Biological significance of a universally conserved transcription mediator in metazoan developmental signaling pathways

Transcriptional mediators are proteins associated with the basal transcription machinery, which are required to integrate diverse gene-specific regulatory signals and to recruit basal transcription machinery to specific promoters (Malik and Roeder,2000EF22). The Mediator complexes were initially identified in yeast as distinct intermediary molecules that mediate signal transfer between gene-specific transcriptional activator proteins and the basal transcription machinery (Kim et al.,1994EF17; Koleske and Young,1994EF18; Thompson et al.,1993EF32). The yeast Mediator complex is composed of the Med proteins (Med1, Med2, Med4, Med6, Med8, Med9,Med10 and Med11), Gal11, Rgr1, Sin4, Hrs1, Rox3 and the Srb family of proteins(Srb2, Srb4, Srb5, Srb6 and Srb7). These Mediator components assemble into several functional modules that regulate distinct groups of genes (Lee and Kim, 1998EF20; Myers et al.,1999EF23).

To date, many mediator complexes have been identified in various multicellular species. These complexes include the human Srb/Med-containing co-factor complex (SMCC) (Gu et al.,1999), the negative regulator of activated transcription (NAT) complex (Sun et al.,1998), mouse and human Mediator complexes (Boyer et al.,1999; Jiang et al.,1998), and the Mediator complex in the nematode C. elegans (Kwon et al.,1999).

Although the biological significance of the mediators at the organismic level is largely unknown in higher eukaryotes, some experimental evidence of a physiological role for these Mediator complexes in development has come from studies in the nematode Caenorhabditis elegans. The med-6,med-7 and med-10 mediator genes were initially identified from a genome search for homologs of the yeast mediators (Lee et al.,1997), and we have previously reported that these mediators are required for regulated transcription of tissue- and stage-specific developmental genes, and that loss of function in any of these genes causes embryonic lethality, confirming their essential roles in development (Kwon et al.,1999). Another mediator gene characterized in the nematode was sur-2. sur-2 was originally isolated as a suppressor of a let-60 Ras gain-of-function mutation(Singh and Han, 1995). Loss-of-function mutations in sur-2 alone resulted in pleiotropic,incompletely penetrant phenotypes, including a vulvaless phenotype in hermaphrodites, defects in development of the male tail, gonadal abnormalities and larval lethality. When sur-2 was cloned, it was found to have no homology to any other gene, but later biochemical studies on human mediator complexes revealed that human SUR-2 was a component of the mediator complex. It was also shown that the adenovirus E1A protein binds to SUR-2, activating transcription in vitro (Boyer et al.,1999). A third C. elegans gene, sop-1, which encodes a transcriptional mediator-related protein, was isolated as a suppressor of a pal-1mutation. pal-1 is a gene that encodes the ortholog of the transcription factor caudal (Waring and Kenyon,1991), and its gene activity is required for both early and late embryogenesis. During late embryogenesis,pal-1 is required for specifying the fate of V6 cells. Expression ofpal-1 in V6 neuroblasts activates expression of mab-5, a gene encoding a homeobox protein, which in turn activates expression ofegl-5 and lin-32, thus defining male ray-specific properties(Costa et al., 1988; Ferreira et al., 1999; Hunter et al.,1999; Wrischnik and Kenyon,1997). One of thepal-1 mutations, pal-1 (e2091), is a tissue-and stage-specific mutation in that it does not perturb any early embryonic development but causes defects in the fate specification of V6 cells. Mutations in the sop-1 gene were isolated in a screen for suppressors of pal-1(e2091), and the sop-1 gene turned out to encode a homolog of TRAP230, which is a component of the human mediator-related protein complex (Zhang and Emmons,2000).

One common feature of the two mediator-related proteins, SUR-2 and SOP-1,found in the nematode, is that these proteins do not have homologs in yeast,but are conserved in the various metazoan species studied so far, including humans, indicating that these mediator-related proteins may relay signals from metazoan-specific transcriptional regulators. As MED-6, MED-7, MED-10 and SRB-7 are conserved not only in the metazoa but also in yeast, it is conceivable that these universally conserved mediators may be the point of convergence at which diverse transcriptional signaling mediated by metazoan-specific transcription factors and mediator-related proteins converge, at least in metazoa. Based on these inferences, we became interested in resolving the issue of the relationship between the metazoan-specific mediator-related proteins and the conserved mediator components. We pursued this by studying the roles of a universally conserved mediator med-6,using Caenorhabditis elegans.

