1. Hypothalamic Circuits for Predation and Evasion
Yi Li, Jiawei Zeng, Juen Zhang, Chenyu Yue, Weixin Zhong, Zhixiang Liu, Qiru Feng, Minmin Luo 
Neuron 
Volume 97, Issue 4, Pages e5 (February 2018)
DOI: /j.neuron
Copyright © 2018 Elsevier Inc. Terms and Conditions

2. Neuron 2018 97, 911-924.e5DOI: (10.1016/j.neuron.2018.01.005)
Copyright © 2018 Elsevier Inc. Terms and Conditions

3. Figure 1 Activating PAG-Projecting LH Neurons Drives Predatory Attack
(A) Schematic showing the injection of AAV-retro-Cre into the periaqueductal gray (PAG) and of AAV-DIO-GCaMP6m into the lateral hypothalamus (LH) in a wild-type mouse. An optical fiber was implanted above the LH.
(B) Schematic of fiber photometry of neuronal activity in the GCaMP6-expressing PAG-projecting LH neurons from hungry C57BL/6 mice while hunting crickets. DM, dichroic mirror; PMT, photomultiplier tube.
(C) Heatmap illustration (upper panel) and mean GCaMP signal (lower panel) aligned to the initiation of cricket hunting from all tested animals during the cricket-hunting task (n = 5 GCaMP6).
(D) Schematic for expressing ChR2-mCherry in PAG-projecting LH neurons in C57BL/6 mice.
(E) ChR2-mCherry expression in the LH. 3v, third ventricle; f, fornix.
(F) Cricket-hunting task and the optogenetic stimulation protocol.
(G) Example picture of the hunting arena with dead crickets after 10 trials of optogenetic activation. Out of total of 10 crickets, 9 were killed (red circle), but not consumed. Insets show close-up views of the dead crickets (blue rectangles).
(H) Heatmap illustration of an animal’s predatory action sequence across 10 stimulation trials.
(I) The probability of predatory attack evoked by optogenetic activation of PAG-projecting LH neurons (n = 5 mGFP; n = 12 ChR2; two-way repeated-measures ANOVA, virus × stimulation, F(1,15) = 132.1, p < , Sidak’s post hoc test).
(J) ChR2-expressing terminals in the l/vlPAG.
(K) Predatory attack probability induced by optogenetic stimulation of axonal terminals from the LH in the PAG (n = 10 ChR2; two-tailed Wilcoxon signed-rank test).
(L) Schematic of slice recording experiments of PAG neurons following optogenetic stimulation of LH terminals.
(M) Optogenetic stimulation of axonal terminals evoked GABAergic IPSC (green; blocked by Gabazine; cell clamped at −10 mV) and glutamatergic excitatory postsynaptic current (EPSC) (red; blocked by 6,7-dinitroquinoxaline-2,3(1H,4H)-dione [DNQX]; cell clamped at −65 mV) in a PAG neuron (n = 6 cells tested).
n.s., not significant; ∗∗p < 0.01; ∗∗∗p < Data are reported as mean ± SEM. See also Figure S1 and Movie S1 first part.
Neuron  , e5DOI: ( /j.neuron )
Copyright © 2018 Elsevier Inc. Terms and Conditions

4. Figure 2 LH GABA Neurons Are Activated during Predation
(A) Schematic of fiber photometry of neuronal activity in the GCaMP6-expressing LH GABA neurons from hungry Vgat-Cre mice while hunting crickets. The blue color indicates the fiber optic. Same conventions are used in the following figures.
(B) Raw traces of Ca2+ signals corresponding to mouse behaviors (attack and consumption) during cricket predation.
(C) Heatmap illustration of neuronal activity changes aligned to the initiation of attack.
(D) Mean GCaMP signal of all animals aligned to the initiation of attack (n = 8 GCaMP6).
(E) Schematic of the computer-controlled food-chasing task. A dish (red) containing food pellets (white) was moved in one of two orthogonal directions (random in each trial) when a mouse entered the trigger zone.
(F) Example images showing animal behaviors during different phases of the food-chasing task.
(G) Example trajectories of a trained mouse.
(H) Raw traces of GCaMP signal corresponding to food chasing and retrieval.
(I) Heatmap illustration of GCaMP signal aligned to food retrieval.
(J) Mean GCaMP signal of all mice aligned to food retrieval (n = 6 GCaMP6).
Data are reported as mean ± SEM. See also Figure S2 and Movie S2.
Neuron  , e5DOI: ( /j.neuron )
Copyright © 2018 Elsevier Inc. Terms and Conditions

