Genetic basis for the evolution of organ morphogenesis: the case of spalt and cut in the development of insect trachea

One of the main processes that facilitates animal habitat adaptation is organ morphology diversification. Morphogenesis is a complex genetically programmed process. Modifications in this genetic program throughout evolution allowed the formation of a wide range of morphologies. Evidence of this phenotypic diversification in evolution has been mainly collected from the study of quantitative traits. However, the genetic mechanisms that explain how simple genetic changes account for the coordinated variations that give rise to modified functional organs is still poorly understood. Here, we address this question by analysing and comparing the tracheal system development of two insects.

The tracheal system is the respiratory organ of the insects. The development of the tracheal system has been deeply studied in Drosophila, where it develops from clusters of ectodermal cells that arises in either side of the embryo from the first thoracic (T1) to the last abdominal segment (A8) (for a review see Manning and Krasnow, 1993). Tracheal cells are specified by the activity of a set of ‘tracheal inducer genes’ that includes the transcription factors trachealess (trh) and ventral veinless (vvl) (for a review, see Ghabrial et al., 2003). Once specified, those tracheal cells invaginate and migrate to form a stereotypical network of tubes. This network consists of a couple of main dorsal trunks (DTs), specified by the transcription factor spalt (sal, also known as salm), which have many ramifications and are connected to the outside through specific structures called spiracles. Posterior spiracles in Drosophila develop from a cluster of epithelial cells in the dorsal side of the eighth abdominal segment (A8) that forms an internal multicellular tube, the spiracular chamber, which links the trachea to the exterior (Hu and Castelli-Gair, 1999). The Hox gene Abdominal B (Abd-B) specifies the formation of the posterior spiracles mainly by the activation of the transcription factors cut (ct), grain (grn) and empty spiracles (ems) (Lovegrove et al., 2006). In contrast, other insects present spiracles in every segment where trachea develops, which goes hand in hand with an overall different tracheal system morphology. Formation of such different tracheal morphologies depending on the development of few or many functional spiracles might imply an evolutionary relationship between both structures (Keilin, 1944). However, the genetic mechanisms underlying these morphologically linked changes remain elusive.

Here, we address this evolutionary question by describing the sequential branching process that gives rise to the Tribolium larval trachea and comparing that with tracheal development in Drosophila. In contrast to Drosophila, the Tribolium tracheal system presents functional lateral spiracles and lacks dorsal trunks (DTs), suggesting that the distinct morphology of the Tribolium tracheal system is coupled with different patterns of ct and sal gene expression. Furthermore, we show that changes in the domains of sal and ct expression are associated with the acquisition of each morphological innovation and that these two genetic modifications are connected both functionally and genetically.

The Tribolium tracheal system starts to develop 27 h after egg laying (AEL). It derives from 10 clusters of ectodermal cells (Tr1-Tr10) on both sides of the embryo, from the second thoracic (T2) to the eight abdominal (A8) segments (Fig. 1A,B). The specification and development of each cluster of cells starts with those in the most anterior segments and progress along the antero-posterior axis. The cells of each cluster invaginate and form an elongated sac of cells connected to the surface by a lateral spiracle (Fig. 1B,B′). A dorsal and a ventral bud then start to form and grow (Fig. 1C,C′). Four additional buds – three dorsal and one ventral – arise later, by 47 h AEL (Fig. 1D,D′). Cells from these buds migrate in various directions and to characteristic lengths and form the main tracheal primary branches that are about two cell diameters wide (Fig. 1E,E′). However, some cells do not migrate and remain clustered around the point of the initial invagination as a sac connected to the exterior by the lateral spiracle, which in Tribolium develops in all tracheal segments except Tr2. By 65 h AEL, the dorsal branches in each segment from the two body sides fuse and form a dorsal anastomosis connecting the right and left sides of the tracheal network. Afterwards, the ventral branches from neighbouring segments fuse and connect the tracheal network along the whole embryo (Fig. 1F,F′ and Fig. S1). In addition, as the tracheal branches elongate, their cells intercalate from a side-by-side arrangement to a one cell thick row organisation, forming narrower and longer branches (Fig. 1F″ and Fig. 2) that make up the final larval tracheal network. Each metamere comprises the following branches: four dorsal branches, two anterior (DBa and DBa′) and two posterior (DBp and DBp′); two lateral branches, one anterior (LBa) and one posterior (LBp) that fuse with the lateral branches of adjacent tracheal segments to form the lateral trunk; and three ventral branches, the visceral branch (ViB) that oxygenates the gut, and the ventral (VB) and ganglionic branch (GB) that migrate towards the central nervous system (Fig. 1G).

