Broad-Complex acts downstream of Met in juvenile hormone signaling to coordinate primitive holometabolan metamorphosis

The majority of animal species on the Earth are holometabolous insects,such as beetles (Coleoptera), flies (Diptera) or moths (Lepidoptera), whose transformation from wingless larvae to flying adults occurs by a transitory stage known as the pupa. This type of metamorphosis enabled higher specialization of the feeding larval stage and the exploitation of new niches,and, consequently, contributed to the Holometabola lineage success. How exactly is the switch from larval to pupal to adult developmental programs coordinated is far from being well understood.

Current models propose that two types of lipophilic hormones regulate the entry to metamorphosis. In the absence of the sesquiterpenoid juvenile hormone(JH), ecdysteroids initiate the pupal program, in which larval structures are transformed into imaginal ones and new pupal cuticle is deposited(Riddiford, 1994; Buszczak and Segraves, 2000; Gilbert et al., 2000; Thummel, 2001; Dubrovsky, 2005; Truman and Riddiford, 2007). The role of JH is anti-metamorphic, as in the presence of JH the larva cannot pupate but molts to another larva. Our knowledge about the molecular events underlying metamorphosis is primarily based on the Drosophila model. Unfortunately, fly development has largely lost dependence on JH, as constant exposure to JH cannot prevent entry to the pupal program(Riddiford and Ashburner,1991; Restifo and Wilson,1998). Thus, in response to an elevated ecdysteroid titer, Drosophila larval epidermis dies and the adult head and thorax, with appendages, develop from imaginal discs, while the abdominal epidermis proliferates from histoblasts (Madhavan and Schneiderman, 1977; Bayer et al., 1996b). Only at this later stage does ectopic JH interfere with Drosophila metamorphosis, most notably with the making of the adult abdomen (Postlethwait,1974; Riddiford and Ashburner,1991; Restifo and Wilson,1998; Zhou and Riddiford,2002). By contrast, both ecdysteroid and JH signals are necessary to coordinate metamorphosis in typical holometabolous larvae, whose polymorphic epidermal cells sequentially produce first larval, then pupal and finally adult structures (Riddiford,1994), while internally growing imaginal discs, if any occur, give rise to appendages only, particularly the wings(Svacha, 1992).

Seminal studies in Drosophila have defined the ecdysteroid signaling pathway. The heterodimeric nuclear receptor consisting of EcR and Usp (Thomas et al., 1993; Yao et al., 1993) regulates primary ecdysteroid-response genes that encode transcription factors(Buszczak and Segraves, 2000). Of these transcription factors, E74(Fletcher et al., 1995) and Broad-Complex (BR-C) are specifically required for metamorphosis.(BR-C has been designated as br by FlyBase, but, in keeping with nomenclature commonly used in the literature and to avoid confusion with the BR-C complementation group br, we use the original name.)

The BR-C gene encodes a Broad-Complex-Tramtrack-Bric-a-brac (BTB)domain with one of four alternatively spliced C2H2 zinc-finger motifs Z1-Z4(DiBello et al., 1991; Bayer et al., 1996a). Mutants in the isoform-specific regions form three complementation groups (br,rbp and 2Bc), and display both specific and partly overlapping defects in the differentiation of adult tissues and the death of larval ones;loss of the entire gene in nonpupariating (npr1) mutants blocks metamorphosis completely (Belyaeva et al., 1980; Kiss et al.,1988; Restifo and White,1992; Fletcher and Thummel,1995; Bayer et al.,1997). The function of BR-C is conserved at least in lepidopterans, as BR-C Z4 from Manduca sexta partially rescues the rbp mutants (Bayer et al., 2003), and corresponding metamorphic defects result from BR-C RNAi in the silkmoth Bombyx mori(Uhlirova et al., 2003). However, because Diptera and Lepidoptera represent advanced and related insect orders, it is of interest to examine the role of BR-C in other holometabolans.

