Components of the transcriptional Mediator complex are required for asymmetric cell division in C. elegans

In every organism, asymmetric cell divisions are crucial to the generation of cell diversity (Hawkins and Garriga,1998; Horvitz and Herskowitz,1992). In Drosophila, asymmetric divisions of neuroblasts cause the Prospero protein to be segregated into only one daughter cell, the one that becomes a ganglion mother cell (GMC). Prospero is a transcription factor that is required for the GMCs to adopt their fates correctly(Betschinger and Knoblich,2004; Roegiers and Jan,2004). In budding yeast, asymmetric cell division contributes to the mating-type switch, which involves the rearrangement of specific DNA segments at the MAT locus. This process is catalyzed by the HO endonuclease,which is expressed in mother but not daughter cells. Thus, mating-type switching occurs only in mother cells(Amon, 1996; Nasmyth, 1993). Transcription of the HO gene is dependent on the SWI/SNF chromatin remodeling complex and the Mediator complex. During telophase, SWI/SNF binds to the HO promoter in the nucleus of mother cells, and recruits the Mediator complex and RNA polymerase to facilitate the transcription of HO(Cosma, 2002; Cosma et al., 1999).

The Mediator complex was first identified in yeast as a complex associated with RNA polymerase that can support activated transcription in vitro(reviewed by Myers and Kornberg,2000). A number of mammalian complexes related to yeast Mediator have since been identified, the TRAP, DRIP, ARC and SMCC complexes, which have nearly identical subunit compositions(Malik and Roeder, 2000). These complexes can mediate the activities of various transcription factors,such as Sp1, thyroid hormone receptor and p53, to activate or repress transcription. The largest Mediator complexes contain about 20 subunits, but they seem to be divided into functional and physical submodules. It has been suggested that yeast Mediator can be divided into four modules: Srb4,Gal11/Sin4, Med9/Med10 and Srb8-Srb11. For example, yeast mutants of the Gal11/Sin4 module components (Gal11, Rgr1, Sin4, Med2 and Pgd1) exhibit similar phenotypes (Jiang et al.,1995; Jiang and Stillman,1995), and the presence of Gal11, Sin4 and Pgd1 in the complex depends on Rgr1 (Li et al.,1995). Under highly stringent conditions, the Srb8-Srb11 module is isolated as a separate entity from the other components of Mediator(Borggrefe et al., 2002), and this module has repressive functions in yeast(Carlson, 1997; Chang et al., 2001; Holstege et al., 1998). CDK8 and cyclin C, the human homologs of Srb10 and Srb11, respectively, also repress activator-dependent transcription in vitro(Akoulitchev et al., 2000). In addition, the human ARC-L complex, a large Mediator complex, is transcriptionally inactive and contains CDK8 and Cyclin C, as well as MED12 and MED13 (homologs of Srb8 and Srb9, respectively)(Taatjes et al., 2002). Therefore, Srb8/MED12, Srb9/MED13, Srb10/CDK8, and Srb11/Cyclin C associate with each other physically and functionally in yeast and human cells. In Drosophila, the MED12 and MED13 homologs are involved in the development of the eye and wing (Janody et al., 2003; Treisman,2001). However, little is known about how these complexes are regulated or contribute to animal development.

In C. elegans, the asymmetric division of certain blast cells,including the T blast cell, is regulated by lin-17/frizzled and lin-44/wnt (Herman et al.,1995; Sawa et al.,1996). In lin-17 mutants, the asymmetry of the division is disrupted, resulting in symmetric division(Sternberg and Horvitz, 1988). In lin-44 mutants, the polarity of the division is reversed(Herman and Horvitz, 1994). It has been proposed that the LIN-44 signal, which acts through the LIN-17 receptor, provides polarity to cells that undergo asymmetric division(Sawa et al., 1996). The Wnt pathway, which controls the polarity of the T cell, shares some components with the canonical Wnt pathway, such as a Tcf homolog POP-1(Herman, 2001). We have previously shown that PSA-1 and PSA-4, components of the SWI/SNF complex, are required for the asymmetric division of the T cell during mitosis, suggesting that distinct cell fates are determined by alteration of the chromatin structure (Sawa et al., 2000). Recently, it was reported that a putative transcription factor, TLP-1, is expressed asymmetrically in the T-cell daughters, and this asymmetric expression is regulated by the Wnt signaling pathway, suggesting that tpl-1 is one of the target genes of the Wnt signal(Zhao et al., 2002). However,it is not clear how the Wnt signal regulates the transcription of its target genes.

