Aversive phototaxic suppression: evaluation of a short-term memory assay in Drosophila melanogaster

Abstract

Drosophila melanogaster is increasingly being used to model human conditions that are associated with cognitive deficits including fragile-X syndrome, Alzheimer's disease, Parkinson's disease, sleep loss, etc. With few exceptions, cognitive abilities that are known to be modified in these conditions in humans have not been evaluated in fly models. One reason is the absence of a simple, inexpensive and reliable behavioral assay that can be used by laboratories that are not expert in learning and memory. Aversive phototaxic suppression (APS) is a simple assay in which flies learn to avoid light that is paired with an aversive stimulus (quinine/humidity). However, questions remain about whether the change in the fly's behavior reflects learning an association between light and quinine/humidity or whether the change in behavior is because of nonassociative effects of habituation and/or sensitization. We evaluated potential effects of sensitization and habituation on behavior in the T-maze and conducted a series of yoked control experiments to further exclude nonassociative effects and determine whether this task evaluates operant learning. Together these experiments indicate that a fly must associate the light with quinine/humidity to successfully complete the task. Next, we show that five classic memory mutants are deficient in this assay. Finally, we evaluate performance in a fly model of neurodegenerative disorders associated with the accumulation of Tau. These data indicate that APS is a simple and effective assay that can be used to evaluate fly models of human conditions associated with cognitive deficits.

Figure 1. Experimental apparatus for testing phototaxic suppression

(a) Flies are placed in a T-maze and allowed to choose between a lighted and darkened alley. Quinine is then placed into the lighted alley to provide an aversive association. The number of visits to the dark alley is tabulated during four blocks of four trials. (b)Female Cs flies (n = 100) learn to select the dark alley more frequently over the course of the 16 trials. A high score indicates learning (performance); one-way anova for blocks (F_3,396 = 132.26, P = 2.3E−59). (c) Frequency histograms for the number of photonegative choices during block 1 and block 4 trials show that scores in block 4 are normally distributed. (d) Flies rarely visit the dark chamber in the absence of quinine (black, n = 12); when quinine is present in both the dark and light alleys (light gray, n = 10) or when quinine is present only in the dark alley (dark gray, n = 10). However, flies visit the dark chamber when quinine is present only in the lighted alley (white, n = 10). One-way anova for condition (F_3,38 = 32.01, P = 1.7E−10). ‘Q’ indicates vials containing quinine * P<0.05.

Figure 2. Control experiments to evaluate the role of desensitization/habituation

(a) Learning score for female Cs flies using the standard protocol (base; n = 9) vs. flies preexposed to a filter paper wetted with quinine for 10 min before the beginning of the 16-trial test (preinc.; n = 9). Normal learning is observed in flies preexposed to quinine; t(16) = 1.74, P = 0.28. (b) Flies experienced 16 unpaired trials after which performance was evaluated for an additional four trials using the standard protocol. For a given trial in the unpaired test, vials were either both dark or both light, and quinine was randomly distributed to either the right or left vial. Thus, the flies would be exposed to dark quinine and light quinine. Performance remained low in flies exposed to the unpaired protocol (n = 8) vs. trained flies (n = 8); t(15) = −6.53, P = 9.8E−06. (c) Flies were trained for three blocks and then transferred to a clean maze for the final four trials to assess potential confounds of chemical traces. Learning during block 4 was similar in transferred (n = 8) and control flies (n = 10); t(16) = 0.4, P = 0.69. (d) Each yoked control fly was exposed to light/quinine or dark/dry for 16 trials at the same interval (50 seconds) and for the same duration (5 seconds) as an experimental fly performing the task (normal training) and then tested for an additional four trials using the standard protocol. Performance remained low in yoked control flies (n = 8) compared with trained flies (n = 10); t(16) = −5.89, P = 2.28E−05. (e) Flies were trained for 16 trials. During 10 of these trials, the yoked control’ fly encountered the same pairing (light/quinine or dark/dry) as during a normal test. However, on six trials (one trial in block 1, two trials in block 2, one trial in block 3 and two trials in block 4) light conditions were immediately reversed after the fly made its choice (n = 11) so that a dark choice would result in light/quinine and vice versa. Flies were then tested for an additional four blocks using the standard protocol. Few visits to the dark vial were observed in yoked control flies compared with controls (n = 8); t(19) = −3.43, P = 0.001. (f) As above, light conditions were modified on 6 of 16 trials. However, in this protocol, the fly would unexpectedly encounter quinine in the dark or dry in the lighted vial when they had made a dark or light choice, respectively. Performance remained low in yoked controls (n = 8) compared with trained siblings (n = 8); t(16) = −3.61, P = 0.001. *P < 0.05, n.s.: non significant, all tests were conducted between ZT0-ZT3:59.

