Figures and data
The larval dopaminergic system is subdivided in two functionally distinct clusters.
(A) The larval dopaminergic neurons (DANs) can be anatomically subdivided into the primary protocerebral anterior medial (pPAM) and dorsolateral 1 (DL1) cluster based on their cell body position. (B) The DL1 cluster (cell bodies in green and purple) consists of four DANs providing input to the c, d, g, and f compartment of the vertical lobe, peduncle and lateral appendix of the MB (in grey). (C, D) Four DL1 DANs (DAN-c1, DAN-d1, DAN-g1, and DAN-f1) are included in the expression pattern of the TH-GAL4 driver line. However, expression of a UAS-mCD8::GFP reporter via TH-GAL4 labels many more neurons (in green, anti-GFP) throughout the entire CNS (in red and blue, anti-N-cadherin and anti-discs large), in total about 100 neurons. (E) To test whether optogenetic activation of the DL1 DANs is sufficient to substitute for a punishment, we used the TH-GAL4 driver in combination with UAS-ChRXXL. Experimental and control larvae were trained by simultaneously presenting and odor and blue light and thus artificial activation of DL1 DANs, whereas a second odor was presented in darkness. Only larvae of the experimental genotype (p > 0.05), but not of the two genetic controls (both p < 0.05), retrieve an aversive associative olfactory memory. The same result was seen after one or three training trials (E and F, respectively). (G, H) To test for the acute function of the DL1 DANs in aversive associative olfactory memory, we expressed GtACR2 via the TH-GAL4 driver. Acute optogenetic inhibition of synaptic output from DL1 and other DANs reduced odor high salt memory. Experimental larvae raised on supplemented food (0.5 mM all-trans-retinal, ATR) and trained in blue light performed on a lower level than control animals kept on standard food (p < 0.05). A memory impairment was seen after one and three training trials (G and H, respectively). All behavioral data is shown as box-plots. Differences between groups are highlighted by horizontal lines between them. Performance indices different from random distribution are indicated below each box-plot. The sample size of each group (N=15) is given below each box-plot. n.s. p > 0.05; * p < 0.05. Scale bars: in C 50 µm and in D 25 µm.
Anatomical single cell analysis of DL1 DAN specific split-GAL4 driver lines.
(A) The larval MB is organized into 11 compartments: CX calyx; IP and LP intermediate and lower peduncle; LA lateral appendix; UVL, IVL, and LVL upper, intermediate, and lower vertical lobe; SHA, UT, IT, LT shaft as well as upper, intermediate, and lower toe of the medial lobe. Single-letter synonyms of compartment names are given as “a–k”. These letters are used to indicate compartment innervation by the MB input and output neurons (Saumweber et al 2018). DL1 cluster DANs are DAN-c1, DAN-d1, DAN-g1, and DAN-f1 that innervate the respective four different compartments of the MB. (B-J) Individual split-Gal4 driver lines were crossed with the reporter strain UAS-mCD8::GFP;mb247-lexA,lexAop-mRFP. Third instar larval brains were dissected, fixed and mounted to visualize the endogenous expression of the MB reporter (mb247-lexA,lexAop-mRFP shown in magenta) and the respective DAN pattern (GFP shown in green). (B-E) SS02160 (DAN-c1), MB328B (DAN-d1), SS02180 (DAN-f1), SS01716 (DAN-g1) each specifically label a single DL1 DAN (cell bodies are highlighted by white arrowheads). (F–K) Two neurons can be seen in MB065B, SS01702, and MB054B split-Gal4 that express in DAN-c1/DAN-f1, DAN-c1/MBIN-e1, and DAN-f1/DAN-g1. Please note that MB065B shows strong expression DAN-f1 but weaker staining in DAN-c1. (H–K) MB054B showed reliable strong expression in DAN-f1 and DAN-g1. In some brains a third weak cell body was visible right next to the other two DANs (I; GFP channel inverted and shown in black; J; cell body highlighted with gray arrowhead). Due to the low expression level we were not able to identify this cell given that only the g and f compartment of the MB were innervated (H). (K) Analysis of the entire brain via native fluorescence expression of GFP (green) and n-Syb (magenta) did not reveal additional cells for MB054B split-Gal4. Scale bars: (B-J) 20 µm, (K) 50µm.
Calcium responses of DANs to gustatory stimulation.
