Wing-to-Leg Homeosis by Spineless Causes Apoptosis Regulated by Fish-lips, a Novel Leucine-Rich Repeat Transmembrane Protein

Abstract

Growth, patterning, and apoptosis are mutually interactive during development. For example, cells that select an abnormal fate in a developing field are frequently removed by apoptosis. An important issue in this process that needs to be resolved is the mechanism used by cells to discern their correct fate from an abnormal fate. In order to examine this issue, we developed an animal model that expresses the dioxin receptor homolog Spineless (Ss) ectopically in the Drosophila wing. The presence of mosaic clones ectopically expressing ss results in a local transformation of organ identity, homeosis, from wing into a leg or antenna. The cells with misspecified fates subsequently activate c-Jun N-terminal kinase to undergo apoptosis in an autonomous or nonautonomous manner depending on their position within the wing, suggesting that a cell-cell interaction is, at least in some cases, involved in the detection of misspecified cells. Similar position dependence is commonly observed when various homeotic genes controlling the body segments are ectopically expressed. The autonomous and nonautonomous apoptosis caused by ss is regulated by a novel leucine-rich repeat family transmembrane protein, Fish-lips (Fili) that interacts with surrounding normal cells. These data support a mechanism in which the lack of some membrane proteins helps to recognize the presence of different cell types and direct these cells to an apoptotic fate in order to exclude them from the normal developing field.

Signals that regulate cell growth, patterning, and apoptosis are interdependent during development (1, 19, 29, 37). In particular, there are likely to be multiple ways to elicit an apoptotic cell fate since apoptosis serves a variety of functions in multicellular organisms (29). In regard to the cell autonomy of apoptosis, a cell-autonomous apoptosis occurs during cell competition, a phenomenon whereby cells that grow slowly due to the mutation of essential genes such as Minute or ras are removed later in development (39, 40, 49, 56). In contrast, it has been reported that nonautonomous cell death is often associated with cell fate changes by altering the morphogen activities (2, 3) of Decapentaplegic (Dpp) (32, 41) or Wingless (Wg) (42, 60). In these cases, apoptotic cell death occurs both within and outside of the abnormal cell population and is referred to as morphogenetic apoptosis. Similarly, when spalt (sal), a target gene of the Dpp signal, is ectopically expressed apoptosis is induced through interaction with the surrounding normal cells (38). Thus, the removal of physiologically abnormal cells by apoptosis can occur by more than one mechanism, although the details of the various processes remain to be elucidated. In order to search for and analyze the ways to remove other types of abnormal cell, we have ectopically expressed various master genes, including homeotic genes, to create cell populations with developmentally abnormal fates. In most cases, cells with abnormal fate induced severe apoptosis in a cell autonomous or nonautonomous manner. Among these, we focus on the master gene ss, of which overexpression in the wing changes the organ identity into a ventral appendage such as a leg or antenna.

Based on sequence identity and splice site conservation, Ss is the closest Drosophila homolog of the mammalian dioxin receptor (arylhydrocarbon receptor [Ahr]) (17, 27). Both mammalian and Drosophila proteins can also bind to the xenotoxin responsive element (XRE) and stimulate transcription from genes containing this cis-acting element (18). Although Ss has not been shown to bind to arylhydrocarbons, it regulates normal morphogenesis of the leg or antenna and bristles, all of which are major Drosophila sensor organs or tissues that respond to environmental chemicals.

Our results reveal that ectopic ss provokes a wing-to-leg and/or antenna homeosis that subsequently elicits apoptosis in an autonomous or nonautonomous manner. This apoptotic response is regulated by a novel transmembrane leucine-rich repeat (LRR) protein, Fili, and may be a common process induced by ectopic expression of various homeotic genes. These results indicate that homeosis elicits a complex set of signals that influence the survival of the transformed cells and their surrounding cells. In addition, these results suggest a hypothesis that different LRR family transmembrane proteins function in different subdomains of a developing field to recognize developmentally misspecified cells and to regulate cell survival.

MATERIALS AND METHODS

Fly strains.

We used two different GAL4 enhancer trap lines of scalloped (sd) in the present study. The (s) and (w) lines induce a strong and weak expression of UAS-driven genes, respectively. The plasmid construct for tgo-RNAi was made by inserting an inverted tgo cDNA into the 5′ region of the standard UAS-tgo construct. When tgo-RNAi was induced singly in the wing, the fly wing showed no phenotype (data not shown).

The name fish-lips was derived from the observation that the expression pattern of fili in the wing disc made a fish lips-like shape. fili is identical to the temporarily assigned hypothetical gene CG4054 in the Berkeley Drosophila Genome Project. The fili-lacZ fly was identified during a screen of lacZ enhancer trap strains with P-lacW, a derivative of the transposon P-element (6). The fili mutant allele Δ102 was isolated by an imprecise excision of P-lacW in fili-lacZ according to the standard procedure. The UAS-fili fly was made by inserting the fili cDNA (2,297 bp, including full-length open reading frame, 18 bp of the 5′ untranslated region and 62 bp of the 3′ untranslated region) into the pUAST vector (7) with a white⁺ marker and transforming the white⁻ fly according to standard procedures.

