The Zic family homologue Odd-paired regulates Alk expression in Drosophila

The Anaplastic Lymphoma Kinase (Alk) receptor tyrosine kinase (RTK) plays a critical role in the specification of founder cells (FCs) in the Drosophila visceral mesoderm (VM) during embryogenesis. Reporter gene and CRISPR/Cas9 deletion analysis reveals enhancer regions in and upstream of the Alk locus that influence tissue-specific expression in the amnioserosa (AS), the VM and the epidermis. By performing high throughput yeast one-hybrid screens (Y1H) with a library of Drosophila transcription factors (TFs) we identify Odd-paired (Opa), the Drosophila homologue of the vertebrate Zic family of TFs, as a novel regulator of embryonic Alk expression. Further characterization identifies evolutionarily conserved Opa-binding cis-regulatory motifs in one of the Alk associated enhancer elements. Employing Alk reporter lines as well as CRISPR/Cas9-mediated removal of regulatory elements in the Alk locus, we show modulation of Alk expression by Opa in the embryonic AS, epidermis and VM. In addition, we identify enhancer elements that integrate input from additional TFs, such as Binou (Bin) and Bagpipe (Bap), to regulate VM expression of Alk in a combinatorial manner. Taken together, our data show that the Opa zinc finger TF is a novel regulator of embryonic Alk expression.


Author summary
The Alk receptor tyrosine kinase is employed repeatedly during Drosophila development to drive signaling events in a variety of tissues. The spatial and temporal expression pattern of the Alk gene is tightly regulated. Identifying factors that influence the expression of Alk is important to better understand how Alk signaling is controlled. In this paper we characterize cis-regulatory sequences in the Alk locus and the transcription factors that bind them to govern Alk expression in the Drosophila embryo. Using a robotic protein-DNA interaction assay, we identified the Zic family transcription factor Odd-paired as a factor that binds to regulatory elements in the Alk locus. Binding of Odd-paired to Alk cis-regulatory elements varies spatially, revealing a requirement for a1111111111 a1111111111 a1111111111 a1111111111 a1111111111
While we and others have previously identified and characterized several important components and targets of the Alk RTK signaling pathway, little is currently understood about the molecular mechanisms regulating the spatial and temporal expression of the Alk receptor itself. Development of the early VM requires the activity of the NK4/msh-2-like homeobox TF Tinman (Tin) for dorsal mesoderm differentiation, as well as the NK3 and FoxF orthologues Bagpipe (Bap) and Biniou (Bin) [12][13][14][15]. Interestingly, the expression patterns of bap and bin in the VM primordia are similar to that of Alk [15]. In addition, ChIP-on-chip studies have shown the region upstream of Alk gene to be occupied by several mesodermally expressed TFs, such as Bin, Bap, Twist (Twi), Tin and Myocyte enhancer factor 2 (Mef2) at different time points during embryogenesis [16,17]. While binding of these factors has been documented, their importance in the regulation of Alk transcription in the VM has only been initially characterized in case of Tin [16,17].
Here we address regulation of Alk expression during embryogenesis. We have employed a combination of in vitro and in vivo approaches to identify and characterize Alk-specific enhancer elements, including high throughput yeast one-hybrid screening (Y1H) with a library of Drosophila TFs [18]. This Y1H screen identified the zinc finger TF Odd-paired (Opa) as binding to an evolutionary conserved cis-regulatory module (CRM) within one of the Alk-associated enhancer regions. In agreement with these findings, opa mutants displayed a complete loss of Alk expression in the epidermis and reduced levels of Alk in the VM. Furthermore, CRISPR/Cas9-mediated deletion of the Opa binding site containing region in the Alk locus resulted in a reduction of VM Alk protein together with loss of Alk expression in both the AS and embryonic epidermis, indicating that Opa plays an important role in tissue-specific Alk expression during embryogenesis. We have also identified additional enhancer regions regulated by the Bin and Bap TFs, likely together with additional TFs, that work with the Opa binding CRM to regulate Alk expression in the VM in a combinatorial manner.

