Vulval development^*

Vulval development is conveniently thought of as occurring in five steps (Figure 1). (1) Generation of VPCs: During the L1 and L2 stages, six vulval precursor cells (VPCs) are specified among the 11 Pn.p cells, which are located in the ventral epidermis. (2) Vulval Precursor Patterning: During the L3 stage, a signal from the gonad and signaling among the VPCs specifies three VPCs to generate vulval cells (Figure 2). The vulval lineages are of two types, 1° and 2°, each of which generate distinct sets of progeny. The uninduced VPCs generate a 3° lineage, which make epidermal cells that fuse with the large syncytial epidermis hyp7. In an intact, wild-type hermaphrodite, the VPCs adopt their fates in a precise spatial pattern: 3°-3°-2°-1°-2°-3° (Figure 3). (3) Generation of the adult cells: The adult vulva comprises 22 nuclei in cells of seven types, vulA, vulB1, vulB2, vulC, vulD, vulE, and vulF, that differ in their patterns of gene expression and in details of their cell biology. (4) Anchor cell invasion: the anchor cell extends process to the center of the vulF cells and forms a hole in the epidermis. (5) Morphogenesis of the vulva: the seven types of vulval cells invaginate, and sequentially form seven distinct toroids, connects to the uterus, and everts as the hermaphrodite molts to adulthood. In addition, two aspects of uterine development are crucial to the development of the vulva. (A) Generation of the anchor cell. The anchor cell is specified from among two somatic gonadal cells during the end of the L2 stage (see AC/VU decision). (B) Uterine Patterning. The anchor cell also patterns the developing uterus, inducing the six pi cells, which generate the utse cell and uv1 cells. The uv1 cells attach to the vulF cells, while the anchor cell ultimately fuses with the utse.

Mutations that affect vulva development often cause defects that are visible in the dissecting microscope. These phenotypes include the inability of hermaphrodites to lay eggs (EGg-Laying-defective; Egl), or morphological defects such as Protruding VuLva (Pvl) or abnormal Eversion of the vuLva (Evl). Specifically, mutants with defective vulval induction, e.g., mutations that decrease activity of the EGF-receptor let-23 (see below) cause a VULvaless (Vul) phenotype. Vulvaless hermaphrodites have fertile eggs that hatch internally and eventually escape from the corpse of their mother. Hermaphrodites with defects in the vulval lineages or in the vulval-uterine connection have protruding vulvae or are egg-laying defective. Mutations that activate LET-23, rendering it constitutive, result in a MUltiVulva (Muv) phenotype. Multivulva hermaphrodites typically have a single functional vulva and additional ventral protrusions, each a pseudovulva formed from vulval tissue. Mutants with reversed polarity of P7.p 2° lineage have a psuedovulva posterior to the normal vulva.

In addition to screens for mutants with vulval defects (Horvitz and Sulston, 1980; Ferguson and Horvitz, 1985; Seydoux and Greenwald, 1993; Eisenmann and Kim, 2000; Palmer et al., 2002), suppressor screens proved useful in identifying new genes and alleles, including members of the LET-23 pathways and genes involved in other steps in vulval development (e.g., Beitel et al., 1990; Han et al., 1990; Aroian et al., 1991; Clark et al., 1992; Clark et al., 1992; 1993; Han et al., 1993; Clark, 1992).

The phenotype of a hermaphrodite observed at the dissecting microscope level can sometimes be misleading, and thus the dissecting microscope phenotypes should always be confirmed by observation of anatomy in a compound microscope. For example, lin-17 mutants often display a vulva (from P5.p and P6.p) and a posterior pseudo-vulva (from P7.p), appearing to be multivulva, but have a normal extent of vulval induction. A number of markers are now available to help distinguish different fates of VPCs (Burdine et al., 1998; Yoo et al., 2004) and of terminal vulval cells (Inoue et al., 2002).

In the intact, wild-type hermaphrodite grown under standard conditions, P5.p, P6.p and P7.p generate vulval cells. Ablation experiments indicated that the six cells P3.p, P4.p, P5.p, P6.p, P7.p and P8.p are competent to respond to intercellular signals and generate vulva. (Sulston and White, 1980; Kimble, 1981; Sternberg and Horvitz, 1986). For example, ablation of P6.p may result in P5.p adopting the 1° fate and P4.p adopting the 2° fate, or, P7.p may adopt the 1° fate and P8.p the 2° fate. Ablation of P5.p may result in replacement by P4.p, whle ablation of P7.p may result in replacement by P8.p. Ablation of all VPCs but P3.p results in P3.p becoming 1°. There is some variability, especially as a function of time at which the ablation occurs.

