The Unc-5 Receptor Is Directly Regulated by Tinman in the Developing Drosophila Dorsal Vessel

During early heart morphogenesis cardiac cells migrate in two bilateral opposing rows, meet at the dorsal midline and fuse to form a hollow tube known as the primary heart field in vertebrates or dorsal vessel (DV) in Drosophila. Guidance receptors are thought to mediate this evolutionarily conserved process. A core of transcription factors from the NK2, GATA and T-box families are also believed to orchestrate this process in both vertebrates and invertebrates. Nevertheless, whether they accomplish their function, at least in part, through direct or indirect transcriptional regulation of guidance receptors is currently unknown. In our work, we demonstrate how Tinman (Tin), the Drosophila homolog of the Nkx-2.5 transcription factor, regulates the Unc-5 receptor during DV tube morphogenesis. We use genetics, expression analysis with single cell mRNA resolution and enhancer-reporter assays in vitro or in vivo to demonstrate that Tin is required for Unc-5 receptor expression specifically in cardioblasts. We show that Tin can bind to evolutionary conserved sites within an Unc-5 DV enhancer and that these sites are required for Tin-dependent transactivation both in vitro and in vivo.


Introduction
Early stages of heart development, both in vertebrates and invertebrates, include the migration of bilaterally paired condensations of cardiac precursors and the formation of a linear tube. The tube is formed once these symmetrical groups of mesodermal cells meet, and attach to each other leaving a luminal space between them [1,2]. The coordinated migration of these mesodermal cells, bilateral interaction and the preservation of a lumen require complex interactions of multiple guidance receptors in Drosophila during DV morphogenesis [3][4][5][6][7][8]. Vertebrate homologs of the same ligand/receptor systems are expressed in the developing heart in many cases with strikingly similar patterns to the ones present in Drosophila [9,10]. Some, like the Robos and their Slit ligands [11,12] or plexins and semaphorins [13,14], have also been identified as key players at different stages of heart development. Nevertheless, how these guidance systems are regulated in place and time during heart morphogenesis is widely unknown.
Cardiogenesis in both vertebrates and invertebrates also requires the key regulatory actions of a core of evolutionarily conserved families of transcription factors (NK2, GATA, and T-box) [15]. They are required early in development during linear tube formation and function again at later stages of heart morphogenesis in vertebrates [16]. For example, Nkx2-5 members and its Drosophila homolog, Tinman (Tin), play an important role in early cardiogenesis starting with the specification of cardiac precursors to remodeling and functionality of the adult heart [1,17]. Given the role of guidance systems in heart morphogenesis, it is likely that they are direct or indirect targets of these families of transcriptional regulators.
To gain a better understanding if these transcription factors (TFs) control heart tubulogenesis through the regulation of guidance receptors, we have studied Unc-5 receptor's regulation in the Drosophila dorsal vessel (DV). The DV develops from mesodermal cardiac precursors. After precursor division, heart cells line up bilaterally into two rows where myocardial cells or cardioblasts (CBs) are positioned dorsally and pericardial cells (PCs) ventrolaterally. Finally, they migrate together towards the dorsal midline of the embryo where CBs fuse to form the tubular heart ( Fig 1A). The CBs will constitute the pumping myocardium and PCs the pericardium. The Drosophila Unc-5 receptor is a repulsive receptor for Netrin A and B [18,19] and in the nervous system has been shown to be required for motoneuron guidance [20,21] and glial migration [22]. Unc-5 is expressed in both major cell types present in the DV and has recently been shown to be required in late dorsal vessel morphogenesis for lumen formation [7,8]. tin is also expressed widely in PCs and most CBs ( Fig 1A). Furthermore, tin and Doc, a Tbox family TF, have been shown to regulate together an early cardiac mesoderm Unc-5 enhancer [23]. In this work we show that tin is specifically required for Unc-5 expression in CBs and is sufficient to induce its expression ectopically in the ectoderm. We identify a unique DV enhancer within the Unc-5 regulatory region that fully recapitulates its expression at late stages of DV fusion. Cardioblast specific expression through this enhancer is strictly dependent on tin as is misexpression in the ectoderm. Additionally, Tin can induce transcription in vitro in a luciferase assay through the Unc-5 DV enhancer but not from other known Unc-5 enhancers. Using ChIP analysis we identify three evolutionary conserved Tin-binding sites within this enhancer that are required in vitro for its activity. Finally, we demonstrate that these sites are the Tin-binding sites required in the Unc-5 DV enhancer for its ectopic regulation in the ectoderm and, more importantly, its specific expression in cardioblasts. Thus, Our work shows how tin regulates Unc-5 receptor expression during late heart tube morphogenesis when Unc-5 is required for lumen formation. Our results provide a regulatory mechanism for a guidance receptor through a direct interaction with three conserved sites within its DV enhancer by one of the core transcription factors during tubulogenesis of the Drosophila DV.