Identification of regulatory regions involved in Alk expression during embryogenesis
To study Alk expression during embryogenesis, we employed transgenic GAL4-lines containing overlapping DNA sequences corresponding to Alk 5-prime upstream regions (Fig 1A, S1  Fig), aiming to identify regulatory elements with activity in the visceral mesoderm (VM). AlkEI6.5-GAL4 was previously described [1] as driving expression in the trunk VM with stronger expression in founder cells (FCs) (Fig 1B, stage 11, arrowhead). We also noted that the AlkEI6.5-GAL4 driver was expressed in the amnioserosa (AS), in keeping with earlier observations that Alk mRNA is expressed in the dorsal-most region of the embryo corresponding to the presumptive AS at the early gastrulation stage (S2A and S2B Fig) [1]. We next analyzed AlkE4-GAL4, which contains 2.4 kb of the AlkEI6.5-GAL4 region and an additional 1.6 kb upstream (4.0 kb in total). This GAL4-driver promotes expression in a similar pattern to AlkEI6.5-GAL4, suggesting this DNA region also contains regulatory elements involved in Alk transcriptional regulation (Fig 1C). In addition, AlkE2.7-GAL4, covering a shorter sequence within AlkEI6. 5 and AlkE4, displays activity in the entire trunk VM, being considerably stronger in FCs (Fig 1D, arrowhead).
To ensure the specificity of our transgenic lines for the Alk locus flanking genes we performed in situ hybridization on both neighboring genes namely CG5065 (upstream) and gprs (downstream) (Fig 1A, S3 Fig). Neither CG5065 nor gprs is expressed in a pattern similar to that of Alk in the VM, suggesting that any VM expressing region identified flanking the Alk locus may be involved in the regulation of Alk transcription.
The elevated level of expression of AlkEI6.5-GAL4 and AlkE2.7-GAL4 in FCs compared with other cells of the developing VM suggests Alk may respond to its own signaling. Since signaling in the FCs is driven by activation of Alk by its ligand Jelly Belly (Jeb), we examined expression of AlkE2.7-GAL4 in either the absence of Alk activity (Alk 1 /Alk 10 ), or upon activation of Alk by overexpression of Jeb in the VM. AlkE2.7-GAL4 expression in the FCs was reduced in Alk 1 /Alk 10 mutants (Fig 1F; arrowhead). In contrast, overexpression of Jeb resulted in robust expression of AlkE2.7-GAL4 in all cells of the VM (Fig 1G; arrowhead). These results suggest that Alk expression in the VM is positively regulated by Alk signaling, representing a positive feedback loop. Thus, we have identified CRMs in the 5' region of the Alk locus that promote Alk expression in the presumptive amnioserosa and developing VM. Additionally, our preliminary GAL4 analysis suggests the presence of inhibitory modules within this region that likely contribute to the overall regulation of Alk expression.
Analysis of Alk enhancer regions in vivo by CRISPR/Cas9-mediated deletion identifies a critical role for the Alk-RB promoter ChIP experiments performed by the Furlong laboratory have identified a 547 bp CRM (MesoCRM-880) overlapping the AlkE2.7 fragment that binds Bin, Bap, Mef, Tin and Twi TFs [16] (shown schematically in Fig 1A, S1 Fig). Later analysis by the Frasch group identified a 1,984 bp region (AlkE301) in a genome wide Tin ChIP analysis that drives expression in the VM [17] (shown schematically in Fig 1A, S1 Fig). Together with our GAL4 analyses these results suggest that the Alk-RB promoter may be important for the VM expression of Alk. To functionally address the role of Alk-RB we generated deletion mutants targeting the Alk-RB isoform with CRISPR/Cas9 [19][20][21], employing two independent single guide RNA (sgRNA) combinations. This resulted in genomic deletions of 1053 bp (represented by Alk ΔRB_1. 22.2 ) or 1325 bp (represented by Alk ΔRB_15. 16.2 ) in the region of the Alk-RB 5'UTR (Figs 1A and 2A;  Table). Both homozygous mutants were embryonic lethal. We further examined the visceral morphology of homozygous Alk ΔRB_1.22.2 mutant embryos and control siblings using Fasciclin III (FasIII) as marker for differentiated VM. In control embryos FasIII was expressed in the visceral musculature surrounding the entire midgut, which at later stages of embryogenesis is subdivided into four chambers (Fig 2B; arrowhead). In Alk ΔRB_1.22.2 embryos, FasIII-FCs (close up, arrowhead). (E-F) AlkE2.7 expression in the FCs is responsive to Alk signaling. lacZ expression in Alk 1 /Alk 10 embryos is weaker when compared to Alk 10 heterozygote balanced controls (arrowhead; compare β-gal heatmaps in E and F; note: epidermal β-gal expression in control (E) is due to presence of lacZ balancer; Alk protein is observed in Alk 1 /Alk 10 animals (F) as these alleles encode non-functional Alk protein truncations detected with anti-Alk). (G) Ectopic expression of the Alk ligand Jeb, leads to activation of Alk signaling in all cells of the VM (arrowhead) and is marked by expression of Org-1 in blue resulting in increased lacZ expression from AlkE2.7 in all cells of the VM (compare β-gal heatmaps in E and G). Close up regions in E-G are indicated with dashed boxes. Scale bars: 50 μm and 10 μm (embryo and close up, respectively).  positive midgut muscles were absent while FasIII-expression could still be detected in the embryonic foregut and hindgut respectively (Fig 2C), resembling the Alk mutant phenotype [2]. In agreement with their mutant phenotype, Alk ΔRB_1.22.2 mutants lacked detectable Alk mRNA and protein in the VM (Fig 2K, S4 Fig) compared to wild-type animals (Fig 2E and 2H,  S4 Fig), while Alk expression levels in the CNS were similar to those observed in control embryos (Fig 2I and 2L; S4 Fig). Alk expression was also lost in the AS and epidermis of Alk ΔRB_1.22.2 mutants (Fig 2J and 2K, asterisks). Therefore, expression from the Alk-RB promotor drives Alk expression in the embryonic VM, AS and epidermis and is critical for proper formation of the midgut musculature.
Identification of potential regulators of Alk expression by high throughput yeast one-hybrid screening A 3.6 kb genomic region that covered the putative VM and epidermal Alk enhancer regions identified in our initial experiments (Fig 1A) was subjected in parallel to high throughput yeast one-hybrid (Y1H) and more detailed reporter gene analyses. Six fragments (denoted AlkEB6 -AlkEB11; S1A Fig; S2 Table) of approximately 700 bp in length, including a~100 bp overlap between neighboring fragments, were analyzed.
Embryonic lacZ reporter activity was observed with only two of the DNA fragments studied, namely AlkEB8 and AlkEB9 (Fig 3A-3E). AlkEB8 displayed weaker activity in the VM than that observed with AlkEB9 ( Fig 3B, 3B", 3D and 3D", arrowheads; quantified in S5 Fig).
In addition to VM expression, AlkEB9 was also expressed in the AS and epidermis where it overlapped with Alk protein (Fig 3E, asterisks; S2D Fig). No expression in the AS and epidermis was observed in AlkEB8 (Fig 3C, asterisks; S2C Fig). To further confirm that AlkEB9 contains important enhancer elements for Alk, we performed rescue experiments using AlkE9-GAL4 (Fig 3F-3H). Ectopic over-expression of Alk (AlkEB9-GAL4>UAS-Alk) in an Alk 1 /Alk 10 mutant background resulted in a rescue of the embryonic gut phenotype (Fig 3H). Therefore, the AlkEB9 genomic region contains sufficient regulatory information to allow rescue of the embryonic Alk VM expression.
High throughput Y1H was carried out on the same six fragments employing a library of Drosophila TFs fused to the yeast GAL4 activation domain [18] (Fig 1A, S1 Fig). Based on our reporter gene analysis we focused on the AlkEB9 DNA bait Y1H data set aiming to functionally characterize novel transcriptional regulators of Alk. A set of TFs was identified to bind to the AlkEB9 DNA bait by Y1H screening (Fig 4A). Among these, Odd-paired (Opa) (Fig 4B), Pointed (Pnt), Side and CG14655 bound to the AlkEB9 DNA bait and promoted growth in selective media in all biological replicates performed. We further investigated a role for TFs binding AlkEB9 in Alk transcriptional regulation in vivo, employing paired (prd)-GAL4, which drives expression in alternating parasegments and offers internal control of Alk expression levels in the epidermis. In this assay both Opa and Pnt were identified as potential regulators of Alk, with Opa inducing and Pnt repressing Alk expression (S6 Fig). Of the TFs tested in this study, Opa was the only one that resulted in an increase in Alk protein. We also overexpressed opa with the engrailed (en)-GAL4 driver which resulted in an increase in AlkEB9-lacZ reporter activity as well as Alk protein levels in the epidermis (Fig 4C-4D'), indicating that Opa is sufficient to promote Alk expression. Therefore we focused on a more detailed investigation of the role of Opa Alk transcriptional regulation.

