Medial preoptic area antagonistically mediates stress-induced anxiety and parental behavior

Abstract

Anxiety is a negative emotional state that is overly displayed in anxiety disorders and depression. Although anxiety is known to be controlled by distributed brain networks, key components for its initiation, maintenance and coordination with behavioral state remain poorly understood. Here, we report that anxiogenic stressors elicit acute and prolonged responses in glutamatergic neurons of the mouse medial preoptic area (mPOA). These neurons encode extremely negative valence and mediate the induction and expression of anxiety-like behaviors. Conversely, mPOA GABA-containing neurons encode positive valence and produce anxiolytic effects. Such opposing roles are mediated by competing local interactions and long-range projections of neurons to the periaqueductal gray. The two neuronal populations antagonistically regulate anxiety-like and parental behaviors: anxiety is reduced, while parenting is enhanced and vice versa. Thus, by evaluating negative and positive valences through distinct but interacting circuits, the mPOA coordinates emotional state and social behavior.

Extended Data Fig. 1. Quantification pipeline for c-fos staining and Vglut2 expression in mPOA. (Associated with Fig. 1)

a. Illustration of force swimming application. The bottom of the test chamber was a metal mesh. For the control experiment, the animal was placed in the same context without being submerged into water (condition on the left). b. Illustration of heat plate application. For the control experiment, the animal was placed in the same context without being touched by the heat plate (condition on the left). c. Protocol for c-fos staining and imaging. Animals were exposed to one of the stressors for 5 min and were sacrificed 3 hours after the treatment. Scale, 100 μm. d. Pipeline for image processing and cell counting. e. Spatial distribution of c-fos+ cells under treatments of three different stressors. LPO, lateral preoptic area; VLPO, ventral lateral preoptic area. Scale, 200 μm. f. Left, Nissl staining; right, tdTomato expression in the same coronal brain section. Images were obtained from transgenic mice by crossing Ai75 (Cre-dependent nucleus-targeted tdTomato reporter) and Vglut2-Cre. g. A more posterior section. BAC, bed nucleus of the anterior commissure; aco, anterior commissure. Scale, 500μm. Images in c,d are representative of n = 9animals, Images in e,f,g are representative of n=3 animals.

Extended Data Fig. 2. Quality control for fiber photometry. (Associated with Fig. 1)

a. Illustration of the fiber photometry setup. A protective cover helps to prevent the optic fiber from bumping against the wall of the test box/chamber. Neurons express Cre-dependent GCaMP6s. b. Example full trace of calcium signals in the control condition for forced swimming. Dashed line indicates the presumptive operation time (no operation was actually applied). c. Example full trace of calcium signals for forced swimming application. Bar represents the exposure duration. d. Example full trace of calcium signals for heat plate application. e. Plot of calcium transients (blue) and concurrent locomotion speed (red, freely moving) in an open field. Z-score = 3 was used as the detection threshold. f. Plot of locomotion speed vs. amplitude of calcium transients. g. Spearman r calculated for each mouse. N= 8 animals. Bars represent mean ± s.d. h. Fluorescence signals in a control animal expressing GFP only. Black bar marks duration of forced swimming exposure. i. Peak calcium transients during the baseline period and stressor application for GFP control animals (n = 4). Statistics can be found in Fig.1. FST, forced swimming test. j. Heat plate exposure in GFP control animals (n = 4). k. Electric shock exposure in GFP control animals (n = 4). (see Supplementary Table 1 for detailed statistics).

Extended Data Fig. 3. Optrode recording. (Associated with Fig. 1)

a. Top, example full trace of fiber photometry recording (left, each black dot represents one application of foot shocks) and average Ca²⁺ response (averaged over trials) to a foot shock application (right). b. Top, raster plot of LED-induced spike responses for an example mPOA glutamatergic neuron. Blue dots indicate the duration of LED pulse (5ms). Bottom, corresponding peri-stimulus spike time histogram (PSTH). Raster plot and PSTH for spikes induced by a single LED pulse (blue line). Thick black line indicates the duration of LED stimulation. Only cells show 1^st spike latency shorter than 4ms were considered as valid optogenetically-identified Vglut2+ neurons and included for analysis. c. Example full trace of single-unit responses to repeated foot shock stimulation. Spontaneous spikes before the first and last electric shock application are shown on top for visualization. d. Spontaneous firing rates of mPOA opto-IDed Vglut2+ neurons at different time points after exposure to foot shocks. N = 19 cells from 2 animals. e. Center time in OFT performed at different time points after exposure to foot shocks. N = 7 animals. f. Heatmaps of single-cell spike responses to heat (top) and electric shocks (bottom) of opto-IDed Vglut2+ neurons. The same cells are shown to demonstrate multimodal responses. g. Population average of PSTHs from cells shown in (f). h. Heatmap of single-cell spike responses to electric shocks for non-optotagged (presumably Vglut2−) neurons. Pie chart shows the percentage of presumably Vglut2− neurons that shows activated, no, or suppressive responses to electric shocks, respectively. Spontaneous firing rates of non-optotagged neurons at different time points after exposure to foot shocks. N = 35 cells from 2 animals.

