Dilp8 requires the neuronal relaxin receptor Lgr3 to couple growth to developmental timing

Abstract

How different organs in the body sense growth perturbations in distant tissues to coordinate their size during development is poorly understood. Here we mutate an invertebrate orphan relaxin receptor gene, the Drosophila Leucine-rich repeat-containing G protein-coupled receptor 3 (Lgr3), and find body asymmetries similar to those found in insulin-like peptide 8 (dilp8) mutants, which fail to coordinate growth with developmental timing. Indeed, mutation or RNA intereference (RNAi) against Lgr3 suppresses the delay in pupariation induced by imaginal disc growth perturbation or ectopic Dilp8 expression. By tagging endogenous Lgr3 and performing cell type-specific RNAi, we map this Lgr3 activity to a new subset of CNS neurons, four of which are a pair of bilateral pars intercerebralis Lgr3-positive (PIL) neurons that respond specifically to ectopic Dilp8 by increasing cAMP-dependent signalling. Our work sheds new light on the function and evolution of relaxin receptors and reveals a novel neuroendocrine circuit responsive to growth aberrations.

Figure 1. Mutation in the Drosophila relaxin receptor Lgr3 leads to increased fluctuating asymmetry.

(a) A neuroendocrine pathway coupling growth and developmental timing. Scheme adapted from Halme et al. (b) Remobilization of Minos element MB06848 positioned between exons 7 and 8 of the Lgr3 locus on chromosome III generated a 3.8-kb deletion, named Lgr3^ag1. Reverse transcriptase PCR assays followed by Sanger sequencing determined that the Lgr3^ag1 deletion leads to usage of an aberrant splicing acceptor within intron 7 and readthrough directly into exon 9 (Supplementary Fig. 2a–d). (c) Scheme of the predicted protein structure of the wild-type Lgr3 protein and the truncated Lgr3^ag1 protein based on vertebrate relaxin receptor structure data. Major domains are depicted: low-density lipoprotein receptor domain class A (LDLa), leucine-rich repeat (LRR) and seven transmembrane (7TM). The first TM/signal peptide (SP) domain is not predicted to be cleaved. The approximate region where ag1 mutation truncates the Lgr3 protein is depicted. (d) Bar graphs of the FAi of the area (left panel) and length (right panel) of the wing pairs of the genotypes indicated. Numbers (N) of the wing pairs scored. F-test P values are shown. Wing area: degrees of freedom (df)_(ag1)=51, df_(+/+)=20, F=3.23; Wing length: df_(ag1)=65, df_(+/+)=28, F=2.08.

Figure 2. Lgr3 couples imaginal disc growth to developmental timing by acting in the Dilp8 pathway.

(a) Mutation of Lgr3 abrogates interorgan communication between regenerating damaged discs (Bx>rpr) and the neuroendocrine centers coordinating the onset of metamorphosis. (b) Lgr3 acts in the Dilp8 pathway. Placing Lgr3^ag1 over a deficiency uncovering the Lgr3 locus suppresses the delay caused by ectopic dilp8 expression (arm>dilp8a). (c) RNAi against Lgr3 (Lgr3-IR) suppresses the delay caused by ectopic expression of either dilp8b or dilp8c transgenes under the control of the ubiquitous tub> driver. (a–c) Box plots (see Methods) showing pupariation time (Time after egg laying (AEL) in h) of (N) larvae obtained from six, two and six egg layings for panels (a–c) respectively. Whiskers are 5 and 95% percentiles. Dots, outliers. P<0.0001, Kruskal–Wallis test for all panels. Genotypes sharing the same letter (blue) are not statistically different at α=0.01, Conover post hoc test. Degrees of freedom, H and C values for the Kruskal–Wallis tests are df=5, H=666.45, C=1.28; df=2, H=54.16, C=0.99 and df=7, H=468.05, C=0.83 for panels (a–c), respectively.

Figure 3. Lgr3 is expressed in a subpopulation of CNS neurons.