In this study, we have isolated and characterized the genetic mutation in the med-6 gene, and have examined the in vivo functions of this mediator gene in the development of C. elegans by establishing biological assay systems involving the Ras and Wnt pathway, in which metazoan-specific mediator-related proteins are known to be involved. Based on our observation that the med-6 mutant animals had a vulval defect,which is due to hypo-induction of the vulval precursor cells (VPCs), we hypothesized that med-6 is involved in transcriptional regulation of genes involved in the Ras pathway. Accordingly we examined the effect ofmed-6 RNAi on the lin-3, let-23, and let-60 genes,which act in the Ras pathway. We also examined the effect of med-6RNAi on male ray development, in which pathway sop-1, a mediator-related gene, is involved. Another interesting finding was that RNAi of med-6 in wild-type background caused more severe phenotypic changes than the putative null mutation of med-6. We examined the possibility that this phenotypic discrepancy was due to maternal rescue by employing a genetic experiment involving the rde-1 mutation.

The C. elegans Bristol strain N2 was used as the standard wild-type strain. The alleles and strains used for cloning med-6were: let-404, BC2590 [dpy-18(e364)/eT1 III; dpy-11(e224)let-404(s119) unc-42(e270) /eT1 V]; let-438,BC2858[dpy-18(e364)/eT1 III; unc-46(e177) let-438(s2114)/eT1 V];let-468, BC3655[dpy-18(e364)/eT1 III; unc-46(e177)let-468(s1533)/eT1 V]; let-339, BC2245[dpy-18(e364)/eT1 III;unc-46(e177) let-339(s1444)/eT1 V]; let-343,BC1919[dpy-18(e364)/eT1 III; unc-46(e177) let-343(s1025)/eT1 V];let-425, BC3060[dpy-18(e364)/eT1 III; unc-46(e177)let-425(s385)/eT1 V]; and let-332, BC1480[dpy-18(e364)/eT1 III; unc-46(e177) let-332(s234)/eT1 V] (Rosenbluth et al.,1988). The alleles used for studying the mediator function in vulval development and their typical phenotypes were: let-60(n1700) IV (multivulva 96%), lin-3(e1417)IV (vulvaless, 10-30% wild type), let-23(sa62) II; him-5(e1490) V(let-23 semi-dominant, over 95% multivulva), and dpy-5(e61) I; syIs1(lin-3 over expression, over 95% multivulva). The deficiency strain we used for examination of hemizygotes of the let-425 mutation was of the genotype dpy-18(e364)/eT1 III; unc-46(e177) sDf20/eT1 V. The strains were obtained from the Caenorhabditis Genetics Center (CGC) and P. W. Sternberg (Caltech, Pasadena, CA). For male tail experiments, pal-1(e2091)III; him-5(e1490) V and pal-1(e2091) III; him-5(e1490) V; sop-1(bx92)X were used (gifts from Scott W. Emmons, Albert Einstein College of Medicine, New York, NY). The strain used for examination of pal-1transcription was BW1851 whose genotype is pal-1::GFP pRF4 integrated(a gift from L. G. Edgar and W. B. Wood, University of Colorado, Boulder, CO)For rde-1 experiments, unc-32(e189) III; rde-1(ne219) V was used (Tabara et al., 1999). The culture of C. elegans has been previously described (Brenner,1974).

