The myogenic potency of HLH-1 reveals wide-spread developmental plasticity in early C. elegans embryos

The transcriptional regulation of vertebrate skeletal muscle development is well understood (reviewed by Buckingham,2001; McKinsey et al.,2001; Pownall et al.,2002; Tajbakhsh,2003; Buckingham et al.,2003). Sonic hedgehog and Wnt signals, from the notochord and neural tube respectively, work in concert with the homeobox gene Pax-3 to turn on the early-acting myogenic regulatory factors (MRFs) Myf-5 and MyoD in the adjacent somites. These two basic helix-loop-helix (bHLH) MRFs initiate a transcriptional cascade, including the activation of two closely related bHLH MRFs, MRF-4 and myogenin, that culminates in the expression of terminal skeletal muscle gene products (e.g. myosin heavy chain, actin) needed for differentiation (McKinsey et al.,2001; Berkes et al.,2004). These events are integrated with growth signals so that proliferative myoblasts exit the cell cycle at the appropriate point in development (reviewed by Sabourin and Rudniki, 2000; Wei and Paterson, 2001).

In C. elegans, the striated body wall muscle cells are comparable to vertebrate skeletal muscle. This musculature provides the locomotive force for the animal and consists of 95 mononucleated cells, arranged in four quadrants along the length of the body(Waterston, 1988). Eighty-one of these cells are born and differentiate during embryogenesis(Sulston et al., 1983), while 14 more are added post-embryonically(Sulston and Horvitz, 1977). Although previous studies have identified several factors required for post-embryonic muscle development (Harfe et al., 1998a; Harfe et al.,1998b; Liu and Fire,2000; Corsi et al.,2000), there has been little progress in understanding embryonic striated myogenesis. One general conclusion of these studies is that embryonic and post-embryonic muscle development are controlled by different sets of transcription factors, an unexpected finding given that muscle cells born during these two different periods of development appear morphologically and functionally equivalent.

In C. elegans, fertilization is followed by a rapid set of embryonic cellular divisions that generate five somatic founder blastomeres called AB, MS, E, C and D, and the germline blastomere P4. The 81 embryonic body wall muscles are derived from four of the five somatic founders AB (1),MS (28), C (32) and D (20). The D lineage gives rise exclusively to body wall muscle whereas the other lineages give rise to multiple cell fates. Maternal effect mutations have been identified that alter the fate of one body wall muscle-producing founder without affecting other lineages. For example, skn-1 mutants lack the 28 MS-derived body wall muscles but body wall muscle from the C and D lineages are present(Bowerman et al., 1992; Bowerman, 1995; Bowerman et al., 1997). Conversely, pal-1 mutants lack the C and D lineage-derived body wall muscle cells without affecting those derived from MS(Hunter and Kenyon, 1996; Ahringer, 1997). This founder blastomere autonomy, with respect to body wall muscle formation, demonstrates that there are several genetically distinct pathways for embryonic striated muscle development. However, it is yet to be determined if these independent genetic pathways converge on a common molecular nodal point to regulate body wall muscle cell fate.

A single C. elegans gene, hlh-1, encodes a transcription factor (HLH-1, a.k.a. CeMyoD) that is related to the vertebrate MRFs. hlh-1 is zygotically expressed in embryonic body wall muscle precursors and their differentiated descendants, beginning at the ∼90 cell stage of development (Krause et al.,1990). This expression pattern suggested that hlh-1 might represent a nodal point for body wall muscle development, analogous to the role of Myf-5 and MyoD in vertebrate myogenesis. Homozygous hlh-1 null mutant animals complete embryogenesis but are paralyzed upon hatching, have severe morphological defects, and usually die during the first larval stage, revealing an essential role for HLH-1 in muscle development and function (Chen et al.,1992; Chen et al.,1994). However, hlh-1 null mutants have 81 embryonic body wall muscle cells and express many terminal muscle products at wild-type levels. Thus, HLH-1 is not essential for body wall muscle cell fate, and one or more additional factors must be involved in myogenic determination. In fact, the hlh-1 knockout allele phenotypes challenge the notion that HLH-1 is itself myogenic, suggesting instead that it may be downstream of myogenic factors in the body wall muscle transcriptional cascade. Similar results in Drosophila for the MRF-related gene nautilus have suggested that there may be a fundamental difference between vertebrates and invertebrates with regard to the regulation of striated muscle development(Michelson et al., 1990; Balagopalan et al., 2001).