We have identified mutations in the let-19 and dpy-22genes that affect the asymmetric division of the T cell. The let-19and dpy-22 mutations cause symmetrical expression of tlp-1in the T-cell daughters. We cloned these genes and found that they encode homologs of MED13 and MED12, components of the transcriptional Mediator complex. LET-19 and DPY-22 also function in the fusion of the Pn.p cells, a process that is also regulated by the Wnt signaling pathway. These results indicate that LET-19 and DPY-22 encode components of the Mediator complex and regulate asymmetric cell division, as the complex does in yeast.

The methods for the culture and genetic manipulation of C. eleganswere as described (Brenner,1974). Transgenic animals were generated as described(Mello et al., 1991). The let-19 and let-425 mutants are sterile and were maintained as heterozygotes over GFP-balancers, mIn1[mIs14] and nT1[qIs51], respectively. The dpy-22(os38) mutants are fertile but semi-sterile. They were in most cases maintained as rescued strains with an extrachromosomal array (osEx89) carrying a genomic subclone of the dpy-22 gene (pPS6.10) and col-10::GFP. Homozygous or non-array-bearing mutants were identified as non-green animals under the fluorescence dissecting scope. Animals were grown at 22.5°C unless otherwise noted. The Psa phenotype was determined as described(Sawa et al., 2000). Expression of tlp-1::GFP in the T-cell daughters was analyzed after the V6 cell division. qIs74 was used for GFP::POP-1(Siegfried et al., 2004).

pAY104 (a rescuing plasmid for let-19) contained both a 9.1 kb PstI fragment of F07H5 (with a 0.4 kb sequence from the Lorist6 cosmid vector) and a 4.1 kb PstI fragment of F07H5 subcloned into the pBSK vector. The let-19::GFP construct (pAY105) was made by inserting a 0.1 kb PCR fragment (from the BstEII site to the C terminus of the let-19 gene) and a GFP fragment from pPD95.79 (a gift from A. Fire)into the BstEII site of the let-19 rescuing plasmid(pAY104). To identify mutations, we sequenced the PCR products amplified from let-19 and dpy-22 mutants using internal primers. The mutations were confirmed by sequencing different PCR fragments. The sur-2::HA construct (pAY106) consisted of a 10.4 kb SacI-BanI fragment of F39B2, a 0.15 kb PCR fragment just upstream of the stop codon and a HA fragment subcloned into the pT7Blue vector. The expression of GFP::POP-1, tlp-1::GFP, let-19::GFP and dpy-22::GFP was analyzed by confocal microscopy (Zeiss LSM510), while that of mab-5::GFP was analyzed by epifluorescence microscopy.

HS490 [harboring SUR-2::HA in a sur-2(ku9) mutant background] and HS518 [harboring SUR-2::HA and LET-19::GFP in a sur-2(ku9) mutant background] strains were grown in liquid culture as described previously(Stiernagle, 1999). To prepare nuclear extracts, the animals were harvested and homogenized essentially as described previously (Mains and McGhee,1999), except that the nuclear pellets were obtained from sonicated homogenates of mixed-stage animals, including embryos, larvae and adults, and the nuclear pellets were extracted with NEB350 [nuclear extraction buffer: 20 mM HEPES (pH 7.6), 350 mM KCl, 2 mM EDTA, 25% glycerol, 0.5 mM DTT,1 mM PMSF, 10 μM E-64, and 0.1% Nonidet P-40]. The nuclear extracts were then co-immunoprecipitated with an anti-Flag antibody (M2, Sigma) or anti-GFP antibody (3E6, Quantum biotechnologies) conjugated to protein A-Sepharose beads (Amersham Pharmacia Biotech) overnight at 4°C. The immunoprecipitates were washed four times with NEB270 (same as NEB350 except containing 270 mM KCl), and eluted with Laemmli sample buffer. For detection of LET-19::GFP and SUR-2::HA, samples were separated by SDS-PAGE (5%), and transferred onto PDVF membranes (Immobilon P, Millipore) by electroblotting for 180 minutes in 10 mM CAPS [3-(cyclohexylamino)-1-propanesulfonic acid; pH 11.0] transfer buffer containing 7.5% methanol. The membranes were immunoblotted with anti-GFP (JL-8, CLONTECH) and anti-HA (12CA5, Boehringer Mannheim), and bound antibodies were visualized with HRP-conjugated antibodies against mouse IgGs (BioRad) using a chemiluminescence reagent (Western Lightning, Perkin Elmer Life Sciences). To detect MED-6, an immunoblot analysis was performed with anti-MED-6, as described previously(Kwon et al., 1999).