BR-C is an attractive target of the as yet poorly characterized JH signaling. Studies in Manduca(Zhou et al., 1998) and Bombyx (Reza et al.,2004) have shown that removal of the JH-producing corpora allata glands causes ectopic BR-C induction as well as precocious pupation,whereas exposure of larvae to JH prevents both BR-C transcription and pupation. BR-C is therefore thought of as the JH-dependent switch between larval and pupal programs (Zhou and Riddiford, 2002; Dubrovsky, 2005), but causal genetic evidence for this idea is missing. In the absence of a bona fide JH receptor, it is also unclear how JH might influence BR-Cexpression.

An excellent model with which to address these problems is the red flour beetle Tribolium castaneum, in which JH exerts its classical anti-metamorphic effect. We have recently shown that perturbed function of the Tribolium ortholog of the Drosophila Methoprene-tolerant(Met) gene causes larvae to metamorphose prematurely, before reaching their final instar (Konopova and Jindra,2007). This phenotype is compatible with Drosophila Metmutation conferring resistance to JH and its mimic methoprene(Wilson and Fabian, 1986; Ashok et al., 1998), and this,together with the high-affinity binding of Met to JH(Shemshedini and Wilson, 1990; Miura et al., 2005), makes Met currently the best candidate for a JH receptor. Interestingly, Metand BR-C mutations interact genetically during Drosophiladevelopment (Wilson et al.,2006).

In this study, we use Tribolium and a neuropteran, lacewing Chrysopa perla, to show that the primary role of BR-C in directing the larval-pupal transition may have been present in primitive holometaboly. We propose that this role is a temporal coordination of tissues during the metamorphic process, which is at least partly achieved by Met-dependent JH regulation of BR-C expression.

Regions encoding the conserved BTB domain and the five zinc-finger domains were identified in the Tribolium genome database(www.bioinformatics.ksu.edu/BeetleBase)by using TBLASTN with Drosophila BR-C sequences. Sequence of the variable core region was obtained by RT-PCR (primers 5′-TCCAGGTGAATAGTGATTACG and 5′-GACTGTCTTTAACCTCGTTC) of the entire open reading frame of the TcBR-C isoform Z4 using total RNA from Tribolium prepupae. Sequences specific to the remaining four isoforms were obtained with a forward primer in the common core region and reverse primers in exons Z1-Z5 (see Table 1). The 5′ and 3′ cDNA ends were amplified by using the GeneRacer Kit (Invitrogen, Carlsbad, CA).

Partial sequence for a putative Chrysopa BR-C Z2 isoform was cloned from late-third (final) instar larvae by using touch-down, nested RT-PCR (Uhlirova et al.,2003), with primers mapping to the BTB and Z2 zinc-finger domains. PCR with a degenerate primer set 5′-AARWSIACICCITGYAARCAYCC and 5′-CKISWRCARTAIACICKYTCRCA was followed by PCR with nested primers 5′-TAYCAYGGIGARGTIAAYGTNCA and 5′-TAYTCYTCYTGICKIWCNGCRTG. BR-C cDNA sequences from both species are available from GenBank(accession numbers EU200752-EU200758).

Wild-type Tribolium castaneum, strain San Bernardino (obtained from G. Bucher, Georg August University, Goettingen) was reared at 32°C and staged as described (Konopova and Jindra, 2007). Under constant conditions, pupation takes place after seven or eight larval instars. Line KS342 carrying a piggyBac insertion within the TcBR-C gene produced by the GEKU screen was kindly provided by S. Brown (Kansas State University,Manhattan, KS). An enhancer-trap line pu11 that marks developing wings (Lorenzen et al., 2003; Tomoyasu and Denell, 2004) was a gift from Y. Tomoyasu (Kansas State University, Manhattan, KS). Chrysopa perla larvae were maintained at 24°C on a culture of live aphids(Acyrthosiphon pisum).

dsRNA of the indicated lengths (see Table 2) was prepared by using T3 and T7 MEGAscript kits (Ambion, Austin, TX). dsRNA concentrated up to 5μg/μl was injected into the abdomen of CO₂ anesthetized Tribolium or Chrysopa larvae as described(Tomoyasu and Denell, 2004; Konopova and Jindra,2007).