Tin is required for Unc-5 expression in the dorsal vessel
The Unc-5 receptor localizes preferentially at the luminal side in CBs at the onset of tubulogenesis and it is required to preserve the luminal space between CBs [7,8]. However, how this receptor is regulated at this late stage is not known. Genome-wide chromatin immunoprecipitation screens to identify Tin target genes in cardiac mesoderm and cardiac precursors have identified Unc-5 as one of its targets [23,24]. Previous studies have established that early cardiac specification in Drosophila is dependent on the homeobox transcription factor Tin [2,25,26].
As a consequence, tin loss-of-function mutants lack a DV due to its early role in the mesoderm to specify cardiac progenitors [26,27]. Nevertheless, in cardiac specific tin mutant animals (tin-ABD;tin 346 /tin 346 ), where tin is re-expressed in a mutant background under the control of enhancer elements (ABD) recapitulating its entire endogenous expression pattern except in the dorsal vessel. Myocardial cells are specified and the DV forms in these mutants [17]. To determine whether tin regulates Unc-5, we analyzed its expression in tin mutant DVs (Fig 1B and  1C). Unc-5 mRNAs is significantly reduced when compared with tin heterozygous DVs (compare 1B with 1C). Thus, tin is required for Unc-5 expression in the DV at the onset of tubulogenesis.

Tinman is sufficient to induce Unc-5 expression ectopically
We further tested tin's sufficiency to induce Unc-5 transcription in vivo by misexpressing tin in ectodermal stripes with an engrailed-Gal4 (en-Gal4) driver. We detected Unc-5 mRNA through in situ hybridization and we confirmed that, indeed, ectopic tin expression in the ectoderm is sufficient to induce Unc-5 in the characteristic engrailed stripes (Fig 2B and 2B') where neither tin nor Unc-5 are normally not expressed (Fig 2A and 2A'). Thus, tin is not only required for Unc-5 expression in the dorsal vessel but it is also sufficient to induce its expression in other tissues.