In lin-17 and lin-44 mutants, the disruption of asymmetric T-cell division results in the absence of phasmid socket cells (Psa phenotype), which are generated by the T.p cell. We identified mutants of the let-19 (Herman, 1978)and dpy-22 (Hodgkin and Brenner,1977) genes in a screen for the Psa phenotype(Sawa et al., 2000)(Table 1). In addition to the Psa phenotype, as shown in Table 1, the let-19 mutants had Dpy (dumpy), Muv (multivulva)and Sterile phenotypes, and the dpy-22 mutants had Dpy, Muv and Egl(egg-laying defective) phenotypes. We determined the T-cell lineage in the let-19 and dpy-22 mutants(Fig. 1). In both mutants,symmetric division was observed, which led to the production of four hypodermal cells, as seen in the lin-17 mutants (type I lineage),indicating that these genes are required for the T cell to divide asymmetrically. To understand how these genes regulate this division, we analyzed the expression of two genes that are asymmetrically expressed between T.a and T.p in wild-type animals.

In the embryo, the Wnt pathway functions through a β-catenin homolog,WRM-1, to downregulate the levels of POP-1/Tcf in the posterior daughter of the EMS blastomere. The level of POP-1 is also lower in the posterior daughters of many cells that divide along anteroposterior axis, including that of the T cell (Herman, 2001; Lin et al., 1998). To determine the localization of POP-1 in the T cell, we used a GFP::POP-1 fusion protein (Siegfried et al.,2004). In wild-type animals, the level of GFP::POP-1 was lower in the posterior daughter of the T cell (n=15, Fig. 2A). As reported previously (Herman, 2001), the level of GFP::POP-1 was symmetric in the lin-17 mutants(n=2, Fig. 2B). In let-19 (n=6) and dpy-22 (n=6) animals, the levels of POP-1 were higher in the anterior T-cell daughter, just as in wild-type animals (Fig. 2C,D). These results indicate that let-19 and dpy-22 do not regulate the POP-1 level, and suggest that let-19 and dpy-22function downstream of pop-1.

Zhao et al. showed that tlp-1 encodes a transcription factor that is required for the asymmetric T-cell division that functions downstream of Wnt signaling (Zhao et al.,2002). In wild-type animals, the tlp-1::GFP fusion gene was expressed in the T.p cell, but not in the T.a cell(Fig. 3A). In lin-17animals, TLP-1 expression was diminished(Fig. 3B). We then investigated the expression of tlp-1::GFP in let-19 and dpy-22mutants to determine whether the let-19 and dpy-22 genes might regulate TLP-1 expression (Fig. 3C,D; Table 2). We observed that GFP was expressed symmetrically in both the T.a and T.p cells in let-19 animals (9/11) and dpy-22 animals (8/13). These results suggest that the asymmetric tlp-1 expression is regulated by let-19 and dpy-22. Furthermore, we observed the symmetrical expression of tlp-1::GFP in a double-mutant between lin-17and let-19, a pattern similar to that seen in the let-19mutant (Fig. 3E; Table 2). These results indicate that let-19 is epistatic to lin-17 and support the idea that let-19 acts downstream of the Wnt pathway.

Despite the defects in the T lineage, the tlp-1 expression in other cells appeared to be normal in the let-19 and dpy-22mutants. Specifically, tlp-1::GFP was not expressed in seam cells other than the T cells and was expressed in the posterior but not the anterior gut cells in the let-19 (n=8) and dpy-22(n=10) mutants, as well as in wild-type animals (n=8). Therefore, these genes regulate the tlp-1 expression specifically in the T-cell lineage.

