The Ras signaling pathway specifies a variety of cell fates in many organisms. However, little is known about the genes that function downstream of the conserved signaling cassette, or what imparts the specificity necessary to cause Ras activation to trigger different responses in different tissues. In C. elegans, activation of the Ras pathway induces cells in the central body region to generate the vulva. Vulval induction takes place in the domain of the Hox gene lin-39. We have found that lin-39 is absolutely required for Ras signaling to induce vulval development. During vulval induction, the Ras pathway, together with basal lin-39 activity, up-regulates lin-39 expression in vulval precursor cells. We find that if lin-39 function is absent at this time, no vulval cell divisions occur. Furthermore, if lin-39 is replaced with the posterior Hox gene mab-5, then posterior structures are induced instead of a vulva. Our findings suggest that in addition to permitting vulval cell divisions to occur, lin-39 is also required to specify the outcome of Ras signaling by selectively activating vulva-specific genes.
Two well-conserved strategies for determining patterns of cellular differentiation during development are the use of Ras-mediated intercellular signals to specify particular cell fates and the use of homeotic selector (Hox) genes to specify regional identity. Ras is a central member of the well-conserved receptor tyrosine kinase/Ras/MAP-Kinase signaling cassette (reviewed by Egan and Weinberg, 1993). In C. elegans and Drosophila, the Ras pathway regulates pattern formation in many tissues, including vulval cell fates in C. elegans, photoreceptor type in the Drosophila eye, and terminal fates in the Drosophila embryo (reviewed by Duffy and Perrimon, 1994; Eisenmann and Kim, 1994; Kayne and Sternberg, 1995; Wassarman et al., 1995). However, it is still not clear what gene products provide specificity to the Ras pathway; that is, what gene products determine the type of structure made in response to activation of the Ras pathway. Furthermore, few genes that act downstream of the conserved components of the pathway have been identified.
The Hox genes are best known for their role in specifying anterior/posterior (A/P) pattern in embryos (reviewed by McGinnis and Krumlauf, 1992; Botas, 1993; Krumlauf, 1994; Lawrence and Morata, 1994; Salser and Kenyon, 1994). These genes are expressed in broad stripes along the A/P axis, where they specify regional identity in all metazoans examined so far. Misexpression can cause homeotic transformations of one body region to another in Drosophila and C. elegans, and can cause homeotic skeletal defects in vertebrates. In both Drosophila and C. elegans, Hox genes are expressed dynamically, and this dynamic expression reflects a requirement for Hox genes at multiple times during development to specify diverse fates (Castelli and Akam, 1995; Salser and Kenyon, 1996). Both the wingless (Hoppler and Bienz, 1995) and Sonic hedgehog (Roberts et al., 1995) signaling molecules have been implicated in control of later Hox expression, but much remains to be learned about how dynamic expression is controlled.
Here we have investigated an interaction that takes place between Hox genes and the Ras pathway during cell fate specification in C. elegans vulval development. The Hox gene lin-39, the C. elegans homolog of Drosophila sex-combs reduced (scr), specifies cell fates in the mid-body of the worm, where the vulva is located (Clark et al., 1993; Wang et al., 1993). The vulva arises from a set of six vulval precursor cells (VPCs), P3.p-P8.p, each of which has the potential to adopt a 1°, 2°, or 3° fate (Sulston and Horvitz, 1977; Sternberg and Horvitz, 1986). 1° and 2° cells divide multiple times to generate cells that will form the vulva; 3° cells are non-vulval: they divide once and join the epidermal syncytium. Vulval fates are specified both by a graded inductive signal from the gonadal anchor cell (AC) and by lateral signals between the VPCs (Fig. 1A) (Katz et al., 1995; Koga and Ohshima, 1995; Simske and Kim, 1995). The AC mediates vulval induction by producing a ligand that activates a conserved EGF-receptor/Ras/MAPK pathway in the VPCs (Fig. 1B) (Aroian et al., 1990; Beitel et al., 1990; Han and Sternberg, 1990; Hill and Sternberg, 1992; Eisenmann and Kim, 1994; Lackner et al., 1994; Wu and Han, 1994; Beitel et al., 1995; Kayne and Sternberg, 1995). A second signaling pathway involving lin-12, a Notch homolog, mediates lateral signaling between VPCs (Greenwald et al., 1983; Greenwald, 1985; Sternberg, 1988; Yochem et al., 1988).
