ABSTRACT
We describe the properties of a new gene, sop-3, that is required for the regulated expression of a C. elegans Hox gene, egl-5, in a postembryonic neuroectodermal cell lineage. Regulated expression of egl-5 in this cell lineage is necessary for development of the sensory rays of the male tail. sop-3 encodes a predicted novel protein of 1475 amino acids without clear homologs in other organisms. However, the sequence contains motifs consisting of homopolymeric runs of amino acids found in several other transcriptional regulators, some of which also act in Hox gene regulatory pathways. The genetic properties of sop-3 are very similar to those of sop-1, which encodes a component of the transcriptional Mediator complex, and mutations in the two genes are synthetic lethal. This suggests that SOP-3 may act at the level of the Mediator complex in regulating transcription initiation. In a sop-3 loss-of-function background, egl-5 is expressed ectopically in lineage branches that normally do not express this gene. Such expression is dependent on the Hox gene mab-5, as it is in branches where egl-5 is normally expressed. Ectopic egl-5 expression is also dependent on the Wnt pathway. Thus, sop-3 contributes to the combinatorial control of egl-5 by blocking egl-5 activation by MAB-5 and the Wnt pathway in inappropriate lineage branches.
INTRODUCTION
Cell fate specification during multicellular development occurs by a progressive series of changes in cell transcriptional states. Through such programs of gene transcriptional regulation, differentiated cells of defined types are generated at specific sites in the body. In the transcriptional switching events that constitute the steps of these developmental programs, gene promoters integrate multiple developmental signals in responding to conditions that may occur in unique combinations in individual cells. What the number and nature of these combinatorial inputs is, and how their combined actions determine transcriptional output, are major questions in developmental biology.
Hox transcription factors constitute one class of transcriptional inputs whose role appears to be to convey positional cell identity (Gellon and McGinnis, 1998). Hox proteins interact with target promoters in combination with the co-factors extradenticle and homothorax, which can function to bring action of the Hox transcription factor under the control of extracellular signals (Mann and Abu-Shaar, 1996; Reickhof et al., 1997). Several additional factors have been identified in Drosophila that contribute to the regulation of Hox target promoters in ways that are less well understood: lines (lin), which encodes a novel protein, is required for Hox gene Abdominal-B (Abd-B) to specify the posterior spiracles and the eighth abdominal denticle belt (Castelli-Gair, 1998; Hatini et al., 2000); teashirt (tsh), which encodes a zinc-finger protein, modulates the function of Hox gene Sex combs reduced (Scr) in the establishment of the identities of the prothoracic and labial segments (Fasano et al., 1991; de Zulueta et al., 1994); and cap-n-collar B (cncB), which encodes a bZIP protein, plays a similar role in specifying the identities of two Deformed (Dfd) expression domains (Mohler et al., 1995; McGinnis et al., 1998). In vertebrates, the retinoic acid receptor heterodimer, RAR/RXR, is a co-regulator of Hox target promoters (Marshall et al., 1996).
Regulatory transcription factors such as Hox proteins and other combinatorial transcription factors exert their effects by influencing one or more of the steps of an assembly pathway that must be traversed before an actively transcribing transcription complex forms on a promoter. These steps include initiation of chromatin opening, chromatin remodeling and histone acetylation, recruitment of holoenzyme to the promoter, and release of the polymerase from a potentially inactive preinitiation complex (Struhl, 1999; Kornberg and Lorch, 1999; Bjorklund et al., 1999). One function of this multistep assembly process is no doubt to ensure fidelity in bringing transcription under the combined control of multiple developmental signals (Cosma et al., 1999).
Postembryonic development of the rays of the Caenorhabditis elegans adult male tail (Fig. 1A) provides an opportunity for studying these issues in the context of a particular developmental transcriptional program (Emmons, 1999). Rays, sensory organs used in mating, develop from postembryonic cell lineages generated by three bilateral pairs of neuroectodermal blast cells (Fig. 2A). Each ray is clonally derived through a stereotyped cell sublineage from a ray precursor cell, and each is grossly similar, constituted of two sensory neurons and a support cell. Yet each ray also has unique characteristics, such as whether or not one of its sensory neurons expresses the neurotransmitters dopamine or serotonin, and the position where it forms in the epidermis. These ray differences are a result in part of a regulated pattern of expression of two Hox genes, mab-5 and egl-5 (Fig. 2A). Mutations that affect expression of these genes in the ray lineages can be identified by their effects on ray morphology. In particular, changes in the levels of MAB-5 and EGL-5 result in homeotic transformations of ray morphological identities that often result in ray fusions (Chow and Emmons, 1994).
MAB-5 and EGL-5 are each expressed in specific subsets of the rays and are required for expression of the unique characteristics of those rays (Chow and Emmons, 1994; Salser and Kenyon, 1996; Ferreira et al., 1999). In the transcriptional program that leads to this expression pattern, mab-5 activates egl-5 expression, but it does so only in a subset of the lineage branches in which MAB-5 is present (Ferreira et al., 1999). Furthermore, MAB-5 is present in the lineage before egl-5 is activated, and is present in the hermaphrodite where egl-5 is not expressed. Thus, egl-5 is regulated by multiple factors so that it is expressed in a cell lineage that contains MAB-5 protein, but only in certain branches of this lineage, only at a specific time and only in one sex. We wish to determine the identities of the additional factors that bring about this expression pattern and resolve how they contribute to regulation of the egl-5 promoter.
