The C. elegans SoxC protein SEM-2 opposes differentiation factors to promote a proliferative blast cell fate in the postembryonic mesoderm

The group C Sox proteins are Sry-related HMG box (Sox)-containing transcription factors (Wegner, 2010). Vertebrates contain three highly conserved SoxC genes, Sox4, Sox11 and Sox12, which play important roles in development, including cell differentiation, proliferation and survival (Penzo-Mendez, 2009). Increasing evidence has also shown that many tumor types in humans are associated with significantly elevated level of SoxC gene expression, suggesting that mis-regulation of the SoxC genes may contribute to tumor formation (Penzo-Mendez, 2009; Moreno, 2010). However, the molecular mechanisms underlying these multiple functions of the SoxC genes are not fully understood, nor are the molecular events that regulate the expression of SoxC genes.

The nematode C. elegans contains a single SoxC gene, C32E12.5 (Phochanukul and Russell, 2010), providing an opportunity to determine the functions of SoxC genes at single cell resolution. In this study, we show that C32E12.5 is the sem-2 gene that is widely expressed and essential for C. elegans embryonic development. Furthermore, sem-2 plays a crucial role in a binary fate decision in the postembryonic mesoderm, the M lineage. The M mesoblast is an embryonically born pluripotent precursor cell (Fig. 1A). During hermaphrodite larval development, the M cell first divides dorsoventrally, then left-right and then twice anterioposteriorly to produce 16 cells (Sulston and Horvitz, 1977). Two of them, M.vlpa and M.vrpa, divide one more time. The anterior cells from these final divisions become sex myoblasts (SMs), the precursors to all the non-striated egg-laying vulval and uterine muscles. The two posterior cells differentiate into striated body wall muscles (BWMs). The SMs then migrate to the future vulval region and further proliferate to produce eight vulval muscles (vms) and eight uterine muscles (ums). We show here that sem-2 is required for the binary fate decision between the SMs and BWMs, and is both necessary and sufficient to promote the proliferative SM fate as opposed to the differentiated BWM fate. This specific function and expression of sem-2 in the M lineage is under the direct regulation of Hox/PBC proteins, MAB-5, LIN-39 and CEH-20. This finding is intriguing in light of the oncogenic roles of Hox and PBC factors (Shah and Sukumar, 2010), suggesting the possibility of Hox regulation of SoxC genes during tumorigenesis.

The following strains were used in this study. LG I: sem-2(n1343); sem-2(ok2422)/hT2[qIs48]; sys-1(q544)/hT2[qIs48]. LG II: hlh-1(cc561ts). LG III: fozi-1(cc609); mab-5(e1239) lin-39(n1760)/hT2[qIs48]; ceh-20(os39)/hT2[qIs48]; unc-32(e189) lin-12(n676n930ts). LG X: sma-9(cc604). Analyses were performed at 20°C, unless otherwise noted.

Integrated lines were as follows. LW0081: ccIs4438 (intrinsic CC::gfp) III; ayIs2 (egl-15::gfp) IV; ayIs6 (hlh-8::gfp) X. LW1066: jjIs1066[pJKL705.1(hlh-8p::mRFP+unc-119(+)]?; unc-119(ed4) III. LW1639: mab-5(e1239) lin-39(n1760) III/hT2[qIs48] (I;III); ayIs6(hlh-8p::gfp) X. LW2466: jjIs1475(myo-3::rfp) I; ccIs4438 (intrinsic CC::gfp) III; ayIs2 (egl-15::gfp) IV; ayIs6 (hlh-8::gfp) X.

sem-2(n1343) was identified in a Tc1 mutagenesis screen (Desai et al., 1988). It was mapped using three factor crosses into a small region between unc-63 and spe-11, 0.2 map units to the right of unc-63. Cosmids spanning that region were tested for rescuing activity. Two cosmids, F47D3 and C32E12, both rescued the egg-laying defect of n1343. By further truncating the overlapping region of the cosmids, the rescuing activity was mapped to C32E12.5. To identify the molecular lesion of n1343, PCR reactions were performed using primers designed to span the genomic region of C32E12.5. Primer pairs JKL-620 and JKL-621 amplified a fragment bigger than expected. Further PCR and sequencing identified a Tc1 insertion in the first intron of C32E12.5 (between −9186 and −9185, Fig. 2A).

Fosmid WRM0623cE02 (Geneservice) was used to generate the gfp::sem-2 translational fusion construct via recombineering (Warming et al., 2005). gfp sequences from the Fire lab vector pPD95.75 were inserted immediately after the ATG start codon. Recombineering was also used to insert Tc1 into the gfp::sem-2-containing fosmid generated above between −9186 and −9185. The insertions were verified by PCR and sequencing.

sem-2 cDNA from yk657g12 was used to generate pJKL776 (hlh-8p::sem-2 cDNA::unc-54 3′UTR).

