In C. elegans, cells of the QL and QR neuroblast lineages migrate with left-right asymmetry; QL and its descendants migrate posteriorly whereas QR and its descendants migrate anteriorly. One key step in generating this asymmetry is the expression of the Hox gene mab-5 in the QL descendants but not in the QR descendants. This asymmetry appears to be coupled to the asymmetric polarizations and movements of QL and QR as they migrate and relies on an asymmetric response to an EGL-20/Wnt signal. To identify genes involved in these complex layers of regulation and to isolate targets of mab-5 that direct posterior migrations, we screened visually for mutants with cell migration defects in the QL and QR lineages. Here, we describe a set of new mutants (qid-5, qid-6, qid-7, and qid-8) that primarily disrupt the migrations of the QL descendants. Most of these mutants were defective in mab-5 expression in the QL lineage and might identify genes that interact directly or indirectly with the EGL-20/Wnt signaling pathway.
HOW migratory cells select one of many possible migratory programs to adopt a specific trajectory is a central question in developmental biology. In Caenorhabditis elegans, the QL and QR neuroblasts are left-right homologs that divide identically to give rise to the same neuron types but migrate in opposite directions (Sulston and Horvitz 1977; Figure 1). On the left, QL and its descendants migrate posteriorly, whereas on the right, QR and its descendants migrate anteriorly. A key step in patterning this left-right asymmetric migration is the expression of the Hox gene mab-5 specifically in the QL descendants and not in the QR descendants. Thus, this system represents a unique opportunity to study how left-right asymmetry and Hox gene regulation intersect in the context of cell migration.
The first step in these migrations involves the polarizations and movements of the QL and QR neuroblasts over a short distance: QL moves dorsally and to the posterior, whereas QR moves dorsally and to the anterior. These migrations are the first visible manifestations of left-right asymmetry between these cell lineages. UNC-40 (the C. elegans netrin receptor; Chanet al. 1996) and DPY-19 (a novel multipass transmembrane protein) are required to ensure that the QL and QR neuroblasts polarize in the correct direction during these migrations (Honigberg and Kenyon 2000).
The QL-specific expression of mab-5 begins during QL’s migration. Correct mab-5 expression may be dependent on the proper polarizations and movements of the Q neuroblasts, because in unc-40 and dpy-19 mutants, QL and QR not only polarize randomly but also express mab-5 randomly (Honigberg and Kenyon 2000). After their migrations, QL and QR divide. mab-5 is expressed at high levels in the QL descendants and is not expressed in the QR descendants (Figure 1B). mab-5 specifies a posterior program of cell migration; in mab-5(-) mutants, the descendants of QL migrate anteriorly like the descendants of QR (Kenyon 1986; Salser and Kenyon 1992; Harriset al. 1996). In contrast, in wild-type animals, the descendants of QR do not turn on mab-5 and therefore migrate anteriorly (Salser and Kenyon 1992). If mab-5 is inappropriately expressed in the QR descendants, they migrate posteriorly (Salser and Kenyon 1992), reinforcing the idea that the expression of mab-5 acts as a critical switch between anterior and posterior migration.
This left-right asymmetry in mab-5 expression is due to a left-right asymmetric response to a C. elegans Wnt signal encoded by the egl-20 gene (Whangbo and Kenyon 1999). The QL descendants are more sensitive to the EGL-20/Wnt signal and respond to this signal by turning on mab-5 expression through the activity of a conserved, canonical Wnt signal transduction pathway (Harriset al. 1996; Sawaet al. 1996; Eisenmannet al. 1998; Maloofet al. 1999; Herman 2001; Korswagenet al. 2002). In contrast, the QR descendants are less sensitive to the EGL-20/Wnt signal and respond by migrating anteriorly (Whangbo and Kenyon 1999). The alternate signaling pathway that mediates this second EGL-20/Wnt response is not well understood. Dose-response experiments indicate that high levels of EGL-20/Wnt activate the canonical Wnt pathway, while low levels signal through the alternate Wnt pathway (Whangbo and Kenyon 1999). By utilizing different EGL-20/Wnt-dependent pathways that specify different migratory responses in the QL and QR lineages, the initial left-right asymmetry between these cells is maintained.
Many molecular details need to be elucidated to fully understand the complex but precise migrations of the cells in the QL and QR lineages. How the migrations of the QL and QR descendants are specified is unknown, as is the question of why cells of the QL lineage are more sensitive to given levels of EGL-20/Wnt than are cells of the QR lineage. How mab-5 expression might be coupled to the left-right asymmetric polarizations of the QL and QR cells is also a mystery. Whether mab-5 expression is targeted to the QL lineage by dedicated pathway(s) or is regulated as part of a process that specifies the entire region-specific pattern of mab-5 expression in the posterior is also poorly understood. Finally, the genes that act downstream of mab-5 to cause the descendants of QL to migrate posteriorly still remain unidentified.
To begin to address these questions, we designed a genetic screen to identify new genes required for these cell migrations. Here, we describe the set of mutants, qid-5, qid-6, qid-7, and qid-8 (for Q is defective), that primarily disrupt the migrations of cells of the QL lineage. Characterization of these qid mutants indicated that we have identified genes affecting the expression of mab-5 in the QL lineage.
