Distinct C. elegans HLH-8/Twist-containing dimers function in the mesoderm

Abstract

Background

The C. elegans basic helix-loop-helix (bHLH) factor HLH-8, the single Twist ortholog in the nematode genome, plays important roles in mesoderm development, including M lineage patterning and differentiation of vulval and enteric muscles. HLH-8 cooperates with HLH-2, the bHLH E/Daughterless ortholog, to regulate downstream target genes, but it is not known whether HLH-2 is an obligate partner for all HLH-8 functions.

Results

Using hlh-2 loss-of-function alleles and RNAi, we discovered that HLH-2 is required in the vulval muscles but not in M patterning or enteric muscle development. Additionally, we found that expressing tethered HLH-8/HLH-8 dimers in hlh-8 null animals rescued M patterning and enteric but not vulval muscle development.

Conclusions

These results support a model whereby HLH-8/HLH-8 homodimers function in M lineage patterning and enteric muscles and HLH-8/HLH-2 heterodimers function in the M-derived vulval muscles. Interestingly, the different dimers function in the same M lineage cells and the switch in dimer function coincides with vulval muscle differentiation. The use of distinct Twist dimers is evolutionarily conserved, and C. elegans provides a paradigm for future dissection of differential promoter regulation by these dimers at a single cell resolution.

Introduction

Basic helix-loop-helix (bHLH) transcription factors are important regulators of gene expression. These proteins function by forming homodimers or heterodimers with other bHLH proteins. The bHLH motif consists of two domains that are highly conserved throughout evolution: a helix-loop-helix domain and a basic domain (Morgenstern and Atchley, 1999). The helix-loop-helix domain promotes dimerization, and the two basic domains in a dimer bind to the DNA of a target gene via a consensus CANNTG sequence, called an E box (reviewed in Murre et al., 1989a, and Rose and Malcolm, 1997). bHLH proteins control multiple processes during the development of many organisms, including yeast, nematodes, fruit flies, and mammals (Massari and Murre, 2000). A subset of bHLH proteins, referred to collectively as E proteins, include E12, E47, E2-2, HEB in mammals, and Daughterless (DA) in fruit flies. The E proteins are expressed in many tissues throughout development. Because of their broad tissue distribution, E proteins are involved in many cell type-specific functions by dimerizing with tissue-specific bHLH proteins (Murre et al., 1989)

In the nematode, Caenorhabditis elegans, the only ortholog of the E proteins is HLH-2 (Krause et al., 1997). HLH-2 is expressed ubiquitously during embryogenesis and its expression becomes restricted to a limited number of cells, including neurons and enteric muscles in first stage (L1) larvae, and the distal tip cells and vulval muscle cells from the L3 to adult stage (Krause et al., 1997). As expected of an E protein ortholog, HLH-2 partners with other tissue-specific bHLH proteins and plays many roles in C. elegans development. HLH-2 functions with LIN-32, which is a bHLH transcription factor of the atonal family, to regulate neuronal development in the male tail (Portman and Emmons, 2000). During somatic gonadogenesis, HLH-2 has several distinct functions: specification and differentiation of the anchor cell (AC) and the distal tip cells (DTCs) (Karp and Greenwald, 2004). However, a tissue-specific partner for HLH-2 in the AC and the DTCs has not yet been identified. HLH-2 is involved in the regulation of programmed cell death during embryogenesis by partnering with an Achaete-Scute bHLH homolog, HLH-3, to regulate a cell death activator gene, egl-1 (Thellman et al., 2003). HLH-2 is also predicted to cooperate with a mesoderm-specific bHLH protein, HLH-8 (Twist homolog), to activate a number of genes (Wang et al., 2006). The function of HLH-2 and its cooperation with HLH-8 in the nongonadal C. elegans mesoderm is the focus of this study.

In many organisms including fruit flies (Thisse et al., 1988), mice (Fuchtbauer EM, 1995, Quertermous et al., 1995) and humans (Wang et al., 1997), the transcription factor Twist is a key regulator in the specification of mesoderm. In C. elegans, HLH-8 is important for the proper patterning and differentiation of myoblast cells derived from a single M cell born during embryogenesis (Harfe et al., 1998, Corsi et al., 2000). This single cell gives rise to all of the post-embryonic nongonadal mesoderm, including the vulval and uterine muscles (collectively referred to as sex muscles) necessary for egg laying. HLH-8 plays a role in the proper formation of these muscles. HLH-8 is also important for the proper formation of the enteric muscles, which are born during embryogenesis and are required for defecation. Consequently, hlh-8 null mutants are constipated (Con) and egg-laying defective (Egl) (Corsi et al., 2000). Several genes have been identified as targets of HLH-8 (Harfe et al., 1998, Corsi et al., 2000, Wang et al., 2006). These targets genes are expressed in a differential pattern throughout the mesoderm, suggesting that HLH-8 regulates subsets of target genes uniquely (Zhao et al., 2007).

Because both E and Twist proteins are playing important roles in metazoan development, we wanted to understand how the orthologs of these proteins were functioning in C. elegans development. Current evidence supports the prediction that HLH-2 forms heterodimers with HLH-8 (Wang et al., 2006; Zhao et al., 2007; Grove et al., 2009). First, HLH-8 and HLH-2 can bind as heterodimers in vitro to DNA containing an E box (Harfe et al., 1998b; Corsi et al., 2002). Second, HLH-8 and HLH-2 are co-expressed in cells such as the sex muscle precursor cells that develop into vulval muscles, and in the enteric muscles (Krause et al., 1997; Harfe et al., 1998). Finally, when both proteins are overexpressed together in wild-type animals, they induce stronger ectopic expression of CeTwist target genes than either overexpressed factor alone (Harfe et al., 1998; Kostas and Fire, 2002; Wang et al., 2006). Recent evidence has also shown in yeast two hybrid assays that the only bHLH partner detected for HLH-8 is HLH-2 (Grove et al., 2009), emphasizing the importance of understanding how the two partners are cooperating in vivo.