To determine the lethal phase in the Tribolium KS342 line, we mated individual males carrying the piggyBac insertion (marked with EGFP) with wild-type females, and then crossed heterozygous KS342/+males and females obtained from their progeny. Almost no lethality was observed in the embryos. The EGFP-positive larvae were reared and scored until adult stage. Whether larvae were homozygous for the KS342 insertion was determined by PCR amplification of crude genomic DNA(Gloor et al., 1993) with primers 5′-GCAAAATTGCATCCGAGAAC and 5′-GTTTGCTTCACCGATATGAC flanking the site of piggyBac insertion.

Early Tribolium prepupae were injected with TcMet or egfp dsRNA and allowed to pupate. These RNAi pupae or intact pupae 0-4 hours after ecdysis were briefly dipped into 0.3 mM methoprene (VUOS,Pardubice, Czech Republic) in acetone, or into acetone alone(Konopova and Jindra, 2007). At the desired stage the pupae were subjected to mRNA expression analysis.

cDNA was prepared from 2 μg of total RNA isolated with TRIzol reagent(Invitrogen, Carlsbad, CA) and treated with DNase (Roche, Mannheim, Germany)as described (Konopova and Jindra,2007). cDNA samples diluted 5-fold with water were subjected to standard PCR reactions with Taq DNA polymerase (Unis, Top-Bio, Prague, Czech Republic); the GC-rich Z3 cDNA was amplified with Phusion DNA polymerase(Finnzymes, Espoo, Finland). For primer sequences and cycle numbers see Table 1.

Samples were fixed in 80% ethanol, postfixed with 1% osmium tetroxide,dehydrated in ethanol, critical-point dried, gold coated, and observed under a JEOL 6300 (Tokyo, Japan) scanning electron microscope.

A putative T. castaneum BR-C ortholog (TcBR-C) has been identified in the Tribolium genome database by the presence of a conserved N-terminal BTB domain and of alternative C-terminal zinc finger DNA-binding domains (Fig. 1A). The BTB domain consists of 115 amino acids of which 90-97% residues are identical to those in BR-C BTB domains from other insects(Fig. 1B). We have isolated multiple transcripts of the gene, including two alternatively spliced 5′untranslated regions and 3′ ends encoding five zinc finger domains alternatively spliced to the common core, comprising the BTB domain and a variable region (Fig. 1A). Four of the TcBR-C zinc finger motifs (Z1-Z4) show homology to those known in other insects (e.g. 84%, 91%, 100% and 93% amino acid identity with Drosophila BR-C Z1, Z2, Z3 and Z4, respectively). Moreover, the order in which the exons encoding the zinc finger domains follow within the genomic DNA (i.e. Z1, Z4, Z2 and Z3; Fig. 1A) is conserved among Tribolium, Drosophila and Bombyx (DiBello et al.,1991; Bayer et al.,1996a; Ijiro et al.,2004; Nishita and Takiya,2004; Reza et al.,2004). Although in these species and in mosquitoes(Chen et al., 2004) no more than four isoforms have been found, an additional exon in Tribolium(Fig. 1A) predicts a fifth zinc-finger domain (Z5), whose sequence does not match any of the known BR-C proteins. All five isoforms produce transcripts with both of the alternatively spliced 5′ untranslated exons. The sequence homology together with the conserved microsynteny of the zinc finger-encoding exons demonstrates that the gene we have isolated is a Tribolium BR-C ortholog.

To identify stages at which TcBR-C is expressed and potentially required, we analyzed cDNA samples prepared from embryos, fifth and eighth(final) instar larvae, pupae and adults. The transcript region common to all TcBR-C isoforms was detected continuously at moderate levels, with a strong peak rising at the onset of the prepupal (i.e. pharate pupal) stage and declining soon after ecdysis in early pupae(Fig. 1C, top row). To see whether there was any stage specificity in the presence or predominance of individual isoform mRNAs, we repeated RT-PCR with primer sets, each of which was able to amplify only one of the five zinc finger-encoding exons. Figure 1C shows that isoforms Z1, Z4 and Z5 were mainly responsible for the sharp rise of total TcBR-C mRNA at 72 hours of the eighth instar, whereas isoforms Z2 and Z3 showed a more gradual increase. Nevertheless, all isoform-specific mRNAs could be detected at all examined stages. These data suggest that TcBR-C may play a role throughout development, although its upregulation during the onset of metamorphosis indicates its requirement for pupal differentiation.