Extended Data Fig. 4. ∣ Locations of somas with viral expression. (Associated with Fig. 2-4)

Schematic coronal sections ranging from 0.98 mm anterior to 1.06 mm posterior to Bregma. a. Left two, Cre-dependent GFP expression at the injection site in a Vglut2-Cre mouse and spatial distribution of expressing cell bodies. Right three, axons in more posterior sections. Same image as in Fig. 5a. Blue, Nissl staining. Scale, 300μm. b. . a, Representative images of GFP-labeled mPOA glutamatergic neurons (left) and their axons in PAG (right). Scale bar: 500 μm. c. Superimposed ChR2-EYFP expressing cell locations for all mice from anterior to posterior sections. Each small red dot represents one cell. d. Superimposed hM3Dq-mCherry expressing cell locations for all mice. e. Superimposed hM4Di-mCherry expressing cell locations for all mice. f. Superimposed ArchT-GFP expressing cell locations for all mice. Images in a are representative of n=6 animals.

Extended Data Fig. 5. Activation of mPOA Vglut2 neurons. (Associated with Fig. 2)

a. Experimental setup for conditioned place preference test. During conditioning, the animal was subjected to LED stimulation whenever it was in the conditioned chamber. b. Conditioned place aversion tested 24 hours after paring photo-stimulation with one chamber. Time spent in the conditioned chamber (Cond) or unconditioned chamber (Un-cond) was quantified. **P < 0.01, Mann–Whitney test, n = 5 animals. c. Quantification of center time in OFT under different light stimulation frequencies (Kolmogorov–Smirnov test with Bonferroni correction, P < 0.001, n = 5 and 5 animals for GFP control and ChR2 groups respectively). Each animal was tested for one session per day with stimulation frequencies randomly selected. d. Quantification of open-arm time in EPM (Kolmogorov–Smirnov test with Bonferroni correction, P < 0.001, n = 5 and 5 animals for GFP control and ChR2 groups respectively). e. The OFT arena was divided into 16 subregions and locomotion speed were calculated for each specific subregion. f. Locomotion speed in center vs peripheral subregions. Each dot is one animal. N.S., non-significant; paired t test. N=12. g. The EPM arena was divided into 9 subregions. h. Locomotion speed in closed-arm, open-arm and center subregions. N=12. i. Expressing hM3Dq receptors in mPOA glutamatergic neurons. j. Raw recorded trace of the membrane potential of a hM3Dq-expessing mPOA glutamatergic neuron in response to CNO application in slice recording. k. Subthreshold membrane potential voltages before and after perfusion of CNO as well as after washing out CNO. **P < 0.01, one-way repeated-measures ANOVA, n = 5 cells from 2 animals. l. Quantification of center time in OFT for mCherry control and hM3Dq expressing animals. **P < 0.01, Mann–Whitney test, n = 6 animals for each group. m. Quantification of open-arm time in EPM for mCherry control and hM3Dq expressing animals. **P < 0.01, Mann–Whitney test, n = 6 animals for each group. n. Experimental timeline: expressing ChR2 in mPOA glutamatergic neurons, optogenetic stimulation for 5-min (20Hz) and anxiety-related behavioral test one hour later. o. Quantification of center time in OFT for GFP control and ChR2-expressing groups. **P < 0.01, Mann–Whitney test, n = 7 animals for each group. Quantification of open arm time in EPM for GFP control and ChR2-expressing groups. **P < 0.01, Mann–Whitney test, n = 7 animals for each group. (see Supplementary Table 1 for detailed statistics).

Extended Data Fig. 6. Strong activation of mPOA glutamatergic neurons. (Associated with Fig. 2)