Here, we describe a new gene, sop-3, that contributes to the regulated pattern of egl-5 expression. sop-3 was identified as a genetic suppressor influencing development of the rays. We previously described another gene, sop-1, identified in the same suppressor screen, which encodes a component of the transcriptional Mediator complex (Zhang and Emmons, 2000). Mutations in both sop-1 and sop-3 suppress a mutation in a cis regulatory element of the C. elegans caudal homolog, pal-1. This mutation prevents expression of pal-1 in the progenitor cell of rays 2-6 and thus results in absence of these rays (Hunter et al., 1999; Zhang and Emmons, 2000). In both sop-1 and sop-3 loss-of-function backgrounds, the crippled pal-1 gene is expressed under the influence of the Wnt pathway. In addition to its effect on pal-1 expression, loss of sop-3 function also results in mis-regulation of egl-5 in subsequent steps of the cell lineage. In particular, egl-5 is expressed in inappropriate lineage branches and this expression is stimulated by the Wnt pathway. Such inappropriate expression remains under the control of mab-5. Thus, SOP-3 appears to be one of the combinatorial inputs that together with mab-5 determines the regulated pattern of egl-5 expression. The similarity in genetic properties of sop-3 and sop-1 suggests that sop-3 acts at the level of the assembly or activation of a transcription complex at the egl-5 promoter.
MATERIALS AND METHODS
Strains
The maintenance of nematode strains, mutagenesis and genetic analysis were handled according to standard procedures (Brenner, 1974). Nematodes were grown at 20°C unless otherwise noted. Most strains carried the him-5(e1490) mutation, which gives a high frequency of males in selfing populations. The following alleles were used in this work. LGI: dpy-5(e61), lin-17(n677) and sur-2(ku9). LGIII: pal-1(e2091), pal-1(ct224), mab-5(e1239), egl-5(n945) and egl-5(u202). LGIV: egl-20(n585). LGV: him-5(e1490). LGX: bar-1(ga80), sop-1(bx92) and nIs118(cat-2::gfp, lin-15) (kindly provided by Hillel Schwartz). The linkage group of bxIs13(egl-5::gfp, lin-15) and bxIs14(pkd-2::gfp, pha-1) is undetermined. pal-1(ct224) is a putative null allele that deletes part of exon 1 and the remainder of the pal-1 gene (Hunter et al., 1999).
Rearrangement: sDp3, a duplication that covers the left portion of chromosome III, including pal-1. sDp3 and egl-20 strains were grown at 25°C.
Isolation, characterization and mapping of bx96
bx96 was isolated as a suppressor of the V6 ray loss phenotype of pal-1(e2091). F2 or F3 males of ethyl methanesulfonate-treated pal-1(e2091) hermaphrodites were screened for the presence of V6 rays. To recover the mutation that restored the V6 rays, sibling hermaphrodites were cloned. 19 mutations that defined more than 10 genes were obtained from about 4000 genomes screened. Four of these 19 lie in the gene sop-1 (Zhang and Emmons, 2000); the remaining are being characterized.bx96 was mapped on linkage group I. Three-factor mapping indicated that sop-3(bx96) is very close to lin-17: 51 out of 52 Dpy nonLin recombinants from dpy-5 + lin-17/ + sop-3(bx96) + cross carried sop-3(bx96).sop-3(bx96) has a maternal effect: 95% of Dpy males from + + / sop-3 dpy-5; pal-1(e2091) hermaphrodites have Pal phenotype (n=200), although most Dpy males should be homozygous for sop-3(bx96).
The cloning of sop-3
sop-3 was mapped between lin-17 and pop-1. YAC Y71F9B contains the sequence between cosmid F32B5 (containing the lin-17 gene) and W03D8 (containing pop-1). Co-injection of two overlapping PCR fragments (fragments 1 and 2, Fig. 3), but not the fragments individually, from this YAC rescued sop-3(bx96). Fragment 1 corresponds to the sequence from 54101 to 71198 of Y71F9B; fragment 2 corresponds to the sequence from 65797 to 81865 of Y71F9B. The DNAs were injected into sop-3(bx96); pal-1(e2091) hermaphrodites at a concentration of 20-50 ng/μl. pRF4, which carries the dominant rol-6(su1006) marker, was co-injected at a concentration of 100-200 ng/μl. F2 Rol males were scored for the presence of V6 rays. Dissecting these two fragments further, we found that one 16 kb PCR fragment, corresponding to the sequence from 58679 to 74891 of Y71F9B, contained sop-3 function.
A cDNA expressed sequence tag from Dr Y. Kohara, yk533h12, was sequenced. RT-PCR was performed and the products sequenced to get the reminder of sop-3 mRNA. sop-3 is SL-1 trans-spliced, suggesting that the cDNA is full length. The primers for the RT-PCR were SL1, 5′GGTTTAATTACCCAAGTTTGAG3′; and Y71F9B, 5′(69698) CCAGTTTGTGTGAATACCGGC 3′(69718).
sop-3 RNAi
RNAi experiments were performed as described by Fire et al. (Fire et al., 1998). The RNA was synthesized using MEGAscript T3 and T7 kit (Ambion). 200 ng/μl dsRNA was injected. The eggs from injected animals are collected from 4 to 48 hours postinjection.
The efficiency of dsRNA in causing the duplication of V6-rays and suppression of pal-1(e2091) was strongly dependent on which part of the gene was used as template for synthesizing RNA. RNA from close to either C-terminal or N-terminal end was much less efficient than RNA from central regions. In this paper, the DNA fragment used as template for synthesizing RNA corresponds to Y71F9B 66012-65569, which covers the ninth sop-3 exon.
