Developmental Cell Article A Systematic Analysis of Tinman Function Reveals Eya and JAK-STAT Signaling as Essential Regulators of Muscle Development Ya-Hsin Liu,1 Janus S. Jakobsen,1 Guillaume Valentin,1 Ioannis Amarantos,1,2 Darren T. Gilmour,1 and Eileen E.M. Furlong1,* 1European Molecular Biology Laboratory, D-69117 Heidelberg, Germany 2Present address: Febit Biomed, Neuenheimer Feld 519, D-69120 Heidelberg, Germany *Correspondence: [email protected] DOI 10.1016/j.devcel.2009.01.006 SUMMARY different family members (Fu et al., 1998), or compensatory roles of other TFs (Habets et al., 2003). Mutations in a single copy of Nk-2 proteins are essential developmental regulators the human Nkx2.5 gene results in congenital heart defects due from flies to humans. In Drosophila, the family to altered cardioblast gene expression (Schott et al., 1998). member tinman is the major regulator of cell fate Although there are obvious interspecies differences in the within the dorsal mesoderm, including heart, visceral, phenotypes of Nkx2.5 mutants, this factor clearly has a and dorsal somatic muscle. To decipher Tinman’s conserved role in regulating cardioblast gene expression. direct regulatory role, we performed a time course Due to the striking parallels between the functions of Nk-2 genes in heart development in flies and mice, the role of the of ChIP-on-chip experiments, revealing a more prom- Drosophila tinman gene has been most extensively studied in inent role in somatic muscle specification than pre- this context. Here, Tinman acts together with the GATA factor, viously anticipated. Through the combination of Pannier, and T-box factors (Doc and mid family members) to transgenic enhancer-reporter assays, colocalization regulate cardioblast specification (Gajewski et al., 2001; Reim studies, and phenotypic analyses, we uncovered and Frasch, 2005), while, later in development, Tinman regulates two additional factors within this myogenic network: cardioblast diversification and heart function (Zaffran et al., by activating eyes absent, Tinman’s regulatory 2006). In contrast, Tinman is only required for the initial stages network extends beyond developmental stages and of visceral muscle development. Visceral muscle primordia are tissues where it is expressed; by regulating stat92E specified through the Jeb/Alk signaling cascade (Weiss et al., expression, Tinman modulates the transcriptional 2001) and the transcription factors Bagpipe (Azpiazu and Frasch, readout of JAK/STAT signaling. We show that this 1993) and Biniou (Zaffran et al., 2001), which act in a combinato- rial manner to coregulate a large group of target genes (Jakob- pathway is essential for somatic muscle development sen et al., 2007). As Tinman directly regulates the expression in Drosophila and for myotome morphogenesis in of both bagpipe and jeb (Lee and Frasch, 2005; Weiss et al., zebrafish. Taken together, these data uncover a 2001), loss of either one of these genes’ expression is sufficient conserved requirement for JAK/STAT signaling and to cause the loss of visceral muscle observed in tinman mutant an important component of the transcriptional net- embryos. work driving myogenesis. In comparison to the heart and visceral muscle, Tinman’s direct regulatory role in somatic muscle development remains very poorly understood. Drosophila somatic muscle or body INTRODUCTION wall muscle consists of a stereotypic pattern of 30 muscles in each hemisegment of the embryo, which can roughly be divided The Nk-2 transcription factors (TFs) are essential regulators of into three groups—the dorsal, lateral, and ventral muscles. All development in all species studied to date. In Drosophila, the dorsal muscles, and at least five muscles in the ventral and single Nk-2 gene, tinman, is essential for all tissues derived lateral regions (LL1, LO1, VL3, VL4, and VT1), are missing in from the dorsal mesoderm. tinman mutant embryos develop tinman mutant embryos due to an apparent block in myoblast no heart, trunk visceral muscle, or dorsal somatic muscle specification (Azpiazu and Frasch, 1993). However, the under- (Azpiazu and Frasch, 1993). Vertebrate genomes contain lying molecular basis for these defects is not understood. Tin- multiple family members with overlapping expression, making a man consensus binding sites are overrepresented in enhancers direct interspecies comparison of the role of Nk-2 TFs difficult. of genes expressed in somatic muscle founder cells, suggesting For example, loss-of-function mutations in the mouse Nkx2.5 a direct regulatory link (Philippakis et al., 2006). However, Tin- gene result in lethality at embryonic day 9.5 due to a block in man’s occupancy on these modules or requirement for their heart looping morphogenesis (Lyons et al., 1995). This is activity has not been demonstrated. Moreover, only 1 of the 11 a much later defect in cardiogenesis than that observed in characterized Tinman direct target genes (see Figure S1 avail- Drosophila, which may reflect functional redundancy between able online) is implicated in somatic muscle development. 