ABSTRACT
Cell migrations play a critical role in animal development and organogenesis. Here, we describe a mechanism by which the migration behaviour of a particular cell type is regulated temporally and coordinated with over-all development of the organism. The hermaphrodite distal tip cells (DTCs) of Caenorhabditis elegans migrate along the body wall in three sequential phases distinguished by the orientation of their movements, which alternate between the anteroposterior and dorsoventral axes. The ventral-to-dorsal second migration phase requires the UNC-6 netrin guidance cue and its receptors UNC-5 and UNC-40, as well as additional, UNC-6-independent guidance systems. We provide evidence that the transcriptional upregulation of unc-5 in the DTCs is coincident with the initiation of the second migration phase, and that premature UNC-5 expression in these cells induces precocious turning in an UNC-6-dependent manner. The DAF-12 steroid hormone receptor, which regulates developmental stage transitions in C. elegans, is required for initiating the first DTC turn and for coincident unc-5 upregulation. We also present evidence for the existence of a mechanism that opposes or inhibits UNC-5 function during the longitudinal first migration phase and for a mechanism that facilitates UNC-5 function during turning. The facilitating mechanism presumably does not involve transcriptional regulation of unc-5 but may represent an inhibition of the phase 1 mechanism that opposes or inhibits UNC-5. These results, therefore, reveal the existence of two mechanisms that regulate the UNC-5 receptor that are critical for responsiveness to the UNC-6 netrin guidance cue and for linking the directional guidance of migrating distal tip cells to developmental stage advancements.
INTRODUCTION
Cell migrations are a key component of development and organogenesis, yet the regulatory mechanisms of cell migrations are poorly understood, particularly the events that cause changes in migratory trajectory responsible for shaping organs and organ systems. Directional information for migrating cells comes from extracellular cues that act through cell surface receptors to regulate the cytoskeletal machinery required for cell polarity and motility. Long-range cell and growth cone migrations typically proceed via a series of sequential, shorter migrations. Transitions between these shorter migrations may involve a change in the direction of movement or in the substratum over which migration occurs. For example, the migrating germ cell precursors of Drosophila associate with, in order, the posterior midgut primordium, the mesoderm and the somatic gonad primordium in a sequence precisely coordinated with embryonic development (Zhang et al., 1996). In principle, the transitions between migratory phases might include changes either in the way the cell responds to its environment or in the environment itself, but how these changes are initiated and executed is not completely understood.
The bilobed gonad of C. elegans hermaphrodites develops during larval development from a 4-cell primordium positioned in the ventral midbody. The shape of the two gonad arms is determined by the migratory path of a cell called a distal tip cell (DTC) at the leading edge of each arm (Hedgecock et al., 1987). DTC migrations proceed through three sequential linear phases, with each phase occurring along one of the natural body axes and at right angles to the previous phase (Fig. 1A-E). The first migratory phase is a centrifugal migration along the ventral band of body wall muscles away from the midbody. One DTC migrates toward the anterior and the other toward the posterior. The second phase begins with an orthogonal (i.e., 90°) turn of the DTCs and consists of a migration along the inner surface of the epidermis from the ventral to the dorsal muscle bands. Finally, during the third phase, the DTCs turn 90° and migrate centripetally along the dorsal muscle bands back toward the midbody. This series of linear migrations on the body wall gives rise to two mirror-image C-shaped adult gonad arms (one for each DTC; Figs 1E, 2A; also see Hedgecock et al., 1987). A similar, triphasic C-shaped migratory path is followed by the male linker cell (MLC) that shapes the single (anterior) male gonad arm in C. elegans and by many neuronal growth cones. For example, the growth cones of the DD motorneurons initially project from their cell bodies longitudinally within the ventral nerve cord, then make a 90° turn to extend circumferentially in a ventral-to-dorsal direction. Upon reaching the appropriate location of the dorsal nerve cord, they turn once more in a retrograde longitudinal direction, resulting in a C-shaped axonal morphology (White et al., 1986).
Circumferential cell and growth cone migrations in C. elegans, including the ventral-to-dorsal second phase of the DTC and MLC migration, are mediated in part by the netrin family member UNC-6 and its receptors UNC-5 and UNC-40 (Hedgecock et al., 1990; Culotti and Merz, 1998). In the DTCs and MLC, UNC-5 and UNC-40 mediate chemorepulsion away from ventrally expressed UNC-6 (Wadsworth et al., 1996). A role for netrins and their receptors in complex cell migrations in mammalian development was demonstrated by the molecular characterization of the murine rostral cerebellar malformation (rcm) locus, which encodes an UNC-5 homologue (UNC5H3; Ackerman et al., 1997). Mutations at this locus disrupt the migrations of granule cell precursors in the external germinal layer of the developing cerebellum (Ackerman et al., 1997; Przyborski et al., 1998).
