Ultraspiracle-independent anti-apoptotic function of ecdysone receptors is required for the survival of larval peptidergic neurons via suppression of grim expression in Drosophila melanogaster

Abstract

In Drosophila melanogaster a significant number of heterogenous larval neurons in the central nervous system (CNS) undergo metamorphosis-associated programmed cell death, termed metamorphoptosis. Interestingly distinct groups of doomed larval neurons are eliminated at different metamorphic phases. Although ecdysone hormonal signaling via nuclear ecdysone receptors (EcRs) is known to orchestrate the neuronal metamorphoptosis, little is known about how this signaling controls such diverse neuronal responses. Crustacean cardioactive peptide (CCAP)-producing neurons in the ventral nerve cord (VNC) are developmentally programmed to die shortly after adult emergence. In this study, we show that disruption of endogenous EcR function by ectopic expression of dominant negative forms of EcRs (EcR^DN) causes premature death of larval CCAP neurons in a caspase-dependent manner. This event is rescued by co-expression of individual EcR isoforms. Furthermore, larval CCAP neurons are largely normal in ecr mutants lacking either EcR-A or EcR-B isoforms, suggesting that EcR isoforms redundantly function to protect larval CCAP neurons. Of surprise, a role of Ultraspiracle (Usp), a canonical partner of EcR, is dispensable in the protection of CCAP neurons, whereas both EcR and Usp are required for inducing metamorphoptosis of vCrz neurons shortly after prepupal formation. As a downstream, grim is an essential cell death gene for the EcR^DN-mediated CCAP neuronal death, while either hid or rpr function is dispensable. Together, our results suggest that Usp-independent EcR actions protect CCAP neurons from their premature death by repressing grim expression until their normally scheduled apoptosis at post-emergence. Our studies highlight two opposite roles played by EcR function for metamorphoptosis of two different peptidergic neuronal groups, proapoptotic (vCrz) vs. antiapoptotic (CCAP), and propose that distinct death timings of doomed larval neurons are determined by differential signaling mechanisms involving EcR.

Introduction

Toward the end of growth period, juvenile animals undergo significant changes in the body as well as in the CNS in order to accommodate adult life style that is different from juvenile one. Most dramatic metamorphic changes are seen in holometabolous insects and amphibians in which obsolete juvenile tissues and organs are degenerated in conjunction with de novo formation of adult ones. It is known that endocrine hormones drive such developmental reprogramming and underlying mechanisms of hormonal actions have been a subject of extensive researches [1, 2].

In holometabolous insects including the fruit fly Drosophila melanogaster, metamorphosis is primarily orchestrated by a steroid hormone, ecdysone. In Drosophila, consecutive pulses of ecdysone at the end of third instar larval and prepupal stages induce massive cell death of larval organs [2]. This metamorphosis-regulated programmed cell death (PCD) has been referred to as metamorphoptosis in order to distinguish it from PCD events occurring during embryonic development that do not involve ecdysone signaling [3]. In contrast to the annihilation of larval organs, the larval CNS continuously functions throughout metamorphosis; however, it is sculptured extensively to form adult CNS. At the cellular level, the larval neurons are subjected to two different fates. Persisting neurons undergo remodeling process that sheds larval-specific neural projections and acquires adult-specific ones, while unnecessary neurons are programmed to die mostly via apoptosis. However, PCD of doomed larval neurons initiates at different phases of metamorphosis depending on the neuronal types [4–8], suggesting that heterogenous molecular death mechanisms operate in distinct neuronal groups. Little is known how ecdysone signaling induces such diverse responses of doomed larval neurons.

In order to assist future investigations on the upstream molecular events that trigger apoptosis in a metamorphic phase-specific manner, we propose to distinguish the doomed larval neurons largely into three classes based on the time of their death. Death class-I includes the neurons dying shortly after puparium formation (APF) that is driven by late larval ecdysone pulse. This class is represented by eight pairs of ventral corazonin neurons (vCrz) [3, 9–11], eleven pairs of CEv motoneurons located in the ventral median area of thoracic and abdominal segments, and two pairs of CEd neurons in the second and third thoracic segments [3]. Death class-II encompasses the doomed neurons in which their death is initiated by a rather small ecdysone peak that leads to pupal formation approximately at 12 h APF. A group of RP2 motoneurons located in the dorsal median area of the VNC represents this class [12]. Death class-III includes the larval neurons that survive throughout metamorphosis and die shortly after adult eclosion. A group of ~300 neurons, previously dubbed as type-II neurons, belongs to this class [13]. These neurons are characterized by a high level of immunoreactivities specific to ecdysone receptor A isoform (EcR-A) throughout pupal development. Majority of these neurons undergo PCD within 24 h after adult emergence [13]. Another group of neurons producing CCAP neuropeptide is also subject to postemergence PCD in a manner similar to the type-II neurons; however, these neurons show weak or moderate EcR-A immunoreactivities, and thus they are branded as type-I neurons [13–15].