A 4554 bp fragment of the sem-2 1st intron (−11650 to −7097) was cloned into L3135 of the Fire lab vector kit to generate pCXT12, which was subsequently used to generate deletion constructs pCXT18-22 and pCXT26-33. pCXT33 was used to generate pCXT97-99 and pCXT173, carrying mutations in the PBC/Hox-binding site. sys-1(T23D8.9) and let-381(F26B1.7) RNAi constructs were obtained from the Ahringer RNAi library provided by Geneservice (Kamath et al., 2003). sem-2(RNAi) construct pCXT9 was made by subcloning the sem-2 cDNA fragment from yk657g12 into L4440 (Timmons and Fire, 1998). pNMA49 and pNMA50 were used for knocking down fozi-1 and mab-5, respectively (Amin et al., 2009).

Transgenic lines were generated using the plasmids pRF4 (Mello et al., 1991), pJKL449 (myo-2p::gfp::unc-54 3′ UTR) (Jiang et al., 2009) or LiuFD61 (mec-7p::mRFP) (Amin et al., 2009) as markers.

The T7 Ribomax RNA Production System (Promega) was used to generate sem-2 dsRNA using yk404e6 as a template. Synchronized L1 animals expressing various M lineage GFP markers were soaked in the dsRNA solution at 20°C for 24-48 hours following the protocol of Maeda and colleagues (Maeda et al., 2001). Animals were allowed to recover at 20°C and adult worms were scored for M lineage phenotypes. Water was used as a soaking control. For fozi-1(RNAi) (Amin et al., 2007), let-381(RNAi) (Amin et al., 2010), mab-5(RNAi) and sys-1(RNAi) (Amin et al., 2009) synchronized L1 animals expressing different M lineage markers were plated on HT115(DE3) bacteria expressing dsRNA for the gene of interest. Bacteria for ingestion were prepared as described by Kamath and Ahringer (Kamath and Ahringer, 2003). RNAi by ingestion was performed at 25°C and animals were scored for M lineage phenotypes 24-48 hours after plating.

Animal fixation, immunostaining, microscopy and image analysis were performed as described previously (Amin et al., 2007). Guinea pig anti-FOZI-1 (Amin et al., 2007) (1:200) and goat anti-GFP (Rockland Immunochemicals; 1:1000) were used.

6xHis-tagged LIN-39 and CEH-20 fusion proteins were purified as previously described (Liu and Fire, 2000). Complementary single-stranded DNA oligos were 3′-end labeled with biotin using the Biotin 3′ End DNA Labeling Kit (Pierce) and annealed at room temperature for 1 hour. Gel shift reactions and detection were performed using the LightShift Chemiluminescent EMSA Kit (Pierce). Oligonucleotides used are: wt, CXT216/217; canonical, CXT218/219; mut 1, CXT220/221; mut 2, CXT222/223; mut 3, CXT224/225.

sem-2(n1343) was identified in a Tc1 insertion mutagenesis screen for egg-laying defective (Egl) mutants in the mut-2(r459) background (Desai et al., 1988). Hermaphrodites homozygous for n1343 are 100% Egl (n>200), lack all vulval and uterine muscles required for egg laying, as monitored by DIC and polarized light microscopy, and fail to express egg-laying muscle specific reporters, such as egl-15::gfp and ceh-24::gfp [expressed in type I vulval muscles, VM1s (Harfe et al., 1998b), or in all the egg-laying muscles (Harfe and Fire, 1998)] (Fig. 1C-F, data not shown). To determine the basis for the missing egg-laying muscles, we followed the M lineage in n1343 hermaphrodites using both Normaski microscopy and the hlh-8::gfp reporter that marks all undifferentiated cells of the M lineage (Harfe et al., 1998b). In n1343 hermaphrodites, the SM mother cells, M.v(l/r)pa as in wild type give rise to two daughter cells, M.v(l/r)paa and M.v(l/r)pap. However, the anterior daughters, normally the SMs, exhibit the morphology and identity of their sister cells, the body wall muscles (BWM; Fig. 1A,B and data not shown). These cells also fail to migrate anteriorly, never divide and never differentiate into the egg-laying muscles. Therefore, n1343 hermaphrodites exhibit a SM to BWM fate transformation. The M-lineage phenotype of n1343 animals may reflect a complete removal of sem-2 function in the M lineage because n1343 over a deficiency of the sem-2 region, sDf4, yielded the same M-lineage phenotype as homozygous n1343 animals (data not shown). The effect of the sem-2(n1343) mutation on SM development appears to be sex specific, as the SMs appear unaffected in n1343 males, and n1343 males mate efficiently (data not shown).