The other lacZ enhancer trap lines we used are sd^ETX4, Dll⁰¹⁰⁹², hth⁰⁵⁷⁴⁵, hid⁰⁵⁰¹⁴, omb^P1, fz3^SW076-A7.3XW, trn^S064117, and puc^E69. hid⁰⁵⁰¹⁴ was also used as a mutant (see Fig. 4E and G). At the late-third-instar larval stage, this hid-lacZ is expressed irrespective of apoptosis around the dorsal-ventral boundary and at the two spots in the dorsal and ventral hinge region (data not shown). However, at the mid-third-instar larval stage, these expressions do not begin, and the expression is associated with apoptosis. Therefore, we observed the hep^CA- or ss-induced hid-lacZ expression by using the mid-third-instar larvae.

FIG. 4.

Involvement of clone shape and hid expression. (A) Clones expressing green fluorescent protein alone. All of the clones show an irregular shape. (B) ss and DIAP1 (IAP)-overexpressing clones. Most of the clones are rounded. (C) Higher magnification of a clone boxed in panel B. JNK activation is found nonautonomously but highly accumulated inside the clone, whereas caspase-3 activation is found only outside of the clone, which may be due to the ability of DIAP1 to breakdown caspase-3. (D) ss-overexpressing clones in a hep mutant background. Most of the clones are rounded. (E) ss-overexpressing clones in a hid mutant background. Most of the clones are rounded. (F) ss-overexpressing clones under the control of tgoRNAi. Most of the clones show an irregular shape and do not induce activation of JNK and caspase-3 significantly. (G and G′) Expression of constitutively active Hep (green) induces hid-lacZ expression (magenta). To create an intact shape of the clone by repressing apoptosis, the enhancer trap mutation hid-lacZ (hid⁰⁵⁰¹⁴) is homozygous in this experiment. (H and H′) ss overexpression (green) induces hid-lacZ expression (magenta) autonomously in the wing blade region (arrow) and nonautonomously in the wing hinge region (arrowheads).

Detection of apoptosis.

Caspase-3 plays a central role in many types of apoptosis, whereas c-Jun N-terminal kinase (JNK) activation elicits a limited group of apoptotic events (14). In the Drosophila larval wing disc, activation of JNK always leads to the activation of caspase-3 (2). Puckered (Puc) is a protein phosphatase specifically inactivating JNK, and its transcription is induced by JNK signal, thereby making a negative-regulatory circuit (36). Most of the endogenous and ectopic puc-lacZ expression observed in the present study was eliminated in a mutant background of the hep gene (20), which encodes the homolog of mitogen-activated protein kinase kinase-7 that activates JNK. Immunofluorescence was carried out according to the standard procedure. The puc-lacZ expression is detected by measuring mouse anti-β-galactosidase antibody (1:200 dilution; Promega Z378) and Cy3-labeled anti-mouse immunoglobulin (1:200 dilution; Jackson Immunoresearch). The rabbit anti-active caspase-3 antibody (CM1) was a gift from Idun Pharmaceuticals. It specifically recognizes the cleaved and activated form of mammalian caspase-3 and has been shown to also recognize the cleaved form of the Drosophila caspase-3 homolog, Drice (59). Thus, active caspase-3/Drice was detected by CM1 (1:4,000 dilution) and Cy5-labeled anti-rabbit immunoglobulin (1:200 dilution; Jackson Immunoresearch).

Antibodies.

The other antibodies we used were rat anti-Al (from G. Campbell [1:1,000 dilution]), rat anti-Sal (from R. Barrio [1:400 dilution]), mouse anti-Dac (MABDAC2-3 from the Developmental Studies Hybridoma Bank [1:10 dilution of supernatant]), rabbit anti-Vg (from S. Carroll [1:200 dilution]), rabbit anti-p-Mad (PS1 from P. ten Dijke [1:200 dilution]), mouse anti-Arm (N2 7A1 from the Developmental Studies Hybridoma Bank [1:10 dilution of supernatant]), mouse anti-Dll (MAb DMDll.1 from D. Duncan [1:500 dilution]), and rabbit anti-Caps (from A. Nose [1:200 dilution]).

Mosaic overexpression analysis.

Discs were prepared from larvae carrying the AyGAL4 (actin promoter-FRT-yellow⁺, terminator-FRT-GAL4 [28]), hs-FLP (yeast FLP recombinase gene driven by heat shock promoter), UAS-ss, and UAS-GFP transgenes. The GAL4-expressing clones were induced by heat treatment at 37°C for 20 min at 48 to 72 h after egg-laying and observed 48 to 72 h after heat treatment. In each experiment, we observed more than 10 wing discs to confirm the results.

RESULTS

Overexpression of ss leads to a disappearance of the wing, which is dependent on tgo.

Using the GAL4/UAS system (7), we overexpressed ss in the wing blade region by scalloped (sd) (s)-GAL4 (Fig. 1A). As a result, no adults developed wings (Fig. 1C). When we expressed ss in a more restricted region using the weaker allele, sd(w)-GAL4 (Fig. 1B), the wing margin structure disappeared (Fig. 1D and E). These results suggest that ectopic expression of ss in the wing leads to massive apoptosis.

FIG. 1.

Activation of ss leads to loss of wing. (A and B) sd-strong(s) and sd-weak(w) expression patterns (green) in the wing imaginal discs. Anterior is to the left, and dorsal is at the top. (C) Wild-type fly (left) and a fly without wing by overexpression of ss under the control of the sd(s)-GAL4 driver (right). (D) Wild-type wing. (E) A wing overexpressing ss under the control of the sd(w)-GAL4 driver shows an incision of the wing margin. (F) Front view of the normal fly head with white⁻ compound eyes for easier recognition of antenna. (G) Front view of the fly head in which tgo is repressed by UAS-tgo-RNAi with Dll-GAL4. The antenna transforms to a leg-like appendage, which is similar to the antenna mutant for ss or tgo (17, 18). (H) A wing overexpressing ss but repressing tgo under the control of the sd(w)-GAL4 driver shows a reversion of the wing margin incision found in the ss overexpression alone.