Figure 3. Phototaxic suppression in memory mutants

(a) Female lio² mutants (n = 9) are learning impaired compared with controls (n = 10); t(16) = −3.91, P = 0.0005. As previously described, lio² mutants have abnormal mushroom bodies with reduced or absent α lobes and fused β/γ lobes. (b) Female lat^p1 flies (n = 8) show performance decrements compared with controls (n = 9). Impairments are still observed in lat^p1/Df(2R)vg-B (n = 7); one-way anova for genotype (F_2,21 = 4.74, P = 0.019). (c) Females flies mutant for pastrel (pst¹, n = 6) and pst¹/Df(3L)pbl-X1 (n = 8) perform significantly worse than controls (n = 10); one-way anovas for genotype (F_2,21 = 8.98, P = 0.001). (d) Both male (n = 8) and female (n = 9) dnc¹ mutants are learning impaired compared with their respective controls (n = 8). A 2(sex) × 2(genotype) anova yielded significant main effect for genotype (F_1,29 = 8.35, P = 0.006). (e) Male rut²⁰⁸⁰ mutants (n = 10) show impairments in learning, whereas rut²⁰⁸⁰ females (n = 10) show normal performance. A 2(sex) × 2(genotype) anova yielded significant sex × genotype interaction (F_1,36 = 10.22, P = 0.003). (f) Cumulative visits to the dark vial over four blocks is similar in female Cs and rut²⁰⁸⁰ flies indicating that they do not differ in the rate of acquisition. (g) Learning impairments can be restored in male dnc¹ mutant flies by expressing a UAS-dnc construct in the MB with the c309 GAL4 driver; one-way ANOVA for genotype (F_2,34 = 3.29, P = 0.049). (h, i) rut₂₀₈₀ males flies bearing the 247-GAL4 driver (rut;247, n = 6) or the UAS-rut construct alone (rut;UAS-rut, n = 8) show learning impairments. rut;UAS-rut/247 flies (n = 11) and rut;UAS-rut/c309 (n = 8) express wild-type rut in the MB and show normal learning; one-way anova for genotype (F_3,49 = 2.37, P = 0.04). *Planned comparisons with Tukey correction (P < 0.05). n.s.: non significant. All tests were conducted between ZT0-ZT3:59.

Figure 4. Retention, acquisition and retrieval

(a) Performance requires <2 min of memory retention: Cs and rut²⁰⁸⁰ flies that achieved a performance score of 0.5 during block 4were retested after a 2-, 5- or 15-min delay. A 2(genotype: Cs, rut²⁰⁸⁰) × 3(delay: 2, 5, 15 min) × 2 (test: test, retest) repeated measures anova showed a significant genotype × test interaction (F_1,32 = 22.87, P = 3.73E−5). Cs flies retained memory of the association for up to 15 min. rut²⁰⁸⁰ flies retain the association for no more than 2 min. (b) Blocking neurotransmitter release in a large subset of Kenyon cells by expressing the UAS-shi^ts1 transgene under the control of MBswitch does not impair learning: 2(RU486 vs. ETOH) × 3(genotype) (F_2,43 = 0.21, P = 0.81). Both RU486 (RU+) and ETOH (vehicle, RU−) fed flies were transferred to the nonpermissive temperature (34°C) for 15 min and then tested at 34°C. (c) MBSwitch>UAS-shi^ts1 flies fed RU+ (n = 5) and RU− (n = 7) were trained at 34°C and evaluated for retrieval after a 5-min delay at 34°C. Blocking neurotransmitter during retrieval impairs performance; t(10) = −1.86, P = 0.046 (left). Both parental lines, MBSwitch/+ (n = 5) and UAS-shi^ts1/+ (n = 5) displayed normal retrieval at 34°C; t(8) = −0.53, P = 0.6 (right). (d) MBSwitch>UAS-shi^ts1 flies fed RU+ (n = 5) and RU− (n = 5) flies were trained at 34°C and evaluated for retrieval after a 5-min delay at 24°C. Retrieval was identical for RU+ and RU− flies at 24°C; t(8) = 0, P = 1. *Planned comparisons with Tukey correction (P < 0.05). n.s.: non significant. All tests were conducted between ZT0-ZT3:59.

Figure 5. Detection of cognitive impairments using phototaxic suppression

(a) Reduced photosensitivity does not impair learning and does not prevent sleep-deprivation-induced cognitive impairments. Female Cs flies with a PI of 0.82 (9500 lx, n = 10) learn as well as flies with a PI of 0.64 (73 lx, n = 9); male blind flies Norp^A36 (n = 10) do not change their behavior over 16 trials. Sleep deprivation only results in a significant learning impairment in flies with vision. A 3(vision) × 2(base, SD) × 4(blocks) repeated measures anova shows a significant vision×blocks interaction (F_6,159 = 6.13, P = 8.42E−6). (b) Female Cs flies learn to suppress phototaxis when electric shock is used as the negative reinforcer (n = 17); electric shock does not produce such a strong association as to prevent learning impairments following sleep loss (n = 15) t(30) = 2.24, P = 0.016. (c) Female flies expressing human tau (c155;UAS-tau^wt, n = 13) show significant reductions in learning compared with parental lines (c155/+ n = 10 and UAS-tau^wt, n = 7 F_2,29 = 4.00, P = 0.029). (d) Learning in 23- to 25-day-old MBSwitch>UAS-Tau^wt (n = 14) flies fed RU486 (RU+) was impaired compared with age-matched siblings fed vehicle (RU−) and 25- to 30-day-old Cs flies (n = 10) (F_2,33 = 3.48, P = 0.042). *Planned comparisons with Tukey correction (P < 0.05). All tests were conduced between ZT0-ZT3:59.