a. Illustration of conflict tests. b. Movement tracing of a GFP control animal (left) and a ChR2-expressing animal (right) in a conflict test. Photostimulation at 15 Hz was applied whenever the mouse was in the light gray marked zone. c. Percentage time spent in the physically harmful side for control and ChR2 animals. *P < 0.05; **P < 0.01, Mann–Whitney test, n = 6 animals for each group. d. Photo of mouse rearing during LED activation of mPOA glutamatergic neurons at 15Hz. e. Frequency of rearing in GFP control (n = 7) and ChR2 expressing (n = 5) animals. **P < 0.01, Mann–Whitney test with Bonferroni correction. f. Photo of mouse jumping during LED activation of nPOA glutamatergic neurons at 15Hz. g. Frequency of jumping in GFP control (n = 7) and ChR2-expressing (n = 5) animals. **P < 0.01, Mann–Whitney test with Bonferroni correction. h. Food intake within 2 hours after being food-deprived for 24 hours. During the 2h test, mPOA glutamatergic neurons were photo-stimulated continuously. **P < 0.01, Mann–Whitney test, n = 5 animals for each group. i. Frequency of rearing in mCherry control (n = 7) and hM3Dq-expressing (n = 6) animals. **P < 0.01, Mann–Whitney test with Bonferroni correction. j. Comparison of rearing frequency between mCherry control (n = 7) and hM3Dq- expressing (n = 6) animals. **P < 0.01, Mann–Whitney test with Bonferroni correction. k. Food intake within 2 hours after being food-deprived for 24 hours. CNO injection (i.p.) was performed 20 min before the test. **P < 0.01, Mann–Whitney test with Bonferroni correction, n = 5 for each group. l. Strategy of viral injection. m. Movement tracing of a GFP control animal (upper) and an ArchT-expressing animal (lower) in a two-chamber place preference test. Continuous LED stimulation was applied whenever the animal stayed in the light gray marked chamber. n. Percentage time spent in the LED-on chamber. **P<0.01, Mann-Whitney test, n = 9 (5 males) for each group. o. Upper, experimental time line. LED light was applied only during the electric shocks. Center time in OFT for GFP control (n = 7) and ArchT-expressing (n = 7; 4 males) animals. **P < 0.01, Mann–Whitney test. p. Open-arm time in EPM for GFP control (n = 7) and ArchT-expressing (n = 7; 4 males) animals. **P < 0.05, Mann–Whitney test, n = 7 animals for each group. (see Supplementary Table 1 for detailed statistics).

Extended Data Fig. 7. Anxiety tests after exposure to social stress. (Associated with Fig. 4)

a. Anxiety-like behaviors of virgin males with or without exposure to male intruders. **P < 0.01, Mann–Whitney test, n = 8 animals for each group. b. Anxiety-like behaviors for virgin males with or without exposure to foreign pups. **P < 0.01, Mann–Whitney test, n = 10 animals for each group. c. Duration of pup grooming for virgin males in LED-off and LED-on conditions. **P < 0.01, Mann–Whitney test with Bonferroni correction, n = 10 animals for each group. d. Anxiety-like behaviors of virgin females with or without exposure to foreign pups. “n.s.”, not significant, Mann–Whitney test, n = 10 animals for each group. e. Anxiety-related tests in virgin males/females not exposed to stress with (green) and without (grey) optogenetic silencing of mPOA Vglut2 neurons. *P < 0.05, **P < 0.01, Mann–Whitney test with Bonferroni correction, n = 7 animals for each group. f. Anxiety-related tests in virgin females not exposed to stress with (green) and without (grey) chemogenetic silencing of mPOA Vglut2 neurons. **P < 0.01, Mann–Whitney test with Bonferroni correction, n =8 animals for each group. g. Left, schematic open field test with a shelter. Right, total time spent in shelter within a 5-min test session. **P < 0.01, Mann–Whitney test, n = 6 (3 males) animals. h. Total time spent in the light side of a light-dark box. **P < 0.01, Mann–Whitney test, n = 7 (4 males) animals for each group. (see Supplementary Table 1 for detailed statistics).

Extended Data Fig. 8. Potential targets of mPOA glutamatergic neurons. (Associated with Fig. 5)

a. Imaging area (top) and a confocal image (bottom) showing EYPF labeling in mPOA (left) and different downstream structures. Scale, 500μm. VMH, ventral medial hypothalamic nucleus; MM, medial mammillary nucleus. b. Schematic terminal stimulation in a potential mPOA target. c. Frequency of jumping induced by activating axonal terminals of mPOA glutamatergic neurons in different target areas. **P < 0.01, One-way ANOVA test, n = 5 animals for each group. GFP control group is for animals expressing GFP in mPOA with fiber implantation in mPOA. d. Two-chamber place preference test when activating axonal terminals of MPO glutamatergic neurons in different target areas. **P < 0.01, One-way ANOVA test, n = 5 animals for each group. e. Left, viral injection strategy to express ChR2 in PAG-projecting mPOA neurons in Ai14 mice. Right, images showing tdTomato- and ChR2-EYFP labeled neurons in mPOA. Scale bar, 500 μm. f. Left, viral injection strategy to transsynaptically label mPOA-recipient PAG neurons. Right, image showing labeled neurons in PAG (with a blow-up image on the right). Scale bar, 500 μm. Images in a,e are representative of n=5 animals. Images in f are representative of n =3 animals. (see Supplementary Table 1 for detailed statistics).

