi-1(rh199); LS292 dys-1(cx18); LS505 dyb-1(cx36);LS721 stn-1(ok292). The following GFP markers were used: DA/DB motoneurons, evIs82 [unc-129::GFP]; early somatic gonad cells/distal tip cell, qIs19 or qIs56 [lag-2::GFP];spermatheca, jcIs1 [AJM-1::GFP]; gonad sheath, tnIs6[lim-7::GFP]; anchor cell and body wall muscle, syIs49[zmp-1::GFP]; mIs11 [myo-2::GFP, pes-10::GFP, gut promoter::GFP], chromosome III integrant; qIs54 [myo-2::GFP, pes-10::GFP, gut promoter::GFP],X chromosome integrant; oxIs12 [unc-47::GFP], X chromosome integrant. dgn-1(cg121) was maintained as a heterozygote balanced by visible X chromosome markers. Alleles of epi-1 were balanced by mIs11. Extrachromosomal arrays of dgn-1 promoter reporter or GFP fusion protein constructs were created by germline injection(Mello et al., 1991). The laminin-β GFP fusion, urEx131 [LAM-1::GFP], and EPI-1 antibody were generous gifts of Bill Wadsworth (Robert Wood Johnson Medical School).

The C. elegans DG homolog (T21B6.2) was identified by BLAST searches and designated dgn-1. Sequencing of the yk671e7 cDNA extended the predicted 5′ end of exon 1 to nucleotide 10315 of cosmid T21B6 (GenBank Z68011), and the 3′UTR to nucleotide 1837. A genomic dgn-1 clone (pJK600) was constructed by inserting 1181-12990 of T21B6 into the XbaI site of BlueScribe M13 Plus (Stratagene). This region includes 2680 bp upstream of exon 1 through the entire 3′UTR of dgn-1. pJK600-containing transgenes rescue dgn-1(cg121)phenotypes, with ∼90% of transgenic animals having restored fertility.

Plasmid pJK602 contains 2680 bp upstream of exon 1 through the middle of exon 2 (4679-12990 of T21B6) inserted between the MluI and HindIII sites of pJJ471. The GFP fusion product contains the first 54 amino acids of DGN-1, a synthetic transmembrane region and GFP. An identical expression pattern was observed with a similar reporter containing only 213 bp upstream of exon 1, suggesting that the major dgn-1 expression controls are in intron 1.

Plasmid pJJ516 was made by inserting GFP from pPD114.38(www.ciwemb.edu/pages/resources.html)into the HindIII site near the end of the dgn-1 coding sequence. The product contains GFP inserted after residue 575 of DGN-1, with the final seven DGN-1 residues at the C terminus. Expression of DGN-1::GFP is identical to that of the dgn-1::GFP promoter reporter, and DGN-1::GFP rescues the sterility of dgn-1(cg121).

DGN-1 regions corresponding to vertebrate α-DG (amino acids 20-379;TRVF...NSFT) and β-DG (amino acids 392-584;VAFS...FIPP) were PCR amplified from yk671e7 and cloned into pGEX-4T1 (Pharmacia) and pMal-c2 (New England Biolabs) to produce glutathione S-transferase (GST) and maltose binding protein (MBP) fusions. Primers used were:

• DGN-1α forward, 5′-CCGCTCGAGACCCGTGTGTTTATTGG-3′;

• DGN-1α reverse, 5′-GGCCTCGAGCTAAGTGAAACTGTTGACTGG-3′;

• DGN-1β forward, 5′-CCGCTCGAGGTGGCTTTCAGCAACAAT-3′; and

• DGN-1β reverse, 5′-GGCCTCGAGTTAAGGAGGAATGAATGG-3′.

• GST-DGN-1 fusion proteins were used as immunogens for the production of rabbit and chicken antibodies, which were affinity purified on MBP-DGN-1 columns.

