Insulin signaling, lifespan and stress resistance are modulated by metabotropic GABA receptors on insulin producing cells in the brain of Drosophila

Abstract

Insulin-like peptides (ILPs) regulate growth, reproduction, metabolic homeostasis, life span and stress resistance in worms, flies and mammals. A set of insulin producing cells (IPCs) in the Drosophila brain that express three ILPs (DILP2, 3 and 5) have been the main focus of interest in hormonal DILP signaling. Little is, however, known about factors that regulate DILP production and release by these IPCs. Here we show that the IPCs express the metabotropic GABA(B) receptor (GBR), but not the ionotropic GABA(A) receptor subunit RDL. Diminishing the GBR expression on these cells by targeted RNA interference abbreviates life span, decreases metabolic stress resistance and alters carbohydrate and lipid metabolism at stress, but not growth in Drosophila. A direct effect of diminishing GBR on IPCs is an increase in DILP immunofluorescence in these cells, an effect that is accentuated at starvation. Knockdown of irk3, possibly part of a G protein-activated inwardly rectifying K(+) channel that may link to GBRs, phenocopies GBR knockdown in starvation experiments. Our experiments suggest that the GBR is involved in inhibitory control of DILP production and release in adult flies at metabolic stress and that this receptor mediates a GABA signal from brain interneurons that may convey nutritional signals. This is the first demonstration of a neurotransmitter that inhibits insulin signaling in its regulation of metabolism, stress and life span in an invertebrate brain.

Figure 1. The GABA_B receptor (GBR) is expressed on insulin-producing cells (IPCs).

A and B. Dilp2-Gal4-driven GFP in IPCs. The IPCs arborize in three regions: lateral branches in upper pars intercerebralis (Lup), median branches (Med) along median bundle above the central complex and in tritocerebrum (Tri in B). The axons projecting to the corpora cardiaca are not seen in this maximum projection. Cb, cell bodies of IPCs. C. GFP driven by GBR2-Gal4 (GABA_Breceptor2) displays the IPCs and additional neurons in the median neurosecretory cell group (arrow). Extensive arborizations can be seen in the pars intercerebralis, a smaller portion of which belong to the IPCs. D1-3. The IPCs colocalize DILP2 immunoreactivity (magenta) and GBR2-Gal4 expression (green). A set of MNCs below the IPCs express GBR2, but not DILP2 (asterisk). E1-3. Punctate immunolabeling with anti-GBR2 can be seen on the cell bodies of some of the IPCs (e. g. at arrows) and on some of the processes both marked with Dilp2Gal4-GFP (projection of 3 optical sections; for single section see Fig. S1B). One cell body expresses GBR immunolabel, but not Dilp2 (similar to cells marked with asterisk in 1D). F1-3. GBR2 immunlabeling of presumed IPC dendrites in the median region above the central complex (region corresponding to Med in 1A).

Figure 2. Insulin producing cells express OK107, but not GABA_A receptors or GABA.

A1-3. The enhancer trap Gal4 line OK107 (green) is expressed in the IPCs as seen by DILP2 immunolabeling (magenta). Some additional median neurosecretory cells express OK107, but not DILP2. B1-3. The GABA_A receptor subunit RDL is visualized here by an rdl-Gal4 driver (green). The rdl-Gal4 expressing cells are not immunolabeled with DILP2 antiserum (magenta). The rdl-positive cells resemble those that express GBR2- and OK107-Gal4, but not DILP2. C1-2. Antiserum to GABA (magenta) labels neuronal processes that superimpose (arrow) the IPCs shown in green (GBR2-Gal4-GFP). D1-2. The biosynthetic enzyme GAD1 is a good marker for GABAergic neurons. The DILP2 immunolabeled IPCs (magenta) do not express Gad1-Gal4 driven GFP, but several neurons can be seen adjacent to IPCs.

Figure 3. GABA_B receptor knockdown on insulin-producing cells diminishes lifespan.

Lifespan was determined for normally fed flies with GABA_B receptor diminished on IPCs (Dilp2-GBRi) compared to controls (Dilp2-w1118 and wildtype flies, wt). A slight, but significant, reduction of lifespan was seen in the Dilp2-GBRi flies (p<0.001 compared to wildtype flies; p<0.001 compared to GBRi-w1118, Log Rank test; n = 82–91 for the different genotypes).

