The retinoic-like juvenile hormone controls the looping of left-right asymmetric organs in Drosophila

The establishment of the two major axes, anteroposterior (AP) and dorsoventral (DV), is central to the patterning of the body plan. This orthogonal, two-coordinate system results in an apparent bilateral symmetry of the body shape in many organisms. In addition to the AP and DV axes, animals have developed a third axis allowing differentiation of the left and right body sides. Left-right (LR) asymmetry arises during development of internal organs, leading to side-specific organ morphology (e.g. lung, brain) or location (e.g. heart, spleen, liver) (for a review, see Burdine and Schier, 2000; Capdevila et al., 2000; Hamada et al., 2002; Mercola and Levin, 2001; Wright, 2001). Another important consequence of LR asymmetry is the directional looping of organs along the AP axis. For example, in normal humans (situs solitus) the heart undergoes a dextral looping leading to its pointing toward the left side. The heart is the organ most sensitive to defects in LR asymmetry and organ looping, which can lead to severe and frequent life-threatening abnormalities(1% of liveborns show heart abnormalities)(Harvey, 1998; Kathiriya and Srivastava,2000). Thus, it is important to develop models in order to analyze how LR asymmetry and looping morphogenesis are established, and how these two processes interact during embryogenesis.

Given the harmful consequences of heterotaxia(Kosaki and Casey, 1998), it is crucial for animals to genetically control LR asymmetry in order to avoid organ randomization. A breakthrough in the molecular analysis of LR asymmetry was the discovery of side-specific expression of nodal, a BMP family molecule, in chick embryos (Levin et al.,1995). The study of nodal homologs in other vertebrate embryos (Collignon et al.,1996; Lowe et al.,1996) indicates that nodal is a central and evolutionarily conserved regulator of LR asymmetry(Burdine and Schier, 2000; Capdevila et al., 2000; Mercola and Levin, 2001; Wright, 2001). In addition to nodal, retinoic acid (RA), a terpenoid compound, has been shown to participate in LR asymmetry in all vertebrates examined so far(Niederreither et al., 2001; Tsukui et al., 1999; Wasiak and Lohnes, 1999). Indeed, either an excess or a reduction of RA activity can lead to LR asymmetry defects (Tsukui et al.,1999). Interestingly, RAinduced LR defects are associated with abnormal expression of the asymmetric markers nodal or pitx2, indicating that RA interacts with the nodal pathway and is an important conserved component of the LR program in vertebrates.

Although there is considerable information about how LR asymmetry is established, very little is known about the mechanisms and molecules controlling looping morphogenesis of LR asymmetric organs. It is important to note that in the absence of LR determination, organs still loop but in a random direction. This indicates that looping morphogenesis lies downstream of the LR pathway and produces a stereotyped directional looping based on interpretation of positional information. One clear candidate in the looping process in vertebrates is RA. Indeed, depending on the species, mode of administration and stage, exogenous RA can induce partial heart looping in mouse and Xenopus (Chazaud et al.,1999; Drysdale et al.,1997; Iulianella and Lohnes,2002; Niederreither et al.,2001; Tsukui et al.,1999; Wasiak and Lohnes,1999; Zile et al.,2000). Furthermore, loss of the Raldh2 gene, which is essential to convert vitamin A into active RA, leads to an incomplete looping of the heart in mouse (Niederreither et al., 2001). These data thus indicate that RA has a dual role,being required both in the establishment of LR asymmetry upstream of nodal and in the downstream morphogenetic control of looping per se. Whether the mechanisms controlling and coupling LR asymmetry and organ looping are conserved is unknown.

In contrast to the prominent sidedness found in vertebrates, LR asymmetry in invertebrates is much less pronounced(Hobert et al., 2002). For example, the Drosophila heart is a symmetrical structure lying at the dorsal midline and running along the AP axis. There are a few stereotyped LR markers in the fly gut, but no mutations so far have been isolated that specifically affect any of these (Hayashi and Murakami, 2001; Lengyel and Iwaki, 2002; Ligoxygakis et al., 2001). In order to establish a genetic model of LR asymmetry and organ looping in Drosophila, we have screened for mutations affecting the asymmetric looping of the spermiduct and genitalia in adult flies (Fig. 1A,B). We use this process as a model to dissect the genetic pathways involved in the looping of LR asymmetric organs.

