Common and Distinct Roles of Juvenile Hormone Signaling Genes in Metamorphosis of Holometabolous and Hemimetabolous Insects

1Winged insects have evolved diverse modes of metamorphosis. Hemimetabolous insects such as grasshoppers, true bugs or cockroaches develop from larvae (also called nymphs) that resemble adults, possess externally growing wing pads, and metamorphose during the final molt by acquiring perfect wings and genitalia. In contrast, larvae of holometabolous insects including flies, butterflies or beetles can differ dramatically from the adults. They undergo a two-stage “complete” metamorphosis (holometaboly), first forming an intermediate called the pupa before changing into a winged adult. Holometaboly has evolved from hemimetaboly and, judging by the number of known species, has become the most successful developmental strategy on land.

2In both hemimetabolous and holometabolous insects, the developmental switch between juvenile and adult forms depends on juvenile hormone (JH), a sesquiterpenoid produced by the corpora allata gland. The presence of JH in pre-final larval instars ensures that the next molt, promoted by ecdysteroids, produces another, only a larger larva,. At an appropriate stage, a natural drop of JH secretion permits metamorphosis. Experimental removal of JH at earlier times activates the metamorphic program prematurely, whereas supply of ectopic JH to final-instar larvae or pupae causes repetition of larval or pupal instars, respectively –.

3Although the anti-metamorphic effect of JH was discovered in the hemimetabolous true bug, Rhodnius prolixus,,, our knowledge on the molecular mode of JH action almost exclusively derives from studies in holometabolans. JH signals through its putative intracellular receptor, the bHLH-PAS protein Methoprene-tolerant (Met), originally identified in the fruit fly Drosophila melanogaster,. In the red flour beetle, Tribolium castaneum, loss of Met triggers pupation of larvae during pre-final instars – a classic precocious metamorphosis phenotype caused by deficiency of JH itself. In response to JH, Met regulates expression of transcription factor genes Krüppel-homolog 1 (Kr-h1) and Broad-Complex (BR-C) –, and loss of Kr-h1 also elicits precocious metamorphosis of beetle larvae. BR-C is dispensable in holometabolous larvae until the onset of metamorphosis, when it specifies pupal features, –. Upon pupation both BR-C and Kr-h1 are naturally down-regulated by the absence of JH to allow adult development –,,. Of the three JH-signaling genes, BR-C has been functionally studied in hemimetabolous insects, where, unlike in holometabolans, it is required for development of the embryonic germ band, and for anisometric growth of the larval wing pads.

4To provide a direct comparison of JH signaling in holometaboly and hemimetaboly, we have examined the function of Met, Kr-h1 and BR-C in the hemimetabolous firebug, Pyrrhocoris apterus (true bugs, Hemiptera). We show that despite the diverse developmental strategies and the vast evolutionary distance between them, transduction of the anti-metamorphic JH signal relies on the common-core elements, Met and Kr-h1. In both insect types, Kr-h1 acts as a strictly JH- and Met-dependent repressor of metamorphosis. In contrast, the function of BR-C has changed from promoting progressive development of hemimetabolous larvae to a new role in specifying the holometabolous pupa.

5As the first step towards functional comparison between holometaboly and hemimetaboly, we have isolated cDNAs encoding the putative JH receptor Met (JN416984) and its target genes Kr-h1 (JN416987) and BR-C (JN416990), from Pyrrhocoris apterus, and Met (JN416985) and Kr-h1 (JN416988) cDNAs from another true bug, Rhodnius prolixus. Alignments amongst the orthologs reveal conservation of the main functional domains (Fig. S1), namely the basic helix-loop-helix (bHLH) region and two Per-Arnt-Sim (PAS) domains in Met, eight zinc-finger motifs in Kr-h1, and a Broad-Tramtrac-Bric-a-brac (BTB) domain followed by one of the alternative zinc-finger isoforms (Z2) in BR-C. Therefore, the three JH signaling genes are common to insect orders developing through holometaboly and hemimetaboly. In fact, their conservation predates the origin of metamorphosis, as we have found Met (JN416986), Kr-h1 (JN416989) and BR-C (JN416991) orthologs in the firebrat, Thermobia domestica, a representative of the primitive lineage of non-metamorphosing wingless insects (Fig. S1).