The timing of the turns made by the DTCs and the MLC are synchronized with progression through the larval stages of development. The progression from the centrifugal first phase to the circumferential (i.e., ventral-to-dorsal) second phase of migration occurs at a precise time late in the second larval stage (L2) for the MLC and the third larval stage (L3) for the DTCs. In mutants of unc-5, unc-6 and unc-40, the second phase of DTC and MLC migration frequently fails, but the first and third phases occur with the proper timing (Hedgecock et al., 1990; see Fig. 2B for resulting gonad morphology). Mutations in two genes, daf-12 (particularly alleles formerly known as mig-7) and mig-8, disrupt the navigational program of the DTCs, such that the second and third migratory phases are delayed or absent (Hedgecock et al., 1987; Antebi et al., 1998). These mutants were isolated because they have linear gonad arms that result from the failure of the DTC program to advance from the second to the third larval stage (Hedgecock et al., 1987; see Fig. 2F). daf-12 encodes a nuclear hormone receptor that is a critical regulator of larval developmental stage progressions for somatic worm tissues (Yeh, 1991). It has been proposed that DAF-12 acts to advance larval stage-specific developmental programs of several different somatic cell types (Antebi et al., 1998). The single existing allele of mig-8 appears to affect only the progression of the navigational program of the DTCs (Antebi et al., 1998).
A detailed in vivo characterization of the mechanisms by which a specific cell migrates may uncover principles broadly applicable to development and cellular biology. It may also reveal basic mechanisms of axon guidance, as directional cues like UNC-6 are utilized by both migrating cells and neuronal growth cones. The purpose of the present study was to determine the molecular mechanisms involved in DTC navigation, particularly the regulation of the first change in direction of DTC movement. The phenotypes of several mutants that affect DTC migrations suggest that the initiation of the ventral-to-dorsal (i.e., second) phase of DTC migration involves the activation or disinhibition of the UNC-6-mediated guidance system by MIG-8 and DAF-12. In principle, this could be achieved by regulating any one or more of the components of the UNC-6 system. Here we report that this second phase of DTC migration is initiated largely by the MIG-8 and DAF-12-dependent upregulation of unc-5 gene expression. A premature high level of expression of UNC-5 using a heterologous promoter induces precocious ventral-to-dorsal DTC migration, indicating that UNC-5 upregulation can be sufficient to drive this migratory phase. However, we also find genetic evidence that UNC-5 function is facilitated at the time of turning, possibly through release from an opposing or inhibiting pathway. Together, these two mechanisms ensure that responsiveness to the UNC-6 directional signal is initiated at the appropriate time in the DTC migration pattern. The regulation by DAF-12 and MIG-8 of the UNC-5 guidance receptor is thus one of the pathways linking this specific cell migration event to the progression of the whole animal through developmental stages.
MATERIALS AND METHODS
Strains
C. elegans were incubated and handled as previously described (Brenner, 1974). Unless otherwise noted, all cultures were grown at 20°C. Most strains not derived in our laboratories were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). Transgenic lines are described below. mig-6(e1931) homozygotes were obtained from the self-fertilization of + mig-6(e1931)/dpy-11(e224) + trans double heterozygotes. All other strains were maintained as homozygotes. Males were obtained from him-5 double mutant strains.
Scoring of DTCs
The morphology of somatic gonad arms in wild-type and mutant worms has been described previously (Hedgecock et al., 1987). Adult gonad morphologies reflect the pathway of DTC migrations. The misshapen gonads produced by DTC or MLC migration defects may be scored either using a dissecting microscope as clear patches produced by displacements of the intestine (Hedgecock et al., 1990), or at higher magnification under Differential Interference Contrast (DIC) optics. Standard errors for the penetrance of DTC defects were calculated assuming a binomial distribution of the actual sample size (Hedgecock et al., 1990; Antebi et al., 1998). To assess the timing of migrations, newly hatched larvae were picked and incubated at 20°C for the desired period of time. Worms were then mounted for examination under DIC optics using 1 mM levamisole as anaesthetic. As a control for accuracy of timing, the developmental state of the vulval precursor cells (VPCs) was also determined. The VPCs (e.g. P6.p) divide at 30 hours after hatching, coincident with dorsalward turning of the DTCs (Sulston and Horvitz, 1977). The timing of VPC divisions was normal in the strains for which timing experiments were carried out, including evIs99, him-5 and mig-6.