Extended Data Fig. 9. Manipulation of the mPOA to VTA pathway. (Associated with Fig. 5)

a. Top, viral injection strategy. Optic cannula was implanted above VTA. Bottom, labeling of dopamine neurons (red) by crossing DAT-Cre and Ai14 mice (left) and mPOA glutamatergic axons in VTA and surrounding regions (right). Scale, 400μm. b. Percentage time spent in the LED-on chamber in PPT for GFP control and ChR2-expressing animals. **P<0.01, Mann-Whitney test, n = 7 (4 males) for each group. c. Center time in OFT. N.S., no statistical difference, Mann–Whitney test, n = 7 animals for each group. d. Open-arm time in EPM. Mann–Whitney test, n = 7 animals for each group. e. Double retrograde dye injection in PAG (green) and VTA (red). f. Representative image showing retrogradely labelled neurons in mPOA. g. Quantification of singly and doubly labeled neurons in mPOA. N=4 animals. h. Viral strategy to label axon collaterals of PAG-projecting mPOA neurons in VTA. i. Images showing many labeled axons in PAG (left two), but extremely sparse axons in VTA (right two). Injection site is shown in Fig. 5b (shared blue channel as reference). Scale, 500μm. Images in a,i are representative of n =3 animals, images in f are representative of n = 4 animals. (see Supplementary Table 1 for detailed statistics).

Extended Data Fig. 10. Control experiments for manipulating neurons in mPOA and proposed circuit model. (Associated with Fig. 6)

a. Open-arm time for GFP control animals in LED-off and LED-on conditions. “n.s.”, non-significant, Mann–Whitney test, n = 10 animals for each group. b. Schematic pup exposure test for virgin males or females. c. Percentage of trials with pup retrieval for GFP control virgin females in LED-on and LED-off conditions respectively. “n.s.”, non-significant, Fisher’s exact test, n = 7 animals for each group. Each animal was subjected to 2-4 trials and all trials were pooled together. d. Duration of pup grooming for GFP control virgin females. “n.s.”, non-significant, Mann–Whitney test, n = 9 animals for each group. e. Percentage of trials with pup attack for GFP control virgin males in LED-on and LED-off conditions respectively. “n.s.”, non-significant, Fisher’s exact test, n = 7 animals for each group. Each animal was subjected to 2-4 trials and all trials were pooled together. f. Left, viral injection strategy and implantation of the optic fiber above PAG. Right, image showing axons of mPOA GABAergic neurons in PAG. Scale, 500 μm. g. Representative images showing injection sites in VLPO (top) and MnPO (bottom). Scale, 500μm. h. Core body temperature measured after 10-Hz LED stimulation for 30 min in animals expressing ChR2 in mPOA, ventral lateral preoptic area (vLPO) and median preoptic area (MnPO) respectively in Vglut2-Cre animals. **P < 0.01, two-tailed t-test, n = 5 animals for each group. bar, s.d. i. Left, photo of a freely moving mouse in an open arena, with a 2x2x2cm 3D object placed in the center. The behavioral test consisted of 3 blocks: LED-off, LED-on, and then LED-off, with each lasting 3 min. 10-Hz LED stimulation was applied during the LED-on block. Right, dislocation of the 3D object by the mouse. No statistical significance was observed between blocks; two-way repeated-measures ANOVA, n = 5 animals. j. CTb488 injection in PAG of either Vglut2-Cre::Ai14 or Vgat-Cre::Ai14 mice. k. Images showing overlap between CTb-labeled and Vglut2+ (top) / Vgat+ (bottom) neurons in mPOA. Scale, 200 μm. l. Quantification of the percentage of PAG-projecting mPOA neurons that are Vglut2+ or Vgat+. **P<0.01, Mann-Whitney test with Bonferroni correction, n = 3 animals for each group. Bar, s.d. m. Representative image showing the CTb injection site in medial PAG. n. Quantification of the percentage of Vglut2+ or Vgat+ mPOA neurons that were labeled by CTb injected in PAG. **P<0.01, Mann-Whitney test with Bonferroni correction, n = 3 animals for each group. Bar, s.d. o. Illustration of the proposed circuit model. Note that due to potentially different inputs, glutamatergic neurons in mPOA respond to physical and social stressors but not to social rewards, while GABAergic neurons are activated by social rewards (e.g. during parenting) but not stressors. Images in f are representative of n = 8 animals. Images in g are representative of n = 5 animals. Images in k, m are representative of n = 3 animals. (see Supplementary Table 1 for detailed statistics).