6The firebug invariantly undergoes four larval molts demarcating five larval instars (L1–L5), followed by a metamorphic molt, which produces an adult possessing external genitals and wings with a specific color pattern. Maintenance of the larval state in Pyrrhocoris requires JH, as has been demonstrated by removal of the JH-producing corpora allata gland (allatectomy) from penultimate (L4) larvae. Conversely, metamorphosis is permitted by a natural decline in JH titer during the last larval instar (L5), as supplying L5 larvae with JH mimics induces an extra larval stage (reference and this work).

7We examined how expression of Met, Kr-h1 and BR-C during Pyrrhocoris development correlates with this regulation by JH. From embryogenesis to adulthood, Met mRNA persisted without major fluctuations through all larval stages (Fig. 1). Such a constitutive presence agrees with the presumed JH receptor role of Met and with the ability of L5 larvae that naturally lack JH to respond to exogenous JH mimics by forming an extra larval instar. Temporal profiles of Kr-h1 and BR-C were more dynamic. Both transcripts reached their highest levels during the second half of embryogenesis, then BR-C oscillated throughout larval development, whereas Kr-h1 was not detected in last-instar larvae, L5 (Fig. 1A). A more detailed profile revealed that upon ecdysis to L5, BR-C mRNA decreased but remained expressed to adulthood, while Kr-h1 expression plummeted from its L4 level by more than 50-fold and was virtually undetectable for the first six days of L5 until the next rise, likely induced by a new surge of JH in pharate adults (Fig. 1B). Allatectomy of adult Pyrrhocoris females drastically reduced Kr-h1 mRNA (494-fold, n = 3), confirming that Kr-h1 transcription absolutely requires the natural source of JH. The expression pattern suggested that Kr-h1 prevents metamorphosis until the final larval instar, when the Kr-h1-free (and JH-free) period may be necessary to initiate metamorphosis.

8To test whether Met and Kr-h1 are indeed required to maintain the juvenile character of hemimetabolous larvae, we utilized systemic RNAi in Pyrrhocoris. Injection of Met dsRNA into early-L4 larvae caused an 86% knockdown of Met mRNA (Fig. 2B), followed by precocious appearance of adult attributes after the ensuing molt, which would have otherwise produced an L5 larva (Fig. 2C and Table 1). Despite their smaller body size compared to normal adults, Met(RNAi) males had external genitals (Fig. 2C). Instead of the solid black and fully attached L5 wing pads, these adultoids developed small but movable, articulated wings, liberated from the scutellum and the tergites. The black melanin pigment disappeared from specific regions of these wings to leave black spots and edges on the red background, a pattern only seen in adults (Fig. 2C). On the thorax, the notum expanded posteriorly and also displayed an adult-specific color pattern. Like in adults, abdominal cuticle tanned with melanin. We did not determine whether particular regions of the cuticle, produced by individual epidermal cells, were purely adult or whether they had a mixed larval-adult identity, so this intriguing possibility remains. Unable to molt again, all L5 adultoids were terminally arrested.

9Kr-h1 mRNA levels were reduced by 90% in L4 larvae subjected to Met RNAi (Fig. 2B), demonstrating that Kr-h1 in Pyrrhocoris was a Met-dependent target gene as in Tribolium,. This reduction paralleled the natural fall of Kr-h1 expression upon ecdysis to L5 (Fig. 1B), suggesting that the normal function of Met during the L4 stage is to respond to endogenous JH by maintaining high expression of Kr-h1, which in turn prevents adult development. Consistent with this prediction, larvae injected with Kr-h1 dsRNA at the L4 stage formed L5 adultoids similar to Met(RNAi) animals (Fig. 2C and Table 1). Occasionally their wings grew larger than upon Met RNAi, and some individuals only showed part of the adult color pattern (Fig. 2C). Clearly, the loss-of-function phenotypes of both genes were in good concert and demonstrated precocious metamorphosis of Pyrrhocoris larvae.

10To verify the function of Kr-h1 in an independent system, we performed RNAi in the blood-sucking bug Rhodnius, the very model in which juvenile hormone had been postulated nearly eight decades ago,. Upon molting to the L4 penultimate instar, all Rhodnius larvae (n = 20) injected with Kr-h1 dsRNA at the previous L3 instar showed accelerated growth of wings and genitals (Fig. 3), both typical hallmarks of adult morphogenesis.

