Sensitive windows, sensitive outcomes: Early or late-postnatal maternal separation differentially impacts puberty and behavior

Abstract

Early-life adversity disrupts the maturation of neuroendocrine, cognitive, and emotional regulatory systems, yet outcomes likely depend on both the timing of development and sex. Maternal separation (MS) in rodents is a widely used model for studying these effects, yet the sex-specific consequences of distinct postnatal MS windows remain underexplored. Here, we investigated how early [postnatal day (P)2–P10], late (P11–P20), or prolonged MS (P2–P20) influences pubertal maturation, spatial memory, and anxiety-like behavior in male and female Sprague–Dawley rats. Pubertal development was assessed through daily visual inspection. Affective behavior was evaluated using dyadic social interaction and light-enhanced startle (LES), and spatial memory was evaluated in the delayed win-shift task during adolescence. Both postnatal timing of adversity exposure and sex influenced the measured outcomes. Females exposed to prolonged or early MS exhibited significantly earlier pubertal initiation, with no effect of MS on pubertal onset in males. Prolonged or early MS exposure also impaired spatial memory, with females affected by prolonged MS and males affected by early MS. Affective measures in males and females were differentially affected by early and late MS exposure. While only late MS exposure yielded more social approach-avoidance behavior in males, females exhibited approach-avoidance changes after both early and late MS. We replicated previous findings that PMS-exposed males display a blunted light-enhanced startle, and now show that late, but not early MS, yields reduced LES in males. Female adolescents, however, responded to early MS with enhanced LES. Together, these findings demonstrate that MS exerts distinct, sex-specific effects on pubertal maturation and behavior depending on the developmental window of adversity. By demonstrating window and sex-dependent effects on pubertal development, anxiety-like behavior, social interaction, and cognition, this work highlights developmental timing and sex as key determinants of early-life stress outcomes and underscores the importance of distinguishing discrete postnatal periods when modeling the consequences of early caregiving disruption.

Highlights

• •Early and prolonged maternal separation accelerates pubertal onset in females.
• •Maternal separation impacts startle modulation in a window- and sex-specific fashion.
• •Social behavior and spatial memory are differentially sensitive to separation timing.
• •Some behavioral effects of maternal separation are buffered by prolonged exposure.

1. Introduction

Early disruptions to the caregiving environment have a lasting impact on emotional, social, and cognitive outcomes across species. Maternal separation (MS) is one of the most widely used rodent models of early-life adversity originating from seminal work demonstrating that early disturbance of mother–infant interactions alters stress responsivity and later-life behavioral patterns (Levine, 1957, 2002). Across both human and animal research, the developmental timing of early adversity has emerged as a key determinant of its long-term impact, supporting the existence of sensitive periods during which early experiences disproportionately shape behavioral trajectories (Aguiar et al., 2020; Biagini et al., 1998; Callaghan and Tottenham, 2016; Gee et al., 2013; Gunnar and Quevedo, 2007; Lehmann and Feldon, 2000; McLaughlin et al., 2021; Roman and Nylander, 2005; Wang et al., 2020; Woodard and Pollak, 2020).

Rodent MS paradigms typically target early (Postnatal day (P)2–P10), late (P11–P20), or prolonged (P2–P20) postnatal windows, which align with distinct windows of neurobiological and behavioral development. The early postnatal period (P2-P10) coincides with rapid maturation of the hypothalamic-pituitary-adrenal (HPA) axis, including the stress hyporesponsive period (Levine, 2001; Sapolsky and Meaney, 1986), during which maternal cues critically regulate glucocorticoid signaling and shape long-term stress responsivity (Gunnar and Quevedo, 2007; Levine, 2005). This window also encompasses early development of limbic and hypothalamic circuits involved in affective processing and neuroendocrine regulation (Andersen, 2003; Bath et al., 2016; Tottenham and Sheridan, 2010). The later postnatal period (P11-P20) corresponds to the emergence of independent sensory processing, increased social exploration (Moriceau and Sullivan, 2006; Spear, 2000; Sullivan and Wilson, 1993) and the refinement of corticolimbic circuitry (Andersen and Teicher, 2008; Callaghan and Tottenham, 2016). It also marks a period when microglial function plays an important role in the development of cognitive function (Fanikos et al., 2024). Comparative human studies similarly show that early life adversities, depending on their nature, differentially affect outcomes such as emotional regulation, social functioning, limbic connectivity and executive control (Gee et al., 2013; Sheridan and McLaughlin, 2014). As a result, identifying timing-specific effects of early-life adversity remains essential for understanding the developmental origins of stress-related phenotypes.

