Functional Roles of Gene Expression During Vulval Differentiation.

Much is known about the signaling network that establishes the pattern of vulval cell differentiation, but our understanding of the GRN that specifies the terminal seven vulval cell types is limited (2). Five transcription factors (lin-11, lin-29, cog-1, egl-38, and nhr-67) are major regulators of cell-fate determination and morphogenesis in the vulva. lin-11 encodes a LIM homeodomain protein, consisting of a homeodomain and two specialized LIM-type zinc-fingers (5). LIM homeodomain family members play roles in differentiation and pattern formation in arthropods and vertebrates (6, 7). lin-29, a C2H2-type zinc finger, plays a role in many events occurring at the larva to adult transition, including terminal differentiation of the seam cells (8), morphogenesis (9), and formation of the vulval-uterine-seam cell connection (10). cog-1 encodes a Nkx6.1/6.2 homeoprotein transcription factor (11); vertebrate Nkx6.1 proteins are involved in neuronal and pancreatic endocrine cell formation (12, 13). egl-38 encodes a Pax2/5/8 protein, which are known to be involved in organogenesis; e.g., mouse Pax2 mediates nephrogenesis (14, 15). nhr-67 is an ortholog of Drosophila melanogaster tailless (tll) (16), a conserved nuclear hormone receptor necessary for Drosophila embryogenesis and neuronal development (17). The pattern of vulval cell types is specified by the differential interactions of the transcription factors that operate within each cell (2).

We have identified 30 genes dynamically expressed in specific subsets of the cells of the mature C. elegans vulva (Table S1 has a complete list). These genes encode transcription factors, guidance cues, proteases, structural proteins, signaling molecules, and novel proteins with unknown function. The physiological relevance is known for several genes. For example, egl-17, which encodes a fibroblast growth factor (FGF), is necessary for migration of the sex myoblasts to the vulva (18). Induction of the uterine uv1 cells depends on the epidermal growth factor (EGF) family member lin-3 (19). bam-2 (neurexin-related transmembrane protein) and syg-2 (transmembrane Ig superfamily protein) are required for vulval innervation (20, 21). sqv-4, which encodes an UDP-glucose dehydrogenase-related protein, is involved in the structural integrity and morphology of the vulva (22). The significance of some of the genes expressed in the vulva is not yet known, including the Drosophila empty spiracles (ems) homolog ceh-2 (23), a cadherin-related protein encoded by cdh-3 (24), and zmp-1, which encodes a MT4-MMP-related zinc metalloproteinase (23, 25). Genetic perturbations that result in altered expression patterns of these effector genes are helpful in elucidating the regulatory network.

Several additional genes with detectable expression in the mature vulva, including two putative transcription factors, have been identified since we last described the vulval GRN (3). The Pax2/5/8 gene pax-2 is expressed exclusively in the vulD cells (16). pax-2 is the result of a recent duplication of egl-38 (26). egl-38 has been identified as a regulator of cell-fate specification in the C. elegans vulva (3, 15, 19). lin-39 encodes a Hox protein that is an ortholog of Drosophila Sex combs reduced (Scr) (27, 28). During the late L3 stage, lin-39 expression increases in the vulA precursor cells; this expression persists in vulA until late L4 (29). The Patched-related protein DAF-6 is expressed in vulE and vulF (30). Hao et al. (31) reported the expression of seven hedgehog-related genes in the vulva: grd-5 in vulB and vulD, grd-12 in vulC, grl-4 in vulA and vulB, grl-10 in vulA and vulB, grl-15 in vulB, vulC, vulD and vulE, grl-25 in vulA, and grl-31 in vulF. Last, nas-37 (32), which encodes a metalloprotease, is expressed in vulB (data not shown).

Conserved Regulatory Strategies in the Vulval GRN.

In ref. 3, we described how the expression patterns of a subset of reporter constructs are affected in transcription factor mutant backgrounds. We have increased the number of known interactions more than twofold, from 15 to 36, and identified many of the regulatory relationships among the transcription factors (Table 1, Fig. S2, and Table S2). This network includes strategies that are shared by other GRNs. For example, the cell-type specific expression of cog-1 appears to be restricted to the vulC and vulD cells by a variety of mutual and autoregulatory controls (16). In a second example, lin-11 is necessary for vulA-specific expression of nhr-67 (16). In turn, nhr-67 is necessary for the expression of the vulA effector gene zmp-1. The differentiated state of vulA may be further stabilized, because nhr-67 is positively autoregulated in vulA. This is an instance where multiple positive inputs, including feedback loops, ensure the maintenance of a terminal cell fate.