MATERIALS AND METHODS
General methods and strains: Strains were cultured using standard methods (Brenner 1974; Wood 1988; Lewis and Fleming 1995). All strains were maintained at 20°, except where indicated. To generate males, we used the him-5(e1490) mutation; this mutation does not affect larval development.
In addition to the mutants isolated in the screen described in Table 1, the following alleles/strains were used in this study and are listed according to linkage group (LG); all alleles are described in Hodgkin (1997) except where otherwise referenced:
LG I: mig-1(e1787), lin-17(n671)
LG II: mig-14(mu71)
LG III: mab-5(e2088)
LG IV: egl-20(n585)
LG V: him-5(e1490)
LG X: mig-13(mu31), bar-1(ga80) (Eisenmannet al. 1998).
The following transgenic strains were used in this study:
muIs3[mab-5::lacZ,rol-6(d)] on LG V (Cowing and Kenyon 1992)
muIs4[mab-5::lacZ, rol-6(d)] on LG I (Cowing and Kenyon 1992)
muIs32[mec-7::GFP, lin-15(+)] on LG II
muIs35[mec-7::GFP, lin-15(+)] on LG V.
The chromosomal insertions, muIs32 and muIs35, were generated by γ-irradiating lin-15(n765ts) animals bearing an extrachromosomal array containing mec-7::GFP and lin-15(+) (a kind gift from Lindsay Hinck and the laboratories of Cori Bargmann and Marc Tessier-Lavigne) as late L4 or young adults. At 25°, lin-15(n765ts) animals have a multivulva (Muv) phenotype that is rescued by a genomic clone of the lin-15 locus in this extrachromosomal array. We screened the F2 generation of these irradiated animals for lines that did not give any Muv progeny at 25°, indicating that the extrachromosomal array was stably inherited. These lines bearing putative chromosomal insertions of the extrachromosomal array were outcrossed three times to the wild-type N2 strain to remove lin-15(n765ts) and other possible mutations generated during the irradiation process. With the exception of a slightly longer generation time, animals bearing muIs32 or muIs35 did not exhibit any obvious developmental defects and had normal neuronal cell migrations (Figures 2 and 3; data not shown).
Mutagenesis and screening: Mutagenesis was performed essentially as described previously (Anderson 1995). Briefly, 15 separate pools of late L4 or young adult P0 animals bearing either muIs32 or muIs35 were mutagenized with 25 mm ethyl methanesulfonate (EMS) for 4 hr at 20°. These animals were then washed several times in M9 to remove the EMS. Their F1 progeny were collected at hatching and stored at 8° in M9, which arrested their development. Batches of these arrested F1’s were plated out over several days and cultured at 20° where they resumed development. These F1’s were allowed to self-fertilize to generate F2 progeny. These F2 progeny were collected and aged until many had reached the L4 larvae or young adult stage. At this point, they were screened for misplaced QL and/or QR descendants (see Figure 2C); ∼100,000 F2 animals were screened. These positive F2’s were placed individually on culture dishes and their F3 progeny were retested; viable lines that still exhibited migration defects in the QR and/or QR lineage were kept.
During the screening process, we used other mec-7::GFP positive cells as landmarks for determining the anterior/posterior orientation of the animal (Figure 2, B and C; Chalfieet al. 1994). These include the two posterior lateral microtubule (PLM) neurons in the tail and the two anterior lateral microtubule (ALM) neurons near the head. By locating the green fluorescent protein (GFP)-expressing posterior ventral microtubule (PVM) and/or anterior ventral microtubule (AVM) neurons in relation to these GFP-expressing landmark cells, we could easily and rapidly determine whether PVM and/or AVM had been misplaced (up to 1000-1200 animals could be examined within 1 hr). Since mec-7 encodes a mechanosensory neuron-specific β-tubulin, its expression also served as a marker for mechanosensory fate (Savageet al. 1989). This allowed us to eliminate mutations that disrupt cell fate determination or differentiation in the QL and QR lineages.
We screened populations of young adults long after the cell migrations had been completed by the end of the first larval stage. This permitted growth in the intervening larval stages to amplify the distance between wild-type and mutant positions by almost 10-fold. Picking mutants as adults also improved the recovery rate of candidate mutations.
Genetic mapping: Mapping against a standard panel of Tc1 markers in the RW7000 strain [hP4 (I), maP1 (II), mgP21 (III), sP4 (IV), bP1 (V)] was performed as described previously (Williamset al. 1992; Williams 1995). muIs32 was mapped against stP100, stP196, stP101, stP50, stP36, stP98, and maP1 to ∼2 cM to the right of maP1 on LG II, placing it at a map position of ∼+6.3. muIs35 was mapped against stP3, stP192, stP23, bP1, stP6, stP18, stP108, stP105, and stP128 to <1 cM from stP6 on LG V at the map position of +5.9.
The mutations isolated in the screen were mapped in a similar fashion. Mutant hermaphrodites were crossed to wild-type males and the hermaphrodite progeny (F1’s) were examined for phenotypes to determine if the mutation were dominant. All the mutations we examined were recessive for the cell migration phenotype in the QL lineage. This enabled us to determine if the mutation was X-linked by examining the male F1 progeny. If the mutation was autosomal, these heterozygous F1 males were mated to the RW7000 mapping strain. The F2 progeny from this cross were picked onto individual plates and their homozygous mutant F3 progeny were isolated from some of these plates. DNA was isolated from these mutants or from their entire set of progeny.