Fig. 1∣. mPOA glutamatergic neurons are activated by physical stress.

a, Experimental timeline for measuring anxiety-like behaviors following 5-min treatment of either forced swimming, heat plate, or 0.5-Hz 0.3-mA electric shocks. b, Tracing of locomotion for representative control (gray) and experimental (dark red) animals exposed to different stressors. Light gray square marks the designated center zone. c, Quantification of percentage time spent in the center zone in OFT (n = 6 animals for each group, 3 males, 3 females; *P<0.05, **P < 0.01, two-way ANOVA and post hoc test). d, Locomotion traces for representative control and stressor-exposed animals. Light gray marks closed arms. e, Quantification of percentage time spent in the open arms of EPM (n = 6 animals for each group, 3 males, 3 females; ** P < 0.01, two-way ANOVA and post hoc test). f, Quantification of the number of c-fos+ cells per 50 μm² in different brain regions. mPOA, medial preoptic area; LHA, lateral hypothalamic area; Amg, amygdala; VMH, ventromedial hypothalamus; BNST, bed nucleus of the stria terminalis; PAG, periaqueductal gray. *P<0.05, **P < 0.01, two-way ANOVA and post hoc test, n = 3 animals. Top inset, experimental timeline for c-fos staining. g, Representative confocal images of c-fos staining in mPOA for a control and an experimental animal exposed to electric shocks. Scale bar: 50 μm. h, Quantification of number of c-fos+ neurons in mPOA for each treatment (n = 3 animals for each group; **P < 0.05, one-way ANOVA and post hoc test; control animals were combined). i, Representative images showing colocalization of Vglut2 (reflected by tdTomato expression in Vglut2-Cre::Ai14 mice) and c-fos (green) signals. Blue is Nissl staining. Scale bar: 25 μm. j, Quantification of percentage of Vglut2+ cells in the c-fos+ population (brown) and percentage of c-fos+ cells in the Vglut2+ population (green). k, Left, experimental setup for photometry. Right, heatmap of Ca²⁺ signals to 20 trials of electric shocks (duration marked by a thick dark line) for an example animal. Bottom panel shows the averaged trace for the shock (solid red) and control (solid black, with no current output) condition, with pale colors indicating individual trials. Dashed line marks the onset of shocks. l, Top, baseline fluorescence signals within 1-s window just before the shock onset over 20 trials (red) for the same animal shown in k. Black is for a control animal. P < 0.001, Kolmogorov–Smirnov test with Bonferroni correction. Bottom, baseline fluorescence signal at the 1^st and 20^th trials for 9 animals. **P < 0.01, two-sided paired t test. m, Quantification of stressor-induced peak ΔF/F (%) for control and exposed animals (n = 5 for each group, 3 males, 2 females; **P < 0.01, one-way ANOVA and post hoc test; control animals were combined). n, Top, diagram for optrode recording from mPOA in head-fixed Vglut2-Cre::ChR2 mice. Bottom, sample recorded traces of spiking of a ChR2-expressing mPOA glutamatergic neuron to pulses of LED stimulation (blue dots). Right inset, comparison of spike waveforms (slightly offset) spontaneous generated (red) and evoked by LED stimulation (blue) of the same unit. o, Spontaneous firing rates across time (bin size: 10 min) before and after exposure to electric shocks (marked by vertical gray bar). P < 0.001, two-sided Kolmogorov–Smirnov test with Bonferroni correction; n = 35 and 29 neurons for control and experimental groups respectively. Bars represent s.e.m. Images in g are representative of n=3 animals. Data in l is representative of n = 5 animals. (Extended Data Fig. 1; see Supplementary Table 1 for detailed statistics).

Fig. 2∣. Activating mPOA glutamatergic neurons enhances anxiety-like behaviors.

a, Left, injection and stimulation configuration. Right, representative image showing expression of ChR2-EYFP in mPOA. Scale bar: 500 μm. b, Left, example traces of spiking of a recorded mPOA glutamatergic neuron to pulses of blue light at 1, 5 and 10 Hz. Right, fidelity of spiking at different stimulation frequencies (n = 10 cells). Bar = s.e.m. c, Left, example membrane potential responses to injections of square (upper) or ramp (lower) currents. Resting membrane potential was −63 mV. Right, average firing rates to injected currents at different amplitudes (n = 10 cells). Bar = s.e.m. d, Representative locomotion tracing for a GFP control and a ChR2-expressing animal in the two-chamber place preference test. e, Quantification of percentage time in the LED-on chamber. **P < 0.01, two-sided Mann–Whitney test with Bonferroni correction, n = 10 animals for each group. Males and females are separately displayed. Error bars, s.d. f, Locomotion tracing for an example ChR2-expressing animal with continuous photostimulation in OFT (upper) or EPM (lower). g, Quantification of center time in OFT. **P < 0.01, two-sided Mann–Whitney test with Bonferroni correction, n = 12 animals for each group. Error bars, s.d. h, Quantification of open-arm time in EPM. **P < 0.01, two-sided Mann–Whitney test with Bonferroni correction, n = 12 for each group. Error bars, s.d. i, Changes in pupil size in 60 trials of photostimulation (LED-on and LED-off trials were randomly assigned). Top, average change of pupil size in the LED-on (solid red) and LED-off (solid black) condition. Pale colors represent individual trials. Bottom, heatmap for change in pupil size in different trials aligned by the onset of LED stimulation (blue bar). Left inset, sample images of the eye at time points a and b. j, Average change in pupil size in LED-off and LED-on conditions. **P < 0.01, two-sided paired t-test, n = 11 animals for each group. Data points for the same animal are connected with a line. k, Traveling distance in an open arena in LED-off (3 min per block) and LED-on (3 min) conditions. **P < 0.01, two-way repeated-measures ANOVA, n = 10 animals for each group. Images in a are representative of n=3 animals. (see Supplementary Table 1 for detailed statistics).