Because the Hox gene lin-39 is expressed in and functions to pattern the body region that generates the vulva, we suspected that it might be required for vulval development, and decided to investigate what its role might be. A recent study of the role of Hox genes in vulval development suggested that lin-39 may be required for the VPCs to respond to Ras activation, and that the posterior Hox gene mab-5 may antagonize vulval development (Clandinin et al., 1997). However, it has been difficult to assess the full role of lin-39 in vulval development because lin-39 is required for generation of the VPCs long before vulval induction takes place. In lin-39 null mutants, the VPCs fuse with the surrounding epidermal syncytium and therefore cannot generate a vulva later in development.
We have used a heat-shock-lin-39 construct to overcome the early fusion defect in lin-39(−) animals. This has enabled us to determine that lin-39 plays a central role in vulval induction. At the time of vulval induction, the Ras pathway and preexisting LIN-39 protein together up-regulate lin-39 gene expression. lin-39 activity, in turn, is absolutely required in order for Ras signals to induce the vulva. If lin-39 activity is not present at the time of Ras signaling, no divisions take place. In addition to permitting vulval cell divisions to occur, lin-39 contributes specificity to the Ras pathway: expression of the incorrect Hox gene causes spatial homeotic transformations in the patterns that are induced in response to Ras activation. Together our findings suggest that in the VPCs, lin-39 functions both in parallel to and downstream of the Ras pathway to selectively activate the expression of genes specific for vulval development.
MATERIALS AND METHODS
General methods and strains
Strains were maintained using standard methods (Brenner, 1974; Wood, 1988). Heat shock strains and mab-5(e1751gf) were maintained at 20°C; other strains were analyzed at 25°C. The following mutant alleles (Wood, 1988; or referenced below), were used:
LGI: pry-1(mu38) (J. N. M., unpublished). LGII: let-23(sy97) (Aroian and Sternberg, 1991). LGIII: dig-1(n1321) (Thomas et al., 1990), mab-5(e1239), mab-5(e2088), mab-5(e1751gf) (Hedgecock et al., 1987; Salser and Kenyon, 1992), lin-39(mu26) (Wang et al., 1993), lin-39 (n709) (Clark et al., 1993), lin-39(n1760) (Clark et al., 1993). LGIV: lin-3(n378), let-60(n1046gf) (Ferguson and Horvitz, 1985; Beitel et al., 1990; Han et al., 1990), unc-22(s7), lin-1(e1026), dpy-20(e1282). LGV: him-5(e1467), him-5(e1490). LGX: lin-15(n309), muIs9[hs-mab-5 unc-31(+)] (Salser et al., 1993). Not yet assigned to a linkage group: muIs23[hs-lin-39 dpy-20(+)] (Hunter and Kenyon, 1995).
PCR was used to delete the majority of the homeodomain (from N-175 to K-225) of a lin-39 cDNA (1514A) (Wang et al., 1993); the resulting fragment was cloned into the (His)10-containing pET16b (Novagen). The fusion protein was expressed and purified on Ni-NTA-agarose (Qiagen), and used to immunize two rabbits. Antibodies were affinity purified using LIN-39-(His)10 protein immobilized on Ni-NTA-agarose; antibodies were eluted with Actisep (Sterogene). Purified antibodies show no reactivity against lin-39(mu26) or lin-39(n1760) larvae, and stain the nuclei of hs-lin-39 worms brightly. Before use, affinity-purified antibodies were pre-adsorbed to lin-39(n1760) larvae as follows: n1760 larvae were rinsed in dH2O, placed in 100 µl aliquots in tubes in a 95°C PCR block for 10 seconds, and frozen in liquid N2. Frozen worms were added to 15 ml tubes filled with methanol, incubated at room temperature for 1 minute, rinsed 3× in TBSTwE (137 mM NaCl, 2.7 mM KCl, 25 mM Tris at pH 7.7, 0.5% Tween 20, 5 mM EDTA), and placed in 2 volumes of block (1% BSA (Sigma #A9306), 5% swine serum (Cappel)). Worms were sonicated, and incubated at 37°C for 30 minutes. Purified antibody was diluted 1:20 (final) and preadsorbed for 2 hours at 37°C. Debris was removed by spinning at 14,000 r.p.m. in an Eppendorf microfuge for 30 minutes, and the supernatant was used for immunostaining.
