|Home | About | Journals | Submit | Contact Us | Français|
It is well established that developmental programs act during embryogenesis to determine animal morphogenesis. How these developmental cues produce specific cell shape during morphogenesis, however, has remained elusive. We addressed this question by studying the morphological differentiation of the Drosophila epidermis, governed by a well-known circuit of regulators leading to a stereotyped pattern of smooth cells and cells forming actin-rich extensions (trichomes). It was shown that the transcription factor Shavenbaby plays a pivotal role in the formation of trichomes and underlies all examined cases of the evolutionary diversification of their pattern. To gain insight into the mechanisms of morphological differentiation, we sought to identify shavenbaby's downstream targets. We show here that Shavenbaby controls epidermal cell shape, through the transcriptional activation of different classes of cellular effectors, directly contributing to the organization of actin filaments, regulation of the extracellular matrix, and modification of the cuticle. Individual inactivation of shavenbaby's targets produces distinct trichome defects and only their simultaneous inactivation prevent trichome formation. Our data show that shavenbaby governs an evolutionarily conserved developmental module consisting of a set of genes collectively responsible for trichome formation, shedding new light on molecular mechanisms acting during morphogenesis and the way they can influence evolution of animal forms.
A general feature of development is the control of tissue and cell morphogenesis, a process whereby each cell acquires a specific shape depending upon its individual identity. Although the mechanisms that permit the patterning of a cellular field are now relatively well understood in different systems, how cell fate becomes translated into cell-shape control remains largely unknown.
Genetic analysis of epidermal differentiation in Drosophila has made this process a paradigm for our attempts to understand the subsequent steps of morphogenesis in animals. During early embryogenesis, cascades of transcription factors progressively refine the different cellular territories and define their respective identity . Signaling pathways then act to establish boundaries between adjacent cell fields (reviewed in [2,3]) and participate in the regulation of the size of these territories through the control of cell survival . Finally, when the epidermis is composed of a monolayer of post-mitotic cells, signaling pathways specify cell fates, ultimately determining morphological differentiation [1,2,5]. Epidermal cells in a given embryonic segment display a stereotyped pattern of two distinct morphological fates: cells with a smooth apical surface, which generate naked cuticle, and cells producing actin-rich cytoplasmic extensions, which eventually become the mature larval microtrichiae or trichomes, generally called denticles (ventral) and hairs (dorsal).
In the ventral region of the abdomen, each segment possesses six to seven rows of pigmented denticles, which are involved in locomotion. Activation of the Wingless (Wg) pathway determines the naked fate , whereas activation of the d-EGF-receptor (DER) pathway promotes denticle formation . In addition, Serrate-Notch signaling also contributes to define the denticle field [8,9] through a fine-tuning of the extent of DER activation . The activities of Wg and DER pathways on epidermal differentiation converge in the transcriptional regulation of shavenbaby (svb) . DER activates svb transcription in denticle-forming cells, whereas Wg acts as a repressor in the naked territories (Figure 1A), leading to the precise expression of svb in denticle cells. In turn, svb is required for denticle formation, since svb mutants display essentially naked cuticle. In addition, the ectopic expression of svb in smooth cells is sufficient to produce cuticular extensions , which nevertheless do not display the typical morphology of denticles, likely because those cells lack auxiliary factors involved in the fine shaping of these structures. svb, which encodes a zinc finger transcription factor [12–14], is the most-downstream regulator so far known to specify the denticle pattern .
Additional evidence supports the hypothesis that, whereas all other upstream regulators act to define epidermal cell fates, svb controls epidermal cell shape changes. Although different mechanisms specify morphological fates in the dorsal epidermis , svb is required for the formation of dorsal hairs as it is for denticles. It has been shown that the modification of svb, leading to the restriction of its expression in a subset of dorsal epidermal cells, is the unique cause of the modification of the trichome pattern in a sibling Drosophila melanogaster species, D. sechellia, in which several rows of dorsal hairs are replaced by naked cuticle . Somewhat surprisingly, given the large number of genes that are theoretically capable of modifying dorsal hair patterning , all cases of dorsal hair–pattern evolution examined in dipterans are the consequence of the modification of svb expression only [16–18]. Thus, developmental and evolutionary studies show that svb is critical for epidermal differentiation and that the presence/absence of svb expression ultimately determines the pattern of denticles and dorsal hairs .
How does svb control the formation of denticles and dorsal hairs during epidermal differentiation? Little is known about the cellular mechanisms underlying trichome formation. Epidermal cells are highly polarized along the apico-basal axis, with actin microfilaments accumulating at the cell cortex where they interact with apical junction complexes [20,21]. The formation of denticles and dorsal hairs is initiated by F-actin accumulation and reorganization, leading to the production of an apical bundle in the posterior region of the cell [14,22,23]. Each bundle then grows perpendicularly to the apical cell surface [22,23] and probably provides the mechanical force for the modification of the apical membrane. At later stages, ecdysteroid hormones induce epidermal cells to synthesize components of the cuticular envelopes, which are modified by the catecholamine biosynthetic pathway to ensure trichome pigmentation and hardening . Despite the large number of genetic screens based upon cuticle observation [25,26], the molecular mechanisms responsible for the different steps of epidermal differentiation still remain to be uncovered.
