Divergent mechanisms for regulating growth and development after imaginal disc damage in the tobacco hornworm, Manduca sexta

Holometabolous insects have been able to radiate to vast ecological niches as adults owing to the evolution of adult-specific structures such as wings, antennae and eyes. These structures arise from imaginal discs – progenitor organs consisting of clusters of epithelial cells at the mesoderm – which differentiate into specific adult structures. In Drosophila melanogaster, these discs have been shown to have remarkable regenerative capacity after damage (Bryant, 1971; Bryant and Fraser, 1988; Halme et al., 2010; Schubiger, 1971; Simpson et al., 1980). A variety of damage and ablation experiments have demonstrated the importance of these discs to the coupling of growth and metamorphosis when injury results in a developmental delay. These delays presumably provide sufficient time for tissue regeneration. Injury to imaginal discs has also been shown to delay development in a variety of holometabolous insects, whether through genetic tissue ablation (Colombani et al., 2012; Hackney et al., 2012), surgical ablation or autotomy (Kunkel, 1977; O'Farrell and Stock, 1953, 1954; Stock and O'Farrell, 1954), or X-ray-induced selective damage (Ely and Jungreis, 1977; O'Brien and Wolf, 1964; Poodry and Woods, 1990; Stieper et al., 2008). A delay to allow growth of appropriate adult proportions and healthy adult organs would ensure the success of holometabolous insects in their new niches.

Development in holometabolous insects depends on tight coordination between three major hormones: juvenile hormone (JH), prothoracicotropic hormone (PTTH) and ecdysone. PTTH and ecdysone drive the larval molts in the presence of JH. During the last larval stage, cessation and clearing of JH is required for the release of PTTH and, subsequently ecdysone, to allow pupation to occur (Nijhout, 2015; Nijhout and Williams, 1974b). Timing of JH cessation during the last larval stage has been linked to a nutrition-dependent developmental milestone called critical weight (Nijhout and Williams, 1974b). The critical weight, defined as the mass at which starvation no longer affects developmental timing under normal conditions (Nijhout and Williams, 1974a), delineates the end of nutrition-dependent development and marks the beginning of the terminal growth period prior to wandering and pupation (Fig. 1). At this point, lack of nutrition can affect growth, but does not affect developmental timing. Altogether, these physiological markers have been used as tools to understand how stress affects the growth and development of an individual (Davidowitz et al., 2016).

Disturbance of imaginal disc homeostasis, by means of damage, is a form of stress that has been suggested to disrupt hormone regulation of developmental timing and growth. In D. melanogaster, X-ray-induced imaginal disc damage delays pupariation by slowing the larval growth rate (Stieper et al., 2008) and shifting the pre-pupal ecdysteroid peak (Hackney et al., 2012). This change in growth rate is also reflected as an increased critical weight in response to imaginal disc damage (Stieper et al., 2008). The increase in critical weight results in larvae committing to pupation at a higher mass but, coupled with a slowed growth rate during the terminal growth phase, ensures that adult mass is not affected. The delays are suggested to be due to the secretion of Dilp8 – a peptide in D. melanogaster named after the insulin-like peptides – from the damaged wing imaginal discs (Colombani et al., 2012; Garelli et al., 2012), as well as a putative retinoid-like factor (Halme et al., 2010). Dilp8 appears to act on both PTTH (Halme et al., 2010; Vallejo et al., 2015) and ecdysone signals (Hackney et al., 2012), and the role of retinoids appears to be ecdysteroid independent (Halme et al., 2010). However, the effects of imaginal disc damage are not exclusive to D. melanogaster, although similar mechanisms have not yet been elucidated in other models.

We have been developing the tobacco hornworm, Manduca sexta (Linnaeus 1763), as a second model organism to study such damage-induced mechanisms. We show that, similar to D. melanogaster, M. sexta development is delayed after imaginal disc damage with a shift in the pre-pupal ecdysteroid titer. Delays are noted predominantly at the pupal-to-adult transition, in an X-ray dose-dependent manner. However, we find that repair to damaged tissues occurs only when levels of damage are mild, leading to emergence of healthy adults. We also note a shift in critical weight and show that this appears to be uncoupled from JH clearance in the body. We discuss the differences in how development is affected in M. sexta and D. melanogaster using these data.

Larvae of M. sexta were raised in individual containers on an artificial wheat germ diet (MP Biomedical, Irvine, CA, USA) under a long-day photoperiod (17 h:7 h light:dark) coupled to a thermoperiod of 27°C:25°C as described by Wells et al. (2006). Pharate fifth instar larvae were staged using external morphological markers, and were treated 1–3 h following brown mandible formation, when the head capsule of the old fourth instar became air-filled [air-filled brown mandible stage (AFBM); see Fig. 1; Copenhaver and Truman, 1982]. AFBM animals typically ecdysed within 4.5±0.4 h (n=50), ensuring that all animals would ecdyse within the same time frame. Larvae were reared individually in their own cups until wandering, at which point they were transferred to wooden blocks without light, to pupate. Afterwards, pupae were transferred to small unsealed plastic bags in larger enclosures, maintained with light and thermal photoperiods.