AlkEB9 contains functional Opa binding sites
Employing the JASPAR online prediction tool [22], we were able to identify a potential Opa binding site (BS) in the AlkEB9 sequence, JASPAR_OpaBS (GACCTCCGGCTG) (Fig 5A and 5B). In addition, we identified another Opa BS similar to the Opa consensus motif previously reported by [23] and therefore referred to as SELEX_OpaBS (GCGGGGATG) (Fig 5A and 5B). Employing the phastCons database, which identifies evolutionarily conserved elements in a multiple alignment, to analyze this sequence, we found that both binding sites are conserved among Drosophila species (Fig 5B; conservation score in green; Opa BS highlighted in yellow) [24,25]. We next assessed the ability of Opa to specifically bind these predicted sites by electrophoresis mobility shift assay (EMSA). EMSA was performed on the SELEX_OpaBS and JASPAR_OpaBS sequences, incubating probes with cell lysates from Opa-expressing HEK293 cells in the presence of poly(dI-dC) to prevent non-specific binding. Addition of Opa lysate to the binding reaction resulted in a shift of both SELEX_OpaBS and JASPAR_OpaBS probes and was reversed by addition of 100 fold molar excess of non-labelled probe (Fig 5C and 5D). In contrast, addition of cold probes that were mutated within the SELEX and JASPAR binding sites, based on published data [23], was unable to compete the shift generated upon addition of Opa to the labelled wild-type probe. Furthermore, labelled mutated SELEX_OpaBS and JASPAR_OpaBS probes did not exhibit a mobility shift upon incubation with Opa (Fig 5C and 5D). The above observations led us to characterize Opa regulates Alk expression the interactions of the Opa with the Alk locus by chromatin immunoprecipitation (ChIP). Consistent with Y1H and EMSA analyses, Opa association is detected with a region upstream of the Alk promoter that spans both the SELEX_OpaBS and JASPAR_OpaBS sequences in chromatin from wild-type embryos ( Fig 5E).
To address the importance of the JASPAR_ and SELEX_OpaBS for in vivo Alk transcription we first attempted to identify a minimal region within the AlkEB9 region that could drive VM expression. This analysis led to the identification of a 154 bp fragment including both SELEX and JASPAR Opa binding sites (AlkEB9_OpaBS; schematically shown in S1 Fig) that drives strong VM and epidermal expression, similar to that observed with the 700 bp AlkEB9 fragment ( Fig 6A-6F'). Quantification revealed that VM expression from AlkEB9_OpaBS was weaker than that of the 700 bp AlkEB9-lacZ reporter ( Taken together, these data show that the AlkEB9 genomic region contains sequence-specific binding sites for Opa that regulate expression from Alk enhancer elements.