The construction of sop-3 reporter genes
Reporter gene EM#300 was constructed by joining the 9.3 kb BamHI-digested sop-3-rescuing fragment (from Y71F9B 74891 to 65620), which contains 507 nucleotides 5′ of the putative AUG start codon and the first 9 exons, to the 4.5 kb HincII-BamHI fragment of pPD95.67, which contains gfp and unc-54 3′ UTR (see Fire laboratory vector kit). The transgenic line, bxEx73, was obtained by co-injection of EM#300 with pBR1, carrying the pha-1 gene, into pha-1 hermaphrodites. A reporter gene consisting of the entire sop-3 gene with a gfp insertion was constructed as follows. A plasmid (EM#301) was made by inserting the 1 kb HindIII fragment from pPD103.87 into the HindIII site of a second plasmid (EM#302), which contained the segment 65587 to 58766 of Y71F9B. The 9.3 kb SplI-digested sop-3-rescuing fragment (from 74891 to 65575) was gel-purified and ligated to SplI-digested EM#301. The transgenic line, bxEx74, was obtained by co-injection of the ligation mixture together with pBR1 into pha-1 hermaphrodites.
RESULTS
sop-3(bx96) is a pal-1(e2091) suppressor
During wild-type male development, the C. elegans caudal homolog pal-1 acts cell autonomously in seam cell V6 to activate the transcriptional program leading to postembryonic ray development (see Fig. 2A) (Waring and Kenyon, 1990; Waring and Kenyon, 1991; Hunter et al., 1999). In V6, pal-1 activates the Antennapedia homolog mab-5, which is expressed continuously throughout most of the postembryonic V6 cell lineage (Salser and Kenyon, 1996). mab-5 in turn activates the Abdominal-B homolog egl-5 in a single lineage branch during L2, and the bHLH gene lin-32 during L3 (C. Zhao, PhD thesis, Albert Einstein College of Medicine, 1995; Ferreira et al., 1999). Proper expression of this transcription factor cascade is required for wild-type ray development (Fig. 1A) (Chisholm, 1991; Zhao and Emmons, 1995; Chow and Emmons, 1995; Salser and Kenyon, 1996; Ferreira et al., 1999; Hunter et al., 1999).
The expression of pal-1 in V6 requires a V6-specific cis regulatory element located in the last pal-1 intron. The regulatory mutation pal-1(e2091) contains a point mutation within this putative enhancer that prevents V6 expression (Hunter et al., 1999; Zhang and Emmons, 2000). As a result, mab-5 and its downstream targets are not activated and the V6 cell lineage is transformed to one resembling an anterior seam cell lineage (Waring and Kenyon, 1990; Hunter et al., 1999); instead of rays 2-6, V6 generates longitudinal cuticular ridges, termed alae, normally found along the body only anterior of the ray domain (Pal phenotype, for posterior alae) (Fig. 1B). To identify genes governing expression of the ray transcription factor cascade, we performed a genetic screen for suppressors of the Pal phenotype of pal-1(e2091) (Zhang and Emmons, 2000). One maternal effect recessive suppressor mutation mapping to LG I defines the gene sop-3 (suppressor of pal-1). In sop-3(bx96); pal-1(e2091) mutants, 85% of V6 lineages produced 5 normal rays, compared with 5% in pal-1(e2091) mutants (Fig. 1C and Table 1, lines 1 and 2).
sop-3(bx96) suppresses the pal-1(e2091) mutation by reactivating pal-1 expression in V6. We demonstrated this by showing that sop-3(bx96) could not suppress ray loss in a pal-1 mutant containing a deletion of pal-1 coding sequences, pal-1(ct224). Because of the embryonic lethality of pal-1(ct224), it was necessary to carry out this test in genetic mosaics. We examined the percentage of males with Pal phenotype segregated by a strain of genotype sop-3(bx96); pal-1(ct224); him-5(e1490); sDp3. sDp3 is a free duplication carrying a wild-type allele of pal-1 that is not completely stable, being lost from the V6 lineage at a frequency of about 3% (Waring and Kenyon, 1991). Thus 3.1% of males segregated from pal-1(ct224); him-5(e1490); sDp3 are Pal (n=420) (Table 1, line 3). If sop-3(bx96) could suppress the pal-1 deletion allele, then the Pal phenotype of pal-1(ct224) mosaics would be suppressed and most sides which lost sDp3 would produce V6 rays. However, we found that this was not the case: the frequency of Pal males was not reduced by introduction of sop-3(bx96) into pal-1(ct224); him-5(e1490); sDp3. 3.6% of V6 lineages produced no rays in sop-3(bx96); pal-1(ct224); him-5(e1490); sDp3 mutants (n=220) (Table 1, line 4). This result shows that pal-1-coding sequences are necessary for expression of rays in a sop-3(bx96) mutant and implies that sop-3(bx96) suppresses pal-1(e2091) by restoring pal-1 expression.
sop-3(bx96) affects the activity of additional components of the ray transcription factor cascade
In wild type, pal-1 activates the ray transcriptional cascade by activating mab-5 (Hunter et al., 1999). We therefore tested whether rays in sop-3(bx96); pal-1(e2091) required mab-5. Consistent with the conclusion that sop-3(bx96) suppresses pal-1(e2091) by reactivating pal-1 expression, we found that V6 rays in sop-3(bx96); pal-1(e2091) mutants are dependent on mab-5 activity. In sop-3(bx96); pal-1(e2091) mab-5(e1239) triple mutants, almost no V6 rays are present (Table 1, line 5).