To track the development of M. sexta (Fig. 1), larvae were checked every 24 h after ecdysis to the fifth instar. Animals were kept in cups until they showed wandering behavior, which typically began 5–6 days after ecdysis. Wandering was assessed by (i) presence of frosted frass, (ii) changes in body wall coloration, (iii) presence of extra moisture owing to excess transpiration, (iv) visibility of the aorta on the dorsal side after loss of pigmentation on the overlying epidermis or (v) purging of the gut (Truman and Riddiford, 1974). Upon reaching the wandering stage, the animals were transferred to wooden blocks without light, simulating M. sexta burrowing conditions in preparation for pupation. These larvae were monitored for pupation one to two times per day, as noted by peristaltic ecdysis behaviors and removal of the larval cuticle.

Pupae were kept inside the wooden blocks for 17 days, and were removed approximately 2 days before the normal or expected time of adult eclosion. They were placed in individual unsealed plastic bags within large glass tanks, and monitored for adult eclosion one to two times per day. The large enclosures were maintained with light and thermal photoperiods to allow for proper adult eclosion, which was first noted by peristaltic ecdysis behaviors in the abdomen and cracking of the pupal cuticle in the mid-body region. Once emerged, adults were allowed to perch on the glass sides of the tanks to allow proper wing and body expansion as the cuticle hardened. No pupae emerged prematurely in blocks.

AFBM larvae were exposed to systemic ionizing irradiation using Faxitron 43855A machines (Faxitron Bioptics LLC, Tucson, AZ, USA) operating at 105 kV and 3.0 mA, delivering a dose of 1.5 Gy min⁻¹, with exceptions. Faxitron RX-650 was used for Acridine Orange labeling and for ecdysteroid sampling, operating at 130 kV and 5.0 mA, delivering a dose of ∼10 Gy min⁻¹. Faxitron TRX5200 was used for TUNEL staining, operating at 125 kV and 3.0 mA, delivering a dose of 1.2 Gy min⁻¹. The larvae were irradiated for sufficient lengths of time to produce total exposures of 20–150 Gy, depending on the experiment. Control larvae were placed on top of the Faxitron machines during irradiation in order to control for temperature shifts in the room, as well as vibrations from the machine. After irradiation, all animals were maintained in the regulated light and thermoperiods described above.

To assess the effects of mechanical damage to the wing imaginal discs (the only discs visible through the body wall), AFBM larvae were randomly sorted into treatment groups. Larvae had their cuticle sterilized with 70% ethanol. Fine forceps were then used to fracture one, two, three or four wing imaginal discs, as described for D. melanogaster (Pastor-Pareja et al., 2008). Treatments consisted of no pinch (untreated), sham fractures to the body wall next to the wing imaginal discs, or one to four wing imaginal disc fractures. One fracture consisted of five pinches with forceps, directly onto the wing imaginal disc tissue that was visible through the cuticle to ensure sufficient disc damage. Thus, four disc fractures in a single animal consisted of 20 consecutive pinches, with shams receiving the same number of pinches to the body wall, without damage to the wing imaginal disc tissue (5 to 20 pinches). Damage to wing discs after fracturing was verified by surgically removing the discs and assessing integrity under a dissecting microscope (data not shown). Larvae that bled or had their cuticle ruptured were not used.

To assess the specificity of irradiation to the imaginal discs, AFBM larvae were irradiated for 6 min (∼60 Gy) with the Faxitron RX-650, or maintained as controls. At 24 or 48 h post-irradiation (Fig. 1), larvae were anesthetized on ice, then imaginal discs and brains were quickly dissected in chilled Manduca saline (Riddiford et al., 1979) and immediately transferred to microfuge tubes containing a 1 µg ml⁻¹ solution of Acridine Orange (Sigma Chemical Co., St Louis, MO, USA). Acridine Orange is a fluorescent dye that accumulates in lysosomes during the condensation of apoptotic corpses, and has been used to detect apoptosis in unfixed tissues of D. melanogaster and M. sexta (Arama and Steller, 2006; Tanaka and Truman, 2005). Tissues were processed using a modified protocol from Tanaka and Truman (2005). In brief, tissues were incubated for 10 min at room temperature, followed by three quick rinses with 1× phosphate-buffered saline (PBS), pH 7.2. Discs were then viewed as wet mounts in saline on a covered microscope slide. Fluorescent images were taken using a NIKON microscope (Eclipse E600, Nikon Instruments Inc., Melville, NY, USA), and full disc images were created from composite images, using Adobe Photoshop (Adobe Inc., San Jose, CA, USA).

Each tissue was assessed for damage using a rubric of Acridine Orange stain, ranging from 0 to 5, from three independent observers blind to the treatment group of the samples they were assessing. Tissues with the most stain were provided as a metric for maximal scoring and all tissues were graded from 0 (no fluorescence) to 5 (maximal fluorescence). The three independent scores were averaged to provide a single value for each sample. Values never differed by more than a score of 1 between individuals.

