A novel C. elegans zinc finger transcription factor, lsy-2, required for the cell type-specific expression of the lsy-6 microRNA

Monod and Jacob proposed more than 40 years ago that the terminal differentiated state of a cell in a multicellular organism is determined and stabilized by feedback loops composed of gene regulatory factors(Monod and Jacob, 1961). The autoregulation of transcription factors has now indeed been recognized as a major feature of gene differentiation programs(Davidson, 2001; Edlund and Jessell, 1999).

We have recently identified a complex autoregulatory feedback loop that controls a cell-fate decision in the nervous system of the nematode Caenorhabditis elegans (Fig. 1A) (Chang et al.,2004; Chang et al.,2003; Hobert et al.,1999; Johnston and Hobert,2003; Johnston et al.,2005). The two morphologically bilaterally symmetric taste-receptor neurons ASEL and ASER develop from a common ground state to express a number of features that are specific for ASEL versus ASER(Fig. 1A). These left/right asymmetric features include the expression of several putative chemoreceptors of the GCY family. For example, in adult animals, the gcy-7 gene is exclusively expressed in ASEL, whereas the gcy-5 gene is exclusively expressed in ASER (Fig. 1). The expression of these two terminally differentiated states requires the activity of a set of gene regulatory factors that interact with one another in a bistable, double-negative feedback loop(Johnston et al., 2005). In this loop, ASEL-specific inducer genes, including the die-1 zinc finger transcription factor and the lsy-6 miRNA, activate other ASEL-specific inducer and effector genes, while repressing ASER-inducers and ASER-effectors in the ASEL neuron. By contrast, in the ASER neuron,ASER-inducer genes, including the cog-1 homeobox gene and the mir-273 miRNA, activate ASER-inducers and effectors, while repressing the ASEL-inducing genes die-1 and lsy-6(Fig. 1A).

What triggers the left/right asymmetric activity of the loop? Are there other, as yet unknown, factors that are components of the regulatory loop,and/or required for the loop to function appropriately? To address these questions, we have isolated and analyzed mutants in which the ASEL and/or ASER cell fates are not appropriately executed. We describe here one factor, lsy-2, that is required for the execution of the ASEL fate. lsy-2 codes for a novel C2H2 zinc finger transcription factor that is not an integral part of the regulatory loop. Rather, lsy-2constitutes a permissive factor that is present in both ASEL and ASER, but is required specifically in ASEL for the execution of the ASEL fate. Furthermore,we show that lsy-2 exerts its activity by regulating the expression of the lsy-6 miRNA.

miRNAs are abundant gene regulatory factors that contribute to the generation of cellular and morphological diversity in a developing organism(Ambros, 2004). Like other gene regulatory factors that contribute to organismal complexity, many, if not most, miRNAs are expressed in a spatially and temporally tightly controlled manner (e.g. Wienholds et al.,2005), yet the mechanisms that control miRNA gene expression are only beginning to be elucidated. Our identification of lsy-2 as a regulator of the lsy-6 miRNA therefore contributes to our understanding of the control of miRNA expression, and of the complex regulatory networks necessary for terminal cell-fate specification.

Previously described strains and transgene: N2 Bristol wild type(Brenner, 1974), CB4856 Hawaiian wild-type isolate (Hodgkin and Doniach, 1997), NL2099 rrf-3(pk1426)II(Simmer et al., 2002), otIs3=Is[gcy-7^prom::gfp; lin-15 (+)],ntIs1=Is[gcy-5^prom::gfp; lin-15 (+)],otIs114=Is[lim-6^prom::gfp; rol-6(d)](Chang et al., 2003), otIs151=Is[ceh-36^prom::rfp; rol-6(d)](Johnston and Hobert, 2003), otIs160=Is[lsy-6^prom::gfp; rol-6(d)](Johnston et al., 2005), syIs63=Is[cog-1::gfp] (Palmer et al., 2002), otEx1759=Ex[ceh-36^prom::gfp::die-1^3′UTR;rol-6(d)] (Chang et al.,2004).