Since svb activity acts cell autonomously to promote denticle formation [11,14], we searched for genes whose expression is controlled by svb as an alternative approach to identify players responsible for the morphological differentiation of epidermal cells. We identified several cellular factors involved directly in subsequent steps of denticle formation. Although actin remodeling is generally regarded as the outcome of post-translational modifications, we show that denticle formation relies on the svb-dependent transcriptional activation of several factors involved in microfilament formation or bundling. We provide evidence that svb directly controls the transcription of miniature (m), which encodes a membrane protein containing a Zona Pellucida (ZP) domain, and we show that m is required for the membrane/cuticle interaction in the denticle. We also identified a svb target responsible for denticle pigmentation, therefore demonstrating that svb coordinates several aspects of denticle differentiation. Finally we show that svb regulates the same set of genes for the formation of dorsal hairs and suggest that the evolution of the dorsal hairs pattern in D. sechellia results from the concerted modification of the expression of cellular effectors regulated by svb.
As a first step towards identifying players involved in denticle formation, we searched for genes expressed in epidermal cells by conducting a systematic survey of expression patterns available from the Berkeley Drosophila Genome Project  and the literature (Table S1). Among the approximately 400 genes reported to be transcribed in the epidermis, we found that only a small number (<10%) are expressed in a segmentally repeated pattern, at a time of epidermal cells morphological differentiation. We further analyzed the epidermal expression of these genes by in situ hybridization and tested for its dependence on svb (Table S1). We first focused on m , because its expression in stage-15 embryos strikingly resembles the trichome pattern, which is specified by svb at previous stages (Figure 1B).
We found that m expression in epidermal cells is abolished in svb mutant embryos (Figure1B), showing that svb is required for m transcription in the epidermis. Detailed examination (Figure 1C) revealed a precise correlation between m expression and the denticle pattern in wild-type embryos, with a six to seven–cell-wide expression stripe in each segment that alternates with a band of m-negative cells. OvoA is a germinal isoform of the ovo/svb gene that acts as a transcriptional repressor , able to counteract svb activity when its expression is artificially directed in the epidermis [13,14]. When expressed in ptc-expressing cells (see Figure 1A), OvoA leads to the replacement of denticle rows 2 and 3 by a stripe of naked cuticle interrupting each denticle belt (Figure 1C). This OvoA expression also represses m transcription in the epidermal cells corresponding to denticle rows 2 and 3, providing evidence that svb activity is required cell autonomously for m expression in denticle cells. To test whether svb is sufficient to induce m expression in the epidermis, we examined the consequence of the ectopic expression of svb in smooth cells. Ectopic expression of svb in wg-expressing cells (Figure 1A) causes the formation of a supernumerary row of cuticular extensions and leads to an ectopic stripe of m expression (Figure 1C), showing that svb is sufficient to trigger m transcription in epidermal cells.
As the evolutionary loss of dorsal trichomes in D. sechellia embryos is the result of the restriction of svb expression to narrow stripes in dorsal epidermal cells , we asked whether m expression was modified accordingly in D. sechellia. When compared to D. melanogaster, m expression in D. sechellia embryos is indeed seen to be restricted to narrow bands in the dorsal region. These data thus bring independent support to the conclusion that m expression is controlled by svb in the embryonic epidermis and show that this regulation has been evolutionarily conserved.
All together, our results demonstrate that svb activity is necessary and sufficient to control m expression specifically in epidermal cells producing denticle and dorsal hairs.
To investigate whether svb directly regulates m expression, we first attempted to identify cis-regulatory elements directing m transcription in the embryonic epidermis.
We generated transgenic lines in which a reporter gene is placed under the control of a 6-kilobase (kb) genomic region encompassing the m transcriptional start site (6Kmin) (see Figure 2A). These 6Kmin reporter transgenes faithfully reproduce m expression in late embryos (Figure 2B and and2C)2C) and display a similar dependence on svb activity (unpublished data). Deletion of most intronic sequences to a 3-kb fragment leads to a strong diminution of the expression (Figure 2D), showing that cis-regulatory elements required for m epidermal expression are located in intronic regions. We focused on a 0.4-kb region from the second m intron, which is highly conserved in Drosophilidae (Figure 2A). This fragment, when placed upstream of a minimal promoter, promotes a robust expression in the epidermis (Figure 2E), similar to that of the endogenous m gene (Figure 2B–2H). Expression of 0.4Kmin lines is lost in a svb mutant background (Figure 2F), whereas an additional stripe of expression appears when svb is ectopically expressed in smooth cells (Figure 2G). Therefore, this 0.4-kb region of the m gene behaves as a svb-responsive element, able to drive epidermal expression under the control of svb.