5. Figure 3 Inhibition of LH GABA Neurons Suppresses Predatory Behavior
(A) Schematic of expressing GtACR1 in LH GABA neurons for optogenetic inhibition.
(B) The effect of optogenetic inhibition (5-s continuous light in the upper panel and 2 s on and 2 s off pattern in the lower panel) on spontaneous firing activity (n = 6 cells tested).
(C) Cricket-hunting task in hungry Vgat-Cre mice.
(D) Number of crickets hunted in a 20-min session (n = 13 GtACR1 versus n = 6 mGFP control; two-tailed Wilcoxon rank-sum test).
(E) Different phases of the food-chasing task.
(F) Relative distance between a trained mouse and food dish in control (gray) and inhibition (green) trials, in which light was delivered from the initiation phase for the “inhibition” trials (n = 30 trials each).
(G) The effect of light delivery (5 s) from the initiation phase on the rate of successful food retrieval (“hit”) (n = 8 GtACR1; two-tailed Wilcoxon signed-rank test).
(H) The effect of light delivery during the consumption phase on the rate of observing gnawing behavior (n = 8 GtACR1; two-tailed Wilcoxon signed-rank test).
n.s., not significant;∗∗∗p < Data are reported as mean ± SEM. See also Figure S3 and Movie S3.
Neuron  , e5DOI: ( /j.neuron )
Copyright © 2018 Elsevier Inc. Terms and Conditions

6. Figure 4 LH GABA Neurons Drive Predatory Attack
(A) Schematic of expressing ChR2 in LH GABA neurons for optogenetic activation.
(B) ChR2 expression pattern in the LH following the injection of AAV-DIO-ChR2-mCherry vectors into a Vgat-Cre mouse. The white triangle indicates the optical fiber tip.
(C) ChR2 expression pattern in the PAG (right panel).
(D) In a brain slice preparation, brief optogenetic stimulation (5 ms) evoked outward currents in PAG neurons, and the currents were largely abolished by GABAzine (n = 6 cells tested).
(E) The probability of predatory attack evoked by optogenetic activation of LH GABA neurons (n = 11 ChR2 versus n = 6 mGFP control; two-way repeated-measures ANOVA, virus × stimulation, F(1,15) = 152.8, p < , Sidak’s post hoc test).
(F) Artificial prey (wax disk; red) chasing task.
(G) Reduction in the mass of the artificial prey wax disk after 10 trials of LH stimulation (n = 8 ChR2 versus n = 6 mGFP control; two-tailed Wilcoxon rank-sum test).
(H–K) Optogenetic stimulation of LH GABA neurons can switch an animal’s behavior from evasion to predation.
(H) Schematic of the evasion-to-predation transition experiment.
(I) The effects of 15-s optogenetic stimulation on the distance between an approaching disk and a mouse and the locomotor speed of the mouse, illustrating its transition from evasion to predation.
(J) Number of evasion events before and during stimulation (n = 8 ChR2; two-tailed Wilcoxon signed-rank test).
(K) The percentage of light-evoked attack on the approaching disk (n = 8 ChR2 versus n = 6 mGFP control; two-tailed Wilcoxon rank-sum test).
n.s., not significant; ∗∗p < 0.01; ∗∗∗p < Data are reported as mean ± SEM. See also Figure S4 and Movies S1, latter part, S4, and S5.
Neuron  , e5DOI: ( /j.neuron )
Copyright © 2018 Elsevier Inc. Terms and Conditions

