Members of the spalt (sal) gene family encode zinc-finger proteins that are putative tumor suppressors and regulate anteroposterior (AP) patterning,cellular identity, and, possibly, cell cycle progression. The mechanism through which sal genes carry out these functions is unclear. The Caenorhabditis elegans sal gene sem-4 controls the fate of several different cell types, including neurons, muscle and hypodermis. Mutation of sem-4 transforms particular tail neurons into touch-neuron-like cells. In wild-type C. elegans, six touch receptor neurons mediate the response of the worm to gentle touch. All six touch neurons normally express the LIM homeobox gene mec-3. A subset, the two PLM cells, also express the Hox gene egl-5, an Abdominal-B homolog, which we find is required for correct mec-3 expression in these cells. The abnormal touch-neuron-like-cells in sem-4 animals express mec-3; we show that a subset also express egl-5.
We report: (1) that ectopic expression of sem-4 in normal touch cells represses mec-3 expression and reduces touch cell function; (2)that egl-5 expression is required for both the fate of normal PLM touch neurons in wild-type animals and the fate of a subset of abnormal touch neurons in sem-4 animals, and (3) that SEM-4 specifically binds a shared motif in the mec-3 and egl-5 promoters that mediates repression of these genes in cells in the tail. We conclude that sem-4 represses egl-5 and mec-3 through direct interaction with regulatory sequences in the promoters of these genes, that sem-4 indirectly modulates mec-3 expression through its repression of egl-5 and that this negative regulation is required for proper determination of neuronal fates. We suggest that the mechanism and targets of regulation by sem-4 are conserved throughout the sal gene family: other sal genes might regulate patterning and cellular identity through direct repression of Hox selector genes and effector genes.
INTRODUCTION
Members of the spalt (sal) gene family encode zinc-finger proteins that both control normal development and appear to function as tumor suppressors in human beings and mice. Drosophila sal is required for many aspects of development, including the establishment of head and tail identity(Jurgens, 1988), and photoreceptor differentiation in the eye(Mollereau et al., 2001). In Caenorhabditis elegans, the sal gene sem-4 controls the fate of several different cell types including neurons, muscle and hypodermis(Basson and Horvitz, 1996). The human gene SALL1 (also known as Hsal1) is mutated in patients with Townes-Brocks syndrome (TBS), an autosomal dominant developmental disorder characterized by sensorineural hearing loss and by malformations of the limbs, anus, kidneys, heart and gonads(Kohlhase, 2000; Ma et al., 2001b; Surka et al., 2001). SALL1 and mouse sall1 act as repressors in cell culture assays(Kiefer et al., 2002; Netzer et al., 2001). Another human sal gene, HSAL2 (also known as SALL2), maps to a region associated with human ovarian cancers, and alterations in the expression of HSAL2 have been found in human ovarian carcinoma(Ma et al., 2001a). The mouse homologue of HSAL2, msal-2, is a target of polyoma tumor virus large T antigen (Li et al.,2001).
Three fundamental questions about the functions of sem-4 and the other sal genes remain unresolved: (1) do sal genes bind DNA; (2) do they function as repressors in vivo; and (3) what are their targets? This work identifies the C. elegans sal gene sem-4 as a regulator of C. elegans homeobox gene expression and examines the mechanism through which sal genes control cellular identity. Like other sal genes, sem-4 is required for multiple aspects of development. The strongest alleles of sem-4 are egg-laying defective (Egl),uncoordinated (Unc), partially sterile and constipated, and have deformed tails (Basson and Horvitz,1996). Animals with sem-4 mutations exhibit abnormalities in a range of different cell types, including neurons, muscle cells,coelomocytes and vulval cells (Basson and Horvitz, 1996; Grant et al.,2000). In particular, sem-4 animals produce additional touch-neuron-like cells that express the touch cell effector gene mec-3 (Basson and Horvitz,1996; Mitani et al.,1993; Mitani,1995).
We report that, in sem-4 animals, the abnormal touch-neuron-like cells and their precursors ectopically express the Hox gene egl-5, an Abdominal-B (Abd-B) homolog. We show that inappropriate expression of egl-5 transforms the fates of these neuroblasts and neurons, that SEM-4 binds to a shared motif in the mec-3 and egl-5 promoters, and that ectopic sem-4 represses mec-3 expression in vivo. Our findings point to three conclusions:first, that sem-4 and other sal genes are repressors that control cellular identity by restricting expression of Hox genes and effector genes;second, that sal genes independently regulate these genes at multiple stages in developmental pathways; and, third, that sal proteins bind directly to a shared motif in regulatory regions of their targets.