We investigated how HLH-2 is involved in regulating target genes of HLH-8 by studying the expression pattern of the two most extensively characterized target genes, arg-1 and egl-15, in hlh-2 mutants and RNAi. The arg-1 gene encodes a predicted ligand of the Delta/Serrate/LAG-2 (DSL) family of transmembrane proteins and is thought to be involved in cell signaling (Mello et al., 1994; Fitzgerald and Greenwald, 1995). The arg-1 gene is one of the HLH-8 targets with the broadest tissue distribution (Wang et al., 2006; Zhao et al., 2007). This gene is expressed in the vulval muscles of the adult hermaphrodite and in the enteric muscles and the head mesodermal cell (hmc) beginning in embryos and continuing through adulthood. The hmc is a mesodermally-derived cell whose function is not known. The second target gene, egl-15, encodes a fibroblast growth factor receptor (FGFR) homolog and plays a role in the migration of the sex myoblast cells that proliferate to become sex muscles (Stern and Horvitz, 1991). The egl-15 gene is expressed in the sex myoblast descendants at the L4 stage of larval development and in the vulval muscles of adults that develop from those cells (Huang and Stern, 2004). Both targets, arg-1 and egl-15, are not activated in hlh-8 (nr2061) null mutants (Corsi et al., 2000, Wang et al., 2006). We evaluated expression of these target genes in animals that had reduced hlh-2 expression to investigate the role of HLH-2 in mesoderm development.

We observed that HLH-2 functions as an activator: to turn on one HLH-8 target gene, arg-1, in the vulval muscles of adult animals and the hmc and to maintain expression of egl-15 in the vulval muscles of adult hermaphrodites. Interestingly, we found that HLH-8 works independently of HLH-2 to regulate patterning of the early M lineage and the proper formation of the enteric muscles, suggesting that these functions may be controlled by HLH-8 homodimers. Support for this hypothesis comes from rescue experiments in which animals missing wild type HLH-8 are rescued by physically-linked HLH-8 dimer molecules. Altogether, our results show that a subset of mesoderm development requires the function of at least two dimer partners: HLH-8/HLH-2 heterodimers and HLH-8/HLH-8 homodimers. Of particular interest, is the finding that the homodimers function prior to and heterodimers function during differentiation of vulval muscles leading a model in which individual HLH-8-containing dimers work in the same cells.

Results

Disrupting hlh-2 gene function leads to defects in vulval muscle patterning and tissue-specific expression of HLH-8 targets

To investigate the role of HLH-2 in the mesoderm, a partial deletion hlh-2 (tm1768) was examined. This allele has also been used in studies investigating HLH-8 function in gonadogenesis and cell invasion (Chesney et al., 2009; Schindler and Sherwood, 2011). The hlh-2 (tm1768) allele has an insertion/deletion leading to a truncated protein that still contains the bHLH domain (Fig. 1A). The hlh-2 (tm1768) animals were sterile (Ste) and many of them had a protruded vulva (Pvl) at 25°C (Fig. 1D,F; Table 1). To confirm that the defects were due to the deletion in the hlh-2 gene, the animals were injected with genomic hlh-2 DNA. The wild type hlh-2 DNA rescued the sterility and Pvl defects in two independent lines (Fig. 1G; Table 1). Therefore, the tm1768 allele is likely to be a loss-of-function allele with defects restricted to the hlh-2 locus.

Figure 1. hlh-2 (tm1768) animals express egl-15::gfp.

(A) The predicted protein products from wild-type hlh-2 and the tm1768 mutant locus. The tm1768 allele contains a 633 bp deletion and an 18 bp insertion leading to a loss of 93 amino acids and gain of 6 novel amino acids near the amino terminus but leaving an intact bHLH domain. (B,D) Nomarski and (C,E) gfp images of wild-type and hlh-2 (tm1768) adult hermaphrodites expressing egl-15::gfp in the vm1 vulval muscles. The wild-type vm1 cells in this lateral view have a characteristic “V” shape surrounding the vulval opening (white arrows). The egl-15::gfp is also expressed in the mutant animals (E), but the vulval muscles appear shorter and are not appropriately connected in comparison to wild-type. (D, F) The temperature-sensitive hlh-2 (tm1768) animals develop into sterile adults with no embryos in the uterus. Additionally, the hlh-2 (tm1768) animals have a protruded vulva (Pvl) phenotype (F) or a severely protruded/everted (Evl) vulva phenotype with the complete expulsion of, presumably, the uterus (not shown). (G) The hlh-2 (tm1768) animals with wild-type hlh-2 genomic DNA (Ex [hlh-2]) were fertile with developing embryos in the uterus. All animals were reared at the non-permissive temperature, 25°C.

Table 1.

Summary of gfp phenotypes in hlh-2 mutant and RNAi-treated animals

	Phenotypes		hlh-8::gfp	arg-1::gfp			egl-15::gfp
Strain	Ste a	Pvl	M lineage D/V	hmc	vms	ents	vms
wild type hlh-2(tm1768) tm1768;Ex[hlh-2] hlh-2 RNAi hlh-2(n5287Δ)/+	0% (>100) b	0% (>100)	100% (>100)	100% (138)	100% (138)	100% (138)	100% (>100)
	100% (>100)	71% (28)	95% (58)	0% (31)	0% (31)	100%c (31)	100% (30)
	43% (63)	19% (37)	nd	60% (30)	83% (30)	93% (30)	nd
	63% (191)	23% (191)	nd	100% (44)	27% (44)	100% (44)	62% (42)
	0% (>100)	0% (>100)	100% (47)	100% (59)	100% (104)	100% (104)	100% (41)

^aAbbreviations used: Ste, sterile; Pvl, protruded vulva; hmc, head mesodermal cell; ents, enteric muscles; D/V, the first division plane divides dorsal/ventral in the M lineage; vms, vulval muscles; +, gfp expression is observed in the indicated tissue; −, no gfp expression is observed in the indicated tissue; nd, not determined because the tm1768 mutant animals were not defective for this phenotype or because the RNAi treatment was not complete until after the M lineage D/V division

^bNumber in parentheses represent number of animals examined.

^cExpression in the enteric muscles is present but fainter than in wild-type animals and the percentage of animals expressing gfp decreases as the animals age (84% and 60% on days 2 and 3 of adulthood in tm1768 versus 100% of animals expressing bright gfp on day 3 of adulthood in wild type).