Derivation of transgenic lines
Transgenes were introduced into C. elegans by germline transformation using the gonad injection method previously described (Mello et al., 1991; Mello and Fire, 1995). The co-injected co-transformation marker was either the pRF4 plasmid containing the rol-6(su1006dm) gene (Kramer et al., 1990), which creates a dominant rolling (Rol) phenotype, or pMH86 (Clark et al., 1995), which contains the wild-type dpy-20 gene and rescues a dpy-20 mutation present in the injected line. Extrachromosomal arrays were integrated by gamma-irradiation (3000 rads) followed by selection for Mendelian inheritance. Integrated transgenic arrays were assigned to linkage groups and were passed into mutant backgrounds using standard genetic methods.
unc-5B reporter constructs
unc-5B reporter constructs were made by fusing a 4.6 kb fragment of genomic DNA 5′ to the start of unc-5 exon 2 to the coding sequence of lacZ, lacZ with nuclear localization sequence (NLS), or gfp reporter genes previously described (Hamelin et al., 1993; Fire et al., 1990; Colavita et al., 1998). pYZ129 (unc-5B∷lacZ) contains the 4.6 kb promoter fused to lacZ, and the evIs54 transgenic strain contains the array [unc-5B∷lacZ;rol-6(su1006dm)] integrated on LGII. pIS21 (unc-5B∷lacZ with NLS) contains the 4.6 kb promoter fused to lacZ with an NLS and the evIs51 line contains the array [unc-5B∷lacZNLS;rol-6(su1006dm)]. The chromosomal site of integration for this line is unknown. pSU15 (unc-5B∷gfp) comprises the 4.6 kb promoter fused to the gfp gene and the nuIs9 line contains the array [unc-5B∷gfp;myo-3∷lacZ;dpy-20(+)] integrated on LGI in a dpy-20(e1282) background. The unc-5B∷HA-tagged reporter construct (pU5 derived from pYZ108 described in Hamelin et al., 1993, corrected to a wild-type coding sequence) was made by fusing a 4.6 kb fragment of genomic DNA 5′ to the start of exon 2 to an unc-5 cDNA with 1 kb of the normal 3′ genomic region appended to its 3′ end. Three in-frame HA tags were cloned into a BglII site after codon D889 (30 codons upstream from but still in frame with the normal stop codon) to derive pU5∷HA. An unc-5∷gfp-tagged reporter construct was made by cloning a full-length gfp gene (from pD9.75 described in Fire et al., 1990) into a BsaB1 site after the codon F909 in pU5.
Cell identifications
We found that lacZ and GFP reporters were distributed throughout cell bodies and neuronal processes (except unc-5B∷NLS-lacZ (pIS21). This facilitated identification of cells. The unc-5B∷lacZ reporter with an NLS (pIS21) sometimes expressed in several additional unidentified neurons in the head and tail that were not stained by constructs lacking the NLS. Because the β-gal staining was faint and axons were not visible, these cells were not identified. The GFP reporters were used for experiments that required careful timing and simultaneous DIC observations in live animals.
Construction of the emb-9∷unc-5 transgene
An XbaI-PvuI fragment containing 1.65 kb of the emb-9 promoter just 5′ to the initiation codon (from pJJ239) was fused to the wild-type unc-5 cDNA previously described (Hamelin et al., 1993) to make emb-9∷unc-5 (pSU16). This construct is missing 15 bp immediately 5′ to the emb-9 initiation codon, but contains the entire unc-5 coding region and 3′ UTR. The emb-9∷lacZ construct (pJJ318) that was co-injected with emb-9∷unc-5 in these experiments was constructed as described (Graham et al., 1997). The evIs99 transgenic line contains a multicopy array of [emb-9∷unc-5;emb-9∷lacZ;dpy-20(+)] integrated on LGI.
β-gal staining
Staining for β-galactosidase activity was carried out as described by Ruvkun and Giusto (1989) with modifications suggested by M. Finney and G. Ruvkun (Mass. Gen. Hospital) and by E. Aamodt and G. Xie (Louisiana State University). Briefly, animals were washed from NGM plates with water and pelleted in a microcentrifuge tube. The supernatant was removed and 500 μl of ice-cold MRWB (160 mM KCl, 40 mM NaCl, 20 mM Na2EGTA, 10 mM spermidine, 30 mM PIPES pH 7.4 and 50% methanol), 100 μl of 20% paraformaldehyde and 400 μl of water were added, mixed and incubated for 35 minutes at 4°C with occasional mixing. The worms were washed twice with 1.0 ml TTB (100 mM Tris-HCl pH 7.4, 1% Triton X-100 and 1 mM EDTA) by pelleting. To the pellet was added 960 μl of TTB, 30 μl of 20% paraformaldehyde and 10 μl of 2-mercaptoethanol. The animals were incubated for 15 minutes at room temperature, pelleted by settling (animals are fragile to centrifugation at this stage), then washed once with 1× BO3 and incubated for 15 minutes at room temperature with 200 mM DTT in 1× BO3. The worm pellet was washed twice with β-galactosidase staining mix (370 μl water, 500 μl 0.8 M sodium phosphate pH 7.5, 2 μl 1 M MgCl2, 4 μl 1% SDS, 100 μl Redox Buffer and 50 mM each of potassium ferrocyanide and potassium ferricyanide), 20 μl 2.5% X-Gal in DMF and 1.0 μl 1 mg/ml DAPI), then incubated overnight at room temperature or for 2 hours at 37°C. A similar permeabilization and fixation protocol was used for immunolocalization of HA-tagged UNC-5 protein using anti-HA antibodies (Wadsworth et al., 1996).