The present study further defines the function of HLH-1 and its relationship to intrinsic and extrinsic factors regulating early development. By ectopically activating HLH-1 in the early C. elegans embryo, we show that HLH-1 alone is sufficient to convert most cells of the early embryo into a body wall muscle-like fate. This is true for cells that would normally give rise to either ectoderm or endoderm, demonstrating that HLH-1 is a bona fide MRF with potent myogenic activity. The ectopic myogenic activity of HLH-1 is limited to undifferentiated blastomeres and spans several hours of early development, including times when non-muscle, lineage-restricted markers are normally expressed. We find that the Caudal-related factor PAL-1 can activate hlh-1, providing a link between maternal factors and HLH-1 activation, and we explore the role of Wnt/MAP kinase signaling in making cells competent for myogenesis. These studies demonstrate a level of developmental plasticity in C. elegans that had previously not been appreciated.

The following strains of Caenorhabditis elegans were used: wild type (N2); heat-shock hlh-1 (strains KM267, KM289), hlh-1null allele (cc450) in the mc6-balanced strain PD4849, hlh-1 temperature-sensitive allele (cc561) strain PD4605, hlh-1::gfp strain PD7963 and myo-3::gfp strain PD4251, all provided by A. Fire (Stanford University School of Medicine, CA, USA), elt-2::gfp strain JM63 from J. McGhee (The University of Calgary,Alberta, Canada), and heat-shock pal-1 strains JA1179 and JA1180 provided by J. Ahringer (University of Cambridge, UK).

The heat-shock hlh-1 construct was made by transferring a fragment containing the full-length hlh-1 cDNA(Krause et al., 1990) into the hsp 16.41 (Stringham et al.,1992) vector pPD49.83 (kindly provided by A. Fire) to yield the plasmid pKM1211. Animals harboring pKM1211 were generated by standard techniques using 100 μg of pKM1211 and 50 μg of the selectable dominant rol-6 plasmid pRF4 (Mello and Fire, 1995). Integrated heat-shock hlh-1 lines were generated by gamma irradiation (Egan et al., 1995) of extrachromosomal transformants and were back-crossed twice with the wild-type strain (N2) prior to use.

One and two cell stage embryos were isolated from transgenic hermaphrodites containing heat-shock expression constructs or from N2 controls. Embryos were heat shocked immediately or incubated for various times at room temperature(∼22°C) prior to heat shock. For all treatments, heat shock consisted of a single 30 minute pulse at 34°C. The heat-shocked embryos were examined over time for the expression of cell type-specific reporter genes or incubated and fixed for antibody staining with markers for muscle, gut,hypodermis and germline (see below). Using the absence of hypodermal marker staining as a sensitive assay for the degree of myogenic conversion, we determined that the optimal conditions for HLH-1 myogenic activity were to isolate one- to two-cell embryos and incubate for 60 minutes prior to heat shock. Heat shock induction of PAL-1 was carried out after 20 minutes incubation of isolated embryos.

Embryos were fixed with 5% paraformaldehyde in phosphate-buffered saline(PBS) on ice for 15 minutes and transferred onto 0.1% gelatin-coated slides,placed on an aluminum block on dry ice, freeze-cracked, methanol fixed at–20°C for 6 minutes, and rehydrated in PBS at room temperature. The primary antibodies used were raised against myosin heavy chain A [MHC A(Miller et al., 1986)], HLH-1(Krause et al., 1990), ELT-2(Fukushige et al., 1998) (a gift from Jim McGhee), LIN-26 (Labouesse et al., 1996) (a gift from Michel Labouesse), a pharyngeal muscle epitope [3NB12 (Priess and Thomson,1987)], UNC-89, UNC-98 (Benian et al., 1996; Mercer et al.,2003) (a gift from Guy Benian), PAT-3(Gettner et al., 1995) (a gift from Don Moerman), and germline P-granules [OIC1D4(Strome and Wood, 1983)]. Secondary antibodies were fluorescein- or rhodamine-conjugated donkey anti-rabbit or goat anti-mouse IgG (Jackson Immunological).

Double-stranded RNA corresponding to the genes mex-1, mex-3, lit-1,wrm-1, skn-1 and pal-1 were amplified from cDNA clones that were generated by reverse transcriptase-polymerase chain reaction (RT-PCR) with the primers listed below. These partial cDNA PCR products were inserted into the vector L4440 (Timmons et al.,2001) and served as a template for in vitro transcription to produce double stranded RNA for injection. The pop-1 RNAi plasmid RL499 was a gift from R. Lin.