(a) CRISPR/Cas9-mediated sfGFP knock-in strategy used to generate the Lgr3 protein reporter allele ag5 (named as sfGFP::Lgr3), which does not contain indels in the exon 2 region, and alleles ag6-9, which contain PTC+ indels (Supplementary Fig. 4). Two thin black lines indicate the sites of the CRISPR gRNAs used. (b) Lgr3 protein scheme depicting the approximate localization of the sfGFP insertion and the protein truncations caused by the PTC+ indel mutations. (c) Sum of confocal z-stack slices of the CNS of a third instar larva stained with anti-GFP (green) to show sfGFP::Lgr3 expression (green) and with anti-nc82 (magenta) and DAPI (blue) counterstains to show the synapses (neuropil) and nuclei, respectively. Arrows point to two bilateral pairs of PIL neurons (top), to the MIL neurons in the midline of the VNC (middle), and the distal VNC pair (bottom). Arrowheads point to the proximal projections of the PIL neurons. sfGFP::Lgr3 is also expressed in ∼170 other cell bodies, but at a lower level than in PIL and MIL neurons. f, oesophageal foramen, seg, subesophageal ganglion. See also Supplementary Fig. 4c for controls on the specificity of this staining pattern. (d) Box plots (see Methods) showing pupariation time of (N) larvae obtained from 11 egg layings. Whiskers are 5 and 95% percentiles. Dots, outliers. P<0.0001, Kruskal–Wallis test. Genotypes sharing the same letter (blue) are not statistically different at α=0.01, Conover post hoc test. Degrees of freedom, H and C values for the Kruskal–Wallis tests are df=7, H=397.84 and C=1.00. Scale bar, 50 μm.

Figure 4. Neuroanatomy of neuronal populations strongly expressing Lgr3.

(a) MIL neuron cell bodies (cb) located deep in the VNC project ventrally and anteriorly for a short distance (red arrowheads). Adjacent anterior projections are depicted (blue arrows). The left and right panels represent views of MIL neurons from different CNS preparations. The left panel is an inset of the sfGFP::Lgr3 channel of the CNS depicted in Fig. 3c. (b) Rostral view of the pars intercerebralis depicting the PIL neurons stained with anti-GFP (green (left) and black (right)) and counterstained with DAPI (magenta). Possible proximal dendritic arborizations are indicated with red arrows. Blue arrows indicate a ramification, likely axonic, with an undetermined terminus. Primary neurites (red arrowhead). (c) Overlap between Cha>myr::tdTomato and sfGFP::Lgr3 expression patterns. Single confocal slices showing the neurons expressing highest levels of Lgr3 (PIL, MIL and distal pair of VNC neurons, arrows in upper, middle and lower panels, respectively) labelled with sfGFP::Lgr3 (anti-GFP, green) and Cha> neurons were visualized with a UAS-myr::tdTomato reporter (magenta). (d) PIL neurons associate closely to IPCs. Z-stack projection of confocal stacks stained with anti-GFP (green) and anti-Dilp5 (magenta). (e) PIL neurons associate closely to Dilp7-producing DMA1 neurons in the pars intercerebralis. Z-stack projection and orthogonal view of confocal stacks stained with anti-GFP (green) and anti-Dilp7 (magenta). Scale bars, 20 μm.

Figure 5. Lgr3 is required in cholinergic neurons to couple growth and developmental timing.

(a) CNS-specific RNAi of Lgr3 rescues the EMS-induced delay. (b) Removal of Lgr3 in cholinergic neurons (using Cha>) rescues the EMS-induced delay. (c) Ring gland-specific RNAi of Lgr3 does not significantly affect pupariation time in EMS assays. (d) FAi in (N) animals expressing RNAi against Lgr3 in neurons using elav> driver (F-test). df_{(elav>Lgr3-IR)}=19, df_(Lgr3-IR)=23, F=4.71 and df_{(elav>Lgr3-IR)}=19, df_(elav>)=23, F=4.65. (e) Ectopic expression of Lgr3 is not sufficient to delay the onset of metamorphosis. Two UAS-Lgr3 transgene insertions (Lgr3a and Lgr3b) were tested. (f) FAi in (N) animals carrying mutations in Lgr3 and rescued with neuronal expression of Lgr3a. df_{(elav>;Lgr3(ag1))}=20, df_{(elav>Lgr3a;Lgr3-IR)}=23, F=3.41; df_{(elav>;Lgr3(ag1))}=20, df_{(elav>Lgr3(ag1/+)}=13, F=2.88. (a–c,e) Box plots (see Methods) showing pupariation time of (N) larvae obtained from six, two, four and six egg layings for a–c,e, respectively. Whiskers are 5 and 95% percentiles. Dots, outliers. P<0.0001, Kruskal–Wallis test for all panels, except for e where P=0.0413. Genotypes sharing the same letter (blue) are not statistically different at α=0.01, Conover post hoc test. Degrees of freedom, H and C values for the Kruskal–Wallis tests are df=5, H=55.06, C=0.99; df=3, H=150.64, C=0.99; df=3, H=37.31, C=0.99; df=2, H=6.32, C=0.90, for a–c,e, respectively.

