Imaginal Discs Secrete Insulin-Like Peptide 8 to Mediate Plasticity of Growth and Maturation

See allHide authors and affiliations

Science  04 May 2012:
Vol. 336, Issue 6081, pp. 579-582
DOI: 10.1126/science.1216735


Developing animals frequently adjust their growth programs and/or their maturation or metamorphosis to compensate for growth disturbances (such as injury or tumor) and ensure normal adult size. Such plasticity entails tissue and organ communication to preserve their proportions and symmetry. Here, we show that imaginal discs autonomously activate DILP8, a Drosophila insulin-like peptide, to communicate abnormal growth and postpone maturation. DILP8 delays metamorphosis by inhibiting ecdysone biosynthesis, slowing growth in the imaginal discs, and generating normal-sized animals. Loss of dilp8 yields asymmetric individuals with an unusually large variation in size and a more varied time of maturation. Thus, DILP8 is a fundamental element of the hitherto ill-defined machinery governing the plasticity that ensures developmental stability and robustness.

Animal size is remarkably constant within species. This constancy is even more striking within the animal, such as in comparing the left and right sides of bilaterian organisms—for example, the symmetry of a human face or the coincidence in size of the left and right hand. Such precision requires growing organs to communicate and coordinate their final sizes, processes that have long remained poorly understood (1).

The imaginal disc epithelia that generate the adult Drosophila structures have a remarkable capacity to regulate their size, although only in larvae (2). The onset of the larval-pupal transition is controlled by pulses of the steroid hormone 20-hydroxyecdysone (20HE), as initiated by brain-derived prothoracicotropic hormone (PTTH), and this transition marks the end of imaginal disc growth (3, 4). Typically, the time of pupariation adapts to accommodate imaginal disc growth (4), and indeed, pupariation is delayed when imaginal discs suffer lesions to allow the missing parts to be restituted (57). The length of the delay correlates with the amount of tissue to be regenerated (6), indicating that the endocrine system fine-tunes organ growth (or regeneration) and adjusts maturation accordingly. Tumor growth in imaginal discs also delays or blocks metamorphosis [(8) and citations therein]. Moreover, larvae with imaginal discs that are damaged or contain tumors metamorphose at the correct size (9). Thus, we speculated that tumors and regenerating discs might emit a common signal to adapt growth and maturation.

We attempted to identify such a signal in oligonucleotide microarrays (fig. S1) by screening for genes encoding signal peptides that are up-regulated in association with tumors in eye discs induced by an oncogenic combination of the Notch ligand Delta and two neighboring epigenetic repressors, pipsqueak and what are collectively called “eyeful” (10), which causes massive overgrowth and metastasis. CG14059 was the most consistently enriched putatively secreted gene product in tumor discs (Fig. 1A and table S1) and was also enriched during transdetermination (11), another process that delays pupariation. Because the gene product had an invariant 6-cysteine motif typical of the insulin-relaxin peptide family (Fig. 1B and figs. S1 and S2), we named this gene Drosophila insulin-like peptide 8 (dilp8). This is the only DILP differentially expressed in tumors (fig. S3). The secretion of DILP8 was confirmed by expressing a carboxy-terminal FLAG-tagged construct, which, consistent with its hormonal nature, was detected in larval hemolymph (Fig. 1C) and in the medium of transfected Schneider (S2) cells (fig. S4).

Fig. 1

Tumor discs produce DILP8, a divergent insulin/relaxin-like peptide. (A) Heat map of putatively secreted genes up-regulated in eyeful tumor eye discs from three independent replicates (1 to 3) of microarray experiments in which expression is color-coded. Relative gene expression levels are color-coded to range from blue (lower levels of gene expression), through white (medium expression), to red (higher levels of expression). CTL, control non-tumor eye discs [+/UAS-Dl GS(2)88A8lola pipsqueak]; TUMOR, eyeful tumor eye discs [+/ey-Gal4 UAS-Dl GS(2)88A8lola pipsqueak]. (B) Alignments of CG14059/DILP8 with representative members of the human insulin-like family of peptides. Alignments of DILP8 with the seven Drosophila ILPs characterized previously are shown in fig. S1. (C) DILP8 is readily detected in the cell-free hemolymph (H) of tub>dilp8::3x-FLAG larvae in Western blots probed with anti-FLAG antibodies. Total (T) hemolymph extract serve as loading controls. Samples are free of cell contamination, as shown by actin and histone H3 controls. (D) Responses of eGFP trap dilp8MI00727 to abnormal imaginal disc growth (arrowheads). Scale bars, 1 mm. Controls (top images in top two panels) are sibling larvae that were either CyO/+ (wild type; top) or heterozygous +/dlgm52 (middle). Representative heterozygous third chromosome Minute mutant, M(3)i55, is shown at bottom. (E) Confocal images show GFP (green) and 4′,6-diamidino-2-phenylindole (DAPI) (blue) staining in wild type (top) and eyeful (bottom) larvae heterozygous for dilp8MI00727. Brain expression is delayed in the eyeful tumor model. rg, ring gland.

