Abstract
In Drosophila, key developmental transitions are governed by the steroid hormone ecdysone. A number of neuropeptide-activated signaling pathways control ecdysone production in response to environmental signals, including the insulin signaling pathway, which regulates ecdysone production in response to nutrition. Here, we find that the Membrane Attack Complex/Perforin-like protein Torso-like, best characterized for its role in activating the Torso receptor tyrosine kinase in early embryo patterning, also regulates the insulin signaling pathway in Drosophila. We previously reported that the small body size and developmental delay phenotypes of torso-like null mutants resemble those observed when insulin signaling is reduced. Here we report that, in addition to growth defects, torso-like mutants also display metabolic and nutritional plasticity phenotypes characteristic of mutants with impaired insulin signaling. We further find that in the absence of torso-like, the expression of insulin-like peptides is increased, as is their accumulation in insulin-producing cells. Finally, we show that Torso-like is a component of the hemolymph and that it is required in the prothoracic gland to control developmental timing and body size. Taken together, our data suggest that the secretion of Torso-like from the prothoracic gland influences the activity of insulin signaling throughout the body in Drosophila.
IN holometabolous insects such as Drosophila, pulses of the steroid hormone ecdysone regulate the timing of developmental transitions, including the metamorphic transition, and hence growth duration (Henrich et al. 1999; Mirth and Riddiford 2007). During the larval and pupal stages of development, ecdysone is produced and released by the major endocrine organ, the prothoracic gland (PG), in response to multiple environmental and developmental stimuli [for review see Danielsen et al. (2013)]. Accordingly, there are many complex cellular signaling pathways involved in coordinating the responses to these signals. A well-studied example is prothoracicotropic hormone (PTTH), a brain-derived neuropeptide that regulates the production and release of ecdysone in response to developmental cues (McBrayer et al. 2007). Prior to each larval molt, PTTH is secreted and activates the Torso (Tor) receptor tyrosine kinase (RTK) in the PG, which signals via the Ras/mitogen-activated protein kinase pathway to upregulate a set of ecdysone biosynthesis genes (Rewitz et al. 2009). Ablation of the PTTH neurons, or loss of function mutations in tor, prolongs the growth period between each developmental transition and results in an overall increase in body size (McBrayer et al. 2007; Rewitz et al. 2009).
Another critical signaling pathway that regulates ecdysone production is the evolutionarily conserved insulin signaling pathway, which acts in the PG to regulate ecdysone biosynthesis in response to nutrition (Caldwell et al. 2005; Colombani et al. 2005; Mirth et al. 2005). In particular, this pathway regulates larval growth rate and the timing of a developmental checkpoint known as critical weight, thereby controlling the timing of the onset of metamorphosis (Mirth et al. 2005; Koyama et al. 2014). In Drosophila, the insulin-like receptor (InR) is activated by a family of insulin-like peptides (dILPs) (Brogiolo et al. 2001). A subset of these (dILPs 2, 3, and 5) are expressed in neurons that innervate the corpora cardiaca, a group of cells neighboring the PG, and ablation of these neurons causes developmental delays and decreased body size (Rulifson et al. 2002).
We and others recently reported that mutant alleles of the Membrane Attack Complex/Perforin-like (MACPF) protein Torso-like (Tsl) exhibit a developmental delay phenotype indicative of defects in ecdysone production (Grillo et al. 2012; Johnson et al. 2013). Tsl is best known for its role in embryonic patterning, where it functions upstream of the Tor receptor to control its activity for patterning the embryo termini (Stevens et al. 1990; Savant-Bhonsale and Montell 1993; Martin et al. 1994). Unexpectedly, however, we reported evidence that Tsl does not act similarly with Tor in the PG to control developmental transitions and body size. Specifically, tsl and tor have opposing effects on body size, and the developmental delay phenotype observed in tsl;tor double mutants is strikingly enhanced compared to either mutation alone, suggesting an additive rather than epistatic interaction (Johnson et al. 2013).
The growth defects of tsl mutants more closely resemble those observed when insulin signaling is reduced; however, Tsl has not previously been implicated in the insulin signaling pathway. Furthermore, it is not clear whether the growth and developmental timing phenotypes of tsl null mutants are due to a role for Tsl in the PG. Here, we report that in addition to growth defects, tsl mutants also display several other physiological and biochemical characteristics of impaired insulin signaling. We further show that Tsl is required in the PG to control developmental timing and body size, and that it influences the expression of dILPs and their accumulation in the insulin-producing cells (IPCs). Finally, we show that Tsl is present in the larval hemolymph, strongly supporting the idea that Tsl is secreted from the PG into circulation, where it acts to regulate systemic insulin signaling.