Within the animal literature, MS timing often differs between studies, making it difficult to determine whether conflicting reports are the result of exposure at particularly sensitive or insensitive windows. For example, females have often been reported to exhibit greater sensitivity to the effects of prolonged MS or early MS on social and affect-related behavior (Arborelius and Eklund, 2007; Farrell et al., 2016; Holland et al., 2014; Knox et al., 2021). In one of the few investigations into the timing of MS, late MS was observed to impair spatial learning memory in male rats (Cao et al., 2014). Prolonged MS (P2–P20), which spans multiple early developmental phases, often induces broad alterations across behavioral domains, including increased anxiety-like behavior, reduced social engagement, and impairments in exploratory and task-relevant behaviors (Abraham et al., 2023; Adjimann et al., 2021; Levine, 2005; Lévy et al., 2003). However, direct comparisons across MS windows remain sparse, limiting our understanding of how specific developmental periods contribute to distinct behavioral outcomes.

In parallel, early-life adversity is often observed to affect reproductive maturation. Both human and rodent studies have shown that early-life adversity is associated with earlier pubertal onset (Belsky et al., 2015; Granata et al., 2024; Grassi-Oliveira et al., 2016), with little known about sensitive windows. Earlier pubertal timing is linked to elevated risk for anxiety and affect-related difficulties, raising the possibility that stress-induced acceleration of puberty may contribute to later-emerging emotional vulnerabilities. MS-exposed rodents consistently exhibit earlier pubertal initiation along with heightened anxiety-like responding, including increased acoustic startle (Granata et al., 2022, 2024).

Because social behavior, cognitive performance, pubertal timing, and anxiety-like behavior develop across early life and may be differentially shaped by the timing of early adversity, examining these outcomes to delineate window-specific influences on behavior is important. We hypothesized that early (P2–P10), late (P11–P20), and prolonged (P2–P20) MS would exert dissociable and sex-dependent effects across behavioral domains. Based on prior work, we predicted that females would be more sensitive to early-life stress in terms of pubertal timing and anxiety-like behavior, while social and cognitive outcomes were expected to vary as a function of the timing of separation rather than showing a uniform directional effect.

2. Materials and methods

2.1. Animals

All experiments were performed in accordance with the 1996 Guide for the Care and Use of Laboratory Animals (NIH) with approval from the Institutional Animal Care and Use Committee at Northeastern University. Animals were housed under standard laboratory conditions in polycarbonate wire-top cages with pine shavings as bedding, a plexiglass tube for enrichment, and food and water available ad libitum. The facility was kept on a 12-h light/dark cycle (light period between 07:00 and 19:00) with regulated temperature (22–23 °C) and humidity (37%–53%).

Male and female Sprague-Dawley rats, originally sourced from Charles River Laboratories (Wilmington, MA), were bred in-house. A single male and female were co-housed for a maximum of seven days, during which pregnancy was assessed daily via the presence of sperm in a vaginal swab. All females were primiparous, and upon confirmation of pregnancy, dams were individually housed. Parturition was monitored daily, with the day of birth designated as postnatal day 0 (P0). On P1, litters were culled to 12 pups, maintaining a 1:1 sex ratio (six males and six females) when feasible. Pup sex was determined on P1 based on anogenital distance. Both males and females were used for this study.