Deletion mutagenesis was performed essentially as described(Barstead, 1999). Progeny of trimethylpsoralen/UV mutagenized animals were screened by PCR for deletions in dgn-1. The cg121 deletion removes nucleotides 2045-4439 of cosmid T21B6. The cg121 mutant strain was backcrossed to wild type at least six times before further analyses.

Embryos and mixed larval stage animals were pulverized in liquid nitrogen and extracted in PBS, 1% NP40 containing protease inhibitor cocktail (Roche Biochemical). Extracts were centrifuged (50,000 g, 20 minutes,4°C) to remove insoluble material. Alternatively, 25 adults were boiled for 15 minutes in 2% SDS containing protease inhibitors and centrifuged(14,000 g, 10 minutes, 4°C). Extracts were subjected to SDS-PAGE and transferred to nitrocellulose. Filters were blocked in PBS containing 5% non-fat dried milk, incubated with affinity-purified anti-DGN-1 antibody followed by horseradish peroxidase-conjugated secondary antibody(Vector Laboratories), and developed for ECL chemiluminescent detection(Amersham Pharmacia Biotechnology).

For deglycosylation experiments, extracts were adjusted to 1% SDS and boiled for 10 minutes. After cooling, nine volumes of 0.5% NP40, 50 mM Tris-HCl (pH 8.0) containing protease inhibitors was added and samples were digested for 16 hours at 37°C with 1 unit of protein-N-glycosidase F(PNGaseF; New England Biolabs).

Animals were mounted on thin pads of 2% agarose in a drop of M9 buffer(Wood, 1988) containing levamisole (0.1 mM) or sodium azide (10 mM) to immobilize them. Immunohistochemistry was performed as described(Kang and Kramer, 2000). Monoclonal anti-MHC-A myosin, polyclonal anti-LET-2, polyclonal anti-NID-1,and polyclonal anti-EPI-1 antibodies were used as described(Kang and Kramer, 2000; Huang et al., 2003). Polyclonal anti-DGN-1 was used at a 1:100 dilution. Images were collected on a Zeiss Axiophot microscope equipped with a CCD camera. Some fluorescent images were deconvolved (VayTek MicroTome) to remove out-of-focus signals.

Assessment of L4 or adult stage dgn-1(0) animals using integrated GFP markers and/or morphology revealed that: 98% have one to two distal tip cells (strong qIs56[lag-2::GFP]); 99% have one to two clusters of presumptive spermathecal cells (strong jcIs1[ajm-1::GFP]); 90% form one to two anchor cells(syIs50[zmp-1::GFP]); 89% have presumptive gonad sheath cells (tnIs6[lim-7::GFP]); and 63% show a peri-vulval lumen in mid/late L4 stage, indicative of uterine tissue (n=95-100 animals scored for each marker).

For activity on plates, individual L4 stage animals were transferred to NGM plates without bacteria and allowed to recover for 1 minute before counting the number of body bends in a 2-minute interval. For thrashing in liquid, L4 animals were transferred to a drop of M9 medium on an unseeded NGM plate,allowed to recover for 1 minute, then filmed for 2 minutes. The number of body bends, defined as a reversal in direction of head movement, was counted from recordings. For defecation assays, the timing of successive posterior body contraction and expulsion steps (Avery and Thomas, 1997) on seeded NGM plates was recorded for 10 defecation cycles. Assays were performed at room temperature (approximately 22°C).

Three C. elegans genes encode DG-like proteins, dgn-1(T21B6.2), dgn-2 (F56C3.6) and dgn-3 (F07G6.1). DGN-1 is the most similar in sequence and structural organization to vertebrate and Drosophila DGs (Fig. 1B,C). DGN-1 contains the N-terminal immunoglobulin-like domain of vertebrate α-DG (Bozic et al.,2004), which is missing from Drosophila DG(Fig. 1B, N-terminal). A threonine-rich, mucin-like region is present in all DGs, although it is much shorter in DGN-1 (Fig. 1B,N-terminal). dgn-1 orthologs in the nematodes C. briggsae(CBG17551) and C. remanei (ORF on Contig1797.2) show strong conservation of sequence (80-90% identity) and domain organization with DGN-1.