To confirm the results obtained from Acridine Orange staining and further assess the specificity of irradiation to the imaginal discs, AFBM larvae were irradiated for 50 min (∼60 Gy) with the Faxitron RX5200, or maintained as controls (0 Gy). At 12 h post-irradiation, larvae were anesthetized on ice for approximately 10–15 min, then forewing imaginal discs and brains were quickly dissected in chilled Manduca saline (Riddiford et al., 1979). TUNEL (terminal deoxynucleotidyl transferase mediated dUTP nick end labeling) was performed from a composite protocol (Halme et al., 2010; Tanaka and Truman, 2005). Tissues were fixed in 3.7% formaldehyde in PBS for 1 h at room temperature followed by three rinses in PBS at 30-min intervals. Samples were then incubated in 20 μg ml⁻¹ proteinase K in 10 mmol l⁻¹ Tris, pH 7.5 for 1 h at 37°C and the reaction was stopped by washing three times in PBS-TX followed by three washes at 30 min intervals in PBS. Tissues were rinsed three times in 10 mmol l⁻¹ Tris, pH 7.5, and processed with a TUNEL reaction for 1 h at 37°C using the In Situ Death Detection Kit Fluorescein-Texas Red (Roche Applied Science, Indianapolis, IN, USA). Samples were rinsed and washed three times at intervals of 10 min with PBS. The specimens were mounted in glycerol.

Manduca sexta hemolymph was collected by cutting an opening in one of the mid-body prolegs and allowing the hemolymph to drain onto a sterilized parafilm strip. For each animal, a 100 µl volume was quickly pipetted into a centrifuge tube containing 300 µl of methanol, and vortexed. Samples were maintained at −20°C until processed. Hormone titers of ecdysteroids were quantified by an enzyme-linked immunosorbent assay (ELISA) as previously described in detail for crustaceans (Abuhagr et al., 2014). The 20-hydroxyecdysone/horseradish peroxidase (20HE-HRP) conjugate and 20-hydroxyecdysone (20HE) antibody were obtained from Dr Timothy Kingan (Kingan, 1989). The 20HE-HRP-conjugated reagent {1:64,000 dilution in assay buffer [AB, 25 mmol l⁻¹ sodium phosphate, pH 7.5; 150 mmol l⁻¹ NaCl; and 1 mmol l⁻¹ EDTA disodium dihydrate with 0.1% bovine serum albumin (BSA; 50 μl)]} was added to all wells of a 96-well plate. Rabbit anti-20HE primary antibody (50 μl; 1:100,000 dilution in AB with 0.1% BSA) was added to all wells, except for the first two wells, which contained only AB plus 0.1% BSA.

Critical weight was determined for healthy control animals and irradiated experimental animals by the method described by Helm and Davidowitz (2015). Briefly, non-irradiated animals were selected at the AFBM stage and allowed to feed ad libitum. Larvae were then grouped in 0.5 g mass bins (e.g. groups of 1.0–1.4 or 1.5–1.9 g) and separated in two treatment groups, fed and starved. Fed animals were provided sufficient food at all times, while starved animals were restricted to a small piece of polyacrylamide water crystal to prevent dehydration. Starved animals were moved to a clean cup with a new piece of water crystal every other day to prevent infections or drowning. Animals were monitored every 24 h until they reached the wandering stage. Irradiated animals (50 min, or ∼75 Gy with Faxitron 43855A) were treated in a similar manner, with both fed and starved groups.

Critical weight was defined as the mass category in which no significant difference was observed for the time to wandering between fed and starved animals (Nijhout and Williams, 1974a). Significance between fed and starved animals was determined by performing a Student's t-test for each of the mass bins.

Total RNA was extracted from tissues at AFBM prior to irradiation and then from both healthy controls and irradiated M. sexta larvae ∼0–2 h after L5 ecdysis, and every 24 h after that for 4 days (prior to wandering). The tissues included a combination of wing imaginal discs, brain, subesophageal ganglion and a total length of three dorsal ganglia, prothoracic glands, gut, fat body, trachea and thoracic muscle. All tissues collected amounted to approximately 30–40 mg of tissue per animal. The tissues were collected by dissecting the larvae in modified Weever's saline (Trimmer and Weeks, 1989), and immediately submerged in 600 µl of RNA-later (Thermo Fisher Scientific, Waltham, MA, USA). Dissected tissues were stored overnight at 4°C, then transferred to a −20°C freezer, where they were stored until the RNA extraction process.

Total RNA was extracted using the RNeasy Mini Kit (QUIAGEN, Germantown, MD, USA) following the methods described in the RNeasy Mini Handbook (2012) for animal tissues. However, one extra step was added: after the first wash with 350 µl RW1 buffer, 10 units of RQ1 RNase-Free DNase (Promega, Madison, WI, USA) were added to the spin column and allowed to incubate at room temperature for 10 min. The DNase was washed away with a second wash of 350 µl RW1 buffer. This extra step was added to ensure that the final product was pure RNA without any DNA contamination. Lastly, the RNA was eluted with nuclease-free water and stored at −20°C.