We observed the hlh-2 loss-of-function phenotype and found that unlike hlh-8 mutants, the animals were not visibly constipated and they were able to lay eggs. We examined the HLH-8 target gene arg-1::gfp in the hlh-2 (tm1768) mutants and found it was expressed in the enteric muscles but not the hmc or the vulval muscles (Table 1). The arg-1::gfp expression in the enteric muscles was expressed in young adults but became very faint and decreased in the number of animals by day 3 of adulthood in the tm1768 background in comparison to the bright gfp expression throughout adulthood of wild-type animals (Table 1). The decreased enteric muscle expression was not due to a depletion of maternal stores of HLH-2 due to the hermaphrodites being reared until the L4 stage at 20°C. The arg-1::gfp expression pattern in the enteric muscles was the same regardless of whether larvae born early or late in the brood at 25°C. The lack of arg-1::gfp expression in the vulval muscles and in the hmc was rescued with extrachromosomal hlh-2 genomic DNA (Table 1). Altogether, the observations with arg-1::gfp suggest that HLH-2 is not required or is required in a lower amount to activate arg-1::gfp in the enteric muscles, but is required for expression in the vulval muscles and the hmc and maintenance of expression in the enteric muscles.

To test whether HLH-2 played a role in egl-15 expression, egl-15::gfp was evaluated in the hlh-2 mutants. The hlh-2 (tm1768) mutants had robust egl-15::gfp expression at 25°C in adults (Fig 1E). However, the gfp expression faded 3 or 4 days after activation in the hlh-2 (tm1768) mutants whereas it persisted in wild-type animals for more than a week. Additionally, since the egl-15::gfp reporter was expressed in the hlh-2 (tm1768) animals, we could observe the morphology of the vulval muscles. In most of the hlh-2 (tm1768) animals, the vulval muscles looked similar to wild type (Fig. 1C) but in approximately a third of the animals the vulval muscles were shorter and less regularly shaped than wild-type vulval muscles (Fig. E). Since hlh-2 (tm1768) is predicted to be a hypomorph (Chesney et al., 2009), these results suggest the HLH-2 plays an important role in the patterning of the vulval muscles and a maintenance role in the expression of egl-15.

hlh-2 RNAi-treated animals showed defects in vulval muscle patterning and CeTwist target gene expression

To confirm the loss-of-function phenotypes we observed for hlh-2 in the mesoderm, we used an RNA interference assay (RNAi). Many functions of hlh-2 have been investigated using RNAi (Thellmann et al., 2003; Karp and Greenwald, 2004; Verghese et al., 2011; Schindler and Sherwood, 2011). Because HLH-2 plays an essential role during embryogenesis (Krause et al., 1997; Kamath et al., 2001), L1 animals were fed the hlh-2 dsRNA and adults were observed. The RNAi phenotypes were similar but less penetrant than the hlh-2 (tm1768) animals (Table 1). A majority of RNAi treated animals became Ste while about a quarter of the animals were Pvl (Fig. 2D; Kamath et al., 2003, Karp et al., 2004, Ceron et al., 2007). None of the hlh-2-RNAi treated animals were visibly constipated although the enteric muscles develop before the RNAi treatment at the L1 larval stage. Additionally, the sterile animals could not be examined for egg-laying defects since they did not contain any embryos. Therefore, we examined the morphology of the vulval muscles using gfp reporters.

Figure 2. HLH-8 target genes are variably affected by hlh-2 RNAi treatment.

(A) Animals expressing gfp reporters were fed bacteria expressing dsRNA from hlh-2 cDNA (hlh-2 RNAi). A second population of animals ingested bacteria harboring an empty vector (Control RNAi). The reporters egl-15::gfp and arg-1::gfp are integrated on C. elegans chromosomes T12D8.9::gfp and cpn-3::gfp are maintained on extrachromosomal arrays and are included as controls since they are not downstream target genes of HLH-8 (Wang et al., 2006). The fraction of the population expressing gfp in the vulval muscles following RNAi treatment throughout the larval stages is indicated on the graph (n=39–54). The egl-15::gfp and arg-1::gfp expression is significantly decreased upon RNAi treatment at p<0.001. (B–E) Micrographs of egl-15::gfp adults following RNAi treatment. (B,D) Nomarski images. (C,E) gfp images. (B,C) Lateral views of the central region of a control RNAi-treated animal showing normal “V”-shaped vulval muscles expressing gfp. (D, E) Lateral views of an hlh-2 RNAi-treated animal showing an abnormal egl-15::gfp pattern in cells near the vulval opening (white arrow). (D) Animal has a protruding vulva (Pvl).

Vulval muscle expression of arg-1::gfp and egl-15::gfp were examined in the hlh-2 RNAi treated animals (Fig 2). As we had previously observed, arg-1::gfp was significantly decreased with hlh-2 RNAi treatment (Fig 2A, Table 1; Zhao et al., 2007). In contrast, the egl-15::gfp expression was still retained in a majority of the animals (Fig. 2A,E, Table 1). The egl-15::gfp reporter also revealed that similar to hlh-2 (tm1768) animals, hlh-2 RNAi treatment led to vulval muscles defects with misshapen GFP-expressing cells located near the vulval opening (Fig. 1E vs. 1C). Two additional reporters (T12C8.9::gfp and cpn-3::gfp) were included as controls since they are not predicted to be downstream of HLH-2 and HLH-2 (Wang et al., 2006). These control reporters continued to be expressed in the adults after the RNAi treatment (Fig 2A). Overall, we found that hlh-2 dsRNA treatment led to defects in the formation of vulval muscles and decreased expression of arg-1::gfp in those cells.