It has been shown that PCD of vCrz neurons and RP2 motoneurons, each representing death class-I and -II neurons, respectively, is triggered by ecdysone signaling and EcR-B isoforms are essential to activate downstream apoptotic pathway [9, 12]. By contrast, PCD of the death class-III neurons is triggered by the reduction of ecdysone titers below the threshold, which results from the degeneration of the prothoracic glands, a source of ecdysone production [13, 16, 17]. When ecdysone titers are maintained high by an injection of ecdysone within 30 min after eclosion, PCD of both type-II and CCAP neurons is aborted along with transcriptional suppression of two major death promoting genes, reaper (rpr) and grim [14, 17]. These data suggest that ecdysone signaling protects the death class-III neurons via suppression of these death genes. However later studies showed that grim plays a major role for the PCD of CCAP neurons [15].

CCAP neurons provide an excellent model system to comprehend the protective role of ecdysone signaling for the death class-III type, as they are readily identifiable using various histological tools as well as malleable to transgenic manipulations [18, 19]. Intriguingly, we found that precocious death of CCAP neurons is brought about by ectopic expression of dominant-negative form of EcR (EcR^DN) that disrupts endogenous EcR function. We further delved into the molecular mechanisms underlying EcR-mediated protection of larval CCAP neurons and differences between antiapoptotic and proapoptotic roles played by EcR signaling.

Materials and methods

Fly Stocks

Flies were raised at 25°C on cornmeal-yeast-agar medium. Canton-S was used as a wild type strain and either yellow white (y w) or w¹¹¹⁸ as genetic controls. Three gal4 drivers, ccap-gal4 [18], bursicon-gal4 [15], and crz-gal4 [9, 20], were used to express various UAS-responders in CCAP and Crz neurons. The following homozygous or trans-heterozygous mutant alleles were used; ecr³¹/ecr ⁹⁹ (ecr^31/99) for ecrB [21], ecr¹¹²/ecr^M554fs (ecr^112/M554fs) and ecr¹³⁹/ecr^M554fs (ecr^139/M554fs) for ecrA [22, 23], X14/hid^P05014 (X14/hid^P) for hid [24] and rpr⁸⁷ for reaper (rpr) [25]. To simultaneously knockdown rpr, hid, and grim, a UAS-miRGH line was used. This line carries micro RNA-based hairpin constructs targeting the three cell death genes simultaneously [11, 26]. Similarly, UAS-migrim [11] and UAS-miusp [27] were used for the silencing of grim and ultraspiracle (usp), respectively. Two UAS-EcR^DN lines were described in Cherbas et al. [28]. To generate vCrz neurons carrying homozygous usp³ mutation by MARCM technique [29], two recombinant fly lines were crossed: (1) FRT^19A, tubP-gal80, hs-FLP; crz-gal4, UAS-mCD8GFP, (2) FRT^19A, usp³/FM7. Embryos collected from 0–8 h after egg laying were heat-shocked for 1 h at 37°C to induce mitotic recombination and then incubated at 25°C.

Construction of UAS-Usp³

To generate UAS-Usp³ construct, genomic DNA was extracted from usp³ heterozygous mutant.The usp³ allele carries a point mutation, substituting a conserved Arg160 for His (R160H) at the base of the second zinc finger domain, which disables Usp³ to bind DNA [30]. PCR was employed to obtain 1657-bp-long usp genomic region (−43 to +1614 relative to translation start site +1) by using Fwd (ACCggtaccTCCAATATACCCA, KpnI site in lower-case) and Rev (AGGGAtctagaGGAGAAATGC, XbaI site in lower-case) primers. The PCR products were subcloned into pGEMTeasy vector, and then a recombinant plasmid clone carrying usp³ allele was confirmed by sequencing analysis (pGEMT-Usp³). Because of the methylation of XbaI site, the pGEMT-Usp³ was amplified in dam-negative JM110 E. coli cells. The purified plasmid was digested with KpnI and XbaI, and then the Usp³ fragment was inserted into the pUAST vector. The resulting UAS-Usp³ was used for germ-line transformation. Two independent lines each inserted in II and III chromosome were used in this study.

Histochemistry of whole-mount CNS

To detect LacZ expression in specific neurons, CNSs were dissected in PBS containing 0.2% Triton X-100 and 10% ethanol, and then fixed in 0.5-mL of 0.2% glutaraldehyde for 5–10 min. Following washing in PBS, the tissues were incubated in 300 µL of X-gal solution (final 2 mg/mL) for overnight at 37°C. The tissues were rinsed in PBS, dehydrated in ethanol and mounted in 90% glycerol. Whole-mount bursicon and Corazonin immunohistochemistry (antibursicon was kindly provided by Dr. Ben White, NIMH) was performed, as described previously[15]. To assess co-localization of EcR isoforms in CCAP neurons, mCD8GFP-reported CCAP neurons were co-labeled with mouse monoclonal anti-EcR-A, anti-EcR-B1, or anti-EcR-COM [31]. Fluorescent images were obtained with Olympus BX61 or Keyence BZ-X710 microscopes.