The expression levels of the JH downstream gene, Kr-h1, were measured using the comparative C_t method. RpL3 was used as an internal reference gene. No significant effects of the time (F=1.094, P=0.298) or dose (F=1.959, P=0.093) in RpL3 expression levels were noted, nor was any significant interaction between these effects (time×dose, F=0.494, P=0.780), as assessed by two-way ANOVA. RpL3 was amplified using the same primers used by Koyama et al. (2008): GAGGATAAAGTAAAGTGGGCCAGGG as the forward primer, and CACGCGCAACGACTTTCTCAAAGTG as the reverse primer. It is worth noting that the protein-coding region of Kr-h1 is not well characterized in M. sexta. Therefore, we designed our primers within an area that is highly similar (>80%) to the exon-1 region of the better characterized Bombyx mori Kr-h1. The primers used to amplify Kr-h1 were: GGAGAACGCCCATACACCTG as a forward primer, and CCGTAGACTGCATGGACTGC as the reverse primer. Gene expression was quantified using the Luna Universal One-Step RT-qPCR Kit (New England Biolabs Inc., Ipswich, MA, USA). RT-qPCR reactions were performed in an Applied Biosystems 7300 machine (Applied Biosystems, Foster City, CA, USA). The reactions consisted of a final volume of 20 µl, containing 10 µl of the kit's 2X SYBR master mix, 1 µl of reverse transcriptase enzyme, 100 nmol l⁻¹ of each primer and 50 ng of template RNA, and RNase-free water was added to complete 20 µl per reaction. The cycling settings consisted of a holding stage at 55°C for 10 min, then another holding stage at 95°C for 1 min, followed by 40 cycles of 95°C for 15 s and 60°C for 60 s. In addition, specificity of the PCR product was assessed by a melting curve analysis from 60°C to 95°C. Each sample was run in technical triplicates and the mean C_t value of each technical triplicate was transformed into a relative quantity using the comparative C_t method (Livak and Schmittgen, 2001). As part of this method, we used the average relative expression of samples collected at the AFBM stage as baseline. Thus, our results show relative expression of Kr-h1 as the fold change in expression compared with untreated animals at the AFBM stage.

Healthy M. sexta were supplemented with a JH analog, methoprene (SPEX CertiPrep, Metuchen, NJ, USA; see Fig. 1), to test whether increased JH activity could trigger developmental delays in healthy animals similar to those caused by irradiation-induced damage. Healthy M. sexta were supplemented with methoprene approximately 1 h after they had ecdysed to the fifth larval stage. As a negative control, animals were treated with the vehicle cyclohexane. As a positive control, untreated AFBM animals were irradiated for 50 min (∼75 Gy) with Faxitron 43855A and monitored. Animals were supplemented with three different doses of methoprene: 0.01, 0.1 and 1 µg. The diluted methoprene was added topically to the dorsal side of unanesthetized animals. After treatment, larvae were placed in individual plastic cups with food ad libitum, and their development was tracked as described above.

Healthy and irradiated animals (75 Gy, Faxitron 43855A) were ligated slightly posterior to the subesophageal ganglion, but not beyond the first spiracle, where the prothoracic gland is located, thus separating the brain/CC/CA complex from the prothoracic glands. Animals were ligated approximately 1 h after ecdysing to the fifth larval stage, or approximately 6 h after irradiation (Fig. 1). Larvae were ligated for only 24 h after ecdysis because by 48 h only 59% had survived (data not shown), likely due to starvation and hypoxia of the head capsule. After the ligation period, ligatures were carefully cut and the larvae were placed in individual plastic cups with food ad libitum and their development tracked as described above.

To determine whether M. sexta responded to tissue damage with delays in metamorphosis and critical weight, as seen in D. melanogaster, imaginal discs of pharate fifth instar larvae were damaged and development was timed. Larvae were irradiated at three increasing X-ray doses (Fig. 2) or had discs damaged mechanically (Fig. 3, Fig. S1). The times to begin wandering, pupating and eclosing were monitored. After irradiation, the time to wander was not affected at any X-ray dose (Fig. 2A). In contrast, both the 50 and 75 Gy doses of irradiation induced 0.5–1 day delays in pupation (Fig. 2B), while all doses delayed eclosion, anywhere from 0.5 to 3 days (Fig. 2C, Fig. S4C).

Mechanical damage of one to four wing imaginal discs (Fig. 3, Fig. S1) did not alter the timing to wander (Fig. 3A) or pupate (Fig. 3B), but produced a small delay in the timing of eclosion (Fig. 3C). Eclosion appeared to be delayed by just over a day after damage to any number of imaginal wing discs. Damage to only two or four discs are shown in Fig. 3 for clarity, while damage to one or three discs are shown in Fig. S1. Results were identical for all situations. To control for possible effects of damage to the cuticle owing to pinching, we used shams, where the cuticle next to the imaginal discs was pinched, but no disc was damaged. No differences in development were noted for 5 to 20 pinch shams (data not shown), and thus 20 pinch sham controls are presented in the figures.

We verified imaginal disc damage produced by whole-body X-ray irradiation by noting phenotypic abnormalities in pupal and adult structures arising from the irradiated discs (Fig. S2), and by assessing and quantifying cell damage/death in the imaginal discs (Fig. 4, Fig. S3). Imaginal discs were clearly damaged upon irradiation, as noted by damage to wings, antennae and eye regions (data not shown) in the pupae (Fig. S2C,D, arrowheads and arrows, respectively) compared with controls (Fig. S2A,B). In most cases, adults showed complete recovery to low doses of irradiation (data not shown), but only marginal repair to these areas after high irradiation doses (Fig. S2D) compared with controls (Fig. S2B).

