Positioning of Longitudinal Nerves in C. elegans by Nidogen

See allHide authors and affiliations

Science  07 Apr 2000:
Vol. 288, Issue 5463, pp. 150-154
DOI: 10.1126/science.288.5463.150


Basement membranes can help determine pathways of migrating axons. Although members of the nidogen (entactin) protein family are structural components of basement membranes, we find that nidogen is not required for basement membrane assembly in the nematodeCaenorhabditis elegans. Nidogen is localized to body wall basement membranes and is required to direct longitudinal nerves dorsoventrally and to direct axons at the midlines. By examining migration of a single axon in vivo, we show that nidogen is required for the axon to switch from circumferential to longitudinal migration. Specialized basement membranes may thus regulate nerve position.

The nervous system is organized into longitudinal and circumferential projections. As the axon scaffold forms, migrating growth cones can switch between circumferential and longitudinal directions. Members of several protein families, including netrin, semaphorin, ephrin, and slit, act as attractive or repulsive guidance molecules to direct axons to their targets (1). Changes in the responsiveness of migrating axons to such guidance molecules could in part underlie the ability of axons to migrate in new directions. Also important for determining where pioneering axons migrate are the physical substrates that support the nervous system, including the basement membranes.

In order to identify genes required for the placement of nerves, we selected mutations that affect the position of nerves, but do not affect axon outgrowth. From a clonal F2 screen of approximately 6000 haploid genomes treated with the mutagen ethylmethanesulfonate, we identified 11 mutations that define several loci (2). We chose one allele, ur41, that disrupts the nid-1 locus and is a loss-of-function allele, causing nerves to form at the wrong positions. Most neural circuits in animals with this mutation are apparently formed normally, as the animals show wild-type behavior instead of the uncoordinated (Unc) phenotype. In general, morphogenesis is normal, and the animals do not have traits that are sometimes associated with cell migration defects. For example, they do not have a dumpy (Dpy) appearance, they are not egg-laying defective (Egl), nor do they have misformed gonads.

In more than 90% of the nid-1(ur41) mutants, the dorsal sublateral nerves are mispositioned to the dorsal midline (Fig. 1, A and B, and Table 1A). Although mispositioned, the axons appear normal, and they form normal bundles (Fig. 2B). We also observe that in 32% of the mutants (n = 100) the ventral nerve axons overpopulate the left fascicle of the ventral nerve cord and underpopulate the right fascicle (Fig. 1, C and D). Thus, instead of the normal asymmetric phenotype of the C. elegans ventral nerve cord, the cord is nearly symmetric (Fig. 2A). We examined the phenotypes ofnid-1(ur41) animals in greater detail (3). Besides defects in controlling the position of specific nerves, we did not observe other defects by electron microscopy. The tissues and basement membranes that support the nervous system are essentially normal (Fig. 2C). Moreover, axons that are not affected by the nid-1(ur41) mutation cross without difficulty the same surfaces that affected axons cross. These observations suggest that the altered positions of nerves are not a secondary consequence of an extensively disrupted substratum.