7. Figure 5
The Effects of Bidirectionally Manipulating PAG-Projecting LH GABA Neurons on Predatory Behavior
(A) Schematic showing the strategy of expressing GtACR1 in PAG-projecting LH GABA neurons. The experimental group received AAV-retro-DIO-Flp injection into the PAG and received AAV-fDIO-GtACR1 injection into the LH. AAV-retro-DIO-Flp was omitted in the control group.
(B) The number of crickets killed in the 20-min test session by starving animals in both the control and the inhibition of PAG-projecting LH GABA neurons experimental groups of Vgat-Cre mice during the cricket-hunting task. (n = 5 for both groups, two-way repeated-measures ANOVA, virus × session, F(2,16) = 7.922, p = , Sidak’s post hoc test.)
(C) Inhibition of PAG-projecting LH GABA neurons starting from the time of food chasing initiation decreased the successful hit rate during the food-chasing task (n = 5, two-tailed Wilcoxon signed-rank test).
(D) Inhibition of PAG-projecting LH GABA neurons starting after the animal has retrieved the food pellet did not affect an animal’s gnawing behavior during the food-chasing task (n = 5, two-tailed Wilcoxon signed-rank test).
(E) Schematic showing the stimulation of the ChR2-expressing terminals from the LH in the PAG of Vgat-Cre mice.
(F) Predatory attack probability evoked by terminal stimulation in the cricket-hunting task (n = 13 ChR2; two-tailed Wilcoxon signed-rank test).
(G) Schematic showing the strategy of expressing ChR2 in PAG-projecting LH GABA neurons. We injected AAV-retro-DIO-Flp into the PAG and AAV-fDIO-ChR2 into the LH. AAV-retro-DIO-Flp was omitted in the control group.
(H) The effect of optogenetic stimulation of PAG-projecting LH GABA neurons on the predatory attack probability in the cricket-hunting test (n = 4 control and n = 6 ChR2, two-way repeated-measures ANOVA, virus × stimulation, F(1,9) = 28.01, p = , Sidak’s post hoc test).
(I) Schematic showing the strategy of expressing ChR2 in PAG-projecting LH neurons and hM3Dq in PAG glutamate neurons.
(J) The effect of applying CNO (1 mg/kg, i.p.) for chemogenetic activation of PAG glutamate neurons on cricket hunting by hungry mice (n = 5 per group; two-way repeated-measures ANOVA, virus × injection, F(1,8) = 9.761, p = , Sidak’s post hoc test).
(K) The effect of chemogenetic activation of PAG glutamate neurons on cricket hunting as evoked by optogenetic stimulation of PAG-projecting LH neurons (n = 6 for mGFP group; n = 5 for hM3Dq group, two-way repeated-measures ANOVA, virus × injection, F(1,9) = 91.92, p < , Sidak’s post hoc test).
n.s., not significant; ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < Data are reported as mean ± SEM. See also Figure S5.
Neuron  , e5DOI: ( /j.neuron )
Copyright © 2018 Elsevier Inc. Terms and Conditions