Further mapping was conducted using Tc1 markers linked to the respective mutation (listed from left to right on their respective linkage groups). qid-5(mu245) was mapped with stP100, stP196, stP101, stP50, stP36, stP98, and maP1 on LG II. qid-6(mu252) was mapped with stP19, stP120, mgP21, stP127, and stP17 on LG III. qid-7(mu327) and qid-8(mu342) were mapped with stP41, stP40, stP156, stP33, stP103, stP129, stP61, stP72, and stP2 on LG X. All recombinants obtained are listed below along with their frequency of occurrence and are designated by the Tc1 markers present in the recombinant chromosome that also carries the qid mutation.
qid-5(mu245): The following recombinants placed qid-5(mu245) on the right of maP1: stP100, stP196, stP101, stP50, stP36, stP98, maP1 (1/17); stP100, stP196, stP101, stP50 (1/17); stP100, stP196 (4/17); stP100 (6/17). Of the 17 qid-5 animals obtained, 5 were nonrecombinants.
qid-6(mu252): The following recombinants placed qid-6(mu252) on the left of stP17: stP17 (4/18). The following double recombinant placed qid-6(mu252) between stP19 and stP17: stP19 and stP17 (1/18). Of the 18 qid-6 animals, 13 were nonrecombinants. No recombination (0/18) was observed between qid-6 and the markers in the cluster (stP120, mgP21, stP127).
qid-7(mu327): The following recombinants placed qid-7(mu327) on the right of stP103: stP41, stP40, stP156, stP33, stP103 (2/27); stP41, stP40, stP156, stP33 (3/27); stP41, stP40, stP156 (2/27); stP41, stP40 (3/27); stP41 (1/27). The following recombinants placed qid-7(mu327) on the left of stP61: stP2 (3/27); stP61, stP72, stP2 (3/27). These double recombinants were obtained: stP41, stP40, and stP2 (3/27); stP41 and stP2 (1/27). Of the 27 qid-7 animals obtained, 6 were nonrecombinants.
qid-8(mu342): The following recombinants placed qid-8(mu342) on the right of stP129: stP41, stP40, stP156, stP33, stP103, stP129 (2/23); stP41, stP40, stP156 (1/23); stP41, stP40 (2/23); stP41 (3/23). The following recombinants placed qid-8(mu342) on the left of stP61: stP2 (7/23); stP61, stP7, stP2 (1/23). One double recombinant was obtained: stP41, stP40 and stP61, stP7, stP2 (1/23). Of the 23 qid-8 animals obtained, 6 were nonrecombinants.
During the mapping process, we retained the mec-7::GFP marker (muIs32 or muIs35) to permit detection of cell migration defects in the QL lineage. Hence the mapping process would indicate linkage to these chromosomal insertions as well as the mutation. Since we knew the chromosomal location of the mec-7::GFP marker, we could infer the location of the mutation. A small number of the mutations showed linkage to the mec-7::GFP marker; in these cases, the mapping process showed only linkage to the one chromosome bearing both the mutation and the mec-7::GFP marker.
Complementation tests: Complementation tests were performed when two or more mutants with similar cell migration phenotypes mapped to the same linkage group. These tests were conducted as follows. For autosomal mutations in which the males were capable of mating, spontaneous males bearing the mutation of interest and the mec-7::GFP marker were crossed into hermaphrodites bearing a known mutation but no mec-7::GFP marker. Cross-progeny were identified by the presence of the mec-7::GFP marker and the positions of the QL descendants were scored using this marker.
For autosomal mutations where the males were not capable of mating, heterozygous males were generated by crossing mutant hermaphrodites also carrying the mec-7::GFP marker with wild-type or him-5(e1490) males. The heterozygous male progeny from this cross were mated with hermaphrodites bearing a known mutation but no mec-7::GFP marker. Cross-progeny were identified by the presence of the mec-7::GFP marker and their QL descendant positions were scored as above, with the expectation that 50% of the progeny would be heterozygous for the cell migration mutation.
We were unable to perform complementation tests between qid-7(mu327) and qid-8(mu342), because these mutations were X-linked and the males were not capable of mating. Using a tra-1 mutation to generate heterozygous XX males for mating also failed, as the low mating efficiency of tra-1 XX males was exacerbated by the low brood size and low hermaphrodite mating efficiency of these two mutants. Since qid-7(mu327) and qid-8(mu342) map to the same general region on LG X and share many phenotypes, it is quite possible that they are allelic. We have designated different names for them as we were unable to show that they correspond to lesions in the same gene.
Whenever a mutation mapped to a particular chromosome, we performed complementation tests against all the known genes that affected the migrations of QL and/or its descendants on the same chromosome as listed in Table 1. Although these tests were fairly exhaustive for published mutations that affect these cell migrations, we cannot rule out the possibility that these mutations could correspond to lesions in known genes whose QL migration phenotype has not been described. These tests were performed as described above, except for those against egl-20. For complementation tests against egl-20(n585), candidate egl-20 alleles were first mated with dpy-20(e1282); him-5(e1490) males. The dpy-20 mutation served as a linked marker for egl-20. The male progeny from this cross, which were heterozygous for the putative egl-20 mutation, were then mated with egl-20(n585) dpy-20(e1282) hermaphrodites. To assay for complementation, the final positions of the Q descendants were determined in the non-dumpy (Dpy) progeny of this cross.
