The Tbx6 Transcription Factor Dorsocross Mediates Dpp Signaling to Regulate Drosophila Thorax Closure

Movement and fusion of separate cell populations are critical for several developmental processes, such as neural tube closure in vertebrates or embryonic dorsal closure and pupal thorax closure in Drosophila. Fusion failure results in an opening or groove on the body surface. Drosophila pupal thorax closure is an established model to investigate the mechanism of tissue closure. Here, we report the identification of T-box transcription factor genes Dorsocross (Doc) as Decapentaplegic (Dpp) targets in the leading edge cells of the notum in the late third instar larval and early pupal stages. Reduction of Doc in the notum region results in a thorax closure defect, similar to that in dpp loss-of-function flies. Nine genes are identified as potential downstream targets of Doc in regulating thorax closure by molecular and genetic screens. Our results reveal a novel function of Doc in Drosophila development. The candidate target genes provide new clues for unravelling the mechanism of collective cell movement.


Introduction
Closure of two separate epithelia is a common phenomenon in development and tissue repair of metazoans. It allows the formation of structures such as the vertebrate neural tube or embryo and thorax of Drosophila. Till now, over 200 genes have been reported to be involved in neural tube closure [1,2]. Movement and fusion of separate cell populations are critical processes for correct morphogenesis. Fusion failure results in an opening or groove on the body surface [3]. For example in humans, neural tube defect (NTD) is a common birth defect due to the failure of spinal column closure. NTD is characterized by a small or missing brain, or the exposure of the spinal cord [4]. Drosophila dorsal closure in the embryonic stage [5,6] and thorax closure in the pupal stage [7,8] are convenient models for investigating the mechanisms of tissue fusion processes.
The adult thorax of Drosophila is derived from the notum parts of two larval wing imaginal discs [9]. Imaginal discs are invaginations of the larval epidermis made up of two opposing epithelia, a columnar main epithelium and a squamous peripodial epithelium. The differentiation into the two types of epithelium arises during larval development and requires continuous signaling input [10][11][12][13]. The two wing imaginal discs develop in spatial separation [14]. At the end of larval development, the discs evert [15,16] and the two heminota move toward the midline and fuse to form a complete thorax in the early pupal stage [8,17,18]. Cells at the junction between main and peripodial epithelium are of prime importance for the fusion process [17].
Decapentaplegic (Dpp), which is a member of the transforming growth factor-β superfamily, is essential for thorax closure. Disrupted Dpp signaling either at the level of its receptors Thickveins (Tkv) and Punt [8] or at the level of the signal transducer Medea [19] results in a thorax closure defect. Lack of Dpp signaling causes a block of filopodia extension and a collapse of the leading edge [7]. In the thorax, pannier (pnr) [20][21][22] and eyegone (eyg) [23] have been demonstrated to be regulated by Dpp in a concentration dependent manner. pnr, encodes a zinc-finger transcription factor of the GATA family, is positively regulated by Dpp signaling to specify the medial fate of notum. This fate specification is regulated either by activating another important morphogen Wingless (Wg) in the most lateral region or by repressing Wg in combination with transcriptional factor U-shaped in the medial region of notum [24]. Ectopic expression of pnr induces extra bristles which is a character of the thorax media patterning [21]. eyg, which encodes a homeodomain Pax protein, is expressed in the anterior region of notum by negative regulation of Dpp and positive regulation of Pnr [23]. The specific eyg expression in the anterior notum together with drumstick (drm) and bowl is required for the notum growth and patterning. Therefore, downstream factors mediating the role of Dpp signaling in thorax closure have not yet been identified.
Embryonic dorsal closure, another model for the standard epithelium fusion, occurs after germ band retraction with the dorsally directed migration of the lateral epidermis over the amnioserosa from both sides of the embryo [25]. Dpp signaling also regulates the dorsal closure by mediating the epidermis adhesion and the contraction of amnisoerosa [26]. The Dorsocross (Doc) genes, encode three functionally largely redundant T-box transcription factors Doc1, Doc2, and Doc3 of the Tbx6 subfamily [27][28][29][30][31], are expressed both in the embryonic dorso-lateral ectoderm and in the amnioserosa under the control of Dpp [30,[32][33][34]. The three Doc genes are collectively referred to "Doc genes" based on the similarities sequence, expression patterns and genetic analysis. Loss of Doc perturbs amnioserosa function during germ band extension and retraction [30] such that a direct role of Doc in embryonic dorsal closure could not be determined. Expression and function of Doc extends beyond amnioserosa and the embryonic ectoderm to e.g., visceral and cardiac mesoderm development [32]. Right now, Doc2 antibody staining was used as a proxy to identify Doc protein expression pattern.
In the wing imaginal disc, there are four Doc expressing domains: two large domains located in the dorsal and ventral regions of the prospective wing blade and two smaller domains in the prospective dorsal hinge and posterior notum regions [30]. The two large wing pouch Doc expression domains are critical for the formation of a fold between hinge and blade regions of the larval disc and for wing disc bending [35]. However, the role of the smaller Doc expression domains is still unknown. Thoracic closure provides a setting in which the involvement of Doc in tissue fusion processes can be tested.
In this study, we investigate the role of Doc expression in the posterior notum. We show that this expression domain corresponds to the future leading edge of the fusion discs. Doc expression required Dpp and Pnr activity. When Doc was reduced in the leading edge cells of the notum, a thorax closure defect was induced similar to the phenotype caused by down-regulated Dpp signaling. Genetic analysis identified nine genes as downstream targets of Doc in regulating thorax closure. This novel function and identified candidates provide clues for future collective cell movement study.

