Drosophila Insulin Pathway Mutants Affect Visual Physiology and Brain Function Besides Growth, Lipid, and Carbohydrate Metabolism

Flies were cultured under standard conditions (25°C, 50% relative humidity, 12/12-h light/dark cycles). Fly stocks used were InR^E19/TM3 (chemically induced hypomorphic mutation not in the coding sequence), InR⁰⁵⁵⁴⁵/TM3 (transposable element insertion in the first coding exon), InR^3T5/TM3, chico¹/CyO (transposable element insertion in the first coding exon), Dp110^2H1/TM6B, Dp110^5W3/TM6B, Dp110^A/TM6 (deletion of the 3′ end of the coding sequence), PKB¹/TM3 (point mutation in the kinase catalytic domain F327I), PKB³/TM3 (point mutation in the coding sequence G99S), dRheb^7A1/TM6B (point mutation in the coding sequence V71 K), dRheb^2D1/TM6B, dRheb^44.1/TM6B (0.6-Kb insertion in the 5′−coding sequence), dS6K^I-1/TM3 (5′−deletion in the coding sequence), and dS6K^P1713/TM3 (transposable element insertion) obtained from E. Hafen (ETH, Zurich, Switzerland). The mutant alleles and balancer chromosomes (CyO, TM3, and TM6B) used in this study are described in FlyBase (http://flybase.org) and the references therein, and with the exception of InR^3T5 and dS6K^P1713, all are characterized and published bona fide mutants.

Homozygous mutant flies for chico, Dp110, and Rheb were made by crossing heterozygous chico¹/CyO, Dp110^2H1/TM6B, Dp110^5W3/TM6B, and dRheb^7A1/TM6B inter se, respectively, and selecting progeny flies without balancer chromosomes. Because mutant alleles of other insulin pathway loci are homozygous embryonic lethal, we used flies with the following viable heteroallelic combinations: InR^E19/InR⁵⁵⁴⁵, InR⁵⁵⁴⁵/InR^3T5, InR^E19/InR^3T5, Dp110^2H1/Dp110^5W3, Dp110^A/Dp110^5W3, PKB¹/PKB³, Rheb ^2D1/Rheb^44.1, and S6K^I-1/S6K^P1713. These last were obtained by crossing heterozygous flies for both mutant alleles in the heteroallelic combination desired and selecting flies without balancer chromosomes in the progeny. These viable alleles and heteroallelic combinations represent unique and different fly models of type 2 diabetes. Most insulin pathway alleles and heteroallelic combinations are lethal and so cannot be studied. Among rare viable insulin pathway mutant combinations, these were chosen to represent a cross section of the pathway at different levels, and to our knowledge, are the sole known Drosophila models of diabetes type 2. To account for differences in genetic backgrounds, pairwise comparisons were always done between mutant combinations and sibling controls.

To determine body weight, wing size, and lipid and carbohydrate content, flies were aged for 2 days after eclosion before use. We measured body weight in cold-killed individual females placed in a microbalance (Cahn C-31; Manasquan, NJ) with 0.1 μg of sensitivity and a range of 0.1 μg to 25 mg. For wing analysis, flies were anesthetized with CO₂, and wings were dissected, placed in absolute ethanol, and mounted in a 6:5 mixture of lactic acid/ethanol (13). Measurements were made directly on digitized images of mounted wings using iVision software (Calgary, AB, Canada).

Lipid determination was as described previously (14). Total lipids were calculated against a calibration curve using vegetable oil solutions as the lipid standard. Carbohydrate content was determined as described previously (15) and was estimated against a calibration curve of standard glucose solutions. Estimates may not reflect total carbohydrate content, because no test was done to ensure complete digestion and reaction of cuticular carbohydrates; nevertheless, all flies were treated in the same way. Results are expressed as the percentage change.