Ablation of the gonad during the L1 stage (Sulston and White, 1980), or of the anchor cell prior to the L3 stage, results in all six VPCs becoming 3° (Kimble, 1981; Figure 6). These experiments also indicate that the anchor cell induces the vulva.

Ablation of all six VPCs in C. elegans did not result in P2.p or P9.p generating vulval progeny (Sulston and White, 1980; compare to Panagrellus redivivus (Sternberg and Horvitz, 1982; see Evolution of development in nematodes closely related to C. elegans), suggesting that only P3.p-P8.p form the "vulval equivalence group." Additional support for this conclusion comes from the observation that mutations that activate the LET-23 or LIN-12 signaling pathways (and hence allow VPCs to generate vulval cells independent of a signal from the gonad) only cause P3.p-P8.p but P2.p or P9.p to adopt vulval fates (Greenwald et al., 1983; Ferguson et al., 1987; Sternberg, 1988). Moreover, when the gonad position relative to the VPCs is displaced by a dig-1 mutation, the position of the 1° cell can shift, no cells other than P3.p-P8.p generate vulval cells (Thomas et al., 1990).

The hox gene lin-39 is a major determinant of the VPC group (Clark et al., 1993). lin-39 is expressed in P3.p to P8.p (Salser et al., 1993). Lack of lin-39 activity results in lack of VPCs because the presumptive VPCs fuse with the hyp7 epidermis. However, overexpression of lin-39 does not cause additional Pn.p cells such as P2.p or P9.p to become competent. lin-39 has at least two roles: preventing fusion and stimulating cell division (Figure 5). A mutation in the eff-1 results in lack of cell fusion (Mohler et al., 2002). Inactivation of both eff-1 and lin-39 results in VPCs that fail to fuse and fail to divide (Shemer and Podbilewicz, 2002). WNT signaling via BAR-1 beta-catenin is necessary to maintain expression of lin-39 in the P(3-8).p (Eisenmann et al., 1998). Expression of lin-39 is sufficient to bypass the requirement for BAR-1 function in generating competent VPCs (Eisenmann et al., 1998). Using a bar-1 mutant to sensitize the VPCs for loss of competence, Eisenmann et al. (1998) showed that let-23 and the somatic gonad were involved in establishing competence of VPCs.

sem-4, which encodes a zinc finger protein, is necessary for full lin-39 expression, and overexpression of lin-39 can rescue the defect of a sem-4 mutant (Grant et al., 2000). CEH-20 and UNC-62 are co-factors of LIN-39 and as such affect vulval development (Yang et al., 2005). Two redundant GATA factors, encoded by the adjacent egl-18 and elt-6 genes, are necessary for vulval competence, and are regulated by LIN-39 and CEH-20 (Koh et al., 2002). The translocation nT1 has a highly penetrant VPC competence defect (Ferguson and Horvitz, 1985; Ferguson et al., 1987). nT1 disrupts the egl-18/elt-6 operon, preventing its expression in the vulva (Koh et al., 2004).

More detailed analysis indicates that the VPCs, while multipotential, are not equivalent in their ability to respond to the major inductive signal (Clandinin et al., 1997). In particular P7.p and P8.p are less sensitive to EGF-R signaling than is P6.p (Figure 4). This difference in sensitivity is dependent on the hox gene mab-5, which is expressed in P7.p and P8.p but not P(3-6).p (Salser et al., 1993). Also, sem-4 is preferentially required in P7.p, and this effect is in part due to mab-5 (Grant et al., 2000). The details of fate specification may differ between P5.p and P7.p. These differences are relatively subtle, but very intriguing in light of vulval development in other nematode species (see Evolution of development in nematodes closely related to C. elegans).

VPC patterning is conveniently and appropriately thought of in terms of VPC level specification (Sternberg and Horvitz, 1986). A specified VPC generates a characteristic 1°, 2°or 3° lineage, and expresses markers that are characteristic of each fate.

