A neural circuit linking learning and sleep in Drosophila long-term memory

Abstract

Animals retain some but not all experiences in long-term memory (LTM). Sleep supports LTM retention across animal species. It is well established that learning experiences enhance post-learning sleep. However, the underlying mechanisms of how learning mediates sleep for memory retention are not clear. Drosophila males display increased amounts of sleep after courtship learning. Courtship learning depends on Mushroom Body (MB) neurons, and post-learning sleep is mediated by the sleep-promoting ventral Fan-Shaped Body neurons (vFBs). We show that post-learning sleep is regulated by two opposing output neurons (MBONs) from the MB, which encode a measure of learning. Excitatory MBONs-γ2α'1 becomes increasingly active upon increasing time of learning, whereas inhibitory MBONs-β'2mp is activated only by a short learning experience. These MB outputs are integrated by SFS neurons, which excite vFBs to promote sleep after prolonged but not short training. This circuit may ensure that only longer or more intense learning experiences induce sleep and are thereby consolidated into LTM.

Fig. 1. vFBs are activated after long but not short courtship experience.

a Mean normalized luminescence traces (±SEM) in vFBs. b Mean normalized luminescence in the 1–3 h time period after training (TP1-3) in a. n_0 h = 33, n_1 h = 29, n_2 h = 37, n_4 h = 35, n_6 h = 35, P_1 h = 0.7824, P_2 h = 0.2996, P_4 h = 2.9e−5, P_6 h = 0.0065 for H₀ N. Lum._exp = N. Lum._nai, P = 0.0014 for H₀ N. Lum._2 h = N. Lum._4hr, P = 0.6311, for H₀ N. Lum._4 h = N. Lum._6 h, two-sided Student T test. c Mean sleep traces of ΔSleep (exp-nai) (±SEM) after training. d Mean ΔSleep (exp-nai) in TP1-3 in c. n_0 h = 45, n_1 h = 42, n_2 h = 41, n_4 h = 44, n_6 h = 48, P_1 h = 0.5480, P_2 h = 0.6338, P_4 h = 0.0022, P_6 h = 1.1e−4 for H₀ ΔSleep = 0, two-sided Wilcoxon Signed Rank test, and P = 0.0251 for H₀ ΔSleep_2 h = ΔSleep_4 h, P = 0.4809 for H₀ ΔSleep_4 h = ΔSleep_6 h, two-sided Wilcoxon Rank Sum test. e LTM, shown as Suppression Index, SI [%] of males trained as indicated in a. n_0 h = 58, n_1 h = 60, n_2 h = 59, n_4 h = 63, n_6 h = 60, P_1 h = 0.2942, P_2 h = 0.3277, P_4 h = 0.0022, P_6 h = 1.2e−4 for H₀ SI = 0, and P = 0.0175 for H₀ SI_2 h = SI_4 h, two-sided Permutation test. f (top) Mean sleep traces of ΔSleep (exp-nai) (±SEM) upon vFBs silencing. (bottom) Mean ΔSleep (exp-nai) in TP1-3 of the data at the top. (left) n_20 °C = 57, n_30 °C = 59, P_20 °C = 1.4e−4, P_30 °C = 0.0581 value is for H₀ ΔSleep = 0, and P = 2.6e−5 for H₀ ΔSleep_20 °C = ΔSleep_30 °C, (middle) n_20 °C = 63, n_30 °C = 46, P_20 °C = 1.6e−6, P_30 °C = 1.4e−6 for H₀ ΔSleep = 0, and P = 0.1399 for H₀ ΔSleep_20 °C = ΔSleep_30 °C, (right) n_20 °C = 59, n_30 °C = 65, P_20 °C = 0.0019, P_30 °C = 1.3e−9 for H₀ ΔSleep = 0, and P = 0.0612 for H₀ ΔSleep_20 °C = ΔSleep_30 °C. Two-sided Wilcoxon Signed Rank test for single groups and two-sided Wilcoxon Rank Sum test across groups. Full genotypes and data analysis details can be found in Supplementary Tables 1–3. Source data are provided as a Source Data file. ns P > 0.05, *P  <  0.05, **P  <  0.01, ***P  <  0.001. n represents independent fly samples with assays repeated at least 3 times (b, d, e and f). Box plots represent median and IQR and whiskers extend to lower and upper adjacent values (b, d and f).