We also assessed the level of damage at the cellular level in wing imaginal discs. Wing imaginal disc cells in M. sexta proliferate continuously and exponentially throughout the late larval and pre-pupal periods (Nijhout et al., 2014), suggesting they should be vulnerable to irradiation. To assess the specificity of damage by whole-body X-ray irradiation, wing imaginal discs from irradiated larvae were dissected and compared with control discs (Fig. 4, Fig. S3, Table 1). Wing discs showed significant increases in Acridine Orange stain (Fig. 4, Table 1) and in TUNEL staining (Fig. S3) after whole-body irradiation, while no significant stain was noted in the brains. Acridine Orange stain was noted in irradiated forewing and hindwing imaginal discs 24 and 48 h after irradiation (Fig. 4C,D,G,H, respectively), but not in controls (Fig. 4A,B,E,F). Brains of non-irradiated and irradiated animals showed no stain (Fig. 4I–L). Blinded scoring of stain intensity, on a scale of 1 to 5, revealed significant differences in this stain intensity in imaginal discs after X-ray compared with their control counterparts (P<0.05; Table 1). The TUNEL assay revealed apoptosis as early as 12 h after irradiation compared with control discs (Fig. S3A,C) and minimal label was noted in non-irradiated and irradiated brains (Fig. S3B,D).

In order to determine whether pre-pupal ecdysteroid titers were altered after X-ray irradiation, as has been noted in D. melanogaster (Hackney et al., 2012), ecdysteroid levels were quantified by ELISA (Fig. 5). Hemolymph from larvae irradiated just prior to L5 ecdysis was collected every 12 h after larvae reached the wandering stage. The ecdysteroid titer of control (0 Gy) larvae peaked around day 2 after larvae started wandering, then declined by day 4 just prior to pupation. In contrast, the ecdysteroid titer of irradiated larvae (75 Gy) peaked ∼1.5 days later than control larvae and had not fully declined by day 4. Sample sizes were lower at the beginning and end of the titers, owing to values often being below the detection level of our assay.

Critical weight has been shown to shift after X-ray-induced imaginal disc damage in D. melanogaster (Stieper et al., 2008). To determine whether the same phenomenon occurred in M. sexta, the critical weight was assessed after significant X-ray damage (Fig. 6). Under control conditions (non-irradiated), the critical weight was determined to be approximately 5.5 g because, after reaching this mass, starvation no longer delayed development (Fig. 6A). In contrast, irradiated animals showed an increased critical weight of approximately 7.5 g (Fig. 6B).

The growth rate of these larvae was also assessed, as the daily wet mass of each animal (Fig. 6C). The growth rate of M. sexta was not affected by irradiation. There were no significant daily differences in mass (P>0.5), even at the peak mass of approximately 12.0 g. Both groups attained their final body size on the same day at wandering.

Critical weight has been considered a proxy for JH levels in M. sexta (Davidowitz, 2016; Nijhout and Williams, 1974b). To determine whether the delay in pupal commitment associated with the irradiation-induced increase in critical weight was also accompanied by a delay in the decline of JH levels, we assessed whether JH levels shifted proportionally to the delayed critical weight after X-ray irradiation. We measured changes in transcription levels of the downstream signal of JH activity, Krüppel Homolog 1 (Kr-h1). Kr-h1 transcription has been used as an indicator of JH activation of its putative receptor, Methoprene-tolerant (Met; Kayukawa et al., 2014; Lee et al., 2015; Minakuchi et al., 2007, 2009). Starting levels of Kr-h1 at AFBM were consistently low, with little variation between animals. As expected, there was a marked increase in transcription of Kr-h1 mRNA around the time of ecdysis to L5, followed by a return to pre-ecdysis levels over the next 72 h (Fig. 7). Notably, there were no significant differences in Kr-h1 mRNA expression between control and irradiated larvae, which were assessed over 4 days at 12 h intervals from the beginning of L5 ecdysis, despite the shift in critical weight.

To confirm our observations of Kr-h1 expression, we tested whether a simple increase in JH levels could result in developmental delays such as those observed as a result of irradiation-induced tissue damage. Supplementation with the JH analog, methoprene, early in the last larval instar did not cause statistically significant changes in time to wandering (Fig. S4A), pupation (Fig. S4B) or eclosion (Fig. S4C). But, as expected, radiation treatment without methoprene did cause delays to pupation and eclosion (Fig. S4B and C, respectively).

We further reasoned that if JH played a role in the damage-induced developmental delays, then removal of JH would rescue normal developmental timing in damaged animals. To test this idea, animals were ligated posterior to the brain/CC/CA complex, but anterior to the prothoracic glands (Fig. 8D–F). This prevented JH from being released into the body, where tissues such as the prothoracic glands, the primary source of ecdysone, were located. Ligations are known to prevent flow of hormones from their source gland to the rest of the body (Hatakoshi et al., 1988; Suzuki et al., 2013; Truman et al., 2006). Larvae were ligated for only 24 h after ecdysis because during that period of time JH activity is at its highest, thus preventing JH from reaching the rest of the body when it was most active. The absence of JH in the body (as well as anterior ganglia including brain and subesophageal ganglion), through ligations, only partially rescued developmental timing in irradiated larvae. That is, ligations partially prevented irradiation-induced delays to pupation (Fig. 8B) by approximately half a day, and fully rescued adult eclosion (Fig. 8C). It should be noted that, although not significant, there was a trend towards a mild increase in developmental timing of adult eclosion after irradiation and ligation, as was noted in pupae. Interestingly, ligated animals wandered ∼7 days after ecdysis whether irradiated or not, compared with non-ligated larvae, which wandered on day 6 (Fig. 8A).

