Neural circuitry for maternal oxytocin release induced by infant cries

Abstract

Oxytocin is a neuropeptide that is important for maternal physiology and childcare, including parturition and milk ejection during nursing^1-6. Suckling triggers the release of oxytocin, but other sensory cues-specifically, infant cries-can increase the levels of oxytocin in new human mothers⁷, which indicates that cries can activate hypothalamic oxytocin neurons. Here we describe a neural circuit that routes auditory information about infant vocalizations to mouse oxytocin neurons. We performed in vivo electrophysiological recordings and photometry from identified oxytocin neurons in awake maternal mice that were presented with pup calls. We found that oxytocin neurons responded to pup vocalizations, but not to pure tones, through input from the posterior intralaminar thalamus, and that repetitive thalamic stimulation induced lasting disinhibition of oxytocin neurons. This circuit gates central oxytocin release and maternal behaviour in response to calls, providing a mechanism for the integration of sensory cues from the offspring in maternal endocrine networks to ensure modulation of brain state for efficient parenting.

Extended Data Fig. 1 |. Identification of ChR2⁺ (OT⁺) and ChR2⁻ (OT⁻) neurons.

a, Sample traces and raster plots of cell-attached recordings from one ChR2⁺ (OT⁺, left) and one ChR2⁻ (OT⁻, right) neuron showing reliable activation of ChR2⁺ (OT⁺) neurons in response to 50 ms pulses of blue light at 5 Hz (20–100% laser power). b, Increase in the number of spikes (left) of ChR2⁺ (OT⁺; pink circles; n = 7 neurons, N = 5 dams, p = 0.002, one-way ANOVA) and spike probability (right; p = 0.009) in response to 50 ms light pulse steps. Number of spikes of ChR2⁻ (OT⁻; black circles; n = 13, N = 3, p = 0.11) neurons, as well as spike probability (p = 0.13) was not modulated. c, Sample traces and raster plots of cell-attached recordings of one ChR2⁺ (OT⁺, left) and one ChR2⁻ (OT⁻, right). ChR2⁺ (OT⁺) neuron was reliably activated in response to 200-ms pulses of blue light (0–100% laser power). d, Increase in the number of spikes (left) of ChR2⁺ (OT⁺; pink circles; n = 9 neurons, N = 6 dams, p = 0.002, one-way ANOVA) and spike probability (right; p = 0.01) in response to 200 ms light pulse steps. Number of spikes of ChR2⁻ (OT⁻; black circles; n = 13, N = 3, p = 0.49) neurons, as well as spike probability (p = 0.62) was not modulated. e, Increase in firing rate of ChR2⁺ (OT⁺; n = 10 neurons, N = 5 dams, p = 0.0001, one-way ANOVA) but not of ChR2⁻ (OT⁻; n = 13, N = 3, p = 0.68) neurons in response to 200 ms light pulse of 100% laser power (‘Opto’) compared to their baseline firing rate immediately preceding (‘Pre’) and immediately after (‘Post’) the light pulse. f, Change of firing rate of ChR2⁺ (OT⁺; n = 10 neurons, N = 5 dams, p = 0.002, Wilcoxon matched-pairs signed-rank two-tailed test) but not of ChR2⁻ (OT⁻; n = 13, N = 3, p = 0.91) neurons in response to 200-ms light pulse of 100% laser power (‘Opto’) compared to baseline firing immediately preceding the light pulse (‘Pre’). g, Box plots (showing the median (line), second to third quartiles (box), minimum to maximum (whiskers) of latency to first spike was significantly shorter in ChR2⁺ (OT⁺; n = 14 neurons, N = 6 dams, p < 0.0001, Mann–Whitney two-tailed test) compared to ChR2⁻ (OT⁻; n = 8, N = 2) neurons in response to 200-ms light pulse of 100% laser power. ‘First spike’ in ChR2⁻ (OT⁻) cells was not light-evoked but occurred spontaneously. Data reported as mean ± s.e.m. *P < 0.05. **P < 0.01.