This dependence of ray development on mab-5, however, was not as complete as in an otherwise wild-type background. Whereas in mab-5(e1239), descendants of V6 generate alae instead of rays in 100% of animals, in sop-3(bx96); mab-5(e1239) double mutants, 14% of male sides lacked alae in the post-anal region, and these males generated an average of 1.5 V6 rays per suppressed side (Table 2, line 3). Generation of these mab-5-independent rays required the function of egl-5 because in sop-3(bx96); mab-5(e1239) egl-5(n945) there were no V5 or V6 rays and alae extended into the tail region in all sides (Table 2, line 5). Therefore, it appears that sop-3(bx96) makes expression of egl-5 at least partially independent of mab-5, and egl-5 can apparently activate the ray developmental program. Whereas in wild type, expression of egl-5 is completely dependent on mab-5 gene function, in a sop-3(bx96) mutant, the stringency of this relationship is weakened. We conclude that in addition to altering the conditions for expression of pal-1 in V6, sop-3(bx96) also affects the activity of one or more additional downstream components of the ray transcription factor cascade.
Loss of sop-3 gene function causes mis-regulation of egl-5
sop-3 was cloned by identifying its genetic map location followed by complementation rescue experiments (see below). Isolation of the sop-3 gene allowed us to test the loss-of-function mutant phenotype by RNAi experiments. sop-3(RNAi) in pal-1(e2091) resulted in suppression of the Pal phenotype, confirming identification of the gene and indicating that loss of sop-3 gene function resulted in pal-1 suppression (Table 1, line 6).
RNAi experiments revealed that sop-3 is also required at additional steps in ray development. After microinjection of sop-3 dsRNA into gonads of him-5 hermaphrodites, adult self-progeny males had duplicated rays and fused rays (Fig. 2B-D; Table 2, line 1). Similar phenotypes were seen but at lower frequency in sop-3(bx96) (Fig. 1C; data not shown), suggesting that sop-3(bx96) may be a loss-of-function mutation but is unlikely to be a null allele.
Ray duplications and fusions suggested that expression and/or function of the Hox genes mab-5 and egl-5 were misregulated in the ray lineages. The most striking class of defects consisted of ray duplications. In the wild-type fan, ray identities, numbered from anterior to posterior, can be unambiguously determined by ray order, morphology and position (rays 1, 5 and 7 open on the dorsal surface of the fan; rays 2, 4 and 8 open on the ventral surface of the fan; ray 3 is thin and extends to the fan margin; ray 6 is thick at the base, extends nearly to the margin and lacks the characteristic ring- and-dot ending that marks the external openings of the other rays; and ray 9 extends to the fan margin). Using these criteria, we found that some sop-3(RNAi) males have such duplicated ray patterns as 1-4-5-6-4-5-6-7-8-9, 1-2-3-5-6-5-6-7-8-9, 1-2-3-4-5-6-4-5-6-7-8-9 (Fig. 2B-D). We injected sop-3 dsRNA into a strain carrying a tyrosine hydroxylase reporter gene (cat-2::gfp) that is expressed in the dopaminergic neurons of rays 5, 7 and 9 (Lints and Emmons, 1999). The ectopic ray 5 in sop-3(RNAi) expressed this reporter, confirming duplication of ray 5 (Fig. 2D).
These duplicated ray patterns in sop-3(RNAi), which involved only rays descended from V6 (rays 2-6), suggested that cell fate transformations had occurred during the V6 lineage. For example, the ray pattern 1-4-5-6-4-5-6-7-8-9, in which rays 2 and 3 are absent and rays 4-6 are duplicated, suggested that in this animal V6.pap had taken the fate of V6.ppp (Fig. 2B). Other duplication patterns suggested additional types of cell fate transformations among V6 descendants (Fig. 2C,D). In all such postulated transformations, an anterior cell assumed the fate of a more posterior cell, a transformation easily accounted for by mis-regulation of egl-5, in particular, its ectopic expression in an anterior lineage branch. For example, expression of egl-5 normally occurs in the V6.ppp branch of the cell lineage but not in the V6.pap branch, and expression in the V6.ppp branch is necessary for this cell to generate three rays instead of two, and for ray 5 to express tyrosine hydroxylase (Chisholm, 1991; Ferreira et al., 1999; Lints and Emmons, 1999). V6.pap might take the V6.ppp fate if egl-5 were inappropriately expressed in this cell, generating a duplication of rays 4-6.
To test the hypothesis that egl-5 was expressed in additional cells in sop-3(RNAi), we asked whether egl-5 activity is required for the duplication of V6-rays. We injected sop-3 dsRNA into an egl-5 strain carrying a pkd-2::gfp reporter gene (Barr and Sternberg, 1999). This reporter is expressed in all rays and was used to overcome the difficulty of scoring the number of rays in an egl-5(—) background, where ray morphogenesis is largely blocked. We found that no extra rays are formed in sop-3(RNAi); egl-5 mutants (0/142 sides had more than four V6 rays), indicating that egl-5 activity is required for the duplicated V6-rays in sop-3(RNAi) males.