We mapped sem-2(n1343) to C32E12.5 based on three-factor mapping and cosmid rescue (see Materials and methods). PCR and sequencing analyses showed that n1343 animals contain a Tc1 transposon insertion located 9185 bp upstream of the predicted ATG of C32E12.5 (Fig. 2A). RNAi of C32E12.5 in wild-type worms by soaking L1s caused 98% (n=500) of worms to be Egl and missing egl-15::gfp expression. RNAi-treated worms in the same experiment also exhibited a multivulvae phenotype, with 48% being bi-vulvae and 5% having three vulvae (n=81; Fig. 1G,H and data not shown).

The n1343 mutation does not appear to be a null allele of sem-2, as a deletion allele of sem-2 generated by the C. elegans knockout consortium, ok2422, resulted in 100% embryonic lethality (n>100). ok2422 has most of the C32E12.5-coding sequences deleted and is probably a null allele (Fig. 2A). Time-lapse video microscopy showed that ok2422 homozygous embryos developed normally until late-comma stage (n=5, Fig. 1I-P). However, these embryos were severely delayed in the elongation process, resulting in noticeable morphological defects, such as enlarged heads relative to control embryos (Fig. 1O). They eventually all reached to, and were arrested at, the threefold stage (Fig. 1L,P; see Movie 1 in the supplementary material). A fosmid (WRM0623cE02) containing C32E12.5 rescued the embryonic lethality of ok2422. RNAi soaking of L4s also resulted in 100% embryonic lethality (n>200). Thus, C32E12.5 is an essential gene required in embryogenesis, vulval and M lineage development, and n1343 is a partial loss-of-function allele of C32E12.5 that specifically affects M-lineage development.

Sequencing of the cDNA clones for C32E12.5 showed that the C32E12.5 locus is alternatively spliced with two different splice isoforms. Splice form 1 is represented by the cDNA clone yk657g12, in which two small exons separated by two large introns; 4.5 kb (containing the Tc1 insertion in n1343 animals) and 7 kb, respectively, are located upstream of the ATG-containing exon (Fig. 2A). Splice form 2 is represented by two independent cDNA clones yk1577b07 and yk1661e08, which have a SL1 splicing leader sequence trans-spliced to the ATG-containing exon (Fig. 2A). Both splice isoforms are predicted to encode the same open reading frame that contains 404 amino acids. The predicted SEM-2 protein contains a DNA-binding domain, the SRY/HMG box (residue 92-156), and a C-terminal serine-rich region (residue 334-354) that is predicted to be the transcriptional activation domain (Fig. 2B). Based on the homology in the SRY/HMG region, SEM-2 is most similar to group C Sox proteins, including Sox4, Sox11 and Sox12 in vertebrates and Sox14 in Drosophila (Bowles et al., 2000).

To understand how sem-2 functions, we generated a N-terminal gfp::sem-2 translational fusion and examined its expression pattern. This gfp::sem-2 translational fusion is functional because it rescued the Egl phenotype of n1343 (1/1 line) and the embryonic lethality of ok2422 (2/2 lines). The GFP::SEM-2 protein was nuclear localized, consistent with its predicted role as a transcription factor (Fig. 2C,E,F,I,L,O).

Expression of gfp::sem-2 was first detectable in a subset of cells of the E and MS lineages in early gastrulating-stage embryos (Fig. 2C,D). A similar expression pattern has been reported by Broitman-Maduro et al. (Broitman-Maduro et al., 2005) using a transcriptional reporter of sem-2. The gfp::sem-2 expression persisted through embryonic and larval development in many cell types, including vulval, hypodermal and intestinal cells (Fig. 2E, Fig. 3D).

To determine the expression pattern of gfp::sem-2 in the M lineage, we also labeled the M lineage cells with hlh-8p::rfp or anti-FOZI-1 immunostaining (Amin et al., 2007). gfp::sem-2 expression in the M lineage was first detectable at the 16-M stage in the SM mother cells, M.v(l/r)pa (Fig. 2F-H), and remained in both of their daughter cells, M.v(l/r)paa and M.v(l/r)pap (Fig. 2F-H). The presence of GFP::SEM-2 in M.v(l/r)pap was transient: GFP::SEM-2 was not detectable after M.v(l/r)pap differentiated into BWMs. However, GFP::SEM-2 persisted in the nuclei of the SM cells and all their descendants until the 8-SM stage, and became undetectable at the 16-SM stage (Fig. 2L-Q; data not shown). The expression pattern of GFP::SEM-2 is summarized in Fig. 2R. Thus, sem-2 expression is turned on in the SM mother cells and is retained in the SMs and their proliferating descendants.