The physical location of med-6 was determined to be in the YAC Y57E12 (Y.-J. Kim, personal communication), on chromosome V. However, no positive cosmids were detected by Southern hybridization using med-6 cDNA as a probe. We reasoned that this was because the med-6 gene resided in a cosmid gap that is located between the cosmid K11C4, encoding unc-68,and VC5, encoding odr-2. As the region was saturated with lethal mutations, and sDf20 and sDf30 delete both unc-68and odr-2 while nDf32 does not, we searched for lethal mutations mapped between the break points of the deficiencies. According to the genetic map available (http://elegans.swmed.edu), let-404, let-438,let-468, let-442, let-339, let-343, let-425, let-346 and let-332reside in this region. We examined the phenotypes of animals containing each mutation by DIC microscopy, and decided that let-425 andlet-332 could be candidates for the med-6 mutation. Alet-425 mutation had been isolated as a lethal mutation with a sterile phenotype, and a let-332 mutation isolated with embryonic lethality (Johnsen and Baillie,1991). We amplified the genomic region containing the entire coding sequence of the med-6gene from either heterozygotes of let-332 or animals homozygous forlet-425. The primers used for amplifying the genomic region from single larvae were CM6F (5′-GGACTAGTATGGGACCTCCAGCAGCTGCAC-3′) and CM6R (5′-TAATTTAATGTTCATTTTTAGTG-3′). The amplified fragments were cloned into pGEM-T easy vector (Promega). We determined the sequences from six independent colonies with T7 sequenase 2.0 DNA sequencing kits (Amersham). To confirm the identity of the mutation, we directly determined the sequence of the PCR-amplified fragments from eight let-425 homozygous animals,eight heterozygotes and eight N2 animals. In an effort to rescue the mutation of let-425 with the wild-type med-6 gene, we amplified a full-length med-6 gene containing the upstream 2 kb region and the entire 3′UTR as well as the coding region. The PCR primers used for this purpose were CM6PF (5′-CGTCGACCACATCCTTCGCCGGAAGC-3′) and CM6PR(5′-CTCTAGACTCATAAACCACAAAGAGGAGC-3′). We microinjected the linear PCR product into wild-type animals with genomic DNA digested withEcoRI, which cuts the coding region of med-6. Heterozygous males of the genotype dpy-18/eT1; unc-46 let-425/eT1 were mated with transgenic animals containing the med-6 transgene, and F₁hermaphrodites that segregated Dpy animals in the next generation were individually selected. Among the siblings of the Dpy animals, rolling UNC-46 animals, whose genotype should be unc-46 let-425; [Ex pRF4 + med-6 +fragmented genomic DNA] were examined for their phenotypes. These UNC-46 animals were normal in vulval development and were able to lay eggs, a phenomenon never seen in the let-425 homozygous animals. This result indicates that the linear PCR fragment containing the med-6 gene complemented the let-425 mutation.

The let-425 mutation is maintained in the heterozygote form in the genotype dpy-18(e364)/eT1 III; unc-46(e177) let-425(s385)/eT1 V. In order to measure the lethality and sterility of the let-425homozygotes, we removed the eT1 balancer by mating dpy-18 homozygote hermaphrodites with N2 males to produce males of the genotype dpy-18/+, and the dpy-18/+ heterozygous males were mated with thelet-425 heterozygotes. In the next generation, Dpy virgins of the genotype dpy-18(e364)/dpy-18(e364) III; unc-46(e177) let-425(s385)/+V were selected, and the phenotypes of their progeny were examined. In order to determine whether the existing allele of let-425 is null, we mated males of the genotype dpy-18(e364)/eT1 III; unc-46(e177)let-425(s385)/eT1 V with hermaphrodites of the genotypedpy-18(e364)/eT1 III; unc-46(e177) sDf20/eT1 V. The number of non-Dpy male progeny and that of Dpy Unc male and hermaphrodite progeny were compared,and the phenotypes of the hemizygotes were examined. The expected ratio of the non-Dpy males to Dpy Unc animals (hermaphrodites and males) is 4:2, if there had been no embryonic lethality (Rosenbluth and Baillie,1981). The observation was that the ratio of the non-Dpy males to Dpy Unc animals was close to 2.45:1(n=913). Among the hemizygous hermaphrodites containing a single copy of thelet-425 mutation, 78% arrested at the L4 stage and the remaining 22%reach adulthood. This number is comparable with that of the homozygouslet-425 mutants. These results indicate that the let-425mutation is a severe reduction-of-function mutation, if not a complete null.