Primers used for PCR amplification of cDNA clones for RNAi:

• MEX-1F (24 bp) 5′-ATGCAATCTTCAAATGGAGAGCAT-3′,

• MEX-1R (24 bp) 5′-TTATCTCGAATAATGATCTTCGTG-3′,

• MEX-3F (33 bp) 5′-AAGGATCCATGAAGGAAGAACAAATCGCCTATA-3′,

• MEX-3R (25 bp) 5′-TTGGCAGATCTTGTTTCGCCGATTG-3′.

• LIT-1F (24 bp) 5′-CATCCGGCCGCCGTCGGCTCTACG-3′,

• LIT-1R (24 bp) 5′-CGGAGCACATGGTTCTTGGCTCCC-3′.

• WRM-1F (23 bp) 5′-ATGGATGTGGATTGCGCAGAAAC-3′,

• WRM-1R (24 bp) 5′-GACTTCGTTTCCGGTCTTCTCAGG-3′.

• SKN-1F (23 bp) 5′-CATCGTCATACGATCGGATCACG-3′,

• SKN-1R (22 bp) 5′-GTAGGCGTAGTTGGATGTTGGG-3′.

• PAL-1F (23 bp) 5′-CTGAGAGAAAAGATGCTGCAACC-3′,

• PAL-1R (22 bp) 5′-AAATGGATCCGTTCAGAGTGGG-3′.

To test the myogenic potential of HLH-1 in the context of C. elegans development, we engineered the full-length hlh-1 cDNA coding region under the control of a heat shock promoter(Stringham et al., 1992) and created integrated transgenic nematode strains harboring these constructs. After incubation at room temperature for various periods of time, isolated transgenic embryos were heat shocked (34°C) for 30 minutes and assayed over time for the expression of cell type-specific markers.

Under optimal conditions, heat shock-induced hlh-1 was robust and high levels of nuclear localized protein could be detected in most cells of the embryos within 30 minutes after the end of heat shock induction. Levels of HLH-1 remained high throughout embryogenesis, although nuclear localization was less pronounced after overnight (16-20 hours) incubation of the treated embryos. Embryos incubated overnight arrested with 400-500 cells and appeared healthy, as assayed by Nomarski optics, indicating that neither the heat shock treatment nor high HLH-1 levels had any obvious deleterious effect on cellular viability or proliferation.

In almost all treatment paradigms, over-expression of hlh-1resulted in widespread myogenesis; most cells had adopted a muscle-like fate. The muscle marker routinely used to assay myogenesis was myosin heavy chain A[MHC A (Miller et al., 1986) Fig. 1]. Although high levels of MHC A were detected, filaments were disorganized and these muscle-like cells fail to contract. To determine the extent to which these cells adopted a true muscle-like fate, we tested several additional muscle markers, including several structural proteins. The muscle-like cells resulting from heat shock-induced HLH-1 activated a myo-3::gfp reporter gene(Fire and Waterston, 1989) and were positive for filamentous actin, UNC-89(Benian et al., 1996), UNC-98(Mercer et al., 2003) and PAT-3 (Gettner et al., 1995; Francis and Waterston, 1985)(data not shown). These muscle cells were not positive for the pharyngeal muscle-specific marker 3NB12 (Priess and Thomson, 1987). The continued presence of HLH-1 and activation of several body wall muscle markers is consistent with cells terminally differentiating and adopting a fate most closely resembling body wall muscle cells.

HLH-1 apparently acts in a feed-back loop to maintain hlh-1 gene expression (Krause et al.,1994). To determine if ectopic HLH-1 was functioning alone to drive myogenesis or acting through the endogenous hlh-1 gene, we crossed the heat-shock-driven hlh-1 strain into a balanced hlh-1(cc450) null mutant background(Chen et al., 1992). Activation of HLH-1 in animals lacking the endogenous hlh-1 gene resulted in widespread myogenesis in all embryos, indistinguishable from non-mutant controls (data not shown). This demonstrated that a single pulse of heat-shock-activated HLH-1 was able to drive myogenesis in the absence of endogenous hlh-1 gene activity.