New transgenes: otEx2044=Ex[ceh-36^prom::lsy-6;elt-2::gfp]; otEx1322 to otEx1324 - three independent lines of Ex[lsy-2^prom::gfp; rol-6(d)]; otEx1325to otEx1328, otEx1777 to otEx1780, and otEx1790 -nine independent lines of Ex[lsy-2::gfp; rol-6(d)]; otEx1945to otEx1947, and otEx1954 - four independent lines of Ex[ceh-36^prom::lsy-2; elt-2::gfp].

lsy-2^prom::gfp, lsy-2::gfp and ceh-36^prom::lsy-2 were constructed using a PCR fusion approach (Hobert, 2002). The gfp constructs are shown in Fig. 4A. lsy-2^prom::gfp was generated by fusing 3 kb upstream (up to the preceding gene) of the lsy-2 gene to the gfp-coding region. The construct was injected at 50 ng/μl together with rol-6(d) as an injection marker (100 ng/μl). lsy-2::gfp was generated by fusing 3 kb of the upstream region and the all of the exons and introns of lsy-2 to the gfp-coding region and the unc-54 3′UTR. The construct was injected at 5 ng/μl together with rol-6(d) as an injection marker (100 ng/μl). ceh-36^prom::lsy-2 was generated by fusing the 5 kb ceh-36 promoter (Chang et al.,2003) to the lsy-2 cDNA, including the coding region and the 3′ UTR. The construct was injected at 5 ng/μl together with rol-6(d) as an injection marker (100 ng/μl).

Primer sequences (5′ to 3′)

lsy-2^prom::gfp

Primer A, GTTGAATCCGACTTCTTCAGGG;

Primer A^*, GTTTCTAGCAATCTGGTTGTTG;

Primer B, CTAGAGTCGACCTGCAGGCCATGACAAAATTTGCCTCAGAC;

Primer C, AGCTTGCATGCCTGCAGGTCGACT;

Primer D, AAGGGCCCGTACGGCCGACTA;

Primer D^*, GGAAACAGTTATGTTTGGTATATTGGG.

lsy-2::gfp

Primer A, GTTGAATCCGACTTCTTCAGGG;

Primer A^*, GTTTCTAGCAATCTGGTTGTTG;

Primer B, CTAGAGTCGACCTGCAGGCAATCAACTGTGGTTCCATCATC;

Primer C, D and D^*, as above.

ceh-36^prom::lsy-2

Primer A, CAAAAATGAGGCTACCAAG;

Primer A^*, CAAAGTAGAGCACTGAGGGTG.

Primer B, CATTTCTTCTGGTTAGCATTTGTGCATGCGGGGGCAGG;

Primer C, CCTGCCCCCGCATGCACAAATGCTAACCAGAAGAAATG;

Primer D, GACTGCAAATGAGACAGTC;

Primer D^*, GACGAAGACGACTCCATAG;

Primers were used as in PCR fusion reactions as previously described(Hobert, 2002).

RNAi was performed using a bacterial feeding protocol(Simmer et al., 2003). NGM agar plates containing 6 mM IPTG and 100 μg/ml ampicillin were seeded with bacteria expressing dsRNA, kindly provided by the Greenwald Laboratory. otIs114; rrf-3 hermaphrodites at the L3/L4 stage were placed onto these plates and grown at 15°C. Adults were then transferred onto freshly seeded plates at 20°C, and their F1 progeny were scored for asymmetry and sterility defects. RNAi was performed in a rrf-3(pk1426) background because of its increased sensitivity to RNAi(Simmer et al., 2003).

Reporter transgenes were crossed into the respective mutant backgrounds. All animals were scored with an Axioplan 2 microscope. When needed, ASE neurons were unambiguously identified through the use of a transgene(otIs151) that expresses DsRedz bilaterally in ASEL and ASER, as well as in AWCL and AWCR, under control of the ceh-36 promoter(Chang et al., 2003).