The med-6 dsRNA used in this study was identical to the dsRNA used in the previous study (Kwon et al.,1999). The size of the RNA used in RNAi was about 800 nucleotides, and contained the fullmed-6-coding region. The concentration of the med-6 RNA used in the experiments was 100 μg/ml. In order to observe the effect ofmed-6 RNAi on vulval development in various backgrounds, vulval invagination at the L4 stage of F₁ progeny laid at 6-24 hours after microinjection was observed. We confirmed the effectiveness of the RNAi by observing that the embryos laid after that time were 100% embryonic lethal. For the RNAi experiments in the rde-1 background, med-6dsRNA was injected into rde-1 mutant animals, and individual animals were mated with three N2 males. After each 12 hour period, we moved the P0 individuals to new plates and provided new males. Individual F₁progeny were transferred to new plates and were examined for their phenotypes after four days of culture. The phenotype `small brood size' was used to describe F₁ animals with fewer than 20 F₂ progeny. Thus,animals classified as normal had fewer progeny than wild-type animals in many cases. Embryonic lethality was calculated by subtracting the number of viable animals from the total number of eggs. In the rde-1 background experiments, we attempted to see if microinjection of dsRNA into the gonad,body cavity or intestine gave different ranges of phenotypes, but we did not see any correlation between the location of injection and severity of the resulting phenotypes. To characterize the role of med-6 in ray development, we likewise microinjected med-6 dsRNA into animals of various mutant backgrounds. When using the rde-1 mutant background,we mated the injected animals with N2 males and examined the phenotypes of the male progeny. When using pal-1(e2091) III; him-5(e1490) V andpal-1(e2091) III; him-5(e1490) V; sop-1(bx92) X backgrounds, we examined the males from the mothers directly. To examine the change inpal-1 transcription by med-6 RNAi, we microinjectedmed-6 dsRNA into the pal-1::GFP integration line, and observed F₁ embryos inside the PO animals. A Zeiss Axioplan2 microscope was used for Nomarski images, and a Zeiss AxioCam digital camera and MC200 camera (Carl Zeiss) were used for taking photographs.

In order to identify genetic mutations in the med-6 gene, we compared the physical map to the genetic map(Fig. 1). The med-6gene was physically mapped to the YAC, Y57E12, by YAC grid hybridization(Y.-J. Kim, personal communication), and later by sequencing data from the genome project (C. elegans Sequencing Consortium,1998). The med-6 gene was located in the cosmid gap between unc-68 and odr-2, and we reasoned that med-6 should be encoded by a lethal mutation that mapped between the break points of the deficiencies, sDf20 andnDf32. Among the lethal mutations that have not been cloned from this region, we found that there was a single base substitution in themed-6 gene from the let-425 mutant animals(Fig. 1; Materials and Methods). The mutation was a C to T transition at nucleotide 292, which we expect causes a truncated LET-425 protein with amino acids missing from the 91st residue. We confirmed the mutation by single worm PCR and sequencing of several homozygous animals containing two copies of the let-425mutation. We were able to rescue the let-425 mutation with a PCR fragment containing the med-6 gene (see Materials and Methods). In addition, the fact that the phenotypes of the let-425 mutation showed striking similarities to the phenotypes caused by med-6 RNAi also supported our sequencing results. For example, the let-425 mutant animals displayed hypo-induction of the vulval precursor cells, similar to animals affected by med-6 RNAi(Fig. 2B,C) as well as larval arrest, a male tail defect and adult sterility. Thus, we identifiedlet-425 as the med-6 gene.

We first determined whether the let-425 mutation was null. As hemizygous animals containing a single copy of let-425 mutation displayed comparable phenotypes with those of the let-425 homozygotes(see Materials and Methods), the existing let-425 mutation is a severe reduction-of-function mutation. The phenotypes associated with thelet-425 mutation were then examined in order to define the biological significance of the med-6 gene in development. We examined thelet-425 homozygous animals laid by heterozygous mothers because homozygous animals are completely infertile. Two phenotypes were apparent in the let-425 homozygous progeny; late larval arrest and adult sterility, but they exhibited no embryonic lethality. In order to check if there was any embryonic lethality associated with the mutation, we examined the progeny from heterozygous mothers with the genotype dpy-18/dpy-18;unc-46 let-425/+. Among the 2145 progeny laid from the heterozygous mothers, we found that 25.7% were let-425 homozygous animals. This percentage is comparable with the expected number of homozygouslet-425 progeny if there had not been any embryonic lethality associated with the let-425 mutation. In another experiment, we let the animals of the genotype dpy-18/dpy-18; unc-46 let-425/+ lay eggs and then transferred all the progeny individually to single plates and allowed them to grow. Among the 194 progeny examined, 25.8% were Dpy Unc larvae or sterile adults, while 47.9% were heterozygotes and 26.3% were dpy-18homozygotes, confirming that there was no embryonic lethality and that thelet-425 mutation caused 100% larval arrest or adult sterility.

