The C. elegans LIM homeobox gene lin-11 specifies multiple cell fates during vulval development

Specification of different cell types during development involves multiple cell-cell interactions mediated by many genes. Studies of the C. elegans hermaphrodite vulva have proven successful in dissecting regulatory networks and understanding the function of some crucial genes. The adult vulva is formed by the progeny of three out of six vulval precursor cells (VPCs) that acquire 1° and 2° fates in a 2°-1°-2°pattern, and undergo three rounds of cell divisions(Sulston and Horvitz, 1977)(Fig. 1). During L4 stage,vulval cells differentiate into seven different cell types and form an invaginated structure (Sharma-Kishore et al., 1999) (Fig. 1). Vulval development thus provides opportunities to study cell fate specification and pattern formation. Mutations that perturb vulval morphology have identified some of the genes that regulate cell differentiation, including lin-17 (frizzled family), lin-18 and lin-11 (LIM homeobox family)(Ferguson and Horvitz, 1985; Ferguson et al., 1987; Freyd et al., 1990; Sawa et al., 1996; Gupta and Sternberg, 2002). Mutations in these genes alter the axes of terminal cell division, suggesting a role in differentiation of a subset of the vulval cells. lin-11 is known to be necessary for the NT portion of the 2° lineage vulval cells because there is a LLLL cell lineage pattern in lin-11 mutant animals compared with the wild-type NTLL pattern (see Fig. 1)(Ferguson et al., 1987). Being a LIM homeodomain family member, LIN-11 is likely to function as a transcriptional regulator of vulval cell-type specific genes.

The LIM homeodomain proteins have two LIM domains and a homeodomain(reviewed by Hobert and Westphal,2000; Bach, 2000). The LIM domains are thought to mediate protein-protein interactions and regulate tissue-specific function. Drosophila Apterous is required for specifying the dorsal region of the wing imaginal disc(Cohen et al., 1992; Diaz-Benjumea and Cohen,1993). Isl1 was identified for its role in the development of pancreatic endocrine cells (Karlsson et al., 1990) and subsequently found to be involved in the formation of motor- and interneurons and pituitary cells(Pfaff et al., 1996; Takuma et al., 1998). Lhx2 is involved in the development of eye, forebrain, erythrocytes and limbs(Porter et al., 1997; Rodriguez-Esteban et al.,1998). Mutations in human LMX1B cause `nail patella syndrome' owing to defects in specification of the patellar mesenchyme of the dorsal limbs (Dreyer et al.,1998; Vollrath et al.,1998).

The C. elegans genome encodes seven LIM homeodomain proteins including LIN-11 (Ruvkun and Hobert,1998). LIN-11 has been shown to be necessary for the development of a subset of vulval cells, uterine π lineage cells and some neurons(Ferguson et al., 1987; Hobert et al., 1998; Newman et al., 1999; Sarafi-Reinach et al., 2001; Gupta and Sternberg, 2002). In this study, we show that the spatiotemporal expression of lin-11confers distinct cell fates. Our experiments reveal at least two distinct functions of lin-11 in vulval cells. lin-11 is first required for setting up the correct pattern of vulval invagination. During this phase, the precursors of vulC and vulD express high levels of lin-11. Later on, lin-11 is expressed in all vulval progeny. Using a conditional RNAi approach, we have examined lin-11 function during vulval morphogenesis and demonstrate that lin-11 is required in vulval progeny for wild-type patterning. Finally, we show that the LIM-binding protein LDB-1 (Cassata et al.,2000) plays a role in vulval differentiation by directly interacting with LIN-11.

C. elegans were grown and mutagenized according to published methods (Brenner, 1974; Wood, 1988). Except for the heat shock strains, for which a 15°C incubator was used, all other strains were maintained at 20°C. Mutations and transgenes used are as follows.

LGI: lin-11(n389) and lin-11(n566)(Ferguson and Horvitz, 1985),and ayIs4[egl-17::GFP + dpy-20(+)](Burdine et al., 1998)

LGII: syIs54[ceh-2::GFP + unc-119(+)]

LGIII: syIs80[lin-11::GFP + unc-119(+)](Gupta and Sternberg,2002)

LGIV: dpy-20(e1282) (Brenner,1974), syIs49[zmp-1::GFP + dpy-20(+)](Wang and Sternberg, 2000)

LGV: nIs96[lin-11::GFP + lin-15(+)](Reddien et al., 2001), syIs53[lin-11::GFP + unc-119(+)](Gupta and Sternberg,2002)

LGX: syIs50[cdh-3::GFP + dpy-20(+)]

Other transgenic strains include syIs78[ajm-1::GFP +unc-119(+)], syEx500[pPHS11.82(hs::lin-11) +pTG96(sur-5::GFP)], syEx530- [pPHS11.82(hs::lin-11)+ myo-2::GFP], syEx552[pPHS11.12(hs::lin-11) +pPHS11.12i(hs::lin-11i) + myo-2::GFP + unc-119(+)], syEx551[pLBP13.3c(ldb-1::GFP)] and syEx565[unc-119(+) + pLBP13.3c(ldb-1::GFP)]. Stably transmitting extrachromosomal array lines and integrants were generated by standard techniques (Mello et al.,1991; Way et al.,1991).

