A novel role for lbx1 in Xenopus hypaxial myogenesis

The ladybird homeobox (lbx1) gene is a member of the NK-class of homeobox transcription factors. lbx1 and other NK-class genes are characterized by a homeobox DNA-binding domain and an engrailed homology domain (Jagla et al., 2001). In amniotes, lbx1 is expressed in a subset of hypaxial myoblasts,which in the mouse migrate into the limbs, tongue, and diaphragm. By contrast, lbx1 is completely excluded from inter-limb body wall hypaxial myoblasts (Dietrich, 1999; Dietrich et al., 1998). In addition to the difference in lbx1 expression, these two types of hypaxial muscles are characterized by different modes of development. Body-wall muscles form as epithelial extensions of the dermomyotome, whereas lbx1-positive limb, tongue and diaphragm myoblasts undergo an epithelial to mesenchymal transition in the dermomyotome before undergoing long-range migration to their target sites(Dietrich, 1999). A null lbx1 mutation in the mouse causes a specific lack of limb musculature attributed to migration defects, while hypaxial body wall muscles undergo normal development (Brohmann et al.,2000; Gross et al.,2000; Schafer and Braun,1999).

The pre-metamorphic X. laevis tadpole does not contain limbs and the entire family of pipid frogs does not develop a tongue(McDiarmid and Altig, 1999),suggesting that early tadpoles might not express lbx1. However, these tadpoles do contain hypaxial muscles, which form the rectus abdominus muscles(Lynch, 1990; Martin and Harland, 2001; Ryke, 1953). These are homologous to the inter-limb body wall muscles of amniotes(McDiarmid and Altig, 1999). The temporal expression of muscle-specific markers in myoblasts that form the ventral body wall muscles is similar to that of amniote limb myoblasts(Martin and Harland, 2001). pax3, which is a marker of proliferative myoblasts, is expressed in cells that exit the somite and migrate ventrally. Later on, pax3-positive myoblasts are seen just ventral to differentiated muscle of the ventral body wall. The expression of myf5, a muscle regulatory factor (MRF), is similar to that of pax3 in the hypaxial myoblasts. However, the MRF myoD is absent from hypaxial myoblasts as they exit the somite. Transcripts of myoD do not appear until cells begin to differentiate, and they overlap with 12/101 staining, a marker of differentiated skeletal muscle (Martin and Harland, 2001).

The expression of myoD has been shown to be associated with the terminal differentiation of skeletal muscle(Hopwood et al., 1989; Hopwood et al., 1992; Montarras et al., 1989; Tapscott et al., 1990; Weintraub et al., 1991). Indeed, myoD has been demonstrated to be an integral factor in cell cycle arrest during the terminal differentiation of skeletal muscle(Halevy et al., 1995). Although Myf5 and Myod1 have been shown to have partially redundant roles in mouse skeletal muscle formation, Myf5 is expressed in proliferative myoblasts that have not exited the cell cycle(Hopwood et al., 1991; Martin and Harland, 2001; Montarras et al., 1991). For example, the inhibition of muscle differentiation in chick limb myoblasts by activated Notch results in a downregulation of myoD, but has no effect on myf5 or pax3 expression(Delfini et al., 2000).

The precise timing of the development of long-range hypaxial muscles requires a fine balance between proliferation and differentiation(Amthor et al., 1999; Amthor et al., 1998). The overlying ectoderm of somites provides a proliferative signal for myoblasts. When ectoderm is removed from chick somites, a transient upregulation of myoD is observed corresponding to a burst of muscle differentiation,but further muscle growth is halted. When ectoderm is removed from somites at limb and tongue levels, the premature differentiation prevents muscle precursors from migrating, and limb and tongue muscles do not form(Amthor et al., 1999).

Work on lbx1 in hypaxial myogenesis has focused mainly cell migration, but lbx1 has also been implicated in cell proliferation. Forced expression of lbx1 in chick mesoderm and neural tube explants increases cell proliferation. In chick limb buds, lbx1 increases myoD expression and the amount of differentiated muscle. This excess can be rescued by inhibiting cell proliferation(Mennerich and Braun, 2001). These results suggest that a role of lbx1 in hypaxial myogenesis may be to expand the population of myoblast cells.

We have examined the expression pattern of lbx1 in X. laevis tadpoles and found that it is expressed in all of the myoblasts that will populate the rectus abdominus and the geniohyoideus, a hypoglossal muscle. We further demonstrate that lbx1 controls myoblast proliferation through the downregulation of myoD and p27. Thus, muscle defects observed after lbx1 loss of function probably result from the reduced capacity of myoblasts to proliferate.

Xenopus laevis embryos were generated and cultured by standard methods (Sive et al., 2000). Embryos were allowed to develop in 0.3× Marc's modified Ringer (MMR)solution and staged according to the normal table(Nieuwkoop and Faber,1967).

The primers TLBX-2 (5′-AGACGGCATGACGATTTTTGGC-3′) and TLBX-3(5′-CCGAGATGGTGAGTACGGCTTC-3′) were designed from Xenopus tropicalis genomic sequence and used to screen an arrayed X. laevis cDNA library. A single clone of lbx1 was found that contains the majority of the coding region. The primers Xlbx5up(5′-CCAAGTCAATGTATCAGTCGCTGTG-3′) and Xlbx5down(5′-TCCTGACTGAGGGCTTGTTTAGG-3′) were designed from X. tropicalis genomic sequence and used to amplify the 5′ end of the sequence from X. laevis genomic DNA.