In order to better understand the mechanisms underlying the ASE left/right neuronal fate decision (Fig. 1A), we undertook a genetic screen to isolate mutants in which expression of the ASEL-specific gcy-7 gene is disrupted. We exposed transgenic animals that contain the chromosomally integrated gcy-7^prom::gfp transgene otIs3 to the mutagen ethylmethanesulphonate (EMS) (Brenner,1974; Chang et al.,2003). F1 progeny of mutagenized L4 animals were singled out onto individual agar plates and the F2 progeny were screened for lateral symmetry(lsy) defects under a stereomicroscope equipped with a fluorescent light source. Screening through 12,200 haploid genomes, we identified a total of 10 mutant alleles that display a `two ASER' phenotype in which the ASEL neuronal fate has switched to an ASER neuronal fate (see Fig. 1B for an example). One mutant, ot71, defines the lsy-6 miRNA locus(Johnston and Hobert, 2003);three mutants, ot69, ot74 and ot78, lie within the lin-49 transcription factor locus(Chang et al., 2003); one mutant, ot79, defines the ceh-36 homeobox locus(Chang et al., 2003); and five mutant alleles (ot64, ot65, ot67, ot72 and ot77) define a novel, single genetic locus on the X chromosome. We focus here on the characterization of this novel locus, which we termed lsy-2(pronounced `lousy two'). In lsy-2 mutants, ASEL no longer expresses the ASEL-specific terminal fate markers gcy-6, gcy-7, lim-6and flp-4, but rather expresses the normally ASER-specific terminal fate marker gcy-5 (Fig. 1B, Table 1, data not shown). All lsy-2 alleles are fully recessive and display a similar range of pleiotropies, including protruding vulva, abnormally migrating gonad arms and completely penetrant sterility.

Using single nucleotide polymorphisms and three-factor mapping(Brenner, 1974; Wicks et al., 2001), we mapped lsy-2 to a ∼80 kb region on the X chromosome(Fig. 2). Because all previously cloned lsy genes coded for gene regulatory factors(Chang et al., 2004; Chang et al., 2003; Hobert et al., 1999; Johnston and Hobert, 2003)(R.J.J. and O.H., unpublished), we examined this genomic region for the presence of predicted gene regulatory factors. We noted one predicted protein,the F49H12.1 protein (Fig. 2),which contains a nuclear localization sequence and five zinc finger domains of the C2H2 type, characterized by a specific composition and spacing of zinc chelating cysteine and histidine residues (consensus C2H2 structure:CX_2-4CX₁₂HX_2-6H)(Iuchi, 2001). Although C2H2 zinc finger proteins can also bind RNA, the vast majority of these proteins are DNA-binding transcription factors(Iuchi, 2001).

The structure of the F49H12.1 gene is confirmed by multiple,full-length EST clones(www.wormbase.org). We sequenced this gene in all available lsy-2 mutant strains and found mutations in it in each of our five lsy-2 strains. These mutations include two premature stop codons, two splice-site mutations and one missense mutation of an invariant cysteine residue in the second zinc finger motif (Fig. 3A; Table 1). The identical early stop alleles lsy-2(ot64) and lsy-2(ot65) were isolated from the same mutagenized population of animals, and possibly arose from one mutagenic event. A second pair of identical alleles, lsy-2(ot67) and lsy-2(ot72), were isolated from independently mutagenized populations of animals. Because of the early stop codon and its fully penetrant lsy and sterility phenotype, we consider the lsy-2(ot64)allele to be a putative null allele, and we used this allele for all of our ensuing genetic analyses.

To further confirm the molecular identity of lsy-2, we performed RNAi experiments and observed both the lsy phenotype and the sterility associated with all lsy-2 mutant alleles(Table 2). Lastly,transformation rescue experiments, described in more detail below, further verified the molecular identity of lsy-2(Table 3).

BLAST searches reveal an apparent lsy-2 ortholog in the genomes of the related nematodes C. briggsae(Fig. 3A) and C. remanei (data not shown). In addition, we found a closely related lsy-2 paralog in C. elegans, F52F12.4, which we have named lsl-1 (pronounced `lizzle one') for lsy-2 like(Fig. 3A). lsl-1 is also highly conserved in C. briggsae(Fig. 3A). Among all of these nematode orthologs and paralogs, the individual C2H2 zinc fingers in the triple zinc finger cluster are significantly more conserved than the fingers in the double zinc finger cluster (Fig. 3A).