Because RNAi treatment is expected to lead to a reduction of hlh-2 function, it would be ideal to confirm these results with a complete inactivation of hlh-2. Recently, a predicted null allele, hlh-2 (n5287Δ), has been identified which deletes nearly all of the hlh-2 locus (exons 2–5) (Nakano et al., 2010). As expected for an hlh-2 null phenotype, n5287Δ homozygotes are embryonic lethal. However, the mutation is semidominant and exhibits some phenotypes when heterozygous (Nakano et al., 2010). Therefore, we constructed strains containing the balanced n5287Δ mutation to examine the expression of egl-15::gfp and arg-1::gfp. Both reporters were expressed in the heterozygous animals and the expression pattern and gross level of gfp did not differ substantially from control animals containing the balancer alone (Table 1). Therefore, the hlh-2 null allele did not allow us to determine whether our RNAi results may be due to threshold effects of the HLH-2 protein levels.

HLH-8 homodimers play a critical role in cells where HLH-2 is less important

Thus far, our results support the hypothesis that HLH-2 plays a critical role in gene expression in the vulval muscles and the hmc, but a less important role in the enteric muscles at least for the arg-1 target gene. We, therefore, wanted to test the possibility that HLH-8 can function independently of HLH-2 in the C. elegans mesoderm, potentially as homodimers. Previous studies investigating vertebrate and Drosophila Twist homodimer function used tandemly-cloned genes to produce artificially tethered dimers (Castanon et al., 2001; Connerney et al., 2005). We made a similar tethered dimer construct, pMCP75, to produce linked HLH-8 molecules that are predicted to favor homodimer formation in vivo (Fig. 3A). The dimer construct was used to rescue egg-laying defective and constipated animals containing an hlh-8 (nr2061) null allele (Corsi et al., 2000). The most striking observation was that the majority of the transgenic animals containing pMCP75 were no longer constipated (Fig. 3B,D,F). The rescued phenotype in six different lines ranged from 54% to 86% of the animals (n>50). Furthermore, all of the transgenic animals were still 100% egg-laying defective (n>00) (Fig. 3C,E,G). As a control, a plasmid containing a single copy of hlh-8 genomic DNA, pMCP7, was injected into the hlh-8 (nr2061) animals. In these transgenic animals, both the egg-laying and constipation defects were rescued (data not shown). Whereas the HLH-8 tethered dimer plasmid rescued only the constipation defect in hlh-8 (nr2061) mutants, the HLH-8 monomer plasmid rescued all of the defects.

Figure 3. Tethered HLH-8/HLH-8 dimers can partially rescue hlh-8 mutant animals.

(A) The plasmid, pMCP75, was engineered to have one copy of hlh-8 genomic DNA tethered to one copy of hlh-8 cDNA via a flexible glycine/serine linker. (B–G) Nomarski images of adult animals. (B,D,F) The constipation in hlh-8 (nr2061) animals is relieved when the animals harbor pMCP75. Notice the expanded intestinal lumen (clear region in the center of the animal) in the hlh-8 mutant (D) that is absent in wild type or mutant animals containing pMCP75 (B, F). (C,E,G) The egg-laying defect of hlh-8 (nr2061) animals is not rescued by pMCP75. The uterus of wild-type adults contains early stage embryos whereas the nr2061 adults, with (G) and without (E) pMCP75, have multiple 3-fold embryos (examples are marked with asterisks) near the vulval opening (white arrows), which are characteristic of the egg-laying defective phenotype. In all micrographs, anterior is to the left. (H) The percentage of adult hermaphrodites expressing arg-1::gfp in the hmc (blue), the vulva muscles (red) and the enteric muscles (yellow) in wild-type, hlh-8 (nr2061), and transgenic animals expressing either pMCP75 or one copy of hlh-8 genomic DNA (pMCP7) is plotted on the graph.

Tethered HLH-8 homodimers rescue arg-1::gfp expression in the enteric muscles of hlh-8 (nr2061) animals

Observations from the hlh-2 mutants suggest that HLH-2 is not required to activate arg-1 in the enteric muscles. We tested whether HLH-8 homodimers might be activating arg-1::gfp, in the hlh-8 (nr2061) mutants, which do not express arg-1:: gfp in any tissues. The transgenic animals containing pMCP75 expressed arg-1::gfp in the enteric muscles of larvae (26%, n=61) and adults (10%, n=30), but not in the vulval muscles or the head mesodermal cell of larvae (n=61) or adults (n=30) (larvae data, not shown; adult data, Fig. 3H). These results complement the phenotypes observed in hlh-2 (tm1768) animals in which arg-1::gfp is expressed only in the enteric muscles of larvae and suggest that HLH-8 homodimers activate arg-1::gfp in the enteric muscles.

To test whether the HLH-8 tethered dimers had any effect in a wild-type background, pMCP75 was injected into wild-type arg-1::gfp animals. Only half of these transgenic animals expressed arg-1::gfp in the vulval muscles and in the hmc compared to all of wild-type animals expressing gfp in both tissues (Fig. 3H). In contrast, nearly all of transgenic animals with pMCP75 expressed arg-1::gfp in the enteric muscles (Fig. 3H). When pMCP7 with the single copy of hlh-8, was injected into wild-type animals, all of the animals expressed arg-1::gfp in all three tissues (Fig. 3H). Altogether, the data indicate that expressing HLH-8 tethered dimers in wild-type animals reduces arg-1 expression in the vulval muscles and the hmc but does not affect expression in the enteric muscles. These results are consistent with the tethered dimers having a dominant negative effect at cell-type specific promoter elements where the homodimer does not normally function to activate transcription.

Tethered HLH-8 homodimers do not rescue egl-15::gfp expression in hlh-8 (nr2061) animals

Since HLH-2 did not seem to play a critical role for egl-15::gfp expression, we tested whether HLH-8 homodimers might be responsible. We introduced pMCP75 into hlh-8 (nr2061) animals containing egl-15::gfp, which is not expressed in the mutant animals. We did not observe gfp expression in any of the transgenic animals (0%, n=54). We also tested whether pMCP75 had any effect on egl-15::gfp in wild-type animals. Only about half of the egl-15::gfp animals containing pMCP75 (47%, n=60) expressed gfp whereas all of the non-transgenic siblings expressed gfp (100%, n=61). Therefore, neither HLH-2 nor HLH-8 homodimers appear to be responsible for activating egl-15 leading to the hypothesis that another HLH-8 containing dimer may play a role. Further evidence to support this claim is that the HLH-8 tethered dimers again are acting in a dominant negative fashion in wild-type animals to reduce expression from a promoter (egl-15) where homodimers normally do not function.