To further investigate the roles of these genes in Wnt signaling, we analyzed the phenotypes of the let-19 and dpy-22 mutants in other developmental events regulated by the Wnt signaling pathway. Wnt signaling is known to regulate cell fusion(Eisenmann et al., 1998). The ventral hypodermal cells, called Pn.p cells (P1.p through P11.p), can assume alternative fates. In wild-type animals, the two anterior and three posterior Pn.p cells fuse with the hypodermal syncytium (F fate), while the six central cells (P3.p through P8.p) do not fuse and become precursor cells for the vulva(VPCs). (The P3.p cell adopts the F fate in about 50% of animals.) In mutants of the bar-1 gene, which encodes β-catenin, cell fusion occurs ectopically, producing fewer VPCs than in wild-type animals(Eisenmann et al., 1998). BAR-1 maintains the expression of LIN-39/Hox, which inhibits cell fusion. In lin-39 mutants, all the Pn.p cells fuse(Clark et al., 1993; Wang,1993). We quantified the unfused ventral hypodermal cells in the let-19 and dpy-22 mutants, using an adherens junction marker, ajm-1::GFP (Koppen et al., 2001). We first found that the Pn.p cells in these mutant animals sometimes underwent an extra division in the late L1 stage, producing extra hypodermal cells. A similar phenotype was reported for lin-25mutants (Tuck and Greenwald,1995). Despite the presence of these extra hypodermal cells in the let-19 and dpy-22 animals, cell fusion occurred less frequently than in wild-type animals (Table 3). Specifically, in five out of 16 let-19(mn19) animals and in two of 26 dpy-22(os38) animals, neither P2.p nor P2.pp fused. In addition, in four out of 26 dpy-22(os38) animals, P9.p (and in one animal, P9.pa, P9.pp and P10.p) did not adopt the F fate. Furthermore, we found that let-19 and dpy-22 mutations efficiently suppressed the bar-1 mutant phenotype(Table 3). Unfused P2.p or P9.p cells were still observed in the let-19; bar-1 or bar-1 dpy-22 double mutants. By contrast, the let-19 mutations did not suppress the lin-39 mutant phenotype. These results suggest that let-19 and dpy-22 function to repress the lin-39/Hox expression that is regulated by bar-1/β-catenin.

egl-20/Wnt and bar-1/β-catenin regulate the posterior migration of the QL neuroblast(Maloof et al., 1999; Rocheleau et al., 1997; Thorpe et al., 1997). Neither let-19 (n=22) nor dpy-22 (n=20) mutant was defective in QL-cell migration and neither significantly suppressed the QL-cell migration defects that occur in bar-1 mutants (defective in 22/22 in let-19; bar-1 and 25/27 in bar-1 dpy-22). Although Wnt signaling and the LIT-1 MAP kinase regulate endoderm induction in embryos(Maloof et al., 1999; Rocheleau et al., 1997; Rocheleau et al., 1999; Thorpe et al., 1997), the RNAi of let-19 did not cause the gutless phenotype (n>50), nor did it suppress the gutless phenotype in lit-1/NLK mutants(n=94). In addition, although mutants of ln-17/frizzled and pop-1/TCF often lack gonad arms because of the absence of distal tip cells (Siegfried and Kimble,2002; Sternberg and Horvitz,1988), let-19 (n=34) and dpy-22(n=23) mutants had the normal number of gonad arms. These results suggest that let-19 and dpy-22 are specifically involved in the Wnt signaling pathway in the T and Pn.p cells.

let-19 was mapped to the right of rol-6 on chromosome II(Sigurdson et al., 1984). let-19 mutants were rescued by cosmid F07H5 and a subclone of F07H5 that contains the predicted gene K08F8.6(Fig. 4A). The RNAi of this gene was embryonically lethal, but escapers mimicked the Psa and extra Pn.p phenotypes of let-19 (data not shown). We sequenced this gene in the let-19 mutants and found mutations in all the alleles, confirming that K08F8.6 was the let-19 gene. All the alleles had nonsense mutations, indicating that they were strong loss-of-function mutants. Consistent with this, all the alleles were fully recessive, and the Muv phenotype of mn19 homozygotes was similar to that of mn19/mnDf46, a deficiency in which the let-19 locus is deleted (data not shown). let-19 encodes a protein of 2862 amino acids that has been reported to be homologous to mammalian MED13, a component of the Mediator complex (Ito et al.,1999).