Other network strategies that are present in the vulva include combinatorial control circuits. egl-17 expression in vulF is perturbed by neither cog-1 nor egl-38 mutations (16). However, egl-17 expression is derepressed in vulF in cog-1; egl-38 double mutants. This redundancy could ensure proper execution of cell fate. Finally, negative autoregulation, as in cog-1 in vulA, vulB, vulE, and vulF and nhr-67 in vulC, vulD, vulE, and vulF, appears to be a fundamental strategy used in the vulval GRN (16). Negative autoregulation has been found to accelerate gene circuit response time and assists in making quick cell fate decisions (33).

LIN-29 and LIN-11 Interact to Determine Vulval Cell Fate.

By examining the effects of specific transcription factor mutations on a wider array of genes expressed in the mature vulva, we have found increased complexity in some of their roles. We had hypothesized that lin-29 is a temporal regulator of gene expression, because it is required for vulval gene expression at the mid to late L4 stage (egl-17 in vulC and vulD, ceh-2 in vulC, and zmp-1 in vulD and vulE) (3, 23), and functions in several developmental processes at the L4 stage to adult transition (8–10, 34). However, in a lin-29 mutant background the mid-L4 expression of lin-3 in vulF cells is not abolished, whereas expression of dhs-31, which is initiated in gravid adults, is abolished (Table 1 and Fig. S2).

LIN-29 is necessary for wild-type levels of lin-11 transgene expression (Table 1) and is thus a key regulator of lin-11. In addition to genes described in ref. 3, lin-11 is required for dhs-31, egl-26, lin-3 and pepm-1 expression (Table 1). Thus, of the known vulval transcriptional regulators, lin-11 exhibits the broadest effect (3). egl-17 during L3 and cdh-3 in vulF are the only vulva expression patterns not abolished in a lin-11 background. Reporter constructs are powerful tools for identifying genes' spatial and temporal expression patterns; however, reporter gene constructs might well lack relevant regulatory motifs. Consequently, we might be missing some regulatory connections.

The interplay between lin-11, lin-29 and their downstream targets is complex. lin-11::GFP expression is not abolished in a lin-29 mutant background, and several of the gene expression patterns perturbed by loss of lin-11 are not affected in a lin-29 deficient background (Fig. S2). LIN-29 might act in concert with another, as yet unidentified, factor to ensure the proper temporal and spatial expression of the general cell-fate regulator LIN-11. Considering LIN-29's role as a regulator of developmental timing, it could serve as a temporal input, whereas another factor serves as the spatial input.

Identification of Additional Components of the Vulval GRN.

Transcription factors identified by forward genetic screens are biased by ascertainment toward those with strong, nonredundant effects, and those that affect multiple aspects of vulval development. Two categories of transcription factors are relatively more difficult to identify in genetic screens. Mutations that result in lethality or other severe defects will not be identified for roles that a partial loss of function mutation might reveal in postembryonic development. Conversely, genes with more subtle developmental phenotypes are also more likely to be overlooked. We identified genes in both these categories by RNA interference (RNAi) screens of 508 transcription factors (Table S3). One of these screens was conducted in a ceh-2::YFP background, a readout for vulB fate during the L4 stage, and identified nhr-67 (tailless) (16). Because nhr-67 deletion mutants die as young larvae, its role in vulval development was not identified. RNAi often causes a partial-loss-of-function phenotype instead of a null phenotype. It can also serve as a temporal or conditional downregulator of gene function. In the case of nhr-67, the larval lethality phenotype was bypassed, because RNAi was administered to L1 larvae, thus making its vulval phenotypes visible.

The second screen was conducted in a zmp-1::GFP background, focusing on perturbations of vulA expression. This screen also identified nhr-67, and the orphan nuclear hormone receptor nhr-113 as positive regulators of zmp-1 expression in vulA cells. nhr-113 might have a narrow role in vulval organogenesis, because nhr-113 RNAi has no effect on the regulation of several other genes: cdh-3, ceh-2, dhs-31, lin-3, or pepm-1 (data not shown). These results, however, show that RNAi screens can identify new components of GRNs.

Differentiation of Discrete Fates in the 1° Vulval Lineage.

The 1° lineage of the vulva generates four vulE and four vulF cells. Signals from the AC and Wnt are required for proper specification of these cell fates (25). The GRN, however, which acts downstream of these intercellular signals to guide differentiation of vulE and vulF fates, is not known. egl-38::GFP is detectable solely in the vulF cells (35); however, analysis of the mutant egl-38 with nhr-67 RNAi and mutant cog-1 shows that egl-38 functions in vulF cells and inhibits both ceh-2 and egl-17 expression in vulE (16) and vulF (3). Consequently, egl-38, cog-1, and nhr-67 enforce spatial boundaries by preventing the 2° cell-fate associated genes ceh-2 and egl-17 from being expressed in 1° cells (Fig. S2, vulE and vulF).

