Abstract

In mammals, Pax3, Six4, Six1 and Six5 genes are co-expressed with Eya1, Eya2 and Eya4 genes during mouse somitogenesis. To unravel the functions of Eya genes during muscle development, we analyzed myogenesis in Eya2-/− and in Eya1−/− embryos. A delay in limb myogenesis was observed between E10 and E13 in Eya1−/− embryos only, that is later compensated. Compound E18 Eya1−/−Eya2−/+ fetuses present a muscle phenotype comparable with that of Six1−/− fetuses; lacking a diaphragm and with a specific absence of limb muscles, suggesting either genetic epistasis between Six and Eya genes, or biochemical interactions between Six and Eya proteins. We tested these two non-exclusive possibilities. First, we show that Six proteins recruit Eya proteins to drive transcription during embryogenesis in the dermomyotomal epaxial and hypaxial lips of the somites by binding MEF3 DNA sites. Second, we show that Pax3 expression is lost in the ventrolateral (hypaxial) dermomyotomes of the somite in both Eya1−/−Eya2−/− embryos and in Six1−/−Six4−/− embryos, precluding hypaxial lip formation. This structure, from which myogenic cells delaminate to invade the limb does not form in these double mutant embryos, leading to limb buds without myogenic progenitor cells. Eya1 and Eya2, however, are still expressed in the somites of Six1Six4 double mutant and in splotch embryos, and Six1 is expressed in the somites of Eya1Eya2 double mutant embryos and in splotch embryos. Altogether these results show that Six and Eya genes lie genetically upstream of Pax3 gene in the formation of ventrolateral dermomyotome hypaxial lips. No genetic links have been characterized between Six and Eya genes, but corresponding proteins activate key muscle determination genes (Myod, Myogenin and Mrf4). These results establish a new hierarchy of genes controlling early steps of hypaxial myogenic commitment in the mouse embryo.

Keywords: Eyes absent/Eya proteins; Six/sine oculis homeoproteins; Pax3; OFC syndrome; Muscle; Somite; Hypaxial

Article Outline

Introduction

Results

Myogenesis is transiently delayed in Eya1−/− embryos

Compound Eya1−/− Eya2−/+ mutant embryos show a muscle hypoplasia similar to Six1−/− embryos

Absence of Eya1 leads to a decrease in Mef3 activity in the epaxial and hypaxial lips of the somites

Limbs of E13.5 Eya1−/−Eya2−/− embryos are muscle less

Six1 expression is maintained in the ventrolateral part of the dermomyotome hypaxial lips of Eya1−/−Eya2−/− embryos

Eya1 expression is maintained in medial and lateral dermomyotomes in Splotch and dSix mutants

Pax3 hypaxial enhancer is bound by Six1 protein in the mouse embryo

MRF4, Myod and Myogenin expression is diminished in Eya1−/−Eya2−/− embryos

Discussion

Migration

Hypaxial lip formation

Genetic loop and Eya/Six genes during evolution

Materials and methods

Mutant and transgenic mice

X-gal staining of mouse embryo

Histology and immunohistochemistry

Whole-mount in situ hybridization

Gel shift assays and chromatin immunoprecipitation

Acknowledgements

References

Introduction

Muscle regulatory factors (MRF) of the MyoD family orchestrate myogenesis in the embryo, and it has been shown that Six1 is required for MyoD and myogenin expression in the limb buds (Laclef et al., 2003). Six1 and Six4 are required for Mrf4, Myod and myogenin expression in the somite, while dispensable for the early somitic activation of Myf5 (Grifone et al., 2005). In the present study, we analyzed the involvement of Eya1 and Eya2 genes during myogenesis in Eya1−/−, Eya2−/− and Eya1−/−Eya2−/− gene-deleted embryos. We show that Eya1 and Eya2 genes have redundant functions during myogenesis and act genetically upstream of Pax3 in the formation of the hypaxial lip of the dermomyotome. We show that Six and Eya genes are activated independently in the ventrolateral part of somitic dermomyotomes and that induction of Pax3 in this region relies on biochemical interactions between Six and Eya proteins. Thus, synergy between Pax, Six and Eya genes in the hypaxial myogenic lineage of the mouse embryo is based on both genetic and biochemical interactions.

