Functional circuit architecture underlying parental behaviour

Abstract

Parenting is essential for the survival and wellbeing of mammalian offspring. However, we lack a circuit-level understanding of how distinct components of this behaviour are coordinated. Here we investigate how galanin-expressing neurons in the medial preoptic area (MPOA^Gal) of the hypothalamus coordinate motor, motivational, hormonal and social aspects of parenting in mice. These neurons integrate inputs from a large number of brain areas and the activation of these inputs depends on the animal's sex and reproductive state. Subsets of MPOA^Gal neurons form discrete pools that are defined by their projection sites. While the MPOA^Gal population is active during all episodes of parental behaviour, individual pools are tuned to characteristic aspects of parenting. Optogenetic manipulation of MPOA^Gal projections mirrors this specificity, affecting discrete parenting components. This functional organization, reminiscent of the control of motor sequences by pools of spinal cord neurons, provides a new model for how discrete elements of a social behaviour are generated at the circuit level.

Extended Data Figure 1. Putative functional roles of brain areas providing monosynaptic inputs into MPOA^Gal neurons

a, Comparison between MPOA^Gal input fractions in virgin males (n = 3) and females (n = 3) after rabies tracing (see Fig. 2d). Sexually dimorphic inputs are highlighted. Two-tailed t-tests, SON: **P = 0.0041, AHPM: ***P = 0.0007, MS: *P = 0.0133. b, Comparison between MPOA^Gal input fractions after rabies tracing was initiated from the right (n = 3) or left (n = 3) hemisphere in virgin females. No significant differences were found (P > 0.05; two-tailed paired t-test). c, Comparison between rabies-injected (ipsilateral, ipsi) and non-injected (contralateral, contra) MPOA of a mother after parental behaviour. Activated (c-Fos⁺) rabies⁺ neurons are shown (upper, arrowheads). c-Fos⁺ neuron numbers are not significantly different between hemispheres (lower, P = 0.43, 95% CI [−4.176, 1.843]; two-tailed paired t-test; n = 6). d, MPOA^Gal neurons receive monosynaptic inputs from magnocellular SON^AVP neurons (mothers, 72.7 ± 9.3% overlap, n = 3; virgin females, 77.4 ± 4.3%, n = 3; fathers, 83.3 ± 3.3%, n = 3) but rarely from SON^OXT neurons (mothers, 4.6 ± 4.2% overlap, n = 2; virgin females, 4.5 ± 1.0%, n = 2; fathers, 2.8 ± 1.8%, n = 2). Scale bars, c, 100 μm; d, 50 μm. Data are mean ± s.e.m.

Extended Data Figure 2. MPOA^Gal projections in males and downstream connectivity

a, Synaptophysin-GFP (Syn-GFP) labelling of presynaptic sites in MPOA^Gal projections. b, Representative MPOA^Gal projections from a virgin male, identified by tdTomato fluorescence. c, Representative MPOA^Gal projections, identified by tdTomato fluorescence, after viral injection into the left MPOA. d, c-Fos⁺ fractions of virally labelled MPOA^Gal projections in fathers (n = 6, 3, 4, 3, 3, respectively, from top). Red line depicts the population average (see ref. 3). Data are mean ± s.e.m. e, Trans-synaptic retrograde rabies tracing from AVPe^TH neurons. f, MPOA^Gal neurons presynaptic to AVPe^TH neurons in females (left, arrowheads, 21.4% Gal⁺, 47/220 neurons, n = 3) and males (right, 16.7% Gal⁺, 4/24 neurons, n = 2). g, Direct and indirect MPOA^Gal→PVN^OXT connectivity. Asterisk, AVPe^TH neurons form excitatory synapses with PVN^OXT in females. h, Conditional monosynaptic retrograde tracing initiated from PAG. i, j, Injection sites with mCherry⁺ starter neurons in PAG of Vgat-ires-Cre (i, left) or Vglut2-ires-Cre (j, left) mice. Presynaptic, rabies⁺/Gal⁺ neurons are detected in MPOA when tracing is initiated from PAG^Vgat (i, right, arrowheads), but not PAG^Vglut2 (j, right), neurons. Scale bars, a, 50 μm; b, 250 μm; c, left, 500 and inset, 250 μm; f, 50 μm; i, j, left, 200 μm and right, 250 μm; i, insert, 50 μm.