Database searches with the lsy-2 sequence using either the full-length protein or individual zinc fingers revealed no clear orthologs of lsy-2 in non-nematode species. However, the existence of a tightly clustered, triple C2H2 zinc finger motif is reminiscent of the DNA-binding domain found in the SP1/KLF proteins, a family of transcription factors with wide-spread functions in growth and development(Kaczynski et al., 2003)(Fig. 3B). The importance of the triple zinc finger motif of LSY-2 is highlighted by the ot77allele, which contains a missense mutation in the second zinc finger that is predicted to affect DNA binding (Fig. 3A,B). This mutant causes an almost complete loss of gene function(Table 1). Apart from the zinc fingers themselves, the linker regions that connect the individual zinc fingers are conserved between the triple zinc finger motifs of LSY-2 and the SP1/KLF family (Fig. 3C). The linker region is important for appropriately spacing the contacts that the adjacent zinc fingers make with DNA, and this region is also engaged in direct contacts with the phosphate backbone via a conserved lysine residue(Iuchi, 2001; Pavletich and Pabo, 1991). Although many multiple-zinc finger proteins display a highly variable length of the linker region (Iuchi,2001), the linker regions of LSY-2 and the SP1/KLF factors have a similar length and share several amino acids. However, searching the C. elegans genome sequence with human SP1 and Drosophila Sp1 proteins, or other KLF proteins, reveals several worm triple zinc finger proteins that have a higher similarity to SP1/KLF proteins than LSY-2 does(Oates et al., 2001) (data not shown). Taken together, our sequence analysis indicates that LSY-2 is likely to be a DNA-binding protein that is not broadly conserved, but is distantly related to the SP1/KLF-family of transcription factors.

In order to analyze the cellular focus of lsy-2 activity, we generated a lsy-2::gfp reporter gene fusion in which the coding region for green fluorescent protein was fused to the lsy-2 locus,including the full-coding region and its complete 5′ region to the next gene (Fig. 4A). Nine independent transgenic lines carrying the lsy-2::gfp extrachromosomal array displayed gfp expression in all cells and tissue types of the worm, including ASEL and ASER (Fig. 4B,C). Expression is first observed at the 50-cell stage, and persists throughout embryogenesis, larval stages and adulthood(Fig. 4B). Notably, LSY-2::GFP was localized to moving punctae in the nuclei(Fig. 4D), which is consistent with a role for lsy-2 in regulating gene expression (see Discussion). This expression pattern relates to endogenous gene function, as lsy-2::gfp transgenes can rescue the sterility and lsyphenotype of lsy-2 null mutants(Table 3). A similar ubiquitous expression pattern is observed in animals expressing gfp under the control of only the 5′ upstream region of the lsy-2 locus(Fig. 4B).

As lsy-2 is widely expressed, we tested whether lsy-2 is required specifically in the mature ASE neurons to determine the stable ASEL cell fate. We fused the lsy-2 cDNA to the transcriptional regulatory region of the ceh-36 gene ('ceh-36^prom'), which is active in two bilateral pairs of head sensory neurons, AWCL/R and ASEL/R(Chang et al., 2003; Lanjuin et al., 2003). In lsy-2(ot64) null mutant animals carrying ceh-36^prom::lsy-2, ASE laterality was at least partially restored (Table 3). As microsurgical ablation of ASER or genetic ablation of AWCL/R has no effect on ASEL development (data not shown), we can conclude that it is the resupplied lsy-2 gene activity in ASEL, rather than ASER, AWCL or AWCR, that is responsible for the rescue of the mutant phenotype. The ceh-36promoter is activated around the time that the ASE neurons are generated and therefore the rescue experiments also indicate that lsy-2 is likely to function postmitotically. Other members of the regulatory loop controlling ASEL/R fate also function postmitotically(Chang et al., 2004; Chang et al., 2003; Hobert et al., 1999; Johnston and Hobert,2003).

Because the provision of extra copies of the lsy-2 gene in ASER with the ceh-36^prom::lsy-2 transgene does not induce the ASEL fate in ASER, we can furthermore conclude that, even upon overexpression, lsy-2 is not sufficient to induce ASEL fate. Together with the normal expression of lsy-2 in both ASEL and ASER, we can conclude that lsy-2 acts permissively rather than instructively to induce ASEL fate in ASEL.