To test more directly for the mis-regulation of egl-5, we carried out sop-3 RNAi experiments in a strain carrying an egl-5::gfp reporter. In wild type, this reporter is strongly expressed in ray precursor cells R5 and R6 and their progeny, weakly expressed in R4 and its progeny, and only rarely can be seen expressed very weakly in R3 and its progeny (Fig. 2E). After microinjection of dsRNA, expression was strong and consistent in R4-R6 and their progeny, and in additional ray precursor cells postulated to be anterior sister nuclei that never express egl-5 in wild type (Fig. 2F,G). These results confirmed that egl-5 was mis-expressed in the V6 lineage.
Ectopic expression of egl-5 in sop-3(RNAi) requires mab-5
Since expression of egl-5 in the V6 lineage normally requires mab-5 gene function (Ferreira et al., 1999), we asked whether the duplicated rays and mis-regulation of egl-5 also required MAB-5 function. We injected sop-3 dsRNA into a mab-5(e1239) strain and found that there were no duplicated rays (Table 2, line 4). Thus, the ectopic expression of egl-5 in sop-3(RNAi) males is dependent on mab-5 and is not due to the mab-5-independent activation of egl-5. However, as described above for sop-3(bx96), alae are absent from 13% of sides and a few rays are generated in sop-3(RNAi); mab-5(e1239) mutants. This is consistent with the observation that 1 or 2 V6 descendants showed the expression of egl-5 in about 15% of male sides when sop-3 dsRNA was injected into a mab-5(e1239) strain carrying the egl-5::gfp reporter gene (data not shown). Thus, this mab-5-independent weak development of rays is also a sop-3 loss-of-function phenotype.
In addition to ray duplications, ray fusions also suggested that MAB-5 and EGL-5 were expressed at inappropriate levels or were acting inappropriately in the later ray lineages. Ray fusion is thought to result when adjacent rays express the same morphological identity (Baird et al., 1991). Such fusions result if either mab-5 or egl-5 is expressed at abnormal levels in the ray lineages (Chow and Emmons, 1995; Salser and Kenyon, 1996). Ray fusions in sop-3(RNAi) suggest that mab-5 and egl-5 expression levels and/or activities are abnormal during the late L3 and early L4 larval stages – when ray cells are generated and their morphogenetic identities expressed.
We show below that sop-3 is widely expressed throughout development, from embryogenesis onwards. It is strongly expressed in proliferating cells of the vulval cell lineages. Consistent with this, about 2% of hermaphrodites have a protruding vulva phenotype in sop-3(RNAi). In view of the expression in many other cells, it is perhaps surprising that no additional strong phenotypes were observed after injection of sop-3 dsRNA. We do not know whether this is because sop-3 has no additional non-redundant functions, or because RNAi was ineffective in eliminating sop-3 gene function.
sop-3 encodes a novel protein with sequence characteristics shared by several transcriptional regulators, including some that affect Hox gene function
sop-3 was cloned by placing it genetically between cloned genes lin-17 and pop-1 followed by complementation rescue (Fig. 3A). Gene identification was confirmed by showing that sop-3(RNAi) suppressed pal-1(e2091). The predicted sop-3 gene structure was verified by sequencing cDNA clones (Fig. 3B). We attempted to identify the bx96 mutation by sequencing all of the exons and exon-intron boundaries and found no differences from wild type. Therefore, we believe sop-3(bx96) is a mutation in a cis-regulatory element.
sop-3 encodes a predicted protein of 1475 amino acids (Fig. 3C). The only related protein sequence found in available databases was the product of the conceptual C. elegans gene T23C6.1. T23C6.1 encodes a predicted protein of 522 amino acids that is 36% identical and 58% similar across its entire length to the N-terminal segment of SOP-3 (Fig. 3D). RNAi studies of T23C6.1 did not reveal any obvious phenotype (data not shown). In particular, dsRNA injection did not suppress the V6 ray-loss phenotype of pal-1(e2091), and the phenotype obtained when dsRNA of sop-3 and T23C6.1 were simultaneously injected was similar to that in sop-3(RNAi) alone.
Although database searches with SOP-3 identified no proteins other than T23C6.1 with overall sequence similarity, the C-terminal segment of SOP-3 contains a number of homopolymeric amino acid motifs found in several other transcription factors. These include proline-repeats, alanine-repeats, serine-repeats, glutamine-repeats and glycine-repeats (Fig. 3C,D). Homopolymeric tracts of Gln, Ser, Ala and Pro residues are present in the Drosophila proteins Cap-n-collar B (CncB), Teashirt (Tsh), Engrailed (En), Fushi tarazu (Ftz) and some Hox proteins (McGinnis et al., 1998; Fasano et al., 1991; Laughon et al., 1985). The presence of homopolymeric runs in Hox proteins and Hox protein modulators such as Tsh and CncB suggests that such motifs may perform a common function in transcriptional regulation. SOP-3 also contains a highly charged domain near the C terminus: 88% (79 out of 90) amino acids are charged (Asp, Lys, Arg or Glu) (Fig. 3C,D). Ala-rich, Pro-rich and highly charged domains have been associated with transcriptional regulatory functions, suggesting that SOP-3 might function as a transcription factor (Hanna-Rose and Hansen, 1996).