280 Developmental Cell 16, 280–291, February 17, 2009 ª2009 Elsevier Inc. Developmental Cell Tinman Directly Regulates eya and JAK-STAT As an essential step to understanding how Tinman regulates scriptional cascade (Figure 1A and Table S3). One-hundred and diverse cell fates, we performed a time course of chromatin eighty targets have been previously characterized genetically immunoprecipitation followed by microarray analysis (ChIP-on- (Table S4); using a literature survey, these genes can be divided chip) to decipher its direct regulatory contribution in vivo. This into two broad classes based on their reported phenotype: class time course identified over 400 regions bound by Tinman at I genes are implicated in mesoderm or muscle development, one or more stages of mesoderm development. Here, we focus although, in the majority of cases, the regulation of their expres- on the regulation and function of a number of direct target genes sion is not known; class II genes are essential for some aspect in somatic muscle development, an understudied aspect of of development, but are not yet reported to play a role in myogen- Tinman function. These include a member of the Eya-Six4 family esis (Table S4). of transcriptional regulators, and components of the JAK/STAT The group of 87 class I genes demonstrates the specificity of signaling pathway. We show that JAK/STAT signaling is essen- Tinman binding, and provides a molecular basis for the observed tial for Drosophila myogenesis, and demonstrate a conserved defects in tinman mutant embryos. Examining their phenotypes requirement for stat activity in myotome morphogenesis in further revealed somatic muscle defects as the most frequent vertebrates. phenotype in these mutants (Table S4). This was unexpected, given the conserved role of tinman in heart development, and RESULTS AND DISCUSSION highlights the regulation of somatic muscle development as a prominent aspect of Tinman function. To explore this further, A Systematic Map of Tinman-Bound Enhancers we focused our analysis on the regulatory connection of Tinman and Direct Target Genes to a number of TFs and signaling cascades in this tissue. ChIP-on-chip experiments were performed at three consecutive We first examined Tinman’s role in regulating known compo- time periods, spanning the stages of mesoderm gastrulation to its nents essential for somatic muscle specification. Figure 1B subdivision into different muscle primordia. To minimize potential represents a literature survey of signaling pathways and TFs false positives due to off-target effects of the antibody, two inde- driving Drosophila somatic muscle specification. Tinman binds pendent anti-Tinman antibodies were used for the ChIPs. The to genomic regions in the vicinity of genes necessary for meso- genomic regions bound by Tinman were hybridized to microar- derm subdivision (components of the wingless and Dpp signaling rays consisting of overlapping 3 kb fragments tiling across 50% pathways), as well as myoblast specification (components of the of the Drosophila genome (Sandmann et al., 2006b). These expe- Notch-Delta pathway; genes outlined in red in Figure 1B; Table riments identified 780 significantly bound regions at one or more S4). Tinman also occupies enhancers of 10 out of 11 somatic developmental time period, with less than 1% estimated false muscle identity genes covered by our array. This in vivo binding, positives (Table S1). Due to the overlapping nature of the array, in combination with the loss of dorsal somatic muscles observed this represents 481 unique Tinman-bound regions, associated in tinman mutant embryos (Azpiazu and Frasch, 1993), indicates with 260 direct target genes (Table S2). that Tinman directly regulates an extensive transcriptional To date, 11 genes are known to be directly regulated by program driving somatic muscle specification within the dorsal Tinman (Figure S1). The loci of three of which were either not mesoderm. Tinman also targets loci of genes specifying lateral covered by our array (Toll, hand), or are regulated by Tinman and ventral muscles, suggesting a more general role in somatic very late in development, outside the time window assayed muscle specification throughout the entire mesoderm. (dSur). Of the remaining eight enhancers assayed, we identified The 93 class II genes have been phenotypically characterized Tinman binding to seven (out of eight) regions, demonstrating the for their role in some other aspect of Drosophila development