We then assayed the ability of the Svb protein to bind in vitro to the corresponding DNA region. Gel retardation assays showed that the recombinant Svb protein binds specifically to the 0.4-kb m element through a single region, further refined to a 200-base pair (bp) fragment (Figure 2I). Sequence analysis revealed the presence of a nucleotide motif matching (7/7) the consensus binding site previously defined for Ovo proteins , which share the same DNA binding domain with Svb . Additional DNA-binding experiments showed that Svb binds to this predicted site with a high affinity (estimated Kd = 2.5 ± −0.8 nM, unpublished data). The sequence of this Svb binding site on the m enhancer (at position chrX:11487686, on the genomic scaffold) is conserved in all sequenced Drosophila species. Introduction of point mutations that substitute two invariant nucleotides found in all Ovo binding sites  eliminates Svb binding in vitro (Figure 2I), showing that Svb interacts specifically with the m enhancer through this single binding-site. To test the consequence of preventing Svb binding in vivo, we introduced these two mutations into the 0.4-kb enhancer construct and generated corresponding transgenic lines (0.4KminKO). All the tested 0.4KminKO lines (12) lost m-like epidermal expression, with only some variable residual staining, variable from line to line (Figure 2J).
These results demonstrate that Svb binds to an evolutionarily conserved element of the m gene, which behaves as a svb-responsive enhancer in the embryonic epidermis. Mutations preventing Shavenbaby binding to this element abolish its enhancer activity, thus providing strong evidence that Svb exerts a direct control on m transcription.
m encodes a membrane protein , which has recently been implicated in actin-based remodeling of wing cell shape during metamorphosis [28,31]. To examine whether m also plays a role in embryonic epidermal cells, we analyzed consequences of m inactivation on denticles.
Although the absence of m does not affect the denticle pattern, detailed examination shows that denticles are altered in m mutant embryos (Figure 3). In wild type, each denticle row has a characteristic morphology, but all denticles share a common organization resembling a rose thorn: a large base, followed by a median constriction forming a narrow neck, which supports the distal hook, which points either anteriorly or posteriorly (Figure 3). In larvae carrying the m1 loss-of-function mutation, or the Df(1)m-MR deficiency that deletes the entire m gene , denticles are small and misshapen (Figure 3 and Table S2); the median constriction is absent, leading to poorly differentiated denticles that display an aberrant triangular shape. These mutant defects are rescued when m is re-expressed using the wild-type 0.4Kmin element as a GAL4 driver (Figure 3A). These results provide additional evidence that this m enhancer is fully functional in the epidermis and demonstrate that the observed denticle defects are due to the loss of m activity.
To further investigate the role of m in denticle formation, we tested the influence of its ectopic expression in wild-type embryos. We used both an EP (target P-element) insertion located 2 kb upstream of the m start site, or the min cDNA placed under the control of UAS sequences (Gal4 binding sites), which both allow a strong m ectopic expression upon crossing with Gal4 drivers (unpublished data). None of the drivers we tested led to cuticular defects. For example, overexpression of m with the ptc-gal4 driver, both in denticle rows 2 and 3 and in a two-cell-wide stripe in naked cells (see Figure 1), does not affect cuticle organization (Figure 3A). Therefore, m expression is required, but not sufficient, for the correct formation of denticles.
These results show that m is directly involved in a specific aspect of denticle formation. All together, these data further demonstrate that svb directly impinges on epidermal cell morphogenesis through the direct control of m transcription.
The inactivation of m is not sufficient to mimic the svb phenotype, indicating that svb must regulate additional targets to achieve denticle formation. Mutations known so far to prevent denticle formation affect either svb or upstream regulators, therefore suggesting that inactivation of a given svb downstream target produces denticle malformation rather than prevents denticle formation. To test this hypothesis, we searched extensively for genes affecting denticle morphology (Table S1) and examined whether they are regulated by svb.
A dozen genes reported to affect denticle formation lead, when inactivated, to general cuticle defects, which are not, however, specific to trichomes (Table S1 and references therein). These genes (e.g., the Halloween family that encodes members of the ecdysteroid pathway) are either not expressed in the epidermis (e.g., are restricted to the ring gland) or display ubiquitous expression in epidermal cells (Table S1 and references therein).
Mutations corresponding to seven genes affect denticles, without altering either the general cuticle organization or embryonic patterning (Table S1). While the expression of fritz, hem, tricornered, and Stubble are ubiquitous or undetectable in the embryonic epidermis, forked (f), singed (sn), and shavenoid/kojak (sha) display a segmental pattern of expression (Table S1). Mutations in both sn and f lead to similar defects , with thin, crooked abnormal denticles, whereas sha mutants display a distinct phenotype [25,32] with small, misshapen, multiply split denticles (Figure 4A and Table S2). Detailed analysis of mRNA distribution revealed that sn, f, and sha are specifically expressed in denticle cells (Figure 4B). The epidermal expression of their transcript is strongly reduced in svb mutants (Figure 4B), revealing that svb is necessary for sn, f, and sha expression in denticle cells. In addition, svb is sufficient to promote sn and sha expression in the epidermis, since their mRNA is robustly expressed in additional stripes following ectopic expression of svb (Figure 4B). The svb-driven ectopic expression of f is barely detectable, indicating that additional factors are required to promote f expression in naked cells. Taken together, these data show that the transcription of genes previously implicated in denticle formation is specifically activated by svb.