sop-3 is widely expressed throughout development
To determine where sop-3 is expressed and to ask whether regulated expression of sop-3 contributed to regulation of egl-5 expression, we constructed two strains containing sop-3::gfp reporter transgenic arrays (Fig. 4A). One array, bxEx73, consists of the same 0.5 kb upstream region as the minimal rescuing genomic fragment (Materials and Methods) plus the first 9 exons of sop-3, fused in-frame to gfp containing a nuclear localization signal and the 3′UTR of the C. elegans myosin gene unc-54. The other, bxEx74, is an in-frame insertion of gfp into the sop-3-rescuing fragment and contains the entire sop-3 coding sequence, including all the introns plus 738 nucleotides 3′ of the predicted sop-3 stop codon. Both reporters have the same expression pattern. During embryogenesis they are widely expressed in many or all cells (Fig. 4B,C). During larval stages, they are expressed in the seam cells, head neurons, ventral cord, male ray cells and other tail neurons (Fig. 4D,E). The reporter genes are also strongly expressed in proliferating cells; for example, they are expressed in the vulval precursor cells (Fig. 4F). In adult animals, the reporters are expressed mainly in neurons. For both reporters, GFP fluorescence was nuclear. In the case of bxEx74, which does not have an added nuclear localization signal, this suggests that SOP-3 is a nuclear protein. Nuclear localization of SOP-3 is consistent with the inference drawn from the SOP-3 sequence that the protein functions as a transcription factor.
In the V6 lineage, both reporter genes were uniformly expressed throughout development in all lineage branches. Therefore regulated expression of sop-3 does not account for the pattern of egl-5 expression in this lineage. We conclude that sop-3 is necessary for conveying regulatory information to the egl-5 promoter, and that its activity is regulated post-transcriptionally and possibly post-translationally.
SOP-3 may function at the level of the transcriptional Mediator complex
Since the sequence and apparent nuclear localization of SOP-3 suggested that it might be a transcription factor, we tested for genetic interactions with other known components of the transcriptional apparatus. First, we investigated whether sop-3(bx96) interacted with mutations in sop-1. sop-1 mutations have a phenotype nearly identical to sop-3. SOP-1 is the C. elegans homolog of TRAP230, a component of the human Mediator complex (Ito et al., 1999; Zhang and Emmons, 2000). Like sop-3(bx96), loss-of-function mutations in sop-1 are recessive, have maternal effects, and restore pal-1 activity in V6 in pal-1(e2091) but not in pal-1(0) (Zhang and Emmons, 2000). We analyzed double mutants of sop-3(bx96) with a strong allele of sop-1 and found that this combination is synthetic lethal. Whereas sop-3 and sop-1 mutations singly have little effect (<5%, n>1000) on viability, in sop-3(bx96); sop-1(bx92) double mutants, 50% of embryos do not hatch (n=2762) and 78% of hatching animals die at an early larval stage (n=1318). Since neither sop-3(bx96) nor sop-1(bx92) appear to be null alleles, this synthetic lethal phenotype, together with the nearly identical genetic properties of the two genes, suggest that sop-3 and sop-1 act in a single pathway. Because SOP-1 is a presumptive component of the Mediator complex, this conclusion implicates SOP-3 as a possible component of the Mediator complex as well, or as a transcriptional co-factor that interacts with the Mediator complex.
Further support for the hypothesis that sop-3 functions at the level of the Mediator complex came from examining genetic interactions with a second Mediator component, SUR-2. SUR-2 is the C. elegans homolog of a protein of the human Mediator complex, hSur-2, that interacts with transcription factors targeted by the Ras/MAPK pathway (Boyer et al., 1999). Mutations in sur-2 suppress the multivulva phenotype of an activated mutation of Ras and are synthetic lethal in combination with weak loss-of-function mutations in several genes acting in the Ras/MAPK signal transduction pathway (Singh and Han, 1995). A mutation in sur-2 cannot suppress the Pal phenotype of pal-1(e2091) (data not shown), indicating that sur-2 does not function in the same pathway as sop-3 and sop-1 in regulation of pal-1. Likewise, mutations in sop-1 and sop-3 do not suppress an activated-Ras pathway Muv mutation (Zhang and Emmons, 2000; data not shown). Therefore, in the ray and vulval pathways, sop-1 and sop-3 on the one hand and sur-2 on the other have separate functions. Nevertheless, we found that a non-null mutation in sur-2 causes synthetic lethality with sop-3(bx96). In sur-2(ku9) sop-3(bx96) mutants, 38% of animals die as larvae. Consistent with the conclusion that sop-3 and sop-1 lie in the same pathway, a sop-1 mutation is also synthetic lethal with sur-2. In sur-2(ku9); sop-1(bx92) mutants, 78% of animals die as larvae. These synthetic lethal phenotypes could be explained if the introduction of mutations simultaneously into multiple components of the Mediator complex has a cumulative effect, possibly disrupting the integrity or function of the complex. Alternatively, unlike in the ray and vulval pathways, elsewhere sop-1, sop-3 and sur-2 may all be required for expression of one or more essential functions, or their effects in separate pathways may be cumulatively lethal.
sop-3 regulates activity of the Wnt pathway
Previous studies have shown that the Wnt signaling pathway is capable of regulating the expression of pal-1 and mab-5 in V6 in some genetic backgrounds (Hunter et al., 1999; Zhang and Emmons, 2000). The Wnt pathway in the male seam has been defined by a gene for a β-catenin homolog, bar-1, a gene encoding a frizzled receptor, lin-17, and two genes encoding Wnt ligands, egl-20 and lin-44. Mutations in lin-17 and lin-44 cause reversals in polarity or loss of polarity at many seam lineage cell divisions, resulting in severe disruption of tail development (Herman et al., 1995; Eisenmann et al., 1998; Maloof et al., 1999). These pleiotropic effects make testing the roles of lin-17 and lin-44 in ray development difficult.