By means of the PH3 mitotic marker, we detected cell division in the tracheal cells from their determination in the ectoderm (25 h AEL) to the formation of primary branches (50 h AEL) (Fig. 2B-E). Upon these divisions, each tracheal cluster contains ∼80 cells, except for Tr1 and Tr2, which contain 140 cells. The extra number of cells in these two clusters is responsible for the formation of the additional cephalic and leg branches on the thoracic segments (Fig. 1F,F′). Tracheal cells are typically distributed among the different branches with a characteristic number of cells (Fig. 2F).

Two main features distinguish the trachea of Tribolium and Drosophila larvae (Fig. 3). First, the lateral spiracles, which in Tribolium connect the tracheal branches at both sides of each segment to the exterior, are not present in Drosophila larvae (Fig. 3A,B). This observation is consistent with the long-standing notion that ‘from a relative primitive, open form of tracheal system of ancient insects originated phylogenetically, the relative secondary, closed form of the present time larvae’ (Keilin, 1944; Palmén, 1877). Instead, Drosophila has two posterior spiracles in the last segment, which open to the exterior through highly specialised structures, namely the stigmatophores (Fig. 3A). During subsequent moults, Drosophila second and third instar larvae also develop anterior spiracles, similar openings at their anterior-most region that become functional at the pupa stage (Manning and Krasnow, 1993). In place of the lateral spiracles, Drosophila larvae have spiracular branches, which are groups of tracheal cells connected to the inner side of the ectoderm but with no opening to the exterior (Manning and Krasnow, 1993). These branches are considered a rudiment of an ancestral tracheal system with segmental spiracles (Palmén, 1877). They consist of aggregates of progenitor adult tracheal cells (Pitsouli and Perrimon, 2010; Weaver and Krasnow, 2008) and do not participate in larval breathing.

The other main feature that distinguishes the trachea of Drosophila and Tribolium larvae is the occurrence of dorsal trunks in the former. These trunks are a pair of prominent branches that run longitudinally along each side of the larva (Fig. 3C,E). In contrast to the other branches, cells in the dorsal trunks do not intercalate and therefore more than one cell contributes to the circumference of the trunks, thus accounting for their greater diameter. Dorsal trunks are not present in Tribolium (Fig. 3D,F) and all tracheal cells, except those connected to the lateral spiracle, undergo intercalation and form branches of a similar diameter in this insect (Fig. 1F″).

To address what might account for the lack of functional lateral spiracles in Drosophila, we compared the patterns of gene expression in the spiracular branches and in the posterior spiracles. In Drosophila, the gene encoding the Ct homeodomain transcription factor plays an important role in the formation of both structures – the spiracular branches and the posterior spiracles (Hu and Castelli-Gair, 1999; Pitsouli and Perrimon, 2010). At the posterior spiracles, ct is expressed in the ectodermal cells that will give rise to the spiracular chamber, the spiracular structure that connects the tracheal cells (trh-expressing cells) to the exterior (Hu and Castelli-Gair, 1999). However, a close analysis shows that ct expression extends beyond the trh expression domain (Fig. 4A,A″). In contrast, the domain of ct expression coincides with that of tracheal cells in the lateral spiracular branches (Fig. 4A,A′). This finding raises the possibility that an epidermal population of ct-expressing cells next to the spiracular branches is required to build a functional spiracle.

Consistent with this hypothesis, we found Tc-ct expression at the sites of the lateral spiracles in Tribolium, in the tracheal cells (Tc-trh-expressing cells), and also in some neighbouring ectodermal cells (Fig. 4B,B″). Remarkably, this specific Tc-ct expression is present in all Tribolium segments but is very much reduced in the third thoracic segment (Fig. 4B,B′) – the only segment in which the lateral spiracles close as development proceeds (Bennett et al., 1999; Lewis et al., 2000). This finding thus reinforces the link between ectodermal Tc-ct expression and functional spiracles. To assess the functional role of Tc-ct in the process, we knocked down its expression by injection of dsRNA (see Materials and Methods for details) and found that the resulting Tribolium larvae lack openings in the position of the lateral spiracles (Fig. 4I-J′).