MATERIALS AND METHODS
Nematode strains and maintenance
Wild-type C. elegans (var. Bristol, N2) and mutant strains were grown as described by Brenner (Brenner,1974). The following mutant strains were used: CB3531[mab-5(e1239)III; him-5(e1490)V](Kenyon, 1986); EM597[him-5(e1490)V; lin-15(n765)X; bxIs12(egl-5::gfp lin-15(+))]; EM783 [pha-1(e2123)III; him-5(e1490)V; bxEx87(egl-5::gfp pha-1(+))]; EM784 [pha-1(e2123)III; him-5(e1490)V; bxEx88(Pv6cregfp pha-1(+))]; EM785[pha-1(e2123)III; him-5(e1490)V; bxEx89(Pv6creδ100gfp pha-1(+))]; MH1346[unc-119(ed3)III; kuIs35 (sem-4::gfp unc-119(+))] (Grant et al.,2000); MT1081 [egl-5(n486)III](Trent et al., 1983); MT1514[lin-39(n709)III] (Li and Chalfie, 1990); MT3179 [sem-4(n1378)I](Desai et al., 1988); MT5825[sem-4(n1971)I] (Basson and Horvitz, 1996); MT5826 [sem-4(n2087)I](Basson and Horvitz, 1996);MT6921 [sem-4(n2654)I] (Basson and Horvitz, 1996); TU2562[dpy-20(e1282)IV; uIs22(mec-3::gfp dpy-20(+))] (Wu et al.,2001). Double mutant strains were created according to standard genetic methods (Brenner,1974).
Phenotypic characterization
Cell lineages were followed as described by Sulston and Horvitz(Sulston and Horvitz, 1977). Expression of gfp reporters was observed at 1000×magnification. The touch sensitivity assay was modified from Hobert et al.(Hobert et al., 1999). Each animal was touched with an eyebrow hair ten times alternately at the head and tail. At least 100 animals were scored for each stable line or mutant strain. To ensure that observations of cells in temperature-sensitive sem-4and Hox mutant strains were not influenced by maternal effects, we scored worms from at least the third generation grown at each particular temperature. At least 45 gravid adults were scored for each strain at each temperature.
Expression studies
Expression studies using sem-4 were carried out using a PCR-amplified fragment of the sem-4 genomic sequence from cosmid F15C11 that was cloned into the PstI and KpnI sites of Fire vector pPD 95.75 (Fire et al.,1990). The fragment contained 5 kb upstream of the sem-4start site and the entire genomic sequence fused at the 3′ end to gfp. This construct was injected into N2 worms and two stable lines were obtained. It was difficult to produce stable lines containing this expression construct because the stable lines exhibited various elements of the sem-4 phenotype, including sterility and deformed tails. In addition, in one instance, T.pppp underwent a necrotic death in a sem-4::gfp transformant, as it did in the sem-4(n1971;mec-3::gfp) strain. We also used MH1346, containing an integrated sem-4::gfp reporter, for sem-4 expression studies.
egl-5 expression studies were carried out using the integrated egl-5::gfp array in strain EM597 (bxIs12), and the extrachromosomal arrays in strains EM783, EM784 and EM785. bxIs12 was generated by integration of a transgene (EM#285) that was constructed to contain 16,027 nucleotides upstream of the egl-5 AUG (beginning at an NruI site at position 23,448 in cosmid C08C3), the full set of egl-5 exons and introns, and 2639 nucleotides downstream of the egl-5 stop codon (ending at position 43,981, the right end of C08C3). GFP was inserted at an ApaI site in the third egl-5 exon at position 40,261 and disrupts the homeodomain, so that the expressed protein is expected to be non-functional. The arrays in EM783, EM784 and EM785 were constructed by insertion of gfp into the ApaI site in exon 3 of egl-5. EM783 (containing cosmid C08C3 bp 36249-43981), EM784(C08C3 35986-36293) and EM785 (C08C3 36088-36293) were generated by PCR fusion(Hobert et al., 1999).
For ectopic sem-4 expression studies, Pmec-7sem-4 constructs were created by PCR amplification of portions of cosmid F15C11. For ease of cloning, we omitted the first, small exon from these clones, which began instead at the second start site (in exon 2) identified by Basson and Horvitz (Basson and Horvitz, 1996). The resulting DNA fragments were cloned into the XmaI and KpnI sites of Fire vector pPD52.102(Fire et al., 1990). The Pmec-7sem-4 (n1378) construct contained an additional point mutation that produced a D506V change and the Pmec-7sem-4 (n2087) construct contained an additional point mutation that produced a T301A change. For studies of the effect of ectopic sem-4 on mec-3 expression, Pmec-7sem-4 constructs were injected into TU2562. For each construct, we scored three independent stable lines and, for each stable line,we scored at least 50 worms. Plasmids were injected with the dominant rol-6(su1006) marker plasmid pRF4 using standard methods(Mello et al., 1992). Injected DNA forms stable extrachromosomal arrays containing multiple copies of the injected construct.
Gel-mobility-shift assays
A PCR-amplified full-length sem-4 cDNA was cloned into the BamHI and NotI sites of pGEX 6P-1 (Amersham Pharmacia Biotech). The construct produced an N-terminal glutathione-S-transferase (GST) fusion. This GST::SEM-4 fusion was overproduced in BL21(DE3)pLysS cells and the fusion protein solubilized as described in Frangioni and Neel (Frangioni and Neel, 1993). SEM-4 was cleaved from the glutathione-bound GST moiety according to the manufacturer's instructions in the following cleavage buffer: 50 mM Tris-HCl pH 7.0, 150 mM NaCl, 1 mM EDTA, 0.1% bovine serum albumin (BSA), 10 mM DTT, 2% Triton X-100, 0.5 mM ZnSO4.