Extended Data Figure 3. MPOA^Gal projections correspond to largely non-overlapping neuronal subpopulations

a, Control injection of a 1:1 mixture of CTB-488 and CTB-647 into PAG results in highly overlapping neuron populations in the MPOA (quantification, see c). b, Strategy to determine collaterals between pairwise injected MPOA^Gal projections in Gal::Cre^+/−; loxP-Stop-loxP-tdTomato^+/− mice. An example with two double-labelled MPOA^Gal neurons is shown after injection of CTB-488 into PAG and CTB-647 into VTA (right, arrowheads). c, Quantification of data in a, b. Data are mean ± s.e.m. (n = 6, 6, 3, 3, 3, 3, 3, respectively, from top). d, Representative image from MPOA of Gal::Cre^+/−; loxP-Stop-loxP-tdTomato^+/− mouse after injection of CTB-647 into PAG. Note high overlap between Gal⁺ and CTB⁺ neurons. e, Frequency of Gal⁺ neurons in individual, CTB-labelled MPOA projections (n = 4, 6, 4, 3, 3, 3, respectively, from top). Red line depicts expected labelling frequency, based on proportion of Gal⁺ MPOA neurons (~20%, ref. 3). Data in c, e are mean ± s.e.m. f, Distribution of cell bodies corresponding to specific MPOA^Gal projections. Individual MPOA^Gal projection areas in Gal::Cre virgin females were injected with Cre-dependent CAV2-FLEx-ZsGreen (see Fig. 2h). Only labelling patterns on the ipsilateral, injected side are shown and only two projection-specific subpopulations per side are displayed for clarity. Distance from Bregma is shown in mm. Mouse brain images in this figure have been reproduced with permission from Elsevier. g, Zones occupied by MPOA^Gal cell bodies projecting to MeA, PAG, VTA and PVN in anterior (left), central (middle) and posterior (right) MPOA. Distance from Bregma is shown in mm. Scale bars, a, b, 50 μm; d, 250 μm and inset, 50 μm.

Extended Data Figure 4. MPOA^Gal projections barely collateralize

a, Strategy to detect brain-wide axon collaterals of specific MPOA^Gal projections. b, Dense labelling of MPOA^Gal neurons after injection of retrograde tracer CAV into PAG and reporter AAV into MPOA. c, Absence of MPOA^Gal labelling in negative control without injection of CAV. d–f, Only minor axon collaterals are detectable from MPOA^Gal neurons projecting to PAG (n = 2) (d), VTA (n = 3) (e) or MeA (n = 2) (f). Note MPOA→MeA fibre tract in BNST in (f). Signal was enhanced using anti-GFP immunostaining (Methods). Scale bars, b, c, 400 μm and insets, 100 μm; d–f, 150 μm.

Extended Data Figure 5. Negative controls for monosynaptic retrograde tracing

a, Absence of rabies⁺ background labelling in the MPOA of AAV- and rabies-injected C57BL/6 control mice (n = 2). b, Labelling of MPOA^Gal neurons after injection of CAV into PAG and starter AAVs into MPOA of Gal::Cre mice (261 ± 19 neurons, n = 4). c, Near-absence of labelling in AAV-only negative control (11 ± 2 neurons, n = 2). d, Background rabies⁺ neurons were present in the following brain areas of CAV-, AAV- and rabies-injected C57BL/6 control mice (n = 3): MPOA, BNST, AH, PVN and SON. These areas were therefore excluded from analysis (see Figure 2k, l and Methods). Scale bars, a–d, 400 μm and insets, 150 μm.