To determine the role of TcBR-C during postembryonic development,we injected Tribolium larvae with dsRNA directed against the variable common region (Fig. 1A) to silence all TcBR-C isoforms simultaneously. When fifth, sixth and final (seventh or eighth) instar larvae were injected, 100% of them(n=100) developed normally and underwent up to three successive larval molts, depending at which instar they had been injected. However, 98%of these animals eventually died during the prepupal stage at the end of the final larval instar (Fig. 2B,B′; see also Table S1 in the supplementary material). The penetrance of this phenotype was reduced to 64% (n=36) when third-instar larvae were injected, perhaps because of a weakening of the RNAi effect.

A prepupa is a pharate pupa in which larval cuticle has been apolysed and the newly deposited cuticle has attained pupal characters. However, removal of the apolysed larval cuticle from the arrested TcBR-C(RNAi) animals revealed recurrence of some larval features and disruption of the normal pupal morphogenesis (Figs 2, 3). These severely affected animals had an overall larval appearance with only rudimentary pupal wings(Fig. 2B′). The same lethal phenotype (Fig. 2C) was observed in 23% (n=260) of progeny produced by crossing beetles heterozygous for a piggyBac insertion KS342 within TcBR-C (Fig. 1A). Similar to the effect of RNAi, the mutant prepupae possessed larval urogomphi and vestigial wings, and they lacked the pupal-specific cuticular structures called gin traps (Fig. 2C-G). As would be expected from the almost 1:3 ratio of dying to normally developing animals, the arrested prepupae were homozygous for the KS342 mutation(see Materials and methods). TcBR-C^KS342 homozygotes had reduced levels of mRNAs encoding isoforms Z2 and Z3(Fig. 2H), whose exons resided farther from the piggyBac insertion than exons Z1 and Z4(Fig. 1A). We speculate that splicing of the Z2 and Z3 mRNA products might be compromised by the insertion,but further analysis is necessary. These results show that BR-C is required for pupal development.

Closer examination revealed that although cuticle in severely affected TcBR-C(RNAi) animals lacked the long larval setae, it remained smooth, without the microsculpture typical for pupae(Fig. 3A, part d; Fig. 3B, part d). The urogomphi retained their larval shape (Fig. 3A) and the gin traps were missing(Fig. 3C) in strongly affected TcBR-C(RNAi) prepupae. Other structures appeared to be less dependent on normal TcBR-C function and continued the pupal or adult program in arrested animals. For instance, distal abdominal segments differentiated pupal characters such as the genital papillae, but abnormalities were clearly visible (Fig. 3A, part d). The compound eyes developed several rows of ommatidia, although these were not evenly spaced as in normal pupae (Fig. 3D; Fig. 2D,E). Although the antennae lacked sensillae, they apparently developed towards the adult fate, with their club shape and clear separation of segments(Fig. 3E). Finally, the legs of TcBR-C(RNAi) animals lost the larval character as they possessed the double claws that normally develop in pupae and are typical for the adult leg(Fig. 3F). Slightly accelerated development was suggested by more distinctly separated tarsal segments and sharper claws in TcBR-C(RNAi) animals relative to normal pupae(Fig. 3F).

For a better insight into how TcBR-C effects the larval-pupal transition, we elicited a milder phenotypic response by diluting the dsRNA up to a 1000-fold. These lowered dsRNA doses were still lethal but allowed prepupae to ecdyse. All such treated animals had noticeably shortened and blistered wings and legs relative to control pupae(Fig. 2A,I). Compared with strong RNAi phenotypes, some pupal characters became more prominent. The animals developed gin traps, albeit aberrant(Fig. 2I; Fig. 3C, part e), their urogomphi were elongated, genital segments were nearly perfectly differentiated (Fig. 3A, part e), and the cuticle surface had a pupal-like microsculpture(Fig. 3B, part e). Except for the absence of sensory bristles, the antennae resembled the adult ones(Fig. 3E, part e). Conversely,legs with less prominent claws and segment borders suggested a weaker acceleration of the adult program than that observed with the strong RNAi effect (Fig. 3F).