The 96 bp mec-3 promoter fragment used as a probe in the Fig. 4A gel-mobility-shift assays extended from positions 1714 to 1809 of the mec-3 promoter(Way and Chalfie, 1988). The m3-1 sequence extends from 1759 to 1764, m3-2 from 1714 to 1718 and m3-3 from 1648 to 1654. The 100 bp egl-5 promoter fragment used as a probe in the Fig. 4B gel-mobility-shift assays extended from position 38515 to 38615 of cosmid C08C3. The e5-1 sequence extends from position 38525 to 38531, e5-2 from 38551 to 38557 and e5-3 from 38600 to 38606. The 105 bp egl-5 promoter fragment used as a probe in the Fig. 4Dgel-mobility-shift assays extended from position 35986 to 36091 of cosmid C08C3. The e5-4 sequence extends from position 35997 to 36003, e5-5 from 36054 to 36060 and e5-T1 from 36070 to 36087.
DNA probes were end labeled using T4 polynucleotide kinase (US Biochemicals) according to the manufacturer's instructions. Labeled probes were separated from free nucleotide on an Amicon Millipore YM-3 column. The following reagents were added to the purified protein, probe and any cold competitor, at these final concentrations: 8.5 mM NaHEPES (pH 7.9), 30 mM KCl,10.4 mM DTT, 0.3 mM PMSF, 2% Triton X-100, 1 mg ml-1 BSA, 4%Ficoll, 8 μg ml-1 polydIdC (Amersham Pharmacia Biotech), 0.5 mM ZnSO4. The purified SEM-4 protein was preincubated in binding buffer with any cold competitor for 10 minutes at room temperature; the probe was then added and the reaction incubated for an additional 15 minutes at room temperature. The reactions were run on 4% acrylamide gels (cross-link acrylamide:bisacrylamide 37.5:1) at 4°C for about 2 hours at 200 V in 0.5× TBE. Before the reactions were loaded, the gels were prerun at 4°C for 2 hours at 200 V in 0.5× TBE.
RESULTS
Loss of sem-4 function causes proliferation of two distinct touch-cell fates
We have continued the characterization of the sem-4 phenotype begun by Mitani et al. (Mitani et al.,1993) and Basson and Horvitz(Basson and Horvitz, 1996). They concluded that sem-4 animals produced two additional touch-neuron-like cells (Basson and Horvitz, 1996; Mitani et al.,1993), instead of PHC neurons, possibly as the result of transformation of a neuronal sublineage (the T.pp lineage)(Fig. 1) into a PLM-like lineage (Basson and Horvitz,1996). The identification of these abnormal cells as touch-neuron-like was based on their expression of several different touch-cell genes, including the touch-cell effector gene mec-3, and the resemblance of their processes to touch-cell processes by electron microscopy (Basson and Horvitz,1996; Mitani,1995; Mitani et al.,1993). Basson and Horvitz(Basson and Horvitz, 1996) also observed that, in sem-4 mutants, the fate of the anterior daughter(T.ppa) of the T.pp neuroblast was variable: it sometimes died, sometimes differentiated and sometimes divided.
We report three additional findings that clarify and extend the initial characterization of T-lineage defects in sem-4 animals. First, we found that the number of ectopic touch-neuron-like cells, as defined by their expression of a mec-3::gfp fusion, ranged from one to five, with more produced at higher temperatures (Table 1): at 25°C, 44% of sem-4 animals produced more than two additional touch-neuron-like cells. Second, the ectopic touch-neuron-like cells displayed two distinct morphologies(Fig. 2): some resembled PLM touch neurons, with cell bodies flattened against the muscle line and processes extending anteriorly and posteriorly; others resembled PVM touch neurons, with rounded cell bodies and processes extending ventrally and then anteriorly. Third, about 80% of the lineages were transformed into a PLM-like lineage (Lineage 1 in Fig. 1);the remaining lineages appear to be variants of the PVM lineage (Lineage 2 in Fig. 1).