Specific functional amino acid residues are conserved in all DGs, including two cysteine pairs that form disulfide bonds in vertebrate DG(Brancaccio et al., 1998; Deyst et al., 1995) and several predicted N-linked glycosylation sites(Fig. 1B). Two regions of sequence divergence are noteworthy. First, neither invertebrate DG shows strong sequence conservation in the vertebrate α/β proteolytic cleavage region (Fig. 1B,core). Second, crucial residues for binding WW and SH3 domain-containing proteins such as dystrophin (Huang et al.,2000) are not conserved in DGN-1(Fig. 1B, cytoplasmic). These differences are also true for the C. briggsae and C. remaneiorthologs (data not shown).

DGN-2 and DGN-3 share only the DG core with vertebrate DG, corresponding to the α-DG C terminus and the β-DG N terminus(Fig. 1B). In addition to the dgn-1 ortholog, C. briggsae (CBG16240) and C. remanei also have other predicted gene products containing a DG core. This region therefore defines a family of DG-like proteins. Drosophila DG contains two adjacent, divergent copies of this core region (Sciandra et al.,2001). Outside of the core, DGN-2, DGN-3 and CBG16240 show no similarity to each other or to other DG family members. Thus DGN-1 is the unique nematode ortholog of vertebrate DG, and additional species-specific DG-like proteins with variable N-terminal and cytoplasmic domains have arisen in nematodes.

Antibodies against DGN-1 regions corresponding to vertebrate α- orβ-DG were used to detect DGN-1 in lysates of embryos and larvae. Both antibodies detected a band of ∼85 kDa, higher than the 68 kDa predicted mass (Fig. 2A). No bands corresponding to separate DGN-1α or β species were detected,indicating that DGN-1 is not processed into separate α and βsubunits. Embryonic DGN-1 migrates as a compact 85 kDa band and a minor 95 kDa species (Fig. 2C, lane 1), and larval DGN-1 as a heterogeneous band centered around 85 kDa(Fig. 2C, lane 3). PNGase F digestion of embryo extracts shifts the 85 kDa species to a 75 kDa band(Fig. 2C, lane 2). The 95 kDa band shifts slightly to ∼90 kDa, suggesting differential N-glycosylation of this species. PNGase F digestion of larval extracts results in widening of the DGN-1 band and a 10-15 kDa decrease in median apparent weight(Fig. 2C, lane 4). Thus, larval DGN-1 is also N-glycosylated, but its heterogeneity is largely determined by other modifications, possibly including O-glycosylation as occurs with vertebrate α-DG.

Sites of dgn-1 expression were determined using reporters driven by dgn-1 upstream sequences. In early (pre-morphological) embryos, dgn-1::GFP expression is evident in many epithelial and neural precursors comprising the outer layer of cells(Fig. 3A). As elongation begins at comma stage, expression becomes most prominent in several specialized epithelial cells, including pharyngeal e2 and marginal cells, excretory cells,the somatic gonad precursors (SGPs) Z1 and Z4, and rectal epithelial cells(Fig. 3B). Weaker expression is apparent in hypodermal precursors and neuroblasts along the ventral midline. Pharyngeal expression persists through the L3 larval stage, whereas excretory and rectal cell expression persists throughout development. SGP expression(Fig. 3C,E) persists in SGP descendants, such as the distal tip cells (DTCs; Fig. 3I), and increases throughout the gonad during the L4 stage(Fig. 3J). Variable, generally weak expression is seen throughout larval development in several neurons,although PVP neurons show strong expression throughout development(Fig. 3C,F). Transient increased expression occurs in new P cell-derived neurons in the ventral nerve cord in late L1/early L2 stage animals(Fig. 3H). Variable weak expression is seen in hypodermal cells, principally hyp5 in the head(Fig. 3D). Preceding the L4/adult molt, expression increases in the vulval epithelium(Fig. 3K).