Doc Is Expressed in the Leading Edge Cells during Thorax Closure
We and others have shown that, in late L3 wing discs, Doc expression includes a stripe of cells in the posterior medial notum [30,35]. We wondered whether this part of the Doc expression pattern was in the future leading edge cells. Thus, Doc antibody staining was performed from the mid 3rd larval instar to early pupal stages. There was no Doc expression in the posterior notum in the mid L3 wing disc ( Figure 1A). Doc expression in the posterior notum was first detectable in late L3 wing discs ( Figure 1B). In a crosssectional view, it was obvious that Doc expression was restricted to the main epithelium and was excluded from the peripodial epithelium ( Figure 1B',B"). At this stage, Doc and puc-lacZ[A251] were expressed along the posterior lateral region in a non-overlapping pattern ( Figure 1C). During early pupal development (6 h After Puparium Formation, APF), Doc and puc-lacZ[A251] were co-expressed in the leading edge cells of the fusing heminota ( Figure 1D-D"). The enhancer trap line puc-lacZ[A251], which in the embryo faithfully reflects JNK pathway activity during dorsal closure [36,37], has a more restricted expression pattern than puc-lacZ[E69] in the wing disc [16,17,38], so that actual JNK pathway activity and Doc expression may overlap already in the L3 wing disc. Our results show how Doc expression expands from the late L3 to the early pupal stage, to finally take up the entire leading edge domain. The co-localization of Doc and puc-lacZ suggests a role of Doc in thorax closure.
Since Dpp signaling is required for correct thorax closure [8,19], we tested for Dpp expression pattern and signaling in the leading edge cells during thorax closure. In tissue fusion, three steps have been discerned: a physiological and structural change of the leading edge cells, the joint movement of epithelia towards the midline, and fusion of the two separate epithelia [3,39,40]. We used dpp-Gal4 driving UAS-GFP as a marker of dpp expression and puc-lacZ, to visualize the leading edge cells [8,17,38]. We found that GFP expression gradually expanded and co-localized with puc-lacZ in the leading edge cells from 5-6 h APF ( Figure 1E,F). At 6-8 h APF, during which stage the heminota began to fuse, the dpp > GFP signal was still present and co-localized with puc-lacZ ( Figure 1G). Similarly, a lacZ reporter of dpp was also observed in the leading edge cells ( Figure 1H). The presence of Dpp signaling in leading edge cells was also demonstrated by PMad antibody staining ( Figure 1I). From the above results, we conclude that Dpp signaling is present in the leading edge cells throughout the thorax closure process.