It was speculated that the vulF cells, which are nearer the AC, have higher levels of nhr-67 activity versus cog-1, whereas vulE cells have higher levels of cog-1 than nhr-67 (16). The recent availability of an egl-38::GFP reporter has allowed us to dissect further the GRN controlling the vulE versus vulF fates. egl-38 expression in the vulF cells is positively regulated by nhr-67 (Fig. 1). This egl-38 expression is necessary for specification of the uv1 fate via regulation of the vulF specific gene lin-3, thus allowing for the proper development of a vulval-uterine connection (19). Conversely, zmp-1 is expressed in vulE but not vulF, and egl-38 is required for inhibiting zmp-1 expression in the vulF cells (3). In vulE cells, where egl-38 expression is absent, lin-3 is not expressed and cog-1 activated expression of zmp-1 is observed (19). egl-38 expression is unaffected in a cog-1(sy275) background, but in an egl-38(n578) background egl-38::GFP expression is decreased, suggesting that egl-38 positively autoregulates in vulF (Fig. 1). Therefore, egl-38 appears to stabilize vulF fate by repressing vulE characteristics and by reinforcing its own expression (Fig. 2).

In the vulE and vulF cells, cog-1 and nhr-67 negatively regulate both each other and themselves (16). We speculated that this might allow 1° cells to rapidly switch their fates upon altered intra- and intercellular inputs. Presumptive vulE cells can induce uterine uv1 fate specification in the absence of vulF cells, thus ensuring the establishment of a proper vulval-uterine connection (19). These observations fit a model in which presumptive vulE and vulF cells are bipotential and positional cues help specify their fates. In this model, vulE, would then be biased for increased nhr-67 activity in the absence of vulF. egl-38 levels would thus increase, and vulF characteristics would be activated, whereas vulE characteristics would be inhibited.

Dissection of zmp-1 Regulatory Elements.

The vulA cells occupy a unique position as they form the outermost ring of cells and fuse to the surrounding hypodermal syncytium. zmp-1 (zinc metalloproteinase) is first expressed in vulD and vulE cells beginning at the late L4 stage and in the vulA cells at the L4 to adult transition (23). vulA-specific expression of zmp-1 is initiated in a different temporal window than its expression in vulD and vulE, which is particularly interesting because, in a lin-29 background, vulD and vulE expression is abolished, but vulA expression is unaffected. Because lin-29 affects early zmp-1 expression (vulD and vulE) but not late expression (vulA), and because lin-29 temporally regulates gene expression, it seems likely that modular cis-regulatory elements drive zmp-1 expression.

A 386-bp fragment of upstream sequence (mk50–51) is sufficient to drive vulA- and vulE- but not vulD-specific expression of zmp-1 (36). We analyzed sequences upstream of zmp-1 and its orthologs in C. briggsae (CBG09053) and C. remanei (CRE04503), using Cistematic (37), which carries out motif-finding and phylogenetic footprinting (Table 2 and Fig. S3) and identified three motifs within the mk50–51 zmp-1 enhancer element and a fourth motif five basepairs upstream of the 5′ end. We deleted the instances of the motifs and analyzed their effects on zmp-1 reporter expression. Deletion of element 103/4 decreased vulA expression and abolished vulE expression. Previous cis-regulatory studies did not identify an element competent to drive vulA expression in the absence of vulE. Deletion of element 107/8 decreased vulA and vulE expression. Deletion of element 105/6, however, resulted in ectopic expression of mk50–51 zmp-1::GFP in vulC, vulD, and vulF. Thus, these elements likely act as positive (103/4 and 107/8) and negative (105/6) regulatory sites for controlling of zmp-1 expression.

Genetics and RNAi.

C. elegans strains were grown and constructed using standard protocols. The nhr-67 RNAi feeding protocol was described in ref. 16.

Generation of Reporter Transgenes.

syEx1009, syEx1091 and syEx1018 were generated by deleting one motif each, 103/4 (5′-CGAGTACGTTTACAC-3′), 105/6 (5′-GTACGTATTGCTT-3′), or 107/8 (5′-AGAAAAAGTAGAAGG-3′), respectively, from the mk50–51 (36) construct and replacing the motif with a SacII restriction site. These constructs were then individually injected into the gonads of unc-119(ed4); him-5(e1490) animals (41), using unc-119(+) (42) and pBSK+ (Stratagene). syEx756 and syEx1012 were generated by injecting the constructs pNP10 (43) and mk50–51 (36), respectively, into a unc-119(ed4); him-5(e1490) background.

Microscopy.

3 mM levamisole was used to anesthetize transgenic animals for observation using Nomarski DIC optics.