Materials and Methods
Drosophila stocks
The following stocks were used: w1118 (BL5905), chico1 (BL10738), Df(2L)BSC143 [chicodf (BL9503), a chromosomal deficiency that deletes the chico coding region], InRA1325D (BL8263, a constitutively active form of InR), and dilp2-3∆,dilp53 (BL30889) from the Bloomington Drosophila stock center; c7-Gal4, obtained from FlyView (Janning 1997); tsl∆, a null mutant of tsl (Johnson et al. 2013); phm-Gal4 (ch2) and UAS-dicerII; phm-Gal4-22, gifts from Michael O’Connor, University of Minnesota, Minneapolis (Ono et al. 2006); and UAS-tslRNAi and gHA:tsl, gifts from Jordi Casanova, Institute for Research in Biomedicine, Barcelona (Jimenez et al. 2002; Furriols et al. 2007). All flies were maintained at 25° on fly media containing, per liter: 7.14 g potassium tartrate, 0.45 g calcium chloride, 4.76 g agar, 10.71 g yeast, 47.62 g dextrose, 23.81 g raw sugar, 59.52 g semolina, 7.14 ml Nipagen (10% in ethanol), and 3.57 ml propionic acid.
Tsl constructs and generation of transgenic lines
To generate the UAS-Tsl:HA and UAS-Tsl:RFP constructs, the open reading frame of tsl followed by a short linker encoding the peptide SAGSAS and either three tandem hemagglutinin (HA) epitopes (for UAS-Tsl:HA) or the open reading frame for RFP (for UAS-Tsl:RFP) was synthesized and subcloned (Genscript) into pUASTattB via BglII and XhoI sites. To generate the phm:Tsl construct, a 1.1 kb fragment of the phm promoter region [from Ono et al. (2006)] was first cloned from genomic DNA (F – 5′-CTG CAG TGA TGC GCT GCT CCT TTG T-3′, R – 5′-AGA TCT CAC TTT CGA TTT CCT CCT GC-3′) into the pGEM-T Easy vector (Promega, Madison, WI), before being sequenced and subcloned into pUASTattB-Tsl:eGFP (Johnson et al. 2017) via PstI and BglII sites to delete the UAS sequence. Transgenic lines were made (BestGene) via ΦC31 integrase-mediated transformation (Bischof et al. 2007), using the ZH-51CE attP-landing site.
Developmental timing and body size analysis
At 24 hr after a 4 hr lay on apple juice agar supplemented with yeast paste, first-instar larvae were sorted by GFP (on a balancer chromosome) into 8–10 groups of 15 or 20 individuals (depending upon experiment) per genotype. Larvae were placed into vials containing fly media (see recipe above) and scored every 8 hr for the time taken to reach pupariation. Following their eclosion, adult flies were sorted by sex and weighed in groups on a microbalance (Mettler Toledo).
Nutritional plasticity
At 24 hr after a 4 hr lay on apple juice agar supplemented with yeast paste, first-instar larvae were sorted by GFP (on a balancer chromosome) and placed into vials containing either standard fly media or one of three low nutrient diets (either 50, 25, or 10% nutrients of standard media). These diets were made by diluting standard fly media with 0.5% non-nutritional agar to the appropriate concentration. Adult flies were collected within 24 hr of eclosion, sorted, and weighed in groups on a microbalance (Mettler Toledo). For each genotype, 10 replicates of 15 larvae were raised on each food type. Because size increases exponentially with increasing nutritional quality, male and female weight data were log10 transformed and analyzed by fitting the log10 transformed weights with linear models, using food concentration and genotype as fixed effects, in R-studio. Significant interactions between food concentration and genotype on body weight indicates that the two genotypes show significant differences in nutritional plasticity for body weight.