Figure 1

Effects of nid-1 on positioning longitudinal nerves and axons. (A, C, E, G, I, and K) Wild-type C. elegans nematodes; (B, D, F, H, J, and L)nid-1(ur41) mutant nematodes. (A andB) The dorsal sublateral nerve (dsl) becomes mispostioned along the dorsal nerve cord (dc) in the mutant. Neurons are identified by GFP fluorescence to show expression of anunc-119::GFP transgene as a pan-neural marker. Lateral aspect. (C and D) Compared with the wild type, in the mutant there are more ventral nerve cord axons in the left fascicle (small arrowhead) and fewer in the right (large arrowhead). Neurons are visualized by pan-neuralunc-119::GFP expression. Ventral aspect. (E and F) Interneuron axons that exit from the head on the left crossover (arrow) to the right fascicle (large arrowhead) at the anterior end of the ventral nerve cord. In the mutant, the axons cross at multiple positions (arrows) and sometime remain in the left fascicle for some distance (small arrowhead). Neurons are identified by expression of glr-1::GFP,a marker for interneurons of the right fascicle. Ventral aspect. (G and H) In the wild type, the PVQR axon is in the right fascicle (large arrowhead) and PVQL is in the left fascicle (small arrowhead). In the mutant, PVQL crosses to the right fascicle. These axons flank the vulva (v). Neurons are identified by expression of sra-6::GFP, a marker for PVQ in the posterior ventral nerve cord. Ventral aspect. (I andJ) In the wild type, the longitudinal processes of the motor neurons are in the right fascicle (large arrowhead); whereas in the mutant, the processes also populate the left fascicle (small arrowhead). Circumferential processes are normal. Neurons are identified by expression of a unc-129::GFPtransgene, a marker for DA and DB motor neurons. Ventral aspect. (K and L) To form the dorsal nerve cord (large arrowhead), axons from the ventral motor neurons migrate along both sides of the body to the dorsal midline. The cord forms along the left side of the midline. In the mutant, axons that migrate dorsally on the right side bifurcate and migrate along the right side of the midline (small arrowhead) before crossing over. Neurons are identified by expression of theunc-129::GFP transgene. Dorsal aspect. Anterior is to the left in all micrographs.

Figure 2

Electron micrographs of nid-1(ur41)larva. (A) Cross section through the ventral epidermal ridge. In the wild type, there are approximately 54 axons in the right fascicle and 6 in the left fascicle (28), in the mutant there are, on average, 25 ± 2 axons in the left and 29 ± 3 in the right (n = 3). Scale bar, 500 nm. (B) Cross section through the dorsal epidermal ridge. The dorsal nerve cord (dc) is positioned along the left side of the epidermal (e) ridge. A mispositioned dorsal sublateral nerves (dsl) is positioned along the right side of the ridge. Axon morphology is normal. Scale bar, 500 nm. (C) As in the wild-type animals, intact basement membranes (between arrow heads) assemble at tissues, including the intestine (int) and epidermis (e). Scale bar, 100 nm.

Table 1

The nid-1 gene is required to position nerves dorsoventrally. Schematic drawings show the midlines as gray lines and axons as solid lines. Anterior is to the left. Thegfp transgenes used to visualize the individual neurons are given in Methods.

View this table:

We observed individual classes of ventral neurons marked with green fluorescent protein (GFP) reporters in thenid-1(ur41) mutant. In wild-type animals, ventral nerve cord interneurons contribute to the normal asymmetry of the cord, as can be seen in a bundle of interneuron axons that exits the nerve ring from the left side and then crosses over to the right fascicle (Fig. 1, E and F, and Table 1C). In nid-1(ur41), the right tract is correctly established, but the crossing of the axon bundle from the left to right side is impaired. The axons still cross but not in the appropriate place, and they cross in small bundles or as individual axons. The PVQ axons were also examined because they help to initiate the extension of the ventral nerve cord from posterior ganglia (4). In nid-1(ur41), the left PVQ axon often fails to initiate the left tract and instead crosses the midline and travels with the right PVQ axon in the right fascicle (Fig. 1, G and H, and Table 1B). The migration of the right PVQ axon is not affected. We also examined the postembryonic HSN neurons, because HSN axon migration to the ventral nerve cord may require PVQ axons (5). We find that, like the PVQ axons, HSNL, but not HSNR, axon migrations are defective in nid-1(ur41) (6). These phenotypes suggest that there is a defect associated with axons approaching the ventral midline from the left side.

In the nid-1(ur41) mutant, longitudinal motor neuron axons that should be in the right fascicle are instead in the left. The motor neurons are arranged in a line at the ventral midline with the two fascicles of the ventral nerve cord flanking the cell bodies (Fig. 1C). Some motor neurons have two processes that grow out in different directions. In wild-type animals, one process joins the right fascicle; the other extends circumferentially to the dorsal midline. In the nid-1 mutant, the circumferential processes are normal, but the longitudinal processes frequently join the left fascicle (Fig. 1, I and J, and Table 1D). This abnormality accounts for the overpopulated left fascicle of the ventral cord, and it suggests that these axons initially fail to distinguish between the left and right sides of the ventral midline. Once the axons adopt a side, however, they apparently respect the midline, for they do not repeatedly cross the midline, and the two fascicles remain apart.