RESULTS
DTC migrations in the wild type have been well characterized (Hedgecock et al., 1987; Antebi et al., 1998). The centrifugal first phase of DTC migration is initiated early in L3 about 25 hours after hatching. The anterior and posterior DTCs migrate away from the midbody (i.e., centrifugally) along the ventral muscle bands (Figs 1, 2A). Late in L3, about 30 hours after hatching, the DTCs reorient and migrate circumferentially toward the dorsal muscle band, where at approximately 33 to 34 hours, they turn once more and migrate back toward the midbody (i.e., centripetally). In worms with an extended body length, such as lon-2, the first turn of the DTCs occurs at approximately the same distance from the midbody as in the wild type (our unpublished data, also see Morita et al., 1999). This indicates that turning occurs at the normal time in development even though the position of turning relative to many other tissues along the anteroposterior axis may be altered.
The ventral-to-dorsal second phase of DTC migration is specifically disrupted by mutations in unc-6, unc-5 or unc-40 (Hedgecock et al., 1990). In these mutants, the DTCs frequently migrate back toward the midbody along the ventral rather than the dorsal muscle bands and do so with normal timing (i.e., after hesitating a few hours on the ventral side) (Hedgecock et al., 1987; see Fig. 2B for resulting gonad morphology). This indicates that the centripetal third phase of the migratory program can proceed independently of the execution of the second phase and that the third migratory phase can occur on either the dorsal or the ventral side of the animal.
The reported phenotype of dig-1 and mig-4 mutants is consistent with this interpretation. In these mutants, the entire gonad primordium is frequently displaced to the dorsal side (Hedgecock et al., 1987; Thomas et al., 1990; see Fig. 2D). Consequently, no ventral-to-dorsal migratory phase occurs, and both centrifugal and centripetal migrations proceed with approximately normal timing along the dorsal muscle bands. These observations suggest that the intrinsic navigational program of the DTCs proceeds independently of the execution of the ventral-to-dorsal second migratory phase and that the centrifugal first phase can occur on the dorsal side as well as on the ventral side.
The leader cell in the development of the male gonad is the male linker cell (MLC), which is known to follow a migration pattern that differs from that of the hermaphrodite DTCs both in sequence and in timing (Sulston and Horvitz, 1977; Hedgecock et al., 1987). MLC migrations begin in the L2 stage with a migration along the ventral muscle band from the ventral midbody toward the head. At the end of L2, the MLC turns and migrates from the ventral to the dorsal muscle band one full larval stage earlier than the hermaphrodite DTCs. Subsequently, the MLC turns toward the posterior and migrates toward the tail, simultaneously descending from the dorsal to the ventral muscle bands. As for the hermaphrodite DTCs, the second, ventral-to-dorsal, migratory phase of the MLC is selectively disrupted by mutations in unc-5, unc-6 or unc-40 (Hedgecock et al., 1990).
unc-5 expression pattern
The products of the unc-5, unc-6 and unc-40 genes act together to mediate ventral-to-dorsal DTC migration (Hedgecock et al., 1990; see Fig. 2B). In principle, the transition from the centrifugal first phase of DTC migration to the dorsally oriented second phase could involve changes in the function of any component of the UNC-6 guidance system. UNC-6 is expressed by ventral epidermiblasts beginning in embryonic development and continues to be expressed at the time of the DTC migrations by neurons in the ventral nerve cord (Wadsworth et al., 1996). UNC-40, which functions cell-autonomously in migrating cells and neuronal growth cones, is expressed in the DTCs at a detectable and approximately constant level throughout their migrations (Chan et al., 1996). We examined unc-5 gene expression using lacZ and gfp as transcriptional reporters (Fig. 3) and we also tagged UNC-5 with GFP and with an HA epitope. The unc-5 gene encodes two SL1-spliced transcripts (A and B) which are identical except for an 82 nt first exon present only in the longer unc-5A transcript (Leung-Hagesteijn et al., 1992).
unc-5B transcriptional reporters were made by fusing 4.6 kb of genomic DNA upstream of exon 2 to gfp or lacZ reporter genes. This fragment contains sufficient regulatory sequence to rescue cell migration and axon guidance phenotypes of unc-5 mutants when fused to an unc-5 cDNA (Hamelin et al., 1993; M.-W. S. and J.G.C., unpublished data). unc-5B∷HA and unc-5B∷GFP translational reporters that rescued the unc-5 mutant defects were also made. Expression was observed in the hermaphrodite DTCs and MLCs as well as in all five classes of motorneurons that exhibit axon guidance defects in unc-5 mutants (DA, DB, DD, VD, AS; Fig. 3A,B). In addition, expression was observed in several classes of neurons in the head that are not visibly affected by unc-5 mutations. These included 12 sensory neurons of the OLQ, OLL and IL2 classes (Fig. 3C), and the interneurons RIA, RIH and ASE.