Vulval cells and lineages were examined using Nomarski optics (see Wood, 1988). GFPexpression was examined using a Zeiss Axioplan microscope equipped with a GFP filter HQ485LP (Chroma Technology), a power source (Optiquip 1500) and a 200 W OSRAM Mercury bulb.

Laser ablations were performed as described by Avery and Horvitz(Avery and Horvitz, 1987). Early- to mid-L3 stage worms were chosen for the study.

hs::lin-11 animals (syEx500 and syEx530) were pulsed at different temperatures (between 30°C to 33.5°C) and duration(15 minutes to 1 hour) during Pn.px and Pn.pxx stages. In general, stronger pulses (1 hour at 31°C or 30 minutes at 33.5°C) caused growth arrest,uncoordinated movement and larval lethality. These are probably the result of interference with the function of other LIM homeobox genes(Ruvkun and Hobert, 1998). Alternatively, high levels of lin-11 expression in neurons may interfere with their normal development(Hobert et al., 1998; Sarafi-Reinach et al., 2001). For the vulval phenotypes, we used a 20 minutes heat shock at 33°C.

lin-11 RNAi animals (hs-dslin-11i) were heat shocked during early Pn.pxxx stage for 1 hour at 33°C. After recovery at 20°C,vulval phenotypes were examined during mid-L4 stage.

To construct pPHS11.82, a 1.5 kb NsiI-KpnI lin-11 genomic fragment from the cosmid ZC247 was cloned into pPD49.83 (Mello and Fire,1995). As the construction deleted 125 bp of the hsp16-41promoter, it was restored as a NsiI fragment by PCR amplification of the vector DNA using primers FBG3 (5′CGGCTCGTATGTT-GTGTGGAATTG3′)and BBG2 (5′CGCGATGCATGATGAGG-ATTTTCGAAGTTTTTTAG3′). The resulting construct was digested with SphI and KpnI to obtain a 1.9 kb DNA fragment. In a separate experiment, a 10.8 kb ZC247 NcoI fragment was inserted in pPD49.83 to obtain pPHS11.108. pPHS11.108 was digested with SphI and KpnI and subsequently ligated with the 1.9 kb SphI-KpnI fragment to obtain pPHS11.82. The beginning and end sequences (18 nucleotides) of the lin-11 genomic fragment are 5′-ATGCATTCTTCTTCTTCG-3′ and 5′-CCATGGTTCCTATGAGGT-3′.

To construct lin-11 RNAi plasmids, lin-11 cDNA (yk452f7;kindly provided by Dr Yuji Kohara, National Institute of Genetics) was separately cloned in sense (pPHS11.12) and antisense (pPHS11.12i) orientations into pPD49.83 (see Gupta and Sternberg,2002).

The cosmid F58A3 was digested with SphI and PstI and a 13.3 kb fragment was subcloned into pPD95.73 (a gift of A. Fire, S. Xu, J. Ahnn and G. Seydoux). The beginning and end sequences (18 nucleotides) of the fragment are: 5′-GCATGC-TTTTTTTTAATT-3′ and 5′-CTGCAGCTGTAGCTTTTT-3′.

ldb-1 cDNA (yk66f4, kindly provided by Dr Yuji Kohara, National Institute of Genetics) was digested with EcoRI and NotI and a 720 bp fragment was subcloned in pBS-SK(+). In vitro RNA was synthesized using the Ambion MEGAscript kit.

ldb-1 RNAi was performed by soaking(Tabara et al., 1998). An equal amount of each RNA strand (20 μl) was mixed to generate dsRNA. For control RNA, the pBS-SK(+) vector with no ldb-1 insert was used. Worms were synchronized by 24 hours L1 starvation in M9 after bleach treatment of the adult hermaphrodites. A small aliquot of the L1 stage worms was mixed with dsRNA solution (5 to 20 μl) and 1-5 μl OP50 bacteria. Worms were incubated for 30-36 hours, at which time they were washed twice with M9 and transferred to regular plates seeded with OP50. L4 stage animals were examined for the vulval phenotype and GFP expression.

Two-hybrid experiments were performed as suggested by the manufacturer(Clontech/BD Biosciences). pGBKT7 (GAL4 DNA binding) and pGADT7(GAL4 activation) vectors were used to subclone lin-11 and ldb-1 cDNA fragments, respectively. Positive control vectors were pVA3 (murine p53 insert) and pTD1 (SV40 large T-antigen insert). To subclone lin-11 LIM domains, a 682 bp product was PCR amplified using primers lin-11-LIM-u1 (5′GGCATATGACCTCACTGGAAGAAGAGGAG3′) and lin-11-LIM-d1 (5′GGGTCGACTCGAGTCATCTGAATTGTCCTTC3′) and lin-11 cDNA as a template. The resulting product was digested with NdeI and SalI and subcloned in pGBKT7. ldb-1 LID region (553 bp) was PCR amplified using primers ldb-1-LID-u1(5′GGCA-TATGGGAAGCAAAAAAGCTACAGCTG3′) and ldb-1-LID-d1(5′GGCTCGAGGTGGCATCCGACTATTCGGCATC3′) and the template ldb-1 cDNA. PCR product was digested with NdeI and XhoI and subcloned in pGADT7. Transformed AH109 yeast cells were grown on SD/–Leu/–Trp and SD/–His/–Leu/–Trp plates at 30°C.