We examined the role of bar-1 and egl-20 in misregulation of pal-1 and egl-5 in sop-3(bx96) and in sop-3(RNAi) mutants. The mutations we examined, bar-1(ga80) and egl-20(n585), are thought to be a null and strong reduction-of-function mutation, respectively. These mutations singly have no effects on the rays, indicating these genes normally do not have essential roles in ray development. However, we found that bar-1 activity is partially required for the V6 rays in sop-3(bx96); pal-1(e2091) mutants. Only 42% of V6 lineages produced rays in sop-3(bx96); pal-1(e2091); bar-1(ga80) triple mutants, compared with 85% in sop-3(bx96); pal-1(e2091) mutants (Table 1, line 7). V6 produces normal rays in sop-3(bx96); bar-1(ga80) double mutants, indicating that sop-3(bx96) does not make ray development sensitive to bar-1 function in a pal-1(+) background. Thus, bar-1 is partially required for suppression of pal-1(e2091) by sop-3(bx96). A similar partial requirement for bar-1 was found for suppression of pal-1(e2091) by sop-1 mutations (Zhang and Emmons, 2000).
Unlike bar-1, egl-20 activity does not stimulate ray development in sop-3(bx96); pal-1(e2091) mutants: 78% of V6 lineages produce 5 rays in sop-3(bx96); pal-1(e2091); egl-20(n585) mutants (Table 1, line 8). Lack of stimulation by egl-20 could be because bar-1 is activated in a ligand-independent manner, or because of redundancy of EGL-20 with LIN-44.
We showed above that sop-3(bx96) renders the activity of egl-5 in the seam lineages partially independent of mab-5. This mab-5-independent activation of egl-5 is also dependent on bar-1 activity. In sop-3(bx96); mab-5(e1239); bar-1(ga80), alae extended into the post-anal region in almost all animals (Table 2, line 6). However, as with pal-1(e2091) suppression, there was no effect of egl-20(n585) (Table 2, line 7).
Finally, we tested whether abnormal expression of Hox genes as evidenced by generation of duplicated rays and fused rays in sop-3(RNAi) required components of the Wnt pathway. We found that the frequency of duplicated rays and fused rays was significantly reduced by introduction of bar-1(ga80) (Table 2, line 8). Thus, all the effects of mutations in sop-3 (i.e. pal-1(e2091) suppression, mab-5-independent activity of egl-5, and mis-expression of Hox genes) require or are stimulated by bar-1. This, together with the evidence that sop-3 acts at the level of the Mediator complex, strongly suggests that SOP-3 plays a role in relaying signaling to the transcriptional apparatus through the Wnt pathway.
DISCUSSION
Mode of sop-3 action
Tightly regulated expression of the C. elegans Hox genes mab-5 and egl-5 is essential for the development and patterning of the male tail rays. Generation of ray 6 and specification of the identities of rays 3-6 requires that EGL-5 expression be initiated in two branches of the postembryonic V6 seam cell lineage. EGL-5 expression requires MAB-5. However, MAB-5 is expressed not only in lineage branches where it activates egl-5, but also in additional branches of the lineage, as well as at times earlier than the initiation of egl-5 expression. Thus, additional genes are required to bring about the regulated pattern of EGL-5 expression. We have shown that one of these genes is the novel gene sop-3. Wild-type sop-3 function is required to prevent the activation of egl-5 in inappropriate lineage branches by MAB-5 and the Wnt pathway.
The structure of the SOP-3 protein, its nuclear localization, the similar mutant phenotypes of sop-3 and sop-1 (which encodes a putative component of the Mediator complex), and the synthetic lethal interaction of sop-3 and sop-1 all point to a relatively direct role of SOP-3 in regulating transcription initiation. As we showed earlier for sop-1 (Zhang and Emmons, 2000), sop-3 loss-of-function mutations suppress the enhancer mutation pal-1(e2091) by allowing activity of an alternative pathway for pal-1 activation in V6. This alternative pathway somehow involves the Wnt pathway, because the degree of pal-1 suppression is strongly decreased, although not eliminated, by mutation in the β-catenin gene bar-1. Mutations in sop-1 that truncate a C-terminal Q-rich domain allow activation of this alternative pathway. One possibility is that SOP-3 interacts directly with SOP-1, possibly with the Q-rich domain, in bringing about repression of the alternative gene activation pathway. However, other models involving less direct modes of SOP-3 action are also possible.
Earlier studies had shown that mab-5, although it is expressed nearly ubiquitously throughout the V6 cell lineage, acts in only a subset of lineage branches to activate egl-5 expression (Salser and Kenyon, 1996; Ferreira et al., 1999). This could be because its action is blocked in other lineage branches, potentiated in particular lineage branches, or both of these mechanisms could be operating. We show here that mab-5 action in some lineage branches appears to be blocked by sop-3, because in a sop-3(lf) background, ray duplications and ectopic expression of an egl-5 reporter gene require mab-5 gene function. The selective action of sop-3 in blocking mab-5 only in certain lineage branches is not due to regulated expression of sop-3, based on the uniform expression of two sop-3 reporter genes in all V6 descendants. Rather, it appears that SOP-3 activity may be regulated in a lineage-specific manner. Alternatively, its inhibitory activity may be ubiquitous, but blocked or bypassed by another pathway activated in branches of the lineage where egl-5 is activated. This would be analogous to pal-1 activation in V6, which normally occurs independently of Wnt signaling by a pathway that is apparently not subject to the inhibitory effects of SOP-1 and SOP-3.