We examined the native and induced expression of dilp8 using an enhanced green fluorescent protein (eGFP) trap in the gene’s first intron [Mi{MIC}CG14059MI00727 (12), hereafter dilp8MI00727 (fig. S5)]. Larval eGFP expression was assessed in mutants with different growth perturbations in the imaginal discs that delay pupariation: fast-growing tumors induced by oncogene activation (10); slow-growing tumors, exemplified by a recessive mutation in the tumor-suppressor gene discs large [dlgm52 (13)]; and slow growth of imaginal discs due to Minute mutations (9). In all cases, there was cell-autonomous induction of eGFP in the affected third-instar imaginal discs (Fig. 1, D and E, and fig. S6), as well as weak and dynamic signals in the normally growing discs and brain (Fig. 1E). Hence, DILP8 is a common response to abnormal imaginal disc growth, and the response is conserved in other Drosophila spp. (fig. S7).

Using the homozygous dilp8MI00727 insertional mutation that reduces dilp8 mRNA expression by 99.4% (Fig. 2A), we investigated whether dilp8 influences pupariation. In synchronous larvae, loss of dilp8 reverted the delay in pupariation caused by eye disc tumors from 26.6 ± 7.5 hours to 5.9 ± 7.1 hours (mean ± SD pupariation time, P = 0.026) (Fig. 2B).

Fig. 2

DILP8 is required for adaptive developmental plasticity in imaginal discs growing abnormally. (A) dilp8 expression relative to rp49 analyzed by means of quantitative reverse transcription polymerase chain reaction (RT-PCR) in virgin females of the wild type (w1118) or homozygous dilp8MI00727 mutant (mean ± SD of three or four independent replicates) (fig. S5). (B) Loss of dilp8 in the eyeful tumor background rescues delayed pupariation. Three replicates were scored, each with two to five tubes containing ~30 larvae. (C) Confocal images of GFP (green) and GAL4 (red) staining in Bx>rpr;dilp8MI00727/+ wing discs (DAPI, blue). Scale bar, 100 μm. (D) Depletion of dilp8 in Bx>rpr animals partially rescues the delay in pupariation (n = 5 replicates of two to five tubes of ~25 larvae per genotype scored). (E) Pupariation time and dilp8 expression analyzed by means of quantitative RT-PCR in animals fed with phosphate-buffered saline or two doses of EMS (10 to 20 mM). (F) EMS (10 mM) activated the eGFP reporter in damaged discs. Scale bar, 1 mm. (G) dilp8MI00727 reduced the delay in pupariation in EMS-fed larvae (four tubes of ~25 larvae were scored per genotype and treatment). ey> and Bx> indicate eyeless (ey)-Gal4/+ and P{GawB}BxMS109/+ and serve as genetic background controls.

To determine whether the dilp8 response is tumor-selective or broadly used, we assayed dilp8 expression and activity during regeneration induced by two forms of damage. First, cell death was induced by overexpressing the proapoptotic gene reaper (rpr) by using Beadex-Gal4 (Bx>rpr), which provokes continuous intrinsic damage and regenerative growth in the wing pouch, and a pupariation delay (14). dilp8 transcripts were up-regulated in third-instar Bx>rpr larvae (fig. S8), and the dilp8MI00727 reporter was activated cell-autonomously in damaged/regenerating cells (Fig. 2C and fig. S8). When dilp8 was diminished in whole Bx>rpr larvae by dilp8MI00727 mutation or by tissue-specifically reducing dilp8 mRNA by 71% through RNA interference (RNAi) (Bx>rpr>dilp8-IR) (fig. S9), the delay in pupariation reverted from 46.2 ± 1.3 hours to 27.8 ± 2.9 hours (P < 0.001) and 29.1 ± 2.5 hours (P < 0.001), respectively (Fig. 2D).