Expression of lacZ or gfp transcriptional reporters in the DA, DB and DD embryonic motorneurons was first detected in late stage embryos at about the time when these neurons are known to extend pioneer axons along the epidermis (aka, hypodermis) to the dorsal side (Fig. 3A). Expression in the DA and DB neurons was transient, and only DD staining persisted into larval and adult stages (Fig. 3B). Expression in larval motorneurons was also consistent with the role of UNC-5 in pioneer axon guidance. The VD and AS classes of motorneurons are born at the L1 to L2 molt, and extend axons toward the dorsal side soon thereafter. The VD motorneurons expressed intensely at this and at later times (Fig. 3B), while the AS neurons expressed faintly and sporadically (not shown). The expression of unc-5B reporters in cells known to require unc-5 for guided migrations is entirely consistent with previous mosaic analysis, which showed that UNC-5 acts cell-autonomously in the DTCs and motorneurons (Leung-Hagesteijn et al., 1992).
unc-5B is expressed in DTCs and MLCs at the time of the initiation of dorsalward turning
Reporter gene expression in the DTCs was never detected during the centrifugal first phase of migration (n=644; Fig. 4A,B). unc-5B∷lacZ and unc-5B∷unc-5HA expression were first detectable in DTCs at the positions where they normally begin to reorient in order to migrate toward the dorsal side (Figs 3D, 4A,B). Similar results were obtained in mig-4(rh51) mutants in which the gonad primordium was displaced dorsally (Fig. 2D, 4C). For the anterior DTC, this turning position is, with some variability, adjacent to the cell body of the VD4 motorneuron. Occasionally, DTC staining could be observed at positions posterior to VD4, but at no greater frequency than that with which the anterior DTC makes its turn at a position posterior to VD4 (data not shown). Staining was relatively intense in DTCs migrating from ventral-to-dorsal muscle bands and decreased thereafter during the centripetal third phase of migration. In the MLCs, β-gal staining was similarly detected beginning at the position at which the first 90° turn is initiated (data not shown). Importantly, both the HA-tagged and GFP-tagged unc-5B∷unc-5 constructs could rescue nearly completely the DTC migration defects of an unc-5 null mutant (M. T. K. and J. G. C., unpublished data).
To assess the timing of unc-5 expression in living animals, unc-5B transcriptional and translational GFP reporters were used. As with lacZ and HA-tagged reporter constructs, expression was never observed in the DTCs during the centrifugal first migratory phase. In contrast, however, expression of GFP was detectable only after the DTCs had begun their second (ventral-to-dorsal) phase of migration, then persisted until the end of the L4 stage. The delayed initial expression of the GFP reporters relative to the lacZ reporter gene may reflect the time required for the activation of the GFP fluorophore.
unc-5 expression is sufficient to induce dorsalward turning
The expression of unc-5B∷lacZ reporter constructs at the time of the initiation of the second phase of DTC migration, and the ability of tagged unc-5B∷unc-5 transgenes (which express at the time of the turn) to rescue unc-5 mutant DTC defects, suggest that unc-5 gene expression may initiate dorsalward turning. If this were the case, expressing the unc-5 cDNA in the DTCs at earlier times could induce precocious turning toward the dorsal side. The unc-5 cDNA was placed under the control of the 5′ regulatory sequence of the emb-9 gene. This promoter drives expression in the DTCs and other somatic cells throughout L3 and L4 (Graham et al., 1997). Most animals that carry the emb-9∷unc-5 transgene in an extrachromosomal or an integrated multicopy array (evIs99) exhibit precocious DTC turning (Fig. 1G-K; summarized in Fig. 2E). Consistent with a precocious ventral-to-dorsal migration superimposed upon the normal first-phase navigational program, DTCs that turn precociously follow an oblique angle (relative to the dorsoventral axis) to reach the dorsal side. They then continue to migrate centrifugally along the dorsal muscle band slightly variable distances, but usually distal to the normal position of the second turn, before initiating the centripetal third migratory phase (Figs 1G-K, 2E). The highest penetrance of this defect was observed in the evIs99 integrated line (Table 1).
In timed experiments (see Methods), it was found that, in evIs99(emb-9∷unc-5) hermaphrodites, 13 of 30 DTCs had turned obliquely by 28-29 hours after hatching. In contrast, in strains not carrying the emb-9∷unc-5 transgene, no DTCs turned prior to 29 hours after hatching (n=16), and ventral-to-dorsal migration was always immediately followed by a centripetal turn. The DTC phenotype in the evIs99 line was largely copy number dependent, as animals with one copy of the integrated array exhibited few precocious DTC migrations (Table 1). This indicates a strong sensitivity of this phenotype to the dose of prematurely expressed UNC-5. Precocious DTC migration was the only visible phenotype caused by the emb-9∷unc-5 transgene in hermaphrodites. In males, precocious MLC migrations were also induced by emb-9∷unc-5, but for reasons that are not understood, male gonad arms always failed to reflex toward the posterior and instead extended toward or into the head of the worm, resulting in sterility. This phenotype was also largely dependent on the copy number of the evIs99 array (data not shown).