Extended Data Fig. 2 |. In vivo responses to individual pup calls.

a, Location and firing rate of cell-attached (n = 19 neurons, N = 8 dams) and whole-cell (n = 1) recordings of ChR2⁺ OT⁺ neurons and ChR2⁻ neurons (OT⁻, cell-attached: n = 16 neurons, N = 8 dams; whole-cell: n = 2, N = 2). b, Sample traces of whole-cell recordings of one ChR2⁺ (OT⁺; upper trace) and one ChR2⁻ (OT⁻; lower trace) neuron during call playback (‘Calls’). Pink bars, individual pup calls. c, Left, Cell-attached recording of one ChR2⁺ (OT⁺) neuron before pup-call onset (1, ‘Pre’), and around 2 min after call onset (2, ‘Post’). Right, Peristimulus time histogram. Bins: 20 s. d, Timeline of ChR2+ (OT⁺; n = 12 neurons) and ChR2⁻ (OT⁻; n = 11) responses. Period before (1, ‘Pre’) and after (2, ‘Post’) pup-call onset. e,f, Sample traces of whole-cell recordings of one ChR2⁺ (OT⁺; e) and one ChR2⁻ (OT⁻; f) neuron showing baseline spiking activity preceding onset of pup calls (1, ‘Pre’, upper trace), activity during playback of a set of pup calls (‘Calls’, middle trace) and activity after pup calls playback (2, ‘Post’, lower trace). Note increased firing rate for ChR2⁺ (OT⁺) neuron but not ChR2⁻ (OT⁻) neuron. g, ChR2⁺ (OT⁺) did not respond to individual pup calls within a set. ‘Pre’, average spiking rate of cell-attached recordings during all baseline periods immediately preceding each call within the set. ‘Call’, average spiking rate during pup-call stimulus for each call within the set. Neither ChR2⁺ (OT⁺; n = 9 neurons, N = 6 dams, p = 0.38, Wilcoxon matched-pairs signed-rank two-tailed test), nor ChR2⁻ (OT⁻; n = 9, N = 5, p = 0.30) neurons increased their firing rate during individual calls (‘Call’) compared to baseline (‘Pre’). h–j, ChR2⁺ (OT⁺) did not respond to presentation of individual pup calls on a trial-by-trial basis. h, Sample traces of cell-attached recordings of one ChR2⁺ (OT⁺), one ChR2⁻ (OT⁻) and one unidentified PVN neuron, as well as whole-cell recording of a PVN neuron during trial-by-trial individual pup-call presentation. i, No increase in the firing rates of either ChR2⁺ (OT⁺; n = 8 neurons, N = 2 dams, p = 0.38, Wilcoxon matched-pairs signed-rank two-tailed test), ChR2⁻ (OT⁻; n = 7, N = 3, p = 0.08), or unidentified PVN (cell-attached: n = 26, N = 13; whole-cell: n = 6, N = 4, p = 0.48) neurons during individual pup calls (‘Call’) compared to baseline (‘Pre’) on a trial-by-trial basis. j, No difference in the z-scores of spiking responses of cell-attached and whole-cell recordings during individual pup calls in ChR2⁺ (OT⁺; n = 8, N = 2, p = 0.30, one-way ANOVA), ChR2⁻ (OT⁻; n = 7, N = 4), and unidentified PVN (cell-attached: n = 21, N = 11; whole-cell: n = 6, N = 4) neurons. Data reported as mean ± s.e.m. *P < 0.05.

Extended Data Fig. 3 |. In vivo responses to auditory stimuli.

a–h, Z-scores of fluorescence activity during fibre photometry recordings of dams and virgins in response to auditory stimuli. Pup calls trigger sustained increase in the activity of oxytocin neurons in dams (a; N = 7; Pre vs Post: p = 0.018; Wilcoxon matched-pairs signed-rank one-tailed test) but not virgins (b; N = 4; p = 0.50; Wilcoxon matched-pairs signed-rank one-tailed test). No responses to adult calls (c; N = 4 dams; p = 0.5966; and d; N = 3 virgins; p = 0.09; Wilcoxon matched-pairs signed-rank one-tailed test) or ultrasound pure tones (e; N = 3 dams; p = 0.38; and f; N = 3 virgins; p = 0.47; Wilcoxon matched-pairs signed-rank one-tailed test). No response to pure tones (g; N = 3; p = 0.11; Wilcoxon matched-pairs signed-rank one-tailed test) or FM sweeps (h; N = 5; p = 0.20; Wilcoxon matched-pairs signed-rank one-tailed test) in dams. Insets represent fluorescence activity during a single trial from two different mice per condition. i, Responses to pup calls across dams and trials (pup calls, dam: n = 28 trials; adult calls, dam: n = 16; FM sweeps, dam: n = 19; tones, dam: n = 12; ultrasound, dam: n = 12; pup calls, virgin: n = 16). j, Average z-score responses to auditory stimuli for individual trials. k, PVN neurons did not respond to individual pure tones. Left, example cell-attached recording of one PVN neuron in response to 23 kHz tone presentation and tuning profile of pure-tone frequency responses in this cell. Right, average tuning profile of pure-tone frequency responses in PVN cells (n = 15 neurons, N = 9 dams). Data reported as median ± 95% CI ( j) or as median or mean ± s.e.m. (k). *P < 0.05.