In Drosophila, some functions are shared by all BR-C proteins,whereas others are fulfilled by a specific isoform(Bayer et al., 1997). To see whether any unique functions apply to Tribolium BR-C isoforms, we injected larvae with dsRNA against each of the five zinc finger domains separately. Probably due to the limited sequence lengths, transcript levels of the targeted isoforms declined only partially(Fig. 4A). All larvae treated for a single isoform ecdysed into pupae displaying a degree of aberrancies increasing in the order of isoforms: Z5<Z1<Z4<Z3<Z2, with the most visible effect being the shortening of the wings and legs(Fig. 4; see also Fig. S1 in the supplementary material). These pupae developed into adults that either eclosed normally or died unable to ecdyse. Targeting of BR-C Z5 had no obvious effect. Interestingly, a simultaneous knockdown of Z1 and Z2 isoforms enhanced the short wing and leg phenotype, such that it resembled the mild effect of the common-region RNAi, including imperfect development of the gin traps(compare Fig. 2I with Fig. 4D). Thus, Z1 and Z2 might together be indispensable for wing elongation. Similarly, the apparent loss of Z2 and Z3 mRNAs in the TcBR-C^KS342 mutants could not be compensated for by the remaining BR-C products(Fig. 2H). Taken together,these results suggest at least some specific roles for the TcBR-C isoforms.

Juvenile hormone (JH) is an anti-metamorphic signal that can block pupation; its removal causes precocious pupal development(Bounhiol, 1938; Wigglesworth, 1954; Nijhout, 1994). Precocious pupation of early-stage Tribolium larvae can be achieved by silencing of the Methoprene-tolerant gene (TcMet), which is necessary for JH signaling (Konopova and Jindra,2007). If TcBR-C is indeed essential for the pupal program as suggested by our results, then TcBR-C mRNA should be upregulated in such untimely TcMet(RNAi) prepupae. To test this hypothesis, we injected early-fifth instar larvae with TcMet dsRNA and examined TcBR-C expression in resulting sixth-instar precocious prepupae. Figure 5 shows that the TcBR-C mRNA level in these TcMet(RNAi) prepupae was high relative to in control early-seventh instar larvae. The increase was comparable to that normally seen during the onset of metamorphosis(Fig. 5 and Fig. 1C). This result links precocious pupal development that ensues from a deficiency in JH signaling with upregulation of TcBR-C. It also supports the premise that TcBR-C is required for pupation, even if premature.

Similar to pupae of other insects, Tribolium pupae respond to exogenous JH or its mimic methoprene by deposition of a second pupal instead of adult cuticle (Konopova and Jindra,2007). In Manduca and Drosophila this ectopic JH treatment induces BR-C expression(Zhou and Riddiford, 2002). We have used this effect to provide evidence that TcBR-C acts in JH signaling downstream of TcMet. First, we established that methoprene treatment indeed caused overexpression of TcBR-C in pupae. Figure 6A shows that, whereas in controls TcBR-C mRNA declines to its basal level after day 1 of normal pupal development, it is highly abundant throughout the pupal stage after methoprene application. As was shown previously(Konopova and Jindra, 2007),these methoprene-treated animals died as second-stage pupae.

As the above effect of methoprene can be averted by the silencing of TcMet (Konopova and Jindra,2007), we next checked whether TcMet RNAi also prevented the upregulation of TcBR-C. Early prepupae were injected either with TcMet or with control dsRNA and, after pupation, the animals were treated with methoprene. Pupae aged 24, 48, 72 and 96 hours were then subjected to RT-PCR. As shown by the 48-hour- and 96-hour-old pupae(Fig. 6B), TcMetdepletion specifically prevented the induction of TcBR-C by methoprene. Consistent with our previous data(Konopova and Jindra, 2007),the methoprene-treated TcMet(RNAi) pupae produced adult beetles. This experiment demonstrates that TcMet is required for JH-induced upregulation of TcBR-C mRNA in Tribolium pupae. It also suggests that TcBR-C is a target of TcMet during JH-induced formation of the ectopic pupal stage.