We have identified a looping-specific mutation in Drosophila,which we call spin. In the spin mutant, direction is normal but looping is incomplete, indicating that this gene controls looping morphogenesis downstream of LR determination. Molecular cloning shows that spin is a novel Fas2 allele; in spin mutants a specific subset of neurosecretory cells innervating the corpora allata is defective. The corpus allatum is a specialized structure of the ring gland,the major endocrine gland in Drosophila, that is involved in the secretion of juvenile hormone (JH), one of the main insect growth regulators. Our pharmacological and genetic data indicate that spin males have an abnormal level of JH during the pupal stage, and that this is responsible for genitalia and spermiduct rotation defects. As the sesquiterpenoid JH is functionally related to the terpenoid RA(Harmon et al., 1995), which plays an important role in organ looping and LR asymmetry in vertebrates,these results indicate that an evolutionary conserved terpenoid pathway controls asymmetric organ morphogenesis in animals.

The original spin mutation was generated in a P-element mutagenesis screen on the X chromosome using a P{GawB} transposon(Bourbon et al., 2002) (S.N.,unpublished). Cloning of spin was carried out by plasmid rescue and sequencing of the insertion site was done using an automated ABI DNA sequencer. The P-element in spin was remobilized using an external source of P transposase in y w spin^{P{Gal4; w+}}/Y; D2-3 transposase/+ males; excisions were selected through the loss of the w⁺ eye marker and balanced using classical approaches and chromosomes (see FlyBase at http://flybase.bio.indiana.edu/).

Fas2^e86, Fas2^EB112 and UAS-Fas2 were kind gifts of C. Goodman(Grenningloh et al., 1991). Fas2^rd1 was a gift of R. Davis(Cheng et al., 2001). UAS-synaptobrevin-GFP flies were provided by M. Ramaswami(Estes et al., 2000). Met flies were kindly provided by T. Wilson. Transgenic lines expressing GAL4 in different subsets of neurons (Aug21, Feb78, Feb170, Feb211,Kurs21) were kindly provided by T. Siegmund(Siegmund and Korge, 2001). All other stocks can be found at FlyBase(http://flybase.bio.indiana.edu/).

Short egg collections (2 hours) were raised on rich medium at 25°C and subjected to a 1 hour heat-shock at 37°C. Only one heat-shock per life cycle was applied. After HS treatment, flies were grown at 25°C until eclosion, and the extent of rotation rescue was scored in adult males. The genotype of the heat-shocked males is Fas2^spinR5/Y; UAS-Fas2/HS-GAL4.

Fas2-expressing clones, marked with GFP, were generated using the`flip-out' technique (Struhl and Basler,1993). Larvae with the genotype Fas2^spinR5/Y; act>y+>GAL4,UAS-EGFP/UAS-Fas2; hsFLP/+ were heat-shocked at 37°C for 1 hour.

Third instar larvae were dissected, inverted inside-out with forceps and antibody stained using standard protocols. After staining, the ring glands were dissected and mounted. The anti-Fas2 (Developmental Studies Hybridoma Bank) primary antibody was used at 1:10; Secondary antibodies used in this study are anti-mouse-FITC (1:400; Molecular Probes), anti-mouse CY3 (1:400;Molecular Probes), anti-rabbit-CY3 or FITC (1:400; Molecular Probes). Phalloidin-TRITC was used at 1:2000 (Molecular Probes). Confocal images were taken using Leica TCS-NT or TCS-SP1 confocal microscopes. Images were processed using Photoshop 6.0 (Adobe).

For scanning electron microscopy (SEM) imaging, adult flies were critical-point dried and coated with 25 nm gold using standard methods.

A JH analog, pyriproxyfen{2-[1-methyl-2-(4-phenoxyphenoxy)-ethoxy]-pyridine} stock solution (10 mg/l in acetonitrile from the laboratory of Dr Ehrenstorfer-Schafers, Germany) was diluted with acetone. White prepupae were collected and 0.25 μl of acetone containing the desired amount of pyriproxyfen was applied to each pupa on the dorsal side, as described by Riddiford and Ashburner(Riddiford and Ashburner,1991).