Identification of the Unc-5 cardiac enhancer
In order to identify regulatory regions required for Unc-5 expression in the DV, we dissected the D. melanogaster Unc-5 genomic locus into overlapping fragments of varying length starting from the preceding gene (Hr51) to the 5th intron within the Unc-5 locus [28] (Fig 3A). All the fragments were fused to a GFP ORF and inserted into the same locus to avoid any variability due to position effect. We identified a unique 1kb minimal fragment upstream of the Unc-5 ATG sufficient to drive GFP expression in the DV at late stages of PCs and CBs migration and during tube formation (stage 13 onwards, Fig 3A) largely overlapping with the early mesoderm enhancer previously described [23]. To characterize the expression pattern of this enhancer we co-stained embryos carrying the DV enhancer driving GFP (GH reporter) with anti-GFP and markers for PCs or CBs, Mef2. The reporter drives GFP expression in all Tin-expressing CBs and PCs (Fig 3B-3D") including Eve-and Odd-positive PCs (Fig 3C-3C") and the Tin-negative Seven Up (Svp)-expressing myocardial cells (SMCs, Fig 3D-3D"). Finally, to confirm that the enhancer faithfully recapitulates Unc-5 endogenous expression in the Drosophila DV we also performed double labeling of the cells expressing the reporter and Unc-5 mRNA by in situ hybridization. Our data shows that the Unc-5 mRNA expression pattern in the DV fully matches that of the enhancer (Fig 3E-3E").   [29]), PCs: all PCs with Zfh1 (B and B", [30,31]) or subsets with Eve and Odd (C-C", [32,33]); and a Svp-LacZ reporter for labelling a set of Tin-negative myocardial cells, also known as Seven Up (Svp)-positive myocardial cells (SMCs) (D-D", [17,34]). GFP expression (green) is present in all CBs (B'; magenta) and PCs (B"; magenta). (C-C") Odd-(C'; magenta) and Eve-positive (C"; magenta) PCs express GFP (green) driven by Unc tin is required for activation of the Unc-5 DV enhancer in cardioblasts Since tin is required for Unc-5 expression in the DV (Fig 1B and 1C). We speculated that tin might exert its regulation through the unique DV enhancer we identified within the Unc-5 genomic locus. Therefore, we examined GFP expression driven by the Unc-5-GH enhancer in tin-ABD;tin 346 mutant embryos at later stages of embryonic cardiogenesis (Fig 4). GFP expression driven by the enhancer was virtually absent in CBs from tin-ABD;tin 346 embryos (from 1.98 ± 0.128 in wild-type CBs to 0.173 ± 0.04 in tin-ABD;tin 346 CBs, P<6 x 10 −14 , Fig 3D). However, GFP expression was still present in PCs (Fig 4B-4B"), where it was slightly reduced (from 3.4 ± 0.2 in wild-type to 2.47 ± 0.21 in tin-ABD;tin 346 , P<6 x 10 −4 , Fig 3E). Importantly, GFP expression from the GH enhancer in a GFP-positive subpopulation of sensory neurons (SNs) where tin is not expressed nor required was not affected in tin-ABD;tin 346 mutants (3.86 ± 0.4 in wild-type and 3.82 ± 0.29 in tin-ABD;tin 346 , Fig 4A" and 4B" [arrowheads], D, E). Furthermore, SMCs, where Tin is not normally expressed, accordingly, still expressed GFP in tin-ABD;tin 346 mutants (S1A-S1A" Fig). These results indicate that tin specifically regulates expression through the Unc-5-GH enhancer in CBs at the developmental stage when they fuse to form the heart lumen.

Tin binds the Unc-5 DV enhancer at three conserved Tin-binding elements
To determine if Tin is sufficient to induce transcription through the Unc-5-GH enhancer we fused different Unc-5 enhancers [28] to renilla luciferase ORF and co-transfected them with tin in Drosophila S2 cells. The only fragment responsive to Tin in this assay was the one containing the GH enhancer (Fig 5A). Therefore, this enhancer is not only under the control of Tin in vivo in the DV, but also responded to it in vitro. Our assay results also suggested that regulation of this enhancer was very likely mediated through a direct interaction with Tin. To test this hypothesis, we performed chromatin immunoprecipitation (ChIP) followed by qPCR using overlapping primers covering the Unc-5-GH enhancer. Our ChIP-qPCR results identified a unique enrichment peak near the 3' end of the Unc-5-GH enhancer covering three consecutive amplified regions (referred to as R8, R9, and R10, Fig 5B). Further analysis of the sequence within the peak revealed three potential binding sites on the Unc-5-GH enhancer that closely match the described consensus Tin-binding sequence [23,24,35]. These motifs are conserved in Drosophila species with a divergence time > 10 7 years, highlighting their functional Schematic of a few enhancers within the 5' region of Unc-5 used to make luciferase constructs used in the luciferase reporter assays. Luciferase activity was normalized to Firefly activity and the only construct presenting activity corresponds to the Unc-5-GH element (magenta). (B) ChIP analysis of the Unc-5-GH locus in S2R+ cells transfected with pAct5C-GFP-tinman. The precipitated DNA was amplified by real-time qPCR using overlapping primers (boxes on the X axis of the graph) designed to fully cover the identified GH enhancer element (magenta line). Enrichments are presented as percentages of total input and error bars represent the standard deviation. ChIP signal is schematically outlined as a curve peaking at R8, R9, and R10. A schematic of the Unc-5 locus is also illustrated below the graph. (C) Alignment of these regions against the 12 sequenced Drosophila species reveals complete evolutionary conservation of the Tin-binding motifs in R8, R9 and R10 regions of Unc-5-GH enhancer (highlighted in red). relevance ( Fig 5C). Thus, our ChIP data has identified evolutionary conserved Tin-binding sites within the Unc-5-GH enhancer that are likely required for its regulation by Tin.