Mutants that complemented all other mutations known to disrupt the migrations of QL and/or its descendants were considered to represent new genes affecting this cell migration. However, because these genes are mostly represented by one allele, it remains possible that they could correspond to known genes whose existing alleles do not display cell migration phenotypes in the QL lineage.
Sequencing egl-20 alleles: Two regions of the predicted egl-20/Wnt gene were PCR amplified from egl-20(mu320) and egl-20(mu241). Exons 1-4 were amplified with JW1, 5′-CTTAAC CAGGCAAATCGGAA-3′, and JW5, 5′-CACACATAAGACAA CACCTG-3′; exons 5-10 were amplified with JW3, 5′-CGT GTCGTTATGAAATACGC-3′, and JW4, 5′-TCTTGTTTTGCT AGGTCCCG-3′. These amplified regions included the entire coding region and all intron/exon boundaries of the predicted Wnt gene. Fragments were cloned and sequenced, as described in Maloof et al. (1999), from two independent PCR reactions.
The mu320 allele contains an opal mutation in the second exon of egl-20. The predicted mu320 protein product has only 48 amino acids in addition to the presumptive signal sequence and thus may be a null allele. The mu241 allele changes an invariant splice-donor GT sequence to AT in the second intron.
Outcrossing and strain construction: The mutations were initially isolated as strains bearing a mec-7::GFP marker (either muIs32 or muIs35). These mutations were outcrossed three times prior to further characterization. Only qid-5(mu245) was retained in the mec-7::GFP background because it was tightly linked to the muIs32 insertion bearing the mec-7::GFP marker. After the outcrossing process, we confirmed that these mutations affected the migrations of cells in the QL lineage by examining the final positions of the QL descendants, PVM and SDQL (Figure 3), and comparing these positions to wild-type or muIs32 controls.
For the first outcross, mutant hermaphrodites were mated with either muIs32; him-5(e1490) or muIs35 males. The heterozygous cross-progeny (F1’s) were cloned out onto individual plates and were confirmed as cross-progeny by scoring the frequency of the mutant phenotype in the next generation (F2’s). F3’s bearing the mutation, a mec-7::GFP marker, and, if possible, the him-5(e1490) mutation were reisolated from these F2’s.
For the second outcross, hermaphrodites from lines reisolated from the first outcross were mated with wild-type males. The heterozygous progeny from this cross were mated with him-5(e1490) males for a third outcross. From these crosses, we reisolated the mutation in the absence of the mec-7::GFP marker and him-5(e1490) as well as in double-mutant combinations with the mec-7::GFP marker or him-5(e1490). The mec-7::GFP and him-5(e1490) double mutants permitted scoring of axon phenotypes and male tail development, respectively (see results). In addition, the heterozygous males from this outcross were mated into a mab-5::lacZ reporter strain (muIs3 or muIs4; Cowing and Kenyon 1992) to generate double mutants bearing both the mutation and the mab-5::lacZ reporter.
In the case of qid-5, it was tightly linked to the muIs32 insertion on LG II and we did not separate it from muIs32 during these outcrosses. Thus, for this mutation, all controls were performed in a muIs32 background.
β-Galactosidase staining: β-Galactosidase staining was performed as described previously (Salser and Kenyon 1992). Staged populations of animals bearing muIs3 or muIs4 (Cowing and Kenyon 1992) along with the mutations of interest were fixed and stained ∼4 hr after hatching, around the time of the first division of QL where mab-5::lacZ is expressed (Salser and Kenyon 1992). In an otherwise wild-type background, the mab-5::lacZ reporters muIs3 and muIs4 were expressed in the QL descendants but not in the QR descendants (Table 5). In addition, we examined the expression of these reporters in the QL descendants in the egl-20(n585) background. Consistent with prior observations (Harriset al. 1996), egl-20(+) is required for mab-5 expression in these cells (Table 5). Unfortunately, we were unable to use these lacZ fusions to examine mab-5 expression in other tissues, as they do not recapitulate the complex expression of mab-5 in these tissues (Salser and Kenyon 1996).
Microscopy and scoring of phenotypes: During screening or strain construction, we identified cell migration mutants that affect the QL lineage by observing the positions of PVM in larvae under a Leica stereo dissecting microscope equipped with a Kramer epifluorescence unit or a Zeiss M2Bio stereo dissecting microscope with a similar epifluorescence unit. For more detailed analyses of cell positions and cell fates, animals were mounted on agarose pads and examined using Nomarski/DIC optics on a Zeiss Axiophot microscope with a ×100 objective. The positions of the Q descendants, AVM, PVM, SDQR, and SDQL were scored as described previously (Harriset al. 1996). Briefly, their positions in relation to the stationary V cells were noted in larvae near the end of the L1 larval stage; animals were deemed to be at the correct point in development when all the P nuclei had descended into the ventral cord, as the Q migrations were complete by this point in wild-type animals. The positions of other migratory neurons, hermaphrodite-specific neurons (HSNs), canal-associated neruons (CANs), ALM, and BDU, were also scored at this time point. Expression of mab-5::lacZ in stained animals was scored using Nomarski/DIC optics with a ×100 objective.