The expression pattern was confirmed and subcellular localization determined using anti-DGN-1 antibody staining and analysis of a rescuing DGN-1::GFP fusion protein. Both approaches yielded similar results, which are described for antibody staining. In pre-morphological embryos, DGN-1 is diffuse around the surface of outer ectodermal cells before BMs form(Fig. 4A), but begins to polarize towards basal surfaces as BMs assemble between germ layers(Fig. 4B). Throughout subsequent development, DGN-1 localizes to the basal surfaces of pharyngeal(Fig. 4C), gonadal(Fig. 4C-G), rectal(Fig. 4H), vulval(Fig. 4E,F) and excretory cell(Fig. 4I,K) epithelia, in close apposition to the underlying BM. DGN-1 accumulation is more variable in the hypodermis (Fig. 4K,L) and neurons (Fig. 4J).

DGN-1 antibodies show no detectable staining of muscle, although DGN-1 is present in the epithelium immediately adjacent to muscles(Fig. 4K,L). Staining was not detected in body wall, pharyngeal or specialized muscles of the alimentary and reproductive systems. Moreover, neither the dgn-1::GFP promoter reporter nor the rescuing DGN-1::GFP fusion show expression in muscle. Thus,DGN-1 is notably different from vertebrate DG, as a role for DGN-1 in muscle function is unlikely.

We isolated a deletion, cg121, that removes most of the coding and some of the 3′ untranslated region of dgn-1. Western(Fig. 2B) and northern blot(not shown) analyses showed that cg121 is a molecular null allele. Homozygous dgn-1(cg121) animals are viable but sterile(Fig. 5), and display epithelial and neural defects.

Sterility results from an early disruption of gonad morphogenesis. The wild-type gonad primordium contains two central primordial germ cells (PGCs)flanked at anterior and posterior poles by the two SGPs. The primordium is compact, with a sharp DIC image boundary, indicating a robust surrounding BM(Fig. 5C). In newly hatched cg121 homozygotes, the primordium is usually compact but often displays bulging of PGCs and a weak DIC boundary(Fig. 5D). SGPs are frequently displaced from the poles, sometimes interposing between PGCs(Fig. 5D,H). In some cg121 larvae, no DIC boundary is detectable and gonadal cells spread dorsally along the body wall (Fig. 5E). Older dgn-1 mutants often show swelling in the mid-body region as a result of body wall muscle cell fusion with, or engulfment of, loose germ cells (data not shown), as is also seen in epi-1 laminin mutants (Huang et al., 2003).

The aberrant gonad morphology of dgn-1 mutants suggests defects in the gonad BM. Antibodies to collagen IV and nidogen, and a LAM-1(laminin-β)::GFP fusion were used to examine the distribution of BM components. In early cg121 L1 larvae with compact gonad primordia,localization of BM components around the primordium appears normal(Fig. 6). In larvae with disrupted primordia, localization of BM components is still seen, but staining is weak and diffusely distributed over the surface of the gonad tissue (not shown). Thus, a BM organizes around the dgn-1(0) gonad but is not maintained. No gross alteration in other BMs is seen in dgn-1mutants.

Failure of gonad primordium BM function in dgn-1 mutants may result from an inability of SGPs to form a stable epithelial layer around the PGCs. Mispositioned SGPs, showing strong association of laminin-β::GFP,can be observed covering one PGC while excluding the other(Fig. 6F,G). Thus, the extrusion of germ cells probably reflects a failure of the somatic gonad epithelium and its associated BM to provide a stable barrier.