A 0.25% stock solution of CellTracker CM-DiI (Molecular Probes) was made up in 100% ethanol. A working solution of 0.1% DiI was diluted in 3 M sucrose. Micropipettes were pulled and broken to a tip of ∼20 μm, backfilled with DiI solution and pressure injected using a Picospritzer (General Valve). Tadpoles were immobilized while injecting and for viewing by immersion in 0.05% benzocaine solution.

Embryos were allowed to develop until the desired stage and then fixed for 2 hours in MEMFA. In situ hybridization was carried out with RNA probes labeled with digoxigenin-UTP using a multibasket technique(Sive et al., 2000). Differentiated skeletal muscle was visualized with the 12/101 monoclonal antibody (Kintner and Brockes,1984). Immunohistochemistry used undiluted monoclonal hybridoma cell supernatant and a goat anti-mouse IgG conjugated to HRP or fluorescein(Jackson) as a secondary antibody at a 1:500 or 1:200 dilution, respectively. In cases where both in situ hybridization and 12/101 staining were carried out on embryos, in situ staining was performed first, followed immediately by immunohistochemistry. The anti-phospho-histone H3 antibody (Upstate Biotechnology) was used at a 1:1000 dilution in 2 mg/ml BSA in PBS plus 0.1%Triton X100. A goat anti-rabbit IgG secondary conjugated to HRP (BioRad) was used at a dilution of 1:1000.

Antisense and control MOs were ordered from Gene Tools. The following sequences were used: splice blocking,5′-GAGTGAGGAACTTACCTTCTGCTGC-3′; translation blocking,5′-TCATCTTTGGAAGTCATAGTGGGAC-3′; control,5′-CCTCTTACCTCAGTTACAATTTATA-3′; and control zebrafish lbx1 translation blocking,5′-TTTAGAGCTGGAGGTCATCTCAGTC-3′. Stock solutions were resuspended at 50 mg/ml. Initially, 17, 33 and 57 ng of antisense MO were injected into one cell at the two-cell stage. An optimal dose of 33 ng was determined and used subsequently.

RNA was extracted from whole tadpoles at stage 28 and used for RT-PCR(Wilson and Melton, 1994). Tadpoles used were untreated, injected with 33 ng of splice-MO into one out of two cells, or injected with 33 ng of MO into two out of two cells. The following primers were used: ef1α, U 5′-CAGATTGGTGCTGGATATGC-3′ and D 5′-ACTGCCTTGATGACTCCTAG-3′, 268 nucleotides; lbx1-intron,U 5′-TCCTAAACAAGCCCTCAGTCCG-3′ and D 5′-CCAACTCATAAATCTGGTGGTTCG-3′, 300 nucleotides (properly spliced). Twenty-one cycles were used to amplify for ef1α and 50 cycles for lbx1 intron.

Synthetic mRNA was made using the mMessage mMachine SP6 kit (Ambion). A zebrafish lbx1 I.M.A.G.E. EST (GenBank AL831789) was ordered from RZPD (Germany), subcloned from pSport to cs107, and digested with AscI for mRNA synthesis. Xenopus myoD was synthesized from the p3 plasmid (Rupp et al.,1994). Mouse Myf5 was synthesized from a full-length PCR insert in cs108 after being digested with AscI. Nuclearβ-galactosidase mRNA (pCS2-Nls-NlacZ, 100-200 pg) was co-injected with test mRNAs as a lineage tracer.

The synthesized mRNA was resuspended as a stock solution in DEPC-treated H₂O at a concentration of 1 mg/ml. Working solutions of lbx1,myoD, myf5, myoD + lbx1 and myf5 + lbx1 were diluted to 0.1 mg/ml in DEPC-treated H₂O. Approximately 4 nl of each mRNA solution was injected into one cell at the two-cell stage. Injections were targeted to the mediolateral region of the embryo.

The 5′ end of the zebrafish lbx1 EST (GenBank AL831789) was amplified with the primers LBXCLAI(5′-CCATCGATGGCGTATGAGGACTAAAGTTCGGGTG-3′) and MLBXR(5′-GACTTCTTAACGGAGAGAGGCTTGTTGG-3′). The 3′ end was amplified with the primers LBXECORI(5′-CGGAATTCCGCCTTGCATTTCAAGTTCTTCCGTG-3′) and MLBXF(5′-CCAACAAGCCTCTCTCCGTTAAGAAGTC-3′). The 5′ and 3′amplified fragments were gel purified and mixed together, followed by amplification using the LBXCLAI and LBXECORI primers. The resulting product was digested with ClaI and EcoRI and gel purified. This was then inserted into the ClaI, EcoRI sites of pCS107.

Tadpoles were embedded in 4% low melt agarose and sectioned using a vibratome (Oxford). The majority of sections were cut at a thickness of 100μm, except for sections in which the anti-phosphohistone H3 antibody was used, which were cut at 200 μm. The anti-phosphohistone H3 stained sections were cleared and mounted in wells cut into Sylgard coated slides and visualized using a Zeiss Axioplan microscope. Stained nuclei in the immediate vicinity of the somite through the entire 200 μm sections were counted using the microscope, and not from subsequent photographs. Sections at approximately the level of the 3rd trunk somite were used for counting. P values were obtained using the paired Student's t-test. Values less than 0.01 were considered to be statistically significant.