In nid-1(ur41), the dorsal nerve cord has a split appearance. In wild-type animals, the dorsal nerve cord is formed by axons from the ventral midline motor neurons that migrate to the dorsal midline, bifurcate, and extend processes longitudinally along the left side of the midline. Motor neuron axons that migrate circumferentially along the right body wall cross the dorsal midline to join the nerve. In nid-1(ur41) mutants, these motor neuron axons bifurcate and migrate longitudinally on the right side of the midline before crossing (Fig. 1, K and L, and Table 1E). Those that reach the tract from the left side appear normal. These results suggest that axons from the right side are not effectively guided across the midline.

Because the axon migration phenotypes at the midline show left versus right differences, we examined other features of the animal that are asymmetric, and we conclude that the left and right differences innid-1(ur41) are confined to the axons at the midline. For example, the stereotyped side that each motor neuron axon and distal tip cell circumferentially migrates across is preserved innid-1(ur41). Also, the anterior migration of the right Q cell and posterior migration of the left Q cell are normal. These results indicate that nid-1 does not affect the overall handedness of the animal.

To identify the protein product of ur41, we mapped the mutation to a region that is one-half map unit in size (Fig. 3A). Using cosmids that corresponded to the region, we cloned the ur41 locus by DNA-mediated transformation rescue of the ur41 dorsal sublateral nerve and ventral nerve cord phenotypes. One cosmid, F54F3, was capable of rescuing the phenotypes. DNA sequence analysis predicts that this cosmid contains the gene that encodes the single C. eleganshomolog of nidogen (entactin) (C. elegans Genome Sequencing Consortium, Fig. 3B). We found that a genomic fragment, enhanced by the polymerase chain reaction (PCR), that contains the predicted nidogen (entactin) coding sequences and 2.5 kb of 5′ flanking sequence can rescue the ur41 phenotypes. Furthermore, animals with the same phenotype as ur41 mutants are observed when loss of nid-1 function is phenocopied by RNA-mediated interference (7). DNA sequence analysis ofur41 identifies a nonsense point mutation introducing a stop in the first third of the predicted protein (8) (Fig. 3B). This is consistent with having the allele behave genetically as a null, that is, the mutant phenotypes are not more severe in animals withur41 in trans to a genetic deficiency that eliminates the nid-1 locus than in ur41homozygous animals.

Figure 3

The nid-1 gene and protein. (A) Molecular cloning. The nid-1 gene maps between lin-25 and odr-3 of linkage group V. The cosmid F54F3 rescues nid-1(ur41) as does an 11-kb PCR fragment. The ur41 mutation introduces aCAA→TAA change in exon 8. Black boxes are exons. (B) Diagram illustrating the domain structures of the C. elegans and human nidogen family members. In C. elegans the motifs are the N1 (amino acids 1–684), the EGF-like (amino acids 685–1320 and 1605–1638), and the low density lipoprotein (LDL)–receptor YWTD (amino acids 1373–1550). The arrow indicates the ur41 mutation, Q550 → stop.

Nidogen (entactin) is a basement membrane component that is highly conserved among human, Drosophila, C. elegans, and ascidians (9). The protein comprises three globular domains (G1, G2, and G3) connected by a flexible linker and a rod-like domain that includes epidermal growth factor (EGF)–like modules (10, 11). It is proposed that nidogen-1 connects laminin and collagen IV networks together to form stable basement membranes (9–11). Vertebrate nidogen promotes cell migrations in vitro, and antibodies that block the laminin-nidogen interaction disrupt tissue morphogenesis (12–15). Although models predict that nidogen is a structural component of basement membranes, basement membranes assemble normally, and in general, tissues and their basement membranes appear to develop normally in theC. elegans nidogen mutant.