Secondly, synchronized larvae were fed with the genotoxic agent ethyl methanosulfonate (EMS) administered from 72 hours after egg-laying (AEL), which produced a dose-dependent delay in pupariation (Fig. 2E), strong caspase activation in imaginal discs, yet only mild defects in adult structures (fig. S10), which is similar to that caused by DNA damage and repair following irradiation (14, 15). In imaginal discs of dilp8MI00727 larvae, eGFP highlighted the damage produced by EMS (Fig. 2F), and this response of dilp8 was dose-dependent (Fig. 2E), indicating that dilp8 is tightly associated with the extent of damage/regeneration. The endogenous dilp8MI00727 mutation shortened the delay induced by EMS by 44.03% ± 13.24 (P < 0.0001) (Fig. 2G), and dilp8 RNAi (tub>dilp8-IR) reduced this delay by 43.24% ± 9.36 (P = 0.0063) (fig. S11). Moreover, dilp8 depletion augmented the pupal lethality associated with exposure to EMS (fig. S11). Thus, DILP8 regulates the timing of pupariation in response to tumor and regenerative growth and increases survival after tissue insult.

We examined the expression of hormone genes regulating the larval-to-pupal transition in relation to DILP8. Cell death–induced damage by Bx>rpr delays metamorphosis inhibiting PTTH synthesis in the brain, a delay that is enhanced by the consumption of provitamin A (ß-carotenoids) in the diet (14). Down-regulating dilp8 attenuated the PTTH delay by some 12 to 24 hours (Fig. 3A), independently of retinoids (fig. S12), indicating that DILP8 is required to delay PTTH synthesis in damaged/regenerating tissues.

Fig. 3

DILP8 delays metamorphosis by regulating the expression of ecdysone biosynthetic genes. (A) PTTH expression relative to rp49 analyzed by means of quantitative RT-PCR in synchronous Bx>rpr larvae with or without dilp8 RNAi silencing induced by expression of inverted repeat (IR) transgene. The mRNA was isolated from approximately five larvae per time point and genotype. (B) dilp8 overexpression delayed pupariation (n = 3 to 4 tubes of ~20 larvae each per genotype). The cysteine-to-alanine mutation at residue 150 (C150A) renders a biologically inactive DILP8 protein. (C) dib (left) and phm (right) expression analyzed by means of quantitative RT-PCR in mRNA isolated from ring gland/brain complexes from synchronous larvae (118 hours AEL; n = 3 replicates, each with RNA isolated from ~20 ring gland/brains, mean ± SD two-tailed unpaired t test). (D) Pupariation time of tub>dilp8 larvae fed 20HE (0.5 mg/mL) or the ethanol (EtOH) as vehicle (n > 40 larvae per treatment). tub> indicates tubulin-Gal4/+ and serves as background control.

We next assessed whether DILP8 is sufficient to delay pupariation in the absence of growth abnormalities. Synchronized larvae overexpressing dilp8 driven by tub-Gal4 (tub>dilp8) initiated pupariation 55.9 ± 7.6 hours later than did control tub> or tub>dilp8C150A larvae that overexpress dilp8 mutated at a conserved cysteine (P < 0.001) (Fig. 3B). Compared with the damaged Bx>rpr animals, dilp8 overexpression caused delayed induction of transcription of the disembodied (dib) and phantom (phm) genes in the ecdysone synthesis cascade (Fig. 3C and fig. S13) without delaying PTTH (fig. S13). The delay in pupariation induced by DILP8 was overcome by feeding larvae 20HE (Fig. 3D), confirming that the effects of DILP8 were a consequence of reduced ecdysone production.

Damaged and regenerating larvae, or those with tumor discs, attain a wild-type size. Similarly, although tub>dilp8 larvae prolong their feeding (longer than the controls), they were no larger (Fig. 4A and fig. S14). However, this extended feeding made tub>dilp8 adults weigh more than controls (Fig. 4B and fig. S15).