Organization of gonad tissue in dgn-1(0) animals was examined following the L1/L2 molt, using lag-2::GFP expression and nuclear morphology to discriminate somatic and germ cells. Early L2 dgn-1(0)animals contained 8.8±1.0 somatic gonad cells (n=25), which is close to the wild-type number of 8. The somatic gonad cells of dgn-1(0) animals cluster together in a central clump, or in a contiguous line (sometimes separated into two clusters) along the ventral surface (Fig. 5J,K). In either organization the somatic cells remain in contact, suggesting somatic cell-cell adhesion is retained. By contrast, germ cells are either not in contact or only in peripheral contact with somatic cells(Fig. 5J). These results further indicate the inability of somatic cells to associate with or ensheath germ cells in dgn-1(0) animals. Monitoring of dgn-1(0)gonads throughout the L2 stage did not reveal any overt re-organization of dgn-1(0) somatic gonad cells, indicating a failure of somatic primordium formation in dgn-1(0) animals. By contrast, wild-type gonads showed reorganization of somatic cells into a central somatic primordium and distal tip cells at the ends of the two emerging gonad arms.

The examination of L4/adult dgn-1(0) animals for the expression of markers for differentiated somatic gonad cell types indicated that DTCs,sheath, spermathecal, uterine and anchor cells form in the majority of dgn-1(0) animals (see Materials and methods). These results indicate that the expansion and differentiation of somatic gonad cell lineages is not blocked in the absence of DGN-1. Variable expansion of the germ lineage cell population was noted in dgn-1(0) animals but the fate of germ cells was not followed

Heterozygous cg121/+ animals form a grossly normal gonad, but produce 15% fewer progeny than do wild type (wild type, 309±41; cg121/+, 263±32; n=18). Twenty-three percent of cg121/+ heterozygotes have defects in gonad arm migration(Fig. 7). The arms migrate on the underlying BM led by DTCs, which express DGN-1. Although a range of migration defects is seen in cg121/+ animals, the occurrence of oblique turns and abnormal midline crossings indicates a failure of DTC responses to BM guidance cues (Su et al.,2000).

Eighty-nine percent of cg121 animals show defects in formation of the tubular excretory cell arms. Missing or short arms are seen in most cg121 homozygotes (Fig. 8B,D). Multiple defects can occur in individual animals, but generally only one to two arms are missing or short. Fourteen percent of mutants also have ectopic arms that generally parallel the corresponding normal arm (Fig. 8C,D). Heterozygous cg121/+ animals show the same range of excretory process defects but at a reduced penetrance (Fig. 8D).

dgn-1 mutants do not have strong movement defects, indicating a severe perturbation of muscle or neural function. However, in 31% of dgn-1 mutants, at least one DA/DB type neuron commissure extends on the wrong side of the body (Fig. 9A). Individual animals also show additional axonal defects, such as defasciculation or abnormal branching(Fig. 9B-D). Heterozygous cg121/+ animals show similar DA/DB guidance defects but at a lower penetrance.

The vulval epidermis shows prominent dgn-1 expression starting when the invaginated epidermis everts and tightens into a slit-shaped opening. Forty-five percent of dgn-1 adults have a protruding vulva (Pvl)phenotype, suggesting a detachment of the vulval epithelium from the underlying tissue (Fig. 5, Table 1). Frequently, rupture at the protruding vulva leads to the herniation of internal organs, indicating that dgn-1 function is important in the anchorage of the vulval epidermis.

Ten percent of dgn-1 mutants have two vulvae, while 12% are vulvaless (n=131), suggesting a defect in specification of the vulva-inducing anchor cell (AC)(Greenwald, 1997). Using the AC marker zmp-1::GFP (Inoue et al., 2002), we found 14% of dgn-1 mutants generate two ACs, while 10% fail to generate an AC (n=95). All wild-type animals produced one AC (n=100). This AC specification defect is likely to result primarily from a disruption of gonad structure in dgn-1mutants.

dgn-1 mutants do not show phenotypes associated with defects in muscle function, such as Pat (paralyzed at two-fold) or Unc (uncoordinated movement). No gross disorganization of body wall muscle was apparent by DIC microscopy. The muscles associated with the alimentary system function normally, while function of uterine and vulval muscles cannot be assessed because of the failure of gonad formation in dgn-1 mutants.