For LIN-39 staining, larvae were rinsed 3 times with dH2O. 20 µl of larvae were spotted onto poly-lysine-coated slides and covered with an 18 mm2 coverslip. Slides were placed on a 95°C aluminum block (on a PCR machine) for 5 seconds, and then allowed to cool on the lab bench for <1 minute. Excess water was removed by aspiration, and then slides were frozen on an aluminum block on dry ice. Coverslips were pried off and the slides were placed in 25°C methanol for 3 minutes, rinsed 2× in TBSTwE, and then blocked for 30 minutes at 37°C. Larvae were incubated with primary antibody for 1-2 hours, washed 3× 20 minutes in TBSTwE, incubated with rhodamine-labelled donkey anti-rabbit IgG (1:50; Jackson Labs) for 1 hour, washed as above, taken through a TBSTwE/glycerol series, and mounted in 80% glycerol, 2% n-propyl gallate, 5 μg/ml DAPI.
For MH27 antibody labelled staining, fixation was as above, except that antibodies were diluted 1:100, and goat anti-mouse IgG (Cappel) was used as the secondary antibody.
Microscopy and laser ablation
Larvae were mounted (Wood, 1988) and observed using DIC (Nomarski) optics. Pn.p cells were observed beginning with the Pn.px or Pn.pxx stage. VPC fate assignment was done as described by Katz et al. (1995).
To observe autofluorescence of the hook, worms were viewed on a Zeiss Axiophot microscope fitted with a mercury lamp, using a standard fluorescein filter set.
Cell ablations were performed with a laser microbeam by standard methods (Bargmann and Avery, 1995). Z1 and Z4 ablations were performed within 4 hours of hatching. Control, unablated, animals were recovered from the same slides. Ablated animals were identified by the absence of germline nuclei as assayed by DAPI staining after fixation.
Heat shock of transgenic animals
A PTC-100 thermal cycler (MJ Research) was used to control heat pulses. 0- to 2-hour old larvae were placed on 35 mm NG plates that had been sanded to remove the ridges from the bottom of the plate for better thermal coupling. A 35 mm plate with a feedback temperature probe embedded in 4 ml of 2% agarose was used to increase reproducibility. Plates were placed directly on the 60-well thermal block (coated with mineral oil), and the machine was programmed as described below.
To assay vulval development in the absence of lin-39 (Table 1B, and Fig. 3, lin-39; lin-1 double mutants), we used the following program (times are given as hours:minutes): (1) 20°C 3:30; (2) 31°C 0:11; (3) 20°C 3:40; (4) back to step 2, 5 more times; (5) 20°C.
Variations of this program were used to provide lin-39 later in development. A seventh, 11 minute, 31°C heat pulse was added either 3 hours and 40 minutes or 8 hours after the 6th pulse (Table 1C), or an 8 minute, 33°C heat pulse was added 2 or 3 hours after the 6th 31°C pulse (Table 1D).
To provide high levels of lin-39 after Z1, Z4 ablation, animals were allowed to develop at 20°C for 21-23 hours (end of L2) or 23-25 hours (early L3) and then were given two 8 minute, 33°C heat pulses with 3 hours separating each pulse.
To provide lin-39 to males (Figs 4B, 5D), we used the program: (1) 20°C 2:00 or 4:00; (2) 31°C 0:11; (3) 20°C 3:40; (4) back to step 2, 2 more times; (5) 32°C 0:10; (6) 20°C 2:40; (7) back to step 5, 2 more times; (8) 31°C 0:11; (9) 20°C 3:40; (10) back to step 8, 3 more times; (11) 20°C.
To provide mab-5 to the central VPCs, and to reduce the amount of lin-39(n709ts) activity at the time of induction (Figs 4D, 5C), we used the program: (1) 15°C 38:00; (2) 20°C 2:00; (3) 31°C 0:11; (4) 25°C 2:45; (5) back to step 3, 1-3 more times. Step 2 was included because shifting from 15°C to 31°C directly was lethal.