One of the let-19 mutants, os36, which has a similar phenotype to the other mutant alleles, contained a nonsense mutation that was predicted to truncate the last 29 amino acids of the protein, indicating that the C-terminal region of the LET-19 protein is essential to its function. The MED13 homologs include C. elegans LET-19, mammalian MED13, D. melanogaster Skuld (Skd), D. discoideum AMIB and yeast Srb9. All these homologs have conserved domains in their C-terminal regions(Fig. 4C)(Boube et al., 2002; Wang et al., 2004). These data imply that the C-terminal region of the MED13 family proteins is important to their function.

We searched for homologs of other components of Mediator in the C. elegans genome and found that a MED12 homolog mapped to the same region of chromosome X as dpy-22. We found that dpy-22 was rescued by cosmid F47A4 and a subclone of F47A4 that contains the MED12 homolog,F47A4.2 (Fig. 4B). The RNAi of this gene mimicked the Dpy Psa and the fertile phenotype of dpy-22(data not shown). This gene was previously identified as the sop-1gene (Zhang and Emmons, 2000). sop-1 was identified from mutations that suppress the pal-1mutant. In contrast to the dpy-22 mutants, which have a variety of phenotypes, as described above, sop-1 mutants did not exhibit any phenotypes by themselves. Most of the sop-1 mutants had nonsense mutations near the C terminus that truncated the glutamine-rich domain of the protein, except for bx103, which contained a splice-site mutation(Zhang and Emmons, 2000). We identified the mutations in three dpy-22 mutants(Fig. 4B). Among them, os38, which had the strongest phenotype, contained a nonsense mutation in the middle of the coding sequence, in addition to a missense mutation near the N terminus, suggesting that it was a strong loss-of-function mutant. sop-1 mutants are likely to be weak loss-of-function mutants of the dpy-22 gene. dpy-22/sop-1 was shown to be expressed ubiquitously during development (Zhang and Emmons, 2000).

sop-1 mutations can suppress pal-1 mutants for the production of rays from the V6 cells in males. However, dpy-22(os38)and let-19(mn19) males without the pal-1 mutation were missing most of the rays [0 rays/sides of animals in let-19(mn19)n=10 and 1.9 rays in average in dpy-22(os38) n=16]. (Both T-derived and V6-derived rays appeared to be similarly affected in the os38 animals.) We then analyzed the expression of the mab-5gene, which acts downstream of pal-1 for ray production. mab-5::GFP was often not expressed in the V6 cells in the dpy-22(os38), let-19(mn19) or pal-1 mutants(Table 4). Therefore, strong loss-of-function mutants of dpy-22 have the opposite effects of weak loss-of-function mutants (sop-1 class) on mab-5 expression in the V6 cells. If both classes of mutations affect the transcription of pal-1, let-19 and dpy-22 are likely to be involved in pal-1 transcription through its intronic enhancer element, which controls pal-1 expression (Zhang and Emmons, 2000). By contrast, Zhang and Emmons suggested that the sop-1-class of dpy-22 mutations activates pal-1transcription through another element, only when the intronic element is defective. Therefore, Mediator may regulate pal-1 expression through two distinct promoter elements. It is also plausible that let-19 and dpy-22 mutations directly disrupt the transcription of mab-5, while sop-1-class mutations affect that of pal-1.

Despite the defects in the V6 cell, the mab-5::GFP expression in other cells did not appear to be significantly affected in the let-19or dpy-22 mutants. At the early L1 stage, mab-5::GFP was not expressed in other seam cells in let-19 (17/17), dpy-22(17/17) or wild-type (11/11) animals, while it was expressed in the P9/10 and P11/12 cells in the let-19 (16/17; no expression in P9/10 in one animal), dpy-22 (17/17) and wild-type (11/11) animals. Therefore, let-19 and dpy-22 are involved in the mab-5expression specifically in the V6 cells.