Extended Data Figure 6. Histology of photometry recording experiments and tuning of MPOA^Gal neurons in other behavioural contexts

a, Specific GCaMP6m expression in MPOA^Gal neurons (90.9 ± 4.3% overlap, n = 3, mothers). b–d, Implantation sites of optical fibres in the MPOA of Gal::Cre^+/−; loxP-Stop-loxP-tdTomato^+/− mother (b), virgin female (c) and father (d). e, Quantification of GCaMP⁺ neuron numbers in MPOA after AAV injection (‘Total’, n = 4) and after injection of HSV into individual projections (n = 5 each). Data for mothers are shown. Data are mean ± s.e.m. Two-tailed t-tests, Total vs. PAG, VTA, MeA: ***P < 0.001, PAG vs. MeA: **P = 0.0033. f–h, Expression of GCaMP6m in MPOA^Gal neurons after bilateral infection of axon terminals in PAG (f), VTA (g) or MeA (h) with Cre-dependent, GCaMP6m-expressing HSV. Insets show fibre implantation sites. i, j, Averaged recording traces from MPOA^Gal neuron activity during sniffing of accessible pups (i) or inaccessible pups enclosed in a wire mesh tea ball (j) in mothers (n = 4), virgin females (n = 3) and fathers (n = 5). k, l, Averaged recording traces from MPOA^Gal neuron activity during sniffing of female (k) or male (l) intruder in mothers (n = 4), virgin females (n = 3) and fathers (n = 5). Two-tailed t-tests, i: ***P < 0.0001, ***P < 0.0001, P = 0.0001, j: *P = 0.0380, k: *P = 0.0219, l: *P = 0.0272. m–q, Averaged recording traces from MPOA^Gal neurons projecting to PAG (left, n = 10), VTA (middle, n = 12) or MeA (right, n = 8) during episodes of maternal behaviour. All traces and bar plots are mean ± s.e.m. Scale bars, a, 50 μm; b–d, 400 μm; f–h, 1 mm and insets, 500 μm.

Extended Data Figure 7. Distribution of parental behaviours in mothers and virgin females

Distribution of parental behaviours during 10 min pup interaction assays in mothers (a, n = 23) and virgin females (b, n = 20). In a, individuals exhibiting high pup sniffing are indicated in blue across plots, and individuals exhibiting high pup grooming are indicated in orange. In b, individuals exhibiting high pup sniffing are indicated in green. Note that y axis ranges are identical between a and b. Lines depict mean.

Extended Data Figure 8. Behavioural specificity of MPOA^Gal projection stimulation

a, Channelrhodopsin-2 (ChR2) expression in MPOA^Gal neurons (97.7 ± 0.2% overlap, virgin female, n = 2). b–g, Effect of activating PAG- (b, c), VTA- (d, e) or MeA- (f, g) projections on time spent in nest in females and males (b, n = 13, 10; d, n = 9, 10; f, n = 10, 10) and number of pup-directed sniffing bouts (c, n = 13, 10; e, n = 9, 10; g, n = 10, 10). h–m, Effect of activating PAG- (h, i), VTA- (j, k) or MeA- (l, m) projections on locomotion velocity (h, n = 13, 10; j, n = 8, 10; l, n = 10, 10) and moved distance (i, k, m). n, q, s, Effect of inhibiting PAG- (n, n = 10), VTA- (q, n = 10) or MeA- (s, n = 11) projections on pup interactions. o, t, Effect of inhibiting PAG- (o, n = 10) or MeA- (t, n = 11) projections on number of barrier crosses. p, r, Effect of inhibiting PAG- (p, n = 10) or MeA- (r, n = 11) projections on chemoinvestigation of a male intruder. u–w, Effect of inhibiting PAG- (u), VTA- (v) or MeA- (w) projections on locomotion velocity and moved distance (n = 10, 10, 11, respectively). Two-tailed paired t-tests, c: *P = 0.0135, f: *P = 0.03, n: *P = 0.0413, q: *P = 0.0264. Scale bar in a, 50 μm.