Ras signaling increases lin-39 expression in the VPCs
The Hox gene lin-39 is known to be required early in development for generation of the VPCs (Clark et al., 1993; Wang et al., 1993). To determine whether lin-39 might also be required later, at the time of vulval induction, we first used anti-LIN-39 antibodies to ask whether lin-39 was expressed in the VPCs at this time. We found that before vulval induction occurs, lin-39 was expressed uniformly in the VPCs. Laser ablation of the AC has shown that the AC signals the VPCs to initiate vulval development during early L3 (Kimble, 1981). We found that at the time of vulval induction, lin-39 expression increased dramatically in P6.p, the VPC closest to the AC, which adopts the 1° vulval fate. P5.p and P7.p, which are further from the AC and which adopt 2° vulval fates, showed lower expression levels. The cells adopting the non-vulval 3° fate, P3.p, P4.p, and P8.p, showed the lowest levels of lin-39 expression (Fig. 2A). Three experiments demonstrated that this expression pattern was governed by AC signals and the Ras pathway. First, the dig-1(n1321) mutation, which displaces the AC anteriorly, displaced the peak of lin-39 expression coordinately (Fig. 2B). Second, ablation of the cells that generate the AC abolished the peak of expression (Fig. 2C). Third, Vulvaless (Vul) mutations that reduce Ras signaling decreased peak levels of Hox gene expression (Fig. 2D), whereas Multivulva (Muv) mutations that ectopically activate the signaling pathway caused strong expression in all VPCs (Fig. 2E).
How do Ras signals control LIN-39 levels? We have observed that hs-LIN-39 protein is expressed and decays uniformly in the VPCs (data not shown), suggesting that Ras signals regulate lin-39 transcription or translation, rather than degradation. Together, these findings showed that lin-39 expression is up-regulated by the Ras signaling pathway, and suggested that lin-39 might be required at the time of vulval induction.
lin-39 and Ras signaling are both required for vulval induction
In lin-39(−) mutants, cells that would normally become VPCs instead fuse with the surrounding epidermal syncytium well before the time of vulval induction (Clark et al., 1993; Wang et al., 1993). Therefore, to determine whether lin-39 activity was also required later, at the time of vulval induction, we used a heat-shock-lin-39 chimeric construct (hs-lin-39) to control lin-39 expression levels in lin-39(−) animals. We gave repeated heat pulses early in development, but stopped the heat pulses to allow LIN-39 levels to drop by the time of vulval induction. Under these conditions, many VPCs remained unfused; however, we found that they adopted non-vulval fates (Table 1B, especially note P6.p descendants). Later and stronger heat pulses restored wild-type vulval development (Table 1C,D). Thus, lin-39 is not only required early to prevent fusion of the VPCs, it is also required later to allow VPCs to generate vulval cell lineages.
Where in the Ras signaling pathway does lin-39 act? In C. elegans, as in other organisms, the Ras signaling cascade leads to modification of an ETS domain transcription factor (Beitel et al., 1995). The ETS homolog that functions during vulval development is the gene lin-1, which acts at the downstream-most position of the conserved signaling cassette to inhibit vulval development. Activation of the Ras pathway by AC signals leads to inactivation of lin-1, which, in turn, promotes vulval development. We found that lin-39 expression was increased in lin-1(−) mutants (Fig. 2E). This finding suggested that control of lin-39 expression by the Ras pathway was mediated, directly or indirectly, by lin-1, and identified lin-39 as a downstream target of the Ras signaling cassette.
Animals lacking lin-1 activity have a Multivulva phenotype: in lin-1(−) mutants, all the VPCs generate vulval cell lineages in an anchor-cell independent fashion. Because lin-39 expression increases in lin-1(−) mutants, it seemed likely that LIN-39 protein would act downstream of lin-1, and thus be required for the Multivulva phenotype of lin-1. To test this, we used early pulses of hs-lin-39 in lin-39(−); lin-1(−) mutants, as described above, to prevent VPC fusion, and then stopped the heat pulses to allow LIN-39 levels to fade. No vulval development took place (68/70 cells adopted the non-vulval, 3° fate). This indicated that lin-39 activity is required for the lin-1 Multivulva phenotype, and thus for aspects of vulval development that take place after the Ras pathway inactivates lin-1.