As sem-2 is an essential gene, the exclusive M lineage defects observed in sem-2(n1343) mutants suggest that the Tc1 transposon insertion may specifically affect sem-2 expression in the M lineage. To test this hypothesis, we introduced the Tc1 transposon back into the functional gfp::sem-2 translational fusion, at the same insertion site as found in n1343 mutants (Fig. 2A), and examined the function and expression of this reporter: Tc1::gfp::sem-2. Tc1::gfp::sem-2 rescued the embryonic lethality of ok2422 mutants (1/1 line). However, it failed to rescue the Egl phenotype of n1343 mutants (4/4 lines, n>100). Thus, Tc1::gfp::sem-2 is functional outside of the M lineage, but not functional in the M lineage. When we examined its expression pattern, we found that Tc1::gfp::sem-2 was not expressed in the M lineage at all (Fig. 3A-C,E-G; data not shown), whereas its expression outside of the M lineage was largely unaffected (Fig. 3D,H). We also forced the expression of sem-2 cDNA in the M lineage of n1343 animals using the M lineage-specific hlh-8 promoter and found that it rescued the missing egg-laying muscle phenotype in sem-2(n1343) animals, as determined by the reappearance of egl-15::gfp expression (6/6 lines). Together these observations demonstrate that the Tc1 insertion located in the 4.5 kb intron specifically disrupts the M lineage expression of sem-2, and that sem-2 is required within the M lineage for proper SM fate specification.

We then tested whether sem-2 is sufficient to promote non-SM cells to adopt the SM fate. To this end, we used the hlh-8 promoter to force sem-2 cDNA expression in all the undifferentiated M lineage cells in wild-type animals. We first assayed for the effect of sem-2 misexpression on the CC and BWM fates using an intrinsic CC-specific reporter CC::gfp and a BWM-specific reporter myo-3p::rfp. Wild-type worms have four embryonically derived and two M lineage-derived CCs (Fig. 3I). Among animals carrying the hlh-8p::sem-2 construct, 62.5% had no M-derived CCs (Fig. 3J,K) and 14.1% had only one M-derived CC (n=64). The animals that lack M-derived CCs also lacked, on average, 10 out of the 14 BWMs derived from the M lineage (n=9, Fig. 3M,N). By contrast, 40.6% of the animals carrying hlh-8p::sem-2 (n=64) had extra hlh-8::gfp-expressing SM-like cells and, later, extra egl-15::gfp-expressing type I vulval muscle-like cells (Fig. 3J). In most cases the vulval muscles born from animals misexpressing sem-2 were not attached to the vulva properly or not even located in the vulval region, and were therefore not likely to be functional (Fig. 3J,K). Taken together, the loss of M-derived BWM and CCs and the appearance of excessive SM- and vulval muscle-like cells are strongly suggestive of a transformation of some (if not all) of the M lineage cell types to SMs. Thus, sem-2 is not only necessary, but also sufficient, to promote the SM fate.

The M lineage-specific function of sem-2 in promoting the SM fate coupled with the presence of the Tc1 insertion in the 4.5 kb intron and the loss of M lineage-specific expression of Tc1::gfp::sem-2 suggest that the 4.5 kb intron may contain element(s) specifically required for sem-2 expression in the M lineage. To test this hypothesis, we first placed the 4.5 kb intron (Fig. 2A) directly upstream of the gfp-coding sequence, but observed no gfp expression in transgenic lines carrying the construct (two lines, n>100). However, the 7 kb intron of sem-2 alone was capable of driving reporter expression in hypodermal, intestinal and vulval cells that express sem-2 even in the n1343 mutants (data not shown). Taking into account the two types of sem-2 transcripts (Fig. 2A), these observations suggest that the 7 kb intron probably has the promoter responsible for the transcription of splice form 2 in hypodermal, intestinal and vulval cells, whereas the 4.5 kb intron has no promoter activity on its own.

To further test whether the 4.5 kb intron contains any M lineage-specific enhancer(s), we placed the entire 4.5 kb intron sequence upstream of the Δpes-10 basal promoter and gfp, followed by the unc-54 3′UTR (pCXT12, Fig. 4A). We observed M lineage-specific expression of the reporter (6/8 lines, Fig. 4B-G), suggesting the presence of M lineage-specific enhancer element(s) within the 4.5 kb intron. However, the gfp expression pattern from this reporter differs from that of the translational gfp::sem-2 reporter described earlier; whereas GFP::SEM-2 was only found in the SM mother and the SM lineage cells, transgenic lines carrying the pCXT12 reporter showed gfp expression in all the undifferentiated cells in the M lineage, including the early M lineage and the SM lineage (Fig. 4B-G). We reasoned that either additional cis-element(s) are involved in restricting sem-2 expression to specific cells within the M lineage, or the SEM-2 protein is unstable in the early M lineage. To this end, we tested 2.8 kb sequences upstream of the 4.5 kb intron, 2 kb sequences of the 3′ UTR and the entire 7 kb second intron by placing each of them in cis with the 4.5 kb intron. None of them was able to restrict the expression of the 4.5kb intron::gfp reporter in the early M lineage (Fig. 4H). Despite this, our data demonstrate that the 4.5 kb intron contains enhancer(s) positively involved in directing sem-2 expression in the M lineage.