Results

Myogenesis is transiently delayed in Eya1−/− embryos

In order to understand the function of Eya1 and Eya2 proteins during mouse myogenesis, we analyzed the expression of several genes specifically expressed in the myogenic lineage in Eya1−/− or Eya2−/− embryos, at different development stages. It was shown previously that Eya1−/− fetuses die at birth due to the absence of rib cage closure and present several organ malformations in particular in the kidney and ear (Xu et al., 1999). Eya2−/− animals present no obvious external phenotype, and are viable and fertile. We crossed Eya1 and Eya2 heterozygous animals with either Six1−/+ animals (Laclef et al., 2003) (knock in allele with the β-galactosidase transgene at the Six1 locus) or with transgenic mice containing the Mlc3f-nlacZ-2E transgene which is specifically expressed in differentiated myogenic cells in the embryo (Kelly et al., 1995). X-gal staining of Eya2−/−Six1−/+ or Eya2−/− Mlc3f-nlacZ-2E embryos did not reveal major myogenic alterations between E10.5 and E18.5 (Fig. 1, and data not shown), and Eya2−/−Six1−/+ animals are viable and fertile. In Eya1−/− embryos on the contrary, analysis of Six1 expression or Mlc3f expression revealed transient myogenic alterations at the somitic level associated with a shorter ventrolateral extension of the dermomyotome/myotome (Fig. 1). Lack of significant Six1 downregulation in Eya1−/− embryos shows that Eya1 is not required to specifically activate Six1 at the somitic level (Fig. 1 and Fig. 3). Transient alterations of Six1 expression, detectable in the limbs until E14, cannot be attributed to a deficiency in the myogenic lineage, since Six1 and Eya1 are expressed in several cell types in the limb bud (Bonnin et al., 2005, Oliver et al., 1995 and Xu et al., 1997a). β-galactosidase staining revealed that the Mlc3f-nlacZ-2E transgene is normally expressed in the forelimb bud of E11 wild type or heterozygous Eya1−/+ embryos but completely silent in the forelimb bud of E11 Eya1−/− embryos, while myogenic Pax3-positive progenitors are present (see below). Furthermore, Mlc3f-nlacZ-2E expression is undetectable in ventral muscle masses of the hindlimb until E13, which is reminiscent of the specific loss of ventral muscle masses reported in Six1−/− embryos (Laclef et al., 2003). In Eya1−/− embryos, we also noticed transient lack of specific axial muscles like the spinotrapezius (Figs. 1G, H), which is reminiscent of the OFC pathology (Estefania et al., 2005 and Rickard et al., 2001) characterized by a hypoplasia of the cervical musculature. By E14, Mlc3f-nlacZ-2E expression is comparable in Eya1−/− and wild type embryos. No major muscle alterations are detected in Eya1−/− or Eya2−/− fetuses at E18, suggesting that these genes, which are coexpressed in the somite, have common functions during myogenesis. Lack of a myogenic phenotype in Eya2−/− embryos suggests either that Eya1 is expressed at a higher level than Eya2, and can compensate for the absence of Eya2, or that Eya2 is expressed later than Eya1, or that Eya2, although expressed in myogenic progenitors, does not control the same myogenic steps as Eya1 during embryogenesis.

Compound Eya1−/− Eya2−/+ mutant embryos show a muscle hypoplasia similar to Six1−/− embryos

To test the possibility that Eya2 and Eya1 have redundant functions, we analyzed double Eya1Eya2 mutant fetuses. We failed to obtain E18 Eya1−/−Eya2−/−dKO (from 40 fetuses analyzed on a mixed 129/C57bl genetic background), but we were able to analyze Eya2−/−Eya1−/+ and Eya1−/−Eya2−/+ fetuses. We did not detect major myogenic defects in E18 Eya2−/−Eya1−/+ fetuses, showing that one allele of Eya1 is sufficient for the formation of most skeletal muscles in the embryo (data not shown). Eya1−/− Eya2−/+ fetuses on the contrary have no diaphragm, and present a severe and selective limb muscle hypoplasia (Fig. 2, and data not shown). This muscle hypoplasia is reminiscent of that observed previously in Six1 null embryos, with remaining hypoplasic dorsal muscles in the distal hindlimb, absence of most ventral muscles and a quasi-absence of muscle in the forelimb (Laclef et al., 2003). These results show that Eya1 and Eya2 have redundant functions during myogenesis and that one allele of Eya1 is sufficient to drive myogenesis in specific/restricted areas of embryos. Comparison of Six1−/− and Eya1−/− Eya2−/+ fetuses suggests either genetic epistasis between Six and Eya genes and/or biochemical interactions between Six and Eya proteins in the myogenic pathway during embryogenesis.