Mammalian Ahr is known to function together with a structurally similar protein, Arylhydrocarbon receptor nuclear translocator (Arnt), by forming a heterodimeric transcription factor (27). To test whether the wing apoptosis phenotype by ss overexpression is mediated by Tango (Tgo) (43), the Drosophila homolog of Arnt, we examined flies in which we simultaneously induced both ss expression and tgo-RNAi. Prior to this experiment, we examined whether tgo-RNAi can remove endogenous tgo function. When tgo-RNAi was induced in the antenna by using Distal-less (Dll)-GAL4, the antenna transformed to a leg-like organ as seen in the mutant of tgo (18) (Fig. 1F and G). Thus, the tgo-RNAi was confirmed to repress the endogenous tgo. When ss was induced together with this tgo-RNAi, the wing phenotype was completely suppressed (Fig. 1H), suggesting that the loss of the wing due to ectopic expression of ss requires tgo activity and is not simply due to a stress caused by ectopic expression of ss.

Overexpression of ss leads to a homeotic transformation from wing to ventral appendage before disappearance of the cells.

ss is known to specify the identities of tarsus (distal leg) and arista (distal antenna) during normal development (17, 18). Thus, we thought that the ss overexpression in the wing may have removed the organ identity as a wing and underwent a homeotic change to a leg, antenna, or other ventral appendage that does not actually exist in the normal fly prior to disappearance of the wing. To investigate this possibility, we next examined the expression of marker genes specific to the wing or haltere or to leg or antenna. The sd gene, which is normally expressed exclusively in the wing and haltere (10) (Fig. 2A), was found to be less strongly expressed in ss-overexpressing clones (Fig. 2C and C′). This result suggests that the ss-overexpressing clones had lost, at least partially, their wing identity. In contrast, Dll, a master gene that directs the identity of the distal leg or whole antenna (13, 45) (Fig. 2A′), was ectopically and cell autonomously induced in the ss-overexpressing clones (Fig. 2D and D′). During normal wing development, Dll expression is induced in response to Wg signaling. However, Dll induction in the ss-overexpressing clones appeared to be independent of the Wg signal since Dfrizzled-3 (fz3) induction and Armadillo (Arm) accumulation did not coincide with the ss-overexpressing region, as demonstrated below. Therefore, we hypothesized that this Dll induction might provoke a different Dll function to redirect the identity of the wing to that of the distal leg or whole antenna. In fact, high-level induction of Dll in the wing has been reported to lead to its transformation into a distal leg (21). Although normal expression of ss in the leg is known to be downstream of Dll (17), Dll expression is regulated by ectopic ss in the wing, indicating the presence of a positive feedback loop between Dll and ectopic ss.

FIG. 2.

Homeotic transformation of organ identity from that of a wing to that of a leg or antenna by overexpression of ss. (A and A′) Expression of sd and Dll in the wild-type imaginal discs. Haltere (halt.), leg, wing, antenna (ant.), and eye discs are shown. (B) Expression of various genes as markers for the leg or antenna segments in the ss-overexpressing clones. A schematic representation at the bottom of the panel shows normal expression in the antenna (left) and leg (right). The open bar in the sal expression in the antenna indicates the broad expression (in the a2 and a3 segments) found only in the early stage. This expression is restricted to a narrower region (only a2) later in development. A table indicating increases (+) or decreases (−) in gene expression in the ss-overexpressing clones in the wing blade and hinge regions is shown at the top. NE, no expression was seen. (C to H and C′ to H′) Wing discs showing an increase (white arrows) or decrease (gray arrows) of gene expression (magenta) in the ss-overexpressing clones (green). ss overexpression was induced, together with DIAP1 (IAP), to delay apoptosis except for panel H. Asterisks in panel G′ indicate normal expression of Dac in the peripodial membrane. (I) An ectopic tarsus-like structure generated in the wing by coexpression of ss and DIAP1 (IAP) with sd(s)-GAL4 driver. (I′) t1 segment of the distal mid-leg in wild type. (J and J′) Normal expression of ptc (green) and Wg (magenta) in the leg and wing. An asterisk indicates a ventral stripe of Wg expression in the leg. (K and K′) Expression of ptc (green) and Wg (magenta) in the wing in which ss is overexpressed by ptc-GAL4 driver. Asterisk indicates an ectopic stripe of Wg expression in the ventral wing. The gray arrow indicates the reduced Wg expression that is not found in the normal wing.

Next, we examined the expression of various genes that are markers for different subdomains of the leg or antenna (Fig. 2E to H) (8, 9, 15, 16, 35, 45; Flybase [http://flybase.bio.indiana.edu/]). In summary, ss-overexpressing clones in the blade region of the wing have most likely transformed their identity to that of a distal leg (t1 area) rather than that of an antenna (Fig. 2B). In the hinge region of the wing, Dll expression was not induced by ss (Fig. 2D). Instead, the expression of hth, a master gene that directs the identity of the proximal leg or whole antenna (12, 44) was enhanced in the ss-overexpressing clones (Fig. 2H and H′). However, in the normal leg or antenna there was no segment that expresses both ss and hth but not Dll (Fig. 2B). Thus, the fate of the ss-overexpressing clones in the hinge region of the wing should have transformed to that of a ventral appendage, although we have not identified which normal appendage segment that the ss-overexpressing clones resemble. In comparison, a leg-antenna intermediate and a hypothetical ancestor appendage have been observed in Dll and ss mutants, respectively (17, 18).