Tinman regulates the Unc-5-GH enhancer in vitro through the conserved Tin-binding elements
In order to determine the requirement of the identified sites to promote Tin-mediated transcription, we compared the transcriptional activity of the wild-type Unc-5-GH enhancer with constructs where each site is changed alone or in combinations (Fig 6). Our in vitro luciferase assay results revealed that mutagenesis of each site lead to reduced transcriptional activity, further confirming that Tin regulates the DV enhancer and its interaction with the conserved binding sites is required to induce Unc-5 transcription.

Tinman activity in vivo is mediated through its binding elements on the GH enhancer
Given that tin is sufficient to induce Unc-5 expression in ectodermal stripes (Fig 2A-2B'), if this regulation is mediated through the DV enhancer, it should also be sufficient to induce ectopic transcription from the GH reporter. As expected, misexpression of the reporter was observed in the tin-misexpressing ectoderm (Fig 7A-7A" and 7E). Thus, tin is sufficient to induce Unc-5 expression from its endogenous locus or from a reporter containing the Unc-5-GH enhancer in vitro or in vivo. As the Unc5-GH enhancer is regulated directly by Tin in vitro ( Fig 5) and in CBs in vivo (Fig 4), we reasoned that it might be mediated through the three identified Tin-binding sites in the GH enhancer (Fig 6). To verify this requirement, we misexpressed tin in en stripes in the presence of a mutant reporter with all three binding sites mutated (R8,9,10-GH). While the wild-type reporter is ectopically expressed in en stripes our  Tin expression in ectodermal stripes is labeled with anti-Tin antibody (magenta). Anti-GFP antibody was used to reveal the expression of the reporter (green). As expected, embryos carrying the R8,9,10-GH mutant reporter display little or no GFP induction in the stripes (B-B"; arrowhead-asterisks, and E). (C-C") The wild-type Unc-5-GH enhancer induces expression of the GFP reporter (green) in all CBs and PCs. Mef2 (red, D or magenta, D') and Zfh1 (blue, D or magenta D") antibodies are used to reveal CBs or PCs, respectively. The R8,9,10-GH enhancer generates a GFP expression pattern similar to that of the wild-type Unc-5-GH enhancer in tin-ABD; tin 346 /tin 346 embryos (Fig 4B-4B") with near complete loss of GFP expression in Tin-positive CBs (D' and F) and a reduction of expression in PCs (D"). (E and F) Quantification of GFP expression by the mutated Unc-5 enhancer (R8,9,10-GH) in ectodermal stripes (E) and CBs (F). Genotypes are indicated on the X axis and fluorescence intensities on the Y axis. For all quantifications GFP expression in sensory neurons (SNs) was used as internal control, as the fluorescence in these cells is not affected. In E, fluorescence is significantly reduced (p<1.2 x 10 −14 ) in engrailed stripes of embryos with the mutant reporter compared to those of embryos carrying the wild-type reporter [from 2.8 ± 0.26 s.e.m. for the wild-type Unc-5-GH reporter to 0.25 ± 0.16 s.e.m. for the results revealed little or no activity in embryos with the R8,9,10-GH mutant reporter (compare Fig 7A-7A" with Fig 7B-7B" and 7E). Therefore, the ability of Tin to regulate Unc-5 in vivo, in the ectoderm, is strictly dependent on the conserved Tin-binding sites identified in vitro. Based on these observations we predicted that Tin regulates Unc-5 through a direct binding to these sites also in CBs. Indeed, GFP expression from the R8,9,10-GH mutant enhancer was also absent from CBs (compare Fig 7C-7C" with Fig 7D-7D" and 7F), indicating that these sites are required by Tin to regulate Unc-5 in CBs. As our internal control we also determined that reporter expression was not affected in cells that never express nor require Tin such as SNs ( Fig  7E and 7F). Together, our results demonstrate that Unc-5 is regulated by tin in cardioblasts through three evolutionarily conserved Tin-binding sites.