To explore whether the essential role of BR-C in pupal development might be common to other orders with less derived holometaboly, we chose to study the lacewing Chrysopa perla (Neuroptera). Chrysopadevelops via three larval instars that are, unlike in Tribolium,easily discernible by distinct cuticle pigmentation and sensillation. Before pupation, Chrysopa larvae spin a cocoon from Malpighian tubules. We have isolated a cDNA fragment of a putative Chrysopa perla BR-Cortholog, hereafter referred to as CpBR-C. At its N terminus, this sequence shows 90% amino acid identity with the last 23 residues of the BR-C BTB domain; the C-terminal 23 amino acids match the Drosophila BR-C Z2 zinc-finger domain with 82% identity. The region between the two conserved domains shows little homology. Like in Tribolium, moderate CpBR-C mRNA levels were detected throughout embryogenesis (data not shown) and all three larval instars, with a marked expression peak before cocoon spinning at the end of final larval instar (see Fig. S2 in the supplementary material).

When first- and second-instar Chrysopa larvae were injected with CpBR-C dsRNA, no developmental defects were observed until the larval-pupal transition. Then, 95% (n=146) of the injected larvae arrested after the third instar at the prepupal stage, and 41% of them failed to complete or even initiate spinning their cocoons (see Table S2 in the supplementary material). The CpBR-C(RNAi) animals looked like imperfect pupae with very short wings and smaller compound eyes(Fig. 7). Although their cuticle pigmentation resembled that of pupae and lacked the long larval setae(Fig. 7), the cuticle showed typical larval thorns and was neither smooth, as in normal pupae, nor did it carry the long bristles seen in adults(Fig. 8A-D). The tarsi in CpBR-C(RNAi) animals became segmented as in pupae and adults, but pretarsi retained the larval character: compared with the pupal leg, they were narrow with hooked claws and ended with a long arolium similar to that found in larvae (Fig. 8E-H). In wild-type pupae, the long antennae pointed dorsally and were coiled on lateral sides (Fig. 7B), whereas in CpBR-C(RNAi) animals they were directed proximally and were twisted above the pupa-like mouthparts (Fig. 7F). Interestingly, differentiation of the compound eye in CpBR-C(RNAi) prepupae was more advanced relative to the smooth eye of control pupae, because it showed development of ommatidial lenses, although defective (Fig. 8I-L,J′-L′). Such premature eye differentiation probably resulted in holes instead of normal lenses(Fig. 8L). These results show that in Chrysopa, like in Tribolium, BR-C is required for pupation and plays important roles in the morphogenesis of certain pupal characters.

In both Tribolium and Chrysopa, BR-C RNAi compromised the larval-pupal transition without affecting earlier development, regardless of the time of dsRNA injection. The TcBR-C^KS342 homozygotes died at the same stage. These data suggested that the moderate levels of BR-C mRNAs, detectable during premetamorphic stages in both species,had no essential role. This scenario would agree with the fact that zygotic BR-C function is not required in Drosophila BR-C null nonpupariating mutants until the onset of metamorphosis(Belyaeva et al., 1980; Kiss et al., 1988), and with recent data on Tribolium (Suzuki et al., 2008). However, as neither RNAi nor the likely hypomorphic TcBR-C^KS342 allele present a complete loss-of-function situation, we cannot exclude a possibility that BR-C plays some additional role, not visualized by our phenotypes. Importantly, the lethal phase correlates with a strong upregulation of BR-C expression. At least in beetles, this stage coincides with a peak of ecdysteroid titer that causes larvae to initiate prepupal development(Hirashima et al., 1995; Aribi et al., 1997).