Figure 6. A restricted subset of Lgr3 neurons relays the Dilp8 signal to the ring gland.

(a) Scheme of the Lgr3 locus depicting the regulatory elements tested in this study. (b) RNAi of Lgr3 in GMR19B09> neurons rescues the EMS-induced delay. (c) Silencing of GMR19B09> neurons by Kir2.1 expression suppresses the EMS-induced delay. (d) GMR19B09> (magenta) drives expression in sfGFP::Lgr3-positive neurons (green): PIL neurons (PIL, arrows); PIL proximal arborizations (PILarb, arrows) and other neurons (arrowheads). (e) PIL neurons (green) are a subset of #5 neurons (magenta) defined by MZ699>. (f) RNAi of Lgr3 in MZ699> neurons rescues the EMS-induced delay. (g) Intersectional pattern of GMR19B09-LexA driving lexAop-mCD8::GFP (green) and MZ699>mCD8::RFP (magenta). Cells expressing both drivers are depicted with arrows in the two overlapping max-intensity projections of confocal z-stack sections and as blue dots in the CNS cartoon to the right. (f) oesophageal foramen. seg, subesophageal ganglion. The midline is depicted with a dashed line. A more detailed image of the PIL neurons in separate green and magenta channels is available in Supplementary Fig. 13. (b,c,f) Box plots (see Methods) showing pupariation time of (N) larvae obtained from two egg layings for each panel. Whiskers are 5 and 95% percentiles. Dots, outliers. P<0.0001, Kruskal–Wallis test for all panels. Genotypes sharing the same letter (blue) are not statistically different at α=0.01, Conover post hoc test. Degrees of freedom, H and C values for the Kruskal–Wallis tests are df=3, H=83.48, C=1; df=3, H=177.75, C=0.97; df=3, H=156.71, C=0.99, for b,c,f, respectively. Scale bars, 50 μm (d,g); 20 μm (e).

Figure 7. Dilp8 activates cAMP signaling in PIL neurons, but not other Lgr3-positive neurons.

Maximum intensity projections of confocal z-stack slices of CNS preparations of larvae carrying a CRE-luciferase reporter (anti-luciferase, magenta) and sfGFP::Lgr3 (anti-GFP, green) in a control background (w¹¹¹⁸, upper panels) or following constitutive ectopic Dilp8 expression (tub>dilp8b background, lower panels). PIL neuron cell bodies (arrowheads) do not express detectable CRE-luciferase (anti-luciferase, magenta), except when activated by Dilp8 expression. Anti-luciferase staining in clock neurons (arrows) serves as an internal control. Background anti-luciferase staining in glial cells is also detectable throughout the CNS. f, oesophageal foramen. Asterisk labels PIL neuron proximal arborizations. Midline is labelled with a dashed line. Scale bar, 50 μm.

Figure 8. Determination of candidate cell surface receptors for Dilp8 and a multiple relay model of the Dilp8–Lgr3 developmental stability pathway.

(a) Scheme of the ligand–receptor capture assay (LRC). TriCEPS reagent coupled with Dilp8 (or with human insulin and glycine as positive and negative controls, respectively) was assayed in triplicate in Drosophila SL2/DL2 cells. (b) Volcano plots (false-discovery rate (FDR)-adjusted P values plotted against the fold changes (FC) between samples) comparing control TriCEPS-bound human insulin ligand against the glycine-quenched TriCEPS reagent control sample. The shaded area represents the receptor candidate space, defined by an enrichment factor greater than 4 and an FDR-adjusted P<0.05. (c) Volcano plots comparing TriCEPS-bound Dilp8 ligand against the glycine-quenched TriCEPS reagent control sample. The shaded area represents the receptor candidate space, defined by an enrichment factor of greater than a factor of 4 and an FDR-adjusted P<0.05. (d) Relative mRNA levels of different receptor genes in Drosophila SL2/DL2 cell line. Values represent the geometric mean±s.e.m. (n=3 biological repeats, except for Drl, where two biological repeats were made) of InR, Drl or Lgr3 mRNA levels relative to rp49 levels (rp49=100). (e) Cartoon depicting the Dilp8–Lgr3 abnormal tissue growth-sensing pathway. Dilp8 either signals directly onto Lgr3 neurons located in the CNS in a process where any of the other candidate direct Dilp8-binding proteins can act as a co-factor to Lgr3, or those receptors/co-receptors could play a role in relaying the Dilp8 signal from the periphery to the Lgr3 neurons in the CNS (Relay 1). The second Relay (Relay 2) would occur from the Lgr3-positive neurons to the ring gland.