2.2. Maternal separation

Litters were randomly assigned to either prolonged maternal separation (PMS), early maternal separation (EMS), late maternal separation (LMS), or control (CON) conditions. 12 litters per rearing paradigm were used (48 litters in total), and no more than n = 2/sex were used from each litter for each behavioral outcome measure. PMS litters underwent daily separations from postnatal day 2 (P2) to P20, EMS from P2-P10, and LMS from P11-P20. From P2 to P10, each pup was individually placed in a plastic cup containing pine shavings from the home cage to maintain olfactory familiarity. Pups were isolated on familiar bedding to preserve olfactory cues while selectively depriving somatosensory and social contact, thereby minimizing novelty-induced stress and modeling disruption of caregiver-derived tactile regulation rather than global sensory deprivation (Moriceau and Sullivan, 2006; Nishi, 2020). Although olfactory familiarity was maintained, removal of maternal contact reliably engages early-life stress systems and their regulation of later life behavior, which we aim to study. These cups were positioned in a circulating water bath maintained at 37 °C for 3.5 h (09:30–13:00 h) as they are unable to thermoregulate at this age. From P11 to P20, when pups have developed sufficient thermoregulatory capabilities, each pup was placed in an individual mouse cage containing clean bedding mixed with a small amount of home cage bedding for 4 h (09:30–13:30 h). A 30 min-longer separation time was used after P11 to account for the typical increase in time spent away from the nest as pups become more mobile and thermoregulate. Throughout the separation period, MS dams remained in the home cage, housed in a separate room.

Pups were identified via ear punches on P8. Control (CON) litters remained undisturbed apart from routine husbandry procedures, including ear punches on P8 and body weight measurements on P9 and P20. To account for experimenter interaction effects without repeated stress exposure, CON pups were briefly handled by an experimenter for 3-5 min on P12 and P15. On P21, all subjects were weaned and pair-housed in standard cages with sex- and condition-matched cage mates. No more than two males and two females per litter were assigned to each experimental group.

2.3. Pubertal status measurement

The pubertal status of rats (n = 9-12/group/sex) was monitored daily by visual inspection from P21-P40, which involved handling the animal for 1-2 min every day until behavioral testing. For females, pubertal development was given a score from 1 to 3; A score of 2 referred to pubertal initiation, defined as the appearance of vaginal puckering, and 3 referred to pubertal completion, or the complete opening of the vaginal pinhole, which typically appeared 2-5 days after pubertal initiation (Table 1). Female puberty initiation was previously validated using uterine weight (Granata et al., 2024). For males, pubertal development was scored on a 0–2 scale, where a score of 1 indicated pubertal initiation, defined by the first observable preputial separation, and a score of 2 indicated pubertal completion, defined by the ability to fully retract the prepuce (Table 2).

Table 1.

Scores and the associated key for visual assessment of pubertal development in females.

Score	Marker	Key
1	A thin line	Pre-pubertal initiation
2	Vaginal puckering	Puberty initiation
3	Vaginal pinhole open	Puberty completion

Table 2.

Scores and the associated key for visual assessment of pubertal development in males.

Score	Marker	Key
0	No preputial separation	Pre-pubertal initiation
1	Slight red ring visible	Puberty initiation
2	Complete preputial separation	Puberty completion

2.4. Light enhanced startle (LES) response

Animals that were monitored for pubertal development were then tested on P40 (Fig. 1A) for LES using the Med Associates Acoustic Startle Reflex System. To minimize handling-induced stress, rats were transported to the behavioral testing room 15 min before the experiment for acclimation. During this period, startle chambers were turned on to stabilize background noise levels. Calibration weights approximately twice the weight of the heaviest animal were used to ensure accurate sensitivity settings across chambers. The startle cabinets were lined with noise-canceling foam on the walls and doors to minimize external noise interference. The chambers contained a grid rod animal holder, which was mounted on a startle platform equipped with a load cell to detect and quantify the startle response. The force exerted on the platform was converted into a voltage signal representing the animal's startle magnitude. The light potentiated startle paradigm was a 2-day paradigm.

Fig. 1.