Fig. 2. MBONs-γ2α’1 activity reflects the length of learning.

a Mean sleep change upon optogenetic activation of MBONs. P value is for H₀ ΔSleep(light-nolight) = 0, two-sided Wilcoxon Signed Rank test, n = 36–54 per group (Supplementary Table 3). b Neural skeletons derived from the FIB-SEM Drosophila brain volume. c LTM, shown as Suppression Index, SI [%], upon MBONs-γ2α’1 silencing. n₁ = 53 and 51, n₂ = 53 and 54, n₃ = 44 and 44 for naïve and trained group respectively, P₁ = 0.8602, P₂ = 0.0003, P₃ = 0.0001 for H₀ SI = 0, and P₁₂ = 0.0105, P₁₃ = 0.0013 for H₀ SI_exp = SI_ctrl, two-sided Permutation test. d (left) Mean sleep traces of ΔSleep (exp-nai) (±SEM) upon MBONs-γ2α’1 silencing. (right) Mean ΔSleep (exp-nai) in the 1–3 h time period after training (TP1-3) on the left. (top) n_20 °C = 70, n_30 °C = 46, P_20 °C = 5.3e−8, P_30 °C = 0.9608 for H₀ ΔSleep = 0, and P = 3.9e−6 for H₀ ΔSleep_20 °C = ΔSleep_30 °C. (bottom) n_20 °C = 68, n_30 °C = 69, P_20 °C = 0.0065, P_30 °C = 0.0002 for H₀ ΔSleep = 0, and P = 0.2388 for H₀ ΔSleep_20 °C = ΔSleep_30 °C. Two-sided Wilcoxon Signed Rank test for single groups and two-sided Wilcoxon Rank Sum test across groups. e Mean GCaMP6s response (ΔF/F) traces (±SEM) in vFBs (top), dFBs (middle) upon optogenetic activation of MBONs-γ2α’1 (red bar) or in vFBs (bottom) upon MBONs-calyx activation (red bar). n_top = 7, n_middle = 4, n_bottom = 7, P_top = 1.0e−5, P_middle = 0.6254, P_botton = 0.4812, for H₀ ΔF/F = 0, two-sided Student T test. f Normalized mean luminescence traces (±SEM). g Mean normalized luminescence in TP1-3 in f. n_0 h = 73, n_1 h = 84, n_2 h = 50, n_4 h = 61, n_6 h = 67, P_1 h = 0.1476, P_2 h = 0.0214, P_4 h = 7.7e−4, P_6 h = 8.0e−8 for H₀ N. Lum._exp = N. Lum._nai, P = 0.0068 for H₀ N. Lum._1hr = N. Lum._4 h, and P = 0.0050 for H₀ N. Lum._2 h = N. Lum._6 h, two-sided Student T test. Full genotypes and data analysis details in Supplementary Tables 1–3. Source data are provided as a Source Data file. ns P > 0.05, *P  <  0.05, **P  <  0.01, ***P  <  0.001. n represents independent fly samples with assays repeated at least 3 times (a, c, d and g). Box plots represent median and IQR, whiskers extend to lower and upper adjacent values and red crosses for outliers (a, d and g).

Fig. 3. MBONs-β’2mp activity peaks after short training.

a Neural skeletons derived from the FIB-SEM Drosophila brain volume. b Mean GCaMP6s response (ΔF/F) traces (±SEM) in vFBs upon optogenetic MBONs-β’2mp activation (red bar). n = 5 flies per group, P_left = 0.0303, P_right = 0.0071 for H₀ ΔF/F = 0, two-sided Student T test. c. Normalized mean luminescence traces (±SEM). d. Mean normalized luminescence in the 1–3 h time period (TP1-3) after training in c. n_0 h = 47, n_1 h = 38, n_2 h = 39, n_4 h = 44, n_6 h = 38, P_1 h = 0.2769, P_2 h = 0.0056, P_4 h = 0.1043, P_6 h = 0.8331 for H₀ N. Lum._exp = N. Lum._nai, and P = 8.6e−4 for H₀ N. Lum._2 h = N. Lum._6 h, two-sided Student T test. e Mean sleep traces (±SEM) of ΔSleep (exp-nai) upon MBONs-β’2mp silencing. f Mean ΔSleep (exp-nai) in TP1-3 in e. n₁ = 54, n₂ = 49, n₃ = 39, n₄ = 42, P₁ = 0.6073, P₂ = 0.0207, P₃ = 0.3553, P₄ = 0.3451 for H₀ ΔSleep = 0, two-sided Wilcoxon Signed Rank test, and P₁₂ = 0.0273, P₃₄ = 0.3559 for H₀ ΔSleep_20 °C = ΔSleep_30 °C, two-sided Wilcoxon Rank Sum test. g LTM, shown as SI [%] upon MBONs-β’2mp silencing. n₁ = 53 and 52, n₂ = 59 and 61, n₃ = 52 and 57, n₄ = 72 and 59 for naïve and trained group respectively, P₁ = 0.2188, P₂ = 0.0019, P₃ = 0.4019, P₄ = 0.2663 for H₀ SI = 0, and P₁₂ = 0.0376, P₃₄ = 0.8575 for H₀ SI_exp = SI_ctrl, two-sided Permutation test. h Mean ΔSleep (exp-nai) in TP1-3 upon MBONs-β’2mp activation. n₁ = 40, n₂ = 40, n₃ = 40, n₄ = 38, P₁ = 0.0025, P₂ = 0.6374, P₃ = 0.0014, P₄ = 1.2e−5 for H₀ ΔSleep = 0, two-sided Wilcoxon Signed Rank test, and P₁₂ = 0.0023, P₃₄ = 0.8768 for H₀ ΔSleep_noact = ΔSleep_act, two-sided Wilcoxon Rank Sum test. i LTM, shown as SI [%], upon MBONs-β’2 mp activation. n₁ = 56 and 57, n₂ = 57 and 56, n₃ = 30 and 40, n₄ = 33 and 37 for naïve and trained group respectively, P₁ = 1.0e−5, P₂ = 0.0517, P₃ = 4.8e−4, P₄ = 1.0e−5 for H₀ SI = 0, and P₁₂ = 0.0403, P₃₄ = 0.9542 for H₀ SI_exp = SI_ctrl, two-sided Permutation test. Full genotypes and data analysis details in Supplementary Tables 1–3. Source data are provided as a Source Data file. ns P > 0.05, *P < 0.05, **P  <  0.01, ***P < 0.001. n represents independent fly samples with assays repeated at least 3 times (d, f, g, h and i). Box plots represent median and IQR, whiskers extend to lower and upper adjacent values and red crosses for outliers (d, f and h).