In contrast to Drosophila npr1 mutants, metamorphosis was not completely blocked by BR-C deficiency in Tribolium or Chrysopa. Instead the arrested prepupae showed a blend of larval,pupal, and partially even adult features. Based on the absence of the pupal-specific gin traps in Tribolium and on the surface microsculpture, the cuticle was apparently larval in both species, thus confirming the requirement of BR-C for the pupal commitment of the epidermis (Zhou et al., 1998; Zhou and Riddiford, 2002). Interestingly, although the thorny cuticle in Chrysopa BR-C(RNAi)animals was distinctly larval, similar to in Tribolium, the body pigmentation resembled that of pupae. We cannot be sure whether this mixed character of the epidermis might be due to persisting CpBR-Cfunction, or might be because CpBR-C is not necessary for the pupal pigmentation.

Pupal characters in BR-C(RNAi) animals included rudimentary wings. In particular, the weak phenotypes in Tribolium (produced with either isoform-specific or diluted common-core dsRNAs) revealed that wing elongation was highly sensitive to BR-C depletion. A similar effect of BR-C RNAi was described for pupal appendages in Bombyx mori(Uhlirova et al., 2003). BR-C silencing prevented the gradual wing enlargement even in larvae of the hemimetabolous milkweed bug Oncopeltus fasciatus(Erezyilmaz et al., 2006). Imaginal discs fail to elongate in Drosophila br mutants with disrupted BR-C Z2 function (Kiss et al.,1988; DiBello et al.,1991; Bayer et al.,1997). The short legs and wings are not due to insufficient proliferation of the disc cells but are due to their inability to change shape in response to the ecdysteroid (von Kalm et al., 1995). This cell shape change requires cytoskeletal components whose mutations enhance the effect of br(Gotwals and Fristrom, 1991; Ward et al., 2003). The rudimentary wings, present even in animals most severely affected by TcBR-C^KS342 mutation or by RNAi, suggest that cell shape changes, rather than cell proliferation may be disrupted by the loss of BR-C in Tribolium as well. Growing wings marked by EGFP in arrested beetle prepupae (see Fig. S3 in the supplementary material) support this idea. The legs in Tribolium BR-C(RNAi) animals were short also but were distally specified as pupal with two tarsal claws. By contrast, the arrested Chrysopa prepupae retained pretarsi with the larval-specific elongated arolium, thus suggesting a stronger requirement for BR-Cfunction in the Chrysopa leg.

Except for small deviations, gross morphology of Tribolium genital segments with the pupal genital papillae was pupal in BR-C(RNAi)animals (see Fig. 3A). In addition, the larval-pupal transformation of the visual system was initiated,as larval stemmata were replaced with ommatidia of the compound eyes. However,as in Drosophila (Brennan et al.,2001), TcBR-C was important for compound eye differentiation. These observations suggested that not all aspects of pupal development were completely blocked by BR-C depletion.

While the above described structures were retarded in their development in BR-C(RNAi) animals, others appeared accelerated in their development towards the adult state, although none could be unambiguously defined as adult. For instance, the antennae in Tribolium or the compound eyes in Chrysopa resembled their adult counterparts, but in fact were intermediates between pupal and adult organs. These heterochronic phenotypes suggest that BR-C may not only be a pupal specifier(Zhou and Riddiford, 2002),but rather a temporal coordinator of the extensive morphogenesis in diverse tissues during metamorphosis.

Drosophila organs require a temporally regulated balance between both inductive and repressive BR-C functions, represented by the individual isoforms (Karim et al., 1993; Crossgrove et al., 1996; Mugat et al., 2000). We therefore see two alternative explanations for the heterochronically advanced phenotypes. First, these structures may require BR-C to repress precocious adult morphogenesis in them, but the inductive BR-C function is dispensable for development beyond larval state. Consequently, loss of BR-C accelerates their development. Second, if both functions are required but the repressive one is more sensitive to reduced BR-C dose, then the inductive function will prevail under an incomplete BR-C knockdown. We favor the first alternative,because progression beyond the pupal stage seems to depend on BR-Cdownregulation (Zhou and Riddiford,2002) (this work).