C. elegans contains single laminin β and γ chains and two α chains, LAM-3/laminin-αA and EPI-1/laminin-αB(Huang et al., 2003). Both laminin-α chains are distributed broadly in BMs, but EPI-1 uniquely localizes to the gonad BM. Gonad epithelialization fails in epi-1mutants (Huang et al., 2003),resulting in gonad rupture like that seen in dgn-1(0)(Fig. 5F). The penetrance of the gonad defect varies between epi-1 alleles, but is 100% in the putative null allele rh199 (Table 1). SGP mispositioning is seen in newly hatched rh199animals, although at a higher penetrance than in dgn-1 mutants(Fig. 5H). The similarity of gonad phenotypes suggests that DGN-1 may be a crucial receptor for EPI-1 in the gonad primordium.

epi-1 mutants show other morphological defects reminiscent of dgn-1 phenotypes, although they often are more severe(Table 1). On average, three to four excretory cell arms are missing in rh199 animals, compared with one to two in cg121, and remaining arms are short, and exhibit aberrant morphology and guidance. epi-1 mutants also show Pvl defects comparable to dgn-1 mutants. These results are consistent with DGN-1 mediating some EPI-1 function in the development of the excretory cell and vulval epithelium.

DG function in vertebrate muscle is mediated by the DAPC complex containing dystrophin and associated proteins(Winder, 2001). We examined C. elegans DAPC mutants of dys-1 dystrophin, dyb-1dystrobrevin and stn-1 syntrophin for morphological phenotypes similar to those of dgn-1. None of these putative functional null mutants show the morphological defects prevalent in dgn-1(0)(Fig. 5, Table 1), indicating that the morphological functions of DGN-1 are largely independent of DAPC function.

C. elegans DAPC mutants display a head muscle hypercontraction phenotype (Fig. 10A) due to excessive acetylcholine neurotransmission(Kim et al., 2004). Hypercontraction is apparent in 100% of dys-1, dyb-1 or stn-1 mutants and occurs in at least 97% of movement cycles in each mutant (n=20 L4 animals, each observed for 50 movement cycles), but is not seen in dgn-1 mutants. Double mutants of dgn-1 with dys-1, dyb-1 or stn-1 show hypercontraction as in the DAPC single mutants (Fig. 10B),indicating that dgn-1(cg121) does not epistatically suppress this phenotype.

DAPC mutants also display hyperactivity on agar plates(Bessou et al., 1998; Gieseler et al., 2001; Grisoni et al., 2003),generating 37%-57% more body bends per minute than wild type(Fig. 10C). The dgn-1(cg121) mutant shows comparable hyperactivity in plate locomotion (Fig. 10C). Both DAPC and dgn-1 mutants also show a 9%-29% increase in thrashing rate when suspended in liquid (Fig. 10C). Double mutants show activity levels comparable to or slightly lower than wild type (Fig. 10C). Cross-suppression of hyperactivity in double mutants is inconsistent with DGN-1 and DAPC acting in a common functional pathway, and indicates that hyperactivity of dgn-1 and DAPC mutants involves genetically distinct mechanisms.

dgn-1 homozygotes (Fig. 10D) show a 27% increase in defecation cycle time(Avery and Thomas, 1997) when compared with wild type (61±6 seconds versus 48±3 seconds). dys-1 mutants display essentially normal periodicity (51±4 seconds), and dys-1;dgn-1 double mutants have periodicity comparable with dgn-1 single mutants (66±4 seconds). Together, these phenotypic comparisons indicate that DGN-1 and the DAPC complex of C. elegans function in different processes and act independently of one another.