We turned to Drosophila to further explore the role of ct in the development of spiracles. ct mutant embryos have abnormal posterior spiracles that lack an open connection (Hu and Castelli-Gair, 1999). Forced expression of ct at the lateral ectoderm, either by broadly expressed drivers such as nullo-Gal4 or by the ectodermal 69B-Gal4 (see Materials and Methods for details), promoted the development of openings at the lateral ectoderm (Fig. S2). However, these embryos were greatly distorted. We thus used grn-Gal4, a construct that drives more restricted expression in tracheal cells and in patches of the lateral ectoderm (Fig. S3). Using this driver, ectopic expression of ct also induced openings at the lateral ectoderm (Fig. 4C-F), which were connected to the internal tracheal cavities (Fig. 4G-H′), supporting the notion that these openings are similar to lateral spiracles (Fig. S4). Thus, ct expression in the ectoderm is necessary for the formation of functional lateral spiracles in Tribolium and both necessary and sufficient to induce the formation of corresponding structures in Drosophila.

Next, we addressed what might account for the development of the prominent dorsal trunks in Drosophila and their absence in Tribolium. In this regard, the morphology of the Tribolium trachea is reminiscent of that observed in Drosophila embryos that lack the two sal genes (Franch-Marro and Casanova, 2002). Indeed, sal acts as a repressor of cell intercalation in Drosophila (Ribeiro et al., 2004) and Drosophila sal mutants do not develop dorsal trunks (Kühnlein and Schuh, 1996), thus supporting the notion that sal expression is associated with the acquisition of this morphological innovation. To address this possibility, we compared sal expression in the two insects. In stage 10 Drosophila embryos, sal is expressed in broad patches of dorsal ectoderm that encompass the dorsal region of each tracheal placode (Kühnlein and Schuh, 1996). As development proceeds, sal expression declines in the ectodermal region between the placodes and becomes restricted to the dorsal region of the tracheal placodes (Kühnlein and Schuh, 1996; Fig. 5A,A‴). As the trachea develops further, sal expression persists in the dorsal branches and in the dorsal trunks and finally accumulates only in the latter (Fig. 5C,C′), where it represses cell intercalation (Kühnlein and Schuh, 1996; Ribeiro et al., 2004). Similar to its expression in Drosophila, Tc-sal is expressed in a broad strip of dorsal ectoderm in Tribolium (Fig. 5B). However, and in contrast to Drosophila, the ectodermal expression of Tc-sal does not include tracheal cells (Fig. 5B-B‴) and Tc-sal is not expressed during the formation of Tribolium tracheal branches. Consistently, we did not observe any evident tracheal phenotype in Tc-sal-depleted Tribolium embryos (data not shown). Nevertheless, we detected Tc-sal expression in the Tribolium tracheal system much later, at 60 h AEL, once all the branches have already formed (Fig. S5).

As mentioned above, both the development of dorsal trunks and the closure of lateral spiracles are adaptations to the breathing physiology of Drosophila larvae. Indeed, it has been reported that dorsal trunks are over-developed in certain insect larvae with few functional spiracles (Keilin, 1944). In many insect taxa, this feature was linked to the aquatic life history of larvae. As early as 1877, Palmén observed an association between the appearance of longitudinal trunks and the loss of segmental spiracles. In this regard, it is interesting to note that ectopic ct expression in Drosophila larvae not only induced the development of lateral openings but was also associated with a failure in the development of dorsal trunks when ectopic ct expression encompassed the tracheal cells (Fig. 5D,D′,F and Fig. S6). This result suggests a link between the lack of functional spiracles and the development of thick tracheal trunks. Furthermore, we observed that ectopic tracheal expression of ct was associated with the downregulation of endogenous sal, which accumulated very weakly or was absent in the tracheal cells that expressed ct (Fig. 5D,D′). A similar repressive effect of ct on sal expression has also been reported in the Drosophila posterior spiracles (Hu and Castelli-Gair, 1999). These results suggest that ectopic tracheal expression of ct interferes with the development of dorsal trunks by downregulating sal expression. However, other outcomes of ectopic ct expression, such as its role in the regulation of the pro-apoptotic gene reaper (Zhai et al., 2012), are probably also responsible for the deleterious effect on dorsal trunk development, as we were unable to revert the defects upon forced expression of ct by co-expression of sal (Fig. S7).

Composite patterns of gene expression are often associated with distinct regions in their promoters that are specific for their regulation in given domains. Such is the case for the Drosophila sal gene, which has a trachea-specific enhancer. Repression of sal by ct in the Drosophila trachea is mediated through its tracheal enhancer (Fig. 5E-H), thus suggesting a mechanism for the acquisition of tissue-specific gene interactions and consistent with the notion of enhancers as an opportunity for the independent evolution of gene regulation (e.g. Averof and Akam, 1995). To further assess the hypothesis that Drosophila might have acquired a distinct enhancer to drive sal expression specifically in tracheal cells, we generated a Tribolium transgenic line bearing a reporter gene under the control of the Drosophila sal promoter region (see Materials and Methods for details and Fig. S8). Unfortunately, and probably because of the phylogenetic distance between the two species, this construct gave rise to a completely unrelated expression pattern, thus impeding us from evaluating this hypothesis further.