Groups of 10 adult flies (aged 2 days) were decapitated, and heads were frozen for at least 24 h. Heads were homogenized in Tris-HCl saline buffer containing protease inhibitors (1 mol/L NaCl, 50 mmol/L MgCl₂, 1 mmol/L EGTA, 1 mg/mL bacitracin, 2 mmol/L benzamidine, 0.1 mg/mL soybean trypsin inhibitor, 10 μg/mL pepstatin, 20 units/mL aprotinin, 20 μg/mL leupeptin, and 10 mmol/L Tris buffer, pH 7.0).

Acetylcholinesterase activity was measured by the Ellman method (16) using 1 mmol/L acetylthiocholine as the substrate and 50 μmol/L iso-OMPA (tetraisopropyl pyrophosphoramide). Iso-OMPA is a cholinesterase inhibitor with relative specificity for butyrylcholinesterase (17). The assay mixture contained 0.05% (w/v) Triton X-100. Acetylcholinesterase activity was determined as described previously (18). One unit of enzymatic activity is equal to 1 μmol of substrate hydrolyzed per hour per milliliter at pH 7.5 and 25°C. Protein content was determined according to Lowry (19) using BSA as the standard.

Electroretinogram (ERG) recordings were performed as described previously (20), with the exception that the recording electrode made electrical contact with the compound eye. The ground microelectrode was placed in the thorax. Flies were adapted to the dark for at least 5 min before recordings, and all recordings were performed in the dark. Stimulating white light was 427 lux for 500 ms. Electrical responses were amplified, visualized, and photographed in a TDS210 oscilloscope (Tektronix, Beaverton, OR). Each fly was tested five times, and individual results were averaged per fly. For each genotype, 10 flies were tested.

Student t test was used for pairwise comparisons between mutant and sibling controls. Data are presented as mean ± SEM.

Most insulin pathway mutant alleles are homozygous lethal. To circumvent lethality and obtain varying degrees of reduction in insulin signaling, we used viable heteroallelic and viable homozygous hypomorphic mutant alleles (Dp110^2H1, Dp110^5W3, and Rheb^7A1). The allele chico¹ is viable, despite being a null (partly because the Drosophila InR has an intracellular extension homologous to insulin receptor substrate genes, allowing the Drosophila InR to function as an insulin receptor substrate) (21).

These flies have some insulin pathway functionality left, typical of type 2 diabetic patients. This allows escape from lethality; because of this, alterations in the parameters studied, although not necessarily present in all mutant combinations tested for the same gene, if seen in at least one mutant combination, is treated to mean that the gene affects the particular function studied.

As a way to quantify insulin-signaling effects on growth, we measured body weight and wing size of mutant flies and control siblings and calculated percentage reductions. All of the mutant combinations had different, but significant, degrees of weight reduction (range 22% for S6K^I-1/P1713 to 65% for PKB^1/3). Wings were similarly significantly reduced (Fig. 1).

To examine effects on metabolism, we determined lipid content in individual flies. We found that mutants generally have a significant increase in lipid content, which was close to 100% in PKB mutants and in Dp110^2H1/2H1. Less dramatic but significant increases were found for the insulin receptor, chico, S6K, and three heteroallelic combinations of Dp110. No significant differences were found for Dp110^5W3/5W3 or for three of four mutant combinations for Rheb. From these results we conclude that all tested proteins are involved in lipid metabolism (Fig. 2A).

We also determined carbohydrate content. Of the mutants tested, fewer showed significant changes. The three heteroallelic combinations of Dp110, InR^3T5/E19, and Rheb^7A1/7A1 exhibited significant increases, whereas Rheb^44.1/7A1 had a discrete decrease (Fig. 2B). Other mutants, such as Dp110^2H1/2H1, PKB^1/3, and chico^1/1 had no significant changes. We interpret these data to mean that at least InR, Dp110, and Rheb are involved in carbohydrate metabolism. Two ways to rationalize that other genes had no effect (chico, PKB, and S6K) is that lipid metabolism may be affected more than carbohydrates by insulin signaling, and/or that the particular combinations tested retained more carbohydrate function.