The excess number of muscle-like cells observed after ectopic activation of HLH-1 could be due to excessive proliferation of myogenic blastomeres,conversion of other cell types to muscle, or both. The total number of cells in HLH-1-activated, terminally arrested embryos was similar to that normally born during embryogenesis, suggesting that there was not a general and widespread hyper-proliferation of blastomeres induced by HLH-1. We therefore assayed the number of embryonic cells adopting one of several cell fates by scoring them for LIN-26 (hypodermis)(Labouesse et al., 1996),ELT-2 (intestine) (Fukushige et al.,1998), 3NB12 (pharyngeal muscle)(Priess and Thomson, 1987) and P-granules (germline) (Strome and Wood,1983) following heat shock induction of HLH-1. Although the majority of activated HLH-1 embryos had the normal number of germline precursors (two), there was a complete elimination of all somatic cell types assayed in almost all embryos (Fig. 1). Similar heat shock of wild-type embryos did not affect normal development and these embryos were positive for all cell-type markers tested. These results suggested that early expression of hlh-1 was able to convert most, if not all, somatic cells of the embryo into muscle-like cells.

It was possible that HLH-1 was not actively converting cells to muscle but instead blocking normal development and revealing a default muscle cell fate program of early blastomeres. To address this, we eliminated two maternal factors needed to specify several founder blastomere fates. In wild-type embryos, the loss of skn-1 and pal-1 gene products prevents the proper specification of the MS, C and D lineages and greatly reduces, or eliminates, the cell types derived from each lineage(Fig. 2)(Bowerman et al., 1992; Bowerman et al., 1993; Hunter and Kenyon, 1996). Such embryos arrest with 400-500 cells, most of which fail to express body wall muscle markers. In HLH-1-activated embryos that have also been depleted of both skn-1 and pal-1 gene products by RNAi, almost all cells adopted the body wall muscle-like fate(Fig. 2). These results further demonstrated the myogenic potential of HLH-1 activity and that myogenesis is a consequence of HLH-1 activity.

As an additional assay of cell fate conversion, we tested the function of ectopic HLH-1 in embryos in which most cells had been specified as intestine. Loss of the maternal factor MEX-1 results in the four granddaughters of AB adopting an EMS-like fate (Mello et al.,1992; Schnabel et al.,1996). If these embryos are also depleted of POP-1/TCF, all daughters of the EMS and pseudo-EMS cells will develop like E, transforming the entire anterior of the embryo into intestine(Lin et al., 1995; Maduro et al., 2001). When we ectopically activated HLH-1 in embryos depleted of mex-1 and pop-1 by RNAi, almost all somatic cells adopted a muscle-like fate(Fig. 2). Even in mex-1,pop-1 double RNAi-treated embryos that had been incubated for 150 minutes(∼3 hours post-fertilization) prior to HLH-1 activation by heat shock, 80%(n=31) of the embryos had widespread myogenesis and no intestinal cells detectable by ELT-2 antibody staining. This result demonstrates that HLH-1 was sufficient to convert cells destined to become intestine into a muscle-like fate.

Stable hlh-1 gene expression is normally detectable in all body wall muscle precursors shortly after they are born, beginning with the daughters of the D founder about 2 hours after fertilization(Krause et al., 1990). HLH-1 is present in body wall myoblasts as they proliferate and in all differentiated body wall muscle cells in embryos, larvae and adults. To determine if the ability of HLH-1 to convert cells to a muscle-like fate was restricted to a specific time of embryogenesis, embryos harboring the heat shock hlh-1 transgene were isolated and incubated for various periods of time prior to induction. The percentage of embryos that were positive for MHC A, and each of several different non-body wall muscle cell type markers,was used to determine the efficiency of HLH-1-induced myogenesis. For embryos incubated for more than 90 minutes prior to HLH-1 induction (∼2 hours post-fertilization), we used the intestine-specific marker elt-2::GFP to count the number of E cell descendants present at the onset of the heat shock under our experimental conditions(Table 1). Embryos that had anywhere between one and eight E cell descendants at the time ectopic HLH-1 activation was initiated showed widespread myogenic conversion and a nearly complete loss of intestine, hypodermis and pharyngeal cell markers(Table 1). At 210 minutes of incubation (∼4 hours post-fertilization), when embryos averaged 10.2 E descendants, markers indicative of other cell fates became evident in most embryos. This defined a window of competence for blastomeres to respond to ectopic HLH-1 to the first 3 hours of development, up to the eight E cell stage of embryogenesis. After this time, the ability of non-myogenic cells to respond to HLH-1 declines rapidly over the subsequent hour of development and is lost completely in terminally differentiated cells.