Experimental timeline. (A) The day of birth was designated as postnatal day (P) 0. Beginning on P2, pups assigned to the prolonged maternal separation (PMS) paradigm were separated from the dams and each other, while the control (CON) litter remained in the home cage. This continued until P20. Early MS (EMS)-reared pups underwent separations from P2-P10, and late MS (LMS)-reared pups underwent separations from P11-P20. From P21 to P40, animals were assessed for their pubertal development by visualization of external markers of puberty. At P40, animals were first habituated to the experimental room before testing on the light-enhanced startle response paradigm. (B) A separate cohort of animals was used to assess the effects of different postnatal windows of separation on social interaction. Animals underwent separations as described above. At P40, they were tested for social interaction in a plexiglass box with an unfamiliar same-sex conspecific. (C) A separate cohort of animals was tested in the delayed win-shift task to assess spatial memory in adolescence.

2.4.1. Day 1

Following initial acclimation, each rat was placed in a calibrated Plexiglas animal holder within the acoustic startle chamber. The session began with a 5-min in-cabinet acclimation period, after which the animals were presented with 30 startle-eliciting stimuli, delivered at 30-s interstimulus intervals. Each stimulus consisted of a 50-ms, 95 dB white noise burst with a 3-ms rise/fall time. Background noise was maintained at 70 dB throughout testing to ensure consistent signal-to-noise conditions. Upon completion of the 30 trials, rats were removed from the chambers and returned to their home cages. Startle boxes and holders were cleaned with 50% ethanol between subjects to minimize olfactory cues.

2.4.2. Day 2

Day 2 of the experiment assessed startle reactivity under conditions of moderate illumination, a mildly aversive environmental stimulus. The light intensity was intentionally selected to create a diffusely threatening context rather than an acutely anxiogenic one. Twenty-four hours after completion of Day 1 testing, rats were transported to the behavioral testing room and allowed to acclimate for 15 min. Following acclimation, each rat was placed in a calibrated Plexiglas animal holder within the acoustic startle chamber. A lamp positioned directly above the Plexiglas chamber was activated before test initiation, producing an illumination level of approximately 100 lux. Rats then underwent the same testing protocol used on Day 1, consisting of a 5-min in-chamber acclimation to background white noise followed by 30 presentations of a 95 dB, 50 ms white noise burst delivered at 30s interstimulus intervals. Upon completion of the session, rats were returned to their home cages.

2.5. Dyadic social interaction

A separate cohort of rats (n = 5/group/sex) was exposed to PMS, EMS, or LMS and assessed for social interaction behaviors at P40 (Fig. 1B), as previously described (Farrell et al., 2016). Individual experimental animals were placed in a Plexiglas open field arena (100 cm × 100 cm) for 10 min to habituate. A naïve conspecific of the same sex and age was marked with an odorless marker and then placed into the arena for a separate 10-min habituation period. The experimental rat was then reintroduced to the arena, facing away from the conspecific. Both rats were monitored for 30 min by a CCTV camera (Panasonic WV-CP500, Secaucus, NJ) suspended directly above the arena. The camera was interfaced with EthoVision (v9.0; Noldus Information Technology, Leesburg, VA), which digitally analyzed nose-to-nose contacts, nose-to-anogenital contacts, locomotion, moving towards conspecific (approach), and moving away from conspecific (avoidance). Experimental animals were designated from conspecifics based on marker color. Each naïve conspecific was only used once. Cumulative duration of behaviors and total distance traveled were recorded with a minimum velocity of 2 cm/s set as the threshold for movement. Duration of approach and avoidance was measured with a threshold of 50 cm minimum distance traveled towards or away from the conspecific. Nose-to-nose and nose-to-anogenital contact were measured with a threshold of <5 cm distance between the nose of the experimental animal and the nose or base of the tail of the conspecific. The arena was cleaned with 30% ethanol between each test.