To analyze the expression patterns of let-19 and dpy-22,we made constructs in which the let-19 and dpy-22 genes were fused in-frame to the GFP (green fluorescent protein) gene at the ends of their coding sequences. Each construct rescued the let-19 or dpy-22 phenotypes, respectively, indicating that the fusion proteins were functional. Using these constructs, we found both let-19 and dpy-22 to be expressed in most cells during embryogenesis and in many if not all cells in developing larvae (data not shown). As shown in Fig. 5, both genes were expressed in the T cell and the T-cell daughters. GFP fluorescence was observed in both of the daughter nuclei, indicating that there was no asymmetry in the expression patterns of let-19 and dpy-22during T-cell division.

We examined the interaction between LET-19 and other putative components of Mediator, LET-425/MED6 and SUR-2/MED23(Kwon and Lee, 2001; Singh and Han, 1995). To this end, GFP-tagged LET-19 and HA-tagged SUR-2 were co-expressed in sur-2mutants. The vulva-less phenotype of the sur-2 mutants was rescued by sur-2::HA, indicating that the SUR-2::HA fusion protein was functional. Nuclear extracts were prepared from mixed-stage animals and protein association was examined by immunoprecipitation (IP) with the anti-GFP antibody, followed by immunoblotting with anti-HA and anti-LET-425. As shown in Fig. 6, LET-425 and the functional SUR-2::HA fusion protein could be co-immunoprecipitated with LET-19::GFP, confirming that these proteins were present in the same complex in vivo.

We also examined whether SUR-2/MED23 and LET-425/MED6 were required for the T-cell division to be asymmetric. We found that sur-2 mutants showed weak defects in the T-cell division asymmetry(Table 1). let-425homozygous mutants obtained from heterozygotes did not show defects in the asymmetry of this division (Table 1) and had very minor developmental defects, probably owing to the maternal contribution, although these mutants are sterile(Kwon and Lee, 2001). However,although RNAi of let-425 causes the embryonic lethal phenotype(Kwon et al., 1999), we found that the escapers of the lethality showed defects in the asymmetry of the T-cell division similar to those of the let-19 and dpy-22mutants. Therefore, SUR-2/MED23 and LET-425/MED6 are also involved in the T-cell division regulated by the Wnt signaling pathway.

When cells divide asymmetrically, the daughter cells are likely to acquire distinct cell fates by transcribing different sets of genes. Mediator complexes are key regulators of transcription(Myers and Kornberg, 2000). We identified let-19 and dpy-22 mutations that affected the asymmetric T-cell division. We showed that let-19 and dpy-22encode proteins similar to MED13 and MED12, respectively. We showed that the LET-425/MED6 and SUR-2/MED23 proteins co-immunoprecipitated with LET-19. Because SUR-2 and LET-425 are also involved in the asymmetric T-cell division,LET-19 and DPY-22 function in the Mediator complex to regulate the asymmetry of the T-cell division. In addition to the Mediator complex, we have previously shown that a chromatin-remodeling complex, SWI/SNF, is involved in this asymmetric division of the T cell(Cui et al., 2004; Sawa et al., 2000). In yeast,both SWI/SNF and the Mediator complex are required for the expression of the HO endonuclease that is transcribed only in mother cells upon asymmetric cell division (Cosma, 2002). Therefore, our results further indicate that the mechanism of asymmetric cell division is conserved between yeast and C. elegans.

Two distinct Mediator complexes have been reported in mammals. The CRSP complex is active for Sp1-dependent transcription, while the larger complex,ARC-L, is transcriptionally inactive(Taatjes et al., 2002). Compared with CRSP, ARC-L has several additional components, including MED12 and MED13, which are homologs of DPY-22 and LET-19, respectively. In yeast,Srb8/MED12 and Srb9/MED13 form a sub-complex and do not always participate in the Mediator complex (Borggrefe et al.,2002; Myers and Kornberg,2000). Similarly, in C. elegans, LET-19/MED13 and DPY-22/MED12 may be present only in the ARC-L-like but not in the CRSP-like complex. Because the let-19 and dpy-22 mutations induce symmetric cell division, similar to lin-17 mutants, activation of the LIN-44/LIN-17 signaling pathway might convert the ARC-L-like complex to the CRSP-like complex, by causing the release of a sub-complex containing LET-19 and DPY-22. Our data suggest that LET-19 and DPY-22 are involved in preventing the expression of TLP-1 in the T.a cell, raising the possibility that the LET-19-DPY-22 subcomplex directly inhibits the expression of tlp-1, a candidate Wnt signal target in the T-cell division. In this case, the ARC-L-like complex may inhibit the expression of tlp-1 in the T.a cell, while the CRSP-like complex may activate transcription of tlp-1in the T.p cell.