Doc Is Regulated by Dpp Signaling in the Notum Region of the Wing Disc
In the embryonic dorsolateral ectoderm, Doc acts downstream of Dpp signaling [34]. Given the co-expression of Doc and dpp in the wing disc notum [30,35] (Figure 2A), we asked whether Doc is a Dpp target during thorax closure. To block Dpp signaling, a dominant negative form of the receptor Tkv (tkv DN ) [41] was expressed in the ap-Gal4 domain. Doc expression was lost from the entire dorsal compartment including the presumptive leading edge cells ( Figure 2B,B').
pnr is induced by Dpp in the embryonic dorsal ectoderm and is required for dorsal closure. However, after embryonic stage 10, dpp becomes Pnr-dependent in the dorsal ectoderm [42]. pnr is also induced by Dpp in the notum region of the wing disc and is required for dorso-ventral patterning [24,43]. Transheterozygous combinations of pnr alleles show thorax fusion defects [43]. By pnr knock-down, we showed that pnr was required for notal Doc expression ( Figure 2C,C'). Another factor specifying notum development is eyg, which is induced by Pnr but repressed by Dpp [23]. Missexpression of eyg in the posterior notum region of the wing disc blocked Doc expression ( Figure 2D,D'). Broad expression of pnr in the notum region of the wing disc did not suffice to extend the Doc expression domain, presumably due to maintained expression of antagonistic factors such as Eyg ( Figure 2E). Co-expression of pnr was sufficient to rescue Doc repression caused by tkv DN expression ( Figure 2F,G) suggesting that Pnr acts downstream of Dpp signaling in the induction of Doc.
However, ectopic expression of dpp ( Figure S1A) or knocking-down of eyg ( Figure S1B) had no effect on Doc expression in the wing disc. This implies other factors, in addition to Dpp signaling, regulate Doc expression in the wing disc notum region.

Doc Is Required for Correct Thorax Closure
The Drosophila adult notum exhibits a smooth surface with regularly arranged bristles ( Figure 3A). Inhibition of Dpp signaling in the medial notum by tkv DN caused a thorax closure defect [8] (Figure 3B). A similar phenotype was obtained by knocking-down pnr ( Figure 3C). Using the adult phenotype, we asked whether Doc plays role in thorax closure. Partial knock-down of Doc by Doc1-RNAi plus Doc2-RNAi ( Figure S2A) or by Doc3-RNAi ( Figure 3D) resulted in a thorax closure defect, demonstrating a requirement for Doc in correct thorax closure. A quantification of defect frequency is shown in Figure S2B. The phenotype penetrance between Doc3-RNAi expression and Doc1 + 2-RNAi co-expression was similar which indicates that the redundant role or the dosage effect may be involved in regulating thorax closure by Doc genes. Recovery of Doc by pnr expression was sufficient to rescue the thorax closure defect when Dpp signaling was suppressed ( Figure 3E,F). The above results demonstrate that Doc acts downstream of Dpp signaling and is essential for thorax closure.

Functional Analysis of Potential Doc Target Genes
In a preliminary molecular screen, nine genes were selected for subsequent phenotype assays based on their differential expression levels in Doc-RNAi wing discs compared with wild type control ( Figure S3A; Table S1). The 2 kb upstream of the nine genes' promoter contained a sequence motif related to the Doc binding consensus TTCACACCT ( Figure S3B). In the following genetic analysis, seven genes (arpc3B, GC12164, CG14456, CG16758, βtub97EF, attc, mec2) which were downregulated in Doc-RNAi manipulation were investigated by crossing with pnr-Gal4 for their effect on thorax development. Severe defects were observed when either of them was knocked-down by RNAi ( Figure 4A-G). pupal cuticle protein (pcp) and secreted decoy of InR (sdr) were upregulated under Doc-RNAi treatment and were selected for gain-of-function assays. Overexpression of pcp or sdr under pnr-Gal4 caused a mild thorax defect ( Figure 4H,I). The thorax closure defect severity was evaluated by the gap between the two middle bristles. The distance of this gap was measured and treated as the thorax closure index which suggest significant difference between the gene manipulation and the wild type control ( Figure 4J). Altogether, nine genes were identified as potential Doc target genes regulating thorax closure.