The 12/101 antibody, which marks differentiated skeletal muscle(Kintner and Brockes, 1984),was used to identify muscles of the X. laevis tadpole. At stage 37(Fig. 1A), the hypaxial rectus abdominus and rectus cervicus muscles can be clearly distinguished from those of the epaxial myotome (McDiarmid and Altig, 1999; Ryke,1953). By stage 45 (Fig. 1B), the cranial muscles can clearly be seen. The geniohyoideus muscles are considered to be of somite origin, and are present as a pair of muscles extending in an anterior-to-posterior manner near the mouth of the tadpole (Hammond, 1965; Hazelton, 1970).

PCR primers designed from Xenopus tropicalis genomic sequence were used to screen an arrayed X. laevis cDNA library. A single clone of lbx1 was found that contains the majority of the coding region minus the 5′ end. The 5′ end of the sequence was determined by PCR of X. laevis genomic DNA using sequence predictions from the X. tropicalis genome. X. laevis lbx1 contains two conserved domains typical of the NK class of transcription factors, including the engrailed homology domain (eh1) and homeodomain. These domains are identical to the corresponding regions of X. tropicalis, mouse and zebrafish lbx1. Overall, the amino acid identities are 94%, 79% and 63% between X. laevis lbx1 and X. tropicalis, mouse and zebrafish lbx1, respectively, while the nucleic acid identities are 88%, 66%and 61%, respectively. The mRNA expression pattern of lbx1(Fig. 1C-H) is first established at stage 17 in the neural tissue, and at later stages defines a dorso-intermediate region of the neural tube, consistent with interneuron specific expression in the mouse (Gross et al., 2002). Expression of lbx1 during gastrula and early neurula stages is not detectable by in situ hybridization (data not shown). Initiation of mesodermal expression can be seen at stage 26 in the ventrolateral region of anterior trunk somites(Fig. 1D, arrowhead). As development proceeds, expression expands posteriorly and ventrally, consistent with the expression of other genes that mark the developing hypaxial body wall(Martin and Harland, 2001). At stage 37/38, a thin stream of cells appears to be migrating into the head(Fig. 1G, arrow, Fig. 3C arrowheads). By stage 42, expression becomes very weak in the ventral domain of the body wall muscles (Fig. 1H, arrowheads)before it is eventually lost in this region.

In order to determine the fate of lbx1-positive cells that appear to migrate into the head, DiI was injected into the ventrolateral domain of the anterior trunk somites (approximately trunk somites 1-3)(Fig. 2A). Labeled cells were visualized at stage 37 (Fig. 2B) and 45 (Fig. 2C). DiI-positive cells can be seen moving ventrally and anteriorly away from the site of injection(Fig. 2B, arrowhead) at stage 37. The path of migration is similar to tongue myoblasts in other organisms,where cells migrate in a thin stream around the hyoid arch and into the head(Mackenzie et al., 1998). At stage 45, the final location of labeled cells in the head can be seen as a thin band (Fig. 2C, arrows). Co-localization of DiI- and 12/101-labeled cells indicates that the migrating population of cells coming from the anterior trunk somites gives rise to the geniohyoideus (Fig. 2D-G). The same population of cells was also found to give rise to the rectus cervicus muscle (data not shown). DiI-labeled cells only give rise to these muscles on the side of the tadpole in which the somites were labeled (control side not shown).

Rectus abdominus and other body wall muscles form as epithelial extensions of the dermomyotome in amniotes and fail to express lbx1(Dietrich, 1999). Because lbx1 is expressed in precursors of the rectus abdominus in X. laevis, we wanted to determine whether these myoblasts migrate in a manner similar to lbx1-positive long-range myoblasts of amniotes. lbx1-positive myoblasts are present at the somite clefts of a stage 37 tadpole, in the posterior-most region of lbx1 expression(Fig. 3A, arrows). This is also true for other hypaxial specific markers such as tbx3(Fig. 3B). In amniotes, the expression of hypaxial markers span the entire width of the somite in the body wall region as they extend as epithelia, whereas lbx1-positive zebrafish fin myoblasts emerge at somite clefts and migrate as mesenchymal cells (Neyt et al., 2000).

If lbx1 has a truly homologous role in X. laevis myoblast migration, we would expect it to be expressed in the long-range type migratory myoblasts such as those that migrate into the limbs and the head. Indeed, lbx1-positive myoblasts migrate into the head(Fig. 3C, arrowheads) in a region that co-localizes with 12/101 staining of the geniohyoideus muscles(Fig. 3D). In addition, lbx1-positive myoblasts are present in stage 53 limb buds(Fig. 3E, arrows) in a region similar to pax3-positive cells(Fig. 3F, arrows). Both genes are expressed in the dorsal and ventral regions, consistent with the development of limb muscle masses in other tetrapods(Uchiyama et al., 2000). Taken together, these results suggest conservation in the type of lbx1-positive myoblast migration in X. laevis, despite expression of lbx1 in myoblasts that populate the body wall musculature.