Fig. 3∣. Silencing mPOA glutamatergic neurons reduces stress-induced anxiety-like behaviors.

a, Viral injection for chemogenetic silencing. b, Top, raw trace of current-clamp recording from a hM4Di-expressing mPOA glutamatergic neuron in the slice preparation. Bottom, average spontaneous spike frequencies before and after perfusion in of CNO as well as after washing out CNO. **P < 0.01, one-way repeated-measures ANOVA, n = 5 cells from 2 mice. Error bars, s.d. c, Movement tracing for an example mCherry control (gray) and a hM4Di-expressing (red) animal in OFT 1 hr after exposure to electric shocks, with CNO injected at 40 min. Top inset, experimental timeline. d, Percentage center time in OFT in mCherry control and hM4Di animals after shock exposure. **P < 0.01, two-sided Mann–Whitney test, n = 7 and 6 mice respectively, 3 males. Saline control experiments were performed on a different day for the same animal. Error bars, s.d. e, Movement tracing for a mCherry control (gray) and a hM4Di-expressing (red) animal in EPM test after exposure to electric shocks. f, Percentage open-arm time in EPM in mCherry control and hM4Di animals after shock exposure. **P < 0.01, two-sided Mann–Whitney test, n = 6 and 5 mice respectively, 3 males. Error bars, s.d. g, Viral injection and stimulation for optogenetic silencing. h, Top, membrane potential response to green LED stimulation in an ArchT-expressing mPOA glutamatergic neuron in the slice preparation. Bottom, average spontaneous spike rates before, during and after LED stimulation (marked by the green rectangle). **P < 0.01, one-way repeated-measures ANOVA, n = 5 cells from 2 mice. i, Movement tracing for an ArchT-expressing animal in OFT within a LED-off (gray) and a LED-on (green) block. j, Percentage center time for GFP control (black) and ArchT (green) animals in LED-off and LED-on blocks of OPT. **P < 0.01, two-way repeated-measures ANOVA, n = 8 and 9 animals (4 males) for control and ArchT groups respectively. Error bars, s.e.m. k, Movement tracing for a GFP control (gray) and an ArchT-expressing animal (green) in the EPM test. l, Percentage open-arm time for GFP control and ArchT animals. **P < 0.01, two-sided Mann–Whitney test, n = 6 animals (3 males) for each group. Error bars, s.d. (see Supplementary Table 1 for detailed statistics).

Fig. 4∣. The mPOA glutamatergic neurons antagonistically regulate social stress-induced anxiety and parental behavior.