Fig. 4

DILP8 reduces inter- and intra-individual phenotypic variations, reflecting greater developmental stability and robustness. (A) Growth rate curve derived by scoring >10 animals of each genotype and time point. Multiple pair-wise two-tailed unpaired t tests with Bonferroni correction showed no significant differences (P > 0.10). Arrowheads indicate the time of 100% pupation for each genotype. (B) Virgin female and male adults overexpressing dilp8 weigh more than controls (***P < 0.001, mean ± SD two-tailed unpaired t test with Bonferroni correction), although their overall size is the same (fig. S14). (C) Thor expression in imaginal discs from synchronous tub> or tub>dilp8 third-instar larvae relative to rp49 and dUba2, analyzed by means of quantitative RT-PCR (feeding larvae, ~115 hours AEL, n = 3 biological replicates, each with RNA isolated from imaginal discs of ~20 larvae per genotype, mean ± SD two-tailed unpaired t test). (D) Box-and-whiskers plot of the wing area of dilp8MI00727 and control of genetic background (RheaMI00296 and PepckMI00026) males (P < 0.0001, f test for unequal distributions). The three Mi{MIC} insertions were generated in the same genetic background (12). (E) Comparisons of the left and right wings of a dilp8MI00727 female. (F) Bar graphs of the FAi of the left and right wings of the genotypes indicated (20). Numbers in brackets in (D) and (F) are the wing pairs scored.

To attain correct final size despite their prolonged larval life span, DILP8 overexpression may also exert control on growth rates to prevent overgrowth. Hence, we quantified the transcription of Thor (d4E-BP), a direct target of the growth inhibitor FOXO, as surrogate measure for imaginal disc growth (1618). Thor expression was selectively up-regulated in tub>dilp8 imaginal discs (Fig. 4C), which is consistent with a slower imaginal disc growth. In contrast, Thor expression in the fat body (fig. S15) showed that insulin/insulin-like growth factor 1 (IGF-1) signaling was not generally impaired, as also evident through the analysis of dilp2 and dilp3 expression (fig. S16). Thus, DILP8 exerts a fundamental influence on an adaptive plasticity of both growth and maturation, either directly or via secondary signals.

In the absence of such plasticity, organisms would be incapable of adjusting the growth of distinct body parts to maintain their overall proportionality and left-right symmetry. Indeed, dilp8MI00727 animals pupate over an extended time scale and are more varied in size than controls sharing the same genetic background (Fig. 4D and fig. S17). Individually, dilp8MI00727 flies reared at 26.5°C display imperfect bilateral symmetry (Fig. 4E), and when intra-individual variation between the left and right wings was assessed by using the fluctuating asymmetry index (FAi) (19, 20), wing FAi was statistically significantly higher in dilp8MI00727 females than in w1118 (Fig. 4F). This higher asymmetry reflects lesser stability (19).

Collectively, our results suggest that DILP8—an insulin/IGF/relaxin–like hormone peptide (Fig. 1)—provides a signal that communicates the growth status of peripheral tissues in order to regulate developmental timing, population robustness, and individual developmental stability [detected by fluctuating asymmetry analysis (19)], as well as local responses to processes such as regeneration and cancer.

Supplementary Materials

Materials and Methods

Figs. S1 to S17

Table S1

References (2127)

References and Notes

  1. Materials and methods are available as supplementary materials on Science Online.
  2. Acknowledgments: We thank E. Hafen, L. Garcia-Alonso, B. Lakowski, and F. Heredia for their comments and suggestions; J. Colombani, D. S. Andersen, and P. Léopold for sharing unpublished data; and I. Gutierrez and E. Ballesta for technical help. V.M. is a Generalitat Valenciana fellow, and A.M.G. was supported by a Marie Curie Fellowship. This work was supported by grants from the Fundación Botin, the Ministerio de Ciencia e Innovación (BFU2009-09074 and MEC-CONSOLIDER/CSD2007-00023), Generalitat Valenciana (PROMETEO/2008/134), and the European Commission (HEALTH-F2-2008-201666) to M.D. The microarray data are deposited in the Gene Expression Omnibus (GEO) repository (accession GSE35471). A.M.G. and M.D. designed the research; A.G., A.M.G., V.M., E.C., and M.D. performed the experiments and analyzed the data. A.G., A.M.G., and M.D. wrote the manuscript.
View Abstract

Navigate This Article