To determine whether lbx1 is required for hypaxial muscle development, a splice-blocking antisense MO was used to inhibit lbx1function. Embryos injected into one cell at the two-cell stage or into both cells at the two-cell stage with 33 ng of MO into each cell were analyzed by RT-PCR. The results indicate a clear decrease in properly spliced transcript(Fig. 4A). Quantification of the radiolabeled bands showed that the one-cell injection caused a 85%decrease in radioactive signal from the properly spliced transcript compared with the control, while the two-cell injection led to a 91% decrease. Embryos were injected into one cell at the two-cell stage with either 33 ng of the splice blocking MO or 33 ng of a control MO and then stained with 12/101. Those injected with the splice blocking MO exhibit a severe decrease in body wall muscle when compared with the uninjected side at stages 37(Fig. 4F,G) and 40(Fig. 4H,I) (87% with defects, n=197). In addition to body wall muscle defects, anterior ventral views of stage 40 tadpoles show a loss of the geniohyoideus muscle on the injected side (Fig. 4O, arrow). By contrast, the control MO injected tadpoles are largely unaffected on the injected side at stage 37 and 40 (Fig. 4B-E,N, arrow) (16% with defects, n=85). Apart from the control MO, which is famously non-toxic, other MOs, such as one directed against zebrafish lbx1 have never caused hypaxial muscle defects(data not shown). As a further control, a second MO was designed to block the translation of lbx1, in order to confirm the knockdown phenotype. When 33 ng of translation blocking MO was injected into one cell at the two-cell stage, the same loss of hypaxial muscle phenotype was seen at both stage 37 and 40 (Fig. 4J-M)(92% with defects, n=40). Later in development, tadpoles injected with either the splice or translation blocking MO exhibit some development of hypaxial muscles (Fig. 4I,M),but the amount of muscle is always significantly less than on the uninjected side (compare Fig. 4I,M with 4H,L). These results indicate that lbx1 is required for the proper development of hypaxial muscles that are derived from lbx1-positive myoblasts, similar to mouse lbx1 (Brohmann et al.,2000; Gross et al.,2000; Schafer and Braun,1999).

The lack of hypaxial muscles observed in lbx1 mutant mice is due to aberrant migration of myoblasts. The lack of body wall muscles in Xenopus tadpoles that have been injected with lbx1-splice MO can also be partially linked to abnormal migration of myoblasts. Fig. 4P-S shows stage 37 tadpoles that have been stained with pax3, a marker of proliferative myoblasts, and the 12/101 antibody, which marks differentiated skeletal muscle. Merged images showing both pax3 and 12/101 staining(Fig. 4Q,S) indicate the position of pax3-positive hypaxial myoblasts relative to the somites from which they were derived (12/101, green). In control tadpoles, pax3-positive cells have moved ventrally away from the somites(Fig. 4Q). In MO injected tadpoles (Fig. 4R,S), anterior myoblasts have begun to migrate away from the somites, but more posterior cells remain adjacent to the ventrolateral edge of the somites (80% with migration defect, n=25).

In addition to the loss of hypaxial muscle, the lbx1-splice MO also causes a specific loss of ventral epaxial musculature in the anterior trunk somites. A transverse section of a stage 40 tadpole that was injected with lbx1-splice MO in one cell at the two-cell stage illustrates this loss (Fig. 5A,B). After the initial onset of myogenesis in the epaxial domain of somites, the musculature of this region expands in the dorsal and ventral domains(Denetclaw et al., 1997; Denetclaw and Ordahl, 2000). This expansion is marked by myf5 expression in proliferative myoblasts at the ventrolateral and dorsomedial lips of the dermomyotome. We examined myf5 expression in injected tadpoles and found that it is specifically reduced in the ventrolateral region of anterior somites, but is expressed at normal levels in dorsal and posterior ventral domains(Fig. 5D, arrow) (96% with reduced myf5 expression, n=26). Injection of a control MO has no effect on myf5 expression(Fig. 5F) (n=6, one with slightly reduced myf5 expression). The first eight trunk somites express lbx1 in their ventrolateral domain(Fig. 1), which normally overlaps with myf5 expression in this region(Fig. 5G-J, arrows). When lbx1 mRNA is overexpressed in somites, a dramatic increase in myf5 expression is observed, as well as an enlarged somite(Fig. 5K) (100% with increased myf5 expression, n=13). Despite the increased size of the somite, there is a lack of differentiated muscle, as determined by 12/101 staining (Fig. 5L). The co-injection of lbx1 mRNA with the lbx1-splice MO rescues the loss of myf5 expression in the ventrolateral domain of the somite(Fig. 5M,N, compare with 5D),in addition to upregulating myf5 in other regions of the somite targeted by the injection (100% rescue, n=7).

Because of the enlarged somites with a lack of differentiated muscle in lbx1-injected tadpoles, we investigated whether lbx1 is controlling myoblast proliferation rather than directly upregulating myf5 expression. An antibody to phosphorylated histone H3 (pH3) was used to visualize mitotic cells (Saka and Smith, 2001). In normal, unmanipulated tadpoles, an abundance of mitotic cells are found at the ventrolateral region of developing somites,where lbx1-positive myoblasts are normally present(Fig. 6A-D, compare Fig. 6B with Fig. 5G).