Extended Data Fig. 4 |. Auditory responses in the PIL.

a, Left, experimental set-up showing in vivo multiunit recordings via tungsten electrode in the PIL of awake head-fixed wild-type dams while playing pup calls from an ultrasound speaker. Right, validation of PIL recording site by coating tungsten electrode tip with DiI. Scale, 500 μm. MGB, medial geniculate body of the thalamus; PIL, posterior intralaminar nucleus of the thalamus; SN, substantia nigra. b–g, In vivo activation of PIL during playback of pup calls and pure tones. Sample trace of stimulus-evoked PIL multiunit spiking activity during individual pup call (b) and 23 kHz tone (c) presentation. Note the increase in PIL activity during the entire duration of the pup call (>1 s) compared to transient activation during pure tones. Pup calls increased the firing frequency of PIL neurons (d; N = 6 dams, p = 0.03, Wilcoxon matched-pairs signed-rank two-tailed test) which corresponded to a significant increase from baseline values (e; N = 6 dams; p = 0.02, one-sample two-tailed Student’s t-test). f, There was no difference in the frequency of multiunit spiking during pup calls and pure tones playback (N = 6 dams; p = 0.33, Mann–Whitney two-tailed test). g, Tuning profile of pure-tone frequency responses in PIL (N = 5 dams). Data reported as mean ± s.e.m. *P < 0.05; ns, not significant.

Extended Data Fig. 5 |. Auditory projections to the PVN.

a, Schematic showing injection of AAV1-hSyn-hChR2(H134R)-EYFP in the PIL of Oxytocin:Cre × Ai9 dams prior to whole-cell recordings from oxytocin neurons (tdTomato⁺) in PVN brain slices. PIL, posterior intralaminar nucleus of the thalamus; PVN, paraventricular nucleus of the hypothalamus. b,c, PVN oxytocin neurons receive mainly glutamatergic input from the PIL. Percentage of optogenetically evoked excitatory (oEPSCs) and inhibitory (oIPSCs) currents in oxytocin neurons triggered by optogenetic stimulation of PIL axons (b) and characterization of oIPSCs (c; n = 12 neurons). d,e, PIL inputs to oxytocin cells are monosynaptic. d, Example traces and summary graph showing oEPSCs in the presence of TTX and 4-AP and their inhibition by DNQX (n = 8 neurons, p = 0.016, Wilcoxon matched-pairs signed-rank test). e, Example traces and summary graph showing oIPSCs in the presence of TTX and 4-AP; DNQX had no effect on oIPSCs amplitude (n = 5 neurons, p = 0.63, Wilcoxon). f–h, Parvocellular PVN oxytocin neurons are the main target of input from the PIL. Magnocellular (MagnOT) and parvocellular (ParvOT) oxytocin neurons were characterized by their signature spiking patterns in current-clamp mode (f; left). 5/16 MagnOT and 11/16 ParvOT cells received inputs from the PIL (f; right). g,h, Characterization of oEPSCs in MagnOT and ParvOT neurons triggered by optogenetic stimulation of PIL axons. g, Example traces from one magnocellular (left) and one parvocellular (right) oxytocin cell. There was no difference in oEPSCs amplitude (h, left; MagnOT: n = 5 neurons, ParvOT: n = 11; p = 0.51, Mann–Whitney two-tailed test) but the latency of oEPSCs in ParvOT cells was longer (h, right; MagnOT: n = 5 neurons, ParvOT: n = 11; p = 0.0275, Mann–Whitney two-tailed test). i, The PVN does not receive input from IC. Left, injection of AAV1-hSyn-hChR2(H134R)-EYFP in IC of wild-type dams. Right, no EYFP staining was found in PVN, suggesting that IC does not project to PVN. Scale, 200 μm. N = 3. IC, inferior colliculus. j, The PVN does not receive input from AuCx. Left, injection of AAV1-hSyn-hChR2(H134R)-EYFP in AuCx of wild-type dams. Right, no EYFP staining was found in PVN, suggesting that AuCx does not project to PVN. Scale, 200 μm. N = 2. AuCx, auditory cortex. Data reported as mean ± s.e.m. *P < 0.05.