Figure 1. MPOA^Gal inputs are activated during parental behaviour in a sex- and reproductive state-specific manner

a, Monosynaptic retrograde tracing from MPOA^Gal neurons. b, Input areas with rabies⁺ neurons. c, Overview of inputs into MPOA^Gal neurons. Hypothalamic input areas in bold. d, MPOA^Gal neurons receive monosynaptic inputs from magnocellular PVN^AVP (37.6 ± 4.1% overlap, n = 3) but rarely PVN^OXT (2.6 ± 0.6%, n = 3) neurons. e, Presynaptic neurons in AVPe are TH⁻ in males (1.9% TH⁺, n = 2) and females (1.8% TH⁺, n = 3). f, Presynaptic neurons in AHPM. g, Identification of activated MPOA^Gal inputs and example of c-Fos⁺ presynaptic neurons. h–j, Activated input fractions in mothers (h), virgin females (i) and fathers (j) (each n = 6, controls n = 6). Green boxes, parent-specific activation, blue boxes, father- and virgin female-specific activation. Two-tailed t-tests (corrected for multiple comparisons, Methods), h: ***P < 0.0001, **P = 0.0267, *P = 0.0196, i: ***P < 0.0001, j: ***P < 0.0001, **P = 0.0035, *P = 0.0104. Data in h–j are mean ± s.e.m. n = number of animals in all figures. Scale bars, b, left, 500 μm, inset, 250 μm; d–g, 50 μm. Abbreviations, see Extended Data Table 1.

Figure 2. Identification of parenting-activated MPOA^Gal projections and input-output logic of the MPOA^Gal circuit

a, Visualisation of MPOA^Gal projections. b, MPOA^Gal projections identified by tdTomato fluorescence. c, Relative synaptic density in MPOA^Gal projection targets (n = 4, Methods). Grey regions could not be quantified due to tissue autofluorescence. Hypothalamic target areas in bold. d, Monosynaptic retrograde tracing from PVN. e–g, MPOA^Gal neurons presynaptic to (e) PVN^AVP (female 15/364 Gal⁺, n = 3; male 46/180 Gal⁺, n = 3) to (f) PVN^OXT (female 26/71 Gal⁺, n = 3; male 7/51 Gal⁺, n = 3) and to (g) PVN^CRH neurons (female 19/72 Gal⁺, n = 3; male 22/45 Gal⁺, n = 3). Significantly more MPOA neurons presynaptic to PVN^AVP and PVN^CRH were Gal⁺ in males than in females (P < 0.0001, P = 0.0170, two-tailed Fisher’s exact test) whereas more MPOA neurons presynaptic to PVN^OXT were Gal⁺ in females than in males (P = 0.0068). h, Labelling strategy for MPOA^Gal projections; example of retrogradely labelled c-Fos⁺ neuron in the MPOA. i, Activated fraction of MPOA^Gal neurons projecting to parenting-relevant brain areas (n = 7, 4, 3, 4, 3, 4, 3, 4, 3, 4, 4, 4, from top). Data are mean ± s.e.m. Red line, population average (ref. 3). Projections chosen for further functional studies are highlighted. j, Strategy for monosynaptic retrograde tracing from projection-defined MPOA^Gal subpopulations. k, l, Map of monosynaptic inputs into VTA-projecting MPOA^Gal neurons (k) and matrix displaying inputs into projection-defined MPOA^Gal subpopulations (l, Methods; n = 5, 3, 4, 4, 4, 4, 5, 5, 4, 4, 3, from top). Tukey post-hoc test assessed whether candidate projections (blue) receive quantitatively different inputs; VTA vs. PAG: *P = 0.0205, PAG vs. PVN: ***P = 0.0002, all other comparisons: ***P < 0.0001. Scale bars, b, left, 500 and inset, 250 μm; e–g, h, 50 μm.