Precocious migrations are suppressed by unc-6 and unc-5 mutations
Genetic, molecular and biochemical results suggest that UNC-6 is the ligand for the UNC-5 receptor, which is required for ventral-to-dorsal migrations on the epidermis (Hedgecock et al., 1990; Hamelin et al., 1993; Leonardo et al., 1997). The effects of unc-5 and unc-6 null mutations on the emb-9∷unc-5-induced precocious DTC migrations were examined. The precocious DTC migration phenotype in the evIs99 line was almost completely suppressed by a null mutation in unc-6 (Table 1). The evIs99; unc-6(ev400) double mutant was indistinguishable from unc-6(ev400) alone, with the centripetal third migratory phase frequently occurring ventrally. Marginally precocious migrations occurred in only 4 of 200 DTCs examined. The precocious ventral-to-dorsal DTC migrations are therefore dependent upon directional information provided by UNC-6.
Surprisingly, a null mutation in unc-5 also reduced the frequency of precocious migrations (Table 1). Whereas in the evIs99 line, 71% (156 of 220) of gonad arms made a precocious dorsalward turn, only 16% (32 of 200) of evIs99; unc-5(e53) gonad arms exhibited this phenotype. These results indicate that, although unc-5 reporter constructs are not expressed in the DTCs during the first migration phase, there is some basal level of expression undetectable with the reporter constructs. This phase one expression, though insufficient to drive ventral-to-dorsal migration, contributes to the precocious migrations in the emb-9∷unc-5 transgenic background. The sensitivity of the precocious DTC migrations in the emb-9∷unc-5 lines to the amount of UNC-5 expressed is also evidenced by the dependence of the phenotype on the copy number of the evIs99 integrated array (Table 1).
The evIs99; unc-5(e53) mutants also exhibited a partial suppression of the ventral misreflexion DTC defects of unc-5. In unc-5(e53) worms, 31% (34 of 110) anterior and 73% (80 of 110) posterior DTCs failed entirely to migrate from ventral to dorsal. In evIs99; unc-5(e53), only 6% (6 of 94) anterior and 20% (15 of 74) posterior DTCs that did not make a precocious dorsalward turn failed entirely to migrate to the dorsal side, indicating that the premature UNC-5 expression that was unable to initiate precocious ventral-to-dorsal migration could nevertheless drive the same migration at the appropriate time in development. This suggests the existence of a facilitating mechanism that acts with UNC-5 at the appropriate time to induce reorientations of DTCs.
unc-5B expression in mutants affecting DTC migrations
The expression of the unc-5B∷lacZ or unc-5B:GFP reporter gene was examined in various mutant strains in which DTC migrations are abnormal. Mutations that disrupt only the execution of the second phase of migration would not be expected to affect the expression of these reporter genes. In contrast, mutations that disrupt the developmental program of the DTCs might exhibit defects in unc-5B reporter gene expression. In mutants of unc-130 (E. B. Nash and J. G. C., unpublished data) unc-5, unc-6 or unc-40 (Hedgecock et al., 1990), the second migratory phase of the DTC frequently fails to occur (Fig. 2B). First and third phase migrations, which occur along the anteroposterior axis, however, are normal in timing and extent. The expression of unc-5B∷lacZ appeared normal in timing and intensity in these mutants, even though the second migratory phase failed to occur (data not shown). As described above, in dig-1 or mig-4 mutants, the entire gonad primordium is often displaced dorsally and the entire DTC migration program is carried out on the dorsal muscle bands. In these mutant backgrounds as well, unc-5B∷lacZ expression appeared normal (Figs 2D, 4C).
In mig-6(e1931) hermaphrodites, the gonad arms do not elongate normally (Hedgecock et al., 1987). There is still, however, ventral-to-dorsal DTC migration in many mutant animals. In 74% (34 of 46) late L4 and young adult hermaphrodites, dorsally positioned DTCs were observed. As in wild-type worms, dorsalward reorientation of the DTCs was never observed prior to Pn.p cell divisions (n=30), indicating that the timing of the ventral-to-dorsal phase of DTC migration was approximately normal. In addition, unc-5B∷gfp expression was frequently observed in these DTCs (Fig. 2C, data not shown). These observations suggest, in this mutant, the DTC developmental program occurs with normal timing even in the absence of DTC migrations along the anteroposterior axis.
In contrast to the above mutants, daf-12 and mig-8 mutants are thought to disrupt DTC migrations by blocking progression between larval stage-specific developmental programs (Antebi et al., 1998). Specifically, entry into the L3 or L4 programs is delayed or blocked. In daf-12(rh84) animals transgenic for the unc-5B∷lacZ reporter construct, β-gal staining was never observed in the DTCs, although neuronal staining appeared normal (Figs 4D, 5; summarized in Fig. 2F). This is consistent with the phenotype of daf-12(rh84), as normal gonad morphologies were observed in only 10/304 (3%) of gonad arms. In the remaining animals, the DTCs either did not turn or did so only at the extreme ends of the animal. It is possible that some low level of unc-5 expression, undetectable with these methods, underlies these infrequent, usually delayed ventral-to-dorsal migrations.