A bistable feedback loop consisting of the miRNAs lsy-6 and mir-273, and their respective target transcription factors cog-1 and die-1, is required for ASE laterality(Fig. 1A)(Johnston et al., 2005). Because the genetic removal of lsy-2 causes a conversion of ASEL to the ASER stable state, we wanted to test what role lsy-2 plays in the feedback loop. We first examined the effect of lsy-2 on the ASEL-specific expression of the miRNA lsy-6. We found a complete loss of lsy-6 expression in lsy-2 null mutant animals(Fig. 5A). lsy-6represses the expression of the cog-1 transcription factor in ASEL and, as a result, expression of cog-1 is biased towards the right cell (Johnston and Hobert,2003). As would be expected from a loss of endogenous lsy-6 miRNA expression, cog-1 expression is de-repressed in the ASEL neuron of lsy-2 null mutant animals(Fig. 5B). The de-repression of cog-1 in ASEL in lsy-2 null mutant is functionally relevant and provides an explanation for the `two ASER' phenotype of lsy-2mutants, as a reduction of cog-1 gene activity in a lsy-2null mutant background significantly represses the adoption of the ASER fate in ASEL (Table 4).

To further characterize the role of lsy-2 in the bistable lsy-6/cog-1/mir-273/die-1 feedback loop, we examined the effect of lsy-2 on the die-1 sensor gene ceh-36^prom::gfp::die-1^3′UTR. This sensor gene provides a means to observe the miRNA-mediated repression of die-1 in ASER, conferred by mir-273 and other, as yet unknown, miRNAs (Chang et al.,2004) (D. Didiano and O.H., unpublished). In lsy-2 null mutant animals carrying the die-1 sensor gene, differential expression of GFP is disrupted (Fig. 5C). Taken together, these results indicate that lsy-2 is required for the proper execution of the ASEL cell fate that is controlled by the previously described bistable feedback loop.

Because components of the regulatory loop regulate the expression of one another (Fig. 1A), the effects of lsy-2 on the expression of individual loop components could be caused either by lsy-2 affecting the expression of only a single component of the loop, or, alternatively, by lsy-2 independently affecting the expression of several components of the loop. We previously reported that die-1 is the likely output regulator of the loop that directly or indirectly controls the expression of effector genes of the loop(i.e. of the terminal differentiation markers)(Johnston et al., 2005). This makes lsy-6 the regulatory gene that is most distal from the effector genes (Fig. 1A). We first asked whether ectopic expression of lsy-6 miRNA, which induces the ASEL fate in both ASE neurons (Johnston and Hobert, 2003), requires the activity of lsy-2. If lsy-2 is required for the expression of a loop component that acts downstream of lsy-6 (i.e. anywhere between lsy-6 and the effector genes), ectopic lsy-6 would not be able to exert its ASEL-inducing effect in lsy-2 null mutants. If the effect of ectopic lsy-6 expression is unaltered in lsy-2 null mutants, lsy-2 would act upstream of lsy-6. If lsy-2 acts on multiple components in the loop, intermediate effects might be expected. We find that the ASEL-inducing activity of ectopic lsy-6 expression is virtually unaffected in lsy-2 null mutants, as measured with two distinct cell fate markers (Table 4).

Another approach to examine the epistatic relationship of lsy-2and lsy-6 is to determine whether the loss of ASEL fate observed in lsy-2 null mutants can be rescued by the expression of lsy-6under the control of a heterologous, lsy-2-independent promoter in ASEL. Indeed, in most lsy-2 null mutant animals examined,heterologously expressed lsy-6 is able to restore the ASEL fate, as measured with two distinct cell fate markers(Table 4). lsy-6 is in fact as efficient at restoring ASEL fate in ASEL in lsy-2 mutants as it is at inducing ASEL fate when ectopically expressed in ASER (∼80%; Table 4). The most parsimonious explanation of these observations is that lsy-2 acts upstream of lsy-6 to regulate ASE asymmetry(Fig. 5D).

lsy-6 is expressed in several neuron types besides ASEL, including labial sensory neurons and the PVQ ventral cord interneurons(Johnston and Hobert, 2003). Although lsy-2 is expressed in all of these neuron types, lsy-6^prom::gfp expression is lost only in the ASEL neuron of lsy-2 null mutants, and not in other head or tail neurons (data not shown). Like many transcription factor interactions with their target genes (e.g. Altun-Gultekin et al.,2001; Tsalik et al.,2003), the genetic interaction of lsy-2 and lsy-6 is therefore cell-type specific.