Extended Data Fig. 6 |. PIL neurons do not exhibit sustained increases in firing after pup calls in vivo.

a, Experimental set-up showing in vivo cell-attached recordings in PIL of awake wild-type dams while playing pup calls from an ultrasound speaker. PIL, posterior intralaminar nucleus of the thalamus. b, Location (depth from pia) and firing rate of PIL neurons (n = 9 neurons; N = 4 dams). c–e, PIL neurons did not modulate their firing rate following playback of a set of pup calls (15 pup calls, 1 s gap in between calls). c, Sample traces from a cell-attached recording of one PIL neuron showing its baseline firing rate immediately preceding (1, ‘Pre’) and at 90 s after the onset of pup calls playback (2, ‘Post’). Firing rates during baseline and after pup calls were calculated over 1–2 min. d, Timeline of responses of PIL neurons. e, PIL neurons (n = 9) did not exhibit persistent increases in baseline firing following pup calls, as calculated between 80–160 s after onset of pup-call playback (e, n = 9 neurons, N = 4 dams, p = 0.77, one-sample two-tailed Student’s t-test). Data reported as mean ± s.e.m.; ns, not significant.

Extended Data Fig. 7 |. PIL inputs to the PVN do not induce postsynaptic spiking or affect the excitability of oxytocin neurons.

a, Schematic showing injection of AAV1-hSyn-hChR2(H134R)-EYFP in PIL of Oxytocin:Cre × Ai9 dams before whole-cell recordings from oxytocin neurons (tdTomato⁺) in PVN brain slices. PIL, posterior intralaminar nucleus of the thalamus; PVN, paraventricular nucleus of the hypothalamus. b, Whole-cell recordings from tdTomato+ oxytocin neurons in PVN slices, optogenetic stimulation of PIL axons and placement of the extracellular stimulation electrode. c,d, Optogenetic stimulation of PIL axons in PVN does not induce postsynaptic spiking in oxytocin neurons. c, Single pulse of optogenetic stimulation triggered postsynaptic potentials but did not induce spiking in oxytocin cells (n = 8 neurons). d, Repeated optogenetic stimulation of PIL axons (‘PIL opto’) did not trigger depolarization of oxytocin cells (n = 16 neurons; p = 0.10, Wilcoxon matched-pairs signed-rank two-tailed test) and did not induce postsynaptic spiking (n = 16 neurons; p = 0.26, Wilcoxon matched-pairs signed-rank two-tailed test). e,f, No change in the number of spikes in oxytocin neurons in response to 20 pA steps of intracellular current injection before (‘Pre’) or after (‘Post’) PIL opto. Sample traces (e) and summary (f; n = 9 neurons, p > 0.44, Wilcoxon matched-pairs signed-rank two-tailed test). g–i, No change in the intrinsic properties of oxytocin neurons after PIL opto, in terms of rheobase (g; n = 9 neurons; p = 0.38, Wilcoxon matched-pairs signed-rank two-tailed test), resting membrane potential (h; n = 6 neurons; p = 0.16, Wilcoxon matched-pairs signed-rank two-tailed test), or input resistance (i; n = 12 neurons; p = 0.06, Wilcoxon matched-pairs signed-rank two-tailed test). Data reported as mean ± s.e.m.

Extended Data Fig. 8 |. Prolonged but not brief optogenetic stimulation of PIL axons triggers iLTD.