M lineage patterning is controlled by HLH-8 homodimers rather than HLH-8/HLH-2 heterodimers

Since both arg-1 and egl-15 are expressed in vulval muscles, which are derived from the M lineage, we tested whether HLH-2 was playing a critical role earlier in the M lineage. In wild-type L1 animals, the M mesoblast initially divides dorso-ventrally (D/V) in the posterior region of the larvae to produce 18 myoblast descendants in four quadrants at the L2 larval stage (Fig 4A–C, Fig 5B). hlh-8 (nr2061) mutants have aberrant M division planes leading to a ventral queue of cells in 75% of the animals (Corsi et al., 2000; Fig. 4J,K). To assess whether HLH-2 plays a role in the early M lineage patterning with HLH-8, we examined hlh-2 (tm1768) animals containing an hlh-8::gfp reporter (Harfe et al., 1998b) that contains only the hlh-8 promoter and no HLH-8 coding sequences. At the non-permissive temperature, hlh-2 (tm1768) animals exhibited normal D/V divisions (Table 1, Fig. 4F, G) suggesting that HLH-2 is not likely to be playing a major role in early M lineage patterning. We also observed the early M lineage patterning with hlh-8::gfp in hlh-2 (n5287Δ)/+ heterozygous animals and found wild-type D/V divisions (Table 1). To test whether HLH-8 homodimers might be patterning these cells, we observed hlh-8 (nr2061) animals expressing the HLH-8 tethered dimers (pMCP75) (Fig. 4N,O). All of the hlh-8 (nr2061) mutant animals containing pMCP75 had a D/V division pattern similar to wild type (n=36). Therefore, the early M lineage defects of hlh-8 (nr2061) animals were rescued by the tethered HLH-8 dimers suggesting that homodimers are functioning to control patterning early in the lineage.

Figure 4. The M lineage is unaffected in hlh-2 (tm1768) animals and is rescued in hlh-8 (nr2061) animals containing tethered HLH-8/HLH-8 dimers.

(A) Schematic representation of wild-type M lineage development in L2 and L4 larvae. At the L2 stage, M descendants include 14 body wall muscles (bwm), 2 coelomocytes (cc) and 2 sex myoblasts (SMs). At the L4 stage, the SMs divided into 16 SM descendants near the vulval opening. (B,F,J,N) Merged Nomarski/gfp and (C,G,K,O) gfp images of the L2 posterior. (C)The wild-type M lineage has four quadrants of descendants (two seen here). (G,O) Both hlh-2 (tm1768) and hlh-8 (nr2061) with pMCP75 have a wild-type M pattern. In contrast, hlh-8 (nr2061) animals frequently have M descendants in a ventral queue (K). (D,H,L,P) Merged Nomarski/gfp and (E,I,M,Q) gfp images of the L4 central region. (E,I) 8/16 SM descendants are in the focal plane of both wild-type and hlh-2 (tm1768) animals. (M) The hlh-8 (nr2061) animals have extra SM descendants. (Q) In the hlh-8 (nr2061) animals with pMCP75, the correct number of SM descendants is restored (6/8 cells in this view). All animals contain a non-rescuing hlh-8::gfp reporter, expressed in all undifferentiated M lineage cells.

Figure 5. Summary of mesodermal mutant phenotypes and model for distinct HLH-8 dimer function in the M lineage.

(A) Summary of phenotypes observed in this study and the predicted HLH-8 containing dimers that contribute to the phenotype or tissue-specific reporter gene expression. (B) In the M lineage, HLH-8 is proposed to function as homodimers in the undifferentiated cells (shown in gray) whereas HLH-8/HLH-2 heterodimers are proposed to function in the differentiated vulval muscles that are derived from this lineage (shown in black). The vertical lines in the lineage represent cells (including sex myoblasts (SM) and vulval muscles) and horizontal lines represent cell divisions (d, dorsal; v, ventral; l, left; r, right; a, anterior; p, posterior). The larval stages (L1–L4) are indicated on the left. Abbreviations used: Con, constipated; Egl, egg-laying defective; hmc, head mesodermal cell; ents, enteric muscles; vms, vulval muscles; nd, not determined since animals did not have any offspring; WT, wild type (e.g. not constipated); mut, mutant (e.g. constipated); Y, tissue-specific gfp expression; N, no gfp expression; dec., decreased gfp expression compared to wild type; ?, unknown partner protein.

To further investigate which HLH-8 containing dimers function early in the M lineage, we examined the phenotypes of the sex myoblasts (SM) and their descendants. In wild-type L3 larvae, two SMs migrate from the posterior to the vulval region where they undergo several divisions to produce 16 SM descendants at the L4 larval stage (Sulston et al., 1983; Fig 4A,D,E, Fig 5). Both HLH-2 and HLH-8 are expressed in these 16 descendants (Krause et al., 1997, Harfe et al., 1998b) that later develop into the vulval and uterine muscles (Sulston et al., 1983). In hlh-8 (nr2061) mutants, more than two cells adopt an SM fate in 57% of animals leading to extra SM descendants at the L4 stage (Corsi et al., 2000; Fig 4L, M). The SMs and their descendants were examined in both the hlh-8 (nr2061) animals containing pMCP75 and hlh-2 (tm1768) animals using the hlh-8::gfp reporter. Both hlh-2 (tm1768) animals (n=45) and hlh-8 (nr2061) animals with pMCP75 (n=41) had the correct number of SM cells (Fig 4H,I,P,Q). Furthermore, the HLH-8 tethered dimers did not have a dominant negative phenotype in the M lineage patterning since 100% of wild-type animals expressing pMCP75 had normal D/V divisions (n=62) and SM phenotypes (n=27). Altogether, these results support the important role for HLH-8 containing homodimers and not HLH-2/HLH-8 heterodimers in C. elegans M lineage patterning.