Previous studies of lin-11 function during vulval development have defined its requirements during differentiation of a subset of the vulval progeny: vulC and vulD (Ferguson et al.,1987). Two experiments suggested that lin-11 might be involved in patterning additional vulval cells. First, in a screen for mutants with altered patterns of egl-17::GFP and ceh-2::GFPexpression, three lin-11 alleles (sy533, sy534 and sy634) were recovered. Second, ablation of the daughters of P5.p and P7.p VPCs in lin-11 mutant animals did not alter the invagination defect of P6.p progeny, suggesting that the effects on 1° lineage are not an indirect consequence of the 2° lineage defect. To analyze the defects in lin-11 mutant vulval cells further, we used the ajm-1::GFP cell-junction marker (formerly known as MH27::GFPand jam-1::GFP) that reveals cell boundaries and allows the in vivo study of cell fusion events (Podbilewicz and White, 1994; Mohler et al., 1998; Raich et al.,1999). Sharma-Kishore et al. have used the marker to describe the process of wild-type vulval development(Sharma-Kishore et al.,1999).

We examined ajm-1::GFP vulval expression in mid-L4 stage animals. Fig. 2A shows a typical wild-type expression (ventral view) in the form of seven concentric rings that arise from specific fusion between vulval cells(Sharma-Kishore et al., 1999). These rings are visible at different focal planes since vulval cells have invaginated significantly (see Fig. 2B for a schematic representation). Each of the vulval rings represents one cell type (vulA-vulF; Fig. 2B). By contrast, ajm-1::GFP expression in lin-11(n389) animals reveals dramatic defects in cell fusion events. In all 12 animals examined, only two or three vulval rings could be seen;moreover, the rings were visible in the same focal plane (Fig. 2C,D,F,G). One of these rings was unusually large (big arrowheads in Fig. 2C,D)and encircled one or two smaller odd-shaped rings (small arrows in Fig. 2C,D). The big ring corresponds to the P5.p and P7.p progeny that have fused together, while the smaller ones belong to the P6.p progeny. In 30% of these animals some of the 2° lineage cells did not fuse with the large syncytium and remained isolated (red star brackets in Fig. 2C,F). Some of these cells (two to six) showed punctate ajm-1::GFPexpression, suggesting partial fusion with the surrounding hyp7 syncytium(Fig. 2C). To confirm that the large syncytium is 2° lineage specific, we ablated the P6.p progeny during early L3 stage and found no change in the formation of large syncytium(n=4; Fig.2E).

We also observed defects in ajm-1::GFP expression in the 1°lineage vulval cells. In these cases, the defect appeared to be the smaller size of the vulE and vulF rings, and their failure to align correctly (small arrows in Fig. 2C,D). To determine whether the 1° lineage defect is due to the autonomous requirement of lin-11, we ablated P5.p and P7.p cells during early L3 stage. Such ablated animals still showed morphological defects in vulE and vulF rings (n=3, Fig. 2H). Together, these results provide strong evidence that lin-11 is necessary for the development of all 1° and 2°lineage vulval cells. We did not observe any cell fusion defect in lin-11(n389) animals at the earlier stages of vulval development.

To study the properties of developing vulval cells in lin-11mutant animals further, we examined cell-type specific markers that are expressed in overlapping subsets of vulval cells. These include egl-17 (fibroblast growth factor receptor homolog), cdh-3(cadherin family), zmp-1 (zinc metalloprotease) and ceh- 2(homeobox family) (Pettitt et al.,1996; Burdine et al.,1998; Wang and Sternberg,2000; Inoue et al.,2002).

egl-17 expression in vulval cells has been shown to be a faithful marker of wild-type development (Burdine et al., 1998; Ambros,1999; Wang and Sternberg,1999). The earliest expression of egl-17::GFP is detected in P6.p. The expression continues in the P6.p lineage during Pn.px (VPC progeny after first cell division, x denotes both anterior and posterior cells) and Pn.pxx (VPC progeny after second cell division) stages. However,after third cell division of VPCs (Pn.pxxx cells) egl-17::GFPexpression is no longer detected in the P6.p lineage but instead is expressed in the vulC and vulD progeny of the 2° lineages(Burdine et al., 1998) (Fig. 3A,B). We find that early expression of egl-17::GFP in P6.p lineage is not altered in lin-11(n389) animals. However, during the Pn.pxxx stage,the 2° lineage cells fail to express a detectable level of GFP(Burdine et al., 1998) (Fig. 3C,D; Table 1). However, the P6.p lineage cells continue to express low levels of GFP, suggesting a defect in their differentiation (Fig. 3C,D). Thus, vulC, vulD cells (2° lineage) and vulE, vulF cells (1° lineage)have not acquired the correct identity.