An open question in C. elegans development concerns fate commitment. I use specification to mean a cell knows what to do but is not irreversibly committed to that fate; commitment is irreversible (within a reasonable set of perturbations). The rapidity of C. elegans development, and the lack of long-term culture have precluded extensive tests of commitment. One can argue that cells are specified and then development proceeds without any need for long-term stability. One experiment with VPCs revealed that the 2° fate VPCs are not committed (Wang and Sternberg, 1999). Expression of high levels of LIN-3 after the VPCs have divided in a LIN-12 activated mutant alters their fate such that they express 1° lineage markers and display the morphological hallmarks of 1° lineages, such invagination; thus cells specified to become 2° are not committed even after they divide. The 3° fate is irreversible after fusion with hyp7, about an hour after their cell division. The 1° fates so far have not been altered after their division (Ambros, 1999; Wang and Sternberg, 1999). Ambros (1999) found that VPCs could become specified to become 1° prior to the end of their S phase but specification of 2° fates required the end of S phase. In this view, the cell cycle helps prioritize the specification process.

Hybrid lineages arise by too little induction (for example ablation of the anchor cell before the VPCs are fully induced) or by late induction acting on the daughter cells (for example if LIN-3 expression is induced after the VPCs divide; Wang and Sternberg, 1999). Fusion of VPC daughters (P3.pa, P3.pp, P4.pa, P4.pp, P8.pa, and P8.pp) occurs about an hour after they are formed, which eliminates their competence to respond to LIN-3.

Vulva development involves cell cycle in two distinct ways. Cell cycle regulation is one important consequence of vulval cell fate specification, and conversely, the cell cycle influences fate specification. One of the major differences between the uninduced 3° and the induced vulval (1° and 2°) VPC fates is in the extent of cell division. Cyclin E, encoded by cye-1 is necessary for proliferation of the VPCs (Fay and Han, 2000). In cye-1 mutants, there are fewer VPC divisions, but differentiation, assayed by expression of egl-17::gfp occurs at the normal time. Moreover, Boxem and van den Heuvel (2002) found that cell division could be uncoupled to some extent from differentiation.

The heterochronic genes lin-4, lin-14, and lin-28 control the G1 to S transition of VPCs (Euling and Ambros, 1996). The cell cycle inhibitor cki-1 is one target of this heterochronic pathway (Hong et al., 1998). cki-1::GFP is expressed in the VPCs during their G1 phase. Inactivation of cki-1 results in VPC division during the L2 stage, and expression of cki-1 in P6.p under control of egl-17 transcriptional regulatory sequences blocks its mitosis (Hong et al., 1998). cki-1 expression in VPCs is reduced in the absence of lin-14, which results in precocious development of the VPCs.

Ablation of the anchor cell or its ancestors prior to the L3 stage completely blocks vulval development, and the VPCs have 3° fates (3°-3°-3°-3°-3°-3°). Ablation of all gonadal cells but the anchor cell results in full induction of the vulva (Kimble, 1981). Males will copulate with these vulvae but not transfer sperm (M. Barr, K. Liu and P. Sternberg, unpublished observations).

lin-3 encodes the inductive signal from the anchor cell that induces the VPCs (Hill and Sternberg, 1992; Figure 9). lin-3 is necessary for vulval induction (Horvitz and Sulston, 1980; Sulston and Horvitz, 1981; Ferguson and Horvitz, 1985; Ferguson et al., 1987; Hill and Sternberg, 1992). lin-3 is expressed in the anchor cell during the time induction takes place. Expression of the EGF domain of LIN-3 directed by a heat shock (hsp-16) promoter/enhancer is sufficient to induce the vulva in the absence of the gonad and hence the anchor cell (Hill and Sternberg, 1992).

The anchor cell signals at a distance from the VPCs. Cell ablation studies reveal that a distal VPC can be induced to a 1° or 2° fate (or some intermediate; Sternberg and Horvitz, 1986; Katz et al., 1995; Sherwood and Sternberg, 2003). In addition, in dig-1 mutants, a dorsally located anchor cell can induce patterns such as 1°-2°-1°, 1°-2°-1°-2°or 2°-2°-2°-2° (Thomas et al., 1990). The dorsally located anchor cell can extend a process ventrally, but this occurs after induction, and is fully consistent with the anchor cell invasion of the vulval epidermis (Sherwood and Sternberg, 2003).

The anchor cell signal is able to act at a distance and appears graded based on a correlation of distance from the anchor cell and the fate of individual VPCs (Sternberg and Horvitz, 1986). An individual VPC, after ablation of other VPCs, can have a 2° fate. Also, in a dig-1 mutant there can be 2° VPCs without 1° VPCs (Thomas et al., 1990). In addition, molecular genetic experiments (Katz et al., 1995; Katz et al., 1996) strongly support the possibility of a graded signal. The dose of LIN-3 was controlled in two ways. The EGF domain of LIN-3 along with a synthetic signal sequence was expressed under the transcriptional control of a heat-shock enhancer/promoter. By varying the duration and temperature of heat shock, different apparent doses of LIN-3 could be produced. A high dose results in 1° fates; an intermediate dose in 2° fates; and a low dose in 3° fates. In most cases there is a distribution of fates as well as intermediate fates. A distinct set of experiments with partially activated LET-23 gave similar results (Katz et al., 1996).