. | Temperature . | . | . | . | . | . | . | . | . | . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 15°C . | . | . | 20°C . | . | . | 25°C . | . | . | 25°C . | |||||||||
mec-3::gfp strains . | <2 . | 2 . | >2 . | <2 . | 2 . | >2 . | <2 . | 2 . | >2 . | PLM-like . | |||||||||
Wild type | 10 | 89 | 1 | 0 | 100 | 0 | 0 | 100 | 0 | 90% (90/100) | |||||||||
egl-5 | 78 | 22 | 0 | 49 | 51 | 0 | 3 | 97 | 0 | 77% (77/100) | |||||||||
<4 | 4 | >4 | <4 | 4 | >4 | <4 | 4 | >4 | |||||||||||
sem-4 | 12 | 80 | 8 | 8 | 64 | 28 | 6 | 50 | 44 | 84% (283/338) | |||||||||
sem-4; egl-5 | 92 | 6 | 2 | 42 | 46 | 12 | 4 | 92 | 4 | 30% (64/213) |
. | Temperature . | . | . | . | . | . | . | . | . | . | |||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
. | 15°C . | . | . | 20°C . | . | . | 25°C . | . | . | 25°C . | |||||||||
mec-3::gfp strains . | <2 . | 2 . | >2 . | <2 . | 2 . | >2 . | <2 . | 2 . | >2 . | PLM-like . | |||||||||
Wild type | 10 | 89 | 1 | 0 | 100 | 0 | 0 | 100 | 0 | 90% (90/100) | |||||||||
egl-5 | 78 | 22 | 0 | 49 | 51 | 0 | 3 | 97 | 0 | 77% (77/100) | |||||||||
<4 | 4 | >4 | <4 | 4 | >4 | <4 | 4 | >4 | |||||||||||
sem-4 | 12 | 80 | 8 | 8 | 64 | 28 | 6 | 50 | 44 | 84% (283/338) | |||||||||
sem-4; egl-5 | 92 | 6 | 2 | 42 | 46 | 12 | 4 | 92 | 4 | 30% (64/213) |
For wild-type and egl-5 worms, at each temperature the percentage of worms with fewer than two, two, or greater than two mec-3::gfp-expressing tail cells is shown. For sem-4 and sem-4; egl-5 worms, at each temperature, the percentage of worms with fewer than four, four, or greater than four mec-3::gfp-expressing tail cells is shown. The percentage of PLM-like cells is shown for wild-type and egl-5 worms at 25°C. The percentage of PLM-like cells is shown for sem-4 and sem-4; egl-5 worms at 25°C. At least 50 gravid adults were scored for each strain at each temperature.
We examined 18 type 1 lineages and all produced a single, mec-3-expressing cell (T.pppaa) that always resembled a PLM cell. In all of these lineages T.pppaa migrated to the ventral muscle and flattened against it (as do the PLM cells). By contrast, we observed three principal characteristics of Lineage 2 that suggested that it had been substantially but not completely transformed to a PVM-producing lineage (the QL lineage). First,the two type 2 lineages that we followed through the division of T.pppaa each generated three mec-3-expressing cells (T.ppap, T.pppaaa and T.pppaap). Five of these cells were PVM-like; only one (a T.pppaap cell) was PLM-like. The PVM-like cells neither migrated toward the ventral muscle nor took on a flattened appearance; these cells had rounded cell bodies and ventral processes similar to those of wild-type PVM cells. Second, the migration of T.ppa in the type 2 lineages resembled the migration of the equivalent cell (QL.a) in the PVM-producing lineage: these cells migrate posteriorly before dividing to produce an anterior daughter that dies and a posterior daughter (QL.ap) that survives. Third, in the QL lineage, QL.ap expresses unc-86; in the type 2 lineages, we conclude that the equivalent cell (T.ppap) also expresses unc-86 because we observed expression of mec-3, which is dependent on unc-86 expression(Way and Chalfie, 1989; Duggan et al., 1998). The transformation to a PVM-producing lineage, however, was incomplete: T.pppaa underwent an extra round of division relative to the equivalent cell in the QL lineage (QL.paa) and T.ppap expressed mec-3 in addition to unc-86 (expressed by QL.ap).
Because Lineage 2 produced more ectopic mec-3::gfp-expressing cells than Lineage 1, we hypothesized that animals with more ectopic mec-3::gfp-expressing cells should produce PVM-like cells. Conversely, animals with fewer ectopic mec-3::gfp-expressing cells should produce mostly PLM-like cells. At 25°C, in 51 sem-4 adults with four mec-3::gfp-expressing tail cells, 93% (189/204) of the cells resembled PLM, 3% (6/204) resembled PVM and 4% (9/204) could not be unambiguously classified as either PLM or PVM. The cells in this last category generally had rounded cell bodies, like PVM, but had processes that were difficult to see or did not extend ventrally and then anteriorly. In 52 animals with more than four mec-3::gfp-expressing tail cells, 67%(181/269) resembled PLM, 10% (28/269) resembled PVM and 22% (60/269) could not be unambiguously classified.
sem-4 restricts proliferation of touch-cell fate through repression of egl-5
We found that loss of sem-4 function caused transformation of the T.pp lineage into two distinct touch-cell lineages: a PLM-like lineage and a PVM-like lineage. The transformation of the posterior T.pp lineage into a mid-body PVM-like lineage is a transformation along the AP axis of the worm. AP transformations often result from defects in the expression or function of Hox genes that control AP patterning and several different aspects of cellular identity, including differentiation, growth and proliferation(Cillo et al., 2001; Veraksa et al., 2000).