cdh-3 belongs to a cadherin superfamily of genes, members of which are known to play various roles in epithelial morphogenesis such as cellular adhesion and cell shape changes (Gumbiner,1996; Pettitt et al.,1996; Costa et al.,1998; Hill et al.,2001). The wild-type cdh-3::GFP (syIs50)expression in vulval cells is detected during L4 stage in the vulC, vulD, vulE and vulF cells (Fig. 3E,F; Table 1). In lin-11(n389) animals, cdh-3::GFP expression is completely abolished in the presumptive vulC and vulD cells (Fig. 3G,H; Table 1). In addition,expression in the 1° lineage cells is considerably affected: presumptive vulF shows very weak GFP fluorescence, whereas vulE shows fluorescence only on rare occasions (Fig. 3G,H; Table 1). Thus, lin-11is required for the differentiation of the vulC, vulD, vulE and vulF cell types.

The zmp-1 gene encodes a zinc metalloprotease and has been used as a marker for the 1° lineage cells(Wang and Sternberg, 2000; Inoue et al., 2002). zmp-1::GFP (syIs49) expression in the vulva begins during the late-L4 stage, first detected in the vulD and vulE and later on in the vulA as well (Fig. 3I,J; Table 1)(Wang and Sternberg, 2000; Inoue et al., 2002). We did not find any zmp-1::GFP expression in lin-11(n389) vulval cells (Fig. 3K,L; Table 1). Thus, lin-11is also necessary for the development of the vulA cell type.

We also examined the expression of a homeodomain family member, ceh-2, which is expressed during L4 stage in the vulB1, vulB2 and vulC cells (syIs54) (Inoue et al., 2002) (Table 1). In lin-11(n389) animals, vulval cells failed to express ceh-2::GFP in any of the cell types(Table 1).

The defects in vulval markers expression and cell fusion in lin-11mutant animals reveal requirements of lin-11 in the specification of both 1° and 2° lineage cells. It is possible that lin-11functions non-autonomously to specify some of the cell fates. To address this possibility, we carried out cell ablation experiments in wild-type and lin-11(n389) animals. We ablated a subset of the vulval cells during Pn.px and Pn.pxx stages, and examined expression of four GFP markers(egl-17::GFP, zmp-1::GFP, cdh-3::GFP and ceh-2::GFP) in the progeny of the remaining cells during L4 stage(Table 1). Four different ablation sets were analyzed [Set 1 (vulA, B1 and B2), Set 2 (vulC and D), Set 3 (vulA, B1, B2, C and D) and Set 4 (VulE and F)]. We found that in wild-type control animals, all four GFP markers are expressed in cell-autonomous manner,i.e. ablation of a subset of the vulval cells did not alter GFPexpression pattern in the progeny of the remaining cells (compare intact and cell ablated animals in Table 1). A similar conclusion for the zmp-1::GFP was drawn earlier by Wang and Sternberg (Wang and Sternberg, 2000). Having examined the autonomy of the GFP markers in wild-type animals, we carried out similar sets of cell ablations in lin-11(n389) animals. The results, summarized in Table 1, demonstrate that lin-11 functions in all vulval cells and specifies their fate in cell-autonomous manner. We can not rule out the possibility of complex interactions. The cell fusion defects are likely to be a secondary consequence of defects in cell fate specification.

Our analyses have revealed broader requirements for lin-11 during vulval development. The vulval expression of lin-11 using lin-11::lacZ reporter assays was previously reported to be in the N and T cells of the 2° lineages (precursors of vulC and vulD) (G. A. Freyd,PhD thesis, Massachusetts Institute of Technology, 1991)(Struhl et al., 1993). This pattern of expression did not provide a suitable explanation for our observations on the lin-11 mutant phenotypes. To determine the spatial and temporal pattern of lin-11 vulval expression precisely,we generated several lin-11::GFP transgenic lines(Gupta and Sternberg, 2002). Two of these, syIs80 and syIs53, were chosen for detailed analysis. Another lin-11::GFP integrant, nIs96(Reddien et al., 2001), was also analyzed. The developmental profile of the lin-11::GFP vulval expression in all three lines is nearly identical, although their fluorescence brightness can be ranked nIs96>syIs80>syIs53. The syIs80 and syIs53 animals reveal dynamic changes in the vulval GFP expression.

The earliest GFP expression in syIs80 vulval cells is detected in one of the two daughters of the 2° lineage precursors (P5.pp and P7.pa cells) (Fig. 4A,B, Fig. 5A). In most cases, GFP fluorescence was detectable only ∼1-2 hours before the VPC daughters were beginning to divide. At this stage, expression in P6.p daughters is much weaker and rarely observed (Fig. 5A). During the Pn.pxx stage, vulval cells begin to reveal brighter GFP fluorescence in both the 1° and 2° lineages (Fig. 4C,D, Fig. 5B). In the 2°lineage, expression is typically seen in only the N and T cells(Fig. 5B). By the Pn.pxxx stage, lin-11::GFP expression is detected in all 2° lineage progeny (Fig. 4E,F, Fig. 5C). In general, vulA has the lowest level of expression compared with others. syIs53 animals reveal a similar pattern of expression, although the overall fluorescence is considerably reduced (compare Fig. 5D with 5B, and Fig. 5E with 5C). The expression of lin-11 in vulval cells is consistent with the cell fusion defects and marker gene expression studies in lin-11 mutant animals. Together,these results further support the hypothesis that lin-11 plays a role in the development of all vulval cell types.

