Inhibition of the medial amygdala disrupts escalated aggression in lactating female mice after repeated exposure to male intruders

Maternal behaviour in rodents depends on extensive genetic, hormonal and neural adaptations that optimize offspring survival. A hallmark adaptation is maternal aggression, an intruder-directed defensive behaviour that is typically robust during early lactation. While dams rapidly attack unfamiliar conspecifics—especially adult males—virgin females co-housed with dams and litters (pup‑sensitized virgins) readily provide pup care but do not show aggression in a first test. This suggests intruder-directed aggression is more dependent on pregnancy/lactation-related endocrine changes than pup care. The authors hypothesised that repeated exposure to intruders might induce or escalate aggression through motivation and learning, potentially enabling pup‑sensitized virgins to develop aggression. Given the central role of chemosensory cues and the medial amygdala (MeA) in processing pheromonal, olfactory and hormonal information and in driving social behaviours, the study asked: (1) does repeated testing (PPD4–PPD6) induce or escalate intruder-directed aggression in dams and pup‑sensitized virgins? and (2) is MeA activity necessary for the experience-dependent escalation of maternal aggression? To test this, they compared behaviours across repeated tests and chemogenetically inhibited MeA with hM4Di DREADDs during intruder encounters.

Prior work shows maternal aggression peaks early in lactation and declines as pups mature. Chemosignals, notably male urinary proteins such as darcin, are key triggers: darcin attracts virgin females but elicits aggression in dams, indicating state-dependent valence shifts tied to hormonal status. The vomeronasal system and MeA are pivotal: MeA receives AOB inputs, expresses steroid and prolactin receptors, undergoes gene-expression changes during lactation, and projects to hypothalamic nuclei (e.g., VMH, BST) that govern social and defensive behaviours. Disruptions of MeA impair social behaviours including maternal care and aggression; aromatase-positive MeA neurons regulate attack initiation in both sexes. Experience-dependent aggression escalation has been demonstrated in males and depends on plasticity (LTP) in MeA→VMH/BST pathways, suggesting motivational and learning components may also underlie maternal aggression escalation. Pup care can be induced in virgins by co-housing, but intruder-directed aggression typically requires hormonal changes of pregnancy/lactation and sustained pup contact (e.g., hormone-treated virgins with prolonged suckling).

Design: Two experiments using CD1 mice tested (1) effects of repeated intruder exposure on aggression in dams versus pup-sensitized virgins (PPD4–PPD6), and (2) effects of MeA inhibition via DREADDs on maternal aggression across two sessions (PPD4–PPD5). All tests were in the home cage, 5 min each, with a different adult male intruder per day. Behaviour scoring was by a blinded observer using SMART 3.0. Experiment 1 (Repeated exposure): Subjects: 16 adult females (8–12 weeks; 8 dams, 8 pup‑sensitized virgins) and 24 adult males. Dams were mated; day of birth = PPD0; litters culled to 8 pups at PPD2. Pup‑sensitized virgins were pair-housed with dams after mating and continuously exposed to pups, sharing pup care. From PPD4–PPD6, each subject was exposed to a different adult male intruder for 5 min. To avoid vicarious learning, the co-housed female (dam or virgin) was moved to a separate cage prior to testing. Pups were removed immediately before testing to prevent injury. Behaviours quantified: number and duration of attacks (bites, chases), latency to first attack, anogenital investigation, and body approaches (female↔male), plus intruder-initiated interactions. Housing: food/water ad libitum, 12 h light:dark, 22–24 °C. Ethical approval and EU Directive 2010/63/UE compliance were stated. Experiment 2 (MeA DREADD inhibition): Subjects: 56 adult females and 82 adult males (8–12 weeks). Females randomly assigned; two DREADD groups received AAV5-hSyn-HA-hM4D(Gi)-IRES-mCitrine (hM4Di) bilateral MeA injections; treatment order counterbalanced: Group 1 received CNO on PPD4 and vehicle (Veh) on PPD5 (n=17 initially), Group 2 Veh on PPD4 and CNO on PPD5 (n=18 initially). Two additional control groups without DREADD received the same CNO/Veh regime to test for CNO-only effects. Only females with confirmed MeA-targeted infections (15 cases) were included in primary analyses. Surgical details: Isoflurane anaesthesia; SC atropine (0.05 mg/kg) and butorphanol (5 mg/kg) for pre/post-op care; AAV titer 3.5×10^12 vg/ml; 0.25–0.3 µl per site at 0.25–0.3 µl/min using glass micropipettes; pipette left in place 10 min; stereotaxic coordinates relative to Bregma (flat skull): AP −1.5 to −1.6 mm; ML ±2.0 to 2.2 mm; DV −4.7 to −5.0 mm. Wounds closed with Histoacryl; 2 days recovery before mating; behavioural testing 3–4 weeks post-infection. Drug administration: i.p. CNO 5 mg/kg (in saline/PBS 0.9%) or Veh given 30 min before each test; each dam tested once with CNO and once with Veh on consecutive days. Post hoc histology/immunohistochemistry for mCitrine confirmed infection sites (DAB/ABC protocol). Statistics: Normality (Kolmogorov–Smirnov with Lilliefors), homoscedasticity (Levene), sphericity (Mauchly) checked; non-normal data log-transformed (log[X+1]). Experiment 1: repeated-measures ANOVA across DAY (PPD4–PPD6), with non-parametric Friedman tests and Wilcoxon post hoc where appropriate; between-group comparisons by Mann–Whitney. Experiment 2: repeated-measures ANOVA with DAY (PPD4, PPD5) as within-subject factor and TREATMENT order as between-subject factor; latency analyses via Kaplan–Meier log-rank when censored (no-attack) cases occurred; Bonferroni corrections applied. Graphs via GraphPad Prism 8.