In addition to the tlp-1 expression, in the fusion of the Pn.p cells, our results indicate that LET-19 and DPY-22 function in transcriptional repression of the lin-39/HOX gene. In this case, the Wnt signal mediated by bar-1/β-catenin may release LET-19 and DPY-22 from the Mediator complex, resulting in the induction of lin-39expression. By contrast, in the absence of the Wnt signal, LET-19 and DPY-22 may participate in the Mediator to inhibit the expression of lin-39,resulting in cell fusion.

Despite defects in tlp-1 expression in the T.a cell, the neural fate of the T.p cell is abnormal in let-19 and dpy-22mutants, rather than the hypodermal fate of the T.a cell being altered. This puzzling contradiction can be explained if let-19 and dpy-22regulate the transcription of other genes required for neural fates in the T.p cell. Another possibility is that the expression of the tlp-1 gene in the T.a cell may affect the fate of the T.p cell, although interactions between the T.a and T.p cells have not been reported.

In yeast, the Srb8-11 subgroup forms a specific module, which is present in holoenzyme preparations from cells growing exponentially in rich glucose medium, but is absent in stationary-phase cells(Holstege et al., 1998). Genetic analyses indicate that the Srb8-11 module is involved in the negative regulation of a small subset of genes(Carlson, 1997; Holstege et al., 1998). In Drosophila, loss of either the skuld(skd)/MED13 or kohtalo(kto)/MED12 gene has exactly the same effect. It was also reported that the Skd and Kto proteins interact with each other(Janody et al., 2003; Treisman, 2001). In C. elegans, we have shown here that mutations in either let-19 or dpy-22 cause similar defects in T-cell division and fusion of the Pn.p cells. They also share the Dpy and Muv phenotypes. A recent paper reported that the male tail phenotype caused by the pal-1(e2091)mutation was suppressed not only by dpy-22/sop-1 mutations, but also by the reduced expression of let-19(Wang et al., 2004). These observations strongly suggest that MED13 and MED12 function as a unit, which is conserved evolutionally. A remaining question is, what are the roles of Cdk8 and Cyclin C, the other components of the Srb8-11 submodule? Do Cdk8 and Cyclin C also have a function similar to MED13 and MED12? Future studies of these molecules will contribute to our understanding of the roles of the Srb8-11 submodule in the Mediator complex.

In yeast, although disruption of Srb4/MED4 affects the transcription of most genes (93% of 5361 genes examined), that of Srb10/CDK8 affects only a small subset of them (3%) (Holstege et al., 1998). In Drosophila, Skd/MED13 and Kto/MED12 are specifically required for proper photoreceptor differentiation(Treisman, 2001), and Skd is involved in the regulation of segment identity(Boube et al., 2002). In C. elegans, disruption of LET-19 at the embryonic stage affects the expression of a subset of developmentally regulated genes(Wang et al., 2004). We have shown that let-19 and dpy-22 mutants have defects in specific developmental events that are regulated by Wnt signaling. These mutations affect the expression of the tlp-1 gene specifically in the T-cell lineage and that of mab-5 in the V6 cell. These results indicate that the Srb8-11 submodule acts on specific genes in specific developmental contexts.

Many of the strains used in this work were provided by the Caenorhabditis Genetic Center, which is funded by the NIH National Center for Research Resources. We thank Judith Kimble for GFP::POP-1; Mike Herman for tlp-1::GFP; Yuji Kohara for providing cDNAs; Jin Mo Park and Young-Joon Kim for the LET-425/MED6 antibodies; and Meera Sundaram, Eva Davison, Erik Andersen, Wally Wang and members of the Sawa laboratory for comments on the manuscript. This work was supported in part by grants from PRESTO (H.S.) and CREST of the Japan Science and Technology Corporation(H.O.), the Japanese Ministry of Education, Culture, Sports, Science and Technology (H.S. and H.O.).