We also obtained other evidence demonstrating that the cells with ss overexpression display characters of the ventral appendages. First, an ectopic appendage-like structure was often generated (Fig. 2I) when ss-induced apoptosis was partially inhibited by coexpression of Drosophila inhibitor of apoptosis protein 1 (DIAP1 [26]). The structure most similar to this appendage that is seen in external morphology of normal fly is the tarsus of leg (Fig. 2I′). Second, during normal development expression of patched (ptc) was dependent on Hedgehog signaling, which induces a band-like expression of dpp in the wing and the dorsal leg region and that of wg in the ventral leg region (5) (Fig. 2J and J′). When ss expression was induced by ptc-GAL4, Wg was ectopically induced only in the ventral half of the ptc-expressing region (Fig. 2K and K′), as found in the normal leg or antenna. However, the level of wg induction in the ptc-expressing region was weaker than that found in the normal leg. Therefore, the ss-expressing wing appears to have partially transformed its identity to that of a leg or antenna. This mixed or conflicted identity may also be responsible for induction of apoptosis. Although it has been shown previously that the ectopic expression of ss by ptc-GAL4 leads to a deletion of the central wing (17), the molecular mechanism was not identified. Our results suggest that the deletion is triggered primarily by homeotic transformation.

Overexpression of ss leads to apoptosis in which cell autonomy is position dependent.

To study the detailed mechanism leading to apoptosis by ectopic expression of ss, we next examined the cell autonomy of apoptosis in ss-overexpressing clones. We monitored caspase-3 and JNK activation by using an antibody that recognizes active caspase-3 and the expression of a reporter gene, puckered (puc)-lacZ, respectively. After 48 h of mosaic induction, we observed two types of apoptosis (Fig. 3A). The ss-overexpressing clones inside the wing blade primordium showed strong activation of caspase-3 and a weak activation of JNK (Fig. 3A′ and A"). Activation of caspase-3 in the vicinity of the dorsoventral compartment boundary (red line in Fig. 3A) was relatively weak and delayed. We can recognize the delay of the cell removal around the dorsoventral boundary more readily at later stage after the clone induction (data not shown). All of these responses were always cell autonomous. In contrast, the ss-overexpressing clones outside of the wing blade region (that is the hinge region) showed weak activation of caspase-3 and strong activation of JNK in cells on either side of the clone boundary (Fig. 3A′′′). In this region, the outline of the clones which were probably induced at an earlier stage, are severely disrupted with strong activation of both caspase-3 and JNK (Fig. 3A′′′′).

FIG. 3.

Overexpression of ss leads to autonomous and nonautonomous apoptosis in a Dll-dependent manner. (A) A wing imaginal disc with mosaic clones overexpressing ss (green). JNK activation visualized by puc-lacZ expression (magenta) and caspase-3 activation (cyan) are shown. Red line denotes the D/V boundary. (A′-A″″) Higher magnification of each clone in panel A. The total number of clones examined is 229. Among these, the number of clones exhibiting autonomous puc-lacZ induction is 30, whereas the number of those exhibiting nonautonomous puc-lacZ induction is 99. (B to B") Wing imaginal disc with ss overexpression driven by the ptc-GAL4 (green). Either cell autonomous or nonautonomous apoptosis was observed symmetrically as a function of the distance of the cells from the dorsoventral boundary. Arrowheads and arrows denote a nonautonomous activation of JNK (magenta) and autonomous activation of caspase-3 (cyan), respectively. (C) Wing subdomains based on the response to ss. The Dll region is white, the vg region is red, and the hth region is blue. (D) Wing imaginal disc with mosaic clones overexpressing ss (green) in the Dll⁵/Dll⁹ mutant background. JNK activation visualized by puc-lacZ expression (magenta) and caspase-3 activation (cyan) are greatly reduced. (E) A wing with ss overexpression by sd(w)-GAL4 in the Dll⁵/+ mutant background shows a strongly suppressed phenotype. (F and F′) Wing imaginal disc with mosaic clones overexpressing Dll (green). DIAP1 (IAP) is coexpressed to delay apoptosis. puc-lacZ expression (magenta). Caspase-3 activation is indicated in cyan. Arrows and arrowheads denote an autonomous and nonautonomous activation of JNK, respectively. (F") High magnification of the boxed area in panel C. (G) A wing with Dll overexpression by sd(w)-GAL4 shows an incision of the wing margin.

The presence of the two types of apoptosis can clearly be recognized in ss overexpression by ptc-GAL4, which is expressed in a band intersecting the dorsoventral boundary (Fig. 3B to B"). Strong, autonomous caspase-3 activation was observed in the blade region, whereas strong, nonautonomous JNK activation was observed around the hinge region. The region around the dorsoventral boundary did not show both JNK and caspase-3 activation. Thus, the wing disc can be divided into three subdomains based on the nature of the apoptotic response to ss overexpression in these regions (Fig. 3C). Interestingly, these subdomains correlate with the expression domains of several key transcriptional regulators. The Dll-expressing subdomain showed an autonomous and delayed response. The vestigial (vg)-expressing subdomain showed an autonomous and acute response. The homothorax (hth)-expressing subdomain showed a nonautonomous response. These subdomains can be recognized by the foldings of the disc cell layer or by distance between the nuclei, through counter staining with DAPI (4′,6′-diamidino-2-phenylindole). The variance of apoptotic responses are summarized in Table 1. Similar traits in position-dependent cell autonomy of apoptosis have also been found in the clones mutant for the Dpp receptor Thick veins (Tkv) (3).