Experiment 1: Dams exhibited robust maternal aggression on all days, while pup‑sensitized virgins showed negligible aggression but engaged in social investigation. Between-group differences were significant for all behaviours (Mann–Whitney U=0; p<0.01): dams had greater total attack duration, shorter attack latency, and lower social investigation than pup‑sensitized virgins. In dams, aggression escalated across days: total duration of attacks increased (repeated-measures ANOVA DAY effect: F2,6=9.988, p=0.012; Bonferroni PPD4 vs PPD5 p=0.009; PPD4 vs PPD6 p=0.01), and latency to first attack decreased (log-transformed; F2,14=4.5, p=0.031; PPD4 vs PPD5 p=0.042). Attack structure also shifted to longer bouts and more long attacks (Supplementary Fig. 1). Anogenital investigation in dams decreased across days as aggression escalated (Friedman χ²(2)=13, p=0.002; trend PPD4 vs later days p=0.058); body approaches did not change significantly (Friedman χ²(2)=4.588, p=0.101). Pup‑sensitized virgins did not show changes across days in aggression (χ²(2)=0, p=1) or social investigation (anogenital: F≈2.60, p=0.967; body: F≈2.63, p=0.097). Experiment 2: Chemogenetic MeA inhibition prevented the expected experience-dependent escalation of maternal aggression. In DREADD-infected dams, repeated-measures ANOVA on log-transformed total attack duration revealed a significant DAY effect (F1,13=8.452, p=0.012) and a significant DAY×TREATMENT interaction (p=0.021). Post hoc tests showed increased aggression on the second day only when Veh was administered on PPD5 (p=0.002); no increase occurred when CNO was given on PPD5. Latency analyses: Group 1 (CNO PPD4, Veh PPD5) showed a trend toward decreased latency on PPD5 (Kaplan–Meier log-rank χ²(1)=3.813, p=0.051); Group 2 (Veh PPD4, CNO PPD5) showed no day difference (χ²(1)=0.071, p=0.790). MeA inhibition did not completely abolish initial maternal aggression on PPD4. Additional non-DREADD control groups showed no significant CNO-only effect on maternal aggression (Supplementary Fig. 3).

The study demonstrates that maternal aggression in dams escalates with repeated intruder exposure over PPD4–PPD6—evidenced by increased total attack time, longer attacks, and shorter latencies—while pup‑sensitized virgins remain non-aggressive despite continuous pup exposure and repeated testing. These results support a dual contribution of learning/motivation and hormonal state to maternal aggression: experience likely enhances recognition and efficiency of defensive responses, while pregnancy/lactation-associated endocrine changes calibrate the valence of male chemosignals (e.g., darcin) from attraction to threat. The MeA, which integrates vomeronasal inputs with steroid/prolactin signals and projects to BST/VMH and other nodes of the social behaviour network, is crucial for this experience-dependent escalation. Chemogenetic MeA inhibition did not eliminate initial maternal aggression but specifically prevented its increase on the second day, paralleling findings in males where aggression escalation depends on LTP in MeA→VMH/BST circuits. This suggests that MeA mediates the plasticity and motivational amplification required for escalating maternal aggression, possibly by gating pheromonal-hormonal integration and relaying to downstream hypothalamic and brainstem effectors. Shared circuitry with intermale aggression is likely, with sex differences arising from receptor and enzyme expression (e.g., steroid receptors, aromatase, prolactin) within MeA and targets.

Maternal aggression in mice increases with experience during early lactation, whereas pup‑sensitized virgin females do not develop intruder-directed aggression even after repeated exposure. The medial amygdala is essential for experience-dependent escalation: chemogenetic inhibition of MeA blocks the typical second-day increase in aggression without fully abolishing initial attacks. These findings highlight MeA as a critical integrator of hormonal and chemosensory information that enables plastic, experience-driven amplification of maternal defence. Future work should dissect specific MeA neuronal populations and projection pathways (e.g., to VMHvl and BST) mediating this plasticity, determine the hormonal mechanisms (prolactin, steroids) modulating MeA-dependent learning and motivation, and refine causal manipulations to probe initial versus escalated aggression components.

- Test duration was 5 minutes (short end of typical 5–15 min assays), potentially missing very long attack latencies, especially in pup‑sensitized virgins. - MeA inhibition did not completely suppress initial aggression; authors note possible limited statistical power and that stronger or more precise inhibition tools might reveal an effect on initial attacks. - Only animals with confirmed MeA-targeted infection were analysed (n=15), and infection extent varied across MeA subnuclei and adjacent regions, which could introduce variability. - Although control groups suggested no CNO-only effect, off-target pharmacology of CNO/clozapine cannot be entirely excluded in general DREADD paradigms. - Findings are in CD1 mice and under specific housing/testing conditions; generalizability to other strains, contexts, or longer postpartum windows may be limited.