a, Whole-cell recordings from tdTomato⁺ oxytocin neurons in PVN slices, optogenetic stimulation of PIL axons and placement of the extracellular stimulation electrode. b, Whole-cell voltage-clamp recordings showing that the amplitude of oEPSCs triggered by single pulse of optogenetic stimulation was not modified following repeated optogenetic stimulation of PIL axons (‘PIL opto’; n = 6 neurons, p = 0.13, one-sample two-tailed Student’s t-test). c, IPSC/EPSC ratio decreased following PIL opto (n = 6 neurons; p = 0.0018, one-sample two-tailed Student’s t-test). d–f, Repeated but brief optogenetic stimulation of PIL axons (‘PIL opto short’) did not induce iLTD: example cell (d; p = 0.74, Mann–Whitney two-tailed test) and summary (e; n = 6 neurons, p = 0.99, one-sample two-tailed Student’s t-test). f, No change in IPSC/EPSC ratio following PIL opto short (n = 6 neurons; p = 0.49, one-sample two-tailed Student’s t-test). g–i, In vivo exposed to pup calls playback occlude iLTD. g,h, Schematic of experimental protocol (g): example cell (g; p = 0.94, Mann–Whitney two-tailed test) and summary (h; n = 4 neurons, p = 0.98, one-sample two-tailed Student’s t-test). i, No change in IPSC/EPSC ratio following PIL opto in slices of dams exposed to pup calls in vivo (n = 4 neurons; p = 0.76, one-sample two-tailed Student’s t-test). Data reported as mean ± s.d. **P < 0.01; ns, not significant.

Extended Data Fig. 9 |. iLTD in oxytocin neurons relies on postsynaptic NMDARs and dynamin signalling.

a, Whole-cell voltage-clamp recordings showing intact iLTD after repeated optogenetic stimulation of PIL terminals in PVN (‘PIL opto’) in presence of bath-applied type-III mGluR antagonist MAP4 (250 μM), for example neuron (top; p < 0.0001, Mann–Whitney two-tailed test) and summary (middle; n = 5 neurons, p = 0.02, one-sample two-tailed Student’s t-test). Bottom, IPSC/EPSC ratio in the presence of MAP4. b, iLTD in oxytocin neurons is NMDAR-dependent. Whole-cell voltage-clamp recordings showing no plasticity after PIL opto in presence of bath-applied AP5 (50 μM), for example neuron (top; p = 0.62, Mann–Whitney two-tailed test) and summary (middle; n = 8 neurons, p = 0.44, one-sample two-tailed Student’s t-test). Bottom, unchanged IPSC/EPSC ratio in the presence of AP5. c, iLTD in oxytocin neurons is dependent on postsynaptic NMDARs. Whole-cell voltage-clamp recordings showing no plasticity after PIL opto when i-MK801 (1 mM) was applied in the recording pipette, for example neuron (top; p = 0.11, Mann–Whitney two-tailed test) and summary (middle; n = 6 neurons, p = 0.18, one-sample two-tailed Student’s t-test). Bottom, unchanged IPSC/EPSC ratio in the presence of i-MK801. d, Whole-cell voltage-clamp recordings showing intact iLTD after PIL opto in presence of bath-applied OXTR antagonist OTA (1 μM), for example neuron (top; p < 0.0001, Mann–Whitney two-tailed test) and summary (middle; n = 8 neurons, p = 0.0008, one-sample two-tailed Student’s t-test). Bottom, IPSC/EPSC ratio in the presence of OTA. e,f, iLTD in oxytocin neurons is dependent on dynamin signalling. e, Whole-cell voltage-clamp recordings showing no plasticity after PIL opto when i-Dynamin inhibitor (1.5 mM) was applied in the recording pipette, for example neuron (top; p = 0.30, Mann–Whitney two-tailed test) and summary (middle; n = 7 neurons, p = 0.63, one-sample two-tailed Student’s t-test). Bottom, unchanged IPSC/EPSC ratio in the presence of i-Dynamin inhibitor. f, Whole-cell voltage-clamp recordings showing intact iLTD in oxytocin cells in the presence of a scrambled dynamin inhibitor in the recording pipette, for example neuron (top; p < 0.0001, Mann–Whitney two-tailed test) and summary (middle; n = 8 neurons, p = 0.0096, one-sample two-tailed Student’s t-test). Bottom, IPSC/EPSC ratio in the presence of a scrambled dynamin inhibitor. Data reported as mean ± s.d. *P < 0.05, **P < 0.01; ns, not significant.

Extended Data Fig. 10 |. Inhibiting oxytocin signalling in the VTA impairs pup-retrieval behaviour.

a, Pup-retrieval protocol. b,c, Wild-type dams infused with the OXTR antagonist OTA (0.5 mg ml⁻¹; b) retrieved less pups compared to saline controls (c; N = 4 dams, P at least <0.02, two-tailed Fisher’s test). Data reported as mean ± s.e.m. *P < 0.05.

