Serpent/dGATAb regulates Laminin B1 and Laminin B2 expression during Drosophila embryogenesis

Transcriptional regulation of Laminin expression during embryogenesis is a key step required for proper ECM assembly. We show, that in Drosophila the Laminin B1 and Laminin B2 genes share expression patterns in mesodermal cells as well as in endodermal and ectodermal gut primordia, yolk and amnioserosa. In the absence of the GATA transcription factor Serpent, the spatial extend of Laminin reporter gene expression was strongly limited, indicating that Laminin expression in many tissues depends on Serpent activity. We demonstrate a direct binding of Serpent to the intronic enhancers of Laminin B1 and Laminin B2. In addition, ectopically expressed Serpent activated enhancer elements of Laminin B1 and Laminin B2. Our results reveal Serpent as an important regulator of Laminin expression across tissues.

postulated co-expression 7,8,15,16 , to characterize gene expression of early embryonic tissues and late embryonic secretion. Both genes show small upstream (upstream enhancer, UE) and large intronic (intronic enhancer, IE) enhancers. Bioinformatic analysis of the large first introns revealed several conserved regions in these enhancers, indicating the presence of cis-regulatory modules (CRMs). Only LanB2 displayed an additional small conserved region in the corresponding UE. Therefore, we generated reporter constructs by fusing the derived CRMs of both Laminin genes to GFP, analyzed the derived tissue-specific expression and compared it to the described mRNA and protein distribution (Supplementary Table S1 and Figs 1 and 2).
The IE 5′ constructs of both Laminins revealed expression in the early mesoderm as well as in the amnioserosa (LanB1 IE 5′ in Fig. 1C,F and LanB2 IE 5′ in Fig. 2C,F, amnioserosa not visible in 2 C). At the end of embryogenesis, expression was found in the mesoderm derived somatic muscles (LanB1 IE 5′ in Fig. 1I,I′ and LanB2 IE 5′ in Fig. 2I,I′) and the visceral muscles (LanB1 IE 5′ in Fig. 1I,I′ and LanB2 IE 3′ in Fig. 2J,J′). Additional expression in endodermal midgut primordia was displayed for the LanB1 IE 3′ (Fig. 1G) as well as LanB2 IE 5′ (Fig. 2F) construct.
Laminin reporter gene expression in hemocytes and fat body cells could be detected from early differentiation and was observed for the LanB2 IE 5′ construct (LanB2 IE 5′ in Fig. 2F,I,I′) as well as in both 3′ constructs (LanB1 IE 3′ in Fig. 1G,J,J′ and LanB2 IE 3′ in Fig. 2G,J,J′). Both Lan IE 3′ constructs showed further expression in some cells of the peripheral nervous system, in the dorsal median cells (both not shown) and the yolk (LanB1 IE 3′ in Fig. 1D,G,J,J′ and LanB2 IE 3′ in Fig. 2D,G,J,J′). The LanB1 IE 3′ construct displayed additional expression in salivary glands and Malpighian tubules (both not shown). Although the LanB2 UE region contains a small conserved region ( Supplementary Fig. S2), the analysis revealed no embryonic reporter gene expression, indicating no CRMs for embryonic expression in this UE. In summary, the reporter constructs reflect the complete known embryonic expression of LanB1 and LanB2 (Supplementary Table S1), so that all CRMs promoting embryonic expression should also be included. anterior and posterior midgut primordia, fb: fat body, fge: foregut epithelium, hem: hemocytes, mes: mesoderm, mge: midgut epithelium, sm: somatic muscles, vm: visceral muscles. Scale Bars = 100 µm.
The comparative Laminin B1 and B2 protein distribution appeared initially in a layer between the mesoderm and ectoderm of embryos at the fully elongated germ band stage and was continued in the somatic and visceral mesoderm as well as in the endodermal midgut primordia. At the end of embryogenesis, LanB1 and LanB2 covered most tissues and were strongly secreted by fat body and blood cells (LanB1 in Fig. 1H,H′ and LanB2 in Fig. 2H, H′ 7,8 ). In conclusion, every tissue in which LanB1 and LanB2 could be detected at the end of embryogenesis seemed to express Laminin itself for its own initial BM assembly.