11In contrast to Met and Kr-h1, injection of BR-C dsRNA into Pyrrhocoris L4 larvae produced no signs of premature adult development (Fig. 2C and Table 1). BR-C RNAi administered early during the L4 instar reduced BR-C mRNA levels to 17.7±2.0% (n = 5 animals) and caused either lethality at the end of the L4 instar or compromised growth of the wing pads in animals that successfully molted to the L5 instar (Table 1). A similar wing defect was observed after ecdysis to the L4 instar in 100% of animals (n = 22) that had been given BR-C dsRNA as L3 larvae. The retarded wing growth was previously shown for BR-C RNAi in the milkweed bug, Oncopeltus fasciatus. When compared to Kr-h1, BR-C mRNA expression was neither as strongly dependent on Met at the L4 instar (Fig. 2B) nor did it completely cease upon ecdysis to L5 (Fig. 1B). Therefore, Kr-h1 but not BR-C functions as a JH- and Met-dependent repressor of hemimetabolous metamorphosis.

12When treated with JH mimics, final instar Pyrrhocoris larvae fail to metamorphose to adults and instead repeat the larval program in the succeeding supernumerary “L6” instar. We achieved the same effect by topical treatment of early-L5 larvae with the JH mimic methoprene (Fig. 4A,B) in 100% (n = 18) of animals. In this background we then tested whether Met and Kr-h1 mediated the response to the exogenous JH activity.

13When Met was silenced by dsRNA injection at the end of the L4 instar, the following application of methoprene no longer perturbed adult development in 96% (n = 23) of treated animals, and these rescued adults were externally indistinguishable from controls receiving no methoprene (Fig. 4A,C). The “status-quo” effect of methoprene was likewise prevented by Kr-h1 RNAi (Fig. 4D) in 12 out of 13 animals. Therefore, both Met and its target Kr-h1 were necessary for the capacity of methoprene to block metamorphosis and induce the L6 larva. In addition, neither Met nor Kr-h1 appeared to be required for the normal L5-to-adult transition, since dsRNA injection to late-L4 larvae, as opposed to RNAi applied early during the L4 stage, did not interfere with metamorphosis. This result corresponded with the natural absence of Kr-h1 expression during most of the final larval instar (Fig. 1B).

14The above data indicated that the Met-dependent reiteration of the larval program, induced by methoprene, occurred via ectopic transcriptional re-activation of Kr-h1 during the L5 instar. Indeed, levels of Kr-h1 mRNA in the epidermis (or in the whole body, data not shown) of mid-L5 larvae, which had been treated with methoprene as L4, exceeded levels normally observed at the L5 instar by 125-fold (Fig. 4E). This induction was the opposite to the natural drop of Kr-h1 expression upon L4 to L5 ecdysis (Fig. 1B), and it was abolished by Met RNAi (Fig. 4E). By contrast, BR-C mRNA was only induced 4.5-fold with methoprene, again in a Met-dependent manner (Fig. 4E). The striking difference in the fold mRNA induction between Kr-h1 and BR-C did not result from difference in the mRNA levels induced by methoprene but from the extremely low expression of Kr-h1 in mid-L5 larvae (Fig. 1).

15The likely JH receptor Met and its targets Kr-h1 and BR-C play key roles during holometabolous insect metamorphosis. Our present data afford a direct comparison with how the three genes function in the ancestral hemimetabolous development (Fig. 5). We provide three lines of evidence that the JH-free and Kr-h1-free period in the final larval instar (L5) of the Pyrrhocoris bug is critical for adult transition: (i) administration of methoprene to L5 larvae induces Met-dependent Kr-h1 expression and an extra larval stage, (ii) this phenotype is averted if Met or Kr-h1 are silenced prior to methoprene treatment, and (iii) premature suppression of Kr-h1, either direct or through depletion of Met, triggers precocious adult development. Therefore, the JH/Met-dependent Kr-h1 activity ensures the larval program, and only when JH disappears from the blood in the last larval instar, transcription of Kr-h1 ceases for six days to create an opportunity for adult transition (Figs. 1B and 5). Interestingly, the same-length time window for adult commitment was previously defined as a methoprene-sensitive period during the final larval instar of the Rhodnius bug. Kr-h1 now provides a molecular determinant of that window.

16Similar regulation applies to holometaboly. In pre-terminal instars of Tribolium larvae, Met and Kr-h1 respond to JH by blocking precocious metamorphosis,. By mid-final larval instar, expression of Kr-h1 declines to reappear at the pupal molt, this time together with BR-C mRNA (Fig. 5). While the transient down-regulation of Kr-h1 permits partial metamorphosis (pupation), its co-expression with BR-C likely ensures that development does not go too far. This is suggested by appearance of not only pupal but also of adult features in Tribolium larvae subjected to Kr-h1 or BR-C RNAi,,,. After the pupal program has been installed, a JH-free period ensures that both Kr-h1 and BR-C, acting downstream of Kr-h1 in this context,, are shut down in order for adult morphogenesis to take place (Fig. 5). Giving exogenous JH or its mimics to pupae will re-activate both genes and block adult development –,.