The lbx1-splice MO was injected into one cell at the two-cell stage and stage 31 tadpoles were stained with the pH3(Fig. 6E) and 12/101(Fig. 6F) antibodies. A specific loss of mitotic cells is seen in the ventrolateral region of the somite on the injected side, as well as a slight loss of muscle in this region. The opposite is found in lbx1 mRNA-injected tadpoles(Fig. 6G,H), where a dramatic increase in mitotic cells is observed in the somitic region of injected tadpoles (Fig. 6G); this also corresponds to a lack of differentiated muscle(Fig. 6H). Injection of a control MO and β-galactosidase mRNA has no significant effect on the number of mitotic cells or the amount of differentiated muscle(Fig. 6I,J; results summarized in Fig. 10). At later stages in lbx1 mRNA injected tadpoles, mitosis decreases, corresponding to the terminal differentiation of these cells(Fig. 6K,L). In some cases, the excess differentiated muscle crosses the midline of injected tadpoles(Fig. 6L). By stage 41, the amount of differentiated muscle on the lbx1 mRNA-injected side of the tadpoles greatly exceeds the amount on the uninjected side, though ectopic muscle is never observed outside the somitic region. Fig. 6M,N shows transverse sections through a stage 41 tadpole at a trunk(Fig. 6M) and tail level(Fig. 6N). An increase in differentiated muscle is seen at both levels. This rules out the possibility that the increase in area of muscle in the section is due to distortion of the embryo (due to defective convergence and extension), and thus confirms that that increase in muscle mass is due to increased cell proliferation.

In many cases, exit from the cell cycle and terminal differentiation of myoblasts can be caused by the expression of myoD(Delfini and Duprez, 2004; Halevy et al., 1995; Hopwood and Gurdon, 1990; Pownall et al., 2002; Weintraub et al., 1991). We therefore investigated whether lbx1 may promote myoblast proliferation and lack of differentiation through the inhibition of myoD expression. One cell of each embryo was injected at the two-cell stage with lbx1 mRNA and analyzed at various stages. At stage 22, a strong downregulation of myoD expression is observed on the injected side (Fig. 7A), while pax3 expression is expanded (Fig. 7F). Transverse sections through stage 24 tadpoles show a complete loss of myoD expression on the injected side, despite the enlarged somite (Fig. 7B), again there is an expansion of pax3 expression on the injected side(Fig. 7G). As late as stage 33,the same results are seen. In regions where lbx1 is expressed (marked by red β-galactosidase staining), myoD expression is downregulated (Fig. 7C-E),while pax3 is expanded (Fig. 7H-J) (100% with reduced myoD expression, n=9;100% with increased pax3 expression, n=10).

Despite the rapid proliferation and pax3 expression in lbx1-injected cells, they do not have a hypaxial identity as shown by repression of endogenous lbx1. A stage 24 tadpole shows a loss of endogenous lbx1 expression on the injected side(Fig. 7L) as opposed to the uninjected side (Fig. 7K). At stage 33, the targeted mis-expression of lbx1 to posterior somites does not induce endogenous lbx1 expression(Fig. 7M). Other markers expressed in hypaxial muscles, such as tbx2 and tbx3 show similar results (data not shown).

If lbx1 normally represses myoD expression, the loss of lbx1 function should result in a transient increase in myoD. This result is observed in stage 26 tadpoles that were injected with the Lbx1-splice MO into one cell at the two-cell stage. A transverse section shows a slight increase of myoD expression on the injected side in the ventrolateral region of the somite(Fig. 7N, arrow) (64% with increased myoD expression on injected side, n=11). In addition, the repression of myoD caused by the ectopic lbx1expression should coincide with the presence of ectopic lbx1 mRNA. In a stage 31 tadpole, the presence of injected lbx1 mRNA can be detected by in situ hybridization (Fig. 7O). At this stage, myoD is repressed on the injected side (Fig. 7P). By stage 41,ectopic lbx1 mRNA can no longer be detected by in situ hybridization(Fig. 7R). This is also the stage at which a much larger differentiated myotome can be detected on the injected side (Fig. 6M,N). A corresponding increase in myoD expression is observed at this stage(Fig. 7S).

As previously mentioned, the expression of myoD can lead to exit from the cell cycle and terminal differentiation. In Xenopus, the cell cycle inhibitor p27 is expressed in developing somites and has been shown to be required for the proper differentiation of the myotome(Vernon and Philpott, 2003). Normal expression is found along the lateral edge of the somite where active differentiation is occurring. We examined the expression of p27 in lbx1-injected tadpoles. At stage 31, when injected lbx1 is present and myoD is repressed, p27 is also repressed compared to the uninjected side (Fig. 7Q, arrow indicates a region of strong expression on the uninjected side, which is absent on the injected side). At stage 41, when injected lbx1 mRNA is no longer present and myoD is upregulated on the injected side, p27 is also upregulated on the injected side, indicating that the cells that were overproliferating are now terminally differentiating (Fig. 7T, right side).

PCR-based mutagenesis was used to delete a 12 amino acid stretch that spans the entire eh1 domain. The over-expression of this mutant construct resulted in a slight dominant-negative phenotype. Instead of the decrease in myoD expression and increase in pax3 expression on the injected side observed after wild-type lbx1 expression, the opposite results were obtained. A transverse section through a stage 24 embryo shows an increase of myoD expression on the injected side(Fig. 8A, arrow) (90% with increased myoD on injected side, n=20), whereas pax3 is downregulated on the injected side of a stage 31 tadpole(Fig. 8B, arrow) (78% with reduced pax3 on injected side, n=9). The amount of cell proliferation does not appear to increase on the injected side of stage 29 tadpoles, as judged with the pH3 antibody(Fig. 8C,E). In addition, the myotome is abnormal on the injected sides of stage 29 tadpoles, having a shorter but wider shape (Fig. 8D,F, arrows).