In silico prediction of putative Srp-binding sites in LanB1 and LanB2. In order to identify transcriptional regulatory mechanisms, we searched for potential regulators controlling LanB1 and LanB2 expression using transcription factor binding profile databases 17,18 . Conservation scores (PhastCons datasets of 14 insect species) 19 were used to identify and eliminate the false positive transcription factor binding sites (TFBSs) enriched in the non-coding regions, based on the assumption that binding sites essential for Laminin expression are strongly conserved across insect phylogeny. We found an overrepresentation of potential binding sites for Srp 20,21 in the intronic enhancers (IE) of LanB1 and LanB2, and one additional hit in the upstream enhancer (UE) of LanB2. These binding sites showed strong conservation scores and were located partially in conserved regions, but also at loci, where only the binding sites displayed peaks of conservation (Supplementary Figs S1 and S2).
LanB1 and LanB2 reporter gene expression in srp mutant background and upon tissue-specific srp knockdown. To test whether expression of LanB1 and LanB2 depends on the GATA-factor Serpent (Srp), we analyzed Laminin reporter gene expression in hypomorphic srp neo45 mutant embryos, in which only the differentiation of hemocytes is affected, and in srp 3 null mutant embryos 13,22 (Fig. 3).
In srp neo45 mutants, hemocytes were missing and within reporter gene expression of hemocytes, whereas expression in all other tissues was still observed (Fig. 3E-H). In contrast, in srp 3 embryos the fat body and the Srp binds directly to the intronic enhancers of LanB1 and LanB2. The experiments described above indicate a potential regulatory role of Srp for LanB1 and LanB2 expression. In order to test a direct binding ability of Srp to LanB1/LanB2, we performed ChIP analysis followed by subsequent qPCR. For LanB1 expression, we analyzed the Srp enrichment at five genomic regions, with seven predicted GATA/Srp-binding sites (Supplementary Fig. S1, with marked genomic location). At the locus, represented by the LanB1 IE 5′ construct, we checked two GATA-transcription factor binding sites (GATA-TFBS 1 and GATA-TFBS 2) for enrichment. The results showed only a slight increase of Srp compared to the control ( Fig. 4B; GATA-TFBS 1: control 0.05% of input (referred hereafter as %), Srp::GFP 0.08%; GATA-TFBS 2: control 0.03%, Srp::GFP 0.14%), whereas the enrichment of Srp at the locus represented by LanB1 IE 3′ was considerably higher than that in the control (Fig. 4B Srp-TFBS 1 + 2: control 0.03%, Srp::GFP 0.36%; GATA-TFBS 3 + 4: control 0.02%, Srp::GFP 0.32%; Srp-TFBS 3: control 0.03%, Srp::GFP 0.36%). The negative controls demonstrated that control primers amplify ChIP-DNA with a slight variation, but do not show enrichment over input ( Fig. 4B control 1: control 0.06%, Srp::GFP 0.08%; control 2: control 0.05%, Srp::GFP 0.07%). Taken together, these results display a weak enrichment of Srp for the GATA-TFBS 2, located at LanB1 IE 5′, and a high significant enrichment of Srp in the LanB1 IE 3′ region. This was also evident in the context of reporter gene expression of these constructs (Fig. 1), because LanB1 IE 5′ display Srp-regulated Laminin expression only in the amnioserosa (Fig. 1C,F). This tissue persists just in the first half of embryogenesis and undergoes programmed cell death thereafter 25 . Therefore, just a small fraction of cross-linked cells could influence the enrichment of Srp in this approach. In contrast, LanB1 IE 3′ showed a high Srp enrichment and strong reporter gene expression in the endoderm, hemocytes and fat body, whose determination and differentiation depends on Srp activation (Figs 1 and 3).