We next investigated whether high levels of lin-39 expression might be sufficient to induce vulval development in the absence of Ras signaling. To test this, we inactivated the Ras pathway by killing the anchor cell, and then administered high, uniform levels of LIN-39 by using hs-lin-39. We found that no vulval cell divisions occurred (n=37). In addition, when high uniform levels of LIN-39 were administered in animals with intact anchor cells, no ectopic vulval induction was seen (Table 1D). These findings indicated that lin-39 alone cannot trigger vulval development. Thus, the Ras pathway must have other functions in vulval development in addition to inducing lin-39 expression. These findings also ruled out a possible role for lin-39 that had been suggested by the graded lin-39 expression pattern; namely, that, different levels of lin-39 are used to specify alternative 1°, 2°, or 3° vulval fates.
lin-39 specifies its own up-regulation in response to Ras signaling
The finding that lin-39 expression was up-regulated by the Ras pathway was surprising, and it also raised an apparent paradox: Ras is required in many cells during C. elegans development (Han et al., 1990; Han and Sternberg, 1990; Chamberlin and Sternberg, 1994), but Ras signaling up-regulates lin-39 only in the VPCs, not in other cells. What specifies that lin-39 is controlled by Ras signaling in the VPCs? lin-39 itself is expressed at low levels in the VPCs before Ras activation (Fig. 3A); thus, one possibility was that basal lin-39 activity may be required for up-regulation of lin-39 in response to the activation of Ras. We tested this hypothesis using n2110, a lin-39 allele that greatly reduces gene activity but still expresses wild-type levels of LIN-39 protein in many cells as detected by immunofluorescence. We expressed hs-lin-39 early to prevent fusion of the VPCs and then examined the pattern of endogenous lin-39(n2110) expression at the time of AC signaling. We found that in n2110, lin-39 expression was nearly uniform in the central VPCs, P5.p, P6.p, and P7.p (Fig. 3B; compare to hs-lin-39 animals with lin-39(+) in Fig. 3C). The n2110 mutation is a single base change in the homeodomain coding region (Clark et al., 1993) and thus probably acts by reducing the activity of LIN-39 protein. Therefore early basal levels of lin-39 activity appear to be required for up-regulation of lin-39 expression in response to Ras signaling.
lin-39 contributes specificity to the Ras signaling pathway
How Ras signaling initiates different developmental programs in different tissues is not well understood. Hox genes function as homeotic selector genes during development, distinguishing between alternative cell fates in a position-specific manner. We therefore hypothesized that, in addition to allowing further development to take place, lin-39 might also influence the type of structure generated by the VPCs in response to activation of the Ras pathway. The male equivalence group, consisting of P(9-11).p, the posterior homologs of the hermaphrodite VPCs, gives rise to the hook, a structure used by the male when mating, and the pre-anal ganglion, a group of neurons important for mating (referred to here collectively as the pre-anal group or PAG). There are many similarities between the development of the PAG and the vulva: both are generated from Pn.p cells that undergo three rounds of division at the end of L3, and both are affected by the lin-12 lateral signaling pathway (Greenwald et al., 1983). In addition, both structures are affected by mutations that perturb Ras signaling, although the role of Ras in PAG development is more limited than in vulval development (Paul Sternberg, personal communication; data not shown).
The vulval precursors express the Hox gene lin-39, whereas the PAG precursors express mab-5, the Hox gene that functions in the body region posterior to the lin-39 domain (Salser et al., 1993). Because these structures develop from similar cells and are patterned by some of the same signaling systems, we reasoned that Hox genes might be required to specify alternative organ types. We tested this hypothesis in two ways: first by mis-expressing lin-39 in the posterior and second by mis-expressing mab-5 in the central body. We expressed hs-lin-39 in mab-5(−) males to determine if lin-39 could cause posterior male cells to adopt vulval instead of PAG fates. We used three vulva-specific features to assay the effects of lin-39 mis-expression. First, morphology: vulval cells invaginate in a unique way (compare Fig. 4A and C); second, plane of cellular division: some vulval cells, known as T cells, divide along the transverse (L/R) axis, whereas PAG cells do not (Fig. 5A,B); and third, detachment from the cuticle: a subset of both vulval and PAG cells detach from the cuticle in characteristic patterns (attached cells are underlined in Fig. 5A,B,D; see legend). We found that ectopic expression of hs-lin-39 in the PAG-precursor cells caused their descendants to form vulva-like invaginations (Fig. 4B), induced some T divisions characteristic of vulval cells, and caused detachment patterns reminiscent of those seen in the vulval secondary lineage (Fig. 5D). Although the transformation was not complete, we conclude that replacing mab-5 with hs-lin-39 can cause posterior male equivalence group cells to adopt vulval characteristics.
