A. Food-deprived npfr1[c01896] flies and those expressing uas-npfr1^RNAi with c061 have impaired 3hr appetitive memory whereas those expressing uas-npfr1^RNAi with c005, 210Y, 104Y, OK107 or MB247 are normal. Asterisk denotes significant difference (P<0.01, ANOVA) from other unmarked groups. Data are mean ± SEM.

Temperature shift protocols are shown above each graph. A. Expressing uas-npfr1^RNAi in all DA neurons with TH-GAL4 or in a subset of DA neurons with c061 impairs 3hr appetitive memory in hungry flies. Expressing uas-npfr1^RNAi in c061 neurons except the DA neurons (c061;THGAL80;uas-npfr1^RNAi) does not affect appetitive memory. B. Feeding flies after training suppresses 3hr memory. All genotypes were food-deprived, trained, fed and tested at the permissive temperature of 23°C. C. Blocking synaptic output from c061;MBGAL80 neurons before testing reveals memory performance in fed flies. Removing uas-shi^ts1 expression from the DA neurons reverses the memory promoting effect (c061;MBGAL80/THGAL80;uas-shi^ts1 flies). All genotypes were food-deprived, trained and stored in food vials for 120min at 23°C. Flies were then shifted to 31°C for 60min and tested for memory. D. Blocking output from c061;MBGAL80 neurons does not enhance performance in food-deprived flies. All genotypes were food-deprived, trained and stored in empty vials for 120min at 23°C. Flies were then shifted to 31°C for 60min before flies were tested for appetitive memory at 31°C. Asterisks denote significant difference (P<0.05, ANOVA) from other unmarked groups. Data are mean ± SEM.

Temperature shift protocols are shown above each graph. A. The permissive 23°C does not disrupt 3hr appetitive memory. All genotypes were starved, trained, stored for 3hr in empty vials and tested for memory at 23°C. B. Stimulating 6, 4 or 2 MB-MP neurons with uas-dTrpA1 before and during testing attenuates memory performance in starved flies. All genotypes were food-deprived, trained and stored in empty vials for 120min at 23°C and were then shifted to 31°C for 60min before and during testing. C. The permissive temperature of 23°C does not disrupt 3hr appetitive memory. All genotypes were starved, trained, stored in empty vials for 3 hr and tested for appetitive memory at 23°C. D. Stimulating 6 MB-MP neurons with uas-TRPM8 before and during testing attenuates memory performance in starved flies. All genotypes were food-deprived, trained and stored in empty food vials for 120min at 23°C and were then shifted to 16°C for 60min before and during testing. Asterisk denotes significant difference (P<0.05, ANOVA) from other unmarked groups. Data are mean ± SEM.

A. Feeding after training suppresses appetitive memory performance. Double asterisk, significant difference (P<0.007); single asterisk (P<0.03) from all other groups. Temperature shift protocols are shown. White bar represents fly storage in empty vials, while yellow bar indicates flies stored with food -figure format used throughout this study. B. dNPF is expressed in neurons that innervate the dorsal and lateral protocerebrum, the SOG and the CC. Immunostaining with anti-dNPF antibody (red), partially overlaps (yellow, merge) with dNPF-GAL4 driven CD8::GFP (green). dNPF stained cells in the SOG are not labeled by dNPF-GAL4. The dNPF antibody only labels the upper layer of the fan-shaped body, consistent with processes in the ellipsoid body and lower fan-shaped body being post-synaptic regions. Scale bar represents 20µm. C. Feeding flies after training suppresses memory performance. All flies were food-deprived, trained, fed and tested at 23°C. D. Stimulating dNPF neurons before testing produces memory performance in fed flies. All flies were food-deprived, trained, and fed for 150min at 23°C. All flies were then transferred to 31°C for 30min and tested for appetitive memory. Asterisk denotes significant difference (P<0.05, ANOVA) from unmarked groups. Data are mean ± standard error of the mean (SEM).