Discussion
The fusion of two separated epitheliums is occurred in nearly all the organisms which provides the possibility of forming a complete structure, such as the vertebrate neural tube, wound healing, complete Drosophila embryo and thorax. In this study, we present a genetic analysis of Doc function in Drosophila notum development to present a potential role of Doc in epithelium fusion. We show that Doc acts downstream of Dpp signaling in regulating thorax closure. Nine potential Doc target genes were identified in regulating thorax closure by an RNAi-based genetic screen with subsequent phenotypic analysis.

Dpp Signaling Promote Thorax Closure
Drosophila thorax closure requires numerous signal transduction pathways. Fos [7] and JNK [7,38], and Dpp signaling [8,19,44,45] are the best characterized pathways in this process. Dpp signaling was activated in the leading edge cells throughout the fusion process ( Figure 1A-E). Inhibition of Dpp signaling resulted in a severe thorax closure defect ( Figure 3A,B). pnr and Doc, acting as the Downstream of Dpp signaling, were essential for thorax closure ( Figure 3C-I).
Dpp and JNK are also required for thorax closure. In contrast to dorsal closure, dpp is not controlled by JNK signaling in thorax closure [7,38]. Dpp is involved in the cell shape change of the leading edge and the spreading and adhesion of the two opposite cell sheets [8], while JNK appears predominantly involved in the leading edge cell shape changes [8]. However, Usui and Simpson [17] noted that in thorax closure, the JNK pathway was activated (puc expression) in two adjacent cell types differing in nuclear size which they termed S (talk) and I (imaginal cells). Based on observation, they attributed different function to S and I cells in the fusion process and assumed that I cells generate the force for disc migration. Doc-positive cells, judged by their location at the junction between main and peripodial epithelium ( Figure 1B') and their exclusion from the disc stalk ( Figure 1C), can be considered as I cells. The fact that Doc knockdown caused a thorax fusion defect lends credit to the hypothesis of Usui and Simpson [17].

Doc Function in Other Epithelia
Doc function was first demonstrated in the embryonic dorso-lateral ectoderm and in the amnioserosa. Doc knock-down perturbs germband extension and retraction and amnioserosa development [30,[32][33][34]. Loss of Doc perturbes amnioserosa function already during germ band extension and retraction [30]. These early effects precluded a direct analysis of Doc involvement in dorsal closure. A detailed analysis of the cell biological effects of Doc loss on germ band movement has not yet been performed. In the flour beetle Tribolium castaneum, the single Tc-Doc is not required for specification and maintenance of its amnion and serosa. This allowed investigation of whether and how Tc-Doc affects morphogenesis of the extraembryonic epithelia in this species. The first rearrangement of extraembyonic tissue (closure of the serosal window) is slowed down or blocked by Tc-Doc-RNAi [58]. Doc is also involved in the reorganisation of the Drosophila embryonic hindgut epithelium which initiates the formation of the four Malpighian tubles (MpT). The two pairs of MpTs develop from the hindgut at the mid/hindgut junction. Doc expression is restricted to where the anterior pair of MpTs buds from the hindgut. In Doc mutant embryos, development of the anterior pair of MpTs is suppressed while the posterior pair develops normally [59]. The precise role of Doc in MpT development is, however, not yet known [60]. We have previously analyzed the morphogenetic function of Doc in Drosophila wing imaginal disc development. Doc is expressed in the wing pouch (destined to develop into the wing blade) at the junction to the future wing hinge. These domains of the wing disc epithelium are separated by a deep fold. Doc promotes fold formation by inducing matrix metalloproteinase 2 expression and by causing a reorganization of the microtubule cytoskeleton [35]. Currently it cannot be judged whether Doc promotes restructuring of different epithelia via common targets.

Potential Doc Target Genes in Regulating Thorax Closure
Based on their Doc dependence and their adult notum phenotype, nine genes were identified as potential Doc downstream targets. The most severe thorax closure defect was observed by knocking down arpc3B and βtub97EF. arpc3B has been shown to play a role in organizing the actin cytoskeleton in yeast and Drosophila [2,[61][62][63][64][65]. βtub97EF is essential for microtubule stability in the Drosophila gut [66]. We propose that the actin and microtubule cytoskeleton is required for thorax closure. Our knock-down analysis also identified three previously unidentified genes (CG12164, CG14456, and CG16758) as potential Doc target genes in thorax closure. pcp and sdr were upregulated in Doc hypomorphic wing discs and caused defective thorax development upon overexpression. While the presence of sequence motifs related to the Doc binding consensus in the upstream region of these genes is suggestive, it does not provide sufficient evidence for direct regulation.