Results

Premature death of CCAP neurons induced by EcR^DN

Three subgroups of CCAP neurons located in the subesophageal (sCCAP), dorsal abdominal (dCCAP), and ventral thoracic (vCCAP) regions of the CNS are programmed to undergo apoptosis within 24 h after adult emergence [14, 15]. Because the apoptosis of these neurons is coincident with lowest ecdysone titers and it is prevented by administration of ecdysone immediately after adult eclosion, it has been postulated that ecdysone signaling through EcR-A isoform provides these neurons with a protective measure against premature PCD during pupal development when EcR-A expression is prevalent.

To delve into the molecular mechanisms underlying EcR-A isoform-mediated protection, we attempted to disrupt EcR function in the CCAP neurons via expression of dominant-negative forms of EcRs (EcR^DN), EcR^W650A and EcR^F645A, by using a ccap-gal4 driver. Expression of EcR^DN caused severely defective pupal development, producing a failure in head eversion and shortened adult appendages such as wings and legs (Fig. 1a-c). Severity of the pupal defects was greater with EcR^W650A than with EcR^F645A (n=50 for both, Fig. 1b, c).

Fig. 1.

Effect of EcR^DN expression by ccap-gal4 on pupal development and CCAP neurons. a Control (ccap>lacZ) shows normally developing appendages, wings (W) and legs (L) as indicated by arrows. b, c EcR^DN expression in CCAP neurons. (b) ccap>lacZ, EcR^W650A and (c) ccap>lacZ, EcR^F645A compromised pupal development, resulting in the defective formation of wings and legs (arrows). EcR^W650A produced more severe phenotypes than EcR^W645A did. d LacZ-reported control CCAP neurons (ccap>lacZ) in the wandering third instar larval CNS (Br, brain; VNC, ventral nerve cord; SEG, subesophageal ganglion). e, f Double transgenic ccap-gal4; UAS-lacZ line was crossed to each UAS-EcR^DN line. Nearly complete elimination of CCAP neurons by EcR^W650A expression (e) and partial removal by EcR^F645A expression (f). g, h Rescue of EcR^DN-mediated defective pupal development by p35 expression. (g) EcR^W650A + p35 and (h) EcR^F645A + p35. i, j Rescue of EcR^DN-induced loss of CCAP neurons by p35 expression (ccap>lacZ, EcR^DN, p35). (i) EcR^W650A + p35 and (j) EcR^F645A + p35. Scale bar in (d), 100 µm

Because the defective pupal phenotypes by EcR^DN expression are comparable to those of flies having CCAP neurons either functionally inactivated [19, 32] or prematurely ablated during larval growth [18, 33], we carried out histological analysis of larval CCAP neurons in the wandering third instar larval stage. In control (ccap>lacZ), CCAP neuronal somata were shown in the protocerebral and subesophageal regions of the brain and in both thoracic and abdominal ganglia of the ventral nerve cord (VNC) (Fig. 1d). However, in EcR^W650A-expressing larvae nearly all of CCAP neurons in both subesophageal and VNC regions disappeared except for a pair of CCAP neurons in each brain hemisphere that seemed largely normal (n=16, Fig. 1e). EcR^F645A expression produced similar pattern, except for a few and weakly stained somata remained in the VNC (n=16, Fig. 1f). Nonetheless, these results indicate that the disappearance of the CCAP neurons by EcR^DN expression was specific to those that normally undergo apoptosis after adult eclosion.

Undetectable CCAP neurons in response to EcR^DN expression could be a result from transcriptional repression of ccap gene or ablation of CCAP neurons. To test if the latter is the case, we employed p35, a virally originated pan-caspase inhibitor. As a result, co-expression of p35 completely prevented EcR^DN-mediated defective pupal development (n=150, Fig. 1g, h).Consistent with the developmental rescue, normal patterns of larval CCAP neurons were restored (n=20, Fig. 1i, j). These data support that the loss of EcR function causes the precocious death of CCAP neurons in a caspase-dependent manner. Therefore, we suggest that EcR function is required to suppress a premature onset of death program in larval CCAP neurons which function is required for normal pupal development.

Developmental timing of EcR^DN-mediated apoptosis

To determine precise developmental timing of EcR^DN-induced apoptosis of CCAP neurons, we performed time-course detection of CCAP neurons labeled with GFP-reporter. To do this, a double transgenic line bearing both ccap-gal4 and UAS-mCD8GFP was generated and crossed it to the UAS-EcR^W650A. Notably, CCAP neurons looked normal in the first instar (L1) and early 2^nd instar (L2) larvae (n>6 for L1 and L2, Fig. 2a, b, ai), but we found a few neurons missing in more mature L2 larvae (n=5, Fig. 2c, ci). In early feeding 3^rd instar (FL3), more neurons became undetectable and nearly all disappeared in later wandering larvae (WL3) (n>7 for FL3 and WL3, Fig. 2d-g). Since GFP expression appeared clearly visible in L1 stage (Fig. 2a, ai), the survival of CCAP neurons in such early larvae is unlikely due to the lack of ccap-gal4 activity. A likely possibility is that EcR^DN has not been accumulated sufficiently to counteract wild-type EcR function at this stage.