Next, we examined the phenotypes of the larvae and adults of thelet-425 homozygous animals in more detail. Among the 132 animals observed under Nomarski optics, 72% were arrested in the L4 larval stage and 28% were adults with 100% sterility. The sterile adults were defective both in oogenesis and spermatogenesis: when the male progeny from RNAi-affectedhim-5 hermaphrodites were mated with unc-101 hermaphrodites,they could not produce any cross progeny (n=20). Additionally, in a reciprocal mating, in which RNA-affected hermaphrodites were mated with N2 males, no progeny were produced (data not shown). Because male tail defects and vulval defects, which may interfere with mating ability by themselves, are not 100% penetrant (see the following sections), enough numbers of matings should have resulted in production of cross progeny if there was no defect in spermatogenesis or oogenesis, respectively. Interestingly, the sterile adult animals showed an additional phenotype of protruding vulvae(Fig. 2F). Accordingly, we examined the let-425 homozygous animals at the L3 molt stage, which is when the vulval precursor cells usually divide to form the vulval tissue(Sternberg and Horvitz, 1986;Sternberg and Horvitz, 1989),in order to explain the cellular defects associated with this phenotype. Out of 15 L3 molt animals, eight had less than wild-type vulval induction, in which we commonly observed induction of P6.p but not P5.p and P7.p (for example, see Fig. 2B). RNAi-affected animals displayed an identical hypo-induction phenotype(Fig. 2C). It is known that hypo-induced VPCs later form protruding vulval tissue, and in our study, we found that the animals confirmed to have hypo-induced VPCs later either arrested as L4 larvae or grew to adulthood with protruding vulvae (data not shown).

A possible explanation for the hypo-induction of VPCs in thelet-425 homozygous animals is that the med-6 mutation causes transcriptional downregulation of genes involved in vulval induction. To test our hypothesis, we examined the effect of reduction of med-6 gene function on lin-3, let-23 and let-60, which encode an epidermal growth factor (EGF), the EGF receptor and Ras, respectively (seeTable 1;Fig. 3; Aroian et al.,1990; Han and Sternberg,1990; Hill and Sternberg,1992). Either gain-of-function or reduction-of-function alleles of these genes were used. Gain-of-function mutations in any one of these genes result in a multivulva phenotype in which more than three VPCs are induced to make additional vulval tissue in addition to a functional vulva. On the other hand, reduction-of-function mutations in these genes cause hypo-induction of VPCs. We reasoned that reduction inmed-6 activity would result in reduction of the transcription levels of these genes, which would suppress the multivulva phenotype of the gain-of-function mutations and enhance the hypo-induction phenotype of reduction-of-function mutations. The alleles of lin-3 that we used were syIs1, an allele generated by integration of multicopies of the wild-type lin-3 gene, and lin-3 (e1417), a reduction-of-function mutation. Reduction of med-6 activity by RNAi did not suppress the multivulva phenotype resulting from overexpression oflin-3 driven by multicopies of this gene(Fig. 3C,D). On the contrary,the hypo-induction phenotype of lin-3 (e1417) was significantly enhanced (Fig. 3A,B). Furthermore, gain-of-function mutations in let-23 or let-60were partially suppressed to the wild-type form by reduction of med-6activity (Fig. 3E-H). From these results, we concluded that reduction of med-6 activity could cause the genes involved in the Ras signaling pathway to be transcribed at lower levels.