Figure 3. Distinct projection-defined MPOA^Gal neuronal pools are tuned to specific aspects of parental behaviour

a, b, Fibre photometry recording strategy (a) and setup (b). c–i, Averaged recording traces from MPOA^Gal population activity during pup sniffing (c), -grooming (d), -retrieval (e), entering nest with pups (f), entering empty nest (g), nest building (h) and crouching (i). Mean peak activity (Z-scores) shown in mothers (n = 4), virgin females (n = 3) and fathers (n = 5). j–l, Averaged recording traces and mean peak activity during control behaviours. m, Strategy for recording projection-defined MPOA^Gal subpopulations. n–p, Mean peak activation for PAG- (n, n = 10), VTA- (o, n = 12) and MeA- (p, n = 8) projections during parenting. q, Tuning matrix for pan-MPOA^Gal (upper) and projection-specific (lower) recordings. Two-tailed t-tests (Methods), c: ***P < 0.0001, ***P < 0.0001, P = 0.0001, d: ***P < 0.0001, e: ***P < 0.0001, ***P = 0.0008, 0.0004, f: ***P < 0.0001, *P = 0.0247, g: *P = 0.0185, 0.0365, 0.0105, j: ***P = 0.0002, ***P < 0.0001, k: **P = 0.0059, n: *P = 0.0362, p: *P = 0.0102, ***P < 0.0001, ***P = 0.0001. Data are mean ± s.e.m.

Figure 4. MPOA^Gal projections mediate discrete aspects of parental behaviour

a, Setup for optogenetic manipulations. b, g, l, Activation of MPOA^Gal projections (left); pup-directed behaviour in virgin females and males without (‘OFF’) or with (‘ON’) activation of PAG- (b), VTA- (g) and MeA- (l) projections. c, h, m, Effect of activating PAG- (c), VTA- (h) or MeA- (m) projections on pup grooming (virgin females, n = 13, 9, 10; males, n = 9, 10, 10). d, i, n, Motivation assay (d) and effect of activating PAG- (d), VTA- (i) or MeA- (n) projections on barrier crossing (virgin females, n = 13, 10, 10; males, n = 13, 10). e, j, o, Intruder assay (e) and effect of activating PAG- (e), VTA- (j) or MeA- (o) projections on male-male aggression. f, k, Effect of MPOA^Gal→PAG (f) or MPOA^Gal→VTA (k) activation on male- (n = 12, 9) or female-directed (n = 10, 10) behaviour. p, Effect of MPOA^Gal→MeA activation on male-directed attack latency (n = 10) and chemoinvestigation (n = 10). q, t, w, Inhibition of MPOA^Gal projections. r, u, x, Pup-directed behaviour in virgin females without (‘OFF’) or with (‘ON’) inhibition of PAG- (r, n = 10), VTA- (u, n = 10) and MeA- (x, n = 11) projections. s, Effect of MPOA^Gal→PAG inhibition on pup grooming (n = 10). v, Effect of MPOA^Gal→VTA inhibition on barrier crossing (n = 10). y, Effect of MPOA^Gal→MeA inhibition male-directed chemoinvestigation (n = 11). Chi-square (b, e, g, j, l, o, r, u, x) or two-tailed paired t-tests (c, d, f, h, i, k, m, n, p, s, v, y), b: **P = 0.0034, c: *P = 0.0273, 0.0374, i: **P = 0.0089, 0.0056, o: *P = 0.0246, p: *P = 0.033, 0.0109, s: *P = 0.0396, v: **P = 0.0038.

Figure 5. Functional architecture of the MPOA^Gal circuit

Broad, state- and sex-specifically activated inputs converge onto largely non-overlapping, projection-defined MPOA^Gal subpopulations that elicit specific aspects of parental behaviour. Asterisk, MPOA^Gal→PVN connections are sexually dimorphic (see Fig. 2e–g).