Periods of JH absence are required first in larvae to initiate the pupal program, and later in pupae to exit it. BR-C in both cases promotes the pupal fate (Zhou and Riddiford,2002), and therefore JH must regulate BR-C differently in larvae and in pupae. In lepidopteran (Zhou et al., 1998; Reza et al.,2004), as well as in Tribolium(Suzuki et al., 2008) larvae,JH prevents BR-C expression until the onset of metamorphosis, and presumably that is how JH prevents pupal differentiation. Conversely, removal of the JH source (allatectomy) causes both BR-C misexpression and precocious pupal development. In pupae, ectopic JH induces BR-C(Zhou et al., 1998; Zhou and Riddiford, 2002; Reza et al., 2004; Wu et al., 2006), and in many insects, including Tribolium(Konopova and Jindra, 2007),such JH application causes reiteration of the pupal stage. In Drosophila,BR-C misexpression alone is sufficient to inhibit adult cuticle formation(Zhou and Riddiford, 2002). BR-C is therefore a prime target of JH signaling, but how JH regulates BR-C expression is unknown.

We showed here that precocious pupation, triggered by interference with the putative JH receptor Met, coincided with precocious TcBR-C mRNA increase in the sixth instar. Thus, disrupted JH signaling induced TcBR-C similarly to allatectomy in lepidopteran larvae(Zhou et al,. 1998; Reza et al., 2004). As expected, TcBR-C not only marked but also was necessary for the untimely pupation, as TcMet; TcBR-C double-RNAi resulted in a phenotype similar to TcBR-C RNAi alone, i.e. entry to a lethal prepupal stage, except one or two instars too early (data not shown). Therefore, although the metamorphic program could be prematurely induced by silencing of TcMet, it could not be completed without TcBR-C. However, loss of Met has been shown to worsen the effect of BR-C mutations in Drosophila, without altering BR-C expression (Wilson et al.,2006). This again might reflect the different response to JH in the fly.

The evidence that TcMet is required for regulation of TcBR-C came from pupae, where the JH mimic methoprene induced TcBR-C mRNA, but not after TcMet knockdown. This result places TcBR-C downstream of TcMet in JH signaling. Importantly, the averting of ectopic TcBR-C expression by TcMet RNAi also rescued the methoprene-treated animals from repeating the pupal stage and allowed them to become adult(Konopova and Jindra, 2007). Together, these findings suggest that, similar to in Drosophila(Zhou and Riddiford, 2002),downregulation of BR-C is required to exit the pupal state in Tribolium.

We propose the following model for BR-C function in holometabolan metamorphosis (Fig. 9). In larvae, JH acts through Met to prevent BR-C induction until the final instar, when JH decline relieves the repression, and BR-C coordinates pupal morphogenesis. Loss of BR-C function causes both retardation and acceleration of development in diverse epidermal tissues, thus producing a mix of larval-,pupal- and adult-like features. In early pupae, low JH titer normally allows BR-C expression to drop, which is necessary for proper adult differentiation. Exogenous JH, again acting via Met, causes BR-Cmisexpression, which in turn promotes another round of pupal, instead of adult, development. Whether Met regulates BR-C expression directly,and what determines whether BR-C will be repressed or activated requires further work.

We are grateful to Y. Tomoyasu and G. Bucher for training us in beetle rearing and RNAi. We acknowledge receiving the KS342 Tribolium line that was generated in the USDA-funded GEKU mutagenesis screen at the Kansas State University from S. Brown, R. Denell and D. Beeman. We thank T. Kimzey and T. Shippy for molecular characterization of the KS342 mutant; Z. Ruzicka for Chrysopa; the Electron Microscopy Lab for help with sample preparation; P. Svacha and M. Friedrich for advice on morphology; and L. M. Riddiford for sharing unpublished information. This work was supported by Projects IAA5007305 and Z50070508 from the Czech Academy of Sciences, and LC07032 from the Czech Ministry of Education. B.K. was supported by Project 204/07/1032 from the Czech Science Foundation.