Fig. 1 |. Delayed persistent activation of dam oxytocin neurons by pup vocalizations.

a, In vivo cell-attached and whole-cell channelrhodopsin 2-assisted recordings and playback of pup calls. b, Oxytocin cells expressed EYFP–ChR2 (white arrowheads). Scale bars, 100 μm (main); 10 μm (inset). 3V, third ventricle. c, In vivo optogenetic identification of ChR2⁺ (OT⁺) neurons with reliable spiking to blue-light pulses (‘Opto’). d,e, Cell-attached recordings of ChR2⁺ (OT⁺) (d) and ChR2⁻ (OT⁻) (e) neurons before the onset of pup calls (1, ‘Pre’) and 80 s after call onset (2, ‘Post’). f, Peristimulus time histograms for ChR2⁺ (OT⁺) and ChR2⁻ (OT⁻) neurons. Bins, 10 s. g, Firing rate of ChR2⁺ (OT⁺) (n = 11 neurons, N = 6 dams; P = 0.0014, Friedman test) and ChR2⁻ (OT⁻) (n = 11, N = 5; P = 0.07) neurons. h, Change in the firing rate of ChR2⁺ (OT⁺) (n = 12; P = 0.01, one-sample two-tailed Student’s t-test) but not ChR2⁻ (OT⁻) (n = 11; P = 0.12) neurons. Cell-attached (circles) and whole-cell (triangles) recordings. i, Injection of AAVDJ-CAG-FLEx-GCaMP6s into the PVN of Oxytocin:Cre dams and fibre implantation. Scale, 200 μm. j, Fibre photometry in awake head-fixed dams and playback of pup calls through an ultrasonic speaker. k–n, Oxytocin neurons responded to pup calls but not pure tones. k,l, Example timelines of responses to five sets of pup calls (k) or ultrasound pure tones (l). m,n, Box plots (showing the median (line), second to third quartiles (box) and minimum to maximum values (whiskers)) of z-scores of fluorescence activity preceding the onset (‘Pre’) of pup calls (m) or tones (n), during auditory stimulus (‘Calls’ or ‘Tones’) and after (‘Post’) (m, N = 7 dams; Pre versus Calls: P = 0.005; Pre versus Post: P = 0.018; n, N = 3 dams, Pre versus Calls: P = 0.21; Pre versus Post: P = 0.34; Wilcoxon matched-pairs signed-rank one-tailed test and correction for multiple comparisons). Data reported as mean ± s.e.m. or median ± minimum to maximum (m,n). *P < 0.05, **P < 0.01.

Fig. 2 |. PVN oxytocin neurons receive projections from the auditory thalamus.

a–e, Rabies-virus tracing. a, Injection of helper virus AAV2-EF1a-FLEX-TVA-GFP followed by Cre-inducible, retrograde pseudotyped monosynaptic rabies virus SAD∆G-mCherry. b, Starter oxytocin neurons in the PVN expressing avian receptor protein (TVA)–GFP and glycoprotein-deleted rabies virus (∆G-RV)–mCherry. Scale bar, 100 μm. c–e, Retrograde rabies infection and mCherry staining were absent in the MGB (c), AuCx (d) and IC (e). Dense rabies-infected field and robust expression of mCherry was found in the PIL (c; N = 4 dams). Scale bars, 500 μm. SN, substantia nigra. f,g, The PIL receives reliable input from the IC. f, Left, injection of AAV1-hSyn-hChR2(H134R)-EYFP into the IC of wild-type dams and whole-cell voltage-clamp recordings in PIL slices. Right, expression of ChR2–EYFP in the IC and IC projections in the MGB and the PIL. Scale bars, 200 μm. g, Characterization of oEPSCs in PIL neurons triggered by optogenetic stimulation of IC axons (10 out of 12 connected neurons, N = 5). h,i, The PIL receives sparser input from the AuCx. h, Left, injection of AAV1-hSyn-hChR2(H134R)-EYFP into the AuCx of wild-type dams and whole-cell voltage-clamp recordings in PIL slices. Right, expression of ChR2–EYFP in the AuCx and dense AuCx projections in the MGB but sparser projections in the PIL. Scale bars, 500 μm (top) and 200 μm (bottom). i, oEPSCs in PIL neurons triggered by optogenetic stimulation of AuCx axons (4 out of 13 connected neurons, N = 3). j,k, PVN oxytocin neurons receive reliable input from the PIL. j, Left, injection of AAV1-hSyn-hChR2(H134R)-EYFP into the PIL of Oxytocin:Cre × Ai9 dams and whole-cell voltage-clamp recordings from oxytocin neurons (tdTomato⁺). Right, expression of ChR2–EYFP in the PIL and PIL projections in the PVN. Scale bars, 200 μm (top) and 100 μm (bottom). k, oEPSCs in oxytocin neurons triggered by optogenetic stimulation of PIL axons (50 out of 67 connected, N = 21). Data are mean ± s.e.m.