sn, f, and sha encode actin-associated factors [32,33], and a recent study has shown that other components of the actin cytoskeleton (Diaphanous, Enabled, and the members of Arp2/3 complex) accumulate in forming denticles . This prompted us to analyze the embryonic expression of these genes, as well as additional actors in actin remodeling (chickadee, gelsolin, and twinstar). All exhibit ubiquitous expression in embryos, which is not influenced by svb activity (Table S1). However, we found that wasp (wsp), which encodes an activator of the Arp2/3 actin nucleation complex , is specifically expressed in denticle cells, in a svb-dependent manner (Figure 4B). The epidermal expression of wsp is strongly reduced in svb mutants and ectopic expression of svb leads to additional bands of wsp mRNA in naked cells, albeit at a much weaker level than in denticle cells (Figure 4B). Cuticle analysis revealed that zygotic wsp mutations  lead to the alteration of denticles, which are abnormally thin and often display bent extremities (Figure 4A). These defects are likely due to the lack of activation of the Arp2/3 complex, since the zygotic inactivation of two ubiquitous Arp2/3 subunits (ArcP34 and ArcP41) causes denticle alterations similar to those observed in wsp mutants (unpublished data).
Consistent with the idea that actin reorganization is required for denticle formation, our results show that Svb activates the expression of a subset of genes controlling actin dynamics, whose individual inactivation produces defects in denticle morphology.
Inactivation of each svb target we identified leads to specific defects in the morphology of denticles, but is not sufficient to inhibit their formation. This suggests either that these genes are involved in nonessential aspects of denticle formation (e.g., fine shaping) or, alternatively, that denticle formation relies on the collective activity of numerous effectors, none of them being absolutely indispensable. To discriminate between these possibilities, we analyzed embryos carrying combinations of mutations in different svb targets.
The phenotype of sn, f double mutants is slightly aggravated when compared to that of sn or f mutants (Table S2). When m mutation is combined with either sn or f, denticles display both the crooked extremities found in sn and f, and the defects characteristic of m. In addition, the inactivation of m and f (or sn) causes denticle splitting, as revealed by a significant increase in the number of abnormal denticles per row (Table S2). Simultaneous inactivation of all three genes leads to small and highly misshapen denticles that are more affected than in double mutants (Figure 5A and Table S2). Adding a fourth mutation, wsp or sha, further impairs the formation of denticles, leading in many cases to their replacement by naked regions (Figure 5A). The cumulated inactivation of svb targets thus progressively leads to tiny and highly malformed denticles, ending in the inhibition of denticle formation (Figure 5A). A similar effect is observed in the dorsal region, where combining mutations in svb targets ultimately prevents dorsal hair formation, in addition to producing cumulative morphological defects (Figure 5B). These results demonstrate that the morphological differentiation of the embryonic epidermis requires the concerted action of multiple genes, all playing a direct role in the formation both of denticles and dorsal hairs.
To better understand the role of svb targets, we analyzed the subcellular distribution of Singed, Forked, and Miniature proteins at the onset of denticle formation. Forked, Singed, and Miniature are sequentially accumulated in denticle cells (Figure S1) and, consistent with their bundling activity, both Singed and Forked co-localize with F-actin in growing denticles (Figure 6A). Forked accumulates in nascent denticles, from the earliest steps of F-actin reorganization, whereas Singed appears slightly later, when the apical bundle is already formed and is continuing to grow. Miniature appears concomitantly with Singed and is initially detected in the cytoplasm of denticle cells (Figure S1). In later embryos (stage 15 and 16), Miniature accumulates in a restricted apical domain that underlines the base of each denticle (Figure 6A). Since Miniature is a membrane protein, this observation suggests the existence of a denticle-specific membrane subdomain. We therefore analyzed mutant defects at the ultrastructural level. In stage 16 wild-type embryos, the apical surface of epidermal cells organizes a dense array of microvilli that contact the cuticle only at their distal tips (Figure 6B). In contrast, there is a continuous and tight interaction between plasma membrane and cuticle layers in the denticle region (Figure 6B). Although naked cells are unaffected, denticle cells do not contact cuticle correctly in m mutant embryos (Figure 6B). The apical membrane is disorganized, with an abnormal gap between the denticle membrane and cuticle layers, indicating that m is required for the membrane/cuticle interaction that is specific to denticle cells. These defects are not observed in sn, f mutants (Figure 6B), confirming that they are specific to the lack of m activity.
Therefore, these data reveal that denticle formation requires, besides actin remodeling, a specific modification of the membrane/cuticle interaction that is directly dependent on the localized activity of Miniature. All together, these results show that svb directs the expression of cellular effectors involved in different steps and aspects of cell shape control, which are collectively responsible for denticle formation.
In addition to the reorganization of cell shape, denticle formation implies pigmentation of cuticle extensions. We noticed that the supernumerary cuticular extensions formed after ectopic expression of svb are as darkly pigmented as normal denticles (Figure 7A), thus suggesting that svb also controls the expression of pigmentation genes in the ventral epidermis.