The graded expression of egl-17, the best-known target of LET-23 signaling in the 1° VPC (Burdine et al., 1998), provides a molecular correlate for a graded inductive signal. egl-17-CFP-LacZ, is initially expressed in P5.p, P6.p and P7.p with higher expression in P6.p, thereby indicating a graded signal (Yoo et al., 2004). Its expression is increased in P6.p and decreased in P5.p and P7.p due to LIN-12 signaling as described below.

In addition to the strong support for a graded signal, there is also strong support for a sequential signaling mechanism in which activation of LET-23 in P6.p leads to activation of LIN-12 in P5.p and P7.p. Certain genetic mosaics lacking let-23 activity suggest that direct induction by the AC is not necessary for the 2° fate: if let-23(+) is present in all cells, there is a wild-type vulva; if all cells are let-23(-) there is no vulva (Figure 10; Simske and Kim, 1995; Koga et al.,1995). If P6.p is let-23(+) but P5.p and P7.p are let-23(-), then in almost all cases there was a normal vulva. Thus, a cell can become 2° without receiving LIN-3. This key result is consistent with the observations that activation of LIN-12 can specify a 2 ° fate in the absence of an anchor cell (Greenwald et al., 1983) or of LET-23 signaling (Sternberg and Horvitz, 1989; Han et al., 1990), and strongly makes the point that the direct induction of 2° by the anchor cell is not necessary. LIN-12 is expected to be the receptor for this 2°-inducing signal: lin-12 activity is necessary and sufficient for the 2° fate (Greenwald et al., 1983), and ligands for LIN-12 are up-regulated in P6.p in response to the inductive signal (Chen and Greenwald, 2004).

The converse mosaic animals had a striking phenotype: if P6.p is let-23(-) but P5.p and P7.p are let-23(+), then additional cells are induced (Simske and Kim, 1995; also see Hajnal et al., 1997). Hajnal et al. argue that LET-23 might be sequestering LIN-3 and preventing its spread to more distal cells.

Dutt et al. (2004) have suggested that lin-3 expression in induced VPCs extends the range of induction, and that this role of lin-3 depends on ROM-1. rom-1 encodes a protease related to Drosophila Rhomboid, which cleaves Spitz, a Drosophila protein related to LIN-3. lin-3 encodes multiple isoforms that differ in the region between the EGF domain and the transmembrane domain (Hill and Sternberg, 1992; Dutt et al., 2004), a part of the protein that would likely be cleaved for LIN-3 in a relay mechanism. The B but not the A isoform of LIN-3 is ROM-1 dependent (Dutt et al., 2004).

Two transcription factors, LIN-31 and LIN-1 are likely targets of the MAPK pathway (Tan et al., 1998). LIN-31 and LIN-1 can both be phosphorylated by MAP kinase MPK-1. Inactivation of either lin-1 or lin-31 leads to a Multivulva phenotype. LIN-1 and LIN-31 form a complex that is disrupted by LIN-31 phosphorylation (Tan et al., 1998). Increased expression of LIN-31 or a non-phosphorylatable LIN-31 protein causes a vulvaless phenotype. During the L3 stage, lin-31 mutants display both vulvaless and multivulva phenotypes (Ferguson and Horvitz, 1985). lin-31 mutants display inappropriate divisions of Pn.p cells during the L2 stage (Ferguson et al., 1987; Miller et al., 2000), suggesting also a role in the proper specification of VPCs.

lin-39 is upregulated by LET-23 signaling (Maloof and Kenyon, 1998) and is an excellent candidate for a direct target. Since overexpression of lin-39 does not result in gonad-independent vulval differentiation, it is unlikely to be the only target (Figure 5). egl-17 is induced early in the induction process (Burdine et al., 1998; Ambros, 1999; Wang and Sternberg, 1999; Cui and Han, 2003; Yoo et al., 2004) and could be an early target of RAS-stimulated transcription. Romagnolo et al. (2002) used RNA from whole animals hybridized to microarrays to search for RAS-regulated genes and found 718; among these might be some that are physiologically relevant during vulval development.

