The long-term memory trace formed in the Drosophila α/β mushroom body neurons is abolished in long-term memory mutants

Abstract

A prior screen identified dozens of Drosophila melanogaster mutants that possess defective long-term memory (LTM). Using spaced olfactory conditioning, we trained 26 of these mutant lines to associate an odor cue with electric shock and then examined the memory of this conditioning 24 h later. All of the mutants tested revealed a deficit in LTM compared to the robust LTM observed in control flies. We used in vivo functional optical imaging to measure the magnitude of a previously characterized LTM trace, which is manifested as increased calcium influx into the axons of α/β mushroom body neurons in response to the conditioned odor. This memory trace was defective in all 26 of the LTM mutants. These observations elevate the significance of this LTM trace given that 26 independent mutants all exhibit a defect in the trace, and further suggest that the calcium trace is a fundamental mechanism underlying Drosophila LTM.

Figure 1.

Long-term memory in 26 LTM mutants. A, An initial screen of mutant flies carrying one copy of c739-Gal4 and one copy of Uas-G-CaMP was performed using a 5× spaced conditioning protocol. Flies received forward conditioning with 1 min exposure to the CS+ odor (Ben) with 12 electric shock pulses (90 V), followed by 1 min exposure to the CS− odor (Oct) without electric shock after a 30 s exposure to fresh air. The training protocol was performed a total of five times with a 15 min intertrial interval (ITI). Flies were transferred to a T-maze at 24 h after conditioning and tested for behavioral memory. Naive groups of flies were carried through the same manipulations as the conditioned animals except that they were not exposed to odor and electric shock. Each conditioned group of flies was tested in parallel with a naive group. Some flies were separated before behavioral testing and analyzed for cellular memory by functional imaging (Fig. 2). The performance gains of the trained flies were measured as ΔPI. This was computed by subtracting the score of a naive group of flies from the corresponding conditioned group. In all cases, we chose experiments wherein the average scores of naive animals for each group were not statistically significant (Wilcoxon test) from zero to prevent floor and ceiling effects. For the c739; Uas-G-CaMP control flies, spaced conditioning was performed a total of three times across the several months that this experiment was conducted with an n = 6 for each experiment. The 24 h memory scores for control groups (∼0.4) were within the normal range of LTM scores we have come to expect and comparable to the LTM scores obtained by differential conditioning procedures that average the performance gains using two different CS+ odors (Tully et al., 1994; Perazzona et al., 2004; Yu et al., 2006). The ΔPI values were subjected to nonparametric tests; i.e., a Kruskal–Wallis test for multiple comparisons with genotype as the main factor, Mann–Whitney U test for comparing two independent samples, and Wilcoxon matched pairs test to compare single performance indices against zero. p values were corrected for the multiple comparisons to the control using the Benjamini and Hochberg false discovery rate. Out of the 26 lines tested, 22 yielded p values that were significantly different from the control (Kruskal–Wallis statistic 64.67, p < 0.0001; corrected Mann–Whitney pairwise comparisons, p ≤ 0.023). One mutant (E3803) was marginally nonsignificant (corrected Mann–Whitney pairwise comparison, p ≤ 0.051). This mutant was not selected for retesting given the marginally nonsignificant probability and because this is a replication experiment of Dubnau et al. (2003). The mutants C0150, E0627, and c0167 yielded p values that were not significantly different from the control (corrected Mann–Whitney pairwise comparisons, p ≥ 0.136. n = 12 for all groups). Error bars are the SEM. B, A second test of the three LTM mutants that failed to reach significance. Flies from each mutant genotype (C0150, E0627, and c0167) were trained using the 5× spaced conditioning protocol and compared to the c739; Uas-G-CaMP control. In each case, the mutant group performed significantly differently from its corresponding control group (Mann–Whitney pairwise comparisons, p ≤ 0.0433; n = 12 for all groups). Significance levels are indicated (*p ≤ 0.05, **p ≤ 0.01). Error bars are the SEM.

Figure 2.

LTM mutants had significantly lower calcium responses than controls in the MB α lobes in response to the CS+ odor (Ben) when tested 24 h after spaced forward conditioning. The calcium response (%ΔF/F_o) of the mutant and control flies (c739; Uas-G-CaMP) to the CS+ and CS− odors 24 h after conditioning. Twenty-four hours after spaced forward conditioning, flies were mounted and tested for calcium responses to the CS+ and CS− odors. As expected, a significant increase in the %ΔF/F_o was detected in the α branch of the α/β MB neurons of control flies to the CS+ (Kruskal–Wallis statistic 73.491, p < 0.0001; Mann–Whitney pairwise comparisons, p < 0.0001). Control flies exhibited odor-elicited calcium responses that were nearly twice as large as any other group. In contrast, all the mutant groups exhibited a blunted calcium response to the CS+ odor, with %ΔF/F_o scores that were similar in magnitude to that for the CS− odor and typical of the responses of naive flies and flies that are trained using conditioning protocols that do not elicit LTM (Yu et al., 2006). The %ΔF/F_o responses to CS− odor in mutants were not significantly different from control (Mann–Whitney pairwise comparisons, p ≥ 0.0974) except for e3947 versus wild type (Mann–Whitney pairwise comparisons, p = 0.0235). Error bars are the SEM. n = 9–28 for all groups.