Quantification of food intake
Early feeding third-instar larvae were transferred to fresh dyed food (4.5% blue food dye) and allowed to feed for 1 hr. After feeding, larvae were removed from food using 20% sucrose solution, washed in distilled water and dried. Replicates of 10 larvae were homogenized in 80 µl of cold methanol and centrifuged for 10 min at 4°. A total of 60 µl of supernatant from each sample was analyzed in a spectrophotometer at 600 nm. As standards, a twofold dilution series of food dye, using a starting concentration of 4 µl dye/ml methanol was used. Five biological replicates were analyzed per genotype.
Hemolymph glucose and trehalose measurements
Hemolymph was pooled from 15 wandering third-instar larvae to obtain duplicate samples of 1 µl for assay. Five biological replicates were performed per genotype. Glucose was measured by adding 99 µl of Thermo Infinity Glucose Reagent (Thermo Scientific) to each sample and processing as per the manufacturer’s instructions. Trehalose was measured using the same reagent after digestion to glucose using trehalase, with a 10-fold dilution due to higher levels of trehalose compared to glucose. For trehalose digestion, 1 µl of hemolymph was incubated in 25 µl of 0.25 M sodium carbonate at 95° for 2 hr, cooled to room temperature, and 8 µl of 1 M acetic acid and 66 µl of 0.25 M sodium acetate (pH 5.2) were added to make digestion buffer. A total of 1 µl porcine trehalase (T8778; Sigma, St. Louis, MO) was added to 40 µl of this mixture and incubated at 37° overnight. The resulting glucose was analyzed using 10 µl of reaction and 90 µl Thermo Infinity Glucose Reagent as above. Glucose and trehalose standards were treated together with samples to quantify sugar levels in hemolymph.
Whole body triglyceride measurements
Triglycerides were quantified in whole wandering third-instar larvae as per Musselman et al. (2011). Ten larvae were homogenized in PBS + 0.1% Tween and then diluted 1:100 with PBS. Samples were heated for 5 min at 65° to inactivate lipases and 2 µl of each sample was mixed with 198 µl Thermo Infinity Triglyceride Reagent (Thermo Scientific) and processed as per the manufacturer’s instructions. The absorbance of samples at 500 nm was used as a relative measure of triglyceride content and was normalized to larval weight. Five biological replicates of 10 larvae were analyzed in duplicate for each genotype.
Immunoblotting
Hemolymph was extracted from ∼80 wandering third-instar larvae on ice. Following centrifugation at 16,000 × g for 5 min at 4°, supernatant was heat-inactivated at 60° for 10 min, re-spun and the remaining supernatant was combined with 1 mM DTT, 10 mM NaF, and complete EDTA-free protease inhibitor cocktail (Roche). For phosphorylated Akt (pAkt) blots, five third-instar larvae were homogenized in 80 µl of lysis buffer [50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2.5 mM EDTA, 0.2% Triton X, 5% glycerol, complete EDTA-free protease inhibitor cocktail (Roche)] and spun at 500 × g for 5 min at 4°. Reducing buffer (containing 6 M urea for Tsl immunoblots) was added to all samples before boiling and separation by SDS-PAGE (any kDa TGX; Bio-Rad, Hercules, CA) followed by transfer onto an Immobilon-P membrane (Millipore, Bedford, MA). Membranes were probed with either 1:1000 anti-HA (12CA5; Roche), 1:1000 anti-phosphorylated Drosophila Akt (4054S; Cell Signaling), or 1:1,000,000 anti–α-tubulin (B-5-1-2; Sigma), washed and incubated with HRP-conjugated secondary antibody (1:10,000; Southern Biotech). Immunoblots were developed using ECL Prime (GE Healthcare) and imaged using a chemiluminescence detector (Vilber Lourmat). pAkt blot images were quantified using ImageJ and differences between genotypes were determined by unpaired t-tests from five biological replicates.