In mig-8(rh50) worms, only 13% (52 of 384) of DTCs migrated from ventral to dorsal at the normal time, as judged by gonad morphologies (data not shown). Of these 52 DTCs, 14 expressed unc-5B∷lacZ. As for daf-12(rh84), most DTCs failed to turn dorsalward at the appropriate time and instead continued toward the head or tail. Some of these (179/334=53%) eventually moved dorsalward at an abnormal anteroposterior position. However, only 18 of these 179 exhibited β-gal staining (Fig. 4E). None of those DTCs that failed to migrate toward the dorsal side exhibited staining. Thus, only daf-12 and mig-8 mutations, which are the only mutations known to block or delay the advancement of stage-specific programs affecting the DTCs, are defective in unc-5 expression. These results are consistent with the hypothesis that daf-12 and mig-8 are components of a developmental timing mechanism, which among other things, regulates expression of UNC-5 in the DTCs to make them responsiveto UNC-6 and change their trajectory. Mutations in other genes that affect only one stage of DTC migration (e.g., unc-5, unc-6, unc-40, unc-130, dig-1 and mig-4) all express UNC-5 with normal timing, consistent with the idea that they affect the execution of specific migratory phases, but do not affect the timing program that regulates DTC turning.
DISCUSSION
Previous work has demonstrated that changes in the expression of extracellular cues or of cell surface receptors can initiate cell migrations or cell shape changes. For example, temporal changes in the spatial expression of the branchless FGF homologue or the Robo receptor in Drosophila direct tracheal and axon extensions, respectively (Sutherland et al., 1996; Seeger et al., 1993; Kidd et al., 1998a,b). Alternatively, Drosophila border cell migrations require the transcription factor encoded by the slow border cells locus (slbo; Montell et al., 1992) to activate directly the expression of the breathless FGF receptor and thereby initiate migration (btl; Murphy et al., 1995). Post-transcriptional mechanisms have been proposed to mediate changes in the repertoire of cell-surface integrins in response to changes in ligand concentrations (Condic and Letourneau, 1997). Finally, ectopic expression of the UNC-5 netrin receptor in neurons that do not normally express that receptor makes the growth cones of these neurons responsive to UNC-6 (Hamelin et al., 1993). We have extended these findings to demonstrate that the UNC-5 receptor is a critical target for regulation of responsiveness to UNC-6 during the first turn of the DTCs. An understanding of the regulatory pathways directing DTC migrations may shed light on complex cell migrations in other systems, including the developing mammalian cerebellum, where granule cell precursors require the Unc5H3 netrin receptor for correct positioning (Ackerman et al., 1997). In addition, the UNC-6 netrin pathway is utilized in the complex migrations of many neuronal growth cones, and principles of guidance may be shared between cells and growth cones.
The triphasic pattern of DTC migrations is tightly synchronized with the progression of larval development (Antebi et al., 1998). The dorsalward turning of the DTCs occurs about 30 hours after hatching, coincident with other developmental events, such as the cell divisions of the Pn.p vulval precursor cells. The coordinated regulation of the development of gonadal and non-gonadal tissues is evidenced by the arrest of DTC migrations upon entry of the worm into the dauer larval stage. Antebi et al. (1998) have proposed that the steroid hormone receptor DAF-12 is part of a mechanism that advances developmental programs in all somatic tissues of C. elegans, including the gonad. Although the identity and source of the DAF-12 ligand are unknown, a diffusible steroid hormone could act simultaneously on many cells to synchronize cell-intrinsic larval stage advancement mechanisms. Mutations in daf-12 or in mig-8 can result in the failure of the DTCs to initiate phases 2 or 3 of migration. This is interpreted as the absence or the delay of entry into the normal L4 developmental program (Antebi et al., 1998). Consistent with this, we have found that, in these mutants, but not in mutants affecting other aspects of DTC migrations, transcriptional upregulation of unc-5 at the appropriate time fails to occur.