To identify the M lineage-specific enhancer(s) within the 4.5 kb intron, we generated a series of deletions in pCXT12 (Fig. 4A), and found a 598 bp region (−11650 to −7097) that is sufficient to drive reporter expression in the M lineage (Fig. 4A). Alignment of this 598 bp region in C. elegans and its homologous sequences in three related Caenorhabditis species, C. briggsae, C. remanei and C. brenneri, which are more genetically different from C. elegans than mouse is from human, identified several blocks of highly conserved sequences (data not shown) that include a site TGATATATCG (Fig. 5A). This site closely matches the consensus PBC/Hox binding sequence TGATNNAT(G/T)(G/A), with TGAT being the PBC-binding site and AT(G/T)(G/A) being the Hox-binding site (Chan and Mann, 1996; Mann and Affolter, 1998). Interestingly, the Tc1 transposon insertion site in the sem-2(n1343) mutants disrupts the putative PBC half site (Fig. 5A). Furthermore, inserting Tc1 into the same location as in n1343 mutants in the reporter construct pCXT33 completely blocked the M lineage expression of the reporter (pCXT173, 6/6 lines, n>100). We further tested the importance of the putative PBC/Hox-binding site by making clustered mutations in each half site and testing the consequences of the mutations on the M lineage enhancer activity. Mutating the PBC half site (mut1) completely abolished the M lineage expression of the reporter (100%, n=45, Fig. 5B). Mutations in the putative Hox-binding site (mut2 and mut3) also led to the loss of the M lineage enhancer activity in all (100%, n=26, for mut2) or most (93.7%, n=63, for mut3) of the transgenic animals (Fig. 5B). Thus, the putative PBC/Hox-binding site, disrupted by the Tc1 insertion in n1343 mutants, is required for the sem-2 M lineage enhancer activity.

Previous studies have found that two Hox factors, MAB-5 and LIN-39, and their co-factor the PBC homolog CEH-20, play essential roles in M lineage development (Liu and Fire, 2000; Jiang et al., 2009). Both mab-5 and ceh-20 are expressed throughout the M lineage (Liu and Fire, 2000; Jiang et al., 2009), whereas lin-39 is expressed in the SM lineage (Wagmaister et al., 2006). Furthermore, MAB-5 and LIN-39 together, and CEH-20 are required for the proper specification and differentiation of M-derived CC, BWMs and SMs (Jiang et al., 2009). We therefore tested whether these three factors are required for the sem-2 M lineage enhancer activity using the minimal pCXT33 reporter. We introduced pCXT33 into the double null mutant lin-39(n1760) mab-5(e1239) and a strong loss-of-function mutant ceh-20(os39) (Liu and Fire, 2000; Arata et al., 2006; Jiang et al., 2009). gfp reporter expression was detected in the M mesoblast in 98.3% (n=56) of lin-39(n1760) mab-5(e1239)/++ animals (data not shown) and 92.6% (n=54) of ceh-20(os39)/+ animals (Fig. 5D,E; data not shown). However, gfp expression was detected in the M mesoblast in only 4.9% (n=82) of lin-39(n1760) mab-5(e1239)animals and in 8.3% (n=72) of ceh-20(os39) animals (Fig. 5F,G; data not shown). Thus, the PBC/Hox factors MAB-5, LIN-39 and CEH-20 are necessary for the sem-2 M lineage enhancer activity.

To test whether MAB-5, LIN-39 and CEH-20 may directly regulate sem-2 expression in the M lineage, we used the electrophoretic mobility shift assay (EMSA) to test whether recombinant LIN-39 and CEH-20 proteins can cooperatively bind to the putative PBC/Hox-binding site in vitro. We were not able to generate recombinant full-length MAB-5 proteins in vitro (Liu and Fire, 2000). As shown in Fig. 5, LIN-39 alone, or together with CEH-20, binds oligonucleotides with a canonical ANTP/EXD composite site (Knoepfler et al., 1996) or oligonucleotides containing the putative PBC/Hox site in the sem-2 M lineage enhancer (Fig. 5B,C). By contrast, mutating the PBC half site (mut1) completely abolished the composite binding of LIN-39 and CEH-20 proteins without affecting LIN-39 binding alone (Fig. 5C). Similarly, mutating the Hox half site (mut2) completely abolished the binding of LIN-39 alone or both LIN-39 and CEH-20. Consistent with the in vivo reporter assay result, mut3 significantly reduced, but did not completely abolish, the binding of LIN-39 alone or LIN-39 and CEH-20 together (Fig. 5C). The composite binding of LIN-39 and CEH-20 to the putative PBC/Hox-binding site can also be competed away using excess of unlabeled oligonucleotides containing the wild-type binding site (Fig. 5C). Taken together, our data suggest that sem-2 is directly regulated by the Hox factors MAB-5 and LIN-39 and their co-factor CEH-20 in the M lineage. Consistent with our finding, ChIP-seq experiments by modENCODE showed that LIN-39 directly binds to this Hox-PBC site in vivo (Gerstein et al., 2010).