By mid-L4 stage, the GFP fluorescence in syIs53 and syIs80 strains begins to fade, and can not be seen by late-L4 stage. However, in nIs96 animals fluorescence can be detected in young adult animals. Expression is also detected in the uterine π lineage cells, VC neurons and a subset of the head and tail neurons in a manner similar to that reported earlier (Hobert et al.,1998; Newman et al.,1999). In addition, we observe expression in the B.pap and its descendents in the developing male proctodeum.

The vulval expression of lin-11 suggests an earliest requirement in VPC daughters (Pn.px cells). In wild-type animals, the vulC and vulD cells of the 2° lineage invaginate during L4 stage(Fig. 1). A high level of lin-11 expression in their precursors at Pn.px and Pn.pxx stages suggests that in wild-type animals, LIN-11 activity could specify the ability of cells to invaginate, consistent with the vulval invagination defect observed in lin-11 mutant animals. To determine whether ectopic expression of lin-11 can alter vulval cells fates and therefore invagination pattern, we generated transgenic animals carrying full-length lin-11 genomic DNA under the control of the heat-shock promoter, hsp16-41. Such animals (hs::lin-11) were heat shocked at different stages (Pn.px, Pn.pxx and Pn.pxxx) and analyzed for the vulval morphology phenotype. Although the heat shock given at the Pn.pxxx stage did not cause a noticeable defect in vulval morphology, heat shocks at the other two stages caused ectopic invagination(Fig. 6). Specifically, the vulA, vulB1 and vulB2 cell types that normally remain adhered to the epidermis in wild type had invaginated (Fig. 6A,C;compare with wild-type in Fig. 7A). The defect was qualitatively similar after the heat shock at Pn.px or Pn.pxx stage, although the penetrance was higher at the Pn.px stage. In most cases, only a subset of the P5.p and P7.p lineage cells showed ectopic invagination (90%, n=19; Fig. 6A), although in one animal all vulval cells were completely invaginated (Fig. 6C). This phenotype suggests that ectopic expression of lin-11 in the precursors of vulA, vulB1 and vulB2 interferes with their normal development,and possibly alters their cell fates. This hypothesis was further supported by our observation that in some cases (two out of six) ectopically invaginated cells showed expression of the egl-17::GFP, a marker for wild-type vulC and vulD cell fates (see Fig. 6B for ectopic expression in P7.p lineage vulA; Fig. 3A,Bshows wild-type pattern). By contrast, no such defect was observed in control heat shock experiment. We conclude that during Pn.px and Pn.pxx stages lin-11 expression in the precursors of vulC and vulD promotes a wild-type vulval invagination.

In addition to the invagination defects, we observed a weak multivulva(Muv) phenotype from the heat shocks given during the early Pn.px stage (16%, n=69; average VPC induction 3.2). Two major signaling pathways that function during vulval induction are the EGF-receptor and LIN-12/Notch mediated pathways (reviewed by Greenwald,1997; Wang and Sternberg,2001). We tested the involvement of EGF-receptor signaling using a hypomorphic lin-3 allele, n378, and observed complete suppression of the Muv phenotype. In addition, hs::lin-11 does not suppress the vulval induction defect of n378 (VPC induction 0.8, n=36 compared with the control n378 heat-shocked animals 0.7, n=35). This epistasis of n378 over hs::lin-11suggests that the effect of lin-11 overexpression is likely to be upstream of lin-3, and possibly at the level of lin-3transcription. We are not convinced that such an effect of lin-11 is physiologically relevant because none of the known alleles of lin-11exhibit defects in the extent of vulval induction, and lin-11::GFPtransgenic animals (nIs96, syIs80 and syIs53) do not reveal GFP fluorescence in the anchor cell (AC) during L2/L3 stages when AC is required for vulval induction. It is more likely that the overexpression of lin-11 mimics the effect of some other homeodomain or LIM homeodomain protein.