a, Exposing a resident mouse to a pup or a younger intruder while imaging ensemble Ca²⁺ activity of mPOA glutamatergic neurons using photometry. b, A representative trace for GCaMP6s fluorescence change in a virgin male when exposed to a pup. Colored bars indicate bouts of interaction with the pup. c-e, Percentage changes in fluorescence in virgin males before (pre-) and after (post-) putting in the home cage a male intruder (c), a pup (d) or a female (e). ***P < 0.01, two-sided Mann–Whitney test, n = 5 animals. Error bars, s.d. f, Percentage changes in fluorescence in virgin females (n = 5) before and after putting in a pup. Error bars, s.d. g, Experimental condition: a virgin male exposed to a younger male intruder. h, Percentage center time in OFT (left) and open-arm time in EPM (right) for intruder exposed resident males in LED-off (gray) and LED-on (green) conditions. **P < 0.01, two-sided Mann–Whitney test, n = 8 ArchT animals. Error bars, s.d. i, Percentage of trials with the resident fighting against intruder in LED-off (gray) and LED-on (green) conditions for GFP control and ArchT animals. **P < 0.01, two-sided Fisher’s exact test with Bonferroni correction, n = 8 animals in each group; “ns”, non-significant. Total number of trials is marked. j, Percentage center times in OFT (left) and open-arm times in EPM (right) without (gray) and with (green) optogenetic silencing of mPOA glutamatergic neurons during exposure to intruder. **P < 0.01, two-sided Mann–Whitney test, n = 8 ArchT animals. Error bars, s.d. k, Experimental condition: a virgin male exposed to a pup. l, Percentage center times in OFT for GFP control (left, n = 9) and ArchT (right, n = 9) animals in LED-off (gray) and LED-on (green) conditions. **P < 0.01, Mann–Whitney test; “ns”, non-significant. Error bars, s.d. m, Percentage of trials with the virgin male showing pup attacks in LED-off and LED-on conditions for GFP (n = 9) and ArchT (n = 9) animals. **P < 0.01, Fisher’s exact test with Bonferroni correction. Total number of trials is marked. n, Percentage center times in OFT (left) and open-arm times in EPM (right) for pup-exposed males without (gray) and with (green) optogenetic silencing of mPOA glutamatergic neurons during pup exposure. **P < 0.01, Mann–Whitney test, n = 8 ArchT animals. Error bars, s.d. o, Experimental condition: a resident virgin female exposed to a pup. p, Percentage center times in OFT for GFP control (left, n = 11) and ArchT (right, n = 11) animals in LED-off (gray) and LED-on (green) conditions. **P < 0.01, Mann–Whitney test with Bonferroni correction. Error bars, s.d. q, Percentage of trials with pup retrieval for GFP (n = 11) and ArchT (n = 11) female mice in LED-off and LED-on conditions. **P < 0.01, Fisher’s exact test with Bonferroni correction. r, Total duration of pup grooming by the virgin female in LED-off and LED-on conditions. **P < 0.01, Mann–Whitney test with Bonferroni correction, n = 11 animals for both GFP and ArchT groups. Error bars, s.d. (see Supplementary Table 1 for detailed statistics).

Fig. 5∣. The mPOA to PAG pathway primarily accounts for the mPOA’s role in regulating anxiety-like behaviors.

a, Representative images of GFP-labeled mPOA glutamatergic neurons (left) and their axons in PAG (right). Scale bar: 500 μm. b, Retrograde labeling of neurons in mPOA (right) by injection of AAVretro-Cre in PAG (left). Scale bar: 500 μm. c, Photoactivation of mPOA glutamatergic axons and recording from PAG neurons in the slice preparation. d, Left, voltage-clamp recording from a PAG neuron showing a light-evoked EPSC. Right, average amplitudes of light-evoked EPSCs in recorded PAG neurons (n = 10). Bar represents s.d. e, Photoactivation of mPOA glutamatergic axon terminals in PAG. f, Percentage time in the LED-on chamber in PPT. **P < 0.01, two-sided Mann–Whitney test, n = 5 animals for both GFP and ChR2 groups. Error bars, s.d. g, Average locomotion speed in an open arena. **P < 0.01, Mann–Whitney test, n = 6 animals for each group. Error bars, s.d. h, Percentage center time in OFT. **P < 0.01, Mann–Whitney test, n = 5 animals for each group. Error bars, s.d. i, Percentage open-arm time in EPM. **P < 0.01, two-sided Mann–Whitney test, n = 5 animals for each group. Error bars, s.d. j, Strategy to label PAG-projecting mPOA neurons with ChR2. Error bars, s.d. k-n, Similar to f-i, for photoactivation of PAG-projecting mPOA neurons. **P < 0.01, two-sided Mann–Whitney test, n = 5 or 6 animals for each group. Error bars, s.d. o, Strategy for photoactivation of mPOA-recipient PAG neurons. p-s, Similar to f-i, for photoactivation of mPOA-recipient PAG neurons. **P < 0.01, two-sided Mann–Whitney test, n = 5 or 6 animals for each group. Error bars, s.d. t, Chemogenetic silencing of mPOA→PAG glutamatergic axon terminals and experimental timeline. u, Left, movement tracing for an example mCherry control (gray) and a hM4Di-expressing (red) animal in OFT after exposure to electric shocks. Right, percentage center times in OFT for mCherry control animals in the control condition (n = 5), control animals after shocks (n = 5) and hM4Di animals after shocks (n = 5). P = 0.0007, two-sided one-way ANOVA; **P < 0.01, post hoc test. Error bars, s.d. v, Left, movement tracing for a mCherry control (gray) and a hM4Di-expressing animal (red) in the EPM test after exposure to shocks. Right, percentage open-arm times in EPM for mCherry control animals in the control condition (n = 6), control animals after shocks (n = 6) and hM4Di animals after shocks (n = 6). P < 0.0001, two-sided one-way ANOVA; **P < 0.01, post hoc test. Error bars, s.d. Images in a,b are representative of n=6,3 animals. (see Supplementary Table 1 for detailed statistics).