The rotation of genitalia takes place during metamorphosis; at this stage,the distal part of the male reproductive apparatus, the genital plate,undergoes a stereotyped 360° clockwise rotation, inducing the spermiduct to loop around the gut in a clockwise direction (see Fig. 1A,B)(Gleichauf, 1936). The rotation of genitalia can thus be compared with the oriented (dextral or sinistral) looping of internal organs in vertebrates, such as gut and heart. These indeed represent specific LR asymmetry markers, as mutations that affect organ positioning also perturb the direction of organ looping (see Introduction). Interestingly, mutations have been previously isolated that resulted in counterclockwise rotation of genitalia both in Drosophila melanogaster and Musca domestica(Milani, 1955; Milani, 1956).

The reversible nature of looping genitalia indicates that this structure represents a suitable LR marker that can be used as a novel genetic model to study both LR asymmetry and looping morphogenesis in Drosophila.

To initiate a genetic characterization of LR asymmetry and organ looping in flies, we screened for viable mutations showing defective genitalia rotation(G.Á. and S.N., unpublished). We focus on a novel viable P-element mutation, spin, for which all the adult males show a characteristic mis-rotation of genitalia and are sterile. In spin males, the extent of rotation varies from ∼30° to 320°, with a large proportion of males (84%; n=195) having their genital plate in a position corresponding to rotation of 135-225°. Fig. 1 shows some characteristic examples of spin genital plates. Because external mis-rotation does not allow discrimination between under-, hyper- or counter-rotation of the genitalia, males of different phenotypes were dissected and the looping of their spermiduct analyzed. All dissected males showed a clear under-rotation phenotype(Fig. 1G,H), indicating that spin is required for the genital plate and spermiduct to undergo complete looping, but has no role in directionality. Dissection of several hundred wild-type males did not reveal any rotation defect (data not shown),indicating the robustness of this process in normal males.

After remobilization of the P-element insert in spin (see Materials and Methods), three different w^–populations were found. In the major class (51/73; 70%), flies reverted to a fully wild-type phenotype. The two other classes included lethal alleles(18/73; 25%) and viable alleles (4/73; 5%), which retained their original spin-like phenotype. Altogether, these results indicate that the P{GAL4} transposon present in spin is responsible for the looping phenotypes and is inserted in or close to an essential gene.

The P-element in spin is inserted in the 5′ UTR of the fasciclin 2 (Fas2) gene(Fig. 2A; see Materials and Methods) (Goodman et al.,1997; Grenningloh et al.,1991), suggesting that spin is a novel Fas2allele (hereafter referred to as Fas2^spin). This conclusion is supported by the following points. First, the expression of the Fas2 protein in the original Fas2^spin allele and in a lethal revertant (Fas2^spinRM1) is strongly reduced or absent in embryos, respectively (Fig. 2B-E). Significantly, in Fas2^spin third instar larvae, the expression of the Fas2 protein in eye imaginal discs or in whole brain extracts is also strongly reduced(Fig. 2F,G; data not shown). Second, we show that expression of a UAS-Fas2 transgene under the control of the original spin^P{GAL4} line can fully rescue the rotation and sterility phenotypes (Fig. 1C).

Because genitalia rotation defects have not been described for any previously characterized Fas2 allele, we wondered whether other viable Fas2 alleles would show rotation defects. Interestingly, we found that Fas2^e86(Grenningloh et al., 1991) and Fas2^rd1 (Cheng et al.,2001) alleles displayed ∼5% and ∼20% genitalia rotation defects, respectively (data not shown).

Taken together, these results indicate that spin is a novel viable, looping-specific Fas2 allele. Fas2^spin is the first fully penetrant defective looping mutant to be described, and thus represents an important tool to study organ looping in Drosophila.

Rotation of genitalia was reported by Gleichauf(Gleichauf, 1936) to take place in 2- to 3-day-old pupae over a period of 24 hours. To establish the temporal requirement for Fas2 function in genitalia rotation, we used a heat-inducible GAL4 line to express Fas2 under the UAS promoter. In this experiment, we rescued the Fas2^spinR5 allele, a viable Fas2^spin excision allele retaining the rotation phenotypes but lacking the GAL4 activity associated with the original Fas2^spin allele (data not shown). Short egg collections were subjected to a single one hour heat-shock (HS) at 37°C. Adult males were then analyzed and the extent of genitalia rotation rescue determined. As shown in Fig. 2H, a single 1 hour HS at day 7 of development is sufficient to rescue genitalia rotation. Indeed, flies that received a HS at day 7 of development showed a very high degree of rescue (up to 90%), while HS applied either before or after this period had little or no effect (Fig. 2H). Control males which have not been heat-shocked show no rescue(data not shown). These results indicate that Fas2 is required during a limited period of time during pupal development for rotation to take place normally. The timing of the rescue is consistent with the previously described period of rotation (Gleichauf,1936).