Cuticle pigmentation results from the activity of the catecholamine biosynthetic pathway, which transforms tyrosine into DOPA derivatives that are responsible for both sclerotization and coloration of the cuticle (reviewed in ). We examined the embryonic expression of ten genes putatively involved in pigmentation ; among them, four are expressed in the epidermis: pale (ple), dopadecarboxylase (Ddc), alpha-methotrexate deficient (amd), and yellow (y) (Figure 7B). While the expression of ple, Ddc, and amd are unaffected in svb mutants, we found that y is regulated by svb. Although y expression is strongly up-regulated in a subset of denticle cells in late wild-type embryos , this expression is undetectable in svb mutants (Figure 7C). Furthermore, ectopic expression of svb turns on y transcription in additional cell stripes (Figure 7C), showing that svb regulates y expression in epidermal cells.
Therefore, in addition to cytoskeletal organization, svb also controls cuticle pigmentation through the up-regulation of y expression in denticle cells.
Using denticle pattern as a readout of developmental cues, generations of fly geneticists have collectively gained a rare insight into the mechanisms acting to specify different cell populations. Quite paradoxically, how this pattern information is connected to cell shape remodeling for denticle formation has so far remained an unsolved question.
We show here that a single regulator, Shavenbaby, previously shown to integrate multiple patterning cascades, governs epidermal cell remodeling through the transcriptional control of several classes of effectors, acting directly in various cellular functions. This svb-regulated set of effectors constitutes a developmental module used in different tissues during development to produce cuticle extensions. Modification of svb expression has allowed the concerted evolution of this developmental module to produce morphological diversification during the evolution of insect species.
One of the first recognizable signs of the morphological differentiation of epidermal cells is the formation of an apical bundle of microfilaments in denticle cells [22,23]. These early steps of denticle formation depend on svb, which is necessary and sufficient to promote the formation of epidermal actin bundles . Our results show that svb controls the transcription of several genes involved in different steps of actin assembly/organization. First, we found that svb directs the expression of shavenoid/kojak, a gene producing strong denticle defects when mutated and recently shown to encode a protein reported to associate with actin , but whose biochemical function is unknown. Second, Svb also directs the expression of singed and forked, coding respectively for the Drosophila putative homologs of Fascin and Espin, two proteins that crosslink parallel actin filaments and promote the formation of bundles of microfilaments [36,37]. The Forked and Singed proteins sequentially accumulate in growing denticles, a situation reminiscent of that of wing hair formation , suggesting that these proteins play similar roles in the formation of adult and embryonic epidermal extensions. Accordingly, the inactivation of sn and f alters denticles, strongly suggesting that denticle formation indeed involves parallel actin bundles, as shown for wing hairs.
Several cytoskeletal regulators such as dAPC, Enabled, Diaphanous/Formin, and the Arp2/3 complex, accumulate in denticles [14,23], suggesting that they are involved in denticle formation, although their respective functions remains to be evaluated. Whereas svb does not control the expression of those ubiquitous actin-associated factors, it is possible that svb regulates their activity, or subcellular localization, indirectly. Consistent with this hypothesis, we have previously shown that dAPC-2 is specifically relocalized in svb-induced ectopic epidermal extensions . In addition, we show here that svb directs the epidermal expression of Wasp, a key activator of the Arp2/3 actin nucleator complex, which is well known to trigger the formation/elongation of actin filaments. Moreover, it has been shown in vitro that Fascin switches the activity of the Arp2/3 complex from the formation of a mesh-like branched network to parallel microfilaments , therefore suggesting that svb targets can regulate both the formation of actin filaments and their reorganization, at least in part, through a tight control of the activity of the Arp2/3 complex during denticle formation.
Taken together, these results show that svb controls the expression of several cytoskeletal factors which, probably by modifying the activity of housekeeping actin-remodeling machinery, act together to trigger the formation of apical cell extensions. Whereas few molecules are sufficient to promote actin organization in vitro, our studies indicate that, in vivo, many players are required to make a simple cellular extension. Pursuing the identification of novel genes regulated by svb should provide a means of identifying additional factors required for actin remodeling in vivo.
A surprising outcome of our studies is that denticle formation requires a specific regulation of the membrane/cuticle interaction. We show that svb directs the expression of m in trichome cells. Miniature is a single-pass membrane protein, with a short cytoplasmic tail and a large extracellular region that contains a conserved Zona Pellucida (ZP) domain . ZP domains were initially identified in the three major proteins of the zona pellucida, the extracellular envelope of mammalian oocytes (see  for a recent review), and they are thought to be components of apical matrices . We show that Miniature is required for the correct formation of denticles, revealing a novel aspect of ZP protein function in the formation of polarized cellular extensions. The absence of m severely impairs the interaction between the plasma membrane and cuticle layers in denticle cells, a defect likely due to a disorganization of the extracellular matrix. In the embryonic epidermis, Miniature is required for the continuous membrane/cuticle interaction that is specific to denticles, whereas only the tips of microvilli contact cuticle in naked regions . The accumulation of Miniature at the base of denticles reveals the existence of a denticle-specific membrane subdomain, suggesting that additional membrane proteins might be involved in denticle formation. We found that two other ZP genes are regulated by svb (Table S1) and analysis of their individual role in denticle formation is under way. Our findings shed light on the importance of membrane proteins and their interaction with extracellular matrices, an aspect of cell-shape control hardly accessible to cell-culture approaches. Future analysis of Miniature targeting to the denticle should help to understand the mechanisms required for localized cell-shape modification during morphogenesis.