Discussion

In this work, we studied the function of the transcription factor HLH-2 in conjunction with its bHLH partner protein HLH-8 in the C. elegans mesoderm. We focused our analysis on cells and tissues where HLH-8 is expressed: in the early M lineage, the M-derived vulval muscles, the head mesodermal cell, and the enteric muscles. Our results showed that HLH-2 is required for activation of arg-1 in the hmc and vulval muscles and for proper formation of vulval muscles. In contrast, for activation of arg-1 in the enteric muscles and for the patterning of the early M lineage, HLH-2 appears to play a less important role. Consistent with these results, we found that tethered HLH-8 molecules can rescue a subset of phenotypes associated with hlh-8 mutants: constipation, arg-1 activation in the enteric muscles, and early M lineage patterning (Fig. 5).

HLH-2 functions in a subset of the mesodermal cells where HLH-8 functions

HLH-2 mesodermal function was studied using mutants and RNAi treatment. Animals with reduced hlh-2 function had at least partially formed vulval muscles (Fig 1E, 2E) in contrast to hlh-8 mutants that do not contain any properly formed vulval muscles (Corsi et al., 2000). The mild vulval muscle phenotype may result from residual HLH-2 activity due to non-null hlh-2 alleles and heterozygous null hlh-2 (n5297Δ)/+ animals. Alternatively, the partial formation of the vulval muscles in the mutants could be explained if HLH-2 is not an obligate partner for HLH-8 to regulate its full complement of target genes. This scenario is supported by the result that HLH-8 tethered dimers rescue hlh-8 null mutant phenotypes that are not defective in hlh-2 mutants such as the M lineage patterning. Therefore, we propose that HLH-2 is playing a major role in only a subset of tissues where HLH-8 functions: the vulval muscles and the hmc (Fig. 5A).

HLH-8 homodimers play an important role in the mesoderm

None of the hlh-2 mutants or RNAi treatment led to constipated animals, and the tethered HLH-8 homodimers rescued the enteric muscle defects in hlh-8 (nr2061) animals. Additionally, in the hlh-2 (tm1768) animals, arg-1::gfp is expressed at all larval stages at the non-permissive temperature, but is not expressed as robustly and is prematurely turned off in the adult enteric muscles. We didn’t find evidence for maternal deposition of HLH-2 at the permissive temperature leading to arg-1::gfp expression in larvae. Therefore, arg-1 regulation is likely to change as the animals transit from the larval stages to adulthood. Although hlh-2 is expressed in the enteric muscles in L1 animals (Krause et al., 1997), HLH-2 may not be required for arg-1 expression at that time. Finally, hlh-8 (nr2061) animals expressing HLH-8 tethered dimers regained arg-1::gfp expression in the enteric muscles of larvae. These results support the hypothesis that HLH-8 functions in the enteric muscles as a homodimer to activate arg-1 in larvae (Fig. 5). Later in development, HLH-8 may dimerize with HLH-2, to play a maintenance role in arg-1 enteric muscle expression during adulthood.

There are several pieces of evidence supporting the conclusion that the HLH-8 tethered dimer phenotypes result from effective dimerization of the two physically-linked monomers in vivo. First, only the constipation phenotype of the hlh-8 (nr2061) mutants was rescued in transgenic animals expressing the linked dimers, whereas constipation and egg-laying defects are rescued when the mutant is expressing a single copy of the hlh-8 gene. Second, in hlh-8 (nr2061) animals the HLH-8 tethered dimers rescued the arg-1::gfp in the enteric muscles but not the vulval muscles or the hmc whereas expression was restored in all tissues with the monomeric HLH-8 expression. Finally, the rescue results with the HLH-8 tethered dimers complement the results with the reduction of hlh-2 function. Although there are alternate interpretations, we suggest the HLH-8 tethered dimers may not effectively dimerize with HLH-2 in vivo. Therefore, the tethered dimers rescue the phenotypes controlled by HLH-8 homodimers whereas the HLH-8 monomers rescue all HLH-8 functions in the mutant animals. Additionally, we observed that HLH-8 tethered dimers acted in a dominant negative fashion on arg-1::gfp expression in the vulval muscles and the hmc but not the enteric muscles. These results are consistent with the HLH-8 tethered dimers having a deleterious effect at promoters elements where HLH-8 should be dimerizing with a partner but not where HLH-8 homodimers function.

A working model for CeTwist and CeE/DA in mesoderm gene regulation

Based on the summation of our data, we propose the following model for the regulation of mesoderm targets by HLH-8 and HLH-2 (Fig. 5). The HLH-8 homodimer may function early in the mesoderm to promote the formation of enteric muscles, to activate arg-1 in these cells, and to regulate M lineage patterning. This hypothesis is supported by the observations that, unlike hlh-8 mutants, hlh-2 mutants do not have defects in the patterning of the early M lineage descendants and do not have defects in the function of their enteric muscles. Furthermore, the hlh-2 mutants have normal arg-1::gfp expression in the larval enteric muscles. Finally, hlh-8 (nr2061) mutants do not express arg-1 in the enteric muscles. For these mesodermal functions, HLH-8 could also be partnering with a bHLH other than HLH-2 as a heterodimer. However, the observation that HLH-8 tethered dimers rescue these phenotypes in hlh-8 (nr2061) mutants suggest that the function is likely to be performed by HLH-8 homodimers.

We propose that HLH-8 partners with HLH-2 later in the M lineage to promote vulval muscle formation and to activate arg-1 in the vulval muscles (Fig. 5). HLH-8/HLH-2 heterodimers maintain egl-15::gfp expression later in adults, but egl-15 activation requires a yet unidentified dimer containing HLH-8 for activation. Indeed, there are three other bHLH potential partner proteins that are expressed in vulval muscles in a large-scale gfp reporter analysis (Reece–Hoyes, et al., 2007). Furthermore, in a microarray study where both HLH-8 and HLH-2 were overexpressed, egl-15 did not show a change in expression whereas arg-1 expression did significantly increase (Wang et al., 2006). These results support the idea that arg-1 responds to HLH-8/HLH-2 heterodimers but egl-15 does not. Altogether, our working model is that HLH-8 homodimers function early in mesoderm development playing an important role in the early M lineage and the enteric muscles, while HLH-8/HLH-2 heterodimers function later in the M lineage to turn on target gene expression in the M-derived vulval muscles and maintain expression of some target genes (Fig 5).