The C. elegans Hox gene that controls patterning of posterior structures is the Abd-B homolog egl-5(Chisholm, 1991; Salser and Kenyon, 1994). In the tail, egl-5 is normally expressed in two pairs of neurons: the PLM touch neurons and the PVC interneurons(Ferreira et al., 1999). Although egl-5 animals are touch insensitive in the tail(Chalfie and Au, 1989; Chisholm, 1991), this defect has been attributed to abnormal development of the PVC interneurons(Chisholm, 1991), which are part of the neural circuit for touch sensitivity(Chalfie et al., 1985). We find that egl-5 is required for normal PLM development: PLM cells in egl-5 animals showed substantially reduced expression of mec-3 and abnormal morphologies(Table 1). We suggest that egl-5 activates mec-3 expression in PLM cells.
The findings that egl-5 is required for correct determination of PLM fate in wild-type animals and that the T.pp lineage is sometimes transformed to a PLM-like lineage in sem-4 animals suggest that loss of sem-4 function produces ectopic expression of egl-5 in the T.pp lineage. We observed ectopic expression in sem-4 animals of an egl-5::gfp reporter first in the T.pp neuroblast and then in T.ppa, T.ppp, T.pppa and T.pppaa (which ectopically expresses mec-3in sem-4 animals). Because we did not observe ectopic egl-5expression in T.pppap (which does not express mec-3 in sem-4animals), ectopic egl-5 expression in T.pppaa was probably not residual gfp expression from T.pppa. In wild-type worms, we observed expression of a sem-4::gfp reporter in T, T.a, T.p and all descendants of T.p. Thus, ectopic egl-5 expression in the T lineage in sem-4 animals began two cell divisions (in T.pp) after sem-4 expression would normally begin (in T). This ectopic T lineage expression was observed with two different egl-5::gfp reporters in the sem-4 background: one reporter contained a 12 kb region immediately upstream of the egl-5 translation start site; the second reporter contained only a 3 kb region immediately upstream of the start site(Fig. 3). We propose that sem-4 acts both in neuroblasts and neurons to restrict egl-5expression.
To investigate whether ectopically expressed egl-5 promoted the PLM fate among the abnormal touch-neuron-like cells in sem-4 mutants,we constructed sem-4; egl-5 animals containing the integrated mec-3::gfp reporter and examined the number and morphology of the ectopic touch neurons produced. Using a relatively strong, partial loss-of-function egl-5 allele, we found that loss of egl-5function decreased the number and significantly altered the morphology of the ectopic touch neurons (Table 1). In sem-4 mutants, most mec-3-expressing tail cells (84%) clearly resembled PLM cells. In sem-4; egl-5animals, the proportion of PLM-like cells decreased to 30%(Table 1).
We investigated whether mutation of the other C. elegans Hox genes, mab-5 and lin-39, could affect the number or morphology of the ectopic touch neurons in sem-4 animals. A partial loss-of-function allele of lin-39 and a putative null allele of mab-5 moderately decreased both the total number of ectopic touch neurons (data not shown) and the number of PLM-like cells: 70% (141/200) of mec-3-expressing tail cells in sem-4; mab-5 animals and 74%(147/200) in sem-4; lin-39 animals were PLM-like.
We conclude that sem-4 normally prevents the expression of egl-5 and possibly of mab-5 and lin-39 in the T lineage. Because egl-5 appears to activate mec-3 expression in normal PLM cells, we suggest that sem-4 restricts mec-3expression in the T lineage indirectly through restriction of egl-5. We found that egl-5 is required for correct expression of the PLM fate in wild-type PLM cells and for ectopic expression of the PLM fate in the abnormal touch cells in sem-4 animals. Negative regulation of egl-5 by sem-4 is therefore necessary to restrict inappropriate proliferation of the PLM fate in wild-type animals.
SEM-4 binds to a shared motif in the mec-3 and egl-5 promoters
Mutations in the mec-3 promoter at a 6 bp site (m3-1 in Fig. 4A), about 300 bp upstream of the translation start produced ectopic expression of a mec-3::lacZreporter in additional cells in the tail (Xue, 1993). The m3-1 region is one of several in the mec-3 promoter that is conserved between C. elegans and the related nematode C. briggsae. The ectopic mec-3::lacZ expression suggested that m3-1 might be a SEM-4 binding site. We found that, although purified SEM-4 protein did not shift a 24 bp fragment with m3-1 at its center (data not shown), it did shift a 96 bp fragment with m3-1 at its center (Fig. 4A). Four complexes were produced that were competed away by unlabeled specific competitor.
Mutation of m3-1 in the unlabeled competitor decreased, but did not eliminate, its ability to compete away complexes 2 and 4(Fig. 4A), and had no effect on its ability to compete away complexes 1 and 3. An oligonucleotide composed of four tandem copies of the 24 bp fragment with m3-1 at the center was not as effective a competitor as the 96 bp wild-type competitor (data not shown). These results suggest that the 96 bp fragment used as a probe contained more than one SEM-4 binding site. The 5′ end of this fragment contains a sequence identical at five out of six positions to m3-1 (m3-2 in Fig. 4A) that we tested for SEM-4 binding. Mutations at both m3-1 and m3-2 substantially reduced the ability of the unlabeled oligonucleotide to compete away complexes 1, 2 and 4(Fig. 4A). Complex 3 is probably formed by SEM-4 binding to a site other than m3-1 or m3-2.