Our experiments so far have defined the function of lin-11 during the Pn.px and Pn.pxx stages in establishing the correct pattern of vulval invagination. lin-11 continues to be expressed at high levels in Pn.pxxx cells and thus might be required for vulval differentiation. To test this hypothesis, we used a conditional RNAi approach to inactivate lin-11 gene function. We generated transgenic hs-dslin-11ianimals (carrying lin-11 cDNA in sense and antisense orientations under control of the hsp16-41 heat-shock promoter) that can be heat shocked at any desired developmental stage to induce the formation of double-stranded RNA. As a control, we heat shocked hs-dslin-11ianimals during early L3 stage (Pn.p cells in Fig. 1) and compared the vulval morphology and egg-laying phenotypes with the lin-11 loss of function alleles. Forty percent (n=8) of the heat-shocked animals showed an egg-laying defective (Egl) phenotype and a weak vulval invagination defect similar to the lin-11(n566) allele. One of them also exhibited the AC migration defect, a phenotype that contributes to the Egl defect in lin-11 mutant animals (Newman et al., 1999). In control heat-shocked animals (no hs-dslin-11i), no such defect was observed (n=20). These results confirm that the RNAi phenotypes of hs-dslin-11i animals are due to the reduction in the wild-type lin-11 gene function. Next, we examined the effect of the lin-11 RNAi on vulval morphology by heat shocking hs-dslin-11i animals during early Pn.pxxx stage (early-L4). In wild-type animals during the mid-L4 stage, vulval nuclei occupy stereotypical positions, such that vulD nuclei of the P5.p and P7.p lineage are located in the same plane of focus(Fig. 7A). Likewise, vulE and vulF nuclei are seen in one focal plane, different from that occupied by vulD nuclei (Fig. 7B). By contrast,the heat-shocked hs-dslin-11i animals showed significant defects in the vulval morphology with misplaced vulval nuclei (30%, n=16; Fig. 7C,D). Specifically, we found that vulC and vulD nuclei were located in wrong focal planes (vulD is shown in Fig. 7C,D). Two out of five defective animals also showed abnormal positioning of the nuclei of 1° lineage cells (see Fig. 7D for vulF position, vulE nuclei are not seen in this plane). Overall, vulval invagination was narrower along the anteroposterior axis compared with the wild type. This abnormal morphology was correlated with a protruding vulva phenotype at the adult stage(Fig. 7E). However, such animals were able to lay eggs normally.

We also examined egl-17::GFP expression in lin-11 RNAi animals during the mid-L4 stage (∼4 hours after the heat shock treatment). Although the overall GFP pattern was qualitatively wild type (n=9;see Fig. 3A,Bfor the wild-type egl-17::GFP pattern), two worms did show moderate reduction in the GFP fluorescence in vulC and vulD. These results reveal a novel function of lin-11 in vulval morphogenesis, distinct from its early role in specifying the invagination pattern.

The LIM homeodomain proteins contain a pair of LIM domains that interact with co-factors and modulate protein activity(Dawid et al., 1998; Bach, 2000; Hobert and Westphal, 2000). Among the co-factors are the LIM-binding proteins represented by NLI/Ldb1/CLIM2, which display highly specific interactions with the LIM domains. The C. elegans LIM-binding protein LDB-1 has been shown to interact with two LIM homeodomain proteins CEH-14 and MEC-3(Cassata et al., 2000). As it is the only known LIM-binding protein in C. elegans, it is likely that LDB-1 regulates the activities of other LIM homeodomain proteins,including LIN-11.

To investigate the role of ldb-1 during vulval development, we examined its expression using a ldb-1::GFP construct. ldb-1::GFP expression is first detected in embryos towards the end of gastrulation (prior to the comma stage) and is seen in the vulval cells among other expression (Cassata et al.,2000) (not shown). In vulval cells ldb-1::GFP expression is observed in both the 1° as well as 2° lineage cells(Fig. 8). Expression during the Pn.px stage was rarely observed (<5%, n=28; Fig. 8I). During Pn.pxx stage,a weak and low penetrant GFP could be detected in 1° and 2°lineage cells (16%, n=25; Fig. 8A,B,I). However, by the Pn.pxxx stage strong and highly penetrant ldb-1::GFPexpression could be detected in all vulval progeny (Fig. 8C-F,I). Expression was also observed during early adult stage(Fig. 8G-I).

We examined the effect of decreasing ldb-1 activity using RNAi. ldb-1 RNAi animals showed uncoordinated movement and failed to respond to touch, a phenotype that has been previously described (Cassatta et al., 2000) (not shown). We also observed defects in vulval morphology and gonad arms at significant frequencies (vulval defect: 26%, n=101, see Fig. 9; gonad defect: 40%, n=18). Vulval phenotypes included abnormal placement of the 1°and 2° lineage cells (Fig. 9A,C,E). Occasionally, animals displayed a protruding vulva phenotype and were Egl. We did not observe a vulval invagination defect in any of the RNAi animals. We also examined the effect of ldb-1 RNAi on vulval markers, egl-17::GFP, zmp-1::GFP, cdh-3::GFP and ceh-2::GFP. In all but the case of zmp-1::GFP, GFP fluorescence was altered(Fig. 9, compare with wild-type patterns in Fig. 3). egl-17::GFP expression was frequently absent in the presumptive vulC(Fig. 9A,B). The cdh-3::GFP showed reduced or no expression (Fig. 9C,Dshows only vulD expression). ceh-2::GFP expression in the presumptive vulB1/B2 cells was variable and weak, whereas presumptive vulC often showed no expression (Fig. 9E,F). In control RNAi animals, no such defects were observed. Thus, LDB-1 plays a role in vulval morphogenesis. In Drosophila, the LIM homeodomain protein Apterous and its LIM-binding partner Chip form a regulatory feedback network to modulate the expression of each other(Milan and Cohen, 2000). We examined the possibility of such feedback regulation between LDB-1 and LIN-11,but found no change in the expression of lin-11::GFP in ldb-1 RNAi animals.