One of the main challenges in evolutionary genetics is to unveil the molecular basis of key phenotypic innovations that ultimately permit the occupation of alternative ecological niches (e.g. Kratochwil and Meyer, 2015). Our results establish a functional correlation between changes in the domains of ct and sal expression and the structure of the larval tracheal system. These changes are likely to be associated with the exploitation of new habitats by the fruit fly larvae. In addition, it has been a long-standing problem to consider the gradual development of such innovations because of the difficulty explaining the selective value of all hypothetical intermediate states. In this regard, it is worth noting that changes in ct expression could be sufficient to account for the differences in the number of lateral spiracles. Moreover, the issue of morphological innovation is even more complex in the case of independent innovations that are nevertheless functionally related. Our results showing crosstalk between ct and sal provide a link between two functionally related innovations, namely the elimination of lateral spiracles and the development of dorsal trunks. All together, these results provide an evolutionary scenario by which a single genetic event contributes to the joint evolution of functionally inter-related organs.

The San Bernardino wild-type strain of Tribolium castaneum was used for dsRNA injections and inmunostaining and was provided by Yoshinori Tomoyasu (Miami University). The Vermilion White strain was used for generating the transgenic line enhancer Dsal (pBac[3×P3-EGFP_Tc‘hsp5’-Dsal enhancer_Gal4]) and was provided by Gregor Bucher (University of Göttingen). All Stocks were reared under constant temperature of 29°C and 60% humidity on wholewheat flour with 5% inactivated yeast.

The following Drosophila lines were used: btl-Gal4 (Shiga et al., 2014), nullo-Gal4 (Kunwar et al., 2003), 69B-Gal4 (Brand and Perrimon, 1993), sal-TSE-lacZ (Kühnlein and Schuh, 1996), grn-Gal4 (Garces and Thor, 2006), UAS-ct (Ludlow et al., 1996) and UAS-sal (Kühnlein and Schuh, 1996). Crosses with GAL4 lines were reared at either 25°C or 29°C.

Tc-ct dsRNA (IB_06353) was synthesised by the Eupheria Biotech Company. A concentration of 1 µg/µl dsRNA was injected into 40 female adults in the abdominal body cavity laterally to avoid damaging genitals according to Tomoyasu and Denell (2004). Injected females were crossed with wild-type males in order to obtain knockdown embryos.

Tribolium embryos were fixed according to Bucher (http://wwwuser.gwdg.de/~gbucher1/tribolium-castaneum-beetle-book1.pdf). For antibody staining, samples were incubated overnight in a primary cocktail in PBX (PBS plus 0.5% Triton-20X and 0.3% BSA) at 4°C, washed three times for 20 min in PBX, incubated for 1 h at room temperature with the appropriate fluorophore-conjugated secondary antibodies, washed three times for 20 min in PBX, and mounted on Vectashield with DAPI (Vector Laboratories) for microscopic analysis. The following primary antibodies were used: rat DE-cadherin (1:20, DSHB, DCAD2), mouse anti-Cut (1:200, DSHB, 2B10), Crumbs (1:10, DSHB, Cq4), α-Spectrin (1:5, DSHB, 3A9) and mAb2A12 (1:1, DSHB); rabbit anti-Spalt (1:200) provided by Rosa Barrio (Barrio et al., 1999) and anti-PKC (1:100, Santa Fe Technologies, H7 sc-8393); rat anti-Trachealess (1:300, generated in the lab.) (Lebreton and Casanova, 2014). Secondary antibodies labelled with Alexa Fluor 488, 555 or 683 were obtained from Molecular Probes and with Cy5 from Jackson ImmunoResearch.

Drosophila staining was performed according to standard protocols using the same antibodies and concentrations as in Tribolium. Micrographs were acquired with a Leica SP5 confocal microscope and images were processed with Fiji and Adobe Photoshop CS4.

First instar larvae cuticles were transferred to a drop of Hoyer's-lactic acid 1:1 on a slide and incubated overnight at 60°C.

We thank R. Barrio, J. Castelli-Gair Hombría, Y. Tomoyasu, and G. Bucher for beetle and fly stocks, reagents and technical support, and M. Averof, C. Desplan, C. Pitsouli and colleagues in the lab for discussions and reading the manuscript. We thank S. Chafino for technical support on rearing Tribolium colonies.