If lbx1 is controlling myoblast proliferation through the downregulation of myoD, then the co-injection of myoD with lbx1 should eliminate the overproliferation of myoblasts. In Xenopus, the overexpression of myoD or myf5 in the somitic region causes an expansion of myotome through the recruitment of non-myogenic cell lineages, rather than an increase in the proliferation of myoblasts (Ludolph et al.,1994). In lbx1-injected tadpoles, there is an increase of mitotic cells (Fig. 9A), a decrease of differentiated muscle (Fig. 9B) and an expansion of pax3 expression(Fig. 9C). This is in contrast to the injection of myoD alone, which causes no change in mitotic cells (Fig. 9D), a large increase in differentiated muscle (Fig. 9E), and a slight decrease in pax3 expression(Fig. 9F). myf5injection causes a slight expansion of the myotome(Fig. 9H), but has no effect on cell proliferation (Fig. 9G) or pax3 expression (Fig. 9I). When lbx1 is co-injected with myoD, a phenotype that is the same as myoD alone is observed. There is an increase in differentiated muscle (Fig. 9K) with no increase in cell proliferation(Fig. 9J). pax3expression is also decreased on the injected side of the tadpole(Fig. 9L). Thus, lbx1cannot promote myoblast proliferation in the presence of myoD. However, co-expression of myf5 with lbx1 produces a phenotype similar to lbx1 alone, where a slightly smaller myotome is observed (Fig. 9N) but there is also an increase in cell proliferation(Fig. 9M) and pax3expression (Fig. 9O).

The summary of the numbers of phospho-histone H3 nuclei in the immediate somitic region of injected versus uninjected halves of tadpoles is shown in Fig. 10. Significantly more mitotic nuclei are observed on the injected sides of lbx1 and lbx1 + myf5 mRNA-injected tadpoles, while significantly fewer are observed on the lbx1-splice MO injected sides of tadpoles(paired Student's t-test, P<0.01. Fig. 10, indicated with an asterisk). All other injections shown, including the control MO +β-galactosidase mRNA-injected tadpoles, do not show a significant difference in the amount of mitotic nuclei on the injected side.

As previously described, the enlarged myotomes of myoD-injected tadpoles is the result of the recruitment of non-myogenic lineages, and is not due to increased cell proliferation(Ludolph et al., 1994). Transverse sections of stage 29 tadpoles that have been injected with myoD (Fig. 11A) or myoD + lbx1 (Fig. 11B), and stained with the pH3 antibody exhibit larger somites on the injected side but no increase in cell proliferation. In situ hybridization for pax8, a pronephric marker, shows that pronephric tubules are lost on the myoD (91% absent, n=11) or myoD + lbx1(80% absent, n=10) injected sides of stage 31 tadpoles(Fig. 11H,J). The pronephric tubules are present on the uninjected sides(Fig. 11G,I, arrows). However,transverse sections through tadpoles injected with lbx1 alone show normal expression of NCAM in the neural tube on the injected side(Fig. 11C,arrow), as well as pax8 in the pronephric tubules (Fig. 11D,arrow) (100% normal, n=8). Similarly, lateral views of a stage 31 tadpole injected with lbx1 show normal expression of pax8 in the pronephric tubules on both the uninjected(Fig. 11E) and injected(Fig. 11F) sides (arrows). These results, combined with the cell proliferation results of Figs 6 and 9, indicate that the enlarged myotomes of lbx1-injected tadpoles are the result of increased myoblast proliferation and not from the recruitment of non-myogenic lineages.

Expression of lbx1 in myoblasts that will form the rectus abdominus muscles in Xenopus laevis is different from its expression in other tetrapods; it is only found in prospective limb and tongue muscles in the mouse and chicken. In amniotes, a very clear distinction is made between myoblasts that undergo long-range migration into the limbs and head, and those that extend as epithelia to form the inter-limb abdominal muscles(Dietrich, 1999). This distinction is based both on the mechanism by which these cells populate target sites (mesenchymal versus epithelial) and by the genes that they express. In amniotes, lbx1 was the first molecular marker that was specific to the long-range type of migratory myoblasts(Jagla et al., 1995). The expression of lbx1 in both limb, head and body wall muscles in X. laevis suggests that the mechanism of epithelial extension to form body wall muscles has been lost or significantly modified in this animal.

The unique expression of lbx1 in body wall muscle precursors raises the issue of whether or not lbx1 plays a prominent role in the development of these muscles. The lbx1 mutation in mouse causes a severe loss, but not depletion of muscles in limbs(Brohmann et al., 2000; Gross et al., 2000; Schafer and Braun, 1999). Likewise, knockdown of lbx1 function in X. laevis using antisense MOs depletes body wall and hypoglossal muscles(Fig. 4). From these data, it appears that the long-range migration mechanism used in amniote and zebrafish limbs/fins is also used for body wall muscle development in X. laevistadpoles. In support of this idea, in situ hybridization patterns show that body wall muscle precursors migrate as mesenchymal cells rather than as epithelial extensions (Fig. 3). Thus, lbx1-positive myoblasts emerge from the somite clefts before moving on to populate the body wall muscle(Fig. 3A,B). This is similar to what is seen in the migration of zebrafish myoblasts into the fin, suggesting conservation in the type of migration(Neyt et al., 2000). Finally, X. laevis lbx1 is also expressed in hypaxial myoblasts that are traditionally considered the long-range type, namely limb and hypoglossal myoblasts (Fig. 3C,E). Taken together, these data support the idea that body wall myoblasts in X. laevis are migrating in a manner similar to limb/fin and tongue myoblasts of amniotes and zebrafish.