Fig. 2.

Gradual loss of CCAP neurons by EcR^DN expression during larval development. Double transgenic UAS-mCD8GFP; ccap-gal4 line was crossed to UAS-EcR^W645A. a-g GFP-reported CCAP neurons are shown from L1 to WL3 stages, as indicated above each panel. Arrows over L2 and L3 indicate aging within the same instar. L3 stages are termed as feeding L3 (FL3) and wandering (WL3). ai, ci Magnified images of a and c, respectively. Loss of CCAP neurons in the VNC began to be detectable in late L2 (c, ci) and early L3 (d). Scale bars, 100 µm

Redundant roles of EcR isoforms for the protection of CCAP neuron death

The ecr locus produces three EcR isoforms, EcR-A, B1 and B2, varying in N-termini but bearing identical DNA and ligand binding domains. Previous studies with isoform-specific mutations and expression suggested differential roles played by each isoform in a stage- and tissue-specific manner during post-embryonic development [21–23, 31, 34]. In general, EcR-B1 isoform is predominant in late larval organs, whereas EcR-A is prevalent in adult progenitor cells [31].

To gain insight into the roles of each isoform for the protection of CCAP neurons, EcR isoforms were detected in the larval CCAP neurons by isoform-specific immunohistochemistry. Consistent with previous report [34], we detected weak expression of EcR-A, but conspicuous EcR-B1 expression broadly in the larval VNC (Fig. 3a, b). Immunoreactivity with anti-EcR-com,which detects all isoforms, was stronger than that of either EcR-B1 or EcR-A, implying a presence of EcR-B2 (Fig. 3c). Co-labeling experiments indicate that both EcR isoforms are present in the CCAP neurons (Fig. 3d-f). These results raise the possibility of protective roles played by all three isoforms in the larval CCAP neurons.

Fig. 3.

Expression of EcR isoforms in larval CNS and CCAP neurons. a-c EcR isoform-specific immunoreactivities in the WL3 VNC. (a) EcR-A, (b) EcR-B1, (c) all EcR isoforms by anti-EcR-com. A doublet of CCAP neurons located from 1^st to 3^rd abdominal neuromeres (indicated by brackets) was selected to show double-labeling for the EcR and ccap>GFP as shown below panels. d-f Detection of EcR isoforms (red signals) in the GFP-reported CCAP neurons (green) by anti-EcR-A (d), anti-EcR-B1 (e), and anti-EcR-com (f). Scale bars for (a) and (d) are 100 µm and 10 µm, respectively.

Next, we tested if co-expression of each EcR isoform is able to rescue EcR^W650A-induced CCAP neuron death. Expression of either EcR-B1 or EcR-B2 fully rescued the death (Fig. 4a-c), whereas EcR-A did so partially (Fig. 4d). Consistent with the neuronal rescue phenotypes,normal pupal phenotype was fully restored by either EcR-B1 or EcR-B2 isoform but to a lesser extent by EcR-A (Fig. 4e-g).

Fig. 4.

Rescue of EcR^DN-induced PCD by EcR isoforms. a Control LacZ-reported CCAP neurons (ccap>lacZ). b EcR^W650A-induced precocious CCAP cell death (ccap>lacZ, EcR^W650A). c, d Complete rescue of EcR^W650A-mediated CCAP neuronal death by co-expression of either EcR-B1 or EcR-B2 (c) and partial rescue by EcR-A (d). ccap-gal4; UAS-lacZ line was crossed to UAS-EcR^W650; UAS-EcR. e-g EcR^W650A-led defective pupal phenotypes are rescued by co-expression of EcR isoforms. Complete rescue by EcR-B1 (e) and EcR-B2 (f) but partial rescue by EcR-A (g). An arrow in (g) points to short legs. Scale bar for (a), 100 µm

To further verify the roles of EcR isoforms for the protection of CCAP neurons, we employed genetic mutations lacking either EcR-A or EcR-B isoforms. Since a majority of CCAP neurons co-expresses another neuropeptide bursicon (burs), we visualized them by using anti-burs [15, 19]. In two of heteroallelic mutations lacking EcR-A isoform expression, ecr^112/M554fs and ecr^139/M554fs, burs-immunoreactivities were comparable to wild-type, implying that CCAP neurons were protected by EcR-B isoforms (n>6, Fig. 5a-c). In ecr^31/99 mutants lacking both EcR-B1 and EcR-B2 isoforms, we observed one or two burs-immunoreactive neurons missing particularly in the thoracic segments of some specimen (n=7, bracket in Fig. 5d), suggesting that CCAP neurons were also mostly protected by EcR-A. Taking these results together supports that all three EcR isoforms redundantly function to maintain larval CCAP neurons alive, but EcR-B isoforms play slightly more prominent roles than does EcR-A particularly during larval stages.