Where SOP-3 blocks MAB-5 activity, it could do so directly as a Hox protein modulator by interfering with MAB-5 binding to target promoters (presumptively, here, the egl-5 promoter) or by interfering with MAB-5 activity in promoting transcription once it is bound. Alternatively, SOP-3 could act indirectly. For example, it could potentiate a repressor that blocked gene activation by MAB-5, or it could block one or more additional components required together with MAB-5 for transcription initiation. For example, it might block the actions of the extradenticle (ceh-20) or homothorax (unc-62) homologs, both of which are required for normal ray development (Y. Teng and S. W. E., unpublished observations). We do not know in general how the effects of multiple factors ‘add up’ in activation of transcription initiation, and it is possible that SOP-3 blocks a necessary combinatorial pathway that is independent of MAB-5.
We have shown that ectopic gene activation in a sop-3(lf) background is stimulated by bar-1. BAR-1, a β-catenin homolog, is thought to activate gene expression through its interaction with the DNA-binding protein POP-1, a member of the Tcf/LEF family of HMG proteins (Korswagen et al., 2000). POP-1 is differentially expressed in anterior versus posterior daughters at many, and possibly all, anterior/posterior cell divisions during C. elegans development, including those occurring in the postembryonic cell lineages of the seam (Lin et al., 1998). Activation of ray development and ectopic expression of egl-5 by bar-1 in a sop-3(lf) background suggests that SOP-3 prevents a putative BAR-1/POP-1 factor from activating gene expression. In Drosophila and vertebrates, Tcf/LEF factors promote gene expression, through interaction with β-catenin, or alternatively inhibit gene expression, through interaction with the general co-repressor Groucho (Cavallo et al., 1998; Roose et al., 1998). This raises the possibility that SOP-3 regulates the activity of POP-1 target genes by promoting recruitment or activation of the C. elegans Groucho homolog, UNC-37. Indeed, we found significant genetic interactions between both sop-3 and unc-37 and sop-1 and unc-37. Double mutants containing sop-3(bx96) or sop-1(bx92) together with the viable missense mutation unc-37(e262) (Pflugrad et al., 1997) are synthetic lethal (H. Z. and S. W. E., unpublished observations). This finding implicates a role of SOP-3 and SOP-1 in regulatory pathways that also involve UNC-37.
A possible family of transcriptional regulators
SOP-3 characteristics exhibit a number of intriguing parallels with those of several known transcriptional regulators. This group of similar proteins includes the products of cncB, tsh, lin, and mastermind(mam) of Drosophila, and lag-3 of C. elegans (McGinnis et al., 1998; Fasano et al., 1991; Hatini et al., 2000; Petcherski and Kimble, 2000). All of these proteins have several or all of following characteristics: (1) they are nuclear proteins involved in regulation of transcription; (2) they contain homopolymeric runs of amino acids, usually involving Q, S, A, P and G; (3) they are involved in Hox gene regulatory pathways; and (4) they act by modulating the Wnt pathway. Strikingly, though they act in pathways involving well-conserved components, none of these proteins has clear orthologs in other organisms. cncB, tsh and lin, like sop-3, are all involved in Hox gene regulatory pathways, affecting the outcome of Hox gene action (McGinnis et al., 1998; Castelli-Gair, 1998; de Zulueta et al., 1994). CncB and Tsh have recognizable DNA-binding domains (a basic leucine zipper domain in CncB, a Zn-finger domain in Tsh) whereas SOP-3 and Lin do not. SOP-3, Tsh and Lin all affect the action of the Wnt pathway (Gallet et al., 1998; Gallet et al., 1999; Hatini et al., 2000). Tsh modulates Wnt signaling by direct binding to the β-catenin homolog Armadillo (Gallet et al., 1998,Gallet et al., 1999). Alone among these proteins, Lin does not contain homopolymeric runs of amino acids. LAG-3 and Mam differ from the other proteins in being involved in the LIN-12/Notch pathway rather than in Hox gene regulatory pathways, and their structures include a greater number and length of homopolymeric runs of Q residues (Petchershi and Kimble, 2000; Smoller et al., 1990).
LAG-3 provides an attractive model for the function of this putative family of transcriptional regulators. It has recently been shown to participate in a ternary complex between the ankyrin-repeat-containing intracellular domain of the LIN-12 receptor, which translocates to the nucleus upon signaling, and the target DNA-binding factor of this pathway, LAG-1 (Petchershi and Kimble, 2000). A possible similar role for SOP-3 as constituent of a multi-protein complex that includes POP-1, BAR-1, UNC-37 and the Mediator component SOP-1 may be a reasonable premise for further investigation. Why such a mode of action would allow for unusual evolutionary variability of protein sequence is not obvious, since the interactions involved are between evolutionarily conserved components. Possibly the mechanism allows for the function of proteins of mixed and variable functional domains that act in the nature of linker proteins. Such a mechanism might provide an important point of evolutionarily flexibility that can lead to variation in the regulatory interactions involved in the combinatorial control of gene expression.
Acknowledgments
We thank L. Edgar for pal-1(ct224). We thank R. Lints and D. Portman for thoughtful discussions and comments on the manuscript, and C. Smith and J. Dimele for technical assistance. Some strains used in this work were received from the Caenorhabditis Genetics Center, which is supported by a grant from the NIH. This work was supported by NIH grant R01 GM39353. S. W. E. is the Siegfried Ullmann Professor of Molecular Genetics.