Several results support a role for lbx1 in mouse hypaxial myoblast migration (Brohmann et al.,2000; Gross et al.,2000; Schafer and Braun,1999; Uchiyama et al.,2000). Using antisense MOs to knockdown lbx1, we have observed a similar effect in Xenopus. However, we also found that lbx1 plays a central role in myoblast proliferation, and that this may be its primary role.

A proper balance between proliferation and differentiation is needed for hypaxial myoblast migration (Amthor et al.,1999; Amthor et al.,1998). In the chick, reducing the proliferative capacity of myoblasts by removing the ectoderm at limb and tongue levels causes a lack of limb and tongue musculature (Amthor et al.,1999). Furthermore, in mouse lbx1 mutants, some limb muscles form, in addition to tongue and diaphragm muscles(Brohmann et al., 2000; Gross et al., 2000; Schafer and Braun, 1999). The precursors to all of these muscles are normally lbx1 positive. Although they form, these muscles are reduced in size, suggesting that myoblasts have not proliferated to the proper extent. If lbx1 is only involved in the migration of myoblasts, one would expect there to be more severe defects in tongue and diaphragm muscles. If the starting population of hypaxial myoblasts has a reduced ability to proliferate, and some myoblasts differentiate at various points along the migratory path, than those sites farthest from the somite would exhibit the most severe defects. This is what it seen in mouse mutants, where the distal regions of limbs completely lack musculature, but the more proximal regions, as well as the diaphragm and tongue, have musculature, except in reduced sizes. This is also consistent with the lack of restriction of myoblast fates, such that muscle compartments are specified at the periphery and filled progressively by migrating myoblasts(Kardon et al., 2003).

The ability of myoblast cells to migrate may also require a critical mass of cells. The migration defects seen in lbx1 knockdown Xenopus tadpoles occur early, and recover in an anterior to posterior progression as the pool of myoblasts expands. Thus, at the time that pax3-positive myoblasts originating from trunk somite 4 may be stuck adjacent to the somite, those originating from trunk somite 1 have migrated ventrally (Fig. 4R,S). This may be the result of those cells having had a longer time to proliferate through non-lbx1 mediated means, such as pax3-dependent proliferation. This may also explain why in mice mutant for lbx1,tongue muscles, which originate from anterior somites, form, whereas hindlimb muscles, which originate from posterior somites, completely fail to form(Brohmann et al., 2000; Gross et al., 2000; Schafer and Braun, 1999).

Finally, evidence that lbx1 is not directly involved in myoblast migration comes from the overexpression of lbx1. When lbx1is overexpressed, there is no increase in number of migratory cells. In fact,there are fewer hypaxial muscles in these tadpoles (data not shown).

At early tadpole stages, Xenopus contain two types of skeletal muscle in the trunk. In the dorsal region, there is the dorsalis trunci,representing the epaxial domain. The ventral region contains the rectus abdominus, which is the hypaxial component of the trunk musculature(McDiarmid and Altig, 1999). In lbx1-splice MO-injected tadpoles, the ventral component of anterior epaxial muscle is reduced, with a corresponding loss of myf5staining in this region (Fig. 5B,D). This indicates that lbx1-positive myoblasts contribute to the formation of epaxial musculature. This is the first example of a case where lbx1-positive myoblasts contribute to both epaxial and hypaxial muscle, suggesting that the developmental programs of these two muscle types are more similar than previously suggested. The role of lbx1 in both types of muscle development also supports earlier arguments that the primary role of lbx1 is to promote myoblast proliferation, and not migration.

Previous work has shown that lbx1 can promote the proliferation of myoblasts in tissue explants (Mennerich and Braun, 2001). Here, we show that this is also true in the context of the whole embryo. When ectopic lbx1 mRNA was targeted to the somitic region, enlarged somites were formed. Initially, these enlarged somites contained a smaller amount of differentiated muscle(Fig. 6H). Later on, after the injected lbx1 had been depleted, the enlarged somites differentiated into a large amount of muscle (Fig. 6M,N). This was the first indication that lbx1 was controlling cell proliferation in the somite, as myoblasts can divide or differentiate, but not do both. However, lbx1 lacks the ability to induce myoblasts, as injections targeted outside the somite did not result in the formation of ectopic myoblasts or muscle (data not shown). The presence of expanded pax3 expression as well as a higher number of mitotic cells in lbx1-injected tadpoles indicates that lbx1 is a major regulator of cell proliferation (Fig. 9A-C).