Fig. 5.

CCAP neurons in ecr mutants devoid of either EcR-A or EcR-B isoforms. CCAP neurons were detected by anti-bursicon. Since bursicon expression is limited to subsets of CCAP neurons, bursicon-negative CCAP neurons in the protocerebrum were not detectable. a-d Larval CNS. (a) Wild-type. (b, c) Heteroallelic ecrA mutants, 112/M554fs (b) and 139/M554fs (c), showed wild-type pattern of bursicon-positive neurons. (d) Heteroallelic ecrB mutant, 31/99. A few bursicon-positive neurons in thoracic neuromeres (bracket) were frequently lost. e-I Bursicon-immunoreactive neurons in the pupal VNCs at 24–48h APF. (e) Control M554fs/+. (f, g) 112/M554fs and (h, i) 139/M554fs mutant. Arrows in (h, i) indicate missing bursicon neurons. Scale bars, 100 µm

EcR-B levels are higher than EcR-A in the CNS during larval development. However, during metamorphosis, EcR-B decreases precipitously while EcR-A gradually increases, thus becoming a dominant isoform in pupal CNS [34]. Since CCAP neurons are immunoreactive only with anti-EcR-A in later pupal stage [14], EcR-A is likely the main protector of CCAP neurons at this stage. If so, it is expected that CCAP neurons might be prematurely eliminated in ecrA mutants during pupal development. To test this, we examined pupal CCAP neurons in ecrA mutants. However, this is a challenge, as both ecrA mutants (112/M554fs and 139/M554fs) died mostly during larval stages and rare escapers were developmentally arrested and died in prepupal and early pupal stages, respectively. We managed to obtain a small number of pupal escapers from both mutant alleles for the immunohistochemical analysis.

During early pupal development, numbers of burs-immunoreactive CCAP neurons are reduced but sizes of somata increase in wild-type [35], because of transcriptional silence of burs in about a half of larval neurons in the VNC. Consistent with this report, we observed about seven pairs of strong burs-immunoreactive CCAP neurons in the abdominal ganglia of control (+/M554fs) at 24–48 h APF (n=14, Fig. 5e). Interestingly both 112/M554fs (n=5) and 139/M554fs (n=3) mutants showed more neurons compared to the control and some of these neurons retained larval features as judging from the small soma size and larva-like projections (Fig. 5f-i). These suggest that transcriptional regulation of burs expression is somehow altered in the ecrA mutants. Because of this factor, it was difficult to recognize dead burs neurons.Nonetheless, we observed one or two missing pupal neurons in 139/M554fs mutant VNC, as indicated by arrows (Fig. 5h-i). We speculate that residual EcR-B is still capable of blocking premature PCD of CCAP neurons to some extent at this stage. The results, however, need to be interpreted cautiously, because it is possible that the missing pupal neurons in the ecrA mutant could be a result from the systemic effect of developmental arrest.

Ultraspiracle (Usp) is required for the normal apoptosis of vCrz, but not for the protection of CCAP neurons

Since EcR is known to form a functional heterodimeric complex with another nuclear receptor Usp, a fly ortholog of vertebrate retinoid X receptor (RXR) [36, 37], we wanted to determine if the EcR-Usp complex provides CCAP neurons with antiapoptotic function. Because EcR is known to be essential for PCD of vCrz peptidergic neurons [9], we first investigated if Usp is required for this event.

As we reported previously, all vCrz neurons died apoptotically by 6–7 h APF and their death was completely suppressed by EcR^DN expression (Fig. 6a-c). Because of embryonic and L1 lethality of the usp-null mutants [38], we employed microRNA-based knockdown of usp using a pre-existing UAS-miusp line [27]. Remarkably, targeted expression of miusp in Crz neurons completely blocked vCrz apoptosis as effectively as EcR^DN did (Fig. 6c, d).

Fig. 6.

Essential role of Usp for vCrz apoptosis. a, b Eight pairs of corazonin-immunoreactive neurons in the wild-type VNC of WL3 indicated by a box (a) are completely removed by 7 h APF (b). c, d Complete blocking of vCrz PCD by expression of EcR^W650A (c) and miusp (d). e-g Mosaic analysis (MARCM) of usp³ in vCrz neurons (FRT^19A, usp³/FRT^19A, hs-FLP, tubp-gal80; UAS-mCD8GFP, crz-gal4/+). Arrows indicate that two vCrz neurons bearing homozygous usp³ mutation did not undergo PCD at 6 h APF. These neurons were labeled by both GFP (e) and Crz-immunosignals (f). A merge of (e) and (f) (g). h Ectopic expression of Usp³ completely blocked vCrz PCD at 7 h APF (crz-gal4/UAS-Usp³). Scale bar, 100 µm

To further corroborate a Usp’s autonomous role, we performed mosaic analysis (MARCM), as described in Lee et al. [29]. vCrz neurons bearing homozygous usp³ mutation, as detected by GFP expression, showed normal-looking appearance in both cell body and neural projections at 6 h APF, indicating that PCD of these vCrz neurons was blocked in the absence of Usp function (arrows in Fig. 6e, n=5). A few vCrz neurons that are weakly positive for corazonin immunoreactivity but GFP-negative in some specimen at this stage (Fig. 6f, g) are normally dying neurons having smaller cell bodies and degenerating axons, as we reported previously [10]. These data together strongly support that metamorphoptosis of vCrz neurons requires ecdysone signaling conveyed by a conventional EcR-Usp complex.