TABLE 1.

Position-dependent variance of apoptotic responses to ss

Subdomaina	JNK variance	Caspase-3 variance	Cell autonomy	Response
Around DV boundary (Dll)	Weak	Weak	Autonomous	Delayed
Rest of blade (vg)	Weak	Strong	Autonomous	Acute
Hinge (hth)	Strong	Weak	Nonautonomous	Variable

^aRepresentative genes expressed in each subdomain are indicated in parentheses.

ss-induced apoptosis is dependent on Dll that is responsible for homeotic transformation from wing to leg or antenna.

To elucidate the role of Dll in ss-induced apoptosis, we placed the ss-overexpressing clones in a strong Dll mutant Dll⁵/Dll⁹ background. As shown in Fig. 3D, the activation of both JNK and caspase-3 was strongly suppressed by this Dll mutation in the wing blade region. Activation was also suppressed in the hinge region, although ss did not induce Dll expression in this region (Fig. 2D). A possible explanation is that broader expression of Dll in the hinge region at an earlier stage of development (data not shown) is required for transformation by ss in later stages. The Dll mutation also suppressed the ss-induced adult wing phenotype (Fig. 3E). To test whether Dll overexpression is also able to lead to apoptosis, we next generated Dll-overexpressing clones in the wing. As shown in Fig. 3F, JNK and caspase-3 activation are observed cell autonomously in the Dll-overexpressing clones in the wing blade region. Other areas show nonautonomy. The wing notching phenotype induced by Dll overexpression is found to be similar to that induced by ss overexpression (Fig. 3G). Thus, ss-induced apoptosis is likely mediated by a pathway that is dependent upon Dll. A discrepancy existed in that the normal expression of Dll in the wing did not induce leg identity or apoptosis. However, the elevated expression of Dll in the wing can induce some leg structure (21). Thus, the level of Dll expression, rather than its timing or place, may be crucial for determining the fate as wing or leg.

Coincidence of round shape and apoptotic response.

Apoptosis in each of these regions was dependent on Hemipterous (Hep, an activator of JNK [20]), DIAP1, Head involution defective (Hid; a proapoptotic protein [23]) and Tgo (Fig. 4B to F). DIAP1 is known to degrade caspase through its ubiquitin-protein ligase activity. Hid can bind to DIAP1 to stimulate autoubiquitination of DIAP1, which prevents the degradation of caspase (58). Interestingly, the ss-overexpressing clones in hep or hid mutant or in DIAP1 overproducer backgrounds retained a round shape, as seen with ss-overexpressing clones in the wild-type background (Fig. 4A to E). In contrast, the ss-overexpressing clones in the tgo-RNAi background lost their round shape and invaded the surrounding normal cells (Fig. 4F). These results indicate that hep and hid only affect the apoptosis pathway, whereas tgo mediated events more upstream, such as the cell fate change induced by ss. Ectopic expression of ss may make a difference in cell affinity between the clones and surrounding normal cells, which leads to a round shape of the clones and subsequent apoptotic response.

Consistently, activation of JNK pathway by constitutively active Hep leads to hid expression (Fig. 4G). Besides, the ss-overexpressing clone induces hid expression autonomously in the wing blade region but nonautonomously in the wing hinge region (Fig. 4H).

Effect of ss overexpression on the intensity of Dpp and Wg signaling.

The JNK-dependent cell death on either side of the clone boundary is quite similar to what has been referred to as “morphogenetic apoptosis,” through which discontinuities in the Dpp and Wg morphogen activity gradients are corrected (2). We next examined the activities of certain morphogen signals in the vicinity of the ss-overexpressing clones. With regard to Dpp signaling, we can find a slight reduction in the expression of the lacZ enhancer trap of optomotor-blind (omb), a target gene of the Dpp signal (24) (see Fig. S1A′ and B′ in the supplemental material). However, the reduction is always found around the center of the clones and does not coincide with the outline of the clones. We also see a similarly moderate alteration of the level of phospho-Mad, an active form of the Dpp signal transducer “Mothers against dpp” (Mad) (48, 52, 57) (see Fig. S1A" and B" in the supplemental material). In the case of Wg signaling, we examined expression of the fz3-lacZ enhancer trap (51, 55), a target gene of the Wg signal, and the level of Arm (46, 54), a Wg signal transducer that is known to accumulate in response to the Wg signal (25, 47). We observed a reduction of fz3-lacZ expression interior to the clone boundary of the ss-overexpressing clones (see Fig. S1C, C′, and D to D" in the supplemental material). However, in the most central cells, cytoplasmic accumulation of Arm was conversely observed (see Fig. S1C, C", D, and D′ in the supplemental material). These results suggested that the level of Wg signal is not uniformly received throughout the ss-overexpressing clones and is not directly regulated by Ss. Similar responses were observed in clones both in the wing blade and in the hinge regions, as seen in Fig. S1C in the supplemental material. Taken together, these results indicate that discontinuities in the strength of Dpp and Wg signal reception does not precisely coincide with the boundary of the ss-overexpressing clones, suggesting that other factors contribute to the nonautonomous JNK activation in this case.

Position-dependent variance in the autonomy of apoptosis is found also in clones ectopically expressing homeotic genes.

