Insulin signaling is acutely required for long-term memory in Drosophila

Abstract

Memory formation has been shown recently to be dependent on energy status in Drosophila. A well-established energy sensor is the insulin signaling (InS) pathway. Previous studies in various animal models including human have revealed the role of insulin levels in short-term memory but its role in long-term memory remains less clear. We therefore investigated genetically the spatial and temporal role of InS using the olfactory learning and long-term memory model in Drosophila. We found that InS is involved in both learning and memory. InS in the mushroom body is required for learning and long-term memory whereas long-term memory specifically is impaired after InS signaling disruption in the ellipsoid body, where it regulates the level of p70s6k, a downstream target of InS and a marker of protein synthesis. Finally, we show also that InS is acutely required for long-term memory formation in adult flies.

Keywords: insulin receptor; insulin receptor substrate; learning; long-term memory; protein synthesis.

Figure 1

Pan-neuronal disruption of Insulin signaling results in memory defects. (A) Learning is significantly defective in Drosophila expressing UAS-RNAi against the insulin receptor substrate, chico pan-neuronally. We expressed 2 different RNAi constructs: [Elav>ChicoRNAi⁷⁷⁷⁶] (P = 0.005, N = 6 PI per genotype) and [Elav>ChicoRNAi⁷⁷⁷⁷] (P = 0.0104, N = 6 PI per genotype). In addition, pan-neuronal overexpression of chico [Elav>UAS-chico] led to a significant (P = 0.0015, N = 8 PI per genotype) defect in learning. The genetic controls ElavGAL4 [Elav/+], chicoRNAi [chicoRNAi⁷⁷⁷⁶], [chicoRNAi⁷⁷⁷⁷] and InRRNAi [InRRNAi⁹⁹²] show no defects compared to wild-type flies [WT]. (B) 1-day memory after spaced training is impaired by loss or gain of function of chico in neurons. RNAi against chico resulted in significant defect in 1 day memory for both constructs [Elav>ChicoRNAi⁷⁷⁷⁶] (P < 0.00001, N = 8 PI per genotype); [Elav>ChicoRNAi⁷⁷⁷⁷] (P < 0.00001, N = 8 PI per genotype). Overexpression of chico [Elav>UAS-chico] also resulted in significant defect in 1 day memory (P = 0.0004, N = 8 PI per genotype) (C) No significant defects in 1 day memory after massed training were observed between Chico RNAi or UAS-chico expressing flies compared to genetic controls (N = 8 PI per genotype). (D) Similarly, pan-neuronal disruption of Insulin receptor leads to learning defects either via expression of UAS-InR RNAi [Elav>InR RNAi⁹⁹²] (P = 0.0082, N = 4 PI per genotype) or with the expression of a dominant negative InR [Elav>InR^DN] (P < 0.00001, N = 4 PI per genotype). There was no significant defect in any of the genetic appropriate controls. (E) Significant defects in 1 day memory after spaced training were also observed in transgenic flies expressing [Elav>InR RNAi⁹⁹²] (P < 0.00001, N = 8 PI per genotype) or [Elav>InR^DN] (P = 0.0052, N = 8 PI per genotype). (F) No significant defects were seen in flies expressing InR RNAi or DN or in the appropriate genetic controls. (G) Representative level of p70S6K in the brain of WT and flies expressing InR^DN pan-neuronally [Elav>InR^DN]. (H) Significant decreased level of p70S6K is observed in Elav>InR^DN flies (N = 5 brains per genotype, P = 0.0022) All graphs depict mean ± s.e.m. ^*p < 0.05; ^**p < 0.001; ^***p < 0.0001.

Figure 2

Insulin signaling participates in different stages of memory formation. (A) Expression pattern of OK107GAL4. (B) Learning is defective with mushroom body (MB) expression of chico RNAi [OK>chicoRNAi⁷⁷⁷⁶] (P < 0.05, N = 4 PI); [OK>chicoRNAi⁷⁷⁷⁷] (P < 0.05, N = 4 PI) or overexpression of chico [OK>UAS-chico] (P < 0.0001, N = 6 PI per genotype). Similarly expression of InRRNAi [OK>InRRNAi⁹⁹²] (P < 0.001, N = 4 PI) and InR^DN [OK>InR^DN] (P < 0.0001, N = 4 PI per genotype) leads to learning defects. (C) Significant defects in 1 day memory after spaced training are observed with MB expression of chicoRNAi [OK>chicoRNAi⁷⁷⁷⁶] (P < 0.001, N = 8 PI per genotype), [OK>chicoRNAi⁷⁷⁷⁷] (P < 0.00001, N = 8 PI per genotype) or chico overexpression [OK>UAS-chico] (P < 0.0001, N = 6 PI per genotype). Similarly, mushroom body expression of InR RNAi⁹⁹² [OK>InRRNAi⁹⁹²] (P < 0.00001, N = 8 PI per genotype) or InR^DN [OK>InR^DN] (P < 0.00001, N = 6 PI per genotype) leads to significant memory defects. (D) No defect is seen in 1-day memory after massed training for the same genotypes. (E) Expression pattern of the FEB170GAL4 illustrated by expressing mCD8::GFP. (F) No significant defect is seen with FEB170 expression of chico RNAi or overexpression as well as with InR RNAi or InR^DN in learning (N = 4 PI per genotype). (G) 1-day memory after spaced training is defective in flies expressing chicoRNAi [Feb>chicoRNAi⁷⁷⁷⁶] (P < 0.001, N = 8 PI per genotype) [Feb>chicoRNAi⁷⁷⁷⁷] (P < 0.0001, N = 8 PI) or with chico overexpression [Feb>UAS-chico] (P < 0.0001, N = 8 PI per genotype). Similarly 1-day memory is defective after spaced training with the expression of InR RNAi⁹⁹² [Feb170>InR RNAi⁹⁹²] (P < 0.0001, N = 8 PI per genotype) or InR^DN [Feb>InR^DN] (P < 0.0001, N = 8 PI) in the central complex. (H) No significant defect is seen in the same genotypes in 1-day memory after massed training. All graphs depict mean ± s.e.m. ^**p < 0.001; ^***p < 0.0001.

Figure 3

Insulin signaling is acutely required for memory formation. (A) Protocol used to express the transgenes acutely in adult. The protocol depicted is for 1 day memory after spaced training. For learning, the testing immediately follows training. (B) Acute expression of UAS-chico [HSP70GAL4>UAS-chico +HS] (P = 0.5425) or InR^DN [HSP70GAL4>InR^DN +HS] (P = 0.4226, N = 2 PI per genotype) does not affect learning. (C) Acute expression of UAS-chico [HSP70GAL4>UAS-chico +HS] (P = 0.0316, N = 8 PI per genotype) or InR^DN [HSP70GAL4>InR^DN +HS] (P = 0.0069, N = 8 PI per genotype) disrupts memory after spaced training. (D) No defect is seen after the same treatment in 1 day memory after massed traininig (P > 0.05, N = 2 PI per groups). (E) The first row shows representative brains immunohistochemistry with p70s6k from HSP70GAL4>InR^DN—HS flies with mock training (presented with odors but no shock) vs. spaced training. The second row shows representative brain immunohistochemistry with p70s6k from HSP70GAL4>InR^DN +HS with mock vs. spaced training. Quantification of the signal strength in the central complex shows a significant (P = 0.0351, N = 4 biological replicates per group) increase after spaced training compared to mock training in HSP70>InR^DN flies only when not exposed to heat shock. All graphs depict mean ± s.e.m. ^*p < 0.05.