The General Character between Dorasl Cloure, Thorax Clousre and Neural Tube Closure
Two processes are involved in the closure process: cell movement and fusion of two separated epithelium [3,39,40]. During the dorsal closure, the leading edge cells stretch along the dorsal-ventral axis by forming filopodial and lamellipodial protrusions [67,68], which are rich in F-actin [3]. The increase of F-actin pulls the whole epidermal forward over the amnioserosa surface. Under this force, the leading edge cells meet at the midline and begin to fuse together. The fusion process is guided by microtubule-dependent filopodia, which make sure the zippering of the opposing epithelial sheets and the correct matching of opposite segment [5,69,70]. The adherens junction and microtubule regulators are delivered to the leading-edge filopodia to make sure the adhesion of two adjoining epithelial sheets [71]. During the thorax closure, the movement of two epitheliums is achieved by the extension of F-actin which pulls the whole epithelium moving forward towards the midline. When two epitheliums are abutting, the adheres-type junctions are accumulated in leading edge cells, preparing for the adhesion of these two separated parts. The cytoskeleton also involved in the processes of thorax closure. Similarly, successful neural tube closure also involves fusion of the opposite neural folds to create the complete neural tube [72]. In all vertebrates, cellular protrusions, such as spike-like filopodia or sheet-like lamellipodia, are formed at the edge of the neural folds for the preparation of the neural fold attachments [73][74][75]. The components of adherens and tight junctions are also involved in the regulation of neural tube closure [76,77].
The potential role of Doc in regulating the cytoskeleton and microtubule dynamic or stability provides the general interests of epithelium fusion, including complete vertebrate neural tube formation, Drosophila embryo dorsal closure, and thorax closure. The collective cell movement and fusion of two separated epithelium [3,39,40] are the general character of forming a complete epithelium. Our study provides nice clue and candidate genes for epithelium formation in neural tube closure and wound healing.

Dissection of Larvae
Wing imaginal discs were dissected according to standard protocol and were fixed for 30 min in 4% paraformaldehyde in PBT (PBS with 0.3% Triton X-100).

Dissection of Pupae
The dissection protocol of pupae was modified from Martin-Blanco et al. [51]. The staged pupae were torn into an anterior and a posterior part. The latter, together with gut, salivary glands, central nervous system, and fat body was discarded. The anterior part with wing discs or notum still inside the pupal case was fixed for 1h in the 4% paraformaldehyde. Then, the ventral pupal case was removed from the fixed thorax followed by the immunohistochemistry steps. After the immunostaining, the fixed tissue was flattened with the prepupal wing and six leg disc orientated distally. Pupal staging is in hours after puparium formation (APF).

Adult Thorax Imaging
Adult thorax images were collected using microscope UV-CTS (Beijing, China).

Potential Target Genes Selection
The genes selected from the transcriptome analysis were used to blast with the Doc transcriptional factor binding site (TFBS, "TTCACACCT", https://mccb.umassmed.edu/ ffs/TFdetails.php?FlybaseID=FBgn0035956; accessed on 11 March 2022) in the 2 kb upstream region of the promotor. Twenty-seven of 177 genes were selected for the genetic screen based on the expression levels. The potential target genes were determined by the genetic screen crossing with the pnr-Gal4 either with the RNAi lines or the UAS lines based on the expression trend.

Thorax Closure Index
The distance between the two middle bristle was measured as the thorax closure index. For each Drosophila, three lines were drawn vertical to the thorax closure fusion mid-line for the measurement and at least five flies were used for quantification.

Conclusions
We characterized the spatio-temporal changes of Dpp signaling in the leading edge cells during thorax closure detailing the role of Dpp in thorax closure. Doc was identified as a gene complex acting downstream of Dpp signaling by analyzing its expression pattern and its dependency on known Dpp pathway components. By molecular and genetic screens, nine genes were identified as potential Doc target genes with a role in thorax development ( Figure 5).