We next examined whether the position-dependent variance in cell autonomy of apoptosis is specific for ss-induced homeosis or a general trait exhibited by ectopic expression of diverse homeotic genes. The identity of each body segment is defined by a series of homeotic genes belonging to ANTP-C and BX-C. However, the Drosophila wing requires no input from ANTP-C and BX-C during its development (11). Instead, these homeotic genes repress wing development in all segments, except for the mid-thorax, where the wing normally develops. In order to produce a misspecified fate in wing cells, we expressed various homeotic genes in the wing discs. When labial (lab), proboscipedia (pb), Antennapedia (Antp), Ultrabithorax (Ubx), abdominal-A (abd-A), or Abdominal-B (Abd-B) were expressed by using sd-GAL4 driver, the adult wings showed abnormal vein patterns and incision of the wing margin (Fig. 5A to F). In the cases of Ubx or abd-A expression, wing incision was less frequently observed. In these cases, the wing has been shown to transform to an organ similar to a haltere (22). When mosaic clones expressing these homeotic genes are generated, JNK activation was found autonomously in the blade region but nonautonomously in the hinge region. All of these responses are quite similar to the results of ss overexpression, as has been described above. Thus, the position-dependent variance in cell autonomy of apoptosis may be a general phenomenon accompanying homeotic transformation.

FIG. 5.

Ectopic expression of various homeotic genes commonly induces autonomous death in the wing blade region and nonautonomous death in the wing hinge region. (A to F) Wings were generated with ectopic expression of the indicated homeotic genes by using sd-GAL4(s) or sd-GAL4(w) drivers. All of the wings show abnormal vein patterns and incision of the wing margin at various frequencies. (A′ to F′ and A" to F") Wing imaginal discs with mosaic clones ectopically expressing the indicated homeotic genes (green). puc-lacZ expression is indicated in magenta, and caspase-3 activation is indicated in cyan. Arrows and arrowheads denote an autonomous and nonautonomous activation of JNK, respectively. The total numbers of clones examined were 30 (lab), 22 (pb) 44 (Antp), 16 (Ubx), 48 (abd-A), and 33 (Abd-B). Among these, the numbers of clones exhibiting autonomous puc-lacZ induction are 15 (lab), 6 (pb) 6 (Antp), 4 (Ubx), 2 (abd-A), and 9 (Abd-B), whereas the numbers of those exhibiting nonautonomous puc-lacZ induction are 10 (lab), 4 (pb) 14 (Antp), 4 (Ubx), 4 (abd-A), and 16 (Abd-B).

Characterization of the novel gene fili that encodes a LRR family transmembrane protein.

Recently, clones with sal-induced misspecified fate were reported to be recognized by the surrounding normal cells through the LRR family transmembrane proteins, Capricious (Caps) and Tartan (Trn), to prevent apoptosis (38). The expression pattern of both caps and trn are complementary to the pattern of sal expression in the late third instar wing (38) (Fig. 6C). Thus, we investigated the role of a newly identified LRR family protein, Fili, which was expressed in the wing in a nearly complementary pattern as that of Dll (Fig. 6A and B), in regulating apoptosis in ss-overexpressing clones.

FIG. 6.

Expression patterns of LRR protein genes. (A to A") Expression of fili-lacZ (magenta) and Dll (green) in the wing, leg, and antenna discs; (B to B") in situ hybridization showing fili mRNA; (C) expression of Caps (blue), trn-lacZ (red), and Sal (green) in the wing disc.

The fly strain fili-lacZ was isolated by a screen of enhancer trap lines that displayed a previously unknown expression pattern. The transposon P-element carrying the lacZ gene was inserted at −56 relative to the presumptive transcription start site, TTAGTT (Fig. 7A). The deduced protein, Fili, was composed of 738 amino acid residues and had a single transmembrane domain and 14 LRR domains (Fig. 7B). These features are common to Caps and Trn, which share 55 and 56% homologies with Fili in the LRR domains, respectively (Fig. 7C and D). The homologies also extend to the amino- and carboxyl-terminus domains (Fig. 7E and F), suggesting that Fili plays a role that is especially related to Caps and Trn among numerous members of LRR family with a variety of functions. A mutant, Δ102, was isolated by imprecise excision of the P-element in fili-lacZ. It lacks 534 bp, including a part of the first exon, and is lethal at late embryonic stages. The main cause of this lethality is unknown. In the wing disc, fili exhibited a pattern of gene expression that was nearly complementary to that of Dll (Fig. 6A and B). In addition, fili expression was complementary to Dll expression in antenna discs (Fig. 6A" and B"). In contrast, in the leg disc, fili seemed to partially overlap with Dll (Fig. 6A′ and B′). However, we found that most of the cells expressing the respective genes were in different layers of the leg disc folding.

FIG. 7.

Structure of fili gene and Fili protein. (A) The horizontal line represents a segment of the genomic DNA around fili gene. The five exons in fili gene are indicated below the DNA. The P-element inserted in the fili-lacZ fly genome is indicated by the downward stippled arrowhead at the position −56 relative to the potential transcription start site. The deleted region in fili^Δ102 mutant is indicated around −56. “cen” and “telo” denote the direction toward centromere and telomere, respectively. (B) Domain organization of Fili protein. The deduced Fili protein possesses two hydrophobic stretches (orange), signal peptide (SP), and transmembrane domain (TM). In the extracellular domain, 14 LRRs are found (green). NF and CF refer to the amino-flanking and carboxy-flanking motifs, respectively. (C) Comparison of 14 LRRs in Fili. The identical amino acid residues are masked with orange. (D) Comparison of LRR consensus among five LRR family members. Variable amino acid residues are shown with hyphens. (E and F) Comparison of NF and CF among Fili, Trn, and Caps. Identical and similar amino acid residues are shown in orange and cyan, respectively.

ss-induced apoptosis can be controlled by Fili.