A recent study has shown that a tissue-specific mediator is involved in regulating transcription of gene(s) involved in ray development (Zhang and Emmons, 2000). Transcription of pal-1, a gene encoding a transcription factor required for proper ray development, is regulated by two separate circuits, one of which is normally silent, owing to sop-1 activity, which represses the transcription of pal-1 from the 5′ upstream regulatory element. This transcriptional regulation is mediated by the Wnt pathway. We wished to examine whether med-6 was also involved in the regulation of genes involved in ray development. We first examined the male tails of a fewlet-425 homozygous animals and found that the male structures were defective (data not shown). We then used RNAi to examine the effect ofmed-6 on male ray development. Because we reasoned that genes required for proper ray development should be functional at the zygotic level,we examined the male tail phenotypes associated with males produced by the mating of N2 males with rde-1 homozygous hermaphrodites that had been microinjected with med-6 dsRNA (see Materials and Methods; forrde-1 experiments, see the next Result sections). We found that 17 out of 45 males observed under Nomarski optics displayed abnormal ray defects(Fig. 4B). A typical phenotype was that the V6-derived rays, 2-6, were partially or completely missing. We observed essentially identical phenotypes when we examined the tail phenotypes of the males produced from N2 hermaphrodites that had first been mated with N2 males, and then subjected to med-6 RNAi (data not shown). It is known that the development of the V6-derived ray structure is dependent on the expression of the caudal homolog pal-1 in the V6 progeny. Because we did not see significant defects in ray number 1, nor in 7-9 in our experimental conditions, it is probable that med-6 may be involved in ray development at the level of pal-1 transcription, although we cannot rule out the possibility that genes downstream of pal-1 were affected. We next examined the ray phenotypes of animals of the genotypesop-1; pal-1 in order to determine whether med-6 is involved in the transcriptional regulation mediated by the Wnt pathway in ray development. In wild-type animals, pal-1 is transcriptionally activated through an intronic regulatory element and the transcriptional activation by the Wnt pathway is repressed by the activity of sop-1. As in the pal-1; sop-1 mutants pal-1 is transcriptionally activated by the Wnt pathway (Zhang and Emmons,2000), it is possible to examine if med-6 is involved in the Wnt-mediated transcription ofpal-1 by using these mutants. If med-6 is involved in the transcriptional activation of pal-1 both by the original intronic regulatory element and by the Wnt pathway, then med-6 RNAi should reduce the development of the V6-derived rays in the double mutant animals. We observed ray defects in the pal-1; sop-1 double mutant animals that were almost identical to those caused by med-6 RNAi in the wild-type background (Fig. 4C-F),indicating that Wnt signal-mediated transcription may also require MED-6 as its transcriptional mediator.

Our previous RNAi result showed that RNAi using a high concentration ofmed-6 double-stranded RNA in the wild-type background caused almost 100% embryonic lethality and that at a lower concentration of RNA, the animals showed both embryonic lethality and adult sterility (Kwon et al.,1999). When we examined the phenotype of the let-425 mutation, we found no evidence of embryonic lethality (see the previous Results section). One possible explanation is that maternal med-6 activity required for embryogenesis may still be provided by the heterozygous mother. We examined this possibility by using therde-1 mutation (Tabara et al.,1999). To test our hypothesis,we designed a genetic experiment in which we could selectively reduce the zygotic med-6 gene function. Fig. 5 shows the mating schemes that we employed. As rde-1homozygous animals are resistant to RNAi both maternally and zygotically, and the rde-1 phenotype is fully recessive (Tabara et al.,1999), we performed RNAi ofmed-6 at a high concentration in the rde-1 homozygous animals and mated them with N2 males. The progeny from this mating should have a zygotic RNAi effect, but not a maternal RNAi effect, because rde-1is now in the heterozygote form. This mating allowed for selective reduction of the zygotic med-6 gene function(Table 2). Unlike wild-type animals, rde-1 mutant animals were indeed 100% resistant to RNAi ofmed-6: we did not see any effect of med-6 RNAi in therde-1 mutant animals. However, after mating the RNA-injectedrde-1 hermaphrodites with N2 males, the progeny showed various degrees of severity in their phenotypes. About 10% of the eggs counted(n=1693) were arrested at the embryonic stage, 6% grew up to adulthood and became sterile, and 28% of the eggs grew to adulthood with no obvious phenotype. About 7% of the adults had limited brood sizes. In addition, males produced in this experiment displayed ray defects and hermaphrodites showed defective vulval induction (data not shown). In control experiments, wild-type animals injected with the dsRNA, whether mated or not,produced nonviable embryos. In summary, we observed phenotypes inrde-1 mutant RNAi that were comparable with, although less severe than, let-425 homozygous mutant phenotypes. From these data, we propose that the maternal med-6 is required for early embryogenesis,and that the zygotic function of med-6 is mostly required for postembryonic development such as vulval development and ray development.