Mediator complex proteins, which link site-specific transcription factors to RNA polymerase, also are necessary for vulval development. lin-25 and sur-2 (mdt-23) are defective in vulval induction and also have defects earlier and later during vulval development (Ferguson et al., 1987; Tuck and Greenwald, 1995; Singh and Han,1995). mdt-6 is positive-acting (Kwon and Lee, 2001). It is striking that other Mediator components have negative effects, e.g., dpy-22 (mdt-12; Moghal et al., 2003; H. Sawa, unpublished observations). There are two extreme interpretations of this observation. First, there might be differential requirements for mediator components on distinct transcriptional targets. Second, there might be one main target for Mediator and there is a regulatory pathway acting among the Mediator components with some components acting in a positive and others in negative manner.

In addition to its role in VPC competence, activation of WNT signaling can increase the extent of vulval induction if the LET-23 pathway is compromised (Gleason et al., 2002). It is not known whether WNT signaling can compensate for the complete absence of LIN-3 signaling.

The heterotrimeric G protein Gq (EGL-30; see Heterotrimeric G-proteins in C. elegans) stimulates vulval induction under certain growth conditions, in particular growth in liquid (Moghal et al., 2003). Expression of activated EGL-30 in neurons stimulates vulval induction. The EGL-30 effect requires the voltage-gated calcium channel encoded by egl-19. The site of action of egl-19 for this effect is in muscle cells. These observations lead to the hypothesis that neuronal and muscle excitation can somehow affect vulval development. The positive signal from Gq acts in parallel to or downstream of RAS. Because the effect is of the magnitude of WNT signaling via BAR-1 β-catenin, it is possible that this is effect is via the WNT pathway (Moghal et al., 2003).

A G protein coupled receptor, SRA-13, acts negatively on vulval development via the G protein GPA-5 (Battu et al., 2003). Starvation decreases vulval induction, and this effect depends upon SRA-13. sra-13 is expressed in chemosensory neurons as well as body-wall muscle (Battu et al., 2003) and gpa-5 is also expressed in chemosensory neurons. The sites of action of SRA-13 and GPA-5 for vulval induction effects are not known. The relationship of the positive EGL-30 (Gq) effect and negative GPA-5 effects is intriguing but not yet known. These findings highlight the importance of nematode physiology, revealed by growth conditions, on vulval development. The effect on vulval induction of growth in liquid (Moghal et al., 2003) and of starvation (Battu et al., 2003) were only revealed by using sensitized genetic backgrounds. I believe that there could be much plasticity lurking beneath the surface, a surface that can be scratched away by careful genetic analysis.

Two genes identified by loss-of-function mutations that suppress activated LET-60 Ras encode cation diffusion facilitators that appear to regulate zinc. CDF-1 is a cation diffusion regulator, and has sites of action in the intestine as well as in the VPCs, consistent with it expression in both tissues (Bruinsma et al., 2002). Exogenous zinc negatively regulates LET-23 signaling (Bruinsma et al., 2002). SUR-7 encodes a distinct cation diffusion facilitator (Yoder et al., 2004). Yoder et al. (Yoder et al., 2004) argue that zinc regulation is at the level of KSR-1 and PAR-1, which somehow regulate LIN-45 Raf signaling (see RTKRas/MAP kinase signaling).

The synMuv genes are three classes (A, B and C) of broadly expressed nuclear proteins that in various combinations inhibit VPCs from adopting vulval fates. Ferguson and Horvitz (1989) described two redundant sets of genes, class A and B, where hermaphrodites defective in both A and B function have a Multivulva phenotype. Mutants defective in only A, or in only B, have normal vulval development. Molecular cloning revealed that many Class B synMuv genes encode proteins with mammalian orthologs that have been well-characterized as transcriptional regulators, including the synMuv genes lin-35 Rb , efl-1 E2F, and hda-1 histone deacytelase (Lu and Horvitz, 1998), as well as a zinc finger protein (Melendez and Greenwald, 2000) and components of the NuRD and SWI/SNF complexes (Lu and Horvitz, 1998; Solari and Ahringer, 2000; Ceol and Horvitz, 2001; Cui et al., 2004). Other synMuv genes encode novel nuclear proteins.