svb also regulates the expression of y, a gene encoding an apically secreted protein that associates with cuticle and is required for the production of black pigments . While pigmentation per se is not related to cell morphogenesis, the role of Yellow in the catecholamine pathway remains elusive; it could be involved in denticle hardening, since y mutant larvae display defects, in the morphology of denticles, that have been proposed to account for their abnormal locomotor activity . In addition, svb could directly regulate the local protein composition of cuticle, as suggested by the identification of an additional target encoding a putative chitin-binding protein (Table S1).
Experimental evidence suggests that svb is situated at the bottom of regulatory cascades determining trichome patterning and is in turn directly responsible for triggering the cellular program of denticle formation (Figure 8). First, svb remains the most-downstream regulator determining the pattern of denticles and dorsal hairs, despite the unprecedented extent of genetic screens based upon cuticle observation (which identified most members of the Wg and DER pathways). Second, among mutations producing trichome defects, svb mutants display the strongest phenotype, in which most denticles and dorsal hairs are replaced by naked cuticle. Third, we show that svb directs the expression of genes involved in various aspects of denticle formation, including control of the cytoskeleton, membrane/matrix organization and cuticle differentiation. Finally, we provide evidence for a direct control of one of the targets (m). We have defined a 400-bp m enhancer reproducing the endogenous expression pattern of m in the epidermis and show that the Svb transcription factor binds specifically to this evolutionarily conserved region. Substituting 2-bp in this cis-regulatory element preventing Svb binding is sufficient to abrogate its in vivo enhancer ability, thus suggesting that direct binding of Svb mediates the control of m epidermal expression. Several putative svb binding sites have been detected in evolutionarily conserved regions of other svb targets. Whether they are all required for svb transcriptional regulation remains to be tested. Further dissection of the m enhancer, as well as those of other svb targets, should lead to the definition of a functional cis-regulatory element responsible for svb control. This outcome should facilitate the identification by bioinformatic approaches of additional svb responsive enhancers and target genes.
We propose that svb directly controls the expression of a set of “effector” genes, all required for a concerted modification of cell shape and cuticle organization to achieve the formation of denticles (Figure 8). How many genes are regulated by svb to promote remodeling of epidermal cells? Our analysis, which covers approximately 25% of the total number of Drosophila genes, has led to the identification of 11 downstream targets, suggesting that svb activates the expression of numerous additional genes to trigger the formation of embryonic epidermal cell extensions.
Several results suggest that this svb-controlled module is used in different developmental programs that produce cuticle extensions. Although signaling pathways act differently in ventral and dorsal embryonic regions, we show that svb is also required to express the same target genes for the formation of denticles and dorsal hairs. These results show that the svb targets identified so far act collectively to promote the formation of various epidermal extensions, despite the fact that they display different shapes. In addition, svb mutations also affect the formation of adult wing hairs and antennae laterals , which are known to require the activity of several svb targets identified in the embryonic epidermis (m, sn, f, sha for adult wing hairs, and sn, f, sha for antennae laterals). How can the same set of svb-regulated effectors participate in the formation of epidermal extensions of diversified morphology? Additional cytoskeletal factors can be differentially expressed in distinct epidermal regions independently of svb. It is also possible that other regulators modulate the response to svb (cf. the weak expression of f and wsp in naked cells expressing ectopically svb, when compared to that of sn, m, sha and y). Finally, upstream signaling pathways certainly contribute to the sculpting of each kind of trichome through the regulation of the expression and/or activities of cellular effectors. Recent studies have shown that members of the planar polarity pathway (PCP) are indeed involved in defining denticle polarity in response to signaling pathways . Interestingly, one of the identified svb targets, shavenoid, has recently been reported to interact with PCP in the adult wing, raising the possibility that such a dialog also occurs during embryonic epidermal cell remodeling. svb thus appears to govern a morphological module responsible for the major switch from smooth surface to trichome, the precise shape of which is finely sculptured by independent intrinsic factors and activities.
The pattern of trichomes has been modified several times during the evolution of insects. Across the genus Drosophila, at least four independent evolutionary transitions have led to the loss (to various extents) of dorsal hairs. Nevertheless, evolution of the pattern of dorsal hairs and denticles results from the modification of svb expression in all studied cases [17–19]. Although the expression of patterning genes is unchanged in all species examined , the difference in the dorsal hair pattern between D. melanogaster and D. sechellia is due only to the modification of shavenbaby's response to signaling pathways . Our analysis further shows that the restriction of svb expression in D. sechellia embryos causes, in turn, the restriction of the dorsal expression of m, sn, f, sha, and wsp genes (Figure S2). Therefore, all these svb targets display a concerted modification of their expression in D. sechellia, bringing additional evidence that together they constitute a developmental module. Consistent with this interpretation, we show that individual inactivation, or experimental modifications of the expression, of any of the svb target genes identified so far are not sufficient to modify the trichome pattern. These data indicate that denticle formation requires multiple factors that act collectively to remodel epidermal cell shape. This requirement for many genes to build a cuticular extension doubtless constitutes a developmental constraint, explaining why modifications of the expression of svb, the factor that governs this entire set of genes, are required for the trichome pattern to evolve.