For analysis of LanB2 regulation, we investigated three genomic loci, including four potential GATA-and Srp-binding sites ( Supplementary Fig. S2, with marked genomic location). The first Srp-binding site is located in the upstream enhancer (Srp-TFBS 1), one is found in the 5′ region of the intronic enhancer (GATA-TFBS 1), and two potential binding sites are located in the 3′ region of the intronic enhancer (Srp-TFBS 2 and GATA-TFBS 2; Fig. 4C). The ChIP analysis detected no enrichment of Srp in the upstream enhancer (Srp-TFBS 1: control 0.06%, Srp::GFP 0.06%), whereas both loci in the intronic enhancer showed a significant enrichment of Srp protein (GATA-TFBS 1: control 0.02%, Srp::GFP 0.15% for LanB2 IE 5′; Srp-TFBS 2 + GATA-TFBS 2: control 0.03%, Srp::GFP 0.20% for LanB2 IE 3′). These data are in agreement with the previous reporter gene analysis (Fig. 2), as we did not find any reporter gene expression of LanB2 UE in the embryo, while LanB2 IE 5′ and LanB2 IE 3′ displayed Srp-dependent Laminin expression patterns (Fig. 2).
Taken together, the strong Srp enrichment in both regions of LanB2 intronic enhancer constructs and the strong enrichment found in LanB1 IE 3′ indicate that hemocyte and fat body regulation show the strongest effect for Srp binding. In addition, the enrichment in the LanB1 IE 5′ region indicates a regulatory role of Srp for Laminin expression in the amnioserosa. www.nature.com/scientificreports www.nature.com/scientificreports/ ectopically expressed srp activates Laminin reporter gene expression. To test whether Srp is able to induce LanB1 and LanB2 expression, we expressed srp 27 ectopically in all muscle cells using the mef2-GAL4 driver in the background of our LanB1 IE 3′ (Fig. 5) and LanB2 IE 3′ (Supplementary Fig. S3) constructs, which displayed no myogenic expression (Figs 1J,J′ and 2J,J′). At the end of embryogenesis, the somatic musculature builds a specific stereotypical pattern. These myotubes were characterized by persistent mef2 expression (revealed by mef2-GAL4 driven UAS-lacZ, Fig. 5A,A″) and a characteristic cell shape in control animals (Fig. 5A,A″′). LanB1 IE 3′ GFP reporter gene expression in control embryos was restricted to fat body and hemocytes, whereas ectopically expressed srp induced reporter activity was detected in several somatic muscles (arrows, Fig. 5B″,B″′) and unfused myoblasts (Fig. 5B″,B″′), displaying ectopic activation of CRMs from LanB1 IE3′. Similar results were obtained with the LanB2 IE 3′ GFP reporter (Supplementary Fig. S3).
This gain-of-function experiments demonstrated that srp expression is sufficient for LanB1 and LanB2 enhancer activation. Interestingly, ectopic expression of srp led to a massive disruption of myoblast fusion, reminiscent of several myoblast fusion mutants [28][29][30][31] . So even at the end of morphogenesis, unfused myoblasts and disarranged fused muscles were clearly visible all over the somatic and the visceral mesoderm (Fig. 5B,B″′), indicating an important role of tissue-specific Srp expression and subsequent gene regulation.

Discussion
Laminins are crucial components for BM construction. Our results indicate a common tissue-specific expression of LanB1 and LanB2. We also identified cis-regulatory modules (CRMs) for this tissue-specific expression. The expression of the Laminin β/γ subunits is essential for subsequent di-and trimerization, and reflects the onset of earliest ECM assembly.