Opa is required for tissue specific Alk expression during embryogenesis
To further dissect the potential role of Opa as a regulator of Alk expression, we examined opa expression during embryogenesis [26]. opa mRNA can be detected at stage 5 in the ectoderm and mesoderm progenitors spanning the presumptive segmented region of the embryo. At stage 9 opa expression decreases slightly and appears in the neuroectoderm persisting until late  We next examined the reporter expression of AlkEB9-lacZ in opa loss-of-function mutants (opa 1 /opa 8 ). While AlkEB9-lacZ is activated in the entire VM and epidermis in wild-type embryos ( Fig 7A and 7A'), opa 1 /opa 8 mutant embryos display only weak reporter activity during embryogenesis (Fig 7B and 7B'; quantified in Fig 7C). The severe developmental defects observed in opa 1 /opa 8 mutants make analysis difficult, however we noted lower levels of Alk protein in the VM and a complete loss of detectable Alk in the epidermis of opa mutant animals, in agreement with the loss of AlkEB9-lacZ activity (Fig 7B). These observations were supported by analysis of RNAi-induced Opa knockdown in the developing mesoderm employing 2xPE-GAL4 (Fig 7D-7F). We observed that embryos expressing opa RNAi (2xPE-GAL4>UASopa RNAi ) displayed a reduction of AlkEB9-lacZ in the VM at later stages when compared with controls (Fig 7E', quantified in Fig 7F).
Since Opa has been reported to be required for proper midgut formation, with opa mutants exhibiting an interrupted VM that fails to form midgut constrictions during embryogenesis [26], we also examined Alk signaling in the VM of opa mutants. opa 1 /opa 8 mutants, examined with the FC-marker Org-1, exhibited Org-1 positive VM FCs, however, the level of Org-1 protein observed was less than in control embryos (S8B and S8C Fig). Since reductions in both Alk and Org-1 protein were seen in opa 1 /opa 8 mutants, we asked whether Opa overexpression was sufficient to drive Alk signaling. As expected, bap3-GAL4 driven expression of Jeb in the VM resulted in an increased expression of the HandC-GFP FC marker reflecting activation of Alk signaling (S8D-S8D" Fig). In contrast, bap3-GAL4 driven expression of Opa did not increase HandC-GFP levels (S8E-S8E" Fig). Thus, while Alk signaling may be reduced in opa mutants, Opa is not sufficient to influence FC specification driven by Alk signaling in the embryonic VM.
As a complement to our analysis of opa mutants, we employed the Opa4opt-lacZ transgene as readout for Opa activity, focusing on the epidermis. Opa4opt-lacZ contains four tandem copies of the SELEX determined Opa-BS [23]. In parallel we analyzed the opa 3D246 lacZ https://doi.org/10.1371/journal.pgen.1006617.g007 enhancer trap which reflects opa expression [26]. We observed expression of both opa 3D246 and Opa4opt-lacZ in the embryonic epidermis, coinciding with Alk protein (Fig 7G and 7H), suggesting that Opa is both expressed and active in these cells. Furthermore, a mutant Opa4opt-lacZ transgene, called Opa4opt-KO-lacZ, in which the Opa binding sites are mutated, no longer displayed expression overlapping with Alk in the embryonic epidermis ( Fig 7I).
Taken together, this data supports an important role for Opa in driving embryonic Alk transcription, particularly in the epidermis, through the AlkEB9 regulatory region. However, in agreement with our earlier analyses, Alk expression in the VM does not depend only on Opa activity, since Alk protein is still observed in the VM of opa 1 /opa 8 loss of function animals ( Fig 7B).
Opa binding sites in the Alk enhancer region regulate Alk protein expression in a tissue-specific context Given the presence of Opa binding sites proximal to the Alk-RB isoform promoter, together with the loss of reporter gene activity after deletion of these sites, we next addressed their in vivo relevance for Alk transcriptional regulation. CRISPR/Cas9 genome editing was again employed to delete the identified Opa binding sites (Opa-BS) in the AlkEB9 enhancer region of the Alk locus (Fig 8A; S1 Fig). This resulted in isolation of two viable Alk ΔOpaBS mutants: Alk ΔO-paBS_10. 28 Table), indicating this region is essential for Alk expression in these tissues. We also observed reduced Alk protein levels in the VM when compared to control embryos at the same stage ( Fig 8H, compare with Fig 8B; quantified in Fig  8N). In close proximity to the Opa binding sites we also observed a cluster of highly scoring JASPAR-predicted binding sites for mesodermal TFs (Bap, Sna and Tin) in the AlkEB9 genomic region, here designated as meso-BS (Fig 8A; S1 Fig). Deletion of this meso-BS region alone, in Alk ΔmesoBS embryos, does not appear to affect either Alk protein levels or the formation of a fully developed gut (S1 and S9 Figs; S1 Table). Interestingly, Alk ΔOpaBS_10.36.1 removes 178 bp including both the Opa-and the meso-BS sites allowing us to functionally address the contribution of the meso-BS region relative to the Opa binding sites. Deletion of both the meso-BS and the Opa-BS regions (Alk ΔOpaBS_10.36.1 ) results in viable animals, albeit with reduced Alk protein levels when compared to those in control embryos (Fig 8F, S1 Table). Reduction of Alk protein levels in the VM was noticeably stronger in Alk ΔOpaBS_10.36.1 when compared with Alk ΔOpaBS_10.28.3 mutants (Fig 8F and 8H; quantified in Fig 8N). However, the reduced Alk protein levels observed in Alk ΔOpaBS_10. 28.3 and Alk ΔOpaBS_10.36.1 were still sufficient to drive Jeb/ Alk signaling in the VM as measured with HandC-GFP reporter expression (Fig 8G and 8I insets), and form a functional gut as visualized by FasIII staining (Fig 8G and 8I).
Since we detected VM expression activity in the overlapping Alk proximal AlkEB8-lacZ reporter (Fig 3B), we explored the contribution of the corresponding region in the Alk locus to regulation of Alk VM expression. To do this we employed CRISPR/Cas9 genomic editing to remove 808 bp covering part of AlkEB9 (312 bp) and the majority of AlkEB8 (647 bp) (represented by Alk ΔEB8 ) (Fig 8A; S1 Fig, S1 Table). These mutants were homozygous viable, with a wild-type VM morphology (Fig 8K; S1 Fig). Investigation of Alk protein levels in Alk ΔEB8 mutants revealed a decrease, but not complete loss, of Alk in the VM (Fig 8J; quantified in Fig  8N), suggesting that CRM(s) within the AlkEB8 region are not essential but contribute to VM expression of Alk. Expression of Alk in the epidermis was not affected, in agreement with a sole epidermal CRM including the Opa binding sites within the AlkEB9 region. To further exclude the possibility that an essential CRM might be located in the overlap between AlkEB8 and AlkEB9, we generated a series of overlapping reporter constructs in this area S1 Fig. We did not observe any VM expression activity in this reporter series (S10 Fig), suggesting that two CRMs, one in the region of AlkEB8 and one in AlkEB9 function together drive VM expression of Alk.
To test the contribution of additional CRMs to VM expression of Alk we extended the Alk ΔEB8 deletion to include the Opa binding sites within AlkEB9. This deletion was denoted Alk ΔOpaBS+EB8 (Fig 8A, 8L and 8M; S1 Fig, S1 Table). Alk ΔOpaBS+EB8 mutants failed to express Alk protein in the VM, epidermis or AS (Fig 8L), and were homozygous lethal due to lack of FC specification (Fig 8M; inset), supporting our hypothesis of several independent CRMs within this area that are critical for Alk expression in the VM.
Taken together, our analysis identifies a CRM proximal to the Alk-RB isoform promotor that contains Opa binding sites as critical for Alk expression in the embryonic AS and epidermis. This region also contributes to Alk expression in the VM. Further deletion analysis reveals additional CRM(s) located within the AlkEB8 fragment that contribute to regulation of Alk VM expression.

