Proposed model for persistent aggressive behavior in females.

Recurrent connectivity between pC1d, pC1e, and aIPg (A) has been proposed to generate a persistent aggressive internal state by prolonging the neural activity of pC1d or aIPg neurons (B) (1, 2). For the connectivity diagram (A), synapse number is noted on each arrow and no thresholds were applied between types.

Brief stimulation of aIPg, but not pC1d alone promotes persistent female aggression.

Cell-specific contributions of aIPg, pC1d, and pC1d+e neurons to persistent aggressive internal state were tested using the sliding door assay (A). Same-genotype testers were separated by a sliding door in the behavioral chamber while receiving a brief 655 nm light stimulation (13+15 seconds). Doors were removed 10 minutes after the stimulation to allow the flies to interact. Level of female aggression (head butting + shoving, red ticks) is shown in the raster plot (B). Fraction of time spent on fighting by each genotype female is compared in the box plot (C). ***p<0.001; **p<0.01; *p<0.05; ns, not significant. Control, empty split-Gal4 driver (BDP-AD; BDP-DBD).

Head butting and shoving JAABA classifier.

pC1d-aIPg recurrent connectivity is not required for aIPg-induced persistent aggressive behavior.

Inhibition of pC1d by Kir2.1 expression did not suppress aIPg-induced persistent aggression. Aggression (head butting+shoving, red ticks) is shown in the raster plot. See Supplementary Figure 5 for quantification. Control, empty split-Gal4 driver (BDP-AD; BDP-DBD).

Neural activity of aIPg neurons is not prolonged by pC1d-aIPg or aIPg-aIPg recurrent connectivity.

A brief 660 nm stimulation that mimics the one used in the behavioral experiment was applied to activate the Chrimson-expressing neurons: pC1d (A or C) and aIPg (B or D). GCaMP responses of pC1d (A or D) or aIPg (B or C) neurons are shown by the blue and green traces, respectively. Dark blue/green line, mean. Light blue/green line, SEM. Tau of GCaMP signals in pC1d neurons is 19.64 seconds and 5.71 seconds when pC1d or aIPg neurons were stimulated, respectively. Tau of GCaMP signals in aIPg neurons is 5.38 seconds and 13.23 seconds when aIPg or pC1d neurons were stimulated, respectively. Area under the GCaMP trace during the 10 second blocks in the baseline (B), stimulation (S), and post-stimulation (P; measured at 3 times Tau after stimulation) period were compared in the box plots shown at right of panel. The difference between B and P in (D) was not significant after multiple comparisons correction. Light blue/green circle, individual data. ***p<0.001; **p<0.01; *p<0.05; ns, not significant.

Comparison of the experimental designs used in Schretter et al. (2020) and Deutsch et al. (2020)

Experimental conditions and results of the behavioral and imaging experiments demonstrated in Schretter et al. (2020) and Deutsch et al. (2020) are compared. For behavioral experiments, the major differences between the two papers are in the stimulation duration (30 seconds vs. up to 5 minutes), gender of the target (female vs. male), and optogenetic effector (CsChrimson vs. ReachR). Based on the cell body location in pan-neuronal GCaMP imaging, (1) conclude that persistent activity was observed in pC1 and aIPg neurons.

Additional circuit components.

Common prominent targets and inputs to aIPg, pC1d, and pC1e neurons. (A) aIPg neurons form direct and indirect connections with pC1e. The most prominent indirect connection goes through the interneuron SMP312. Previous connectomic analysis highlighted SMP054 neurons in a circuit indirectly linking aIPg with LC10 neurons, a specific type of lobula columnar visual output neurons tuned to small moving objects (2, 31, 32). In addition to being common target of aIPg, pC1d, and pC1e, SMP054 receives indirect visual input through SMP312 possibly indicating the presence of another fly. Because our attempts to generate cell-type-specific driver lines for SMP054 have failed, we cannot test the behavior generated by SMP054 activation. Synapse number is noted on each arrow and neurotransmitter predictions (30) are indicated by color (blue, GABAergic; orange, cholinergic). A threshold of 25 synapses was applied, except for the connectivity between pC1d and pC1e. (B) Distribution of synaptic inputs from aIPg (green), pC1d (blue), and SMP312 (yellow) on the neuronal skeleton of pC1e. (C) Distribution of synaptic inputs from aIPg (green) and LC21 (pink) on the neuronal skeleton of SMP312. (D) Distribution of synaptic inputs from aIPg (green), pC1d (blue), pC1e (purple), and SMP312 (yellow) on the neuronal skeleton of SMP054. Select brain regions (SIP, superior intermediate protocerebrum; SMP, superior medial protocerebrum; PVLP, posterior ventrolateral protocerebrum; AOTU, anterior optic tubercle) are outlined in light grey. Dorsal side of the model brain is up, and the lateral is towards the left. Data are from NeuPrint hemibrain v1.2.1.

Stimulation of aIPg, but not pC1d alone promotes persistent attacks toward a wild-type target.

(A) Testers remained aggressive toward a group housed wild-type target after the cessation of aIPg stimulation, but aggression was time-locked to the stimulation period when pC1d neurons were stimulation. Aggression (head butting+shoving, red ticks) was shown in the raster plot, and the fraction of time spent on attacks was compared in the box plot. ****p<0.0001; ***p<0.001; **p<0.01; *p<0.05; ns, not significant. Control, empty split-Gal4 driver (BDP-AD; BDP-DBD).

Density influences the persistence of aggression.

Aggression increased and persisted longer after stimulation with increased fly density in the behavioral chambers. For each experiment, 15 (top) or 40 (bottom) flies were placed in the behavioral chamber. Percentage of flies performing aggressive behavior at a given frame is shown by the color-coded traces. Green, aIPg-activated flies. Purple, pC1d-activated flies. Grey, Control flies (empty split-Gal4 driver, BDP-AD; BDP-DBD). (A) Control > CsChrimson, n = 1 experiment; aIPg-SS > CsChrimson, n = 7 experiments; pC1d+e-SS > CsChrimson, n = 4 experiments; (B) Control > CsChrimson, n = 4 experiments; aIPg-SS > CsChrimson, n = 11 experiments; pC1d+e-SS > CsChrimson, n = 7 experiments.

Inhibition of pC1d activity does not alter aIPg-induced aggressive behavior.

Blocking pC1d activity by Kir2.1 expression did not significantly change persistent behavior induced by aIPg activation. Fraction of time spent performing aggressive behaviors as shown in raster plots in Figure 3. ns, not significant. Control, empty split-Gal4 driver (BDP-AD; BDP-DBD).

A 5-minute stimulation of pC1d neurons does not induce prolonged activity in aIPg neurons.

pC1d (A or C) or aIPg (B or D) neurons were activated for 5 minutes. The blue and green traces showed the GCaMP signals in either pC1d or aIPg neurons, respectively. Dark blue/green line, mean. Light blue/green line, SEM. Area under the GCaMP trace during the 10 second blocks in the baseline (B), stimulation (S), and post-stimulation (P; 3 times Tau after stimulation) period were compared in the box plots. Light blue/green circle, individual data. ***p<0.001; **p<0.01; *p<0.05; ns, not significant.