PAL-1 is a Caudal-related homeobox protein that is required for the development of the posterior founder blastomeres. Maternal PAL-1 functions to specify both the C and D lineages (Hunter and Kenyon, 1996) while zygotically expressed pal-1 is needed for proper development of their descendants(Edgar et al., 2001). The C lineage gives rise to hypodermis and body wall muscle, whereas the D lineage gives rise exclusively to body wall muscle cells(Sulston et al., 1983). MEX-3 is a negative regulator of PAL-1 and mex-3 mutations or administration of mex-3 RNAi results in anterior blastomeres (the granddaughters of AB) adopting a C-like fate(Draper et al., 1996; Hunter and Kenyon, 1996; Bowerman et al., 1997; Huang et al., 2002). Although there are no other known MEX-3 targets, lineages not expressing PAL-1 are also affected in MEX-3 mutants suggesting additional functions beyond PAL-1 regulation (Draper et al.,1996).

To determine if hlh-1 is a downstream target of PAL-1 in myogenic lineages, we ectopically activated PAL-1 in early embryos harboring an hlh-1::gfp reporter gene. In mex-3 RNAi-treated embryos,ectopic PAL-1 activated the hlh-1 reporter within 3 hours in a subset of blastomeres that differentiated as body wall muscle-like cells. The extent of differentiation of muscle-like cells in mex-3 RNAi-treated embryos was indistinguishable from that observed after heat shock induction of HLH-1 activity. The embryos also had ectopic hypodermal cells that were LIN-26 positive, as was expected from previous characterizations of mex-3mutants (Draper et al., 1996; Bowerman et al., 1997). To determine if the effects of mex-3 RNAi were attributable to PAL-1 mis-regulation, we also ectopically over-expressed PAL-1 using a heat shock promoter-driven pal-1 cDNA clone, kindly provided by Julie Ahringer. We found that heat shock-induced PAL-1 activity in early embryos also resulted in widespread hlh-1 reporter gene activation and myogenesis(Fig. 3), as well as hypodermal development (data not shown). Although similar, the effects of mex-3RNAi and heat shock-induced PAL-1 were not identical and could be distinguished with the gut cell marker ELT-2. All mex-3 RNAi-treated embryos (100%, n=16) were positive for ELT-2 demonstrating that the gut lineage was largely unaffected in these embryos, consistent with earlier studies (Draper et al., 1996). However, only 58% (n=268) of heat shock-induced PAL-1 embryos were ELT-2 positive, suggesting that over-expression of PAL-1 interfered with normal gut development.

The cells of the C lineage that give rise to hypodermis and body wall muscle are the result of anterior-posterior cell divisions(Sulston et al., 1983) and have different levels of nuclear POP-1/TCF(Lin et al., 1998). Studies of Wnt/MAP kinase signaling in the early embryo have shown that LIT-1/MAP kinase phosphorylation of POP-1/TCF, in a WRM-1/beta-catenin-dependent manner,results in the nuclear export of POP-1(Kaletta et al., 1997; Lin et al., 1998; Ishitani et al., 1999; Meneghini et al., 1999; Rocheleau et al., 1999; Shin et al., 1999; Lo et al., 2004). Both LIT-1(Kaletta et al., 1997) and POP-1 (Mickey, 2000) are involved in cell fate regulation in the C descendants; hypodermal precursors have high POP-1 levels whereas myogenic precursors have low POP-1 levels reflecting the actions of Wnt/MAP kinase signaling(Fig. 4).

To test the effects of Wnt/MAPK signaling in our experimental system, we repeated our ectopic PAL-1 activity experiments in embryos in which POP-1 levels were knocked down in all blastomeres by RNAi. Ectopic PAL-1 activity in these experiments was achieved using mex-3 RNAi. As predicted,reduction (or loss) of POP-1 greatly enhanced the ability of ectopic PAL-1 to activate HLH-1 and the muscle-like fate(Fig. 3). Muscle-like conversion in these treated embryos was comparable to embryos in which ectopic HLH-1 alone had been activated; almost all blastomeres adopted a muscle-like fate. Conversely, when we blocked the down-regulation of POP-1/TCF using RNAi knockdown of LIT-1/MAP kinase (or WRM-1/β-catenin; data not shown), we found that ectopic PAL-1 activated a hypodermal-like program in almost all somatic blastomeres (Fig. 5). To determine if the effects of Wnt/MAP kinase signaling were POP-1-dependent,we heat shock-induced ectopic PAL-1 in embryos depleted of both LIT-1 and POP-1 activity by RNAi. We assayed myogenesis using an integrated hlh-1::gfp transgene. As seen with pop-1 RNAi alone, lit-1 and pop-1 double RNAi, in combination with ectopic PAL-1, resulted in hlh-1::gfp reporter gene activation in most somatic cells for 84% (n=44) of the embryos. These results demonstrated that PAL-1 is myogenic and activates hlh-1 in somatic cells, provided that they lack POP-1 or that POP-1 has been down-regulated by Wnt/MAP kinase signaling. PAL-1-positive blastomeres with high POP-1 activity adopt a hypodermal-like fate (see Fig. 4).