During the examination of the male tail, animals were immobilized with 1 mm levamisole. Cells and other anatomical features were identified as described in Sulston and Horvitz (1977). Wild-type or him-5(e1490) animals exhibited a low frequency of ray defects in which one ray derived from either V5 or V6 (V rays) is missing as previously reported (Sulstonet al. 1980). Only mutants with defects significantly more penetrant than this background were considered to have a male tail phenotype.
Axon morphology was visualized using the mec-7::GFP fusions muIs32 and muIs35 under the ×20 objective of the Zeiss M2 Bio stereo dissecting microscope. The worms were partially immobilized by chilling at 2°-4° in a cold room for 5-15 min to facilitate scoring of the axons. This process did not affect axon morphology; worms treated this way did not differ significantly in their axon morphology from control worms (data not shown). We note that a small number of animals bearing the muIs32 and muIs35 insertions display defects in their axon morphology. We considered only mutants with significantly more penetrant defects as having a bona fide axon outgrowth phenotype.
Screening for mutations that disrupt cell migrations in the Q lineages: Previous genetic screens for mutants defective in the migrations of cells within the QL and QR lineages relied on screening at high magnification to locate the positions of the neurons generated by QL and QR (Wanget al. 1993; Harriset al. 1996; Symet al. 1999; Du and Chalfie 2001). These screens were laborious because they required either mounting animals on slides for viewing at high magnification or β-galactosidase staining.
To screen for mutants that disrupt the migrations of QL, QR, and/or their descendants rapidly, we used animals bearing a chromosomally integrated mec-7::GFP fusion that expressed GFP at high levels specifically in six mechanosensory neurons (Chalfieet al. 1994), including one QL descendant (PVM) and one QR descendant (AVM; Figure 2). These GFP-expressing neurons were visible under a dissecting microscope with epifluorescence capability, which permitted rapid and direct identification of mutants with misplaced QL and/or QR descendants on a culture plate.
We generated two chromosomal insertions, muIs32 (on LG II) and muIs35 (on LG V), that contain the mec-7::GFP marker. These insertions did not affect the migrations of the QL or QR lineages (Figure 3; data not shown) and were used as starting strains for screens for mutants with misplaced QL and QR descendants. We classified the mutants isolated by their most prominent phenotype. In the first class, the primary defect was misplaced QL descendants (some of these mutants also show subtle defects in the positioning of the QR descendants; see below). In the second class, the primary defect was misplaced QR descendants. The third class exhibited defects in both QL and QR descendants. In this study, we focused on the first class of mutants whose most prominent phenotype is a cell migration defect in the QL lineage.
Identification of new genes that disrupt cell migrations in the QL lineage: From ∼100,000 F2 animals screened, we isolated a total of 35 mutants with misplaced PVMs, indicative of a migration defect in the QL lineage. A number of genes required for the migration of cells in the QL lineage have already been described. Thus, we wanted to determine which mutations identified in this screen were lesions in these known cell migration genes and which mutations corresponded to new loci. We assigned these mutations to individual chromosomes by mapping them against a panel of Tc1 transposon insertions with known chromosomal locations (see materials and methods). For mutations that mapped to the same chromosome as known QL migration genes, complementation tests were performed to determine if they were alleles of known genes.
This process determined that 31 mutations represented new alleles of previously identified genes. By this process of elimination, we found that 4 mutations corresponded to new genes involved in the migrations of cells in the QL lineage (see materials and methods). These mutations were all recessive and therefore mostly likely represent reduction or loss-of-function mutations. Further complementation tests among mutations that map to the same chromosome and additional mapping that refined the position of these mutations (Table 2; materials and methods) indicated that these 4 mutations likely corresponded to lesions in four genes that we have named qid-5-qid-8, as listed in Table 2. For technical reasons (see materials and methods), we were not able to perform complementation tests between qid-7(mu327) and qid-8(mu342). Since these mutations map to a similar region on the X chromosome and share similar phenotypes (see below), it is still possible that they represent lesions in the same gene.
Further characterization of the new mutants with abnormal cell migrations in the QL lineage: What are the function(s) of these new genes in cell migration and other processes? We addressed these questions by determining how these mutations affect the migrations of the cells in the QL lineage and other related developmental events. In this section, we list the characterizations performed; the summary of these conclusions is detailed in the next section of the results.
We determined the final positions of both the QL and QR descendant cells (AVM, PVM, SDQR, and SDQL—collectively known as the Q.pax cells) at high magnification and compared their distributions to those of wild-type or muIs32[mec-7::GFP] controls (Figure 3). This confirmed the cell migration phenotypes in the QL lineage and provided additional information about the penetrance and expressivity of these phenotypes. This analysis also revealed additional positioning defects for the QR descendants that were not obviously visualized with the mec-7::GFP marker (see below).
Next, we examined these mutants for other selected developmental phenotypes listed below. The spectrum of observed phenotypes in a mutant, along with the specificity or pleiotropy of the defects, could suggest a potential developmental function for the corresponding gene. In addition, we could compare the set of phenotypes in the new mutants to that of existing mutations in known genes to determine if the new genes are likely to function in a common pathway.
Other neuronal migrations: We examined the final positions of other migratory neurons in these mutants to determine whether the migration defect was specific to the QL and QR lineages or whether there were more general migration defects (Figure 4). These neurons included the HSNs, the CANs, the ALMs, and the BDU neurons (Sulstonet al. 1983). All of these migrations occur during late embryogenesis in the directions indicated in Figure 4. These neurons are all bilateral and show no apparent left-right asymmetry in their migrations.