We have previously examined transcription mediator functions by using RNAi and showed that med-6 was required for regulated gene transcription but not for constitutive gene transcription (Kwon et al.,1999). We also showed that mediators are essential for embryonic development as RNAi of any one of the mediators caused embryonic lethality. The next goal was to characterize the biological functions of these mediators by identifying mutations in their genes. Accordingly, we identified let-425 as the med-6 gene by sequencing the genomic region of med-6 in let-425 mutant animals. We were unable to rescue the let-425 mutation by introducing high copies of a genomic fragment that contained the full-lengthmed-6 gene. A possible reason for not being able to rescue the phenotype would be that the transgene, introduced by microinjection, may have interfered with endogenous mutant RNA, causing co-suppression (for example,Dernburg et al., 2000). We therefore tried low-copy transformation by co-injecting the PCR fragment and genomic DNA digested with a restriction enzyme. Using this method, we were able to rescue the let-425 mutation, apparently by overcoming the silencing of multicopy genes (Kelly et al.,1997).

Mediators were shown to be required for regulated gene transcription in yeast and in the nematode (Kwon et al.,1999; Lee and Kim,1998). In yeast,med-6 mediates transcription of about 10% of the whole genome(Holstege et al., 1998). It is therefore conceivable that med-6 also is involved in the regulated transcription of many genes in the nematode C. elegans. In order to establish biological assay systems to study the role of transcription mediators in development, we examined genes acting in two signaling pathways of the nematode: the Ras and the Wnt pathways. The Ras signaling pathway is one major signaling pathway in hermaphrodite vulval development, and the Wnt pathway is involved in ray development. Because vulval and ray development should occur at the appropriate time in development at the correct place, the genes involved in those pathways should be regulated temporally and spatially. Therefore, it is conceivable that mediators may be involved in regulated transcription of the genes in these pathways. In this report, we showed that nematode med-6 is involved in the regulation of genes associated with the Ras pathway and the Wnt pathway.

We first examined vulval development because we observed abnormal vulval morphogenesis in the let-425 homozygous mutant animals. From the animals produced at early time points after injection of med-6 dsRNA,we were able to show that med-6 is indeed involved in the Ras pathway: reduction of med-6 activity lowered the gene activity of both the gain-of-function mutant genes and a reduction-of-function mutant gene. The fact that embryos laid at later time points after RNAi were arrested at the embryonic stage indicates that the animals produced earlier might have had residual med-6 activity, sufficient to enable the animals to survive up to adulthood. We were unable to obtain direct evidence showing that the transcription levels of the genes were reduced. However, from the phenotypes of the RNAi-affected animals, we could see the effects caused by the reduction of med-6 activity. It was not surprising that reduction of med-6 activity did not suppress the multivulva phenotype caused by multicopies of lin-3. We interpret this as follows: the level oflin-3 activity is known to be crucial in determining how many VPCs are induced (Katz et al.,1995), and it is probable that reduction of med-6 activity by RNAi was not effective enough to reduce the transcription level of lin-3 to a point where fewer VPCs were induced. However, it is obvious that lin-3 is directly or indirectly activated by med-6, because the reduction-of-function mutation of lin-3 was enhanced by the med-6 RNAi. We do not know at this point whether transcription of all three of the genes we examined, lin-3, let-23, and let-60, is directly regulated by med-6 activity. We cannot rule out the possibility that other genes downstream of these are affected by the med-6 RNAi, resulting in the effects we saw in the experiments. It would be necessary to establish an in vitro transcription assay system to test this directly.

It was recently reported that components of the nucleosome remodeling and histone deacetylase (NURD) complex in C. elegans antagonize Ras-induced vulval development by deacetylating specific target genes to repress vulval development (Solari and Ahringer,2000). Now that it is evident that both the mediators and the chromatin remodeling complex are involved in the strict temporal and spatial regulation of sets of genes for proper development, the C. elegans vulval development system can serve as a model in which one can study the biological significance of the mediators and chromatin remodeling complexes.