Four different split-Gal4 lines and the R58E02 driver line were crossed with UAS-GCaMP6m to express a calcium reporter in DANs. The responses of each DAN towards 100 mM NaCl, 1 M NaCl, and 500 mM fructose was tested in intact larvae using a microfluidic chip-based setup. (A) 1 M NaCl (red) and 500 mM fructose (green, p < 0.05 for both), but not 100 mM NaCl (orange, p > 0.05), induced a calcium response in DAN-c1. (B) DAN-d1 calcium responses were only seen after 1 M NaCl stimulation (red, p < 0.05), but not after 100 mM NaCl (orange, p > 0.05) and 500 mM fructose stimulation (green, p > 0.05). (C) Stimulation with 100 mM (orange, p > 0.05) and 1 M NaCl (red, p > 0.05) did not induce calcium responses in DAN-f1. However, stimulation with 500 mM fructose reduced the obtained calcium signal (green, p < 0.05). (D) 1 M NaCl (red, p < 0.05), but not 100 mM NaCl (orange, p > 0.05) and 500 mM fructose (green, p > 0.05), induced a calcium response in DAN-g1. (E) pPAM DANs calcium responses were only seen after 500 mM fructose stimulation (green, p < 0.05). Stimulation with low (orange) and high salt concentrations (red) did not increase calcium signals, however both reduced pPAM activity (p < 0.05). Each graph shows the mean calcium signal plotted as the relative response strength ΔF/F and the related standard error of the mean on the y axis. The time in seconds is given below each graph on the x axis. The grey box indicates the duration of the stimulus application. The sample size of each group (N=5-7) is given above each row. n.s. p > 0.05; * p < 0.05.
Ablation of individual DANs does not impair aversive olfactory memory.
In all panels aversive associative performance indices are shown for tests immediately after odor-high salt classical conditioning. In the upper panels (A-D) larvae are trained once, whereas in the lower panels (E-H) three training cycles were applied. Schematic overviews for both conditioning protocols are shown on the left. The four different DL1 DAN specific split-Gal4 driver strains SS02160, MB328B, SS02180, and SS01716 were crossed to the effector UAS-hid,rpr to induce apoptosis (A-H). Cell-specific ablation of distinct DL1 DANs did not impair odor-high salt memories after either one or three cycle training. In no case were associative performance indices upon DL1 DAN ablation lower than in both genetic controls (A-H, in all experiments at least one or even both control groups are compared to the experimental group p > 0.05). All behavioral data is shown as box-plots. Differences between groups are highlighted by horizontal lines between them. Performance indices different from random distribution are indicated below each box-plot. The sample size of each group (N=15, 20, or 25) is given below each box-plot. n.s. p > 0.05; * p < 0.05.
Ablation of DAN-f1 and DAN-g1 together impairs aversive olfactory memory.
In all panels associative performance indices are shown for tests immediately after classical conditioning. In the upper panels (A-C) larvae are trained once by pairing an olfactory stimulus with high salt punishment, whereas in the lower panels (D-F) three training cycles were applied. Schematic overviews for both conditioning protocols are shown on the left. The three different DL1 DAN specific split-Gal4 driver MB054B, MB065B, and SS01702 that each label two neurons were crossed to the effector UAS-hid,rpr to induce apoptosis (A-F). (A) With MB054B used as driver strain to ablate the DL1 DAN combination DAN-f1/DAN-g1, the aversive associative performance index of the experimental group was decreased compared to both controls (p < 0.05). (B-F) In all other experiments ablation of different DL1 DAN combinations did not reveal a phenotype. (in all experiments at least one or even both control groups are compared to the experimental group p > 0.05). Please note, that this also includes MB054B crossed with UAS-hid,rpr tested after three training trials (D). (G) To verify the memory phenotype of MB054B crossed with UAS-hid,rpr tested after one trail conditioning, we repeated the experiment using the odor pair hexyl acetate (HA) and benzaldehyde (BA). Again, experimental larvae tested after one trial learning showed a robust decrease in odor-high salt memory when compared to both genetic control groups (p < 0.05). The memory phenotype was not seen after three training trials (J, p > 0.05). (H, K) With MB054B used as driver strain to ablate the DL1 DAN combination DAN-f1/DAN-g1, aversive odor-quinine memory was not impaired after one or three cycle conditioning (p > 0.05 when comparing experimental and control groups). Similarly, appetitive odor-fructose learning after one and three cycle conditioning was not impaired when ablating DAN-f1/DAN-g1 (p > 0.05 when comparing experimental and control groups). All behavioral data is shown as box-plots. Differences between groups are highlighted by horizontal lines between them. Performance indices different from random distribution are indicated below each box-plot. The sample size of each group (N=15, 20, or 25) is given below each box-plot. n.s. p > 0.05; * p < 0.05.
Optogenetic DL1 DAN activity can substitute for salt punishment.