Discussion
Cardiac mesoderm specification is strongly dependent on the combined actions of several transcription factors including the Doc family of T-box transcription factors and tin [2,23,24]. Early mesodermal expression of Unc-5 is also dependent on the combined actions of tin and Doc [23] and Unc-5 cardiogenic mesoderm enhancers are bound by tin, Doc and Pnr [24]. However, at later stages of cardiogenesis their expression pattern segregates; tin is restricted to CBs and becomes the major regulator in these cells while Doc expression is restricted to SMCs (reviewed in [2]). One of tin functions in CBs is to repress Doc, and consequently activate only tin dependent genes [17]. Our results show that at this developmental stage CB-specific expression of Unc-5 is strictly dependent on tin (Fig 1). In tin-ABD;tin 346 / tin 346 mutants all CBs ectopically express Doc [17]; however, it does not seem sufficient to promote Unc-5 expression on them (Fig 1) or through the Unc-5-GH heart enhancer (Fig 4). In SMCs where tin is not expressed but Unc-5 is (Fig 3) Unc-5 is still expressed in tin mutants. It is very likely that its expression in these cells is dependent on Doc and svp. In fact, cardiac mesoderm specification is strongly dependent on the combined action of several transcription factors including the Doc family of T-box transcription factors and tin [2]. At this stage Unc-5 expression is dependent on both, tin and Doc [23]. Thus, tin specific regulation of Unc-5 in CBs when the tubular DV assembles could represent a mark of the original cardiogenic transcriptional code owing to its developmental lineage. It would be interesting to determine if Doc is regulating Unc-5 in SMCs to confirm the segregation the expression pattern of the transcriptional regulators is reflected functionally. Enhancer regulation in CBs, where expression is virtually absent in tin mutants, contrasts with that of PCs where is still moderately active (Fig 4) indicating a partial requirement for tin. Interestingly, some PCs express eve, a known regulator of Unc-5 in motoneurons [28,36]. tin may work combinatorially with eve and other regulators in PCs as shown for Unc-5 regulation by eve in motoneurons [37].
It has been recently shown that Unc-5 receptor's role during heart morphogenesis is to preserve the luminal space between opposing CB membranes during heart tube lumen formation [7,8]. Accordingly, Unc-5 and the Unc-5-GH reporter are expressed during tubulogenesis ( Fig  3A-3E") and its expression in CBs is strictly dependent on tin (Fig 1). Thus, there is a perfect match between Unc-5 expression in CBs and tin regulation. Furthermore, the elimination of the Tin-binding sites in the DV enhancer renders it unresponsive to Tin in vitro (Fig 6) and in R8,9,10-GH reporter (with unchanged SNs' fluorescence of 2.52 ± 0.192 s.e.m and 2.8 ± 0.177 s.e.m, respectively)]. (F) GFP fluorescence is also significantly reduced (p<7 x 10 −18 ) in CBs from 1.98 ± 0.127 s.e.m. for the wild-type Unc-5-GH reporter to 0.28 ± 0.052 s.e.m. for the R8,9,10-GH mutant reporter. In F, fluorescence in SNs is not affected with unchanged SNs' fluorescence of 3.75 ± 0.44 s.e.m in Unc-5-GH embryos and 3.79 ± 0.36 s.e.m in R8,9,10-GH embryos, respectively. All panels are lateral views of stage 14-15 embryos with dorsal side up and anterior to the left. A magnification of the regions delineated by insets is shown for each panel.
doi:10.1371/journal.pone.0137688.g007 vivo (Fig 7). Therefore, our results strongly suggest that Unc-5 is specifically regulated by tinman, through a direct interaction with three evolutionary conserved sites within its regulatory region at later stages of DV tubulogenesis.
Given the high degree of conservation on the molecular pathways controlling heart morphogenesis in vertebrates [38,39] the NK2, family of transcription factors is a very likely candidate to drive this process, in part, through a direct regulation of guidance receptors.