We examined ray development because it was reported that another mediator protein was involved in repression of the Wnt pathway in ray precursor cells(Zhang and Emmons, 2000), and we subsequently found that med-6 was also involved in regulation of the gene(s) involved in ray development. Reduction of med-6 activity caused V6 progeny to undergo abnormal development, in that ray structures produced by these cells were missing or reduced in number. We also demonstrated that med-6 is required for proper regulation of the genes in the Wnt signaling pathway by examining the RNAi effect in thepal-1; sop-1 mutant animals. The data presented in this report suggest that med-6 is involved in the development of V6-derived ray structures by regulation of pal-1 transcription. Consistent with this, we found that med-6 RNAi lowered pal-1 transcription during embryogenesis (data not shown). However, it is formally possible thatmed-6 is involved in regulated transcription of genes downstream ofpal-1, such as mab-5.

The fact that SUR-2 and SOP-1 proteins do not have homologs in yeast, but are conserved only in metazoa, while MED-6 is conserved from yeast to humans,raises the possibility that MED-6 is a more general mediator of transcriptional regulatory signaling. Consistent with this, SUR-2 has been shown to have roles in the Ras pathway, but not in ray development, while SOP-1 has been shown to be involved in the Wnt pathway, but not in the Ras pathway. On the contrary, med-6 has been shown to be involved both in the Ras pathway and in the Wnt pathway. Our model for the action of the mediator-related proteins in the nematode is shown inFig. 6. We propose that the metazoan-specific mediator-related proteins interact with metazoan-specific transcription regulators to generate signals that converge at a more conserved mediator complex containing MED-6, which in turn transmits the signal(s) to the basal transcriptional machinery, leading to developmental specific gene expression.

The finding that the let-425 mutant phenotypes were less severe than the effects of RNAi prompted us to examine the differences between the maternal and zygotic gene functions of med-6 in development. If mutations are introduced in essential genes that are required both maternally and zygotically, the phenotype of homozygous mutant animals from heterozygous mothers would be masked because of maternal rescue. We were able to show by using the rde-1 mutant background that one reason why the phenotypes of the let-425 mutation are less severe than the effects of RNAi is due to the maternal contribution of wild-type med-6 gene activity. As RNAi in the wild-type background does not distinguish between the reduction in maternal or zygotic functions of genes that are expressed both maternally and zygotically, we employed the rde-1 mutant background. By using therde-1 mutant background, we were able to examine the phenotypes caused by the reduction in zygotic gene expression only, leaving the maternally expressed genes intact. From the data we presented above, we could infer that the maternal contribution of med-6 is required for normal embryogenesis and that its zygotic gene activity is mostly required for normal larval development and fertility, as well as normal ray development.

Additionally, we were able to confirm the limitations of RNAi, one of the main ones being that the effect of RNAi is not as severe as the null mutation of the gene, particularly for genes acting late in development. Therde-1 RNAi experiment we described above is an example of this. Sterility and larval arrest are late phenotypes compared with embryonic lethality, and quite a few progeny from the RNAi-affected rde-1animals were able to escape the RNAi effect, while the effect of thelet-425 genetic mutation was persistent through adulthood.

In summary, we have shown in this report that MED-6 is encoded bylet-425, that the maternal and zygotic let-425 have distinct roles in development. Maternal let-425 has been shown to be required for early embryogenesis and zygotic let-425 has been shown to be required for late development, including hermaphrodite vulval development and male ray development. As for the action mechanism of the mediators, we suggested that MED-6, a universally conserved transcription mediator, is required for actions of metazoan-specific mediator-related proteins, such as SOP-1, for appropriate transcriptional regulation of development-specific genes. The next step will be to examine whether and how med-6 is involved in regulated transcription of other genes in development. One way to pursue this issue on a large scale would be to perform a microarray experiment, as has been done in the yeast system.

We thank Drs P. W. Sternberg, S. Emmons, L. G. Edgar and W. B. Wood, and the Caenorhabditis Genetics Center (MN, USA) for providing the nematode strains. We also thank to M. Choi, Dr J. Park and Dr Y.-J. Kim for sharing unpublished data, and for their critical reading and advice on this work. We especially thank an anonymous reviewer for constructive suggestions such as low-copy transformation for the rescue experiment. This work was supported by the NON DIRECTED RESEARCH FUND, Korea Research Foundation, 1998 to J. H. L.