The DG-like genes in C. elegans and related nematodes define a family of proteins sharing a core region corresponding to the C terminus of vertebrate α-DG and the extracellular region of β-DG. The vertebrate DG core contains the site of post-translational α/βcleavage and sites within α and β subunits mediating their association (Sciandra et al.,2001). DGN-1 and Drosophila DG(Deng et al., 2003) are not processed into separate α and β subunits, although interaction of the corresponding regions may occur. Interestingly, a mutant DG in whichα/β cleavage is disrupted dominantly produces altered DG glycosylation and muscular dystrophy in mice(Jayasinha et al., 2003),suggesting a regulatory role for vertebrate α/β cleavage. The core region also contains conserved potential N-linked glycosylation sites important in intracellular trafficking of DG(Holt et al., 2000). The core domain may mediate conserved interactions with as yet unidentified extracellular or transmembrane factors.

Distinct N-terminal and C-terminal domains distinguish the members of the DG family. DGN-1, Drosophila and vertebrate DGs comprise an orthologous subfamily with conserved roles as BM receptors(Deng et al., 2003; Parsons et al., 2002; Winder, 2001) (this work). Multiple Caenorhabditis species contain additional species-specific,more divergent, DG-like genes whose roles remain to be elucidated. C. elegans dgn-2 and dgn-3 are expressed in a few neural and epithelial cells, and deletion mutations cause no overt phenotypes (J.M.K.,unpublished). It is possible that the N-terminal and C-terminal domains of these divergent DG-like proteins mediate distinct transmembrane linkages between specific extracellular ligands and intracellular effectors.

BMs form and most are maintained in dgn-1(0) mutants, although the gonad BM is not maintained in the absence of DGN-1. Similarly, vertebrate DG is not generally essential for BM assembly(Li et al., 2002) but may play an important role in some contexts (Henry and Campbell, 1998). Mouse embryos lacking DG do not form Reichert's membrane, but form the epiblast BM in the embryo proper(Williamson et al., 1997). In a brain-specific DG knockout, the pial BM forms but contains focal discontinuities (Moore et al.,2002). Thus, in both nematodes and vertebrates the DG ortholog has roles in the maintenance of specific BMs but is not required for all BM assembly.

The similar gonad primordium defects in epi-1 and dgn-1mutants suggest that DGN-1 is a likely mediator of EPI-1 function in the early gonad. The single C. elegans β integrin PAT-3 is another potential laminin receptor, but dominant interference of PAT-3 function does not appear to produce the same early gonad phenotype(Lee et al., 2001). DGN-1 is not required for the initial localization of laminin to the surface of the primordium, indicating that other factors mediate laminin recruitment. EPI-1 functioning through DGN-1 appears to be essential for promoting the epithelial function of the SGPs, although the nature of this nascent epithelium is unclear. Somatic cells of the early gonad do not display junctional complexes indicative of mature polarized epithelia and do not express identified junctional components (Miskowski et al.,2001). A BM signal through DGN-1 may play a role in the apicobasal polarization of somatic gonad cells in the early gonad, before a mature epithelium has differentiated.

Early gonad disruption in dgn-1 and epi-1 mutants results in the escape of germ cell precursors into the body cavity. The gonad BM may be important structurally to maintain ensheathment by the somatic gonad cells,but it is also possible that an unidentified DGN-1-mediated signal from the gonad BM promotes germ-soma interaction. DGN-1 function is not required for the adhesion of somatic gonad cells with one another. Somatic gonad cells cluster together in dgn-1(0) L2 stage animals, and differentiated spermathecal and uterine cells displaying AJM-1-containing cell junctions and lumen formation are found in L4 stage animals.

Migrations of axons and the tubular arms of the excretory cell involve the extension of cell processes between the hypodermis and its BM(Wood, 1988), and several cell adhesion and cytoskeletal factors, including laminin, are involved in their guidance (Buechner, 2002). The excretory cell and axonal guidance phenotypes of dgn-1 and epi-1 mutants suggest that DGN-1 partially mediates laminin function in guiding cell processes along the hypodermal BM.