We examined physical interaction between LIN-11 and LDB-1 using a two-hybrid interaction assay (Fields and Song, 1989). For this, we designed two separate expression constructs. One construct expresses LIM domains of LIN-11 as a fusion protein with the GAL4 DNA-binding domain, whereas the other expresses LDB-1 LIM-interacting domain (LID) fused to the GAL4 activation domain (see Materials and Methods). Fig. 10B shows that fusion proteins involving LIN-11 LIM domains and LDB-1 LID interact with each other thereby leading to the growth of yeast cells on plates lacking histidine, leucine and tryptophan. Neither one alone(Fig. 10B-2,B-3)is sufficient for HIS3 expression. In a control experiment(Fig. 10A), all transformants grew on plates lacking leucine and tryptophan. These results demonstrate a physical interaction between LIN-11 and LDB-1.

Earlier studies on lin-11 using cell lineage(Ferguson et al., 1987) and lin-11::lacZ expression (G. A. Freyd, PhD thesis, Massachusetts Institute of Technology, 1991) (Struhl et al., 1993) have shown its function in specifying the vulC and vulD cell types. Our results extend these findings and demonstrate that lin-11 is necessary for the differentiation of all vulval cell types,including the 1° lineage cells.

In wild-type C. elegans during the L4 stage, vulval cells initiate the process of invagination (Fig. 1) and specific cell fusion to give rise to the adult structure. The cell fusion events are ordered and occur only between the homologous cell types, e.g. P5.p lineage vulA fuses only with the P7.p lineage vulA(Sharma-Kishore et al., 1999). By contrast, lin-11 mutant vulval cells exhibit defects in cell fusion events. Often all 2° lineage vulval cells fuse together, suggesting that they have acquired a common fate. However, this fate is distinct from any of the wild-type cell fates as none of the examined markers (egl-17::GFP,zmp-1::GFP, cdh-3::GFP and ceh-2::GFP) is expressed in lin-11 mutant vulval cells. Using cell ablation experiments, we have shown that defects in cell fusion events and marker gene expression in lin-11 mutant animals result from cell-autonomous requirements of lin-11. The phenomenon of a cell fusion defect is similar to that observed for ray fusion in C. elegans mutants affecting male tail development (Baird et al.,1991; Chow and Emmons,1994).

Analysis of the ajm-1::GFP expression in 1° vulval cells in lin-11 animals has revealed defects in the morphology of vulE and vulF rings, consistent with the abnormal patterns of egl-17::GFP,cdh-3::GFP and zmp-1::GFP expression (see Fig. 3). Hence, the 1°lineage cells are not likely to form a functional vulval opening. These results are consistent with our previous findings on tissue-specific regulation of lin-11, where we used vulval- and uterine-specific regulatory elements of lin-11 to demonstrate that wild-type egg-laying requires lin-11 function in both the vulva and the uterineπ lineage cells (Gupta and Sternberg,2002).

The vulval cell fusion defects in lin-11 mutant animals could arise because lin-11 directly regulates the process of cell fusion or as a consequence of abnormal differentiation. Two sets of results support the latter possibility. First, lin-11 expression in vulval cells is detected beginning at the Pn.px stage (see Fig. 4). Second, induction of lin-11 RNAi (using hs-dslin-11i) during mid-L4 stage causes no significant effect on the formation of vulval rings. Thus, during terminal differentiation, lin-11 mutant vulval cells might fail to express cell-type specific genes, leading to the defects in fate specification. The abnormal cell fusion is the consequence of cells failing to acquire their unique identities. Such a role of lin-11 in the vulva is similar to that of C. elegans LIM homeobox gene, mec-3, in touch receptor neurons. In mec-3 mutant animals, the presumptive touch neurons are generated but fail to acquire the correct identity(Way and Chalfie, 1988).

LIM homeobox genes have been shown to express in highly restricted spatial and temporal manner (Bach,2000; Hobert and Westphal,2000). Wing development in Drosophila requires dynamic expression of the LIM homeobox gene apterous. The level and domain of apterous expression are highly regulated and help define the dorsoventral boundary leading to wing growth and patterning(Diaz-Benjumea and Cohen,1993; Milan and Cohen,2000).

The dynamic expression of lin-11 in vulval cells can be classified into two distinct patterns: an initial polarized expression (during Pn.px and Pn.pxx stages) where only a subset of the cells express lin-11, and a broad pattern of expression during terminal differentiation (Pn.pxxx cells)where all 2° lineage cells express lin-11. We hypothesized that these two different patterns of lin-11 expression may have different functions and tested the hypothesis experimentally. First, a hs::lin-11 system was used to express lin-11 ectopically in all vulval cells during Pn.px and Pn.pxx stages. This led to defects in vulval invagination caused by the failure of presumptive vulA, vulB1 and vulB2 to remain adhered to the epidermis (Fig. 6; wild-type pattern in Fig. 1). Second, using a RNAi approach, we inhibited lin-11function during early Pn.pxxx stage when lin-11 is expressed in all 2° lineage cells. The lin-11 RNAi animals showed defects in vulval morphology and vulval nuclei failed to occupy stereotypic positions. This phenotype is likely to result from a differentiation defect in vulval progeny.