Larvae stained with anti-nidogen antiserum show that nidogen is localized to body wall basement membranes (16). Intense staining is associated with the basement membrane of the body wall muscle (Fig. 4, A and B). Staining is not detected in nid-1(ur41) animals [Web fig. 1 (17)]. Nidogen can bind collagen type IV, fibulins, laminin, and perlecan (10,18, 19). It is interesting that both collagen type IV and perlecan in C. elegans are selectively distributed to the basement membranes between the epidermis and body wall muscles (20, 21). Each sublateral nerve runs longitudinally along the center margin of a muscle quadrant, sandwiched between the basement membrane and the epidermis; nerve cords run longitudinally along the edges of the interface between muscle and basement membranes and epidermis. Possibly nidogen is incorporated differently into this membrane and has distinct regulatory functions at these sites. This would be consistent with a model that multiple guidance cues, distributed in gradients and associated with the different tissues, and specialized basement membranes, form unique combinations at each dorsoventral position to specify where the different longitudinal nerves form (22, 23).

Figure 4

Nidogen is associated with body wall basement membranes. (A) Larva stained to reveal anti-nidogen antibodies. Intense staining is associated with the basement membrane of the body wall muscle. (B) Muscle cells in the same larva visualized by using an anti-UNC-54 myosin monoclonal antibody. The muscle cells form four sublateral rows extending the length of the body. One muscle quadrant is in the plane of view. Scale bar, 10 μm.

To examine the expression pattern of nid-1, we used in situ hybridization analysis and GFP expression under the control of thenid-1 promoter sequence (24). Expression of the GFP transcription reporter is first detected in late gastrulation by the cephalic, inner labial, and outer labial cells (Fig. 5). As the embryo elongates and morphogenesis begins, expression in body wall muscle cells is detected by in situ hybridization [Web fig. 2 (17)] and is observed by using the GFP reporter. In addition, GFP is observed in the left and right lateral ALM neurons and the anal depressor and intestinal muscle cells. As embryogenesis continues, expression in the body wall muscle declines and is not observed in the larval stages. During the larval and adult stages, GFP expression by the PLM neurons, the intestinal cells, and the distal tip cells of the gonad is observed. In addition, transient expression is observed in HSN neurons during the L3 and L4 stages and in ventral nerve cord neurons during the early L2 stage. Although these observations are limited by the sensitivity of in situ hybridization and the ability of the nid-1 promoter sequence to reproduce the wild-type nid-1 expression pattern, they suggest that the nid-1 gene is not expressed by the axons affected in the nid-1(ur41) mutant.

Figure 5

Expression of the nid-1 gene detected by using the nid-1 promoter to drive GFP expression. Embryos (A, B, C, D, and E) and larvae (F, G, H, and I). (A and B) Shortly before (A) and at the beginning (B) of embryonic elongation, expression is detected in the Cep (cephalic), IL (inner labial) and OL (outer labial) neurons (arrows). (C, D, and E) During succeeding stages of elongation until hatching, expression is detected in body wall muscle (arrowheads), the intestinal and anal depressor muscle cells (*), and the ALM neurons. (F, G, H, and I) In the larva, expression is detected in the head Ceh, IL, and OL neurons (F), the excretory cannel cell (exc) and anterior intestine (int) (G), the distal tip cells (DTC) and the HSN neurons (H), and posteriorly in the PLM neurons and the anal depressor and intestinal muscle cells (mu int and mu anal) (I).

Because nid-1 encodes a secreted matrix protein and is required to position axons, we asked whether it might have activities similar to those of known extracellular guidance molecules. We expressed nid-1 in neurons, intestinal cells, and body wall muscles to determine how ectopic expression affects nerve development (24). Staining of these strains with the anti-nidogen antiserum shows wild-type protein distribution (25), suggesting that the incorporation of nidogen is specific to the local assembly of the body wall basement membranes. We find that ectopic expression from any source rescues the nid-1mutant phenotypes, causing the nerves to be correctly positioned. In contrast, ectopic expression of netrin UNC-6 or transforming growth factor–β UNC-129 causes anomalous axon migrations that are consistent with the idea that the distributions of attractant and repellent guidance molecules are altered by the ectopic expression (22, 23, 26). Together these observations indicate that nidogen and chemotropic guidance factors function differently to position nerves.