Bin and Bap contribute to VM expression of Alk
Previous studies identified CRMs binding Bin, Bap, Twi, Tin and Mef2 in the Alk locus [16,17]. In particular the ChIP and reporter gene analyses performed by Jin et al. (2013) suggested Tin binding to be important if not essential for Alk expression. We studied expression of Alk protein and the AlkEB9-lacZ reporter in tin346/ED6058 mutant embryos (S11 Fig). Both Alk protein and reporter gene expression, could be observed in the dorsal epidermis and amnioserosa at stage 10/11 and in the epidermis at stage 14 (S11 Fig), indicating that regulation by Tin is not critical for Alk expression outside the VM. In contrast, Alk and AlkEB9-lacZ reporter gene expression as observed in the VM of control embryos (S11 Fig) was not observed. However, this analysis was inconclusive since it is difficult to address if, and to which extent, VM formation proceeds in tin mutant embryos. Bin and Bap TFs are known to have a critical function during Drosophila VM development [12,15]. In our initial experiments we were unable to see any effect on Alk expression on ectopic expression of either Bin or Bap alone in the epidermis employing en-GAL4 as driver (S12 Fig), however this may reflect a lack of tissue competence in our experimental approach. Therefore, we analyzed both Alk protein and AlkEB9-lacZ expression in bin and bap mutants focusing on VM expression. Although VM development does not proceed normally in either bin or bap mutants, we could observe Alk protein and AlkEB9-lacZ expression in the VM in both cases (S13B-S13C' Fig). We next investigated AlkE-B8-lacZ expression, which was reduced in both bin mutants and bap mutants (S13E-S13F' Fig). On closer inspection of the AlkEB8 region we identified four putative Bap binding sites, which we deleted to create AlkEB8ΔBapBS-lacZ. AlkEB8ΔBapBS-lacZ failed to exhibit reporter expression suggesting that Bap may be involved in Alk expression in the VM through binding sites within the AlkEB8 region (Fig 9A and 9B). Based on these findings, we analyzed Alk protein levels in Alk ΔOpaBS_10.28.3 ;bin 1 /BSC374 and Alk ΔOpaBS_10. 28.3 ;bap 208 /ED6058 double mutant backgrounds to test whether Alk expression was affected in a combinatorial manner. We observed a strong reduction of Alk expression in the VM of both Alk ΔOpaBS_10. 28.3 ; bin 1 /BSC374 mutants (Fig 9C and 9D; quantified in Fig 9G) and Alk ΔOpaBS_10. 28.3 ;bap 208 /ED6058 mutants lacking detectable Alk protein in the VM and epidermis (L, stage 11), and failing to specify FCs (M, inset, stage 11 Alk ΔOpaBS+EB8 mutant embryo) or develop a midgut (M, stage 16, arrowhead). (N) Alk ΔOpaBS_10.36.1 , Alk ΔOpaBS_10. 28.3 and Alk ΔEB8 mutants show a significant decrease in Alk protein in the VM when compared to control embryos (n = 10 animals per genotype, **** p 0.0001). Scale bars: 50 μm.
https://doi.org/10.1371/journal.pgen.1006617.g008 (Fig 9E and 9F; quantified in Fig 9H), with loss of Bap appearing to have a stronger impact. These results suggest that additional factors, including Bin and Bap, contribute to regulate Alk expression in the VM through the AlkEB8 region of the Alk locus (Fig 9I).

Discussion
In this study we report the identification of Alk cis-regulatory elements and TF binding sites that control the expression of Alk during embryogenesis. We have been able to identify regions that regulate transcription of Alk in the AS, the VM and the epidermis. We further identify the Opa TF as well as Bin and Bap as regulators of Alk transcription in these tissues during embryogenesis. Taken together our results shed light on the regulatory mechanisms Alk expression in AS and epidermis is under control of a critical CRM proximal to Alk-RB that binds Opa (in orange). Regulation may be by Opa alone, or in combination with uncharacterized additional factors (denoted in grey) as have been described previously by ChIP. VM expression of Alk involves a more extensive organization of CRMs that appear to act in a combinatorial fashion. These include the Opa binding region in AlkEB9 also employed for the AS and epidermis, and an additional more proximal CRM(s) within the AlkEB8 region that appears to mediate input from Bin and Bap (in green and purple, respectively) that cooperate to drive Alk expression in the VM in combination with additional factors (in grey). Scale bar: 50 μm. controlling Alk transcription and identify important cis-regulatory sequences required for regulation of Alk gene expression.