We next asked whether the reciprocal situation, expressing mab-5 in the normal lin-39 domain, might cause vulval cells to adopt PAG fates. One of the male equivalence group descendants generates the hook, an arrowhead shaped structure used by males in mating. The hook is autofluorescent when illuminated with light of the appropriate wavelength (Link et al., 1988; see Materials and Methods) (Fig. 4F) whereas the vulva is not (Fig. 4E); thus, autofluorescence can be used as a marker for the hook. We looked for autofluorescence in the vulval region of hermaphrodites in two strains that ectopically express mab-5 in the mid-body (Salser and Kenyon, 1992; J. N. M. and C. K. in preparation). About 4% of mab-5(e1751gf) and 24% of pry-1(mu38) mutant hermaphrodites show autofluorescence, in some cases hook-shaped, near their vulva (Fig. 4G; data not shown). The autofluorescence is likely due to mab-5, since no autofluorescence is seen in a pry-1(mu38); mab-5(e2088lf) double mutant (J. N. M. and C. K., in preparation). Furthermore, no hook-like autofluorescence was seen in mab-5(e1751gf) hermaphrodites after ablation of the AC precursors, which shows that AC signals are needed for ectopic hooks (n=165, P<0.003), and suggests that mab-5 specifies ectopic hook development by changing the fates of cells generated in response to the Ras pathway. We did not see any changes in the pattern of vulval divisions in these animals. However, these experiments were performed in the presence of wild-type lin-39, which we thought might be interfering with the ectopic mab-5.
To ask whether ectopic mab-5 expression might cause a more complete transformation in the absence of lin-39, we decided to remove lin-39 activity and then examine the effects of ectopic mab-5 expression on vulval development. To do this, we used lin-39(n709), a temperature-sensitive allele, along with a hs-mab-5 construct. Young worms were grown at the permissive temperature (15°C) to allow lin-39 activity to keep the VPCs unfused. Shortly before vulval induction, the temperature was raised to 25°C to reduce lin-39 activity, and then 31°C heat pulses were given to induce hs-mab-5 expression. This regime resulted in a dramatic transformation of vulval to PAG fates as assayed by several criteria. Unlike wild-type vulval cells, the transformed cells did not invaginate (Fig. 4D). In contrast to wild-type or control (n709) vulval lineages, no T divisions occurred; instead, hs-mab-5 promoted a pattern of L divisions similar to that seen in PAG lineages (Fig. 5C). In three of four hs-mab-5 animals, P6.p underwent a division pattern characteristic of P11.p in the PAG: one daughter divided obliquely (* in Fig 5), and the other daughter generated a descendant that did not divide again (N in Fig. 5). Two animals produced a P5.p granddaughter that underwent an additional division, resulting in a P5.p lineage identical to that of a normal P10.p in the PAG. These division patterns were never seen in wild-type or control vulval lineages. Thus replacing lin-39 with mab-5 causes a transformation of vulval cell fates to PAG cell fates. Together with the hs-lin-39 data, these results suggest that lin-39 is required in the vulva to specify that vulval fates are adopted as an outcome of Ras signaling.
In this study, we have examined the role of the Hox gene lin-39 in the development of the C. elegans vulva. We have found that this Hox gene is a downstream target of the Ras pathway, that it is required for all vulval cell divisions, and that it imparts specificity to the signaling system.
Ras signaling can up-regulate lin-39 expression
Mechanisms for controlling Hox expression in fields of cells late in development are now beginning to be identified. For example, in the chick, expression of Hox genes in the hindgut is under control of Sonic hedgehog (Roberts et al., 1995) and, in Drosophila, Wnt signaling controls the expression of the Hox genes Ultrabithorax in the visceral mesoderm and labial in the midgut endoderm (Thuringer and Bienz, 1993; Hoppler and Bienz, 1995). Our findings identify a different signaling pathway, Ras, as being able to control Hox expression late in development. Given that the Ras pathway controls lin-39 expression during vulval development, and the prevalence of the EGF/Ras pathway in cell fate specification, it seems possible that Hox genes will be found under EGF/Ras control in other organisms as well.