Immunostaining and fluorescence quantification
Newly molted L3 larvae were collected and allowed to age on standard media for 24 hr before brains were dissected and fixed in PBS containing 4% paraformaldehyde for 40 min. Tissues were extensively washed in PBS containing 0.3% Triton X-100 and then blocked for 1 hr in PBT containing 2% normal goat serum (Sigma-Aldrich). Primary antibodies [rat anti-dILP2 and rabbit anti-dILP5; gifts from Dr. Pierre Léopold (Géminard et al. 2009)] were diluted to 1:800 in fresh block solution and incubated overnight at 4°. After extensive washing with PBS containing 0.3% Triton X-100, secondary antibodies (anti-rat Alexa Fluor 488 and anti-rabbit Alexa Fluor 568 conjugated; Molecular Probes, Eugene, OR) were diluted to 1:500 and incubated overnight at 4°. Brains were mounted in Fluoromount-G mounting medium (Southern Biotech) and Z series of the IPCs were obtained using a spinning disk confocal microscope (CV1000; Olympus), maintaining a 1 µm step size and identical imaging settings across all genotypes. ImageJ software was used to generate maximum projection images of the Z stacks and to quantify total fluorescent intensity across the IPCs. This was achieved by drawing an area of interest around each group of IPCs and calculating the raw grayscale values in this region of interest. The total fluorescence was normalized to IPC area to account for the size discrepancy between genotypes.
dILP gene transcript quantification
For each biological replicate, 10–15 third-instar larvae (anterior end only) or 20–25 dissected third-instar larval brains were snap frozen before RNA was extracted using TRIsure reagent (Bioline) and treated with DNAse (Promega). Complementary DNA was synthesized using Tetro reverse transcriptase (Bioline) by priming either 5 µg (for anterior ends) or 1 µg (for dissected brains) of RNA with oligo (dT) and random hexamers. Quantitative PCRs were performed in triplicate on a Light Cycler 480 (Roche), using SensiMix SYBR (Bioline) and primers specific for dilp2 (forward: 5′-ACG AGG TGC TGA GTA TGG TGT GCG-3′; reverse: 5′-CAC TTC GCA GCG GTT CCG ATA TCG-3′), dilp5 (forward: 5′-TGT TCG CCA AAC GAG GCA CCT TGG-3′; reverse: 5′-CAC GAT TTG CGG CAA CAG GAG TCG-3′), and Rp49 (forward: 5′-GCC GCT TCA AGG GAC AGT ATC T-3′; reverse: 5′-AAA CGC GGT TCT GCA TGA G-3′). Fold changes relative to Rp49 were determined using the ΔΔCT method and means and SE calculated from three to five biological replicates per genotype.
Data and reagent availability statement
Data and reagents are available upon request. The authors state that all data necessary for confirming the conclusions presented in the article are represented fully within the article.
Results
In addition to its role in regulating growth and developmental timing, the insulin signaling pathway is also critical for regulating glucose and lipid metabolism in Drosophila [for review see Garofalo (2002)]. Thus, in addition to growth defects, mutants with reduced insulin signaling are unable to regulate their blood sugar levels (Rulifson et al. 2002). This results in increased levels of glucose in the hemolymph (but not of trehalose, a glucose disaccharide that is synthesized from intracellular glucose in the fat body and secreted into circulation), and increased triglyceride content (Böhni et al. 1999; Rulifson et al. 2002; Ugrankar et al. 2015). To determine if tsl mutants also have such defects we performed metabolic analyses. This revealed that tsl null mutant (tsl∆) larvae have significantly elevated hemolymph glucose levels (Figure 1A, P = 0.0002), despite consuming less food than heterozygous controls over a 1 hr period (Supplemental Material, Figure S1 in File S1, P = 0.0040). By contrast, the concentration of circulating trehalose was unaltered in tsl∆ larvae (Figure 1B, P = 0.7232). In addition to the observed increase in circulating glucose, tsl∆ larvae had increased triglyceride content per milligram of body weight compared to heterozygous controls (Figure 1C, P = 0.0040). Taken together, the metabolic phenotype of tsl∆ larvae is consistent with previous studies on chico and other insulin pathway mutants (Böhni et al. 1999; Rulifson et al. 2002; Ugrankar et al. 2015), and supports the idea that Tsl regulates the insulin signaling pathway.
torso-like null mutants phenocopy mutants with reduced insulin signaling. (A–C) tslΔ larvae have significantly elevated hemolymph glucose levels (A, P = 0.0002), unaltered hemolymph trehalose levels (B, P = 0.7232), and increased total triglyceride content (C, P = 0.0040) compared to heterozygous controls (tslΔ/+). n = 5 groups of 10 larvae for all means. (D and E) The variation of both male (D) and female (E) adult body weight over different food concentrations is significantly smaller for tslΔ animals compared to heterozygous controls (tslΔ/+). Regression lines that differ significantly in their slopes, indicating differences in nutritional plasticity for body size between genotypes, are marked with different letters. n = 6–10 groups of at least three individuals for each food type. (F) tslΔ larvae show reduced levels of pAkt. (G) Levels of pAkt were quantified from four biologically independent experiments, using Tubulin as a loading control. pAkt/Tubulin densities were standardized by fixing the values of tslΔ/+ to 1. For all bar graphs, error bars represent ±1 SEM and genotypes sharing the same letter indicate that they are statistically indistinguishable from one another (P < 0.05, two-tailed t-tests). The data used to generate each graph can be found in Supplemental Data File 1 in File S3.