In the chick, lbx1 overexpression in the limb region leads to an increase in myoD expression and limb muscle formation. This process requires cell proliferation, indicating that lbx1 may be expanding the population of myoblasts (Mennerich and Braun, 2001). We have found that the increase in myoDexpression in these experiments is the end result of a period of myoblast proliferation in which myoD is not expressed. The overexpression of lbx1 in the somitic region strongly inhibits the expression of myoD, so we suggest that the increase in myoD expression in the chick experiments may follow a period of myoblast proliferation in which myoD is not expressed. This would fit with the putative role of Lbx1 protein as a transcriptional repressor. In addition to having a homeobox DNA-binding domain, it also has an engrailed homology region, which has been shown to mediate transcriptional repression(Jagla et al., 2001). Indeed,our results demonstrate that after the injected lbx1 can no longer be detected by in situ hybridization, the intensity of myoD staining on the injected side is actually greater than the amount on the uninjected side(Fig. 7S), similar to what was seen in the chick experiments.

We found that while lbx1 represses myoD, and causes proliferation; the proliferative effect can be reversed by restoring myoD expression with injected mRNA (Figs 9, 10). Thus the enlarged somites caused by lbx1 expression are different in origin from enlarged somites caused by ectopic expression of myoD or myf5. The overexpression of myoD and myf5 in Xenopus embryos has previously been shown to produce enlarged myotomes, as the result of the induction of myogenesis in cells that would otherwise have a different fate(Ludolph et al., 1994). We confirmed this observation by showing that overexpressed myoD causes no change in the number of mitotic cells, a slight decrease in pax3expression, and an increase in differentiated muscle(Fig. 9D-F). When we co-expressed myoD and lbx1, the same result was obtained,indicating that lbx1 does not have the ability to control cell proliferation independent of myoD levels(Fig. 9J-L). This indicates that the control of cell proliferation by lbx1 is achieved through the inactivation of myoD expression.

In addition to the gain-of-function experiments showing that lbx1inhibits myoD expression, we also find evidence in MO loss-of-function experiments that links lbx1 to the inhibition of myoD transcription. If lbx1 has a direct role in repressing myoD, loss of lbx1 function should result in a transient upregulation of myoD. Stage 26 tadpoles that have been injected with the lbx1-splice MO exhibit an increase in myoD expression on the injected side (Fig. 7N),suggesting that the repression by endogenous lbx1 has been relieved.

We have also shown that the putative repressor domain of lbx1 is essential for its function. The removal of a 12 amino acid stretch that encompasses the entire eh1 domain changes the function of lbx1. When overexpressed, the mutant construct no longer inhibits myoD. Instead,a weak increase in myoD expression is seen on the injected side(Fig. 8A). Furthermore, rather than pax3 being upregulated on the injected side, it is inhibited(Fig. 8B). These results indicate that lbx1 normally functions as a repressor and that the removal of the eh1 domain produces a weak dominant-negative effect.

The opposite activity of lbx1^eh- relative to wild-type lbx1 indicates that it is acting as dominant negative. Contrasting with the lbx1-splice MO, the lbx1^eh- injections also have an effect outside of the normal lbx1 expression domain. This result suggests that lbx1^eh- interferes with the normal function of a protein other than lbx1. This may include another NK-class transcription factor, or perhaps even pax3, which is expressed throughout the entire dorsoventral axis of the somite. Pax3contains both a homeobox DNA-binding domain as well as an eh1 repressor domain, similar to lbx1. The Lbx1^eh- protein may bind to homeobox-binding sites and prevent the binding of endogenous homeobox containing proteins. Owing to the lack of the eh1 domain, it would not be able to repress the target gene. A similar phenotype to lbx1^eh-overexpression is observed when the bagpipe homologue koza is overexpressed in Xenopus (Newman and Krieg, 2002). Bagpipe is a member of the NK-class of transcription factors. Importantly though, koza contains a non-conservative amino acid substitution in the eh1 domain. This may in effect cause it to be a natural dominant-negative protein.

The cell cycle inhibitor p27 is expressed in developing Xenopus somites. When its function is reduced using antisense MOs,enlarged undifferentiated somites result, similar to lbx1-injected tadpoles (Vernon and Philpott,2003). These results indicate that p27 is a central player in the precise timing of myotome differentiation and that lbx1may inhibit its expression. We tested this by examining p27expression in lbx1-injected tadpoles. At stage 31, when injected lbx1 mRNA is still present, p27 expression is inhibited on the injected side (Fig. 7Q). At stage 41, when injected lbx1 mRNA is no longer detectable, p27 expression is enhanced on the injected side(Fig. 7T). These results are similar to myoD expression in the same tadpoles and indicate that cell-cycle inhibition is disrupted in lbx1-injected tadpoles.

The enhanced cell proliferation seen in lbx1-injected tadpoles suggests that this is the cause of enlarged somites, but there may also be a recruitment of non-myogenic lineages similar to what is seen in myoD-injected tadpoles (Ludolph et al., 1994). We examined tissues surrounding the somite in lbx1-injected tadpoles and found that both the neural tube and pronephros form normally (Fig. 11C-F). However, the pronephros is absent in both myoDand myoD + lbx1-injected tadpoles(Fig. 11G-J). These results demonstrate that the enlarged somites in lbx1-injected tadpoles are the result of enhanced cell proliferation and not from the recruitment of non-myogenic lineages.

We thank members of the Harland group, Marvalee Wake, and Sharon Amacher and members of her group for helpful comments and advice. This work was supported by the NIH (GM42341) and the Muscular Dystrophy Association(015164). B.L.M. was also partially supported by a NIH grant (GM061952) to Sharon Amacher.