Epistasis experiments suggest that synMuv genes inhibit LET-23 pathway output (Ferguson et al., 1987; Huang et al., 1994). Site of action studies with synMuv genes suggest that they control a pathway of intercellular signaling from hyp7 to the VPCs, although no signal has been identified (Figure 11; Figure 12). lin-15AB and lin-37B mosaics suggested action not in the VPCs, and possibly in hyp7 (Herman and Hedgecock, 1990; Herman and Hedgecock, 1995); Mosaic analysis and tissue-specific rescue experiments indicate that the focus of lin-35 is hyp7 (Myers and Greenwald, 2005). However, lin-36B mosaics suggest a site of action in the VPCs (Thomas and Horvitz, 1999), and lin-15A mosaics and tissue-specific rescue experiments also suggest a VPC focus (A. Gonzalez-Serricchio and P. W. Sternberg, unpublished observations).

Other negative regulators were identified as suppressors of hypomorphic mutations in LET-23-RAS pathway components. These genes do not interact strongly with the synMuv genes, but do interact with each other. sli-1, unc-101, ark-1, gap-1 and lip-1 (Jongeward et al., 1995; Lee et al., 1994; Yoon et al., 1995; Hajnal et al., 1997; Berset et al., 2001; Hopper et al., 2000) are of this class. In addition, Yoo et al. (2004) found four additional negative regulators, dpy-23, lst-1, lst-2, lst-3, and lst-4. These four regulators, and lip-1 (Berset et al., 2001) are targets of LIN-12 lateral signaling.

unc-101 and apm-1 each encode medium chains of the AP-1 adaptin, and these are inhibitory on vulval induction (Lee et al.,1994; Shim, 2000); while LET-23 is a plausible target for these membrane trafficking proteins, the targets are not known. dpy-23 encodes a medium chain of the AP-2 adaptin and is induced by LIN-12 (Yoo et al., 2004; see below).

lin-12 is necessary and sufficient for specification of the 2° VPC fate (Greenwald et al., 1983). Null alleles of lin-12 lack 2° VPCs. Gain of function lin-12 mutants cause all six VPCs to generate 2° lineages, even in the absence of inductive signaling. As described in LIN-12/Notch signaling in C. elegans, lin-12 encodes a Notch-type transmembrane receptor (Yochem et al., 1988).

lin-12(lf) hermaphrodites have additional anchor cells, and this complicated somewhat the analysis of vulval induction (Greenwald et al., 1983). With two (or more) anchor cells, P5.p, P6.p and P7.p are 1° and P4.p and P8.p occasionally are 1° or one daughter is 1° like (Sternberg and Horvitz, 1989).

The lateral signal involves at least three redundant genes, apx-1, lag-2 and dsl-1 (Chen and Greenwald, 2004; see LIN-12/Notch signaling in C. elegans). APX-1 and LAG-2 are transmembrane ligands for LIN-12. DSL-1 is a secreted ligand for LIN-12. Expression of lag-2 occurs in all VPCs prior to induction and is upregulated in P6.p. Expression of apx-1 and dsl-1 appears to be initiated by induction of P6.p. The main effect is to induce neighbors to become 2°, thereby inhibiting them from becoming 1°. The redundancy of lag-2, dsl-1 and apx-1 for VPC patterning appears to have prevented the discovery of their role by genetic screens. The targets of LIN-12 include a number of redundant negative regulators of LET-23 signaling (Yoo et al., 2004; Figure 14).

The pattern of VPC fates in a lin-15 multitvulva mutant (1°-2°-2°-1°-2°-1°; Figure 15) is reminiscent of spacing patterns such as that of bristles in the insect abdomen and thus suggests a lateral inhibition mechanism whereby a specialized cell type (1° VPC) inhibits adjacent cells from also becoming of that type and instead differentiating as some alternative (2° VPCs). While adjacent 2° VPCs are common, adjacent 1° VPCs are rare. In a lin-15 strain, which appears to have activat ed LET-23 signaling, a single VPC (e.g., P7.p) will become 1° in the absence of the gonad and of its neighbors (Sternberg, 1988). If there are two adjacent VPCs (P7.p and P8.p), they will become 1°-2°or 2°-1°, with approximately equal probability. Thus, these cells signal each other.

In the presence of the gonad, P7.p becomes 1° and P8.p becomes 2°, indicating that the inductive signal can bias the stochastic lateral signaling among induced VPCs (Sternberg, 1988). Double mutant combinations among inductive signaling pathway genes and lin-12 mutations support this view: a lin-12 loss-of-function mutation eliminates the lateral inhibition in a lin-15 mutant resulting in all 1° VPCs (Sternberg and Horvitz, 1989).