Absence of Eya1 leads to a decrease in Mef3 activity in the epaxial and hypaxial lips of the somites

To test the possibility that Eya proteins can synergize with Six proteins on MEF3 DNA motifs during myogenesis in the embryo, we generated a transgenic line expressing a nls-lacZ transgene under the control of a promoter composed of a multimerized MEF3 element cloned in front of a TATA box of the aldolase A gene (MEF3-nlsLacZ transgene) (Grifone et al., 2004). In this context, X-gal staining revealed a conspicuous expression of the transgene from E10 in the epaxial domain of rostral somites in the transgenic embryos (data not shown). In E11 embryos, the transgene is activated in both the epaxial and hypaxial lips of the lombar and more rostral somites (Fig. 3 and data not shown), but is not detected in the anterior and posterior lips of the dermomyotome, and hypaxial lips of sacral and caudal somites, while Six1 and Six4 proteins (Fig. 3 and data not shown), and Eya1 and Eya2 mRNAs (Fig. 6) are expressed in all somites. These expression data suggest either that in these specific dermomyotomal somitic cells a higher level of Six cofactors is reached that can specifically activate transcription of the MEF3-nlsLacZ transgene, or that Six and their cofactors can be activated by post-translational modifications taking place specifically in epaxial and hypaxial dermomyotomal lips of those somites. This MEF3-nlsLacZ expression is maintained throughout embryogenesis in the muscles of the trunk and the head and later in adult trunk and head muscles, but is never detected in the myogenic precursors migrating into the limbs (analysis of four different transient transgenic embryos, and one transgenic line) (Fig. 3 data not shown; Grifone et al., 2004). In 4 out of the 5 MEF3-nlsLacZ transgenic embryos, β-gal activity was detected specifically in the dermomyotomal lips of the somite and head and trunk myogenic cells. To test whether Eya1 protein was required for this elevated MEF3 transcriptional activity level, we analyzed the MEF3-nlsLacZ transgene activity in an Eya1−/− background. While expressed in E11.5 and E13.5 Eya1−/+ embryos, the MEF3-nlsLacZ transgene activity was downregulated in Eya1−/− embryos to undetectable levels (Fig. 3). This result suggests that either (i) the absence of Eya1 precludes robust MEF3 activity in the epaxial and hypaxial lips, (ii) that the Six1 gene is no longer expressed in these lips or (iii) that these lips are not formed in Eya1−/− embryos. We found that Six1 is still expressed in dermomyotomal hypaxial lips of Eya1−/− embryos (Fig. 1 and Fig. 3), and we did not detect major morphological alterations of the hypaxial dermomyotomal lips of Eya1−/− embryos (Fig. 1 and Fig. 3). Altogether these observations rather suggest a specific defect in the recruitment of the strong coactivator Eya1 by Six homeoproteins on MEF3 elements (Grifone et al., 2004) in Eya1−/− embryos, impairing the transactivation of the MEF3-nlsLacZ transgene in the epaxial and hypaxial lips of the dermomyotomes. To ascertain that Eya1 is correctly produced in these structures in wild type cells, we further performed in situ hybridization that revealed a strong Eya1 expression in both epaxial and hypaxial somitic lips (Fig. 5), confirming the possible involvement of Eya1 protein in MEF3 activity. In conclusion, we show here that Six proteins are able to recruit Eya1 protein in the somites to produce a strong transcriptional complex able to activate gene transcription through the MEF3 sequence. Absence of Eya1 precludes high MEF3 reporter activity and slows down normal hypaxial myogenesis (Fig. 1).

Limbs of E13.5 Eya1−/−Eya2−/− embryos are muscle less

From the experiments presented above, Eya proteins stand out as important actors of myogenesis during mouse development. We attempted to analyze Eya1−/−Eya2−/− embryos at earlier stages of development, and obtained several E13.5 double Eya1−/−Eya2−/− Mlc3f-nlacZ-2E embryos. These mutant embryos showed major muscle deficiencies (Fig. 4). In particular, as observed earlier in double Six1Six4 deficient embryos, ventral hypaxial muscles are lacking at the trunk level, and no muscle is detected in the forelimbs and hindlimbs of double Eya1Eya2 deficient embryos. Since Mlc3f-nlacZ-2E is a marker of differentiated muscle fibers, we also tested the expression of desmin, which is expressed both in proliferating myoblasts and differentiated myofibers, and observed the same strong reduction of the number of desmin expressing cells in the hypaxial interlimb region and in the limbs, as observed with Mlc3f-nlacZ-2E detection (data not shown). Epaxial myogenesis on the other hand is relatively preserved in Eya1−/−Eya2−/− (dEya) null embryos (Fig. 4), as also shown in the past for Six1−/−Six4−/− (dSix) null embryos (Grifone et al., 2005).