Fig. 4. SFSs integrate MBONs-γ2α’1 and MBONs-β’2mp inputs onto vFBs.

a Neural skeletons derived from the FIB-SEM Drosophila brain volume. b Confocal image of SFSs expression pattern extracted from the Multi Color Flip-Out (MCFO) of the VT033828-GAL4 line. Scale bar = 50 μm. (Details in the Methods). c, d Traces of mean GCaMP6s responses (ΔF/F) (±SEM) in SFSs upon optogenetic activation (red bar) of MBONs-γ2α’1 (c, n = 9, P = 0.048), MBONs-β’2mp (d, n = 5, P = 0.0002). P value is for H₀ ΔF/F = 0, two-sided Student T test. e. Traces of mean GCaMP6s responses (ΔF/F) (±SEM) in vFBs upon SFSs focal activation (red bar). n = 6, P = 0.0075 H₀ ΔF/F = 0, two-sided Student T test. f. Mean ΔSleep (exp-nai) in the 1–3 h time period after training (TP1-3) upon SFSs silencing. n = 39 flies per group, P₁ = 0.0046, P₂ = 0.1208, P₃ = 1.1e−4, P₄ = 0.0445 for H₀ ΔSleep = 0, two-sided Wilcoxon Signed Rank test, and P₁₂ = 0.0091, P₃₄ = 0.2711 for H₀ ΔSleep_20 °C = ΔSleep_30 °C, two-sided Wilcoxon Rank Sum test. g LTM, shown as SI [%], after SFSs silencing in TP1-3. n₁ = 52 and 52, n₂ = 61 and 64, n₃ = 62 and 51, n₄ = 65 and 62 for naïve and trained group respectively, P₁ = 1.0e−5, P₂ = 0.4976, P₃ = 0.0028, P₄ = 2.0e−4 for H₀ SI = 0, and P₁₂ = 0.0115, P₃₄ = 0.7954 for H₀ SI_exp = SI_ctrl; two-sided Permutation test. h Mean calcium traces (±SEM) in SFSs, measured for 1 min at multiple time points spanning TP1-3. i Mean calcium levels in TP1-3 in h. n_0 h = 28, n_1 h = 17, n_2 h = 18, n_4 h = 19, n_6 h = 26, P_1 h = 0.3431, P_2 h = 0.6126, P_4 h = 0.0312, P_6 h = 0.0131 for H₀ F_nai = F_exp, P = 0.0466 for H₀ F_2 h = F_4 h, and P = 0.8225 for H₀ F_4 h = F_6 h, two-sided Student T test. j Model of the circuit mechanism to generate post-learning sleep for LTM consolidation. Solid lines indicate monosynaptic connections. Full genotypes and data analysis details in Supplementary Tables 1–3. Source data are provided as a Source Data file. ns P > 0.05, *P  <  0.05, **P  <  0.01, ***P  <  0.001. n represents independent fly samples with assays repeated at least 3 times (c, d, e, f, g and i). Box plots represent median and IQR, whiskers extend to lower and upper adjacent values and red crosses for outliers (f and i).