Next, we investigated if Usp is also essential for the survival of larval CCAP neurons. To our surprise, miusp expression did not induce premature killing of the CCAP neurons (Fig. 7a-c). This result can be due to insufficient reduction of endogenous usp expression in the CCAP neurons, although it is unlikely based on the miusp results shown for the vCrz neurons. Anyway, we designed a putatively dominant negative form of Usp as an alternative way to interfere with endogenous Usp function like EcR^DN. A candidate is Usp³ mutant protein that carries R160H substitution in the second zinc finger motif [30]. This study has shown that Usp³ can compete effectively with wild-type one for the dimerization with EcR, but the EcR-Usp³ complex fails to bind its target DNA binding sites.

Fig. 7.

Dispensable role of Usp for the protection of larval CCAP neurons. UAS-mCD8GFP;ccap-gal4 line was crossed to indicated UAS lines. a GFP-reported larval CCAP neurons in control (ccap>mCD8GFP). b Premature CCAP neuron death by EcR^W650A. c-e CCAP cell death was not caused by either miusp (c) or Usp³ expression (d, e). Two independent UAS-Usp³ lines inserted in 2^nd (II) and 3^rd (III) chromosomes showed the same results. Scale bar, 100 µm

To test Usp³’s dominant negative action in vivo, we generated UAS-Usp³ transgenic lines,and then the results from ectopic Usp³ expression driven by various tissue-specific gal4 drivers were compared to those from miusp and EcR^DN expression. Ubiquitous expression of miusp or Usp³ using a tub-gal4 caused embryonic lethality (Table 1) and similar results were found for EcR^W650A as reported previously [28]. Expression of miusp or Usp³ in the fat body using r4-gal4 driver [39] similarly caused pupal lethality, whereas EcR^W650A severely retarded larval growth leading to larval death at L1-L2 stage (Table 1). Expression by A9-gal4 [40] also showed different effects between EcR^DN and Usp³ expression on larval and pupal development (supplemental Fig. 1). EcR^W650A caused lethality broadly ranging from early larvae to pupae.Dying larvae frequently showed melanotic spots in the digestive system; pupal development was severely compromised and all pupae failed to reach adulthood (supplemental Fig. 1a-e). By comparison, both Usp³ and miusp expression similarly led to the developmental arrest at late pupal stage and majority of pupae died inside the pupal case (Table 1, supplemental Fig. 1f).Comparable developmental arrest phenotypes produced by ectopic Usp³ and miusp expression in various gal4 drivers suggest that Usp³ mutant form effectively competes with wild-type Usp protein to disrupt normal Usp functions. Because wild-type EcR and Usp expression patterns do not completely overlap in some tissues (modENCODE tissue expression data; http://flybase.org/reports/FBlc0000206.html), different effects by EcR^DN and Usp³ (and miusp) on the development suggest that some EcR functions are independent of Usp, or vice versa.

Table 1.

Comparison of developmental defects by ectopic expression of EcRW^650A, usp³ and miusp.

gal4	expression	EcRW650A	Usp3	miusp
tubulin	ubiquitous	emb. lethal	emb. lethal	emb. lethal***
r4	fat body	L1-L2 lethal	pupal lethal**	pupal lethal****
A9	imaginal discs	larva-pupal lethal*	pupal lethal*	pupal lethal****

^*see also supplemental Fig. 1

^**During early pupal development, fat bodies do not seem to be distributed to the head region, making it transparent. Also, normally granule-shaped fat bodies are not clearly developed. Survival rate into adults was ~12%.

^***Very few larval escapers were found, but none survived beyond L2 stage.

^****Most of them were died in pharate adult stage. Survival rate into adults was~10%.

In addition to the gross developmental effects, we tested if Usp³ expression suppresses the metamorphoptosis of vCrz neurons. As expected, targeted expression of Usp³ in the Crz neurons completely inhibited vCrz cell death (n=12, Fig. 6h), which is consistent with the results observed from miusp-mediated knockdown. These results verify that Usp³ is capable of disrupting EcR-Usp function for PCD of vCrz neurons by acting as a dominant negative form.In contrast, Usp³ did not cause premature death of larval CCAP neurons (Fig. 7d, e), which was consistent with miusp expression. Based on these data, we conclude that the EcR signaling plays a role, independently of Usp, for the prevention of premature onset of CCAP apoptosis.

grim is the cell death gene mediating EcR^DN-induced CCAP-neuron death

In Drosophila cells, a death-inducing signal promotes transcriptional activation of a single or a combination of cell death genes rpr, grim, and head convolution defective (hid), commonly referred to as RGH. These death promoters antagonize endogenous caspase inhibitor Diap1, thereby leading to the activation of caspases [41]. Since EcR^DN-mediated CCAP-neuronal death also involves caspase activation (Fig. 1), it is conceivable that disruption of EcR signaling by EcR^DN triggers expression of the cell death genes. The RGH genes have been shown to work in a cell-type specific manner [e.g., ref. 2], prompting us to identify specific death genes responsible for the EcR^DN-induced CCAP death.