Fig. 3 |. Activation of PIL–PVN inputs decreases inhibition in oxytocin neurons through postsynaptic NMDARs and the dynamin-dependent internalization of GABA_ARs.

a, Whole-cell recordings from tdTomato⁺ oxytocin neurons in PVN slices, optogenetic stimulation of PIL axons and extracellular stimulation. b,c, Short-term enhanced spiking probability of oxytocin neurons after PIL opto. Improved spiking after 5-Hz electrical stimulation 5 min after PIL opto (‘Post’) versus baseline (‘Pre’). b, Example. c, Summary (n = 9 neurons, pulses 1–5: P = 0.016, 0.023, 0.098, 0.094, 0.047, Wilcoxon matched-pairs signed-rank two-tailed test). d, Long-term enhanced spiking probability of oxytocin neurons after PIL opto after a single electrical stimulation pulse. Left, example. Middle, spiking probability (n = 6 neurons). Right, spiking significantly increased 20 min after PIL opto (n = 6; P = 0.03, Wilcoxon). e,f, PIL opto induced iLTD in oxytocin neurons. Ra, access resistance. e, No change in the magnitude of EPSCs after PIL opto. Left, example cell (P = 0.15, Mann–Whitney two-tailed test; scale bar, 20 ms and 50 pA). Right, summary (n = 7 neurons; P = 0.14, one-sample two-tailed Student’s t-test). f, Decreased magnitude of IPSCs after PIL opto. Left, example cell (P < 0.0001, Mann–Whitney; scale bar, 20 ms and 100 pA). Right, summary (n = 10 neurons; P = 0.0003, one-sample Student’s t-test). g, NMDAR-mediated currents in oxytocin cells with a single optogenetic stimulation pulse of PIL axons (n = 7 neurons; P = 0.02, Wilcoxon). h, iLTD is blocked by AP5 (n = 8; P = 0.002, Mann–Whitney), i-MK801 (n = 6; P = 0.02) and i-Dynamin inhibitor (n = 7; P = 0.007), and occluded by the presentation of pup calls (‘Occlusion’; n = 4; P = 0.014), but not by MAP4 (n = 5; P = 0.37), OTA (n = 8; P = 0.08); PIL opto versus PIL opto short (P = 0.02), or PIL opto versus dynamin inhibitor control (P = 0.90). i, Possible mechanisms for postsynaptic NMDAR-dependent decreased inhibition. Data are mean ± s.e.m. (c,d (right), g,h) or mean ± s.d. (d (middle), e,f). *P < 0.05, **P < 0.01; NS, not significant.

Fig. 4 |. Chemogenetic inhibition of PIL projections to PVN impairs the detection of pup calls and pup-retrieval behaviour.