Generation of constructs
Unc-5 locus dissection was carried out by PCR-amplifications using genomic DNA as template to amplify overlapping fragments of random sizes. The PCR products were cloned using TOPO TA Cloning (Invitrogen), sequenced and recombined into destination vectors: pGateway-nlsVenus-attB and/or pGateway-Gal4 and integrated into the attP2 site [41]. The pGateway-Rluc vector was used for luciferase assays. PCR amplified tin was cloned into pActC-GFP or pAct5C-FLAG plasmids to generate GFP-Tin and FLAG-Tin used in ChIP or luciferase assays, respectively. For site directed mutagenesis of Tin binding sites the most conserved nucleotides within the CACTTGA consensus motif, the "CA" dinucleotide and the first "T" [24], were mutated to "GT" and "A", respectively. The following primers were used for mutagenesis: CACGGTATAGAGGCAACGG and CCGTTGCCTCTATACCGTG for R8, GTTCGTCTACA GGGCAGTCAC and GTGACTGCCCTGTAGACGAAC for R9, and TGCTGTCTAGTTTTGTGTGT TCTG and CAGAACACACAAAACTAGACAGCA for R10.
Stacks of images were obtained using Zeiss Confocal LSM700 Microscope and 20X or 40X oil-immersion. ImageJ was used for quantification of fluorescence within regions of interest (ROI). For GFP fluorescence quantification, all controls and samples were fixed together using the same procedure, stained with GFP antibodies and imaged using the same configurations. Samples and controls were mounted in the same slide for imaging. Image analyses were done using ImageJ software. Background correction was performed individually for each embryo and the intensity for GFP ROIs was divided by the intensity of control areas and finally averaged for each genotypic group.

Statistical analysis
Statistical significance of alterations in luciferase activity levels for reporters with different mutations or fluorescence intensity in different samples were calculated using one-tailed t-test for pair-wise comparisons and histograms were generated using Microsoft Excel 2013.

Chromatin immunoprecipitation
ChIP was performed and analyzed essentially as described previously [42]. In summary, extracts from S2R+ cells transfected with either pAct5C-GFP-Tin or pAct5C (as mock control) were fixed in 1% formaldehyde for 10 minutes at room temperature and then lysed. Following shearing the chromatin by sonication, lysates were incubated with rabbit anti-GFP (ab290; Abcam) for 2 hours at 4°C followed by incubation with protein A-sepharose (P9424; Sigma) for an additional 2 hours. Beads were then washed and the immunoprecipitated material were eluted at 70°C overnight. Phenole-chloroform DNA extraction was performed the next day to purify the precipitated DNA. The immunoprecipitated DNA was subsequently quantified by real-time qPCR.
Luciferase reporter Assays S2R+ cells were used for luciferase assays. Approximately 10 5 cells were transfected (using FuGENE 1 HD Transfection Reagent, E2311) with the transcription factor plasmid, Rluc construct, and PolIII-Fluc (as internal control). Cells were analyzed for luciferase activity 36 hours post transfection using the Dual-Glo Luciferase Kit (Promega) according to manufacturer's instructions. Samples were assessed in triplicate.