Transcription from the Alk-RB promoter is essential for Alk expression in the visceral mesoderm
The importance of Jeb/Alk signaling in vivo in the embryonic VM for FC specification is well established [2][3][4][5]. From this earlier work we know that activated Alk in the VM triggers not only transcriptional activation but also post-translational modifications that promote the specification of the FC fate [3][4][5][6]11]. In contrast, very little is known about factors that mediate Alk transcriptional regulation. In this study we aimed to identify CRMs and TFs important for Alk transcription. The Alk-RA and Alk-RB transcripts encode the same protein, but differ in their 5' non-coding regions which employ alternative promoters [1]. This potentially allows differential expression of the Alk-RA and Alk-RB mRNA isoforms both temporally and spatially. Such regulation has been described previously for genes such as the Drosophila DOA kinase [27] and the BBG PDZ-protein [28], among others. Embryos in which the promoter of the Alk-RB isoform has been disrupted fail to express detectable Alk protein in the VM, AS and epidermis, and exhibit an Alk loss of function phenotype, revealing that this promoter is critical for Alk expression in these embryonic tissues. However, expression of Alk in the embryonic CNS is not compromised by the removal of the Alk-RB promoter and upstream sequences, suggesting that CNS expression of Alk is independent of the VM, AS and epidermal enhancers identified here. Taken together, our results indicate a critical requirement for Alk-RB expression to ensure sufficient Alk protein levels in the VM for signaling and founder cell specification, as well as for Alk expression in the AS and epidermis where the function of Alk is currently uncharacterized.

Enhancer elements upstream of the Alk locus regulate expression in the amnioserosa, visceral mesoderm and epidermis
Previous reports have studied sequences within the Alk locus either by reporter activity assays [1,17] or ChIP-on-chip analyses [16,17]. Our analysis of reporter activity has identified regions upstream of Alk that are active in the AS, VM and epidermis. These coincide temporally with Alk protein expression, allowing us to define Alk VM, AS and epidermal enhancers located proximal to the Alk-RB promoter. High-throughput Y1H screens performed in this study identified a number of TFs that potentially bind to and regulate these regions of the Alk locus. In addition, a genome-wide ChIP-on-chip screen for mesodermal TFs occupancy identified a CRM upstream of the Alk locus that is active during mesoderm development [16]. This CRM maps to 2R:16,639,969..16,640,341 (relative to Dmel_Release_6 sequence assembly) and was described to be bound by mesodermal TFs including Bin, Bap, Mef2 and Tin and Twi [12,15,16,29,30]. However, none of these factors were found in our Y1H analysis. This may reflect additional requirements for binding of some TFs, which would preclude their identification by Y1H, such as heterodimerization with co-factors or post-translational modifications. Interestingly, homozygous mutants for bin and bap still express Alk protein in the VM, suggesting that while they may be involved in the modulation of Alk expression, additional factors are also important in the regulation of Alk expression in the VM.
One such factor could be the NK-4/msh-2 TF Tinman (Tin) which has been previously reported to bind CRMs at the Alk locus [16,17]. Indeed, expression of Alk in the VM is affected in tin mutant embryos ( [17], this study) however it is not clear if this occurs due to direct regulation of Alk expression by Tin or a general lack of induction of the VM lineage. Moreover, our analysis of Alk ΔOpaBS ;bap double mutants uncovers a severe decrease in Alk protein in the VM suggesting only a minor direct contribution of Tin. Interestingly, opa has been reported to be directly regulated by Tin during heart development [17,31] and Tin is critical for the expression of two key VM TFs bin and bap [4,15]. Therefore it is likely that the importance of Tin for Alk expression relies on its activating potential for these Alk-regulating TFs. Interestingly, loss of tin does not affect Alk expression in the epidermis.
Reporter gene expression analyses suggest the VM Alk enhancer is located upstream of the Alk-RB isoform, in agreement with previously reported AlkE301-lacZ reporter spanning 1,984 bp (Fig 1A; S1A Fig) [17] and the 547 bp MesoCRM-880 [16] that both cover the AlkEB9 region. Our data suggests that Alk-RB expression can be activated through an upstream enhancer that is bound by Opa located within AlkEB9. We were also able to identify additional nearby enhancer elements in AlkEB8 that integrate information from factors such as Bin and Bap that are critical to ensure precise and robust VM expression of Alk. Taken together with the earlier ChIP analyses from the Furlong and Frasch groups, our data suggest that Opa, along with mesodermal TFs such as Bin, Bap, Mef2 and Tin and Twi function in a combinatorial manner to drive robust expression of Alk in the VM (Fig 9I).

Opa activates Alk transcription through the AlkEB9 enhancer
Our efforts to identify novel TFs involved in Alk transcriptional control by in vitro Y1H assay resulted in a cluster of TFs potentially binding the AlkEB9 sequence. Of those TF hits for which UAS-transgenes were available to test, only Opa was observed to induce cell autonomous expression of Alk when ectopically expressed. opa is a pair-rule gene [32] that encodes a zinc finger protein important during embryonic segmentation and midgut formation [26,33,34], as well as adult head morphogenesis by direct regulation of decapentaplegic (dpp) transcription [23,35]. opa transcript is expressed in a spatially and temporally dynamic pattern, starting from stage 5 in a broad expression domain and from stage 11 onwards in two discrete domains in the VM corresponding to the first and third midgut constrictions [26,33].
While Opa plays a role in the differentiating midgut musculature, with opa mutants exhibiting an interrupted VM unable to form midgut constrictions during embryogenesis [26], its role during segment formation presents a challenge when attempting to decipher the contribution of this TF more precisely. One component of this may be the regulation of Alk by Opa shown here. While we observed that opa mutants display lower levels of Alk protein in the VM, Jeb/Alk signaling is not abrogated, suggesting that while reduced, Alk protein levels are not reduced to levels under the threshold critical to drive Alk signaling. The lack of a critical role for Opa in the VM expression of Alk may reflect the importance of Alk signaling in this tissue for survival of the fly, where a more complex network of TFs may be employed to ensure rigorous Alk expression.