We used the migration of the SDQR axon to study the positioning of dorsal sublateral axons. In the first larval stage, the SDQR axon is repelled dorsally from ventral netrin UNC-6 sources until it reaches the dorsal sublateral position. There, the responsiveness to UNC-6 changes, and the axon migrates anteriorly (22). Inunc-6 mutants, SDQR migrates ventrally, presumably in response to some other cues that are unmasked by the absence of UNC-6. We now find that nid-1 is also required to position SDQR dorsoventrally. However, although unc-6 is required for the SDQR axon to be directed dorsally, nid-1 is necessary to prevent the axon from migrating past the dorsal sublateral position to the dorsal midline. In 54% of the nid-1(ur41) mutants (n = 100), the SDQR axon migrates to the dorsal midline (Table 2). This phenotype is never observed in unc-6 mutants or when unc-6 is ectopically expressed (22, 23). We also note that the SDQR axon, as well as other circumferentially migrating axons, migrate directly across the dorsal sublateral basement membrane in thenid-1(ur41) mutant, indicating that the lack of nidogen has not affected the ability of axons to migrate across this surface.

Table 2

The unc-40 gene is epistatic tonid-1 for SDQR axon positioning. Schematic drawings show the midlines as gray lines and axons as solid lines. Filled boxes represent basement membranes where nidogen is localized. Not shown are SDQR axons that joined other lateral longitudinal nerves. The SDQR cell body is sometimes positioned further ventrally in unc-5,unc-6, and unc-40 mutants (22). Axon migrations were scored from cell bodies in either position. Anterior is to the left. The unc-119::gfp transgene was used to visualize the SDQR neurons. Alleles scored were nid-1(ur41)and unc-40(e1430).

View this table:

Because the netrin receptor genes, unc-5 andunc-40, also affect SDQR migrations, we examined the relationship between nid-1 and these genes. In unc-5, unc-6, or unc-40 mutants, the SDQR axon never migrates past the dorsal sublateral position to the dorsal midline [n = 100 for each; see also (22)]. However, in a few unc-5; nid-1 and nid-1; unc-6 double mutants where the SDQR axon does reach the dorsal region, the axon will migrate to the dorsal midline (for unc-5; nid-1 mutants, 5% turn at the dorsal sublateral tract and 3% migrate to the dorsal midline and for nid-1; unc-6 mutants, 5% turn at the dorsal sublateral tract and 2% migrate to the dorsal midline; n = 100 for each). In contrast, the SDQR axon never migrates to the dorsal midline in unc-40; nid-1 mutants (Table 2), but instead is guided anteriorly at the dorsal sublateral position (65%, n= 100). This migration again indicates that this basement membrane, even when lacking nidogen, can support axon migrations. Furthermore, this shows that unc-40 is epistatic to nid-1, that is, the double mutants display the unc-40 phenotype instead of the nid-1 phenotype, indicating that these genes function together to direct the circumferential to longitudinal turn of the SDQR axon. This unexpected relationship suggests that nidogen has an affect on UNC-40–mediated signaling.

Our results show that nidogen affects the switch from circumferential to longitudinal axon migration and that it helps determine where longitudinal nerves form along the dorsoventral axis. We propose that nidogen alters the SDQR axon response to guidance cues at the dorsal sublateral position. Nidogen or a nidogen/laminin complex could act directly as a ligand for axonal guidance receptors, including UNC-40. This would be consistent with the recent description that extracellular matrix molecules can in vitro modify the behavior of growth cones in response to netrin-1 (27). It is also possible that nidogen creates a different basement membrane configuration that allows new interactions between the migrating axon and the basement membrane of the body wall muscle. To direct the switch from circumferential to longitudinal migration, nidogen may cause a weaker response to UNC-40–mediated circumferential guidance signals so that the axon can then be guided by longitudinal guidance cues. Alternatively, nidogen may be required for the axon to have a response to longitudinal guidance cues that is strong enough to overcome the effects of UNC-40–mediated signals at the dorsal sublateral position. At the dorsal and ventral midlines in the embryo, the nerve cords form along the edge of body wall muscle basement membranes and nidogen at this interface could influence midline crossing by affecting axon responses to midline guidance cues.

  • * To whom correspondence should be addressed. E-mail: wadswort{at}


View Abstract

Stay Connected to Science

Navigate This Article