Taken together, these results argue that growth and both lipid and carbohydrate metabolism is partially regulated by insulin. This has been shown in mammals and diabetic patients, making the fly model similar.

Insulin signaling has been implicated in retinal (22) and brain function (23), so we looked for alterations in the fly nervous system. We assessed retinal physiology using ERGs. Most mutant genotypes tested, but not all, showed a decrease in amplitude of the ERG-sustained component (Fig. 3A). The off transient, a consequence of light-evoked synaptic activity at the brain lamina (24), also showed amplitude reductions (Fig. 3B).

ERG decreases do not correlate with the extent of metabolic alterations. Dp110^5W3/5W3 shows severe ERG defects but normal lipid and carbohydrate content. PKB^1/3 has defective ERGs (Fig. 3B), a dramatic increase in body lipids, but no change in carbohydrates. InR^5545/3T5 exhibits no change in ERG and carbohydrate levels but an increase in lipid content. Dp110^2H1/A has no ERG defect but does have severe lipid and carbohydrate alterations. Finally, Rheb^7A1/7A1 has normal ERG and lipid levels but a significant reduction in carbohydrates. As suggested for diabetes, retinal defects here arise independently of microvasculature defects.

A general and global way to assess nervous system function is to use an enzymatic activity common in nervous tissue as an activity reporter. We measured cholinesterase activity from fly head extracts because acetylcholine is the main excitatory neurotransmitter in fly brains (25). Most mutant phenotypes studied have significantly diminished cholinesterase activity, implying faulty brain acetylcholine metabolism (Fig. 4). Diabetic retinas in rats also show significant changes in cholinesterase activity (18).

Insulin signaling affects lipid and carbohydrate metabolism and size in flies, showing evidence of the pathway’s two-prong control over metabolism and growth. Moreover, nervous system defects exhibit alterations on at least two neurotransmitter systems: acetylcholine and histamine. Histamine is the fly photoreceptor neurotransmitter (25). Reduced photoreceptor depolarization (amplitude reductions of sustained ERG component) leads to reduced neurotransmitter release, and whether this solely explains reduced off transients (brain electrical responses to histamine) does not belie the fact that neuronal communication is compromised. Reduced cholinesterase activity, as a general brain readout, implies widespread nervous system effects.

Our results, in general, argue that nervous system phenotypes are partially independent of metabolic and growth phenotypes, strongly implying an independent origin. This has the unforeseen benefit of allowing the study of insulin-signaling defects in relative isolation: mutant conditions occur without other effects in nervous system physiology in Dp110^5W3/5W3, lipid metabolism in InR^5545/3T5, and carbohydrate metabolism in Rheb^7A1/7A1.

In summary, partial loss-of-function insulin pathway fly mutants have reduced growth, lipid and carbohydrate alterations, and abnormal nervous system function. They represent viable type 2 diabetes models, allowing study of several typical insulin-signaling defects of patients with diabetes, some in relative isolation.

This research was funded by Consejo Nacional de Ciencia y Tecnología (grant 81864) and by Universidad Nacional Autónoma de México laboratory budget and Programa de Apoyo a Proyectos de Investigación e Innovación Tecnológica program (grant IN202403) (to J.R.R.-E.).

No potential conflicts of interest relevant to this article were reported.

J.M.M.-M. and J.R.R.-E. researched data, contributed to discussion, wrote the manuscript, and reviewed and edited the manuscript. R.S. researched data, contributed to discussion, and reviewed the manuscript. G.S.-C. researched data and contributed to discussion. L.M.S. reviewed the manuscript and contributed to discussion.

The authors thank María Teresa Peña-Rangel, Departamento de Neurobiología del Desarrollo y Neurofisiología, Instituto de Neurobiología, Universidad Nacional Autónoma de México, for technical assistance. Information in FlyBase was important for the results reported.