In order to identify the tissue(s) and cells that require Fas2function for genitalia rotation, we used the GAL4-UAS system(Brand and Perrimon, 1993) to drive tissue-specific expression of a UAS-Fas2 transgene in Fas2^spinR5 males. Because Fas2 is required in many aspects of neuronal development(Brunner and O'Kane, 1997; Goodman et al., 1997), and is expressed mostly in neurons, we first asked whether Fas2 function was required in the nervous system for rotation. Surprisingly, we found that the elav-GAL4 line, which drives expression specifically in the CNS during development, is able to rescue Fas2^spinR5 rotation defects fully (Fig. 4A). This result prompted us to examine in detail the expression pattern of Fas2 protein in the brain and to look for potential nervous system phenotypes in Fas2^spin. Our analysis uncovers a previously unknown function of Fas2 in the ring gland (RG). The RG is a composite neuroendocrine organ made of three different specialized regions: the prothoracic gland (PT), the corpora cardiaca (CC) and the corpora allata (CA; Fig. 3B). The CC probably plays a role in the regulation of blood sugar levels in larvae through adipokinetic hormones (Noyes et al., 1995; Rulifson et al., 2002). The PT and CA are specialized cells responsible for the secretion of the two primary insect hormones ecdysone and juvenile hormone (JH), respectively(Riddiford, 1993). Interestingly, Fas2 expression is restricted to the CC and to specific axonal processes innervating the CA (Fig. 3A,C). These neurosecretory cells (nCA; see Fig. 3A) control JH level and can be easily identified using a specific GAL4 line expressed in all CA neurons (Kurs21-GAL4) (Siegmund and Korge,2001). As Kurs21-GAL4 driven GFP expression and the anti-Fas2 staining overlap precisely, we conclude that Fas2 is expressed in all nCA terminals (Fig. 3A). In Fas2^spin, the overall morphology of the CA synapse is abnormal (Fig. 3C,D), showing fused terminal boutons and a reduced number of presynaptic nerve terminals(compare Fig. 3E,F). This result is consistent with the previous finding that the bouton number is reduced in the neuromuscular junction in strong hypomorph Fas2 flies(Stewart et al., 1996).

Both the expression pattern of Fas2 in the nCA and the abnormal synapses in Fas2^spin mutants suggest that Fas2 may be required for JH synthesis, which in turn may be important for appropriate genitalia and spermiduct looping during metamorphosis.

In order to demonstrate a direct link between Fas2, the CA, and genitalia rotation, we used the UAS-GAL4 system to express Fas2 in specific subsets of neurons innervating the RG. In a recent study, Siegmund and Korge (Siegmund and Korge,2001) identified and mapped the few neurons innervating the RG in Drosophila. The PT is innervated by two neurons from each brain hemisphere, whereas the CA is innervated by three neurons(Fig. 3A)(Siegmund and Korge, 2001). Importantly, neurons innervating the PT and CA are different and map to distinct regions of the brain. We used a collection of GAL4 lines expressed in different populations of neurons innervating the RG(Fig. 4)(Siegmund and Korge, 2001) to induce neuron-specific expression of Fas2 in Fas2^spinR5 mutants. When GAL4 is expressed strongly in the neurons innervating the CA (nCA) using the Kurs21-GAL4 line(Fig. 4A,B), the rotation of genitalia is completely rescued, just as observed with an elav-GAL4driver. Using another line in which GAL4 is only weakly expressed in CA neurons (Feb78), we found that the rotation, although only weakly and partially rescued, exceeds that seen in control Fas2^spinR5males. Using another nCA-GAL4 line (Feb170) in which GAL4 shows variegated expression (i.e. some larvae express GAL4 in nCA, while in others the expression is absent) (Siegmund and Korge, 2001), rescue is either complete or absent, respectively(Fig. 4A). By contrast, when GAL4 is expressed in the neurons innervating the prothoracic gland (nPT;Feb211; Fig. 4A,C) or in the CA itself (Aug21), no rescue was observed. These results show that Fas2is required in the nCA for normal genitalia rotation. In addition, they indicate that the rotation defects associated with Fas2^spinR5, and probably also with other viable Fas2 alleles, are linked to a defective neuroendocrine function leading to abnormal synthesis of JH during pupal development(Fig. 2H).