We also looked for putative SEM-4 binding sites in the egl-5promoter. We found a cluster of three sites (e5-1, e5-2, e5-3), one identical to m3-1 (although on the opposite strand) and two identical to m3-2 (with one on the opposite strand), within an 81 bp region located about 900 bp upstream of the egl-5 translation start site(Fig. 4B). Similar clusters, of three sites within 125 bp, occur infrequently in the region we analysed. Specifically, in the 31 kb separating the translation starts of egl-5and mab-5 (these genes are transcribed in opposite directions, with their 5′ ends facing one another, separated by one predicted gene), only one other similar cluster occurs. This second cluster, of three sites within a 121 bp region, is located about 5 kb upstream of the mab-5translation start site. We reexamined the region of the mec-3promoter that contained the two sites we identified and found a third site(m3-3) about 60 bp upstream of the 5′ site(Fig. 4A). Thus, this region contains a cluster of three sites within 117 bp. A 206 bp region in the lin-39 promoter, about 10 kb upstream of the translation start site,also contains a cluster of three SEM-4 sites.
SEM-4 shifted a 100 bp fragment containing e5-1, e5-2 and e5-3(Fig. 4B). Two complexes were produced that were competed away by unlabeled competitor oligonucleotides. Mutation of all three sites in the wild-type competitor significantly decreased, but did not eliminate, its ability to compete away both complexes.
We identified a second region of the egl-5 promoter to which SEM-4 binds. About 3.5 kb upstream of the egl-5 start site, there is a 300 bp sequence (V6CRE) that mediates expression of egl-5 in the V6 hypodermal lineage. We found that a reporter containing V6CRE fused to gfp was not expressed in the T lineage. Occasionally, some faint expression was detected in a couple of T lineage cells. In a sem-4background, however, expression of Pv6cregfpincreased significantly (Fig. 4C). T lineage expression in wild-type worms of a reporter lacking the first 100 bp of Pv6cregfp(Pv6creδ100gfp) was similarly strong(Fig. 4C).
The region deleted in Pv6creδ100gfpcontains a consensus TRA-1 binding site (e5-T1) and two sites that contain five out of six bases of SEM-4 binding sites (e5-4 and e5-5). We tested binding of SEM-4 to a probe composed of the first 105 bp of V6CRE. SEM-4 shifted this probe (Fig. 4D),forming four complexes that were competed away by unlabeled competitor oligonucleotides. Mutation of e5-4 and e5-5 in the wild-type competitor produced a small but consistent reduction in competition(Fig. 4D). Mutation of e5-T1 had no effect on the ability of the competitor to compete away the complexes(Fig. 4D).
We conclude that SEM-4 binds to a shared motif in the mec-3 and egl-5 promoters and suggest that these interactions repress mec-3 and egl-5 expression in the T lineage. We propose that sem-4 restricts mec-3 expression both directly, by binding to the mec-3 promoter, and indirectly, through repression of egl-5.
Ectopic sem-4 represses mec-3 expression in vivo
The mec-3 gene is normally expressed in only ten cells: the six touch neurons, a pair of neurons in the head (the FLP cells) and a pair of mid-body neurons (the PVD cells). To test whether ectopic sem-4 could repress mec-3 in vivo, we expressed sem-4 in the touch cells under the control of the mec-7 promoter(Pmec-7sem-4). mec-7 encodes a β-tubulin that is expressed strongly in all six touch neurons during their terminal differentiation and less strongly in several other cells(Hamelin et al., 1992; Mitani et al., 1993; Savage et al., 1989). We analysed the effect of this ectopic sem-4 activity on mec-3expression by transforming Pmec-7sem-4 into worms containing an integrated mec-3::gfp reporter.
Transformation with Pmec-7sem-4 decreased the proportion of mec-3::gfp-fluorescent PLM cells from 100% (102/102) in the control line to 48±20% in three transformed lines (350 cells)(Fig. 5A). (In C. elegans, transformed DNAs form extrachromosomal arrays that are often not present in all cells.) We also tested the ability of several truncated versions of SEM-4 to decrease expression of the mec-3::gfp reporter(Fig. 5B-F). The N-terminal half of SEM-4, truncated after zinc finger 3, was a relatively effective repressor: transformation with this construct decreased the proportion of fluorescent PLM cells to 74±18% in three transformed lines (312 cells)(Fig. 5B). When SEM-4 was truncated after zinc finger 2, however, the resulting fragment did not decrease mec-3::gfp expression(Fig. 5C).
We found that the reduction in mec-3 expression produced by ectopically expressed sem-4 was sufficient to decrease touch-cell function significantly. Wild-type animals transformed with Pmec-7sem-4 were considerably less touch sensitive than animals transformed with Pmec-7sem-4(Q321ocher), which encodes a null allele of sem-4(Fig. 5G). The null allele is an important control because transformation with genes driven by touch-cell promoters sometimes produces partial touch insensitivity, as it did in our experiments (Fig. 5G).