Fig. 6∣. The GABAergic neurons play an opposite role in regulating anxiety state and parental behavior.

a, Expressing ChR2 in mPOA GABAergic neurons. b, Left, representative movement tracing of a GFP control (gray) and a ChR2 (blue) animal in PPT. Right, percentage times in the LED-on chamber for GFP (n = 11 males and 11 females) and ChR2 (n = 8 males and 8 females) mice. **P < 0.01, one-way repeated-measures ANOVA. Error bars, s.d. c, Left, movement tracing for a ChR2 animal in OFT within a LED-off (gray) and a LED-on (blue) block. Right, percentage center times for GFP control (black) and ChR2-expressing (blue) animals in LED-off and LED-on blocks. **P < 0.01, two-way repeated-measures ANOVA test, n = 7 animals (4 males) for each group. Error bars, s.d. d, Percentage open-arm times for ChR2 animals (n = 7; 4 males) in LED-off and LED-on conditions. **P < 0.01, two-sided Mann–Whitney test. Error bars, s.d. e, Percentage of trials with the virgin female exhibiting pup retrieval in LED-on and LED-off conditions. **P < 0.01, two-sided Fisher’s exact test with Bonferroni correction, n = 7 ChR2 animals. f, Duration of pup grooming by the virgin female in LED-off and LED-on conditions. **P < 0.01, two-sided Mann–Whitney test with Bonferroni correction, n = 9 ChR2 animals. Error bars, s.d. g, Percentage of trials with the virgin males exhibiting pup attacks in in LED-off and LED-on conditions. **P < 0.01, two-sided Fisher’s exact test with Bonferroni correction, n = 7 ChR2 animals. h, Recording from mPOA Vgat− neurons while stimulating Vgat+ neurons in the slice preparation. i, Top, traces of voltage-clamp recording from a mPOA glutamatergic neuron under two holding potentials. Blue vertical line indicates the onset of light stimulation. Bottom, average amplitudes of EPSCs and IPSCs in 15 mPOA glutamatergic neurons. Error bars, s.d. j, Recording from PAG neurons and photostimulation of ChR2-expressing GABAergic mPOA axons. k, Top, a light-evoked IPSC recorded in a PAG neuron. Bottom, average amplitudes of EPSCs and IPSCs from 9 PAG neurons. Error bars, s.d. l, Top, stimulation of GABAergic mPOA axons in PAG. Bottom, movement tracing of an example animal in PPT. m, Right, percentage times in the LED-on chamber for GFP and ChR2 animals. **P < 0.01, two-sided two-way repeated-measures ANOVA test, n = 8 animals for each group. Error bars, s.d. n, Percentage center times in OFT for GFP and ChR2 animals. **P < 0.01, two-sided Mann–Whitney test, n = 6 animals (3 males) for each group. Error bars, s.d. o, Percentage open-arm times in EPM for GFP and ChR2 animals. **P < 0.01, two-sided Mann–Whitney test, n = 7 animals (4 males) for each group. Error bars, s.d. p, Percentage of trials with the virgin female exhibiting pup retrieval in LED-off (gray) and LED-on (blue) conditions. **P < 0.01, two-sided Fisher’s exact test with Bonferroni correction, n = 8 animals for both GFP and ChR2 groups. q, Percentage of trials with the virgin male exhibiting pup attacks in LED-off and LED-on conditions. **P < 0.01, two-sided Fisher’s exact test with Bonferroni correction, n = 8 animals for both GFP and ChR2 groups. (see Supplementary Table 1 for detailed statistics).

Fig. 7∣. Monosynaptic inputs to mPOA glutamatergic neurons.

a, Strategy for cell-type specific tracing of monosynaptic inputs using pseudotyped rabies. b, Example images of retrogradely labeled neurons in different brain regions. Scale bar: 500 μm. LSv, lateral septum ventral; LSr, rostral lateral septum; BST, bed nucleus of the stria terminalis; MS, medial septum; NAc, nucleus accumbens; aco, anterior commissure; PVT, periventricular nucleus of the thalamus; PT, parataenial nucleus; BLA, basolateral amygdala; CEA, central amygdala; MEA, medial amygdala; AHN, anterior hypothalamic nucleus; VMH, ventromedial hypothalamic nucleus; PVN, paraventricular hypothalamic nucleus; PVi, periventricular hypothalamic nucleus; SuM, supramammillary nucleus; MM, mammillary nucleus; CS, superior central nucleus raphe. c, Quantification of numbers of retrogradely labeled cells in different regions in contralateral and ipsilateral sides of the injected mPOA (n = 3 animals; bar = s.d.). PCG, pontine central gray; GRN, gigantocellular reticular nucleus; LPO, lateral preoptic area. Images in b are representative of n=3 animals.