If fili is involved in the control of apoptosis in ss-overexpressing clone, its expression should be downregulated by ss. As expected, fili expression is repressed in the ss-overexpressing clones (Fig. 8A ′). Interestingly, in contrast, expression of caps (Fig. 8A") and trn (data not shown) is enhanced by ss. Similar responses in fili and caps can also be observed in Dll-overexpressing clones (Fig. 8B to B").

FIG. 8.

Ss and Dll regulate the expression of fili and Caps. (A to A") fili-lacZ expression (red) is repressed, whereas Caps expression (blue) is induced in the ss-overexpressing clones (green). (B to B") fili-lacZ expression (red) is repressed, whereas Caps expression (blue) is induced in the Dll-overexpressing clones (green). Areas of repression (gray arrows) and induction (white arrows) are indicated.

When ss was induced together with fili, JNK and caspase-3 activation was greatly reduced (Fig. 3A and 9A to A "). Residual JNK and caspase-3 activation was observed only in the most peripheral hinge region, where fili is not expressed in the normal wing. Similarly, in the central wing region, where fili is not expressed in the normal wing, caspase-3 activation was not suppressed by fili coexpression. As a control experiment, we observed that coexpression of both caps and ss did not suppress the ss-induced apoptotic response (Fig. 9B and B′). Consistently, ss overexpression does not repress but rather elevates the caps expression (Fig. 8A"). Thus, we conclude that fili expression is sufficient for survival of ss-overexpressing cells in the wing region where endogenous fili is expressed, probably through recognition by surrounding normal cells transmitting a survival signal, as shown for Caps and Trn (38).

FIG. 9.

Fili regulates ss-induced apoptosis. Areas of puc-lacZ expression (red), caspase-3 activation (cyan), and Caps expression (blue) are indicated. (A to A") Wing imaginal disc with mosaic clones overexpressing ss and fili (green). JNK activation and caspase-3 activation are suppressed. (B to B") Wing imaginal disc with mosaic clones overexpressing ss and caps (green). JNK activation is not suppressed by Caps. (C to C") Wing imaginal disc with mosaic clones overexpressing sal (green). Nonautonomous activation of JNK is found at the region where endogenous sal is not expressed normally. (D to D") Wing imaginal disc with mosaic clones overexpressing sal and fili (green). JNK activation is not suppressed.

We further carried out a control experiment to test the effect of fili on sal-induced apoptosis because sal-induced apoptosis was reported to be suppressed by coexpression of caps or trn. At first, we observed that the sal-overexpressing clones displayed a nonautonomous activation of JNK in the wing region where sal is not normally expressed (Fig. 9C to C"). This nonautonomy was quite similar to the morphogenetic apoptosis and ss-induced apoptosis, suggesting that the regulation of all of these apoptotic responses may share similar molecular mechanism(s). When sal was induced together with fili, JNK, and caspase-3 activation was not affected significantly (Fig. 9D to D"). Thus, these results indicate that fili does not suppress all cases of nonautonomously induced apoptosis.

Shape and apoptosis of clones mutant for or overexpressing fili.

Next, we examined the relationship of the clone shape and apoptosis in a simple fili mutant or in its overexpression. The fili mutant clones showed an irregular shape irrespective of their position in the wing disc (Fig. 10A) and did not display any apoptotic signs. These results are not always discrepant from the other results and are similar to observations in mutants for caps or trn alone (38). The other LRR family proteins might function with Fili in a redundant manner in the generation of affinity with surrounding normal cells. In contrast to the fili mutant, the fili-overexpressing clones showed a smooth outline (Fig. 10B). The smooth outlines were found even where endogenous fili was expressed, which may be due to the immense amount of its expression by using UAS/GAL4 system. However, the clone in the central region where endogenous fili was not expressed exhibited the most rounded shape (Fig. 10B′). Interestingly, unlike the case of caps and trn, the overexpression of fili alone induced apoptosis in a nonautonomous manner. When UAS-fili was induced by ptc-GAL4 (Fig. 10D), nonautonomous activation of JNK was apparent in the cells around the posterior edge of ptc expression but not in the cells around the anterior edge (Fig. 10D to D"). The posterior edge is known to create a sharp discontinuity of ptc expression levels, whereas the anterior edge does not (5). More interestingly, the area where JNK was activated did not overlap the area where endogenous fili was expressed. As a result, the areas with JNK activation were divided into three regions (arrowheads in Fig. 10D"). Also, forced expression of fili in the central wing blade in the adult wing generated a severely notched phenotype by apoptosis (Fig. 10C). These data strongly suggest that the occurrence of a large discontinuity in the fili expression level activates JNK nonautonomously.

FIG. 10.

Phenotypes of fili mutant and ectopic expression. (A) Wing disc with twin spots generated by somatic recombination of filiΔ¹⁰². Clones without green fluorescent protein are the homozygous filiΔ¹⁰² mutant. (A′) High magnification of the boxed area in panel A. (B) Wing disc with clones with fili overexpression (green). (B′) High magnification of the boxed area in panel B. (C) Adult wing where UAS-fili is driven by ptc-GAL4. (D to D") Wing disc where UAS-fili is driven by ptc-GAL4 (green). Areas of puc-lacZ expression (magenta) and caspase-3 activation (cyan) are also shown. Arrowheads indicate nonautonomous induction of puc-lacZ expression in places where endogenous fili is not expressed.