We tested the effect on touch-cell function of ectopic expression of two partial loss-of-function sem-4 alleles(Fig. 5G). The n2654`neuronal' allele (containing a mis-sense mutation that changes one of the zinc-chelating histidines in zinc finger 2 to a tyrosine) exhibits more defects in neurons than in mesoderm (Basson and Horvitz, 1996); the n1378 `mesodermal' allele(containing a nonsense mutation, Q569ocher, which truncates zinc fingers 5, 6 and 7) exhibits more defects in mesoderm than in neurons(Basson and Horvitz, 1996). We found that transformation with Pmec-7sem-4(n1378), the Q569ocher `mesodermal' allele, produced touch insensitivity, although not as effectively as wild-type sem-4(Fig. 5G). By contrast,transformation with Pmec-7sem-4(n2654), the `neuronal'allele, did not produce touch insensitivity(Fig. 5G). We also found that the neuronal allele was more touch insensitive than the mesodermal allele(Fig. 6). These results are consistent with the hypothesis that sem-4(n2654) does not function properly in neurons.
We conclude from these data that ectopically expressed sem-4represses mec-3 in vivo. Two observations indicate that the zinc-finger pair 2 and 3 is crucial for this repression. First, a mutant version of SEM-4 truncated after zinc finger 3 was a relatively effective repressor, whereas a version truncated between zinc fingers 2 and 3 did not repress. Second, the mesodermal allele (truncated after zinc finger 4)produced touch insensitivity, whereas the neuronal allele, containing a defective zinc finger 2, did not.
Mutant versions of sem-4 display a gain-of-function phenotype
Production of three mutant versions of SEM-4 under the control of the mec-7 promoter produced mec-3::gfp expression in additional cells in the tail (Fig. 5C-E). Because mec-7 is expressed in cells other than the touch cells(Hamelin et al., 1992; Mitani et al., 1993), the Pmec-7 constructs probably expressed the truncated SEM-4 proteins in these additional cells. These gain-of-function mutant proteins contained C2H2 zinc finger 1 and various portions of zinc finger 2(Fig. 5C-E). These truncated SEM-4 fragments might have interfered either with endogenous SEM-4 or with other SEM-4 interacting partners in these additional cells.
We observed a gain-of-function phenotype in the sem-4(n2087)allele, which contains a nonsense mutation in zinc finger 2(Basson and Horvitz, 1996). We found that n2087 worms were more touch insensitive than n1971 worms, which contain an early splice-site mutation N-terminal to zinc finger 1 (Fig. 6)(Basson and Horvitz, 1996). The SEM-4 protein fragment encoded by n2087 is the same as that encoded by the Pmec-7sem-4 construct (shown in Fig. 5D). Worms transformed with this construct had the most ectopic mec-3::gfp-expressing cells(Fig. 5D). The fact that homozygous n2087 worms exhibit a gain-of-function phenotype suggests that truncated SEM-4 proteins interfere with the function of a protein other than SEM-4.
DISCUSSION
Function of sal genes is evolutionarily conserved
Several lines of evidence indicate that the function of sal genes has been evolutionarily conserved. The sal genes appear to function as cell fate determinants, regulators of Hox genes and AP patterning, and transcriptional repressors. The targets of and mechanism of regulation by sal genes appear to be conserved. The sal genes are involved in determination of precursor and differentiated cell fates. In Drosophila, sal is required in neuronal precursors and differentiated neurons to restrict neuronal fate to the proper cells. It restricts the fates of neuronal precursors during development of a sensory organ in the peripheral nervous system(Rusten et al., 2001). In the developing Drosophila eye, sal and salr(spalt-related) are required late in pupation for terminal differentiation of particular photoreceptor cells(Mollereau et al., 2001). Loss of sal and salr resulted in transformation of certain photoreceptor cells into other photoreceptor cells, as judged by rhabdomere morphology and opsin gene expression(Mollereau et al., 2001). The sal genes are probably involved in determining neural fates in mouse and human beings: human SALL1 is expressed in specific areas of the fetal brain(thalamus) and adult brain (corpus callosum and substantia nigra)(Kohlhase et al., 1996); mouse sall1 is expressed in particular embryonic neural tissues (tissues surrounding some of the ventricles and specific layers of the neural tube)(Buck et al., 2001).
Hox genes appear to be targets not only of sem-4 but also of other sal genes. Drosophila sal might negatively regulate Sex combs reduced (Scr) and other Drosophila Hox genes. Loss of sal function in Drosophila BX-C- embryos produced some limited ectopic expression of the Hox gene Scr(Casanova, 1989). Mutations in sal enhanced the phenotypes of Polycomb group (PcG) mutants. These genes are known to be negative regulators of Hox genes(Landecker et al., 1994). Loss of sal function affects AP patterning in Drosophila. Mutations in sal incompletely transform both head and tail structures into trunk-like structures; sal activity has been shown to promote head development (Jurgens,1988). Hox genes in mammals might also be targets of sal family genes. Patients with TBS, which is caused by mutations in SALL1,display characteristic features of syndromes associated with mutations in HOX genes (Powell and Michaelis,1999; Surka et al.,2001; Veraksa et al.,2000).