Our data support a model in which Fas2-expressing nCA neurons control JH titers which in turn remotely control the rotation of the genital plate (Fig. 7A). This model has two main predictions: first, if JH is a mediator of Fas2 function during rotation, then Fas2^spin should function cell non-autonomously; second, JH itself should have the potential to perturb rotation when its level is altered experimentally.

The cell non-autonomy of Fas2^spin is already manifest in the previous rescue experiments using GAL4 lines expressed in subsets of neurons in the brain (Fig. 4). If Fas2 is cell non-autonomous for rotation, then it is expected that mosaic animals, in which some cells are mutant while others are wild type, should be rescued. Interestingly, we found that almost all (367/368; 99.7%)transposase-induced mosaic males (Fas2^spin /Y; D2-3 transposase/+) have normal, fully rotated genitalia. Consistent with a complete rescue of the external genitalia rotation phenotype, these mosaic males showed normal looping and fertility (data not shown). Importantly, the comparison of the morphology of the CA synapse in wild type, Fas2^spin and mosaic males (Fas2^spin/Y;D2-3 transposase/+) indicates that all rescued mosaic males had restored a normal CA synapse (Fig. 5A-C).

In further support of the non-autonomy and nCA origin of spinphenotypes, we performed a different mosaic experiment in which expression of a UAS-Fas2 transgene is induced randomly using the `flip-out'technique (Struhl and Basler,1993). In this experiment, random clones of cells expressing a wild-type Fas2 transgene in a Fas2^spinR5background are generated and positively marked with GFP (see Materials and Methods). After clone induction, the developed adult males are sorted depending on their genitalia phenotypes. Only two different categories were found, corresponding to males that either are completely rescued or not rescued at all. Males from each class were dissected and analyzed for GFP(hence Fas2) expression in the RG. During metamorphosis, the RG migrates posteriorly toward the proventriculus and the prothoracic gland degenerates (Dai and Gilbert,1991). In adults, the RG is thus made only of CA and CC cells. We found that none of the males (n=5) in which genitalia is not rescued show any GFP expression in the adult RG, while all (n=7) dissected males showing full rescue of genitalia rotation had GFP expressed in the RG(Fig. 5D,E). Altogether, our mosaic analysis indicate a non-autonomous function of Fas2 during organ looping.

The second prediction of our model is that JH, which is proposed to control rotation, should induce looping defects when its level is modified during pupal development. Interestingly, it has been shown that the JH analogs methoprene and pyriproxyfen produce rotation defects at low doses, after topical application to white pre-pupae(Riddiford and Ashburner,1991). Higher doses of these JH analogs induce abdominal defects and lethality. Because the extent and direction of rotation in the previous study were not determined, we applied varying doses of pyriproxyfen to white pre-pupae and monitored the effects on genitalia rotation. Application of half-lethal doses of pyriproxyfen (0.25 pM/pupae) to wild-type pupae induces rotation defects that are very similar to Fas2^spinphenotypes. Indeed, dissection of the posterior abdomen of unhatched adults(pharate adults) indicates that pyriproxyfen-treated males have an under-rotated phenotype, with some males showing a complete absence of rotation (Fig. 6A). At higher doses (0.4 pM/pupae), the treatment is lethal; as animals die as pharate adults, their morphology can be analyzed. Dissection reveals a shift toward a no rotation phenotype, with a larger proportion of males having their genital plate in its initial position. Thus, increasing the level of a JH analog produces looping defects ranging from partial to a complete loss of circumrotation.