Axon guidance and outgrowth: We also examined these mutants for defects in axon guidance and outgrowth to determine whether the migration defects were specific to cell migrations or whether they also affected axonal outgrowth (Table 3; Figure 5), as some existing mutations are known to affect both axonal and cell migrations. We used the mec-7::GFP fusion to visualize the axons of the touch neurons in these mutant backgrounds because it permitted us to examine axons that are guided in different directions and along different axes: the axons of AVM and PVM grow ventrally and then anteriorly, while those of the ALMs and PLMs grow anteriorly.
Male tail development: Many existing mutants with misplaced QL descendants, such as mab-5 and egl-20, show defects in anteroposterior patterning or cell fate determination that result in defective male tails. The Hox genes mab-5 and egl-5 that control the migrations of the cells in the QL lineage are also responsible for specifying the fates of many cells in the male tail. Thus the presence of certain characteristic male tail phenotypes could suggest defects in these Hox pathways (Kenyon 1986; Chisholm 1991; Salser and Kenyon 1992). We assessed the new mutants for similar defects that could reveal a function in the same developmental processes (Table 4).
The male tail is generated by a vast array of developmental events, including cell migration, cell-fate determination, neurogenesis, anteroposterior patterning, and morphogenesis. By examining this one structure we could survey a diverse set of developmental events (Sulstonet al. 1980). Moreover, defects in specific structures in the male tail could be traced to particular cell lineages, aiding diagnosis of the phenotype (detailed in Sulstonet al. 1980). We focused on the following structures: (1) the nine rays on each lateral side, which are neuronal structures derived from a set of lateral epidermal cells located on the corresponding side near the tail of the male C. elegans; (2) the hook and its associated cells derived from posterior ventral epidermal cells; and (3) the mating spicules generated by a male-specific blast cell (the normal straight-spicule morphology also requires the function of posterior sex muscles derived from migratory myoblasts).
Expression of mab-5 in QL: mab-5 expression in the QL but not QR lineage is a critical step during the left-right asymmetric migrations within these lineages because mab-5 expression in these cells is necessary and sufficient to direct a program of posterior migration (Salser and Kenyon 1992; Harriset al. 1996). Since mab-5 is normally expressed only in the QL descendants, it also serves as a molecular marker for the left-right asymmetry between QL and QR. By examining the expression of a mab-5::lacZ reporter in the Q descendants of these mutants (Table 5), we could infer whether the cell migration defect is caused by defective or inappropriate mab-5 expression or is due to an inability of the migrating cells to respond properly to mab-5.
Phenotypes of the mutants with cell migration defects in the QL lineage: qid-5(mu245)II: In ∼50% of the animals examined, the QL.pax cells were found in the anterior (Figure 3), consistent with a defect in mab-5 expression in the QL daughters (Table 5). In this mutant, the ALM cell bodies also stopped prematurely (∼40% penetrant) and were positioned more anteriorly than normal (Figure 4). Other cell migrations appeared normal in qid-5(mu245) animals. (See Figure 4.)
qid-6(mu252)III: This mutant developed fluid-filled blisters along its body. It was unlikely to be a lesion in bli-5 (Hodgkin 1997), a known mutation located on the same chromosome with a blistering phenotype (Bli), because qid-6(mu252) complemented bli-5(e518) and the QL descendants were not misplaced in bli-5(e518) animals (data not shown). In addition to the Bli phenotype, qid-6(mu252) animals were also somewhat sluggish (uncoordinated, Unc) and slightly short and fat (Dpy). In this mutant, the QL.pax cells were found in the anterior ∼40% of the time, most likely due to a defect in mab-5 activation (Figure 3; Table 5); other neuronal migrations appeared normal. In the male tail, cell-fate and morphogenesis defects were observed, with animals showing missing rays or inappropriate ray fusion (Table 4). The positions of the rays also appeared slightly disorganized.
qid-7(mu327)X: This mutant had pleiotropic phenotypes. In a population of qid-7(mu327) animals, the QL.pax cells were variably displaced along the body axis (Figure 3). This was consistent with the incompletely penetrant absence of mab-5::lacZ expression in the QL descendants (Table 5); even when mab-5::lacZ was expressed, the levels of staining were also often reduced in comparison to the wild-type controls. On the right side, the distribution of the QR.pax cells was consistently shifted posteriorly (Figure 3), a phenotype reminiscent of egl-20/Wnt mutants (Harriset al. 1996). qid-7(mu327) animals were very sick and exhibited low brood sizes. They were sluggish (Unc), slightly Dpy, and egg-laying defective (Egl), often bloated with unlaid eggs. This Egl phenotype may be partially accounted for by potential defects in vulval development suggested by the protruding vulva (Pvl) phenotype (Eisenmannet al. 1998; Eisenmann and Kim 2000) and the strong HSN cell migration phenotype (Desaiet al. 1988); >50% of the HSNs either were located in more posterior positions than wild type or were not found in the body at all (Figure 4). In the axons of the touch neurons, we observed a variety of defects, including ectopic branching and aberrant trajectories (Figure 5, B-D). Male tail development was severely disrupted (Table 4); rays were missing or disorganized. Crumpled spicules and missing or displaced hooks were also observed. Together, these defects in the male tail indicated developmental defects in the posterior lateral/ventral epidermis and the sex muscles.