In all panels associative performance indices are shown for tests immediately after classical conditioning. In panels (A-E) larvae are trained once by pairing an olfactory stimulus with artificial blue light activation, whereas in panel (F) three training cycles were applied. Schematic overviews for both conditioning protocols are shown to the left of (A) and (F). (A-D) To test whether optogenetic activation of the individual DL DANs DAN-c1, DAN-d1, DAN-f1, and DAN-g1 is sufficient to substitute for a punishment, we used the split-Gal4 lines SS02160, MB328B, SS02180, and SS01716 in combination with UAS-ChR2XXL. (E, F) For simultaneous optogenetic activation of DAN-f1/DAN-g1 we used MB054B. Larvae of the experimental genotypes for DAN-c1, DAN-f1, DAN-g1 and the DAN-f1/DAN-g1 combination (for all p < 0.05), but not for DAN-d1 and all genetic controls (for all p > 0.05), showed an aversive associative memory. The results imply that in the tested conditions, the punishment signal can be mediated by the artificial activation of all individual DL1 DANs, with the exception of DAN-d1. All behavioral data is shown as box-plots. Differences between groups are highlighted by horizontal lines between them. Performance indices different from random distribution are indicated below each box-plot. The sample size of each group (N=15) is given below each box-plot. n.s. p > 0.05; * p < 0.05.
Optogenetic inhibition of DL1 DAN activity impairs aversive olfactory memory.
In all panels associative performance indices are shown for tests immediately after classical odor high salt conditioning. In panels (A-E) larvae are trained once by pairing an olfactory stimulus with an aversive high salt stimulation, whereas in panel (F) three training cycles were applied. Schematic overviews for both conditioning protocols are shown to the left of (A) and (F). (A-D) To test whether optogenetic inhibition of the individual DL DANs DAN-c1, DAN-d1, DAN-f1, and DAN-g1 during training impairs punishment signaling, we used the split-Gal4 lines SS02160, MB328B, SS02180, and SS01716 in combination with UAS-GtACR2 and blue light stimulation during the entire training phase. (E, F) For simultaneous optogenetic inhibition of DAN-f1/DAN-g1 we used MB054B. (A-C) Larvae with inhibited DAN-c1, DAN-d1, or DAN-f1 function during training showed no impairment of odor-high salt memory comparable to controls that were kept on standard food without supplemented all-trans-retinal (ATR, 0.5 mM) and received the same protocol (p < 0.05) (for all p > 0.05). (D-F) In contrast inhibition of DAN-g1 alone, or the combination of DAN-f1/DAN-g1 after single trial and three trial conditioning impaired odor-high salt memory compared to controls (for all p < 0.05). This shows that DAN-g1 function is of central importance for signaling a salt punishment teaching signal. All behavioral data is shown as box-plots. Differences between groups are highlighted by horizontal lines between them. Performance indices different from random distribution are indicated below each box-plot. The sample size of each group (N=15) is given below each box-plot. n.s. p > 0.05; * p < 0.05.
Interneurons and hub analysis of sensory to DAN pathways.
(A, B) Schematic graph representation of individual DL1 (A) and pPAM (B) DANs. The outer ring at the bottom of each scheme represents the sensory composition of neurons targeting sensory-to-DANs interneurons (orange circle). The type of sensory information is encoded by the respective color (ACa: anterior central sensory compartment, AVa: ventral anterior sensory compartment, ACal: lateral anterior central sensory compartment, ACp: posterior anterior central sensory compartment, ACpl: posterior-lateral anterior central sensory compartment, VM: ventromedical sensory compartment, TD CO2: tracheal dendritic neurons responding to CO2, ORNs: olfactory receptor neurons). DAN input neurons are shown that get no direct sensory input (grey circles). Individual DANs are shown in the middle of the scheme as black circles. They are connected to MB KC (purple circles), which in turn connect to mushroom body output neuron (MBONs; dark red circles). Arrows indicate the direction of the synaptic connection and its strength (coded by arrow thickness). Numbers in circles indicate number of neurons. The percentages indicate the proportion relative to the total input that the cell receives from the specific neuronal partners. For example, DAN-f1 receives 17% of its input from six different sensory-to-DAN interneurons (yellow circle), 17% from 19 other, non-sensory interneurons, 41% from 64 MB KCs, and 2% from two MBONs. (C) Dot plot showing the importance of interneurons acting as sensory to DAN hub. Dot size was calculated using the fraction of total input an interneuron receives from sensory neurons multiplied by the fraction of total input this interneuron gives to an DAN. Colored backgrounds of dots are highlighted in orange for the connections with a hub size of 0.001 or above. (D, E) Schematic of graph representation. The outer ring represents the sensory composition of neurons targeting upstream neurons of DANs. The type of sensory information is encoded by the respective color. Synaptic threshold for upstream interneurons of DANs = 3 and of upstream sensory neurons = 1. Line thickness to interneurons and targets represents the percentage of synaptic input. White or orange circles connected to the outer ring represent the interneuron layer. The inner ring represents individual target neurons, DL1 and pPAM DANs. The identity of each DAN and interneuron is given by the label in its related circle. (F) EM reconstruction of DL1 and pPAM DANs (grey) highlighting their presynaptic sites in red (for DL1 DANs) and green (for pPAM DANs). At the top a horizontal view of the brain is shown. At the bottom a frontal view of the brain is shown.