Because pyriproxyfen mimics spin phenotypes in a dose-dependent manner, we conclude from these experiments that Fas2^spinpupae may have an elevated level of JH. To support this conclusion further, we modified the dosage of the Methoprene-tolerant gene (Met),which encodes a (bHLH)-PAS family of transcriptional regulators(Ashok et al., 1998). Flies that are mutant for Met are more resistant to JH analogs, indicating that the Met gene participates in JH signal transduction. Conversely,elevated expression of the Met⁺ gene results in flies with higher susceptibility to JH analogs (Ashok et al., 1998). If JH is elevated in Fas2^spin,then increasing the dose of the Met gene should enhance Fas2^spin genitalia rotation defects, while mutations in Met should suppress them. We tested the effect of Met^K17, a P element-induced amorphic Metmutation, on Fas2^spin rotation defects. Strikingly, the loss of Met function completely rescues the spin rotation defects, indicating that spin and Met genetically interact and antagonize each other during this process(Fig. 6B). Conversely, the increase of the dose of the Met+ gene in a Fas2^spinR5 background results in lethality at the pharate adult stage, confirming the strong interaction between the two genes. Dissection of the genital plates reveals exacerbation of rotation defects,with ∼25% having a complete lack of rotation of genitalia(Fig. 6B). This phenotype is very similar to the phenotype of pupae treated with high doses of pyriproxyfen(Fig. 6A). Altogether, our results indicate that Fas2^spin males have an elevated level of JH causing genitalia rotation defects. This allatotropic (promotion of JH synthesis) phenotype of Fas2^spin is consistent with studies in several insects indicating a role of the CA nerves in negatively controlling levels of JH during metamorphosis.

In this study, we identified and characterized the first looping specific mutant in Drosophila, allowing uncoupling of the process of LR determination from that of looping morphogenesis. The molecular and genetic analysis of Fas2^spin reveals a link between organ looping and the retinoic-like juvenile hormone(Fig. 7A). Because of the chemical similarity between JH and RA (see below), these data provide a parallel between vertebrate and invertebrate asymmetric organ looping(Fig. 7B). Our study also indicates that Drosophila is a promising genetic model for the study of LR asymmetry and organ looping.

We show that asymmetric organ looping in Fas2^spin males is impaired and is due to an abnormal endocrine activity of the ring gland during the pupal stage. The effects on genitalia and spermiduct are mediated by an excess of JH, which then modifies looping morphogenesis downstream of LR determination (Fig. 7A). How do these results relate to vertebrate organ looping and LR asymmetry? The fact that JH affects looping morphogenesis in Drosophila suggests an important evolutionary conservation of the role of terpenoids in this process,downstream of LR determination. Like the retinoid hormones, JH is synthesized from the common isoprenoid precursor farnesyl diphosphate via the mevalonate biosynthetic pathway (Harmon et al.,1995). Furthermore, JH is a sesquiterpenoid that is chemically related to the vertebrate terpene group, represented by retinoic acid(Jones and Sharp, 1997). The common terpenoid nature of JH and RA has thus led to the proposal that these molecules may bind a common family of nuclear hormone receptors that might play similar functions in different organisms(Moore, 1990). In this respect, it is important to note that the JH analog methoprene, the topical application of which leads to genitalia rotation defects(Riddiford and Ashburner,1991), can specifically bind and activate the RXR receptor in mammalian cultured cells (Harmon et al.,1995). RA signal transduction in vertebrates requires the binding to and the activation of heterodimers composed of RAR and RXR nuclear receptors isoforms. Interestingly, the only insect homolog of vertebrate RXR is encoded by the Drosophila ultraspiracle (usp) gene, which has been shown to bind to JH in vitro(Jones and Sharp, 1997; Jones et al., 2001). Altogether, these data thus suggest that JH and RA have a related activity.

In addition to sharing common chemical features, both RA and JH, when present in excess, have strikingly similar effects on organ looping. In conditions of excess RA, a series of heart defects has been observed,including reversal of symmetry or incomplete looping. In Xenopus, the heart tube fails to loop after continuous exposure to low doses of RA(Drysdale et al., 1997), and incomplete looping of the heart is also observed in mice treated with RA over a long period (Chazaud et al.,1999). It is important to note that the effects of excess RA on heart looping are dose sensitive and stage specific, as are the effects of JH analogs (methoprene and pyriproxyfen) on the looping of genitalia in flies(Fig. 6A)(Riddiford and Ashburner,1991).

In addition to blocking organ looping, excess RA can also induce a reversal of LR asymmetry in several vertebrate models (see Introduction). Such a reversal of asymmetry has not been observed after topical application of pyriproxyfen in Drosophila (Fig. 6). This apparent discrepancy may be explained by species- and/or stage-specific responsiveness to excess terpenoids, as is found among vertebrates for RA. Another possibility is that JH in flies may have a function restricted to organ looping, not sharing the dual role of RA seen in vertebrates (Fig. 7B).