17While the function of Kr-h1 as a JH-induced repressor of adult morphogenesis is clearly a common trait of holometaboly and hemimetaboly, the role of BR-C is not. Studies on the hemimetabolan Oncopeltus bug have revealed BR-C requirement during embryogenesis and for the anisometric growth of larval wing pads but no BR-C expression or function connected with metamorphosis,. Conversely, BR-C is essential for pupal development but not at earlier stages in representatives of four holometabolan insect orders, –. The delay of BR-C activity until the pupal stage in Holometabola has been ascribed to an early surge of JH during embryogenesis, which is thought to preserve the seemingly undeveloped (“embryonic”) nature of holometabolous larvae, –. However, while BR-C is necessary for hemimetabolous embryogenesis,, its function has not been causally linked with JH in insect embryos.

18Unlike in Oncopeltus, BR-C expression continues, albeit at a lower rate, throughout the last larval instar of Pyrrhocoris or another hemimetabolan, the cockroach Blattella germanica, in the absence of JH (Fig. 5). Consistent with this pattern, our data show that compared to Kr-h1, expression of BR-C much less depends on JH and that in contrast to Kr-h1 or Met, removal of BR-C cannot accelerate metamorphosis in bug larvae. The changed need for BR-C function from hemimetabolous embryos and larvae to holometabolous metamorphosis suggests that during the evolution of holometaboly, BR-C has been recruited for the new function in specifying the pupal state.

19Kr-h1 is intimately regulated by JH and Met to safeguard juveniles of both hemimetabolous and holometabolous insects against precocious and hence fatal metamorphic changes. Therefore, regardless of the disparate life histories, insects undergoing both types of metamorphosis use a common signaling pathway to commit to adult development. The parallel timing of the critical down-regulation of Kr-h1 in the final-instar Pyrrhocoris larva and in the holometabolous pupa (Fig. 5) supports the view that these stages represent ontogenetically homologous units,, rather than hypotheses building on the assumption that the pupa has originated via compression of all hemimetabolous larval instars into one,,.

20Pyrrhocoris apterus (short-winged form) was maintained at 25°C and a photoperiod of 18 h light to 6 h dark, on dry linden seeds and was supplemented with water. Eggs were collected daily and larvae of particular instars were identified based on the size of the body and wing pads; staging within instars relied on measuring time after ecdysis.

21Partial sequences for Pyrrhocoris Met, Kr-h1 and BR-C genes were isolated by using touch-down nested RT-PCR with degenerate primers (Table S1), mapping to conserved domains. cDNA ends of selected genes were amplified with the GeneRacer Kit (Invitrogen, Carlsbad, CA).

22Total RNA was isolated from Pyrrhocoris embryos, whole larvae or abdominal epidermis with the TRIzol reagent (Invitrogen, Carlsbad, CA). After TURBO DNase (Ambion, Austin, TX) treatment, 2 µg of RNA were used for cDNA synthesis with Superscript II reverse transcriptase (Invitrogen). Relative transcript levels were measured by quantitative RT-PCR using the iQ SYBR Green Supermix kit and the C1000 Thermal Cycler (both from Bio-Rad Laboratories, Hercules, CA). All data were normalized to the relative levels of ribosomal protein (Rp49) mRNA as described. Primer sequences used for qRT-PCR are listed in Table S2.

23dsRNAs comprising 952 bp (Met), 844 bp (Kr-h1) and 1026 bp (BR-C) of the Pyrrhocoris cDNA sequences and control dsRNAs encoding the EGFP and MalE proteins (720 bp and 901 bp, respectively) were synthesized by using the T3 and T7 MEGAscript kit (Ambion, Austin, TX). Approximately 2–5 µg of dsRNA (depending on the size of larvae) were injected into the abdomen of CO2-anesthesized bugs. For JH mimic treatment, late-L4 stage Pyrrhocoris larvae were injected with dsRNA and within 2–3 hours after ecdysis to the L5 instar, a 4-µl drop of acetone-diluted 0.3 mM methoprene (VUOS, Pardubice, Czech Republic) or acetone alone (control) was applied on their dorsal side.