We first employed a UAS-miRGH that knocks down rpr, grim, and hid simultaneously [11, 26]. As a result, we found that EcR^W650A-induced death was rescued effectively by miRGH (n=8, Fig. 8a, b). Furthermore, knockdown of grim alone by migrim [11] rescued CCAP death as comparably as miRGH did (n=12, Fig. 8c). In contrast, no roles were found for hid and rpr, as the death was not rescued in either hid-null mutants (hid^P/X14) or rpr-null homozygous mutants (rpr⁸⁷) (n>7 for both, Fig. 8d, e). These findings suggest that grim is the major cell death gene whose expression is suppressed by EcR-mediated function in larval CCAP neurons.

Fig. 8.

Essential role of grim for EcR-mediated protection of larval CCAP neurons. a Control (UAS-mCD8GFP; ccap-gal4 x y w). b, c EcR^W650A-caused premature cell death was blocked by knockdowns of RGH (UAS-mCD8GFP; ccap-gal4 x UAS-EcR^W650A, UAS-miRGH) (b) and by grim knockdown (UAS-mCD8GFP; ccap-gal4 x UAS-EcR^W650A, UAS-migrim) (c). d, e EcR^W650A-caused cell death was not blocked in either a heteroallelic hid-null mutant (hid^P/X14) (UAS-mCD8GFP; ccap-gal4; X14/TM6B x UAS-EcR^W650A; hid^P/TM6B) (d), or rpr⁸⁷ homozygosity (UAS-mCD8GFP; ccap-gal4; rpr⁸⁷/TM6B x UAS-EcR^W650A; rpr⁸⁷/TM6B) (e). f-h Rescue of EcR^W650A-induced CCAP cell death by Diap1 expression. Control (UAS-mCD8GFP, burs-gal4 x y w) (f). EcR^W650A expression by burs-gal4 (UAS-mCD8GFP, burs-gal4 x UAS-EcR^W650A) (g). Diap1 expression significantly rescued EcR^W650A-mediated premature cell death (UAS-mCD8GFP, burs-gal4; UAS-diap1 x UAS-EcR^W650A) (h). Scale bar, 100 µm

Since Grim is known to antagonize Diap1 to activate caspases, we further tested if Diap1 also plays an important role for the protection of larval CCAP neurons. To do this, we used a burs-gal4 to co-express Diap1 and EcR^W650A because of the convenience of transgenic manipulations. burs-gal4 activity is limited to one pair of subesophageal and abdominal CCAP neurons in the larval CNS (n=14, Fig. 8f). Expression of EcR^W650A alone prematurely eliminated most burs-expressing CCAP neurons (n=12, Fig. 8g), whereas Diap1 overexpression blocked it in many neurons (n=7, Fig. 8h). These results suggest that grim acts as an upstream death inducer required for the activation of caspases by antagonizing Diap1.

Discussion

The final steps in insect metamorphosis include adult emergence, expansion of the adult wings,and cuticular hardening and darkening. CCAP neurons play important roles for these events by producing multiple neuropeptides and a neuropeptide receptor [42–44]. Then as soon as their roles are complete, these neurons undergo apoptosis. Although CCAP neurons in the subesophageal brain area are also programmed to die, two particular CCAP subgroups, ventral CCAP and dorsal CCAP (a.k.a. bursCCAP), are relatively well studied with respect to their apoptotic events [3, 14, 15]. Falling ecdysone titers toward the end of pupal development [13, 14, 17, 45] are considered to be a key developmental cue inducing the PCD of both CCAP and type-II neurons. It was initially thought that changes of ecdysone titers per se trigger the PCD event, but further studies suggest that it is initiated when insufficient ecdysone signaling fails to represstranscription of the death promoting genes, thus implying ecdysone signaling acting as a protective agent [14, 17]. Since EcR-A is the only detectable isoform in type-II and CCAP neurons in pharate adults, this isoform has been suggested to repress cell death genes [13, 17].

In this study, we further pursued anti-apoptotic aspects (protection theory) of ecdysone signaling by disrupting EcR function in CCAP neurons. From these experiments, we anticipated premature induction of CCAP neuronal death during pupal development when EcR-A levels are moderate. The expression of EcR^DN indeed led to the premature death; surprisingly however, it took place much earlier during larval development. In addition, the results from rescue experiments and isoform-specific mutants suggested that not only EcR-A but also both EcR-B isoforms protect CCAP neurons in a redundant manner. Since CCAP neuronal functions are required for larval molting, pupal development, and pupal ecdysis in addition to adult eclosion and maturation [18, 44], the functional redundancy of EcR isoforms seems important in order to provide continuous protection for CCAP neurons from larval development and onward.