Wnt/MAP kinase signaling has also been implicated in the formation of body wall muscle cells from the MS lineage from previous studies of lit-1(Kaletta et al., 1997). To explore this possibility further, we increased the number of MS-like blastomeres using mex-1 RNAi, which causes a transformation of AB granddaughter blastomeres to an MS-like lineage, in addition to other lineage defects (Mello et al., 1992; Schnabel et al., 1996). As a consequence of the reiteration of the MS lineage there is a large excess(four- to fivefold) of body wall and pharyngeal muscle that can be distinguished from each other using antibodies to MHC A and 3NB12,respectively. Consistent with previous studies(Kaletta et al., 1997),blocking MAP kinase signaling using lit-1 RNAi in a mex-1RNAi background resulted in a severe loss of body wall muscle and a large increase in pharyngeal muscle. To confirm that this effect was also WRM-1/beta-catenin dependent, we assayed mex-1, wrm-1 double RNAi embryos for muscle production and obtained similar results to those seen with mex-1, lit-1 double RNAi (Fig. 6). Most embryos had a small cluster of body wall muscle cells(∼10) near the posterior of the embryo; these are probably descendants of D, which should be largely unaffected by lit-1 or wrm-1 RNAi as D has no detectable POP-1 (Lin et al.,1998) and only low levels of POP-1 are detected in D descendants(data not shown). mex-1 RNAi perturbs both the C and D lineages, in addition to affecting AB (Schnabel et al.,1996), explaining why we did not observed the normal number (20)of D descendants. Our results confirm that Wnt/MAPK signaling is required to knock down POP-1 levels in cells destined to be body wall muscle within the MS lineage.

To determine if HLH-1 function was also dependent on Wnt/MAP kinase signaling, we compared ectopic myogenesis after heat shock induction of HLH-1 in a wild-type versus lit-1 RNAi background. Loss of lit-1gene activity had no effect on the ability of HLH-1 to induce widespread myogenesis, demonstrating that HLH-1 can function independently of Wnt/MAP kinase signaling (data not shown).

In all experiments in which ectopic PAL-1 activity, in concert with low or no POP-1 activity, resulted in a muscle-like fate, HLH-1 was activated and localized to the nucleus of myogenic cells prior to the expression of terminal muscle markers. Previous genetic studies have demonstrated that HLH-1 is not necessary for cells to adopt a body wall muscle fate during embryogenesis(Chen et al., 1992; Chen et al., 1994). To determine if the PAL-1-induced muscle-like fate was dependent on HLH-1, we repeated the ectopic PAL-1 experiments in an hlh-1 null mutant background. Hermaphrodites heterozygous for the balanced hlh-1(cc450)null allele (Chen et al.,1992) were treated with mex-3, pop-1 double RNAi and progeny embryos collected. If HLH-1 was required for PAL-1 to induce muscle,25% of the embryos (corresponding to the homozygous hlh-1(cc450)animals) should not respond to PAL-1. However, all treated embryos resulted in widespread body wall muscle-like myogenesis demonstrating that HLH-1 activity was not required for PAL-1-induced myogenesis(Fig. 7). We did notice that 22% (n=196) of the embryos showed less robust myogenesis, based on MHC A filament formation (Fig. 7B). These experiments were repeated using the temperature-sensitive hlh-1 allele cc561(Harfe et al., 1998a), or hlh-1 RNAi for which the genotype of each embryo was unambiguous and the same results were obtained (Fig. 7C); myogenesis occurred but was not as robust in the absence of HLH-1. These results demonstrated that ectopic PAL-1-induced myogenesis was slightly more robust with, but not dependent on, HLH-1.

The hallmark feature of myogenic regulatory factors (MRFs) is their ability to convert cultured non-muscle cell types into striated muscle-like cells. This study provides the first evidence in C. elegans that HLH-1 is myogenic and clearly demonstrates that this factor is a bona fide MRF. Induction of HLH-1 activity throughout the early C. elegans embryo is sufficient to convert most, if not all, somatic cells into a body wall muscle-like fate as assayed by the expression of several cell-type-specific markers.