Navigational programs of migrating cells may simultaneously regulate more than one guidance mechanism. For example, transgenic expression of btl in Drosophila border cells under a conditional promoter rescues the slbo mutant phenotype (Murphy et al., 1995); however, the rescue is incomplete, suggesting that btl is probably not the sole target of regulation by the slbo protein. Similarly, the advancement in developmental stage mediated by daf-12 and mig-8 does not affect the DTC migrations exclusively through unc-5. The ventral-to-dorsal migration defects in null mutants of daf-12 or mig-8 are of higher penetrance than those of null mutants of unc-5 or unc-6 (Hedgecock et al., 1990; Antebi et al., 1998). Thus, unc-5- and unc-6-independent mechanisms of guidance along the dorsoventral axis must also be affected. In addition, as noted above, unc-5 and unc-6 mutations, unlike those in daf-12 and mig-8, do not disrupt the third, centripetal phase of migration. We propose that, at the time of DTC re-orientation, guidance systems acting in both the longitudinal and the dorsoventral axes must be turned on or off. As summarized in Fig. 6A, the DAF-12/MIG-8-dependent turning involves the activation of both UNC-6-dependent and UNC-6-independent dorsoventral guidance systems. In addition, DTCs alter their responses to longitudinal cues and re-orient from centrifugal to centripetal migration. These other guidance mechanisms in DTC migrations are not yet understood and their regulation need not occur in the same manner as the UNC-6 guidance system. For example, some heterochronic mutants that, unlike daf-12, affect larval stage advancements only of extragonadal tissues, do not themselves cause DTC migration defects, but act as strong enhancers of the DTC migration defects caused by unc-5 null mutations (D. C. M., unpublished data). In contrast to the DTC-intrinsic mechanism by which DAF-12 affects the UNC-6 guidance system, temporal regulation of UNC-6-independent dorsoventral guidance may involve changes in some DTC-extrinsic guidance factor, possibly a directional cue.
Mosaic analysis has previously demonstrated that UNC-5 is required cell autonomously by cells and neurons that are repelled by UNC-6, including the DTCs (Leung-Hagesteijn et al., 1992). The shorter transcript of the unc-5 gene begins with exon 2, and this transcript (unc-5B), along with its upstream regulatory sequence, is able to rescue completely the known cell migration and axon guidance defects of unc-5 mutants (Hamelin et al., 1993). Analysis of the timing of reporter construct expression suggests that expression of the unc-5B transcript by the DTCs begins as the DTCs reorient from a longitudinal to a dorsally oriented trajectory. However, genetic evidence indicates that some low level of UNC-5 is actually present in the DTCs prior to the initiation of turning, and that endogenous UNC-5 contributes to the precocious migrations caused by premature (emb-9∷unc-5) transgenic UNC-5 expression. We propose that the unc-5B reporter constructs contain an enhancer element that is, directly or indirectly, responsive to a signal from DAF-12. The phase one expression of unc-5 that we detect genetically may not be visible using our reporters since the reporter constructs, which are based on 5′ regulatory sequence, are lacking large intronic fragments that could contain additional regulatory elements. Thus, either unc-5 transcript (A or B) could normally be expressed in the DTCs prior to their ventral-to-dorsal (second) phase of migration. We interpret this as evidence that inhibiting or competing pathways must act during the first phase of migration to prevent precocious responses to UNC-6 through UNC-5 and UNC-40. This is consistent with the observation that premature transgenic expression of UNC-5 in the DTCs causes precocious dorsalward migrations in an UNC-5 dosage-sensitive manner. The presence of high levels of UNC-5 is required to initiate turning, while low levels are insufficient to cause precocious turning.
Constitutively expressed transgenic UNC-5 largely rescued the DTC defects of a null unc-5 mutant, although the frequency of precocious migrations was reduced. Therefore, UNC-5 expression that was insufficient to initiate precocious dorsalward turning during the first migration phase is able to do so at the normal time of reorientation. Therefore, as summarized in Fig. 6B, in addition to transcriptional regulation of unc-5, DAF-12 activity may also induce a mechanism that facilitates UNC-5 function during the initiation of turning. The nature of this mechanism is unknown. It presumably is not involved in the transcriptional regulation of unc-5 but it could affect UNC-5 post-transcriptionally. For example, it could be a mechanism that inhibits the UNC-5 inhibitory (or opposing) pathway that operates during phase 1 as proposed above. Alternatively, it may be a novel guidance mechanism that acts in parallel with UNC-5 and UNC-6.
These studies have thus far focussed only on the transition between the first and second phases of DTC migrations. The initiation of motility in the DTCs, the transitions between the second and third migratory phases and the cessation of motility should also be accessible to genetic analysis. The combined analyses of cell migration mutant phenotypes and reporter construct expression may be used to distinguish mutations affecting the migration program (or its advancement) from those affecting the execution of the program. It will thus be possible to test the model for the regulation of the UNC-6/netrin system proposed above for its applicability to other transitions in DTC migrations. If such a scheme is generally applicable to cell and growth cone migrations, it may shed light on the mechanisms of cell migration in development and organogenesis in general.
We thank J. Kaplan for the gift of the nuIs9 strain, and Lou Siminovitch, S. Cordes, C. Todoroff, N. Levy-Strumpf and R. Ikegami for comments on the manuscript. This work was supported by postdoctoral fellowships from the National Cancer Institute of Canada to D. C. M., the MRC of Canada to M.-W. S. and M. T. K., and grants from the SCRF (Spinal Cord Research Foundation), MRC and NCI of Canada to J. G. C.