Previous studies found expression of grim and rpr in dying CCAP neurons and n4 type-II neurons [14, 17], implying important roles of both genes for the PCD of these neurons. However, data from gene-specific mutations or knockdown of individual death genes have clearly demonstrated that it is not rpr but grim that is primarily responsible for post-emergence PCD of CCAP neurons [15]. As shown here, grim is also a key player for inducing premature death of larval CCAP neurons by EcR^DN. Therefore, we suggest that EcR^DN-mediated premature death of CCAP neurons mirrors postemergence PCD of them and premature onset of CCAP apoptosis is prevented by continuous suppression of grim expression by redundant function of EcR isoforms either directly or indirectly. Interestingly, grim is also the major death gene responsible for the metamorphoptosis of another doomed larval peptidergic neuronal group, vCrz neurons [11]. For this event, however, grim expression is induced by EcR signaling.

How does the same ecdysone signaling induce grim expression in one type of larval neurons but represses it in other type? Our studies provide an insight into the differential molecular mechanisms of EcR action. One major difference is a role played by Usp. Metamorphoptosis of vCrz neurons is blocked by either Usp³ or miusp expression as well as EcR^DN, indicating that Usp is an essential component for the vCrz death as a canonical EcR-Usp complex. In contrast, Usp function appears to be dispensable for the protection of larval CCAP neurons, as disruption by a dominant-negative Usp³ did not cause CCAP death. One might argue that EcR-Usp³ complex is still capable of repressing grim expression because DNA binding domain in Usp is not needed for such negative action. However, similar results obtained from a miusp-mediated knockdown support that Usp is not an essential component for the EcR’s repressive function. Hence, we propose that EcR acts independently of Usp to render protection for larval CCAP neurons and perhaps type-II larval neurons that similarly die after adult eclosion. One possible mechanism of Usp-independent function of EcR is that EcR functions with an unknown partner. Having said this, Usp was reported to form a dimer with DHR38 other than EcR [46, 47]. Additional difference between pro-and anti-apoptotic roles of EcR signaling comes from a differential use of EcR isoforms. While only EcR-B isoforms are necessary for vCrz PCD [9], all three EcR isoforms are redundantly required for anti-apoptotic function in larval CCAP neurons. EcR-A might provide an additional layer of protection to maintain the CCAP neurons alive in larvae. And then, because EcR-A is only detectable isoform during later pupal development [13], its role is likely to protect CCAP neurons in the pupal CNS until their destined demise at post-eclosion. Therefore the antiapoptotic role of ecdysone/EcR signaling plays an important role in determination of developmental PCD timing of CCAP neurons.

Lastly, our studies raise the possibility that accidental loss of steroid-mediated endocrine signaling can cause abnormal CNS development by altering fates of doomed postembryonic neurons, as exemplified by continued survival of vCrz neurons and premature death of CCAP neurons in Drosophila. In this respect, it is worthy to note that many artificial chemicals, collectively termed endocrine-disrupting chemicals (EDCs), mimic steroid hormones. In vertebrates, these chemicals are well known to disrupt development and reproduction by interfering with endogenous steroid hormones. Neurons in the brain are not immune to EDCs either. A growing body of evidence shows that steroid hormones play important roles in neuroprotection following brain injuries via various mechanisms including suppression of cell death [48]. Hence, such neuroprotective roles can be compromised by EDCs, as demonstrated by in vitro and in vivo studies [49–51]. Alternatively, EDCs can interfere with normal course of developmentally-regulated neuronal cell death. In the primary visual cortex, male rats have 19% more neurons than do females due to a greater degree of steroid-mediated cell death in the females [52, 53]. Sex-biased apoptosis also contributes to sexually dimorphic development of other brain areas [54]. Either premature degeneration or interference with developmentally regulated normal apoptosis can result in the formation of abnormal neural architectures and/or activities, which potentially cause various impairments of the neurological conditions such as defective cognition, abnormal behaviors or mental retardation. Further studies will shed more lights on the effects of EDCs on the steroid-associated neuronal apoptosis and neuroprotection.

Supplementary Material

Supplemental Fig. 1

Developmental phenotypes by A9-gal4 driven expression of EcR^W650A or Usp³. a-e EcR^W650A expression caused various developmental defects in larval stages (a-b) and in pupal stages (c-e), whereas Usp³ expression did not affect larval growth but led to death of pharate adults inside the pupal case (f).

Abbreviations

AE

After eclosion

APF

After puparium formation

CNS

Central nervous system

PCD

Programmed cell death

VNC

Ventral nerve cord

CCAP

Crustacean cardioactive peptide

Crz

Corazonin

EcR

Ecdysone Receptor

USP

Ultraspiracle