The myogenic activity of HLH-1 is robust, possibly reflecting its ability to auto-activate its expression in a positive feedback loop(Krause et al., 1994). We find that induction of HLH-1 as late as the eight E cell (>100 total cells)stage of embryogenesis still results in almost all somatic cells adopting a muscle-like fate. Within the E lineage, several tissue-specific markers (e.g. end-1 and elt-2) are being expressed by the eight E cell stage, indicating that these cells have already initiated the intestinal cell fate (reviewed by Maduro and Rothman,2002). The over-expression of HLH-1 is able to extinguish the gut program and redirect these early intestinal cells into body wall muscle-like cells. This distinguishes our studies from previous work in which blastomere fate-switching was induced much earlier in development (two E cell stage)(Fukushige et al., 1998; Zhu et al., 1998). We did not observe cells expressing terminal markers of multiple fates, suggesting that cell fate decisions are mutually exclusive. How competing transcriptional factors result in all-or-none developmental fate decisions at the mechanistic level is an interesting and unanswered question. Regardless of the mechanism,the potency of HLH-1 reveals a remarkable level of developmental plasticity in cells that have already initiated a cell fate program; within the first 3 hours of development somatic blastomeres are not irreversibly committed to a single fate.

The decision to adopt a body wall muscle cell fate can be cell autonomous. The blastomeres adopting the muscle-like fate are not related by lineage and are not in a fixed location within the embryo. If exogenous signals are required for HLH-1-mediated myogenesis, they must originate in the germline precursors as that is the only non-muscle lineage that remains identifiable in these HLH-1-activated embryos. Such signals would have to be far-reaching to affect the most anterior blastomeres that do not physically contact the P cell lineage after the four cell stage. Consequently, we think it is unlikely that signals from outside the muscle lineage are required for HLH-1 to activate the muscle program, although we can not exclude a `community effect'(Gurdon, 1988) among muscle cells.

One question that arises from our current work is the extent to which HLH-1 is able to drive myogenesis. That is, are these muscle-like cells exhibiting most of the characteristics of terminally differentiated cells, or have they merely initiated a small part of the muscle program? Studies in both Xenopus (Hopwood and Gurdon,1990; Hopwood et al.,1991) and the mouse (Miner et al., 1992; Faerman et al.,1993) have demonstrated that ectopic MRF activity in vivo is able to activate some genes of the skeletal muscle program but fails to drive terminal muscle differentiation. However, more recent work has shown terminal differentiation of skeletal muscle after transfection of the chicken embryonic neural tube with Myf5 or MyoD expression transgenes(Delfini and Duprez, 2004). We have assayed five major markers of body wall muscle in C. elegans and found that all are present in the muscle-like cells generated by ectopic HLH-1. This includes gene products encoding structural components needed for terminal differentiation. However, these muscle-like cells lack clearly defined sarcomeres and contraction has not been observed. This may reflect the fact that these embryos also lack somatic non-muscle cell types that might be important for normal sarcomere assembly and function. This includes the hypodermal cells, which play a role in sarcomere organization (reviewed by Rogalski et al., 2001; Labouesse and Georges-Labouesse,2003) and neurons needed for innervation. Additional studies will be required to determine if the failure to make functional sarcomeres reflects the lack of an appropriate cellular environment, a failure in expression in all requisite muscle cell genes, or a combination of these factors.

Our results demonstrate that PAL-1 is sufficient to activate hlh-1and that this is part of the mechanism of body wall muscle development in the C and D lineages. Interestingly, this function of PAL-1 is not completely HLH-1 dependent, demonstrating that one or more factors must act redundantly with HLH-1 in driving body wall myogenesis. It is not clear if PAL-1 directly activates hlh-1. The hlh-1 promoter has been extensively characterized and essential cis-acting elements for expression delineated(Krause et al., 1994) (J. Liu,personal communication). In addition, the DNA-binding site preferences for Caudal and related factors have been defined in Drosophila and mammalian tissue culture studies (Dearolf et al., 1989; Suh et al.,1994; Charite et al.,1998; Xu et al.,1999). However, we have yet to uncover a direct interaction between PAL-1 and the hlh-1 promoter using either bioinformatic or experimental approaches. This analysis is complicated by the AT-rich binding site preferences of Caudal-related factors and our lack of understanding of which, if any, co-factors act in concert with PAL-1 to regulate transcription in C. elegans. This is an important question that needs to be answered in future studies.