LIM homeobox genes, such as mec-3, might also be conserved targets of sal genes. The closest mammalian homolog to mec-3 is the human LIM homeobox gene Lhx5 (Zhao et al.,2000). Lhx5 and the human SALL1 gene appear to be expressed in different sets of cells in the developing thalamus, which constitutes a very small portion of the entire brain(Kohlhase et al., 1996; Nakagawa and O'Leary, 2001). SALL1 and Lhx5 are not expressed in most other regions of the fetal brain. Their expression in separate thalamic cells could indicate that SALL1 restricts Lhx5 expression in the thalamus.
The mechanism through which sem-4 negatively regulates its targets is probably conserved. SEM-4, SALL1 and mouse sall1 are transcriptional repressors. SALL1 and mouse sall1, fused to heterologous DNA binding domains,behaved as repressors in mammalian cell culture assays(Kiefer et al., 2002; Netzer et al., 2001). We suggest that these genes bind directly to regulatory regions of their targets.
Particular mutant versions of SALL1, like certain SEM-4 truncations, might act as gain-of-function proteins. TBS is an autosomal dominant disorder caused by mutations in SALL1 (Kohlhase,2000). No deletions of the entire gene or mutations that truncate the protein upstream of the first zinc finger have been detected. Thus, all 21 SALL1 mutant alleles encode truncated protein products that contain at least the first zinc finger. We found that truncations of SEM-4 containing the first zinc finger acted as gain-of-function proteins. We suggest that TBS could result, at least in part, from interference by these truncated proteins with wild-type SALL1 or other proteins.
SEM-4 negatively regulates genes at multiple levels of a developmental hierarchy
Very little is known about the pathways that lead from Hox proteins to determination of the fates of particular structures or individual cells. Recent evidence has suggested that Hox proteins can act independently on genes that function at different points along a particular developmental pathway(Veraksa et al., 2000). For example, the Drosophila Hox gene Ultrabithorax(Ubx) negatively regulates diverse genes throughout haltere development (Weatherbee et al.,1998). These genes encode signaling molecules, their immediate targets (including sal and salr) and proteins further downstream, including transcription factors. We propose that negative regulators of Hox genes, like the Hox genes themselves, also function at different levels in a given developmental hierarchy. We found that sem-4 negatively regulates both the Hox gene egl-5 in precursors and differentiating cells and the LIM homeobox gene mec-3in differentiating cells. Furthermore, we discovered that egl-5positively regulates mec-3 in normal PLM cells. Because SEM-4 binds to a shared motif in the promoters of both mec-3 and egl-5,we conclude that sem-4 negatively regulates each gene independently and also inhibits mec-3 expression through inhibition of egl-5.
No evidence for a global repression system for Hox genes in C. elegans has been reported. Polycomb group (PcG) and trithorax group (trxG) genes were originally identified in Drosophila as repressors and activators, respectively, of Hox gene expression (Brock and van Lohuizen,2001). PcG genes are now known to function together in a chromatin repressive complex (Francis and Kingston,2001). Although the roles of Hox and trxG genes in patterning are conserved in C. elegans, a role for PcG genes has not yet been established. We speculate that sem-4 might function as a member of a general repressive complex akin to the PcG complex.
Drosophila and mammalian studies have suggested that sal genes might function as PcG genes. Casanova(Casanova, 1989) found that sal mutations caused limited ectopic expression of the Hox genes Ubx and Scr, and Landecker et al.(Landecker et al., 1994) found that sal mutations enhanced mutations in the PcG genes polyhomeotic and Polycomb-like. Human SALL1 localizes to chromocenters in mammalian cells (Netzer et al., 2001) and mouse sall1 interacts with components of chromatin remodeling complexes (Kiefer et al., 2002). One additional speculation is that Drosophila sal might bind to a 138 bp silencing sequence in the Polycombresponse element in Abd-B, the egl-5 ortholog(Busturia et al., 2001). We have identified two sites that match the SEM-4 binding sequence in this Drosophila silencing element.
PcG genes might play a role in positive, in addition to negative,regulation of Hox genes (Brock and van Lohuizen, 2001): mutations in some PcG genes enhance trxG mutant phenotypes. sem-4 also appears to have a positive regulatory role in Hox gene expression in certain tissues: sem-4 might activate lin-39 in vulval lineages (Grant et al., 2000) and egl-5 in hypodermal lineages (Y. Teng et al., unpublished).
Acknowledgements
We thank M. Basson for sem-4 cDNA clones, J. McDermott for advice on gel shift assays, R. Horvitz for the sem-4(n2654) strain, W. Hanna-Rose for MH1346 and D. Xue for identifying m3-1 as a possible SEM-4 binding site. Some nematode strains used in this work were provided by the Caenorhabditis Genetics Center, which is funded by the NIH National Center for Research Resources (NCRR). This research was funded by NIH grants GM30997 to M.C. and GM39353 to S.W.E., and by postdoctoral fellowships from the American Cancer Society to A.S.T. and from the Brazilian National Council of Scientific and Technological Development (CNPq) to H.B.F. S.W.E. is the Siegfried Ullmann Professor of Molecular Genetics.