a, Injection of AAVrg-ENN.AAV.hSyn.Cre.WPRE.hGH into the PVN and AAV8-hSyn-DIO-hM4D(Gi)-mCherry or AAV8-hSyn-DIO-mCherry into the PIL. b, Expression of mCherry in PVN-projecting PIL neurons. Scale bar, 200 μm. c–e, Bath application of CNO (1 μM) reduced the firing rate in PIL slices from dams expressing hM4Di. c, Example whole-cell current-clamp recording from a PIL neuron. d, Summary (n = 6 neurons; P = 0.03, Wilcoxon matched-pairs signed-rank two-tailed test). e, Normalized spiking for same neurons as in d (P = 0.003, one-sample two-tailed Student’s t-test). f,g, Dams expressing hM4Di and injected with CNO had no preference for pup calls. f, Schematic of speaker approach protocol and number of calls played for approach (31 trials). g, Approach probability (43 trials; N = 4 dams; P = 0.002, two-tailed Fisher’s test; two-tailed binomial test; saline: P = 0.0054, CNO: P = 0.1173). h–m, Dams expressing hM4Di and injected with CNO had impaired pup-retrieval behaviour. h, Pup-retrieval protocols. i, Latency to approach a speaker playing pup calls (31 trials) versus a crying pup (158 trials; P < 0.0001, Mann–Whitney two-tailed test). j–l, One pup, ten trials test. j, Performance of all dams across trials. k, Dams expressing hM4Di and injected with CNO retrieved with a lower probability compared to control saline-injected dams at trials 6 to 10 (N = 11 dams; P = 0.0055, Fisher’s test); there was no difference in dams expressing mCherry (N = 6 dams; P > 0.9999, Fisher’s test). l, No difference in time to retrieval for trials 6 to 10 (hM4Di: N = 11 dams; P > 0.10 for all, Mann–Whitney; mCherry: N = 6 dams, P > 0.3939, Mann–Whitney). m, Five pups test. No difference in time to retrieval (hM4Di: N = 6 dams; P > 0.0556, Mann–Whitney; mCherry: N = 4 dams, P > 0.31 for all, Mann–Whitney). Data are mean ± s.e.m. or median ± 95% confidence intervals (CI) (g; l,m as marked). *P < 0.05, **P < 0.01.

Fig. 5 |. Pup calls trigger the release of oxytocin in the VTA through the PIL-to-PVN pathway.

a, Injection of AAV5-hSynapsin1-FLEx-axon-GCaMP6s into the PVN of Oxytocin:Cre dams and GCaMP6s⁺ axons in the VTA. Scale bar, 200 μm. N = 2 dams. b, Schematic of the OXTR-iTango2 genetic strategy. TEV-seq, TEV protease cleavage recognition sequence; TEV-C, C terminus of TEV protease; TEV-N, N terminus of TEV protease; v2-tail, C-terminal tail of the vasopressin receptor 2. c, Injection of OXTR-iTango2 viral constructs and fibre implantation in the VTA. d, Dams were exposed to pup calls paired with blue light (‘Blue light + pup calls’; ‘BL + PC’), pure tones paired with blue light (‘Blue light + tones’; ‘BL + T’) or blue light alone (‘Blue light only’; ‘BL’). e, Images of VTA neurons expressing OXTR-iTango2 constructs for the BL + PC group (left; scale bar, 200 μm) and magnified images from the area marked by a square (right; scale bar, 50 μm). f–h, Distribution patterns and percentage of OXTR-iTango2-labelled neurons in the VTA of dams (f, BL + PC; g, BL + T; h, BL). Scatter plots of individual cells (black dashed lines indicate threshold): cells with no viral infection (tdTomato⁻EGFP⁻; black, ‘B’); cells with only a TRE-reporter signal (tdTomato⁻EGFP⁺; green, ‘G’); cells with only a β-Arrestin–tdTomato signal (tdTomato⁺EGFP⁻; red, ‘R’); and cells with both a red and a green signal (tdTomato⁺EGFP⁺; yellow, ‘Y’). Average red (R) and green (G) fluorescent signals were calculated for each region of interest (ROI) and were divided by the mean background value (R0 and G0 for red and green channels, respectively) outside the ROIs for normalization. Scale bars, 50 μm. i, Injection of AAVrg-ENN.AAV. hSyn.Cre.WPRE.hGH into the PVN, AAV8-hSyn-DIO-hM4D(Gi)-mCherry into the PIL and OXTR-iTango2 viral constructs into the VTA together with fibre implantation in the VTA. Dams were injected with CNO and exposed to pup calls paired with blue light (BL + PC + CNO). j, Distribution patterns and percentage of OXTR-iTango2-labelled neurons in the VTA of dams expressing hM4D(Gi) in PVN-projecting PIL cells and injected with CNO. Scale bar, 50 μm. k, Density quantification of OXTR-iTango2-labelled cells. There was a significant number of yellow (tdTomato⁺EGFP⁺) cells in the BL + PC group (N = 5) compared with the BL + T group (N = 4; P < 0.0001, one-way ANOVA; P < 0.0001, post hoc Bonferroni correction), BL group(N = 5; P = 0.0001), or BL + PC + CNO group(N = 4; P = 0.0008). There was no difference in the number of yellow cells between BL + T, BL and BL + PC + CNO. Data are mean ± s.e.m. **P < 0.01; NS, not significant.