We confirmed the efficacy of ligations by measuring Kr-h1 expression in tissues collected from ligated animals. As we expected, tissues collected from ligated animals showed a significant decrease in expression of Kr-h1 (data not shown), which confirmed that ligations prevented the flow of JH from the CA to the rest of the body.

We show that the holometabolous insect M. sexta can be used as a model to study the regulation of growth and development after selective imaginal disc damage when whole-body X-ray irradiation is used to directly damage imaginal discs (Fig. 2) or when discs are fractured manually (Fig. 3). Even though we only compared wing imaginal disc damage with that of brains after whole-body irradiation to verify selective damage of the discs (Fig. 4, Fig. S3), other tissues such as the gut, salivary glands, lymph glands and fat tissue have likewise been shown to be unaffected by irradiation in D. melanogaster when wing discs are damaged, using both TUNEL and caspase staining methods (Halme et al., 2010). The extent of damage to imaginal discs appears to go beyond the wings because other discs such as eyes (data not shown) and antennae also appeared damaged in M. sexta, as noted phenotypically (Fig. S2). Thus, whole-body irradiation appears to be an effective strategy for studying the developmental effects of imaginal disc damage in this newer, larger insect model.

Imaginal disc damage after whole-body irradiation in M. sexta resulted in delays in the pupal and adult stages similar to those seen for D. melanogaster after whole-body irradiation (Halme et al., 2010; Stieper et al., 2008) and after genetically induced imaginal disc damage (Colombani et al., 2012; Hackney et al., 2012). Delays have also been described in other insects after injury (Bourgin et al., 1956; Bryant and Simpson, 1984; Ely and Jungreis, 1977; Halme et al., 2010), suggesting that the developmental delay response to imaginal disc damage may be conserved amongst insects.

The developmental delays noted at pupation and eclosion in irradiated M. sexta indicate that both the larval and pupal stages are extended after imaginal disc damage (Figs 2 and 3). Presumably, these delays provide an extended developmental window for growth plasticity while imaginal discs regenerate before development continues (Madhavan and Schneiderman, 1969; Jaszczak et al., 2015; Boulan et al., 2019). This response appears to be dose dependent: greater damage by irradiation produces longer delays (Fig. 2). This may reflect secretion of delay factors by the damaged discs, where greater damage results in greater secretion. In contrast, we noted that there were only limited delays after selective mechanical damage to just the wing imaginal discs. That is, only eclosion was delayed, even after damage to four wing discs (Fig. 3, Fig. S1). We suggest that the longer delays after irradiation may have been due to imaginal disc damage beyond just the wings, such as to antennae, proboscis and eyes (Fig. S2). We cannot rule out the possibility that fracturing the wing discs compromised tissue integrity rather than inducing cellular damage such as apoptosis, and as such, different mechanisms may have been at play in delaying eclosion rather than pupation. This seems less likely, given that integumentary mechanical damage to another holometabolous insect, Galleria melonella, delayed pupation (Malá et al., 1987). These differences nevertheless merit further investigation, possibly through localized irradiation, by shielding some imaginal discs during the irradiation process. This was done successfully in D. melanogaster, where Jaszczak et al. (2015) were able to localize damage to wing imaginal discs.

Although both M. sexta and D. melanogaster delayed their development in response to imaginal disc damage, they appeared to do so using different mechanisms. In M. sexta, we noted the greatest delays during the pupal stage, ranging from 0.5 to 3 days, while the larval stage was typically extended by 1 day. The extension of the larval stage in M. sexta was short in comparison to the 3–5 day extensions in Drosophila larvae after imaginal disc damage (Hackney et al., 2012), especially given the longer life cycle in M. sexta. A 2-day delay in D. melanogaster comprises approximately 15% of its life cycle in comparison to only 5% in M. sexta. Moreover, the extent of repair after damage appeared less in M. sexta as well (Fig. S2). A compelling explanation for the less extensive delays in M. sexta arises from its apparent lack of Drosophila insulin-like peptide 8 (Dilp8), a peptide secreted from damaged discs that modulates developmental delays and growth plasticity in D. melanogaster (Colombani et al., 2012; Garelli et al., 2012; Jaszczak et al., 2015). The most recent publicly available M. sexta genome does not include a sequence representative of Dilp8 (Table S1), which appears to be lacking in other holometabolous insects as well.

Nevertheless, there appear to be conserved mechanisms in place to delay development by delaying the secretion of ecdysone in both M. sexta (Fig. 5) and D. melanogaster (Hackney et al., 2012; Halme et al., 2010). This suggests that dysregulation of developmental hormones may be a mechanism of action of damaged imaginal discs to delay development and enhance the probability of coordinated adult growth (Boulan et al., 2019). We cannot discount the possibility that a completely unique delaying factor may be secreted from damaged imaginal discs in M. sexta. The putative retinoid signaling pathway previously suggested to work in D. melanogaster (Halme et al., 2010) could very likely still act to regulate developmental timing after imaginal disc damage in M. sexta. Thus, there may be multiple secreted factors affecting development, some of which are conserved across species while others are not. It will be important to assess hormone titers at all developmental stages following irradiation to see where the effects are most noted.