The arg-1 promoter paradigm

Our results are consistent with previous work revealing that tissue-specific arg-1 expression is regulated by individual promoter elements (Zhao et al., 2007). Of the three E boxes (E1, E2 and E3) in the minimal arg-1 promoter region, mutations in E1 and E2 abolish arg-1 expression in the vulval muscles and the hmc, while maintaining arg-1 expression in the enteric muscles (Zhao et al., 2007). Mutations in E2 and E3 abolish arg-1::gfp in the enteric muscles (Zhao et al., 2007). These results are consistent with the idea that arg-1 is regulated by two different HLH-8 dimer partners, where HLH-8 homodimers may bind to E3 to activate arg-1 in the enteric muscles, and HLH-2/HLH-8 heterodimers may bind to E1 to activate arg-1 in the vulval muscles. However in vitro gel shift assays showed that the homodimers and heterodimers can bind to both E1 and E3 (Zhao et al., 2007). Further investigation of the arg-1 promoter in vivo might reveal that in the context of chromatin or in the presence of other factors, the two E boxes will favor being bound by only one of the dimers. Regulation of a single gene by two different dimers was reported in vertebrates, where Twist homodimers and Twist heterodimers differentially regulate FGFR2 expression (Connerney et al., 2006). However, the specific binding sites in FGFR2 for the dimers have not yet been identified. We expect that due to its broad expression pattern and modular promoter that arg-1 will continue to be an interesting paradigm for understanding HLH-8 transcriptional regulation.

HLH-8 function relative to other organisms

The pathway in which HLH-8 functions shows significant homology with the Twist pathway in Drosophila and in vertebrates (Harfe et al., 1998b). For instance, both arg-1 and egl-15 have homologs that are also targets of Twist in these organisms. In addition, the Twist–containing homodimers versus heterodimers are an evolutionarily conserved theme. In vertebrates, Twist1 homodimers and Twist1/E heterodimers differentially regulate fibroblast growth factor receptor 2 (FGFR2), the homologue of EGL-15 (Connerney et al., 2006). Twist also activates an FGFR homologue during Drosophila gastrulation (Shishido et al., 1993). In both of these organisms, the Twist containing dimers function in a distinct time or in distinct cells during development (Castanon et al., 2001; Connerney et al., 2006). For example, in Drosophila, Twist homodimers play a role in early embryogenesis in patterning the entire mesodermal layer and later in development Twist-Daughterless heterodimers are involved in specifying the muscle type that will develop from the mesodermal layer (Castanon et al., 2001). In vertebrate skull development, the sutures containing undifferentiated bone precursor cells have Twist1/E heterodimers in the mid-suture mesenchyme and Twist1 homodimers in the peripheral cells at the osteogenic front that are differentiating into bone (Connerney et al., 2006). In the osteogenic front, E proteins dimerize with the HLH protein Id to promote Twist homodimer formation (Connerney et al., 2006). In the C. elegans genome, there does not appear to be an Id homolog (Ledent and Vervoort, 2001). Therefore, another mechanism is likely to drive HLH-8 homodimer formation. However, this is the first time a Twist homolog has been shown to function with different dimer partners in the same cells (Fig. 5).

We are particularly intrigued to find that different HLH-8 containing dimers are functioning in the same cell lineage because it fits with what we know about hlh-8 gene regulation. An element in the hlh-8 promoter controls expression in all undifferentiated cells of the M lineage (Harfe et al, 1998b). Two hlh-8 intron E boxes control expression in the differentiated sex muscles by autoregulation through HLH-8 containing dimers (Meyers and Corsi, 2010). This differential regulation may reflect the need for a higher level of HLH-8 in the vulval muscles where HLH-8/HLH-2 heterodimers are playing an important role and HLH-8 may be functioning with other partners to activate genes such as egl-15. It is currently not well understood how the level of Twist is controlled in vertebrates. It will be interesting to determine whether autoregulation contributes to variation in the level of Twist in specific cell types and contributes to cell-type specific dimer selection. In the meantime, we expect that C. elegans will continue to be a useful model for understanding the evolutionarily conserved nature of Twist transcriptional regulation by allowing single cell resolution studies with relevance to human diseas.

Experimental Procedures

hlh-2 deletion mutants

The allele hlh-2 (tm1768) was isolated and provided by the National BioResource Project of Japan. The strain was backcrossed three times with wild-type (N2) animals and the hlh-2 locus was PCR amplified and sequenced to confirm the deletion (Chesney et al., 2009). A truncated mRNA of the expected size was detected by RT-PCR. We sequenced the entire hlh-2 locus in the mutant animals and found no other mutations. The hlh-2 (tm1768) animals were analyzed at 20°C and 25°C. The allele hlh-2 (n5287Δ) was obtained from the Horvitz laboratory (Nakano et al., 2010). This strain is homozygous embryonic lethal and was maintained as a balanced line with hT2[bli-4(e937) let-?(q782) qIs48] (I;III). These animals were maintained and analyzed at 20°C.

Rescue of hlh-2 (tm1768) mutants with hlh-2 genomic DNA

Wild-type hlh-2 genomic sequence was used to rescue the hlh-2 (tm1768) mutants. Two overlapping PCR fragments were amplified from a C. elegans lysate (Williams, 1995). These fragments consisted of an 8 kb sequence amplified from the promoter region of hlh-2 and overlapping by 595 bases with the second PCR product consisting of hlh-2 coding DNA, and the 3’UTR. The hlh-2 (tm1768) mutants were injected with these two fragments along with a dominant marker rol-6 (su1006) (pRF4) (Mello et. al, 1991). Two independent transgenic lines were generated.