Six1 expression is maintained in the ventrolateral part of the dermomyotome hypaxial lips of Eya1−/−Eya2−/− embryos

We have shown that myogenesis is severely impaired in dEya Knockout (KO) embryos. This phenotype is similar to the one obtained in dSix KO embryos, in which Pax3 expression is abolished in the hypaxial somitic territories anterior to the hindlimb, resulting in altered hypaxial myogenesis. We now show that Pax3 expression is also severely reduced in the hypaxial lips of most somites of Eya1−/−Eya2−/− embryos, while Pax3 expression is not affected in the posterior lips of the dermomyotome (Fig. 5). This similarity between the muscle phenotype of the dEya KO and dSix KO, prompted us to check for Six1 expression in dEya KO embryos. Interestingly, Six1 is still expressed in somites of dEya KO embryos (Fig. 5), showing that Eya1 and Eya2 genes are not required to activate Six1 transcription in the ventrolateral dermomyotome. Thus, absence of Pax3 expression in epaxial and hypaxial lips of dEya KO embryos is not the result of a decrease in Six1 expression. This demonstrates that Pax3 cannot be activated by the Six1 protein in the absence of Eya proteins and that Six and Eya genes are both required upstream of Pax3 in the somitic medial and lateral lips. At the hindlimb level, we also noted that somitic dermomyotomal ventrolateral cells that express Six1 in the dEya mutant are rerouted ventromedially (Fig. 5). Thus, instead of migrating in the limb as myogenic progenitors, these cells which express Six1, but not Eya and not Pax3, lost their identity and fail to migrate correctly and to activate the myogenic program, that is reminiscent of the behavior of somitic dermomyotomal ventrolateral cells in dSix mutants (Grifone et al., 2005).

Eya1 expression is maintained in medial and lateral dermomyotomes in Splotch and dSix mutants

To explore the possibility that Six and Eya genes control each other specifically in the hypaxial lips of somitic dermomyotomes, and to test whether one gene is genetically upstream of the others, we assayed the expression of Eya1 and Six1 in Pax3−/− embryos and the expression of Eya1 and Eya2 in Six1−/−Six4−/− embryos. As shown on Fig. 6, Eya1 and Eya2 expression is maintained in splotch and in the dSix KO in the ventrolateral somites as well as in the mediolateral somites, while Eya expression is reduced at the limb bud level (Xu et al., 1997a). While absence of Pax3 precludes the formation of a hypaxial lip structure (Goulding et al., 1994) (which is therefore absent in dEya and in dSix Ko embryos; Grifone et al., 2005), we observe in splotch as well as in dEya and in dSix mutant embryos, the formation of an epaxial lip structure, showing that formation of this lip is not under the control of these genes. This conclusion is further supported by the presence of epaxial muscles in all of these mutants. These results suggest that Eya1 is activated in ventrolateral dermomyotomal cells by surrounding tissues, independently of Pax3 and Six proteins. Eya1 expression does not seem to be upregulated in the posterior lips of dSix null embryos contrary to Pax3 expression which is upregulated in this structure in dSix KO embryos as well as in dEya KO (Fig. 6). Table 1 summarizes the expression of Pax3, Six1 and Eya1 genes in ventrolateral dermomyotomes in different Pax/Six/Eya mutant embryos.

Pax3 hypaxial enhancer is bound by Six1 protein in the mouse embryo

It was shown recently that a 291 bp fragment of the Pax3 gene contains a somitic hypaxial enhancer able to drive the expression of a β-galactosidase transgene in most hypaxial lips of the mouse somites (Brown et al., 2005). To assess the possibility that Pax3 expression could be directly under the control of Six1 protein, we checked for the presence of a conserved MEF3 site in this enhancer sequence. A TtAGaTTTC sequence was present in Pax3 hypaxial enhancer, which is very close to the MEF3 TCAGGTTTC consensus. Gel mobility shift assays performed with in vitro translated Six1 protein shows that Six1 is able to recognize this site (Fig. 7, and data not shown). Chromatin immunoprecipitation experiments performed with Six1 antibodies on E11 trunk and limb embryos confirmed the binding of Six1 protein on the Pax3 hypaxial enhancer in vivo (Fig. 7). These results in keeping with the in vivo results (Fig. 5, Grifone et al., 2005) confirm that Pax3 hypaxial lip expression is under the control of a Six1 transcriptional complex binding to the MEF3 site present in its specific hypaxial enhancer.