Evolutionary modifications of cis-regulatory elements of lin-48, the putative svb homolog in worms, were responsible for the difference in the position of the excretory duct between Caenorhabditis elegans and C. briggsae . Although accumulated data thus demonstrate the particular role of svb genes in morphological diversification between relatively close species, how they are related to the evolution of animal forms across more distant phyla remains an open question. Several features of svb function have been conserved in mammals, including the role of one of its homologs, m-ovo1, in the differentiation of epidermal derivatives  and its regulation by the Wnt pathway . Our identification of genes regulated by svb in flies now opens the way to evaluate the contribution of the different parameters contributing to the role of svb in morphological evolution, from the modification of its response to signaling pathways to that of its cellular targets.
We used UAS-svb; UAS-ovoA; Ptc-Gal4; Wg-Gal4; svb1; btd1, svb1; svbR9; w,m1; sn3; f3; f5; f36a; sha1; shaV15; sn3,f36a; m1,f36a; m1,sn3; m1,sn3, f36a; wsp1; wsp3; m1,sn3, f36a; m1,sn3, f36a;sha1; m1,sn3,f36a;shaV15; m1,sn3, f36a;wsp3; w, Df(1)m-MR; P(EP)EP345, and a w D. sechellia stock from the D. Stern Laboratory. Stocks with multiple mutations were obtained by recombination, and mutant chromosomes were kept over balancers carrying Kr-Gal4>>UAS-GFP transgenes . Transgenic lines were generated using standard P-element transformation. Based on the 2003 genome release, the 6kmin construct corresponds to genomic position chrX:11585555–11591094 and 11596605–11597139, 3kmin to chrX:11585555–11588410; positions, both inserted into pCasPER β-gal. Wild-type, or site-directed mutagenized, 0.4Kmin enhancer was amplified from genomic regions (11487355–11487753) and cloned into the pCβ vector (carrying a nuclear-LacZ reporter) and in the pGal4 vector for rescue experiments. A full-length m cDNA inserted into the pUASt vector was used to generate pUAS-min transgenic lines. Details of cloning strategy are available upon request.
Antisense probes derived from cDNAs or genomic fragments were synthesized in vitro and processed for in situ hybridization following standard procedures. To identify svb mutant embryos, we used either a svb mutant chromosome carrying the btd1 mutation (leading to head defects; a gift from E. Wieschaus), or embryos sorted using the GFP balancer. For high-resolution imaging, ventral epidermis was hand-dissected. Immunostaining was performed according to , with anti-Miniature  at 1/400, anti-Forked  at 1/300, anti-Singed (Development Studies Hybridoma Bank, Iowa City, Iowa, United States) at 1/2, anti β-galactosidase (Cappel, Solon, Ohio, United States), Alexafluor488-labeled antibodies (Molecular Probes, Eugene, Oregon, United States) and TRITC-phalloidin (Sigma, St. Louis, Missouri, United States). Embryos were mounted in Vectashield (Vector Laboratories, Burlingame, California, United States) and photographed with a Leica TSP2 confocal microscope (Wetzlar, Germany).
Live mutant embryos (lacking GFP fluorescence from balancers) were hand-selected and cuticles prepared in hoyers/lactic acid (1/1). Abdominal segments 3 and 4 were imaged using a Zeiss Axioplan microscope equipped with phase contrast (Zeiss, Oberkochen, Germany). All high-magnification pictures correspond to the midline region of the fourth abdominal segment. For scanning electron microscopy, stage 16 embryos were fixed 15 min in heptane saturated in glutaraldehyde, devitellinized by hand and, after several water washes, dehydrated through an ethanol series. They were then dried using CO2 in a BOMAR SPC-900/EX critical-point dryer (Bomar, Tacoma, Washington, United States), sputtered (MED 020, Bal-Tek, Balzers, Liechtenstein) with a 20-nm gold palladium coat and examined with a Philips XL30 field emission microscope. For transmission electron microscopy, stage 16 embryos, fixed and devitellinized as above, were post-fixed for 1 h in para-formaldehyde 3%, glutaraldehyde 0.5% in PBS and then in 1% osmium-tetroxide for 1h. After dehydration, samples were embedded in Epon. Sections were stained with uranyl acetate and photographed with a Leo 510 (Zeiss) microscope.
DNA binding assays were performed as previously described . Wild-type or site-directed mutagenized m DNA probes (position 11487606–11487768) were synthesized by PCR, end-labeled and incubated with recombinant proteins, produced in bacteria, and purified according to the manufacturer's specifications (Pharmacia, Uppsala, Sweden). We used GST-SVB, a fusion protein with the 450 C-terminal aa of Svb encompassing the DNA binding domain, and GST alone as control. Electrophoresis gels were analyzed using a Bio imaging analyzer (Fuji, Tokyo, Japan).