Insulin signaling is also required for coupling nutrition and growth, such that body size is adjusted according to nutritional availability (Tang et al. 2011). For example, in wild-type flies kept under low nutritional conditions insulin signaling is downregulated, resulting in a reduced larval growth rate and decreased adult body size. Accordingly, mutants with impaired insulin signaling are unable to adjust their body size in response to nutrition to the same extent as wild-type flies, as signaling is downregulated even in highly nutritious environments (Tang et al. 2011). To determine whether tsl mutants also share this feature with insulin signaling mutants, we examined their adult body size when grown as larvae on foods with varying nutrient content. Diluting the larval diet to 50, 25, and 10% of control food resulted in progressively smaller adults for both male and female heterozygous controls (Figure 1, D and E). At the lowest food concentration tsl∆ animals showed little difference in body size compared to heterozygous controls. However, as the food concentration increased, tsl mutants exhibited significantly reduced plasticity for body size (Figure 1, D and E; P = 0.0104 for males, P = 0.0038 for females; Table S1 in File S2) as would be expected if insulin signaling is impaired. The reduction in plasticity for tsl∆ animals was not found to be as severe as that observed in chico mutant animals, which appeared smaller across all food concentrations (Figure S2 in File S1, P < 0.0001 for both males and females; Table S2 in File S2).
Next, we used immunoblotting to measure the levels of pAkt, a biochemical readout of insulin signaling pathway activity. Consistent with a reduction in insulin signaling, we found that tsl∆ larvae had significantly lower pAkt levels compared to heterozygous controls (Figure 1, F and G, P = 0.0440). Taken together, these data show that many aspects of the tsl mutant phenotype parallel those seen in insulin signaling mutants, supporting the idea that Tsl is required for insulin signaling in response to nutrition.
To provide further evidence that Tsl acts in the insulin signaling pathway, we conducted genetic interaction studies. We first asked if the delay caused by loss of tsl is epistatic or additive to that caused by loss of the dILPs. Consistent with previous studies (Grönke et al. 2010), removing dILPs 2, 3, and 5 resulted in a severe developmental delay, with a delay of ∼341 hr compared to heterozygous controls (Figure 2A, P < 0.0001). Loss of tsl alone resulted in an 18 hr delay (Figure 2A, P < 0.0001). In larvae mutant for tsl and dilps 2, 3, and 5, the observed developmental delay was similar to the delay seen for dilp2-3∆,dilp53 mutants alone (Figure 2A, P = 0.2192). This result suggests that Tsl and dILPs 2, 3, and 5 act via the same signaling pathway to regulate developmental timing.
Torso-like genetically interacts with the insulin signaling pathway. (A) Larvae deficient for both tsl and dilps 2, 3, and 5 (dilp2-3Δ, dilp53, tslΔ) show a similar delay in time to pupariation (∼325 hr) to loss of dilps 2, 3, and 5 alone (P = 0.2192). (B) The reduced time to pupariation caused by expression of InRCA in the prothoracic gland (∼42 hr, P < 0.0001) is not suppressed by removal of tsl (P = 0.5015). Error bars represent ±1 SEM for all graphs. Genotypes sharing the same letter indicate that they are statistically indistinguishable from one another (P < 0.05, ANOVA and pairwise t-tests). n ≥ 10 for all means and ≥37 individuals were tested per genotype. The data used to generate each graph can be found in Supplemental Data File 1 in File S3. hAEL, hr after egg lay.