hlh-2 RNAi treatment

hlh-2 dsRNA was introduced into C. elegans strains harboring gfp reporter constructs by feeding the animals E. coli expressing dsRNA. In order to bypass the embryonic lethality resulting from early loss of hlh-2, L1 larvae were subjected to a feeding protocol (Karp and Greenwald, 2003). Animals were synchronized at the L1 stage by collecting embryos after hypochlorite treatment followed by hatching on nematode growth medium (NGM) agar plates in the absence of food. The feeding RNAi was performed according to Kamath and colleagues (Kamath et al., 2000). Briefly, an E. coli strain HT115 (DE3) containing either an empty L4440 vector (Timmons and Fire, 1998) [Control RNAi] or the hlh-2 full-length cDNA [hlh-2 RNAi] was grown on 0.35 mM IPTG to induce the dsRNA. After overnight incubation at room temperature, the L1 animals were added to the feeding plates and further incubated at 20°C. The animals were moved to a new plate every 24 hours for 2 days, and then were observed for vulval muscle gfp expression.

hlh-2 mutant gfp reporter strains

Transcriptional GFP reporter strains expressing known targets of HLH-8 were maintained at 20°C. The strains are as follows: NH2447 ayIs2 [egl-15::gfp;dpy-20(+)]IV, shows gfp expression in the vm1 vulval muscles (Harfe et al., 1998); PD4444 ccls4444[arg-1::gfp; dpy-20(+)] II, shows gfp expression in the head mesodermal cell (hmc), vm1 vulval muscles, and four enteric muscles (Kostas and Fire, 2002); PD4667 ayIs7 [hlh-8::gfp] IV expresses in the M mesoblast cell, the 18 myoblast descendants and the sex myoblast descendants (Harfe et al., 1998). These integrated gfp reporter constructs were crossed into hlh-2 mutant animals via standard mating techniques. The homozygous reporters were confirmed through observing gfp expression in 100% of the animals. To confirm that hlh-2 (tm1768) was homozygous, we used PCR amplification and observed phenotypic changes at the non-permissive temperature. The hlh-2 (n5287Δ) mutation was maintained with a balancer that could be followed due to a pharyngeal gfp marker. The n5287Δ allele was confirmed to be present in the final strains by detecting the deleted locus with PCR. The GFP reporters were also crossed into heterozygous balancer strains without hlh-2 mutations to control for effects of the balancer on gfp expression. The followings strains were generated from these crosses: AK130 ccls4444 [arg-1::gfp; dpy-20(+)]II; hlh-2 (tm1768); AK115 ayIs7 [hlh-8::gfp] IV; hlh-2 (tm1768); AK114 ayIs2 [egl-15::gfp;dpy-20(+)]IV; hlh-2 (tm1768); AK164 ayIs7 [hlh-8::gfp] IV; hT2[bli-4(e937) let-?(q782) qIs48] (I;III)/+; AK165 ayIs7 [hlh-8::gfp] IV; hlh-2 (n5287Δ)/hT2; AK166 ayIs2 [egl-15::gfp;dpy-20(+)]IV; hT2/+; AK167 ayIs2 [egl-15::gfp;dpy-20(+)]IV; hlh-2 (n5287Δ)/hT2; AK168 ccls4444 [arg-1::gfp; dpy-20(+)]II; hT2/+; AK169 ccls4444 [arg-1::gfp; dpy-20(+)]II; hlh-2 (n5287Δ)/hT2.

hlh-8 mutants

hlh-8 (nr2061) mutant animals have been previously characterized (Corsi et al., 2000). The hlh-8 (nr2061) allele contains a 1267 base pair deletion that removes the entire HLH domain. hlh-8 (nr2061) animals were used to generate the following transgenic animals: AK107 thEx8 [pMCP75; pRF4];hlh-8 (nr2061);arg-1::gfp, expressing the CeTwist tethered dimer construct pMCP75; AK113 thEx9 [pMCP7; pRF4]; hlh-8 (nr2061); arg-1::gfp, expressing the single copy of hlh-8 genomic DNA, pMCP7. See next sections for a description of these plasmids. The dominant marker rol-6 (pRF4; Mello et al., 1991) was used to identify transgenic animals. In addition, transgenic animals harboring the pMCP75 plasmid were crossed to hlh-8 (nr2061) animals with previously integrated gfp reporters egl-15 or hlh-8 (Corsi et al., 2000) and scored for gfp expression.

HLH-8 monomer and homodimer plasmid construction

The pMCP7 plasmid containing the hlh-8 genomic DNA was generated from pBH64 plasmid. pBH64 was previously used to rescue all phenotypes of hlh-8 (nr2061) animals (Corsi et al., 2000). The 9 kb upstream promoter contained in pBH64 was reduced to 4 kb in pMCP7. The final pMCP7 plasmid included the promoter region, the hlh-8 coding sequence and 3 kb downstream genomic DNA. In addition, this plasmid was modified to generate 2 unique BstXI sites flanking the genomic DNA in order to insert the tandem copies of hlh-8 for constructing the HLH-8/HLH-8 tethered dimer construct pMCP75. pMCP75 was prepared according to previous methods used for other tethered dimers (Castanon et al., 2001 and Neuhold and Wold, 1993), with some additional modifications. First, a Sequence Overlapping Extension (SOE) PCR (Hobert, 2002) fragment was generated to join the one copy of hlh-8 genomic DNA with one copy of hlh-8 cDNA. We included hlh-8 genomic DNA because there are intron enhancer elements required for complete hlh-8 expression (Meyers and Corsi, 2010). The 4 kb upstream promoter and hlh-8 genomic DNA was amplified from pBH64. The hlh-8 cDNA was amplified from pAC2 (Corsi et al, 2000). Because the hlh-8 genomic DNA and the cDNA share the same sequence at the 3’ end, we introduced several silent base pair changes at the 3’ end to allow a unique PCR fragment to be amplified. The two monomers were connected via a flexible, 16 amino acid long (Gly₃-Ser₁) ₄ polylinker, which was added between the hlh-8 genes during PCR amplification. The 8.7 kb HLH-8/HLH-8 tethered dimer fragment was cloned into the BstXI sites of pMCP7, which contained the hlh-8 4 kb promoter and 3’UTR. The final plasmid pMCP75 was confirmed by restriction digestion and sequencing.

Bullet Points (major take-home points).

• HLH-8/HLH-8 dimers function in M lineage

• HLH-2/HLH-8 dimers function in vulval muscles

• HLH-8 dimer switch coincides with differentiation