qid-8(mu342)X: This was also a very pleiotropic mutant with sluggish (Unc), Egl, Pvl, very weak Dpy phenotypes as well as very small brood sizes. In qid-8(mu342), the final positions of the QL.pax cells were distributed along the body (Figure 3); this could be explained by the fact that mab-5 expression is variably reduced in the QL daughters (Table 5). An egl-20/Wnt-like posterior shift was also seen in the distribution of the QR.pax cells (Figure 3). Among other migratory neurons, HSN showed a strong phenotype (Figure 4). The very small populations of CAN and ALM neurons that were located more anteriorly than normal (Figure 4) raise the possibility that there might be very subtle migration phenotypes in these neurons. Inappropriate axon branching and guidance defects were seen in the touch neurons (Figure 5, E and F). The male tail was also severely compromised with ray and hook defects that were likely due to lineage or cell-fate defects in the posterior lateral/ventral epidermis (Table 4). Also, the presence of crumpled spicules indicated defects in the sex muscles (Table 4).
lin-17/Frizzled is required for normal PLM axonal outgrowth: In this screen, we isolated mu243, an allele of lin-17/Frizzled (Sawaet al. 1996) that exhibited axon outgrowth defects in the PLM touch neurons. In wild-type animals, the PLM neurons normally send out axons that extend slightly past the mid-body along the ventrolateral surface of the worm. These PLM axons were often truncated in lin-17(mu243) mutants (Table 3; Figure 5H). This phenotype was also detected in lin-17(n671) (Table 3; Figure 5G), a strong loss-of-function allele (Sawaet al. 1996). Mutations that disrupt other components of the conserved Wnt pathway in C. elegans, such as egl-20, mig-14, or bar-1, appeared wild type for PLM axon outgrowth (data not shown).
lin-17 has been implicated in cell-fate and cell-lineage decisions in C. elegans (Sternberg and Horvitz 1988; Sawaet al. 1996). However, in lin-17(mu243) and lin-17(n671), the PLM touch neurons still expressed mec-7::GFP (Figure 5, G and H), a very specific differentiation marker, suggesting that they were still adopting the correct touch-cell fate. Moreover, among the touch neurons that we scored, this defect was specific to PLM. Thus, it is possible that mutations in lin-17 disrupt axonal outgrowth in the PLM neurons in a step distinct from the adoption of a touch-neuron fate.
The left-specific expression of the Hox gene mab-5 in the QL lineage and its outcome are controlled by at least three different sets of genes. The first set governs the polarization of the QL neuroblast properly toward the posterior. The second set allows the QL neuroblast and/or its progeny to receive the EGL-20/Wnt signal and specifies the left-right asymmetric responses to this EGL-20/Wnt signal so that this signal is interpreted correctly. Finally, the third set of genes is turned on by mab-5 and acts more directly to guide cells to the posterior. Here, we have described new mutants that disrupt the migrations of the cells in the QL lineage by affecting the activation of mab-5 in these cells. The corresponding genes are likely to function in this step during the migration of the cells in the QL lineage.
The cell migration defects in these mutants allowed us to place them in different steps of a large pathway that controls and coordinates these complex left-right asymmetric cell migrations. This is summarized in Figure 6 and detailed below. With the caveat that most of these genes are represented by only one allele, the overall spectrum of phenotypes in these mutants (Table 6) allowed us to speculate about their broader roles in development and how these developmental processes might intersect with the migrations of the cells in the QL and QR lineages.
Genes isolated in this screen for cell migration mutants: In this large-scale screen, we were able to isolate several new mutations that disrupt the cell migrations within the QL lineage. This was done rapidly by labeling these cells with a GFP reporter that could be seen under a dissecting microscope equipped with epifluorescent capabilities. However, this screen was not saturating, as many of the new genes presented here were represented by only single alleles. Hence our interpretation of their role in cell migrations (below) must remain tentative until additional alleles are isolated.
In this screen, ∼45% of the candidate mutants we isolated were sterile or not viable. In the future, F2 clonal screens that rely on examining a population of siblings for misplaced QL descendants might be more useful in isolating cell migration mutants of this class. Such a screen could also identify mutants with weak phenotypes that might be more obvious when observed in a population.
Possible role(s) of these genes in regulating mab-5 expression: After the initial migrations, an EGL-20/Wnt signal triggers mab-5 expression in the QL descendants through a conserved Wnt signaling pathway; this causes them to migrate posteriorly (Salser and Kenyon 1992; Harriset al. 1996; Sawaet al. 1996; Eisenmannet al. 1998; Maloofet al. 1999; Herman 2001). Most of the mutants described here act at this step: the mutations in qid-5, qid-6, qid-7, and qid-8 all compromised mab-5 expression in the QL descendants. This likely explains the anterior migrations of the QL descendants in these mutants. This phenotype indicated that these new genes intersect with the egl-20/Wnt pathway directly or indirectly during the activation of mab-5 expression in the QL lineage.