We next overexpressed a constitutively active and ligand-independent form of InR (InRCA) in the PG (using phm-Gal4) and asked whether Tsl is required for its function. We chose to manipulate InR activity specifically in the PG because it is well established as the key tissue involved in the InR-mediated regulation of ecdysone production (Mirth et al. 2005), and because we know that tsl is expressed there (Grillo et al. 2012). As expected (Walkiewicz and Stern 2009), overexpression of InRCA in the PG markedly reduced the time to pupariation (Figure 2B, P < 0.0001). This phenotype was not suppressed by loss of tsl (Figure 2B, P = 0.5015), suggesting that Tsl activity is not required for insulin signaling downstream of InR in the PG. Taken together, the results of these two genetic interaction experiments further support the idea that Tsl acts in the insulin signaling pathway and, if so, it does so either upstream or at the level of InR.
In the early embryo our work has led us to hypothesize that Tsl is required for the secretion of the Tor ligand, Trunk (Henstridge et al. 2014; Johnson et al. 2015). We therefore reasoned that it might regulate secretion of the dILPs, the ligands for InR. In mutants that affect dILP secretion an accumulation of dILP2 and dILP5 is observed in the IPCs (Géminard et al. 2009; Rajan and Perrimon 2012; Sano et al. 2015; Koyama and Mirth 2016). We therefore immunostained the IPCs in tsl∆ larvae for both dILP2 and dILP5. This revealed that tsl∆ larvae had a significant increase in the accumulation of both dILP2 (Figure 3, A and B; P = 0.0019) and dILP5 (Figure 3, C and D; P < 0.0001) in the IPCs compared to controls, with dILP2 accumulating to a lesser extent than dILP5.
torso-like mutants show increased insulin-like peptide expression and accumulation in the insulin-producing cells. (A–D) Removal of tsl increases the accumulation of dILP2 (A and B) and dILP5 (C and D) in the IPCs. dILP protein levels were standardized by fixing the values of w1118 to 1. n ≥ 35 for all means. (E and F) Expression of dilp2 (E, P = 0.0240) and dilp5 (F, P = 0.0016) is significantly increased in tslΔ larvae compared to controls (tslΔ/+). (G and H) dilp2 expression in the larval brain is not significantly altered in tsl mutants (G, P = 0.4243); however, expression of dilp5 is significantly elevated (H, P = 0.0009). Expression levels were normalized using an internal control, Rp49, and then standardized by fixing the values of tslΔ/+ larvae to 1. n = 3–5 for all means and ≥75 individuals were tested per genotype. For all graphs, error bars represent ±1 SEM and genotypes sharing the same letter indicate that they are statistically indistinguishable from one another (P < 0.05, ANOVA and pairwise t-tests). The data used to generate each graph can be found in Supplemental Data File 1 in File S3.
Although the observed accumulation of dILP2 and dILP5 could reflect a defect in their secretion, it was also possible that the observed accumulation reflects elevated expression of these peptides. Increased dilp2/5 expression is commonly observed when there is a systemic reduction in insulin signaling caused by insulin resistance in peripheral tissues (Musselman et al. 2011; Pasco and Leopold 2012). To determine whether the observed accumulation of dILP2 and dILP5 results from their elevated expression we quantified dilp2 and dilp5 messenger RNA levels. This showed that the expression of both dilp2 (Figure 3E, P = 0.0240) and dilp5 (Figure 3F, P = 0.0016) was elevated in tsl∆ larvae compared to heterozygous controls. As low levels of dilp expression have previously been detected in larval tissues other than the IPCs (Brogiolo et al. 2001), we also quantified dilp2 and dilp5 messenger RNA levels specifically in the larval brain. We found no significant difference in the expression of dilp2 (Figure 3G, P = 0.4243) in tsl∆ brains compared to heterozygous controls; however, expression of dilp5 was significantly elevated (Figure 3H, P = 0.0009). Taken together, these findings suggest that Tsl influences dilp expression during larval development.
We next asked if the role of Tsl in insulin signaling is due to a function in the PG. Previously, Grillo et al. (2012) showed that tsl is expressed in the PG and that RNA interference (RNAi) knockdown of tsl specifically in this tissue results in a significant developmental delay. Although we were unable to reproduce this result using the publicly available RNAi lines (Johnson et al. 2013), when we used the same RNAi line used in the Grillo et al. (2012) study [originally generated by Furriols et al. (2007)], we did observe a developmental delay phenotype (Figure 4A, P = 0.0018). By contrast, no phenotype was observed when we knocked down tsl expression specifically in the fat body using c7-Gal4 (Figure 4B, P = 0.9201).