Previous studies have shown that LIN-12/Notch signaling is required for promoting the SM fate on the ventral side, whereas the antagonism of TGF-β signaling by SMA-9 is required for the CC fate on the dorsal side, of the M lineage (Greenwald et al., 1983; Foehr et al., 2006; Foehr and Liu, 2008). We therefore tested whether SM fate and sem-2 expression in the M lineage is under the control of these dorsoventral patterning mechanisms. As shown in Fig. 6A-C, no gfp::sem-2 expression was observed in M.v(l/r)pa cells at the 16-M stage (n=44) in unc-32(e189) lin-12(n676n930ts) animals at the restrictive temperature, in which the SMs are transformed to CCs (Greenwald et al., 1983; Foehr and Liu, 2008). By contrast, gfp::sem-2 expression was detected in both M.d(l/r)pa and M.v(l/r)pa, as well as in their descendants at the 16-M stage (n=18, Fig. 6D-F) and the 18-M stage (n=4; data not shown), respectively, in sma-9(cc604) mutants, which show a M-derived CC to SM fate transformation (Foehr et al., 2006). Furthermore, sem-2 is required for specifying the dorsal SMs in sma-9(cc604) mutants, as both the endogenous and the ectopic SMs in sma-9(cc604) mutants were lost upon sem-2(RNAi) treatment (97.3%, n=75) or in sem-2(n1343); sma-9(cc604) double mutants (100%, n>200). Thus, the M lineage expression of sem-2 is downstream of both LIN-12 and SMA-9.

We have recently shown that the FoxF/C forkhead transcription factor LET-381 also functions downstream of SMA-9 and LIN-12 to promote CC fate specification (Amin et al., 2010). let-381 is expressed in M.d(l/r)pa and their mothers, and let-381(RNAi) leads to the fate transformation of M-derived CCs to the SM mother cells (Amin et al., 2010). In let-381(RNAi) animals, gfp::sem-2 was detected in both M.d(l/r)pa and M.v(l/r)pa, as well as in their descendants (Fig. 6G-I, n=23/45). Furthermore, no SMs were produced in sem-2(RNAi) let-381(RNAi) animals (94.1%, n=118), suggesting that sem-2 functions downstream of let-381. Thus, the exclusive ventral expression of sem-2 in the M lineage is due to the negative regulation of LET-381, the expression of which is repressed by LIN-12 on the ventral side and activated by SMA-9 on the dorsal side (see Fig. 7).

Along the anterior-posterior axis, the Wnt/β-catenin asymmetry pathway is required for proper M lineage fate specification at the 16-M stage (Amin et al., 2009). Specifically, SYS-1/β-catenin is enriched in the nuclei of posterior cells, and mutations of sys-1 lead to a posterior-to-anterior fate transformation and the production of extra SMs and CCs. POP-1/TCF, however, has a reciprocal localization and loss-of-function phenotypes (Amin et al., 2009). Because sem-2 expression is first detected only in M.v(l/r)pa cells but not in their posterior sister cells M.v(l/r)pp at the 16-M stage, we looked to see whether sem-2 expression in the M lineage is under the control of the Wnt/β-catenin asymmetry pathway. As shown in Fig. 6J-L, gfp::sem-2 was expressed in both M.v(l/r)pa and M.v(l/r)pp (33.3%, n=26), and all the endogenous and ectopic SMs (85.7%, n=14) in sys-1(RNAi) animals. sem-2 is also required for both the endogenous and the ectopic SMs produced in sys-1(q544) animals, as sem-2(RNAi) sys-1(q544) animals produced no SMs at all (100%, n=20). Thus, sem-2 acts downstream of, and is negatively regulated by, SYS-1.

Previous studies have shown that the MyoD homolog HLH-1 functions redundantly with the zinc-finger protein FOZI-1 to specify M-derived BWMs and CCs while repressing the SM fate (Harfe et al., 1998a; Amin et al., 2007). We found that gfp::sem-2 was expressed in both the endogenous and the ectopic SMs in fozi-1(RNAi) animals (98.6%, n=70, Fig. 6M-O) and hlh-1(RNAi) animals (100%, n=25). Furthermore, no SMs were produced by sem-2(RNAi); fozi-1(cc609) animals (100%, n=71) or sem-2(RNAi); hlh-1(cc561ts) animals at the restrictive temperature (100%, n=30). These observations suggest that sem-2 expression is repressed in the M-derived BWMs and CCs by HLH-1 and FOZI-1.