egl-20/Wnt has a second role in the cell migrations of the QR descendants: it signals these cells to migrate anteriorly to their fullest extent and this appears to be mediated by unknown genes distinct from those required for the QL migrations (Harriset al. 1996; Eisenmannet al. 1998; Maloofet al. 1999; Whangbo and Kenyon 1999). In egl-20 mutants, the QR descendants migrate in the anterior direction but often stop short of the wild-type position. Among this set of mutants with mab-5 expression defects, only qid-7 and qid-8 exhibited a posterior shift in the final positions of the QR descendants, similar to that observed in egl-20 (Harriset al. 1996). Also, there was considerable overlap between the other phenotypes of egl-20 and these mutants, such as the HSN migration defect and the presence of crumpled spicules in males (Desaiet al. 1988). This raised the possibility that these genes might function in a more integral part of the egl-20/Wnt pathway. However, we also note that qid-7 and qid-8 also exhibited many other phenotypes not seen in egl-20 mutants, such as the axon outgrowth defects, and thus may act in additional pathways. Also, the egl-20 mutant has a defect in the asymmetric V5 cell division, a phenotype that we did not detect in any of the new mutants in this screen, including qid-7 and qid-8 (data not shown).
The different roles of egl-20/Wnt in QL and QR migration highlight the complexities involved in interpreting this signal. Not only is it likely that different downstream signaling components are used in QL and QR, but also there must be factors that generate these distinct responses of QL and QR to a given level of EGL-20/Wnt (Whangbo and Kenyon 1999). Mutations in qid-5 and qid-6 impaired one egl-20-dependent process (mab-5 activation in QL descendants) but not the other (anterior migration of QR descendants). With the caveat that most of these genes are each defined by only one recessive mutation that might not be null, this suggests that these genes do not function in all aspects of EGL-20/Wnt signaling. This specificity is very intriguing. We speculate that these genes could act in three ways. First, they could represent links that couple the initial migrations of the QL neuroblast to the QL-specific activation of mab-5. Second, they could contribute to the left-right asymmetric response to egl-20/Wnt by acting to transduce the EGL-20 signal only in the mab-5 activation pathway. Third, these genes could function by predisposing the QL descendants to respond to an EGL-20 signal with mab-5 activation instead of anterior migration.
qid-6(mu252) animals also exhibited a blistered cuticle phenotype, suggesting a defect in cuticle formation. Genes encoding extracellular matrix molecules such as bli-1 and bli-2, as well as proteases postulated to process these collagens (Peterset al. 1991; Thackeret al. 1995), can mutate to cause a similar blistered cuticle phenotype (Johnstone 2000). A Bli phenotype was not sufficient to confer a cell migration defect in QL or its descendants, as bli-5 animals exhibited normal placement of the QL descendants (data not shown). Perhaps qid-6 affects cuticle synthesis or disrupts the extracellular matrix in a different way or at a different time than bli-5 does. It is possible that the mab-5 activation defect could be due to related defects in extracellular matrix molecules that are required for the distribution or presentation of signals like EGL-20/Wnt, as similar examples have been described in Drosophila and zebrafish (Hackeret al. 1997; Walsh and Stainier 2001).
qid-6, qid-7, and qid-8 also affected male tail development with phenotypes that resembled mab-5 mutants, such as crumpled spicules, missing rays, and missing hooks (Kenyon 1986). This suggests that they might also function as regulators of mab-5 during male tail development, raising the possibility that they represent more general regulators of mab-5 activation or function in C. elegans.
Conclusion: In summary, the collection of mutants identified here provides additional entry points into the study of the migrations of QL and its descendants as well as other cell migrations in C. elegans. Many of the known mutants that disrupt cell migrations in the QL lineage also affect other developmental processes in a manner similar to those affected by some of these new mutants. However, the additional phenotypes are not identical. Apart from the cell phenotypes in the QL lineage, some of these new mutants showed only a subset of the phenotypes displayed by known mutants that affect cell migrations in the QL lineage. This suggests that these new genes might act at specific times and places, instead of being obligate factors in previously characterized pathways. Other mutants such as qid-6 exhibit phenotypes not seen in known cell migration mutants that affect the QL lineage; these new mutants might therefore act in new pathways that intersect with some of the known pathways during the migrations of QL and/or its descendants. Discerning their molecular identities and their relationships with the known pathways that control the migration in the QL lineage will be important in understanding the complex developmental processes that result in the precise control of these intriguing cell migrations.
We thank past and present members of the Kenyon Lab for discussions and advice. We are indebted to Cori Bargmann, Lucie Yang, Scott Alper, Phil Anderson, and anonymous reviewers for critical and helpful comments on the manuscript. We are also grateful to Lucie Yang for her assistance in mapping the mec-7::GFP insertions, Lindsay Hinck, Marc Tessier-Lavigne, Cori Bargmann, and the Caenorhabditis elegans Genetics Center for strains and to Tom Kornberg and Cori Bargmann for use of equipment. Q.C. was a Howard Hughes Medical Institute Predoctoral Fellow, L.W. and J.S.W. were National Science Foundation Predoctoral Fellows, M.S. was a Postdoctoral Fellow of the Jane Coffin Childs Memorial Fund for Medical Research, Y.S.L. is a Postdoctoral Fellow of the Cancer Research Fund of the Damon Runyon-Walter Winchell Foundation (DRG 1669), and C.K. is the Herbert Boyer Professor of Biochemistry and Biophysics. This research was supported by the National Institutes of Health (grant GM-37053) and by grants of the Virtual Research Institute of Aging of Nippon Boehringer Ingelheim.
Communicating editor: P. Anderson
- Received May 10, 2002.
- Accepted April 16, 2003.
- Copyright © 2003 by the Genetics Society of America