Torso-like is required in the prothoracic gland to regulate both developmental timing and body size. (A) Knockdown of tsl specifically in the prothoracic gland (using phm-Gal4) results in a significant developmental delay (∼11 hr, P = 0.0018) that is similar to the delay observed for tslΔ. (B) No developmental delay is observed when tsl is knocked down specifically in the fat body (using c7-Gal4). (C and D) Expression of an UAS-Tsl:RFP (UAS-tsl) transgene specifically in the prothoracic gland (using phm-Gal4) rescues both the developmental delay (C) and reduced adult body size (D) of tsl∆ homozygotes (P < 0.0001 for both delay and body size compared to tsl∆). (E and F) The developmental delay (C) and reduced adult body size (D) or tsl mutants is partially rescued by the phm:Tsl:3Myc:eGFP (phm:Tsl) construct (P < 0.0001 for both delay and body size compared to tsl∆). Error bars represent ±1 SEM for all graphs. Genotypes sharing the same letter indicate that they are statistically indistinguishable from one another (P < 0.05, ANOVA and pairwise t-tests). n ≥ 10 for all means and ≥37 individuals were tested per genotype. The data used to generate each graph can be found in Supplemental Data File 1 in File S3. hAEL, hr after egg lay.
We also asked if PG-specific expression of tsl could rescue the growth defects of tsl mutants. When we expressed an UAS-Tsl:RFP transgene (UAS-Tsl) in the PG using phm-Gal4, we found this completely rescued both the developmental delay (Figure 4C, P < 0.0001) and small body size (Figure 4D, P < 0.0001) of tsl mutants. However, this transgene also partially rescued the tsl∆ phenotypes in the absence of the Gal4 driver, most likely due to leaky transgene expression. To overcome this problem, we generated a genomic rescue construct in which the phm promoter sequence (from phm-Gal4) was fused to the tsl coding sequence C-terminally tagged with three tandem Myc epitopes and the eGFP coding sequence (phm:Tsl). We found that this transgene partially rescued both the developmental delay (Figure 4E, P < 0.0001) and small body size (Figure 4F, P < 0.0001) of tsl null mutants. Together with the RNAi experiments, these data strongly suggest that tsl expression is required in the PG for regulating developmental timing and body size. However, we are unable to rule out the possibility that tsl is also required in other tissues for these roles.
How might Tsl function in the PG to regulate systemic insulin signaling? Because Tsl is a secreted protein, this could be explained if Tsl is secreted from the PG into the hemolymph. We therefore asked if Tsl is found in the larval hemolymph. To do this we used a genomic rescue construct that carries ∼3 kb of promoter and the tsl coding sequence N-terminally tagged with three tandem HA epitopes (gHA:Tsl; Jimenez et al. 2002). This construct has previously been shown to completely rescue the developmental delay and reduced body size of tsl∆ animals (Johnson et al. 2013). Using immunoblotting, we were able to clearly detect gHA:Tsl in the larval hemolymph (Figure 5A). We further asked if PG-produced Tsl enters the hemolymph by expressing a functional C-terminally tagged Tsl transgene (UAS-Tsl:HA) in the PG (phm-Gal4) and performing Western blots on protein extracted from larval hemolymph. This revealed that Tsl:HA protein was present in the hemolymph (Figure 5B). Although this is an overexpression situation, we reason that because Tsl is endogenously expressed in the PG, it is likely that at least a proportion of the total Tsl in circulation originates from the PG. However, it remains possible that the Tsl we detect in the hemolymph with the genomic construct is produced and secreted from another tissue.
Torso-like is secreted from the prothoracic gland into the larval hemolymph. (A) Immunoblot (anti-HA) of Tsl expression in the larval hemolymph. (B) Tsl is also detected in the hemolymph via immunoblotting (anti-HA) when expressed in the prothoracic gland, using phm-Gal4.
Discussion
Our data presented here provide compelling evidence that Tsl is secreted into the hemolymph and regulates growth and developmental timing via the insulin signaling pathway. Although the tsl mutant phenotypes described here closely resemble those observed when insulin signaling is reduced in the entire organism, it should be noted that loss of tsl has a less severe effect on the pathway compared to mutations in other genes. For example, loss of function mutations in InR are homozygous lethal, and only a few heteroallelic combinations produce viable adults in which growth defects can be observed (Chen et al. 1996). By comparison, mutations in the adaptor protein Chico do not result in lethality, but rather cause severe growth and metabolic defects (Böhni et al. 1999). Here, we find that the defect in nutritional plasticity for body size is not as severe in tsl mutants as it is in chico mutants. Our findings therefore suggest that Tsl regulates, but is not essential for insulin signaling.