Because the SMs are transformed to BWMs in sem-2(n1343) mutants, we also examined whether fozi-1 and hlh-1 are expressed in the ectopic BWMs in n1343 animals. As shown in Fig. 6P-S, fozi-1, which is transiently expressed in all M-derived BWMs prior to their differentiation in wild-type animals (Fig. 6P-Q), is expressed in the ectopic BWMs in sem-2(n1343) animals (Fig. 6R,S). Similar results were also obtained for hlh-1 expression in sem-2(n1343) animals (data not shown). As the expression of sem-2 in the SM mother cells (M.v(l/r)pa) overlaps with those of hlh-1 and fozi-1, the above results suggest that the BWM-specifying factors FOZI-1 and HLH-1 and the SM-specifying factor SEM-2 mutually repress the expression of each other to maintain their proper expression pattern in the respective daughter cells derived from the SM mothers.

We have demonstrated that the single SoxC protein SEM-2 is both necessary and sufficient to specify the SM fate in the C. elegans postembryonic mesoderm. SMs are precursors that have the potential to proliferate and then differentiate into 16 non-striated egg-laying uterine and vulval muscles. Disrupting sem-2 expression specifically in the M lineage leads to a transformation of SMs to cells that differentiate into striated BWMs. sem-2 is specifically expressed in Mv(l/r)pa, a bi-potent precursor cell that asymmetrically divides to give rise to a SM and a BWM. sem-2 expression is retained in the SM, perdures in its descendants that remain proliferative and ceases when these cells switch from a proliferation state and differentiate into mature egg-laying muscles (Fig. 2R). Forced overexpression of sem-2 throughout the M lineage led to the loss of M-derived BWMs and CCs and to the gain of extra SMs (Fig. 3J-N). These experiments demonstrate that SEM-2 is both necessary and sufficient for promoting the egg-laying muscle precursor SM fate while preventing the sem-2-expressing cells from differentiating into BWMs. Consistent with this, ectopic sem-2 expression was detected in mutant animals with extra SMs, and removing sem-2 in these animals blocked the formation of both ectopic and normal SMs.

sem-2 is also broadly expressed during embryogenesis and essential for embryonic development. sem-2 null mutant embryos were arrested at the threefold stage. Time-lapse movies showed that this arrest is preceded by delays in hypodermal migration and elongation. Besides the embryonic phenotype, RNAi knockdown of sem-2 postembryonically caused a multivulval (Muv) phenotype, which may result from a fate specification defect or a proliferation defect in the vulval precursor cells. Further characterization of the embryonic and vulval phenotypes of sem-2 mutants will help shed light on how sem-2 functions in these processes.

Our findings provide an example of how lineage information and positional information are integrated to activate the cell type-specific expression of a cell fate determinant (in this case sem-2) in the specification of the sex myoblast cells. The M lineage expression of sem-2 is controlled by a cis-acting M lineage enhancer located in the 4.5 kb intron (Fig. 4A). Disruption of this site by the transposon Tc1 in n1343 animals specifically disrupted the expression and function of sem-2 in the M lineage without affecting its expression in other cells. We showed that the Hox factors MAB-5 and LIN-39 and their co-factor, the PBC protein CEH-20, directly bind this site to activate sem-2 expression in the M lineage (Fig. 5B,C). However, these three genes are also expressed in other cell types in C. elegans (Costa et al., 1988; Wagmaister et al., 2006; Jiang et al., 2009). In fact, it has been shown that MAB-5/CEH-20 and/or LIN-39/CEH-20 complexes directly activate the expression of hlh-8 throughout the M lineage before terminal differentiation (Liu and Fire, 2000), of egl-18, elt-6, eff-1 and lag-2 in the developing vulva (Koh et al., 2002; Shemer and Podbilewicz, 2002; Takacs-Vellai et al., 2007), and of egl-1 in the P11 lineage and the VC neurons (Liu et al., 2006; Potts et al., 2009). A Hox-independent role of CEH-20 has also been found in the activation of mls-2 in the early M lineage (Jiang et al., 2008). Thus, additional factors must cooperate with Hox/PBC proteins to refine the cell type-specific expression of Hox/PBC targets.

Our results demonstrate that, in addition to the Hox factors and CEH-20, the integration of dorsal-ventral (D/V) and anterior-posterior (A/P) positional information is crucial in dictating the specific localization of sem-2 within the M lineage. Previous studies have shown that LIN-12/Notch signaling and SMA-9, which antagonizes the Sma/Mab TGF-β signaling pathway, work independently to control D/V patterning of the M lineage, and that they both regulate the expression of the FoxF/C transcription factor LET-381 for its dorsal M lineage expression (Foehr et al., 2006; Foehr and Liu, 2008; Amin et al., 2010). We showed that the proper pattern of sem-2 expression is due to the presence of LIN-12/Notch signaling and the absence of SMA-9 along the D/V axis. Furthermore, sem-2 functions downstream of let-381 and is negatively regulated by let-381 in the dorsal M lineage (Fig. 6G-I). Currently, it is not clear whether this negative regulation of sem-2 by let-381 is direct or not.