A. Projection view of a c061;uas-CD8::GFP brain. Filled yellow arrows mark the heel region of the MB resembling that in TH-GAL4 labeled brains (Figure S4A). B. Combining MBGAL80 with c061;uas-CD8::GFP eliminates MB expression and reveals neurons that innervate the heel of the MB (filled yellow arrows). Also see Movie S1 and Movie S2. C. Combining THGAL80 with c061;uas-CD8::GFP eliminates expression in the neurons innervating the region between the MB lobes and MB heel (hollow arrows) but leaves expression elsewhere intact. D. Projection view of a c061;MBGAL80/THGAL80;uas-CD8::GFP brain. THGAL80 removes expression from the neurons innervating the dorsal protocerebrum and MB heel (hollow arrows) and leaves expression in the fan-shaped body and elsewhere intact. Scale bar represents 20 µm. E. Higher magnification single confocal section views of the MB heel and peduncle region in a c061; uas-CD8::GFP brain. Top panel shows innervation of the MB heel. Bottom panel, innervation in the base of the peduncle. Inset, section through the peduncle showing zones occupied by the αβ,α´β´ and γ MB neurons. The MB is co-labeled with MB-DsRED. F. Confocal section through a c061;MBGAL80;uas-CD8::GFP brain at the level of the MB calyx (outlined). GFP (green) labels 3 cell bodies at the side of the calyx and 5 more lateral cell bodies. Counter staining with anti-TH antibody (red) labels 12 cell bodies in the PPL1 cluster and 3 of them overlap (merge, yellow) with c061;MBGAL80 driven GFP. Scale bar represents 10 µm. G. Confocal section through a PPL1 cluster in a c061;MBGAL80/THGAL80;uas-CD8::GFP brain. GFP (green) labels 5 cell bodies and none overlap with anti-TH staining. Scale bar represents 10 µm.

A. Projection view of a c061;MBGAL80;uas-CD8::GFP brain showing MB-MP neurons (white arrows) and neurons in the subesophageal ganglion (also see Movie S1 and Movie S2). B. Projection view of a NP2758;uas-CD8::GFP brain showing MB-MP neurons (white arrows) and neurons in the subesophageal ganglion (also see Movie S3 and Movie S4). C. Projection view of a MBGAL80; krasavietz/uas-CD8::GFP brain showing MB-MP neurons (white arrows). DA neurons innervating the MB α-stalk (blue arrows, also see Figure S4C), and neurons in the fan-shaped body and local neurons in the antennal lobe are also labeled (also see Movie S5 and Movie S6). The MB is labeled with MB-DsRED. D. Cartoon illustrating the gross structure of MB-MP neurons and the expression pattern of each GAL4. The MB is shown as an outline. DA neuron cell bodies (red, anti-TH) of a single PPL1 cluster are shown with the labeling of each GAL4 overlayed. The cell body organization is not stereotyped and it is difficult to distinguish the projections of each MB-MP neuron. No order or detail is inferred here. At least one MB-MP neuron sends a contralateral projection to the other MB (green arrow head). E. krasavietz and NP2758 label a subset of c061 MB-MP neurons. Each column shows the separate and merged channels from confocal images of a PPL1 cluster in brains co-labeled with GAL4 driven GFP (green) and anti-TH (red). Double-labeled neurons are marked with an arrow in the merged images. The quantification of neurons is shown in Figure S4B. Scale bar represents 20 µm (A, B, C) or 10 µm (E).

Left panels illustrate the state of the inhibitory control exerted upon the MB in the fed state (top) and starved state (bottom). When fed flies are exposed to the conditioned odor during memory testing (right panels) the appropriate projection neurons and MB neurons are activated (yellow). However, the signal only propagates beyond the MB neurons in hungry flies when the MB-MP neuron ‘gate’ is open. Red lines denote inhibition and green lines relief from inhibition. See discussion for more detail.