Critical weight is a developmental milestone in fifth instar larvae that has been identified as an endocrine proxy, most specifically to JH levels (Davidowitz, 2016; Nijhout and Williams, 1974b). By definition, critical weight is the point in development when lack of nutrition no longer delays time of PTTH release, required to stimulate the increases in ecdysone in the hemolymph necessary for wandering and pupation (Nijhout and Williams, 1974a). Release of PTTH is dependent on the cessation of JH secretion, which appears to be the critical endocrine event triggered when critical weight is reached, at least under typical conditions. Once JH titers fall, the hormonal cascade that stimulates PTTH release, ultimately causing wandering behaviors and eventually pupation, is initiated (Nijhout and Williams, 1974b). Thus, under normal conditions in M. sexta, critical weight is thought to link nutrition-sensing mechanisms, such as insulin signaling, with the cessation of JH, where it ultimately defines developmental timing and final body size (D'Amico et al., 2001; Davidowitz et al., 2003; Hatem et al., 2015; Nijhout and Williams, 1974b). Yet in this system, imaginal disc damage in M. sexta increased the critical weight, without significantly altering growth rate, and delayed pupation but not wandering. Moreover, even though critical weight was increased by irradiation, there was apparently no accompanying change in the pattern of JH secretion, as assessed by Kr-h1 expression levels (Fig. 7). Thus, even though irradiated animals reached their critical weight at a later age than control animals, their JH levels appeared to decline at the same rate and on the same day of development as control animals, and they wandered at the same time. This indicates that JH titers became uncoupled from critical weight after imaginal disc damage, and that the effects of irradiation worked downstream of JH activity.

It will be important to determine the JH levels in M. sexta and D. melanogaster directly, as comparisons, and what pre-wandering ecdysteroid titers look like. JH activity was assessed by Kr-h1 expression, thus we cannot discount the possibility that JH titers remained high but were acting through a different receptor pathway than the typical Met receptor. It has been suggested that JH can act on different signaling pathways (Kayukawa and Shinoda, 2015; Wilson and Fabian, 1986; Zhao et al., 2014). For instance, high concentrations of JH have been shown to bind to the orphan receptor Ultraspiracle (Jones et al., 2001), and membrane-bound receptors have been suggested to exist in insect ovaries (reviewed by Davey, 2007). This does not appear likely in our experiments, however, given that ligations, placed posterior to the head capsule to inhibit JH activity in the body, only partially rescued normal developmental timing (Fig. 8). Likewise, supplementation with methoprene did not delay pupation or eclosion (Fig. S4). The doses of methoprene we used did not induce erroneous development, such as supernumerary larval molts or hybrid larval–pupal molts (Hatakoshi et al., 1988). At higher doses, we did note similar larval–pupal hybrid transitions (data not shown). These data therefore suggest that a factor from the head capsule appears to at least partially delay development in irradiated larvae, but what this is remains unclear, given the lack of change in JH signaling in irradiated larvae. It will be important to determine the JH, PTTH and ecdysteroid titers in both M. sexta and D. melanogaster to determine what roles they all play in the developmental delays noted after imaginal disc damage.

Understanding how damage to imaginal discs affects critical weight in M. sexta is an important first step to defining the endocrine signals that may be disrupted after imaginal disc damage, resulting in delayed development. Although a shift in critical weight without any changes in wandering time (Fig. 2B), growth rate (Fig. 6C) or kr-h1 expression levels (Fig. 7) may warrant caution in regards to the sample size used in this experiment, the effect of food availability on developmental timing had strong statistical power for both healthy and irradiated animals (β<0.1 for both data sets).

We had expected delays in the onset of wandering following such a large shift in critical weight without a concomitant change in overall growth rate, as was noted in D. melanogaster (Stieper et al., 2008). The method we used has been well established in M. sexta, but it was not the same method used with D. melanogaster. The critical weight of the fly is determined by a bi-segmented linear regression, where the breakpoint represents the critical weight (Stieper et al., 2008). Even using a bi-segmented linear regression in our data set, we observed a shift in critical weight of M. sexta (data not shown). These differences in methodology shed light on the inconsistency of critical weight as an endocrine proxy across insect species, highlighting the importance of confirming endocrine inferences through more direct measurements. It will be imperative to assess the pre-wandering ecdysteroid titers, which we assume will not be shifted after irradiation in M. sexta, given the lack of delay in wandering. These peaks have not been assessed in D. melanogaster either, although pre-pupal shifts have been noted in both M. sexta (Fig. 5) and D. melanogaster (Hackney et al., 2012), which were accompanied by delays in pupation.

As a side note, there was an interesting significantly faster rate of development in control starved (non-irradiated) animals within the 7.0 g group (Fig. 6A) that has not been previously reported (D'Amico et al., 2001; Davidowitz et al., 2003; Helm and Davidowitz, 2015; Nijhout and Williams, 1974a). This may have been due to overcompensation as a result of the delays by starvation, but requires further investigation.