A role for Bin and Bap in regulation of Alk transcription in the VM
Additional VM enhancer elements 5' of AlkEB9 in the Alk locus are regulated in part by Bin and Bap, two TFs that are critical for VM development. Thus multiple partially redundant enhancer regions are employed to safeguard VM expression of Alk, a phenomenon that has been observed in numerous genes expressed in the Drosophila embryonic muscle [36]. Moreover, while we have tested the role of Opa and the Opa binding sites in the AlkEB9 region of the Alk locus in this work, we have done so under standard laboratory conditions, and as a result have not tested whether either Opa itself, AlkEB9 or AlkEB8 VM enhancers may play an increasingly critical role in Alk expression in more demanding environmental conditions, as it has been described for some Drosophila loci [37]. Although Alk is expressed in bin and bap mutants, our experiments combining deletion of the Opa binding region in Alk in a bin or bap mutant background suggest a combinatorial role for Bin, Bap and Opa driving VM expression of Alk [12][13][14][15]. Opa, Bin and Bap potentially act in combination with other TFs to control Alk transcription in the VM, as has been described for sloppy paired-1 (slp1) activation in the somatic blastoderm in response to Opa and Runt [38]. In addition to direct regulation of Alk expression, Opa may also impact Alk expression via indirect mechanisms during embryogenesis.
Further complexity arises when the regulation of opa itself in the VM is considered. It is known that Dpp signaling restricts the VM spatial expression pattern of opa to PS6-8, with dpp mutants showing continuous opa expression throughout the VM [26]. Opa is also known to regulate dpp expression during adult head development [23]. In addition, opa is broadly expressed in the mesoderm at stage 6 potentially driving Dpp signaling. The Dpp mesodermal response consists of up-regulation of tin and bap, important regulatory genes in the dorsal mesoderm that essentially contribute to the specification of the VM [12,39]. Similarly, Alk activity, the FoxF forkhead domain TF Bin and the Tbx1 Org-1, are also critical factors for expression of dpp in the VM and subsequent activation of Mad signaling in the midgut endoderm [40,41]. Moreover, loss of org-1, whose expression is maintained by Alk signaling in the VM, results in decreased opa VM expression [4,41], revealing a complex interplay of regulation where both Alk and Opa control each other's expression in a spatially and temporally regulated manner.
Surprisingly, in addition to a non-essential role for Opa in the regulation of Alk transcription in the VM, in this work we have been able to identify a critical role for Opa in Alk expression in the AS and epidermis. Here, in contrast to the VM, Opa appears to be required and sufficient to drive Alk expression, although the functional significance of Alk in these tissues remains uncharacterized. Expression of the AlkEB9-lacZ reporter and derivatives in which the Opa binding sites have been mutated indicate that Opa has an important function in Alk transcription through the predicted Opa BS. This is supported by the absence of detectable Alk protein in the AS and epidermis of Alk ΔOpaBS mutants, where the Opa binding sites within the AlkEB9 enhancer have been deleted. Given that Alk ΔOpaBS mutants are viable, it may be that Alk signaling is employed in a small population of non-essential cells that remain to be identified. Further work will be required to characterize the role of Alk in this context.
We have focused here on the regulation of Alk expression during embryonic development, however, Alk is also observed in larval and adult stages. Although Alk signaling does not seem to be critical for viability post-embryogenesis, a number of important roles in the nervous system have been described [42][43][44][45][46][47]. While we have not investigated the role of Opa, Bin or Bap in Alk expression at these other stages, nor in the CNS in this study, this would certainly be of interest to address in future experiments.

Immunohistochemistry
Embryos were stained as described [1]. Primary antibodies used were: guinea pig anti-Alk Images were acquired with a Zeiss LSM800 confocal microscope or Axiocam 503 camera, processed and analyzed employing Zeiss ZEN2 (Blue Edition) imaging software. For analysis of protein levels, the laser, pinhole and PMT settings were adjusted on control siblings subsequently employed for imaging of mutant embryos.
Fluorescence intensity measurements were quantified using Zeiss ZEN2 (Blue Edition). In brief: mean fluorescence values were acquired from regions of interest (ROI), corresponding to the VM or epidermis (Alk staining) selected in confocal sections of stage 11 embryos. This mean fluorescent intensity was corrected using a background ROI chosen from a non-stained area. Measurements were taken from 10 embryos per sample analyzed. For statistical analysis we performed a one-way ANOVA using GraphPad Prism 6 software, where n.s. stands for non-significant, ÃÃÃ p 0.001 and ÃÃÃÃ p 0.0001. All plots are visualized as mean ±S.D.

Generation of Org-1 antibodies
Recombinant N-terminal Org-1 protein was produced from pET30a-Org-1-N as generated by [50] was purified by His affinity chromatography and injected into rabbits for antibody generation (Genscript).