lin-39 is required for vulval development
Little is known about how activation of the conserved Ras signaling cassette leads to specific development of the C. elegans vulva; this work shows that the Hox gene lin-39 plays an important role in this process. We have shown that lin-39 is absolutely required in order for Ras signaling to induce a vulva: without lin-39 activity, no VPC cell division or vulval development takes place. Our genetic analysis suggests that, at a functional level, lin-39 acts in parallel to the Ras pathway during vulval development. lin-39 activity is still required for vulval development even if lin-1, the transcriptional regulator that is inactivated by Ras signaling, has been eliminated by mutation. Conversely, the Ras pathway is still required for vulval development even if high, uniform levels of lin-39 are administered artificially. In addition, since lin-39 expression is also up-regulated by the Ras pathway, lin-39 can also be considered to be a downstream target of the Ras pathway.
lin-39 imparts specificity to Ras signal transduction during vulval development
lin-39 can promote vulva-like lineages when expressed ectopically in the male tail, and the posterior Hox gene mab-5 can transform the vulval cells, causing them to make posterior structures instead. These findings indicate that lin-39 instructs the VPCs to adopt vulval fates, rather than an alternative fate, in response to Ras signals. It is important to emphasize that the homeotic transformations that we observed when lin-39 was expressed in the posterior body region were not complete. For this reason, we believe that additional specificity factors are likely to be required as well.
How might lin-39 confer specificity to the Ras pathway? As described above, it was known previously that AC signals lead to the inactivation of lin-1, and that lin-1 inhibits vulval development, probably by acting as a transcriptional repressor. Our findings indicate that lin-1 represses lin-39. In addition, lin-1 must repress other genes, since ectopic lin-39 expression cannot overcome the requirement for AC signals. One attractive model to explain these findings is the following: in the absence of AC signals, lin-1 represses expression of three types of genes: (i) lin-39, (ii) genes directly involved in vulval development, and (iii) genes with unrelated functions (for example, those involved in PAG fates) (Fig. 6A). When lin-1 is inactivated by AC signaling, all three types of genes become competent for expression. Some LIN-39 protein is already present even in the absence of Ras signaling, and this basal LIN-39 is able to activate strong expression of the lin-39 gene in the absence of lin-1. In addition, LIN-39 is now able selectively to activate those genes that are vulva-specific, thereby initiating vulval development (Fig. 6B). In this way, AC signaling limits gene activation to the central VPCs, and lin-39 activity ensures that only vulva-specific genes are activated. This model is supported by the finding that lin-39 is required for its own up-regulation in response to Ras signaling. Specifically, this result identifies lin-39 itself as an example of a vulva-specific gene whose expression requires both Ras activation and lin-39 activity, and thus clearly demonstrates that lin-39 can act to select vulva-specific genes for activation by the Ras pathway. We imagine that when we replaced lin-39 with mab-5, a different set of genes was activated in response to lin-1 inactivation.
In summary, we have found that the Hox gene lin-39 plays two roles in vulval development. First, this Hox gene is absolutely necessary for vulval cell divisions to occur following activation of the Ras pathway. Interestingly, blocking expression of specific Hox genes or Ras can inhibit division of human melanoma cells (Care et al., 1996; Ohta et al., 1996), suggesting that Hox genes may permit Ras-mediated cell division in higher organisms as well. Second, lin-39 function contributes specificity to the Ras signaling pathway, probably by selecting vulva-specific genes for transcriptional activation. Homeotic selector genes, rather than novel factors, may impart specificity to signaling pathways in other cases as well.
We thank Raffi Aroian, Cori Bargmann, Stacey Harmer, Lee Honigberg, Craig Hunter, Gregg Jongeward, Steve Salser, Mary Sym, and Ilan Zipkin, for useful discussion and comments on the manuscript. Helen Chamberlain and Paul Sternberg kindly provided us with sy97, n1046, and n378. Our reviewers provided motivation to do some very interesting experiments. Some strains were obtained from the Caenorhabditis Genetics Center, which is supported by the National Center for Research Resources of the National Institutes of Health. J. N. M. was supported in part by a HHMI pre-doctoral fellowship. This work was supported by NIH grant RO1 GM37053 to C. K.