The effects of PAL-1 on cell fate specification are altered in the presence of POP-1 activity. Cells lacking POP-1 (e.g. the D lineage) respond to PAL-1 by activating the body wall muscle program. However, in the presence of POP-1 activity, PAL-1 instead promotes hypodermal fate. Our results show that POP-1 activity must be down-regulated in cells that become body wall muscle in both the C and MS lineages. The down-regulation of POP-1 via Wnt/MAP kinase signaling is, therefore, an important component of embryonic body wall muscle development in all lineages except D. The source(s) of Wnt/MAP kinase signaling to descendants of the C lineage is unknown. However, once hlh-1 is activated within body wall muscle precursors, Wnt/MAP kinase signaling is dispensable.

Mechanistically, the combination of PAL-1 and Wnt/MAP kinase signaling in the C lineage acts in a manner that is analogous to E and MS founder fate specification. The transcription factor SKN-1 is required for both the MS and E founder fates that give rise to mesoectoderm and endoderm, respectively(Bowerman et al., 1992; Bowerman, 1995). In the absence of POP-1, SKN-1 initiates a gut cell fate, analogous to PAL-1 initiating a muscle-like fate. In the presence of high POP-1 activity, SKN-1 results in the MS fate, analogous to PAL-1 resulting in a hypodermal fate. In both cases,Wnt/MAP kinase signaling is responsible for the down-regulation of POP-1(Rocheleau et al., 1997; Thorpe et al., 1997; Lin et al., 1998; Meneghini et al., 1999).

The body wall muscle cells are the only striated musculature in C. elegans. All 81 embryonically born body wall muscle cells are arranged along the length of the animal in one of four parallel quadrants(Waterston, 1988). These cells are morphologically nearly identical to each other, making C. elegansbody wall muscle one of the simplest striated muscle systems under study. Despite this anatomical simplicity, these 81 cells arise by a surprisingly complex number of different genetic programs. There are at least two maternal transcription factors, SKN-1 and PAL-1, regulating embryonic myogenesis in a manner that is distinct from each other and distinct from the regulation of post-embryonic body wall muscle development(Bowerman et al., 1992; Hunter and Kenyon, 1996; Edgar et al., 2001). In addition, ablation experiments reveal a complicated interplay between different founder blastomere lineages in regulating myogenesis(Schnabel, 1995). Finally,blastomere culture experiments reveal that cell-cell interactions within the C lineage influence cell fate decisions(Mickey, 2000). Taken together, these studies reveal a surprising level of complexity for the genesis of an anatomically simple striated musculature.

Previous studies of the transcriptional regulation of body wall myogenesis in C. elegans have highlighted numerous differences between the vertebrate and nematode systems. In vertebrates, MRFs heterodimerize with members of the broadly distributed E protein family to activate transcription of muscle-specific genes (reviewed by Weintraub et al., 1991; Weintraub, 1993). In C. elegans, the only E-related factor, E/DA, is not detected in striated muscle cells and HLH-1 appears to function as a homodimer to activate transcription (Krause et al.,1997). In addition, members of the Twist and MEF-2 transcription factor families, which play important roles in vertebrate mesoderm specification and muscle differentiation (reviewed by Black and Olson, 1998; Castanon and Baylies, 2002),play little or no apparent role in embryonic body wall muscle development in C. elegans (Corsi et al.,2000; Dichoso et al.,2000). The transcriptional cascade regulating myogenesis in C. elegans embryogenesis is clearly distinct from that operating in the vertebrates.

However, the current study also highlights the similarities between nematode and vertebrate striated myogenesis. HLH-1 is a bona fide MRF with potent myogenic activity in vivo, demonstrating conservation of function throughout evolution. It would appear that function is redundant to other, as yet unidentified, factors in C. elegans and Drosophilawhereas vertebrates may rely predominantly on the redundancy provided by multiple MRFs.

The role of Wnt signaling in C. elegans body wall muscle specification also suggests parallels to other systems. During vertebrate embryogenesis, Wnts are needed for the dorsal neural tube and surface ectoderm to induce myogenesis in the adjacent somites in the trunk region(Munsterberg et al., 1995; Tajbakhsh et al., 1998). Wnt signaling is also important for activating myogenesis in resident stem cells during muscle regeneration following injury in mice(Polesskaya et al., 2003). During both of these vertebrate developmental events, Wnt signaling precedes the activation of the MRFs. C. elegans Wnt/MAPK signaling similarly precedes expression of hlh-1, suggesting that at least some aspects of the regulation of MRF genes may also be evolutionarily conserved.

We thank Julie Ahringer, Guy Benian, Andy Fire, Michel Labouesse, Rueyling Lin, Jim McGhee and Don Moerman for useful strains and reagents. Thanks to Andy Golden, Joan McDermott, Jim McGhee, Geraldine Seydoux and anonymous reviewers for comments on improving the manuscript.