The two distinct requirements of lin-11 in vulval cells are likely to be mediated by different target genes. The vulval invagination defect in lin-11 animals suggests that one potential target of lin-11could be the genes that regulate epithelial morphogenesis. Our reporter gene expression studies have identified a cadherin family member, cdh-3,that functions downstream of lin-11. In lin-11 mutant vulval cells cdh-3::GFP expression in the presumptive vulC and vulD is abolished (Fig. 3; Table 1). Cadherins are known to regulate epithelial morphogenesis by mediating adhesions between cell-cell and cell-extracellular matrix (Gumbiner,1996). However, the function of cdh-3 in vulval development is not essential, perhaps owing to redundancy.

The early expression of lin-11 is polarized and confers identity on cells to give rise to progeny (vulC and vulD) that invaginate during L4 stage (see Fig. 1). This conclusion is also supported by the roles of lin-17 (a frizzled family member) (Sawa et al., 1996) and lin-11 during vulval development(Gupta and Sternberg, 2002). In lin-17 mutant animals, lin-11 expression in P7.p lineage cells is often reversed, i.e. LL lineage cells begin to express lin-11 instead of the wild-type NT lineage cells. This reversal in the polarity of lin-11 expression correlates with the opposite orientation of invagination of the P7.p lineage cells. A similar role for lin-11 has also been demonstrated in the specification of the ASG and AWA neurons (Sarafi-Reinach et al.,2001). Although during embryonic stages both neurons express lin-11, expression in AWA is lost by the L1 larval stage and persists only in the ASG neuron. This later stage expression of lin-11 in ASG is necessary for its wild-type development. If lin-11 is ectopically expressed in AWA during post-L1 larval stages, the AWA adopts partial ASG-like features. Similar functions of other LIM homeobox genes in determining polarity or asymmetric cell fates have also been demonstrated. C. elegans lim-6, another LIM homeobox gene, generates functional differences in the chemosensory behavior between a pair of neurons ASEL/R(Pierce-Shimomura et al.,2001). In mouse embryonic axis formation, the role of lim1 in anterior-posterior polarity is also suggestive of such a biological function (Perea-Gomez et al.,1999). In this case, lim1 expression in the anterior region cells of the visceral endoderm confers anterior identity and makes them different from the posterior cells. Thus, the role of the LIM homeobox genes in generating cellular asymmetry appears to be a conserved biological function.

How does lin-11 play different roles at different stages of vulval development? One possibility could be that LIN-11 interacts with stage- and cell-type specific factors to bring about the different outcomes. The LIM domains of the LIM homeodomain proteins are known to serve as protein-protein interacting interface that promote the formation of multimeric complexes and influence DNA-binding affinity of the homeodomain(Dawid et al., 1998). Many studies have revealed the presence of LIM domain-binding proteins(Dawid et al., 1998; Bach, 2000; Hobert and Westphal, 2000). Although a majority of them belong to the NLI/Ldb1/CLIM2 family, others such as POU homeodomain factor Pit1 (Bach et al., 1995), WD40 repeat containing factor SLB(Howard and Maurer, 2000) and bHLH factor E47 (German et al.,1992) have also been identified.

The C. elegans LIM-binding protein LDB-1 was previously shown to be required for the wild-type functioning of several neurons(Cassata et al., 2000). We find that ldb-1 is expressed in both the 1° as well as 2° lineage vulval cells, a pattern that overlaps with lin-11 expression. Analysis of the ldb-1 vulval expression has revealed some differences from lin-11. During Pn.pxx stage, lin-11::GFP shows alternating low and high pattern of expression in P5.p and P7.p lineage cells(LLHH-HHLL, respectively, from anterior to posterior; L, low; H, high). ldb-1::GFP, however, shows no such pattern and is detected at uniform level in all 1° and 2° lineage cells. In addition, ldb-1::GFPcontinues to be expressed at high levels in 1° vulval progeny during L4 and young adult stages, whereas lin-11::GFP expression in 1°lineage cells is significantly weaker compared with the 2° lineage cells. Hence, LDB-1 may regulate only a subset of the LIN-11 functions. Consistent with this, ldb-1 RNAi animals did not show defects in vulval invagination but in vulval morphology (Figs 7, 9). However, it is possible that ldb-1 RNAi effect is weak because of the partial elimination of gene activity. Our findings that LIN-11 and LDB-1 physically interact support the hypothesis that LIN-11 and LDB-1 function together to regulate vulval differentiation. Furthermore, these results suggest that lin-11 may use other mechanisms during earlier stages of vulval development.

We thank S. Cameron, D. Sherwood and T. Inoue for integrated strains. Robert Oania provided assistance in two-hybrid experiments. Some of the strains were obtained from Caenorhabditis Genetics Center. We also thank D. Sherwood, T. Inoue and anonymous reviewers for comments on the manuscript. This work was supported by funds from the HHMI and USPHS grant HD23690 to P.W.S. B.P.G. was supported by a long-term fellowship from HFSP and as an associate of the HHMI. M.W. was an AMGEN fellow. P.W.S. is an investigator of the HHMI.