In situ hybridization
For in situ hybridization, fragments of Alk, gprs, CG5065 and opa were amplified from genomic DNA with the primer combinations shown in S3 Table. PCR products were cloned into the dual promoter PCRII TOPO vector (Invitrogen) and used as template to generate DIG-labeled in situ probes with SP6/T7 polymerases (Roche). In situ hybridization of antisense probes to embryos was carried out as previously described [51]. Samples were mounted in in situ mounting media (Electron Microscopy Sciences).
High-throughput yeast one-hybrid screening pMW2-vectors containing the different Alk putative CRMs were generated by regular cloning techniques (primer combinations shown in S4 Table) and integrated into the yeast genome as described [18]. Each DNA bait yeast strain was then transformed with a library of 647 Drosophila TFs fused to GAL4. Interaction was assessed by growing transformant yeast strains on selective plates followed by data analysis as previously described [18]. Briefly, selective growth of diploid yeast colonies was analyzed by the Matlab-based image-analysis program TIDY which quantifies bright spots, representing yeast colonies to the dark background. For every biological replicate in the screen, each bait-TF interaction was analyzed in four technical replicates resulting in quandrants of yeast colonies as shown for AlkEB9 DNA bait in the results.

Generation of transgenic flies
For generation of lacZ reporter flies, DNA sequences of for AlkEB6 to AlkEB11 were PCR amplified (S4 Table) and cloned into the eve.p-lacZ.attB vector [52]. In addition, the AlkEB9 DNA bait was cloned into pPT-GAL vector (1225, DGRC) to generate the AlkEB9-GAL4 construct. DNA sequences for AlkEB8ΔBapBS-lacZ, AlkEB9_OpaBS-lacZ and AlkEB9_OpaKO-lacZ were assembled by Genscript and cloned into eve.  Table). Constructs were sequenced (GATC Biotech) and injected into w 1118 flies, except for attB constructs which were injected into Bloomington 24482 and 24485, for PhiC31 directed integration at 51C and 68E respectively (BestGene Inc.).

Generation of Alk mutants
Deletions within the Drosophila Alk enhancer region were generated with CRISPR/Cas9 [53]. The sgRNA targeting sequences used (listed in S1 Table) were cloned into pBFv-U6.2 expression vector (Genome Engineering Production Group at Harvard Medical School). Constructs expressing sgRNA were injected into vasa (vas)-Cas9 (Bloomington 51323) embryos by BestGene Inc. Screening of deletion events was performed by PCR and further sequencing (GATC Biotech). For additional complementation tests we employed balanced Alk 10 or Df(2R) Exel7144 flies.

Electrophoretic Mobility Shift Assay (EMSA)
DNA coding sequence of opa was synthesized (Genscript) in frame with carboxy-terminal OLLAS and 6xHis tags and cloned into the pcDNA3.1(+) mammalian expression vector. Binding of Opa to the AlkEB9 was analyzed by a DNA binding assay on dsDNA oligonucleotides with cell lysates from HEK-293F cells expressing Opa-OLLAS. Binding reactions were performed as described in [54] containing 10 mM Tris-HCl (pH 8.0), 25 mM KCl and 1 mM DTT, 1 μg poly-dIdC (Sigma-Aldrich), 2.5% glycerol, 0.05% Triton X-100, 0.2 mM MgCl2 and the indicated 3'-end biotin labelled probe. After 20 min incubation at room temperature, reactions were separated on a 6% native TBE-PAGE in 0.5x TBE buffer at 100V. DNA was transferred to nylon+ membranes (Amersham), UV cross-linked to the membrane and detected by Chemoluminiscence Nucleic Acid Detection Module (Pierce) according to manufacturer's indications. Competition assay was performed by addition of 100 fold molar excess of unlabeled competitor DNA to the reaction mix. Wild-type probes used for band shift experiments were Opa_SELEX and Opa_JASPAR. Mutated version were made according to for Opa_SELEX mutant [23], and in a similar manner for Opa_JASPAR mutant. All four EMSA probe sequences are shown in S3 Table.

Chromatin immunoprecipitation
Chromatin was prepared from approximately 100 mg of pooled collections of fixed 3-4 hour embryos. The embryos were homogenized for 1 min in 10 mM EDTA and 50 mM Tris (pH 8.1). After addition of SDS to a final concentration of 1% and incubation on ice for 10 min, glass beads (150-200 μm) were added and the homogenates were sonicated to give sheared chromatin preparations with an average DNA size of 300-400 bp. Chromatin immuno-precipitation was performed largely as described previously [55] using an affinity-purified anti-Opa antibody raised against a truncated recombinant protein spanning from amino acids 125-507 of Opa, a region containing the DNA-binding zinc-fingers at a concentration of 0.5 μg/ml with 100 μg of chromatin in 1 ml of 0.01% SDS, 1% TritonX-100, 1 mM EDTA, 20 mM Tris, pH 8, 150 mM NaCl and 1x Protease Inhibitor Cocktail (Roche). After overnight incubation of the chromatin and antibody at 4˚C, the mixture was incubated with Protein-A Agarose (Millipore) for 2 hours at room temperature, followed by low-salt, high salt and LiCl washes as used in the Chromatin Immunoprecipitation Assay Kit (Upstate Biotechnology). After heat reversal of protein-DNA crosslinks, protein digestion, phenol chloroform extraction and purification of the nucleic acids by ethanol precipitation the amount of recovered DNA was quantified using qPCR and a standard curve generated for each primer pair with a sample of nucleic acid purified from the input chromatin. The control primer pair produces a 115 bp amplicon located 12.4 kb upstream of odorant receptor 42b, a region devoid of modEncode hallmarks of cis-regulatory DNA sequences. The DESE-Opa primer pair produces a 209 bp amplicon from a central region of the slp1 DESE enhancer that requires Opa for expression [56]. The Alk primer pair produces a 140 bp amplicon that extends from 21 bp downstream of the SELEX_OpaBS to 53 bp upstream of the JASPAR_OpaBS. The ChIP values that are reported are percent precipitation relative to input DNA with error bars representing the mean ± S.D. from three technical replicates of the qPCR. The sequences of the primers are summarized in S3 Table and are