Hormone-mediated neural remodeling orchestrates parenting onset during pregnancy

Abstract

During pregnancy, physiological adaptations prepare the female body for the challenges of motherhood. Becoming a parent also requires behavioral adaptations. Such adaptations can occur as early as during pregnancy, but how pregnancy hormones remodel parenting circuits to instruct preparatory behavioral changes remains unknown. We found that action of estradiol and progesterone on galanin (Gal)-expressing neurons in the mouse medial preoptic area (MPOA) is critical for pregnancy-induced parental behavior. Whereas estradiol silences MPOA^Gal neurons and paradoxically increases their excitability, progesterone permanently rewires this circuit node by promoting dendritic spine formation and recruitment of excitatory synaptic inputs. This MPOA^Gal-specific neural remodeling sparsens population activity in vivo and results in persistently stronger, more selective responses to pup stimuli. Pregnancy hormones thus remodel parenting circuits in anticipation of future behavioral need.

Fig. 1. Hormone action on MPOA^Gal neurons is critical for pregnancy-induced onset of parental behavior.

(A to C) Testing pup-directed behavior in repeatedly pup-exposed pregnant females (A, Preg, n = 10), pregnant females exposed to pups twice (B, Dual, n = 9) and repeatedly pup-exposed ovariectomized females (C, OVX, n = 10). Day of pregnancy (A and B) or relative to pairing with male (C) shown. (D and F) Parental behaviors in Preg group. Within-group (Preg, virgin [Vir] vs each subsequent timepoint, red asterisks) and between-group (Preg vs OVX, black asterisks) shown. (E and G) Comparison of Vir and D18 timepoints across groups. Note that virgins from Preg and Dual groups are pooled (fig. S1M). (H) Behavioral state transition diagrams for Vir and D18 females (Preg, n = 10). Average transition probabilities (P_T) between behaviors are shown, and differences between Vir and D18 highlighted if P < 0.05 (U test, see materials and methods). (I) AAV-mediated ablation (KO) of Esr1 or PR in MPOA, and control (ctrl). (J to L) Effects of MPOA-wide KO of Esr1 or PR on pup-directed behaviors (n = 7, 8, 9 mice). (M) KO of Esr1 or PR in MPOA^Gal neurons. (N to P) Effects of MPOA^Gal- specific KO of Esr1 or PR on pup-directed behaviors (n = 8, 5, 13 mice). Kaplan-Meier survival analysis with log rank test in (D), (E), (J) and (N), Fisher’s exact test with Benjamini-Hochberg adjustment for multiple comparisons in (F), (G), (K), (L), (O) and (P). Shaded area in (D) is SEM. ***P < 0.001, **P < 0.01, *P < 0.05.

Fig. 2. Hormonal remodeling of MPOA^Gal neurons.

(A) Whole-cell recordings from wild-type (upper panel) and receptor-deleted (bottompanel, KO) MPOA^Gal neurons. (B) Cumulative distribution of baseline firing frequency (Vir, D18; 33, 21 cells from n = 15, 7 mice). (C) Resting membrane potential of control and receptor-deleted MPOA^Gal neurons, and recordings in presence of GIRK channel blocker Tertiapin-Q (Tert-Q) (38, 32, 15, 18, 26 cells from n = 15, 9, 3, 3, 5 mice). (D) Example current clamp recording traces of cells with (Vir) and without (D18) depolarization block. (E) Percentage of neurons exhibiting depolarization block (34, 30, 18, 25 cells from n = 15, 8, 3, 5 mice). (F) Example voltage clamp recording traces with sPSCs. (G) sPSC frequency (21, 23, 18, 26 cells from n = 9, 6, 3, 5 mice). (H and I) EPSC (H, Vir, D18; 31, 30 cells from n = 5, 4 mice) and IPSC (I, 31, 28 cells from n = 5, 4 mice) frequency. (J) Dendritic segments of MPOA^Gal neurons with spines. (K) Spine density (14, 10, 8, 15 cells from n = 10, 4, 3, 4 mice). (L) Summary scheme for hormonal remodeling of MPOA^Gal neurons. U test in (B). One-way ANOVA with Dunnett’s post hoc test in (C), (G) and (K). Fisher’s exact test with Benjamini-Hochberg adjustment in (E). K-S test in (H) and (I). Scale bars, 20 μm (A), 10 μm (J). ***P < 0.001, **P < 0.01, *P < 0.05.

Fig. 3. Reorganization of MPOA^Gal population activity during pregnancy.

(A) Recording setup for miniature microscope recordings. (B) Gal-Cre animals were injected into the MPOA with AAV-FLEx- GCaMP7s and implanted with a GRIN lens. GCaMP7s expression and GRIN lens position shown. (C) Experimental design (see materials and methods). (D) Sample recording frames with detected neurons and example activity traces from a virgin. (E) Number of detected (non-silent) neurons per animal (n = 5 mice). (F and I) Temporal profile of MPOA^Gal responses during pup retrieval (F) or sniffing (I) in virgins, at D18 and D50 (162, 77, 93 neurons from n = 5 mice). Dashed lines indicate action onset. Order based on hierarchical clustering sorted by mean cluster response onset. (G and J) Fraction of neurons with positive evoked response during pup retrieval (G) or sniffing (J, n = 5 mice). (H and K) Averaged Z score for neurons activated during pup retrieval in virgins, at D18 and D50 (H, 115, 41, 63 neurons from 5, 5, 4 mice) or sniffing (K, 122, 51, 86 neurons from 5, 5, 4 mice). Two-way ANOVA with Tukey post hoc test; gray bars indicate periods of significant difference for Vir vs D18 and Vir vs D50. (L) Correlation between normalized tuning index for responses to pup sniffing and normalized mean baseline activities at D18 (r² = 0.202, P < 2.4 × 10^-5). (M) Selectivity of chemoinvestigation-associated responses for indicated stimulus pairs at Vir, D18 and D50 (142, 35, 108 cells from n = 4, 3, 4 mice) compared to pups. A selectivity score of 1 means the neuron is only activated during pup sniffing, a score of 0 means selective activation during sniffing of other stimulus, and 0.5 equals a non-selective response (see materials and methods). (N) Example MPOA^Gal neuronal activity at Vir and D18 during object investigation in LDA space (int, intruder; obj, screw, dummy pup). Temporal bins were used as features. Ellipsoids represent 95% confidence area of neuronal activity to each stimulus. (O) Separability of indicated stimulus combinations by the MPOA^Gal population (RI, Rand Index, n = 4, 3, 4 mice). (P) Correlation between separability and activated fraction of neurons during pup retrieval (r² = 0.56, P < 6.1 × 10^-21). Paired t tests in (E), (G) and (J), mixed linear model with mouse ID as group in (M). Linear regression in (L) and (P), unpaired t tests in (O). Scale bar in (B) 500 μm. ***P < 0.001, **P < 0.01, *P < 0.05.