Along the AP axis at the 16-M stage, POP-1/TCF functions as a repressor in M.v(l/r)pa due to low nuclear levels of SYS-1/β-catenin and is converted to an activator in M.v(l/r)pp due to high nuclear levels of SYS-1/β-catenin (Kidd et al., 2005; Liu et al., 2008; Amin et al., 2009). We showed that sys-1(RNAi) caused ectopic expression of gfp::sem-2 and ectopic production of SMs (Fig. 6J-L). As sem-2 is expressed in M.v(l/r)pa where POP-1 acts as a repressor, these results suggest that POP-1 must activate the expression of a repressor of sem-2 in M.v(l/r)pa to restrict sem-2 expression along the anterior-posterior axis in the M lineage. We hypothesize that an additional factor(s) must be present either in M.v(l/r)aa to repress sem-2 expression or in M.v(l/r)pa to promote sem-2 expression in order to ensure specific expression of sem-2 in M.v(l/r)pa (see model in Fig. 7).

Immediately after the SM mother cell M.v(l/r)pa divides, both its daughter cells express gfp::sem-2 (Fig. 2I-K). However, gfp::sem-2 perdures in the anterior SM lineage, but not in the posterior BWMs. hlh-1 and fozi-1 exhibit a reciprocal expression pattern: they are turned off in the SMs, but remain expressed in the BWMs (Harfe et al., 1998a; Amin et al., 2007). The initiation of the asymmetric expression of sem-2 versus hlh-1 and fozi-1 may be due to the nuclear POP-1 and SYS-1 asymmetry along the anterior-posterior axis (Amin et al., 2009). Once initiated, sem-2 and hlh-1 and fozi-1 appear to mutually repress each other to maintain their proper expression, as sem-2 is expressed in all the ectopic SMs in hlh-1 and fozi-1 mutant or RNAi animals, while fozi-1 and hlh-1 are expressed in the ectopic BWMs in sem-2(n1343) animals.

SEM-2 is a member of the SoxC subfamily, which includes Sox14 in Drosophila and Sox4, Sox11 and Sox12 in vertebrates (Bowles et al., 2000). Drosophila Sox14 exhibits a dynamic expression pattern during development and regulates dendrite severing and other unknown processes essential for fly metamorphosis (Sparkes et al., 2001; Kirilly et al., 2009; Ritter and Beckstead, 2010). Functional studies on vertebrate SoxC members have shown that SoxC proteins play essential roles in multiple lineages such as oligodendrocytes, B lymphocytes, osteoblasts and others to regulate cell fate specification, cell differentiation and cell survival (Penzo-Mendez, 2009). In humans, Sox4 overexpression has often been detected in cases of prostate cancer (Dhanasekaran et al., 2001; Ernst et al., 2002; Lapointe et al., 2004; Liu et al., 2006; Luo et al., 2001; Magee et al., 2001; Rhodes et al., 2002; Welsh et al., 2001). Sox4 is also overexpressed in many other types of cancers, including leukemias (Andersson et al., 2007), melanomas (Talantov et al., 2005), glioblastomas (Sun et al., 2006), medulloblastomas (Lee et al., 2002), bladder cancer (Aaboe et al., 2006) and lung cancer (Friedman et al., 2004). Similarly, Sox11 is highly expressed in medulloblastomas (Lee et al., 2002), gliomas (Weigle et al., 2005), non-B cell lymphomas (Wang et al., 2008) and epithelial ovarian tumors (Brennan et al., 2009). However, the underlying mechanism by which the Sox proteins may contribute to cancer is not fully understood. We have found that sem-2 is specifically expressed in the proliferating SM mother cells, the SMs and their descendants prior to terminal differentiation (Fig. 2R). Furthermore, increasing sem-2 level throughout the M lineage is sufficient to transform other M lineage cells into proliferating SM-like cells, even though they express normal levels of CC- and BWM-specifying and differentiating factors. Thus, it appears that increasing sem-2 level is sufficient to tip the balance between proliferative and differentiative factors and allows for proliferation. This pro-proliferation function of SEM-2 may be a conserved role of the group C Sox proteins.

Hox and PBC proteins have also been implicated in tumorigenesis. There are numerous examples of Hox gene overexpression in various tumor types (Argiropoulos and Humphries, 2007; Moreno, 2010; Shah and Sukumar, 2010). Given our finding that sem-2 expression in the M lineage is directly activated by Hox-PBC proteins and that there is overexpression of Sox4 and Sox11 in a variety of cancers, it is possible that the oncogenic activity of Hox genes in some cases is due to their direct activation of SoxC gene expression.

We thank the C. elegans Genetics Center, the knockout consortium and Yuji Kohara for strains and yk clones; Nirav Amin for examining sem-2 expression in let-381(RNAi) animals; Alex Beatty and Rich McCloskey for help with preparing embryos for the time lapse movies; Nirav Amin and Jared Hale for helpful discussions and critical comments on the manuscript. This work was supported by NIH R01 GM066953 (to J.L.). Deposited in PMC for release after 12 months.