Figure 4. GABA_B receptor knockdown on insulin-producing cells affects DILP levels.

A. Relative DILP immunofluorescence in IPCs in fed and starved flies with and without GABA_B receptor knockdown (GBRi) in IPCs (Dilp2-Gal4/UAS-gbr2-RNAi). The DILP antiserum used is likely to cross react with DILP2, 3 and 5 . Control flies (GBRi-w1118) display significantly lower levels of DILP-immunofluorescence than the flies with GBRi (Dilp2-GBRi), both in fed flies (p<0.001; Anova with Tukey's comparison) and after starvation (p<0.001). A smaller, but significant, increase in DILP fluorescence is seen at starvation for both genotypes (p<0.05 in both cases). 19–35 cells were measured in 5 specimens of each genotype. B–E. representative confocal images of DILP-immunolabeled IPC of the fed and starved control (GBRi-w1118) and experimental (Dilp2-GBRi) flies.

Figure 5. GABA_B receptor knockdown on IPCs increases sensitivity to starvation and desiccation.

We tested GABA_B receptor knock-down with two Gal4 drivers, compared to parental controls, in starvation (flies kept on aqueous agarose) and desiccation (no food and no water). All experiments were run in at least three replicates, unless specified. A. Using a Dilp2-Gal4 driver to knock down the GABA_B receptor (Dilp2-GBRi) we obtained flies that display significantly reduced survival at starvation (p<0.001, Log rank test; n = 210–270 for each genotype). B. The same genotypes were tested for survival during desiccation. (p<0.001; n = 120 for each genotype, 2 replicates). C. The OK107 Gal4 driver is also expressed in IPCs and was used for GABA_B receptor knock-down (OK107-GBRi). At starvation survival is significantly decreased in OK107-GBRi flies (p<0.001; n = 155–193). D. The same genotypes were tested at desiccation. Again, a significant reduction was seen after receptor knockdown (p<0.001; n = 80–120; 2 replicates). E. As a control for the OK107 driver, that includes mushroom body Kenyon cells, we utilized a distinct driver for Kenyon cells (MB247) that is not expressed in median neurosecretory cells (IPCs). MB247-driven GBRi does not affect survival at starvation compared to the two parental controls (n = 100–184 for each genotype). F. We have no evidence for expression of the GABA_A receptor subunit RDL in IPCs. Driving Rdl-RNAi in IPCs with the cross Dilp2-RdlRi did not alter survival at starvation. We found no significant difference between the three genotypes (n = 136–181; 2 replicates).

Figure 6. Knockdown of an inward rectifying K-channel mimics GABA_B receptor knockdown.

Two different UAS-Irk3-RNAi lines (Irk3Ri^KK and Irk3Ri^GD) were crossed to Dilp2-Gal4 flies and tested for survival at starvation. Both crosses resulted in flies that survived significantly shorter than parental controls. A. Dilp2-Irk3Ri^KK flies (p<0.001; n = 169–180 for each genotype; three replicates). B. Dilp2-Irk3Ri^GD flies (p<0.001; n = 140–158; 3 replicates).

Figure 7. GABA_B receptor knockdown affects levels of trehalose and lipid at starvation.

A. Trehalose levels were determined in fed flies and flies starved for 5 and 12 h (0, 5, 12 in bars) in flies with GABA_B receptor knockdown in IPCs (Dilp2-GBRi) and parental controls. In controls the trehalose levels gradually diminish (significantly) after 5 and 12 h starvation, whereas in GBRi flies there is no significant difference between 5 and 12 h starvation (n = 120 for each genotype). Two way Anova [ns, not significant (p>0.05), * p<0.05, ** p<0.01, *** p<0.001]. B. Lipid levels were determined in flies of the same genotypes after 0, 12 and 24 h starvation. In controls there is a significantly more drastic decrease in lipids between 12 and 24 h, whereas in GBRi flies there is a much more drastic (and significant) decrease between 0 and 12 h starvation (n = 120 for each genotype). Two-way Anova [ns, not significant (p>0.05), * p<0.05, ** p<0.01, *** p<0.001].