Laminins are expressed in all germ layers (this study 7,8,15,16 ) and an initial Laminin network exists prior to further ECM component secretion (this study 32 ), which is responsible for complex developmental processes such as cell migration and organogenesis 7,8,33,34 , suggesting complex regulatory mechanisms. While the loss of Laminin reporter gene expression in srp mutants (Fig. 3) could also be explained by lack of differentiation of these tissues, ectopic activation of Srp in muscle cells (Fig. 5) indicates a regulatory role for Laminin expression. Based on the identified CRMs (Fig. 4), our results indicate that Srp the main GATA-transcription factor in Drosophila, act as Laminin regulator in distinct tissues and germ layers (Figs 1, 2, 5). In this context, it seems evident that Srp, owing to its expression in multiple tissues, is also able to regulate Laminin expression in a variety of tissues.
While, in Drosophila, Srp is the single most important GATA-factor for development, in vertebrates, multiple GATA-orthologues play important and tissue-specific roles during differentiation 24 . In this context, it seems interesting that Srp binds at the predicted GATA-TFBS (b in Fig. 4B,D), as this binding site was found with a PWM for mouse GATA1. The role of Srp for hemocyte/blood cell differentiation may be conserved throughout vertebrate development 35 . Therefore, it appears that there is a Srp-binding site closely related to the vertebrate GATA1-TFBS, while the Srp-TFBS reported here is similar to the vertebrate GATA4-and GATA6-TFBS 17 .
In cultured mouse embryonic stem cells, forced vertebrate GATA4 and GATA6 expression strongly enhances and silences laminin-1 production, respectively 36,37 . These experiments reveal no direct regulatory role of GATA-factors for Laminin regulation, but in the context of this study, it would be interesting to investigate, which GATA-factor takes over the regulatory role of Srp in vertebrates. Interestingly, the reciprocal loss of GATA4 or GATA6 in epithelium-derived tumor cells reveals a subsequent loss of Laminin expression 38 . In the context of our results and with the assumption that GATA-factors in vertebrates take over the role of Laminin regulation, it makes sense that epithelial cells show a strong expression of GATA-factors.
A recent study supports this notion, revealing a common role of Srp and human GATA4/6 in epithelial-mesenchymal-transition (EMT) 39 . Although, functions in epithelial maintenance and EMT seems to contradict each other, Laminin activation is required in mesenchymal cells for the migration-related processes 8,40 . As the presence of Srp disturbs the epithelial character in Drosophila 39,41 , we assume that the initial Laminin activation by Srp is essentially required in all Srp-dependent tissues, and that subsequent Laminin regulation for epithelial tissue assembly and maintenance must be ensured by other GATA-factors. Maintenance of gene expression in the differentiated midgut epithelium has already been demonstrated for dGATAe 42 , along with the fact that amnioserosa differentiation is dependent on dGATAa/pannier 43,44 .
Bioinformatics. Evolutionary conservation in non-coding regions of LanB1 and LanB2 were identified by using PhastCons datasets of 14 insect species 19 . The intergenic regions upstream of the first transcription start site and the intronic sequences of LanB1 and LanB2 were analyzed by CIS-BP Database (PWMs log Odds, Threshold: 0.8) 18 and JASPAR (Threshold: 0.8) 17 for potential TFBSs. Further manually separations were performed according to the conservation score 19 in the UCSC genome browser 51 using BDGP R5 dm3.
Chromatin immunoprecipitation. For preparation of chromatin, crosslinked 0-24 h old embryos (w 1118 and srp::GFP) were used. Embryos were collected and dechorionated as described in 45 . Fixation was performed as previously described 52 . For sonification embryos were treated with a Branson 250 with four bursts for 30 s and 20-40% power in 3 ml RIPA buffer and 0.5 ml acid-washed glass beads (Sigma, G8772). Chromatin purification was performed as in 53 . 1 ml chromatin extract were preabsorbed with 20 µl Protein A/G agarose beads (Merck, IP05), while anti-GFP (Abcam, ab290, 1:500) antibody was incubated with 20 µl agarose beads in 1 ml RIPA buffer at 4 °C for one hour with an overhead shaker. The preabsorbed chromatin extract was incubated with the antibody-coupled agarose beads over night at 4 °C with an overhead shaker. DNA was purified with spin columns and used as template for qPCR.

Data availability
All data generated or analyzed during this study are included in the manuscript.