8. Figure 6
The Activity of PAG-Projecting LH Glutamate Neurons Is Necessary for Making Predictive Evasion
(A) Schematic showing fiber photometry recording during the artificial attacker evasion task.
(B) Raw traces of the GCaMP signal of LH glutamate neurons engaged in the evasion task. The arrow indicates that GCaMP6 signal increased before the onset of evasion behavior.
(C) Heatmap illustration of GCaMP signals across trials for one mouse.
(D) Mean GCaMP signal of all mice tested (n = 6 GCaMP6).
(E) Schematic for expressing GtACR1 in LH glutamate neurons for optogenetic inhibition.
(F) Real-time plotting of the relative distance (solid line) between the artificial attacker and the mouse and the speed (dashed line) of the mouse from a control trial (black) and a GtACR1 inhibition trial (green) during the evasion task. The red horizontal line indicates the nearest distance between the attacker and the mouse in control trials. The trained mouse could predict the upcoming danger and evade it before attack in control trials. The mouse failed to predict the upcoming danger in GtACR1 inhibition trials.
(G) During the evasion task, inhibition of LH glutamate neurons shortened the distance between the artificial attacker and the mouse at the point of mouse evasion (n = 8 GtACR1; two-tailed Wilcoxon signed-rank test).
(H) Optogenetic inhibition of LH glutamate neurons did not affect evasion speed (n = 8 GtACR1; two-tailed Wilcoxon signed-rank test).
(I) Schematic showing the strategy of expressing GtACR1 in PAG-projecting LH glutamate neurons. The experimental group received AAV-retro-DIO-Flp injection into the PAG and received AAV-fDIO-GtACR1 injection into the LH. AAV-retro-DIO-Flp was omitted in the control group.
(J) Median value of the minimum distance between the artificial attacker and the mouse in both control and inhibition experimental group in Vglut2-ires-Cre mice during the artificial-attacker-evasion task (n = 5 for both control group and experimental group, two-way repeated-measures ANOVA, virus × session, F(1,8) = 8.939, p = , Sidak’s post hoc test).
(K) Median value of the peak evasion speed during the 15-s evasion bout in both control and the inhibition of PAG-projecting LH glutamate neurons experimental group during the artificial-attacker-evasion task. (n = 5 for both control group and experimental group, two-way repeated-measures ANOVA, virus × session, F(1,8) = 5.144, p = , Sidak’s post hoc test)
n.s., not significant; ∗∗p < Data are reported as mean ± SEM. See also Figure S6 and Movie S6.
Neuron  , e5DOI: ( /j.neuron )
Copyright © 2018 Elsevier Inc. Terms and Conditions

9. Figure 7
Activating PAG-Projecting LH Glutamate Neurons Drives Evasion Behavior
(A and B) Schematic (A) of optogenetic activation of ChR2-expressing LH glutamate neurons (B) from a Vgult2-ires-Cre mouse.
(C) The effect of optogenetic stimulation on the locomotor speed of a mouse (n = 10 trials). Blue line indicates the stimulation period.
(D) Mean animal speed before, during, and after activation (n = 8 ChR2; one-way repeated-measures ANOVA, F(2, 14) = 19.04, p = , Dunn’s multiple comparisons test).
(E) Schematic showing the behavioral switch from predation to evasion upon optogenetic stimulation of LH glutamate neurons at the time of food retrieval.
(F) Example trajectories of a hungry mouse during the food-chasing task (blue, stimulation trial; black, control trial).
(G) Mean distance between a mouse and the food dish during the food-chasing task (blue, stimulation trials; black, control trials).
(H) ChR2-expressing terminals in the l/vlPAG following the injection of AAV-DIO-ChR2-mCherry vectors into the LH of a Vglut2-ires-Cre mouse.
(I) Brief optogenetic activation of LH terminals evoked EPSC in a PAG neuron from a ChR2-expressing Vglut2-ires-Cre mouse. The EPSC was blocked by DNQX (n = 6 cells). The test neurons were held at −65 mV in voltage clamp mode.
(J) The changes in an example mouse’s speed upon optogenetic stimulation of the terminals of LH glutamate neurons in the PAG. The black line indicates the mean value of the ten-trial stimulation (gray line).
(K) Animal speed before, during, and after activation of glutamatergic terminals in the PAG (n = 6 ChR2, one-way repeated-measures ANOVA, F(2,10) = 6.226, p = 0.0175, Dunn’s multiple comparisons test).
(L) Schematic of expressing ChR2 in PAG-projecting glutamate neurons in the LH.
(M and N) The effect of optogenetic stimulation of PAG-projecting LH glutamate neurons on locomotor speed from an example (M) and grouped data (N). The control group received AAV-fDIO-ChR2 injection into the LH but did not receive injection of AAV-retro-DIO-Flp into the PAG (n = 4 for control group and n = 6 for experimental group; two-way repeated-measures ANOVA, virus × stimulation, F(2,16) = 6.064, p = , Sidak’s post hoc test).
n.s., not significant; ∗∗p < 0.01;∗∗∗p < Data are reported as mean ± SEM. See also Figure S7 and Movie S7.
Neuron  , e5DOI: ( /j.neuron )
Copyright © 2018 Elsevier Inc. Terms and Conditions