Embryos were collected for 1 h at 18 °C, then allowed to develop at 25 °C for a period corresponding to 11, 12, and 14 h, before being processed for actin and immunological staining. Confocal views of abdominal segments A3 and A4 were recorded using a Leica TSP 2 microscope.
(2.6 MB JPG)
Epidermal expression of wasp, forked, singed, and shavenoid in stage 15 embryos from D. melanogaster (left) and D. sechellia (right) species. All pictures are lateral views, corresponding to the same abdominal segments. Anterior is to the left and dorsal is up.
(513 KB JPG)
Candidates were selected by three independent criteria. (1) Expression pattern: we searched for genes transcribed in a subset of epidermal cells, by conducting a survey of the project of systematic determination of patterns of gene expression in Drosophila embryogenesis that runs in the Berkeley Drosophila Genome Project (BDGP; http://toy.lbl.gov:8888/cgi-bin/ex/insitu.pl), and from published literature. (2) Denticle phenotype: we selected genes reported to specifically affect denticle formation when mutated. (3) Putative function: we selected candidates known to be involved in a biological process that is implicated in denticle formation, either actin remodeling, cuticle pigmentation, or cuticle formation.
We confirmed or established the expression pattern of candidates by in situ hybridization, and determined whether their expression depended upon svb activity by comparing their expression between wild-type and svb mutant embryos. Each candidate showing a strong reduction of its epidermal expression in svb mutants was further validated through the analysis of consequences both of svb ectopic expression (wgsvb) and of the inhibition of svb activity in a subset of denticle cells (ptcOvoA). The different columns refer to the name of candidates, their synonym, their respective Flybase identifier (http://flybase.bio.indiana.edu/), their putative function, a summary of their embryonic expression at the time of epidermal differentiation and their dependence on svb activity.
CNS, central nervous system; Ep, epidermal; NT, not tested.
(35 KB PDF)
The denticle belt of the fourth abdominal segment of wild-type and simple or multiple mutant embryos was photographed at high resolution (using a 63× phase contrast objective, NA [numerical aperture] = 1.4, and a digital Nikon CCD camera). Pictures were processed and analyzed using the ImageJ software (http://rsb.info.nih.gov/ij). We focused on the fourth row of denticles, which are relatively large and display a characteristic stereotyped morphology. Analyses were performed on five to ten individuals of the same genotype, to obtain a number of denticles superior to 70.
(A) Summary of the consequences of svb target inactivation on denticle formation. We measured several geometrical parameters, including denticle area, perimeter, width, and height. We also determined the number of denticles that are present along the fourth row (*, we focused on a 75 μm-wide region, corresponding to the most ventral part). Statistical analyses (3 way ANOVA, with denticle being nested in individuals, which is nested in genotype on a given variable) confirm that despite intragroup variations, differences between genotypes can be considered as highly significant (p < 0.001). The only exception is between m1 and Df(1)mr, which is not statistically different, in a good agreement with the fact the m1 mutation is reported to be a null m allele.
(B) Histograms plot the respective effects of svb targets on denticle size (as estimated by the area) and number along the row. Inactivation of each individual target decreases denticle size to various extents, a phenotype aggravated when mutations are cumulated. The absence of shavenoid causes a severe reduction of the size resulting from multiple split denticles, as indicated by the increase of the apparent denticle number (and confirmed by scanning electron microscopy analyses). Combination of m and sn (and/or f) also leads to denticle splitting, albeit at a lesser extent. The simultaneous inactivation of m, sn, f, and sha results in a strong reduction of the number of denticles that are replaced by naked regions, indicating a functional interaction between these genes for denticle formation.
(1.1 MB JPG)
We thank J. Smith, P. Blader, A. McGregor, D. Stern, I. Delon, L. Waltzer, and other members of our laboratory for discussions and critical reading of the manuscript. We are grateful to Y. Latapie and P. Valenti for their help during the latest stages of this work. We are grateful to G. Guild, D. Stern, E. Wieschaus, the Bloomington and National Institute of Genetics (NIG-Fly) fly stock centers, and Developmental Studies Hybridoma Bank (DSHB) for providing us with fly stocks and antibodies. We very much thank M. Affolter, in whose laboratory part of this work was done, and Daniel Mathys and Ursula Sauder (Zentrum Mikroskopie der Universität Basel) for the skilful electron microscopy pictures.
Author contributions. HCD, FP, and SP conceived and designed the experiments. HCD, IF, FP, and SP performed the experiments. HCD, IF, FP, and SP analyzed the data. FR and FP contributed reagents/materials/analysis tools. HCD, FP, and SP wrote the paper.
Funding. This work was supported by grants from the Association pour la Recherche contre le Cancer (3832) and Fondation pour la Recherche Médicale (programme équipe 2005). HCD was supported by a fellowship from the Ligue Nationale Contre le Cancer.
Competing interests. The authors have declared that no competing interests exist.