We have described a novel C2H2 zinc finger protein that plays a permissive role in controlling the execution of the ASEL cell fate. Although lsy-2 is ubiquitously expressed, its spatially and temporally restricted function in the ASEL/R cell fate decision indicates that lsy-2 is a specific regulator of a select number of target genes in different cell types, rather than a global facilitator of gene expression. How does lsy-2 regulate gene expression? The subnuclear localization of lsy-2 to moving speckles suggests a role in splicing or transcriptional regulation. Nuclear speckles were initially characterized as`supply houses' of splicing factors to sites of active transcription(Lamond and Spector, 2003). However, there is a growing body of evidence indicating that subnuclear structures may also contain specific core components and regulators of transcription. For example, nuclear speckle localization has been reported for transcription factors such as ALL-1/Trithorax, steroid receptor co-activator(SRC-1), steroidogenic factor 1 (SF-1) and Pnn/DRS(Alpatov et al., 2004; Amazit et al., 2003; Chen et al., 2004; Zeng et al., 1997). Although we cannot exclude a role for lsy-2 as a splicing factor, the now well-established localization of at least some transcription factors to speckles, together with the triple C2H2 zinc finger motif architecture of LSY-2 that is shared with bona fide DNA-binding transcription factors, strongly suggests a role of LSY-2 in transcriptional regulation.

A recent genome sequence analysis has revealed the presence of a conserved 8 bp motif located ∼200 bp upstream of virtually all known nematode miRNA(Ohler et al., 2004). This observation suggested that a ubiquitously expressed transcription factor, such as lsy-2, might act through this motif to enable or facilitate global miRNA gene expression. However, we consider this possibility to be unlikely,as deletion of this motif does not affect the expression of lsy-6(data not shown). Loss of lsy-2 also does not affect lsy-6expression in cells other than ASEL. A role for lsy-2 as a permissive regulator of miRNA expression is also ruled out by the observation that none of the morphological defects that are associated with the loss of the lin-4 or let-7 miRNAs [lin-4: elongated, vulvaless(Ambros and Horvitz, 1984); let-7: lethal because of bursting vulva(Reinhart et al., 2000)] can be observed in lsy-2 null mutants (data not shown). LSY-2 is therefore unlikely to be a general activator of miRNA gene expression.

In genetic terms, lsy-2 exerts its function by regulating the cell-type specificity of lsy-6 miRNA gene expression, thereby making lsy-2 one of the very few factors known to be involved in the spatial control miRNA gene regulation. Although lsy-2 is expressed throughout the nervous system, it nevertheless regulates lsy-6 in only one of the three neuron classes that express lsy-6. Such cell-type specificity in regulatory interactions is a common theme in transcriptional regulation. One of the many prominent examples in C. elegans is the LIM homeobox genes that regulate specific target genes in some cell types but not in others, even though they are co-expressed(Altun-Gultekin et al., 2001; Tsalik et al., 2003).

Our genetic screens have not only retrieved instructive, left/right asymmetrically expressed factors that are both required and sufficient to induce either the ASEL or ASER fate, but also permissive factors that control the ASEL/ASER fate decision (Chang et al.,2004; Chang et al.,2003; Johnston and Hobert,2003). Instructive factors, such as lsy-6, cog-1, mir-273(in conjunction with other miRNAs; Dominic Didiano and O.H., unpublished) and die-1, are expressed asymmetrically in either ASEL or ASER and are not only required to induce either the ASEL or ASER fate, but are also sufficient to do so if misexpressed in the opposing cell. By contrast,permissive factors are expressed in both ASEL and ASER, and are therefore only required but not sufficient to induce the respective fate. As these permissive factors are not expressed in a left/right asymmetric manner, they are not intrinsic components of the bistable feedback loop shown in Fig. 1A and Fig. 5D, but are permissively required to confer the activity of left/right asymmetric factors in the loop. These permissive factors include the Groucho-like co-repressor UNC-37, the PHD/bromodomain protein LIN-49, the OTX-type homeodomain protein CEH-36(Chang et al., 2003), and, as we describe here, LSY-2. unc-37 is required for the execution of the ASER fate, whereas lin-49, ceh-36 and lsy-2 are required for the ASEL fate. What are the specific features of these permissively required factors and how may their cell-type specific activities be explained?