The responses in M. sexta differ in some ways from changes noted in D. melanogaster after irradiation (Stieper et al., 2008; see Fig. 9). In D. melanogaster, Dilp8 signaling modulates hormone secretion to serve two different functions: to regulate developmental timing and to coordinate growth of undamaged tissues. Through the modulation of hormone secretion, Dilp8 acts as a delay factor to downregulate the insect's growth rate to delay development (Vallejo et al., 2015), as well as a size sensor to coordinate the growth of healthy tissues as the damaged tissues recover (Jaszczak et al., 2015; Boulan et al., 2019). Together, both mechanisms serve to alter the growth rate and critical weight (Stieper et al., 2008), consequently altering developmental timing in D. melanogaster.

We observed a similar shift in critical weight in M. sexta (Fig. 6B), but we did not see changes in growth rate (Fig. 6C). This contrast suggests that M. sexta regulates developmental timing in response to imaginal disc damage independent of mechanisms that regulate growth. The differences noted between M. sexta and D. melanogaster indicate that some of the mechanisms uncovered in D. melanogaster are not redundant in M. sexta. This could explain the limited delays and regenerative capacity observed in M. sexta. These differences may arise from differences in the regulation of growth and development under normal conditions. For instance, in D. melanogaster, ecdysteroid biosynthesis depends on coordinated signals from insulin and PTTH, but this does not appear to be the case in M. sexta (Mirth et al., 2005; Smith et al., 2014). This suggests that the limitations in the duration of developmental delays in M. sexta may arise from differences in regulation of hormone secretion. In addition, D. melanogaster and M. sexta clearly respond differently to a variety of environmental stresses and hormonal shifts. For instance, D. melanogaster goes through premature metamorphosis under starved conditions after attaining critical weight, resulting in smaller adults (Stieper et al., 2008), whereas M. sexta shows normal developmental timing, which allows it to maintain a relatively normal body size. Likewise, application of JH in the last larval instar results in a supernumerary molt in M. sexta (Cymborowski et al., 1982; Hatakoshi et al., 1988), which is not the case with D. melanogaster (Mirth et al., 2014).

Our data provide a model for how regulation of growth and developmental timing varies between D. melanogaster and M. sexta (Fig. 9). Although both M. sexta and D. melanogaster showed increased critical weight after X-ray, growth rate only changed in D. melanogaster. Increasing critical weight may be a mechanism to ensure enough nutrient storage necessary to satisfy the growth and repair of damaged imaginal discs during the extended developmental period, but the different responses in growth indicate different regulatory mechanisms. For instance, the terminal growth period is delayed in D. melanogaster, resulting in a delay to wandering, but is shortened in M. sexta, resulting in a constant wandering period, but an extended post-wandering phase prior to pupation. This reduced terminal growth period may also account for the shorter delay noted to pupation in M. sexta compared with D. melanogaster, relative to the lengths of their life cycles. Thus, it is not entirely surprising that these two insect species respond differently to the same tissue insults. Using M. sexta as an alternative model to understand the regulation of growth and development in response to tissue damage may shed light on the evolutionary divergence of mechanisms designed to generate similar physiological responses. Therefore, it would be interesting to expand this investigation to other well-characterized holometabolous insects, such as B. mori.

Differences in life history of these two insect species may account for some of the differences in the regulation of growth and development. For example, larval D. melanogaster becomes more visible to predators as it pupates, with its dark coloration providing contrast to the lighter medium it rests on. Thus, extending the larval stage during imaginal disc repair for days may be more advantageous than extending the pupal phase to the same extent. In contrast, M. sexta pupates below ground, providing optimal cover for an extended pupal developmental period. Whether the delays noted in this study indicate an evolutionary adaptation based on life history strategies may become apparent when we eventually compare delays in a variety of other holometabolous insects.

Developmental delays associated with tissue damage appear to be conserved in many insects and are presumed to be a mechanism to allow time for repair of the damaged imaginal discs. Life histories of different insects may dictate where the majority of the delays occur – at the larval or pupal stages. In M. sexta, the developmental delays caused by imaginal disc damage occur in the stages of development surrounding metamorphosis, when tissue size regulation is critical to ensure proper adult development. This is in contrast to D. melanogaster, when development appears particularly prolonged in the larval stage. The fact that M. sexta has some shared and some unique responses with D. melanogaster suggests that it may be a strong complementary model to understanding the cellular mechanisms regulating these delays. Our results suggest that disruption of tissue development homeostasis in M. sexta through irradiation-induced imaginal disc damage may uncouple hormone-driven physiological milestones, presumably by attempting to prioritize tissue repair to ensure proper adult development. In particular, we now have a system to assess the role of JH independent of nutritional signals that define critical weight in irradiated larvae. It will be important to understand what cues are regulating development without altering growth in these damaged animals. The next steps are therefore to define the mechanisms that change critical weight and induce developmental delays.

We thank Iswar Hariharan for the use of his Faxitron machine at the University of California, Berkeley, and his willingness to welcome M.F. to his lab meetings; Adrian Halme for his insights and introduction to this research area; Annette Chan and the Cell and Molecular Imaging Center for overseeing use of the confocal microscopes; Sharon A. Chang for laboratory assistance of ecdysteroid measurements; and Michael Lin, Cleopa Omondi and Cheikh Njie for technical assistance with animals.