Heterozygous dgn-1(0)/+ animals show defects in excretory cell and axon migration, as well as in DTC guidance along BMs. DGN-1 may have an essential role in DTC migration that cannot be assessed directly because of the gonad disruption in dgn-1(0) homozygotes. The defects in excretory cell, DTC and axon migration in cg121/+ heterozygotes must represent haploinsufficiency, in which the reduced DGN-1 levels in heterozygotes are insufficient for normal function. DGN-1 overexpression from extrachromosomal arrays can also cause dgn-1(0)-like defects in wild-type animals, particularly in excretory cell morphology (R.P.J.,unpublished). The appearance of similar phenotypes from increased or decreased expression suggests that DGN-1 activity may be required dynamically and/or in precise stoichiometry relative to other components for normal function.

DGN-1 functions in a variety of epithelia and neurons but is not expressed in muscle. Vertebrate DG also functions in the nervous system(Moore et al., 2002; Saito et al., 2003) and in at least some types of epithelia (Durbeej et al., 1995; Durbeej et al.,2001), as well as in muscle. In Drosophila, DG is important in the polarization of follicular epithelia and oocytes(Deng et al., 2003), although it is expressed and may have additional roles in muscle and neural tissue(Dekkers et al., 2004). These findings suggest that either the DGN-1/DG subfamily originated as an ECM receptor in epithelial/neural tissue and that muscle function was acquired later, or that an original muscle role for DG was not retained in nematodes. Notably, each sarcomere in nematode muscle is anchored to the underlying BM via integrin-containing dense bodies(Moerman and Fire, 1997), and these numerous attachments may preclude the need for further sarcolemmal stabilization by DGN-1.

DGN-1 functions in epithelia and neurons do not depend on the conserved DAPC complex, which is consistent with the poor conservation of the dystrophin-binding site in the DGN-1 cytoplasmic domain. The major site of DAPC function is in muscle, where it regulates contraction intensity via the acetylcholine transporter SNF-6 (Bessou et al., 1998; Gieseler et al.,2001; Grisoni et al.,2003; Kim et al.,2004). It is unclear whether the hyperactivity of DAPC mutants(Bessou et al., 1998; Gieseler et al., 2001; Grisoni et al., 2003) is also due to altered cholinergic stimulation. dgn-1 mutants show similar hyperactivity that is genetically distinct from that of the DAPC mutants and may reflect neuronal DGN-1 function.

The divergence of DGN-1 and DAPC complex function in C. eleganshas intriguing implications for the normal and pathological roles of their vertebrate homologs. The progressive muscle degeneration in C. elegans DAPC and snf-6 mutants is reminiscent of vertebrate muscular dystrophy (Bessou et al.,1998; Gieseler et al.,2001; Grisoni et al.,2003; Kim et al.,2004; Cohn and Campbell,2000) and may reflect a conserved muscle function of the DAPC that is separable from its interaction with DG. Conversely, evidence from vertebrate systems suggests that non-muscle DG may have roles not requiring DAPC function. Several studies indicate that vertebrate DG has important functions in the early embryo and in non-muscle tissues that do not appear to depend on DAPC or sarcoglycan complex function, but do depend on the ECM ligand-binding activity of α-DG(Durbeej et al., 1995; Durbeej et al., 2001; Michele et al., 2002; Moore et al., 2002; Rafael et al., 1999; Imamura et al., 2000; Saito et al., 2003; Williamson et al., 1997). Thus, non-muscle DG probably mediates important ECM interactions that are wholly or partly independent of DAPC function, suggesting that other intracellular factors transduce DG function in these contexts. The DAPC-independent roles of DGN-1 in C. elegans may thus help to elucidate conserved non-muscle roles of DG as a BM receptor and to identify novel downstream partners of DG.

We thank Y. Kohara for providing dgn-1 cDNA clones. Some strains used in this work were provided by the C. elegans Genetics Center,which is funded by the NIH National Center for Research Resources (NCRR). This work was supported by NIH grant HD27211.