DISCUSSION

Ectopic ss induces a homeosis-dependent apoptosis that may cause a rare frequency of wing-to-leg transdetermination.

Homeosis is a naturally occurring rare phenomenon with formation of a local body part having characteristics that are normally found in a related part at another location in the body. This phenomenon had been studied exclusively in Drosophila melanogaster but can also be found in a variety of insects and other organisms (53). Although homeosis is rarely found in wild-type flies, transdetermination, a similar phenomenon, can be easily induced by in vivo culture and subsequent transplantation of imaginal discs between different individuals or by directly manipulating functions of various master and/or homeotic genes. Although such developmentally abnormal cells occurring in homeosis might a priori be expected to be removed during development to guarantee normal morphogenesis, apoptosis has not been previously observed to accompany homeosis. Our results suggest that the cells undergoing homeosis do in fact induce apoptotic responses to help prevent them from developing abnormal structures. The rare frequency in wing-to-leg transdetermination, which was observed about 30 years ago (30), may be derived partly from the high frequency of apoptosis associated with this particular transformation.

ss-induced autonomous apoptosis is related to cell competition.

Cell competition is a phenomenon whereby cell clones with a slow growth rate are eliminated during development. The autonomous cell death in cell competition has been demonstrated to be caused by an impaired reception of extracellular survival factors. As representative examples, clones with cells heterozygous for the M(2)60E mutation show an autonomous death as a result of reduced Dpp reception (40). Thus, the autonomous death in ss-overexpressing cells in the wing blade might be a type of cell competition response. In this case, however, the extracellular factor that the ss-overexpressing cells fail to receive is not likely to be Dpp, since we observed that the Dpp signaling level is not strongly affected in the ss-overexpressing cells (see Fig. S1A and B in the supplemental material). Also, ss-overexpressing cells can be removed from the lateral region of the wing blade where the level of Dpp signaling is normally low. In contrast, cells heterozygous for M(2)60E can survive in this area due to the loss of competition for Dpp (40). Therefore, Wg or other extracellular survival factors might be affecting the survival of ss-overexpressing cells, as discussed below.

Position-dependent variance in cell autonomy is found in the apoptotic response to ss overexpression.

We have not yet addressed the question of why the variance in cell autonomy of ss-induced apoptosis is related to dorsoventral patterning or blade/hinge subdomains. However, the variance suggests an interesting model in which signals that normally regulate dorsoventral appendage patterning or blade/hinge subdomains also directly affect the apoptotic response or specification of organ identity as cryptic mechanisms. Interestingly, ectopic expression of most of the segment-specifying homeotic genes commonly shows a similar position-dependent variance in cell autonomy (Fig. 5). In this view, Wg, the morphogen controlling the dorsoventral patterning (42, 60), or Vg, the selector of wing subdomain (31, 34), may be involved in the regulation of position-dependent variance in the cell autonomy of apoptosis. However, definitively demonstrating the involvement of Wg or Vg is difficult because changing Wg signal or Vg activity alone also induces a severe unrelated apoptosis (2). Thus, the simultaneous manipulation of ss and Wg/Vg activities may lead to a complicated result that is difficult to interpret.

Implication of various LRR family transmembrane proteins in the regulation of apoptosis to remove misspecified cells.

A recent study concerning the relation of apoptosis in sal-overexpressing cells and LRR family proteins Caps/Trn has provided a new insights into how cell survival of misspecified cells is controlled (38). Cells that overexpress sal lack caps/trn expression and are removed from the area where sal is not normally expressed. However, cells expressing both sal and caps can survive. In order to control cell survival in the wing, Caps and Trn presumably provide position-dependent recognition cues along the anteroposterior axis, whereas Fili appears to provide such cues to neighboring cells along the dorsoventral axis. These results raise the possibility that cells can recognize the type of neighboring cells using distinct LRR family members, depending on the position of the cells in the primordial tissues. The Drosophila genome contains at least 10 members of LRR family, most of which have not been examined for function (Flybase).

Functional relationship between Drosophila Ss and mammalian Ahr.

In the present study, several cellular responses to ss overexpression have been described. Clonal overexpression of ss induces a sequential response composed of an induction of leg- or antenna-specific genes, autonomous homeosis, rounding, and autonomous or nonautonomous apoptosis. The activation of mammalian Ahr is known to affect apoptosis, cell proliferation, differentiation, and morphogenesis in various tissues (50). Polycyclic aromatic hydrocarbon, a ligand for Ahr, induced apoptosis accompanied by the activation of JNK (33). It should also be noted that dioxin-induced responses such as P450 expression in the liver are not observed in all cells exposed to dioxin but are observed in a subset of centrilobular cells with the highest sensitivity to dioxin (4). Thus, the mosaic activation of Ss and subsequent JNK activation in surrounding cells in Drosophila induce circumstances similar to that of liver cells exposed to dioxin. With regard to morphogenesis, the type of morphogenetic signals that are modified or distorted during dioxin action is unknown, despite many examples of dioxin-induced mismorphogenesis that have been reported. Therefore, our studies on Ss in Drosophila may provide at least in part, a common molecular and cellular basis for understanding Ahr-induced apoptosis and mismorphogenesis. It could involve homeotic transformation of organ identity, recognition by surrounding cells through LRR protein, and apoptosis-mediated large deletion of tissues that can cause a malformation during organogenesis. Interestingly, all of the genes involved in this sequential process are evolutionarily conserved in mammals. Similar and detailed analysis in mammalian system will resolve these issues.