Based on the genetic interactions of cog-1 and unc-37/Groucho, and the presence of a Groucho-binding EH1 domain in the COG-1 protein, the cell-type specific activity of the ubiquitously expressed UNC-37 protein can be explained by its physical association with the ASER-specific COG-1 homeodomain protein(Chang et al., 2003). In a conceptually similar manner, the cell type-specific activity of UNC-37 in regulating VA motoneuron specification can be explained by its association with the VA motoneuron-specific homeodomain protein UNC-4(Pflugrad et al., 1997).

This Otx-type homeobox gene is only expressed in two pairs of head neurons(Chang et al., 2003; Lanjuin et al., 2003), but its bilateral expression in ASEL and ASER still classifies the gene as a permissive factor required for ASEL cell fate. One potential model that may explain the cell-type specificity of ceh-36 proposes that ceh-36 activity in ASER is competed for by the ASER-inducing cog-1 gene, which is exclusively expressed in ASER(Chang et al., 2003).

This gene codes for a PHD/bromodomain protein that is required for the induction of the ASEL fate, as well as for the regulation of a variety of other cell fate decisions (Chamberlin and Thomas, 2000). We have previously shown that a complete loss of lin-49 function in lin-49 null mutant animals can be partially suppressed by lowering cog-1 gene activity(Chang et al., 2003). This effect would support a role of lin-49 upstream of cog-1. Consistent with this notion, lin-49 mutants display a failure of downregulation of cog-1 in ASEL in adult animals, a concomitant loss of lsy-6 expression in ASEL, and, as a likely consequence of ectopic cog-1 expression, a misregulation of the die-1 3′UTR(data not shown). These are phenotypes that closely resemble those observed in lsy-2 mutants, and it is conceivable that both proteins act together to regulate the expression of lsy-6 and other ASEL-inducing components of the feedback loop.

Of the above-mentioned cases, the case of unc-37 provides the clearest example for how a bilaterally expressed, permissively acting factor can confer cell-specific activity through physical association with cell-specific, instructive regulatory proteins. By analogy to this case, a good candidate to confer functional specificity to lsy-2 (and also lin-49, which acts in a genetically similar manner to lsy-2)is the ASEL-inducing zinc finger transcription factor die-1. Like lsy-2, die-1 is required for the expression of the miRNA lsy-6 (Chang et al.,2004). Genetically, the key difference between lsy-2 and die-1 is that die-1 is left/right asymmetrically expressed and can act instructively; that is, it can induce ASEL fate if it is misexpressed in ASER. A common architectural feature of several well-characterized cis-regulatory regions is the presence of binding sites for both cell-type specifically expressed factors and broadly expressed transcription factors, such as, for example, SP1 (to which LSY-2 is distantly related) (e.g. Falvo et al.,2000; Xiao et al.,1987). Broadly expressed transcription factors appear to be required for baseline promoter activity, and functionally synergize with factors that provide spatiotemporal specificity. A similar scenario may hold true for LSY-2, which may synergize with DIE-1 to efficiently activate lsy-6 expression. An ongoing analysis of the lsy-6 promoter may identify cis-regulatory elements that could be directly targeted by DIE-1 and LSY-2 proteins.

Factors that control the ASEL versus ASER cell fate decision and their regulatory interactions are summarized in Fig. 5D. Ongoing genetic analysis in our laboratory has uncovered even more factors that are involved in this cell fate decision. The gene regulatory network controlling the diversification of the ASE neurons therefore appears to be unusually complex at first sight. However, only if systematic and extensive genetic approaches similar to those that we have taken with the ASE neurons are applied to other neuronal fate decisions, can one assess whether such complexity in regulatory networks may be the rule or the exception.

We thank Q. Chen for expert technical assistance, Y. Kohara for providing EST clones, the Greenwald Laboratory for dsRNA clones, and members of the Hobert Laboratory for discussion and comments on the manuscript. This work was funded by an NSF pre-doctoral fellowship to R.J.J. and by NIH R01 NS050266-01 and NS39996-05 to O.H.