The chemical, and, as shown in this study, the functional and phenotypic similarities associated with JH and RA in flies and vertebrates, respectively,show that terpenoids play an evolutionarily conserved role in handed looping(Fig. 7B).

In Drosophila, genetic control of the establishment of the two major body axes has been well described. LR asymmetry has attracted little interest and thus remained an elusive process for several reasons. First,there are only few and mostly transient (i.e. present during embryonic stages only) LR organs (Hayashi and Murakami,2001; Lengyel and Iwaki,2002; Ligoxygakis et al.,2001), leading to the view that flies may not represent a good model to study LR axis like vertebrates. In the case of genitalia rotation,only one dedicated study has been published(Gleichauf, 1936), and, yet,this or another LR process have not been clearly validated as candidate LR markers. Second, to our knowledge no mutations have been isolated so far that showed a fully penetrant and rotation-specific defect. Though some studies have reported rotation defects in specific allelic combinations, the rotation phenotypes are poorly penetrant and are associated with other developmental defects (e.g. Abbott and Lengyel,1991; Holland et al.,1997; Santamaria and Randsholt, 1995; Yip et al.,1997).

In this study, we developed an approach to identify genes that are involved in two main processes underlying directional organ looping: the determination of directionality (LR asymmetry) and looping morphogenesis. Interestingly, our work reveals the existence of a zygotic control for genitalia LR asymmetry that is apparently distinct from the maternal control of LR development during embryogenesis (Hayashi and Murakami,2001; Lengyel and Iwaki,2002; Ligoxygakis et al.,2001). Additional work will be necessary to understand the mechanisms underlying asymmetric development in different tissues and developmental stages. Together with previously described mutations(Milani, 1955; Milani, 1956) (see Introduction) and our recent identification of a situs inversus mutation leading to a complete reversion of genitalia rotation (counterclockwise)(G.Á. and S.N., unpublished), these data suggest that just like in vertebrates, the earliest steps of symmetry breaking are accessible to genetic analysis in flies. (Mochizuki et al.,1998; Morgan et al.,1998; Yokoyama et al.,1993). Altogether, our results indicate that Drosophilahas all the basic elements to make it a genetic model to study organ looping in the context of LR asymmetry. In this respect, it is interesting to note that asymmetric organ looping, rather than asymmetric organ localization, is a more general and common LR marker among animals. For example in vertebrates,the first appearance of LR asymmetry is indicated by embryo turning and heart tube looping. Furthermore, organ looping can also be observed in plants through the helical growth of stalks and stems(Thitamadee et al., 2002).

We have shown a novel parallel between the programs underlying Drosophila and vertebrate asymmetric organ looping. Is this parallel more general? In order to address this question, future work will have to be focused on the identification of new genes involved in genitalia rotation in Drosophila, using genetic screens and reverse genetic approaches. One major goal of future work in Drosophila will be the identification of asymmetrically expressed genes and/or proteins. The fact that no such gene has been identified so far may be due to the lack of appropriate data on the developmental aspects of LR asymmetry in Drosophila. In this respect,our study now allows identification of the male genital disc(Sanchez and Guerrero, 2001)as a clear candidate tissue for looking at asymmetrically expressed molecular markers. The use of Drosophila and the comparative analysis of the LR asymmetry programs in vertebrates and invertebrates will help provide insights into the molecular mechanisms that underlie the question of symmetry breaking in animals.

We thank R. Davis, C. Goodman, M. Ramaswami, T. Siegmund, T. Wilson, the Bloomington Stock Center and DSHB for providing us with fly stocks and other materials; B. Baum for discussions; and C. Tabin, P. Léopold and members of our laboratories for critical reading of the manuscript. Work in S.N.'s laboratory is supported by grants from Centre National de la Recherche Scientifique (ATIPE-CNRS), Association pour la Recherche sur le Cancer (ARC),Fondation pour la Recherche Médicale (FRM) and Action Concertée Incitative (ACI), NATO and EMBO YIP (to S.N.). G. A. is supported by a grant from CNRS (« poste rouge »). N.P. is an investigator of the Howard Hughes Medical Institute.