How might Tsl regulate insulin signaling throughout the body? One possibility is that Tsl regulates the insulin response in all tissues by acting in conjunction with InR (Figure 6A). Loss of insulin response in tsl mutants could then result in the increased dilp expression that we observe. Alternatively, Tsl may act to influence the activity of the dILPs, which in turn regulate systemic insulin signaling. There are two main tissues that are known to regulate dILP activity [for a complete review of the regulation of dILP production and secretion see Nässel and Vanden Broeck (2016)]. One is the fat body, which produces and releases important regulators of dILP secretion in response to intracellular nutrient levels (Figure 6B; Géminard et al. 2009; Rajan and Perrimon 2012; Sano et al. 2015; Delanoue et al. 2016; Koyama and Mirth 2016). Interestingly, a recent study found that knocking down tor specifically in the fat body led to a decreased body size, leading the authors to suggest that Tor acts in the fat body to influence insulin signaling via an unknown mechanism (Jun et al. 2016). Given the key role of Tsl in the regulation of Tor activity during early embryogenesis, it is therefore possible that Tsl acts in the Tor pathway in the fat body. However, fat body-specific knockdown of tor does not result in a developmental delay (Jun et al. 2016), thus this would seem unlikely to be the only role for Tsl in regulating growth and developmental transitions. In addition, experiments that we have performed to detect Tsl or knockdown its expression in the fat body have not provided any evidence of Tsl expression or function in this tissue. Determining if Tsl acts in the fat body to regulate dILP activity, either with Tor or with another pathway, requires further investigation.
Possible mechanisms of Torso-like function from the prothoracic gland. (A) Tsl is secreted from the PG into circulation where it acts in conjunction with InR in all tissues to modulate insulin signaling. (B and C) Tsl influences the activity of the dILPs in the IPCs, which in turn regulate systemic insulin signaling. In this model, Tsl could act on either (B) the fat body (FB), which is responsible for producing and releasing important regulators of dILP secretion (R, brown oval) such as CCHamide-2, Unpaired-2, Growth-blocking peptides 1 and 2 and Stunted, or (C) the IPCs to directly control dILP activity.
An alternative possibility is that Tsl acts directly on the IPCs to control the activity of the dILPs (Figure 6C). Given the close proximity of the IPCs to the PG, this perhaps fits better with the known role of Tsl in the early embryo, where it is secreted from the follicle cells and acts locally on Tor signaling (Jimenez et al. 2002; Stevens et al. 2003). It is therefore possible that the systemic effects of Tsl are due to a role in regulating dILP expression and/or secretion in the IPCs. These ideas could be tested in future by experiments such as examining the kinetics of dILP secretion in tsl mutants following starvation, or testing if artificially stimulating dILP release can rescue the tsl mutant phenotype. Understanding the exact role of Tsl in this system will provide fundamental insights into the mechanisms that regulate the evolutionarily conserved insulin signaling pathway, as well as the role of MACPF proteins in developmental signaling events.
Acknowledgments
We thank Karyn Moore, Lauren Forbes Beadle, Katherine Shaw, and the Australian Drosophila Biomedical Research Facility for technical support; Jordi Casanova and Michael O’Connor for providing fly stocks; and Pierre Léopold for the dILP2 and dILP5 antibodies. M.A.H. is a National Health and Medical Research Council Early Career Fellow. This work was supported by an Australian Research Council grant to C.G.W., C.K.M., and T.T.
Author contributions: C.G.W. conceived the experiments, interpreted the data and led the work. M.A.H. conceived the experiments, performed the experiments and interpreted the data. L.A., T.K., and T.K.J. performed experiments. J.C.W., T.T., and C.K.M. interpreted the data. M.A.H. and C.G.W. wrote the manuscript with assistance from all authors.
Footnotes
Supplemental material is available online at www.genetics.org/lookup/suppl/doi:10.1534/genetics.117.300601/-/DC1.
Communicating editor: D. Greenstein
- Received December 6, 2017.
- Accepted February 12, 2018.
- Copyright © 2018 by the Genetics Society of America