2.6. Delayed spatial win-shift task

A separate cohort of rats (n = 5-8/group/sex) was exposed to PMS, EMS, or LMS and tested in adolescence in an 8-arm radial arm maze with a win-shift task to measure spatial memory (Fig. 1C). At P36 rats were food restricted to 15g of their daily Prolab® diet (RMA-3000, 5P00, TestDiet®, St. Louis, MO) to motivate the rats to consume the sucrose reward pellets (181135, TestDiet®, St. Louis, MO) presented within the maze. The food-restricted subjects’ body weights were maintained at 80% of ad libitum-fed. On P38–P39 rats (n = 5-7/group/sex) were habituated to an 8-arm maze for 5 min each day. From P40, rats were trained in the win-shift task. In Phase 1, rats were introduced into the center of the maze and given 5 min to consume food rewards from four open arms, with the other four arms closed. After a 5-min delay period in the home cage, rats were reintroduced in Phase 2 to the maze with all 8 arms open but with rewards only available in the arms that had previously been blocked. An entry into an arm that did not contain a reward was marked as an error. After reaching the criterion of two consecutive days with ≤1 error in Phase 2, subjects were tested with a 30-min between-phase delay. Subjects that had not reached criterion by P55 (11% of all subjects, spread across groups) were designated as taking 16 days to reach criterion for data analysis purposes and were not subject to further testing. The maze was cleaned with 30% ethanol between each test, and the configuration of open and closed doors was changed daily.

2.7. Statistical analysis

All analyses were performed with GraphPad Prism version 10 (GraphPad Software, San Diego, California, USA) and RStudio (RStudio Team, 2023). Two-way ANOVAs (2 x 4; sex x rearing) were used to test the effects of sex and rearing on pubertal initiation, social interaction, which included: approach duration, avoidance duration, nose-to-nose duration, nose-to-tail duration, and distance traveled. Spatial memory was tested using the win-shift maze paradigm, and a two-way ANOVA (2 x 4; sex x rearing) was used to test the effects of sex and rearing on days to reach criterion and errors. Post hoc multiple comparisons were carried out using Tukey's or Dunnett's multiple comparisons test. In all tests, alpha was set at p = 0.05. For ANOVAs, partial eta square was determined using the following calculation: (Sum of Squares (SS) of the effect/SS effect + SS error) from the GraphPad Prism output files. Cohen's d was calculated using an online calculator (Social Science Statistics, 2025). Before analysis, data were assessed for outliers using Grubbs' Test and removed from the analysis. The number of outliers removed are mentioned in Supplementary Table 1. Normality of data distribution was analyzed for all variables using the Shapiro-Wilk Test of Normality.

3. Results

3.1. Postnatal timing of maternal separation differentially alters pubertal onset in a sex-specific manner

The pubertal development of males and females was monitored daily from P21-P40 via visual inspection. A significant two-way interaction of rearing x sex (F (3, 84) = 16.38, p < 0.0001, η² = 0.369) and main effects of rearing (F (3, 84) = 18.56, p < 0.0001, η² = 0.398) and sex (F (1, 84) = 61.22, p < 0.0001, η² = 0.421) were found on pubertal initiation (Fig. 2). Tukey's post hoc tests revealed that PMS-exposed females (p < 0.0001, d = 3.155) and EMS-exposed females (p < 0.0001, d = 3.402) displayed earlier pubertal onset when these groups were compared to their CON counterparts (Fig. 2).

Fig. 2.

Effects of rearing and sex on puberty initiation. In females, PMS-reared and EMS-reared groups showed accelerated puberty as compared to their CON counterparts; n = 9-12. Mean ± SEM with individual data points is shown. (****p < 0.0001).

3.2. Postnatal timing of maternal separation differentially alters light-enhanced startle in a sex-specific manner

To understand the effects of rearing and sex on situational anxiety, animals were tested in the light-enhanced startle response paradigm on P40. There was no sex × rearing interaction (F (3, 86) = 1.262, p = 0.2924, η² = 0.042) and no main effects of sex (F (3, 86) = 0.8639, p = 0.4631, η² = 0.029) or rearing (F (1, 86) = 0.0135, p = 0.9078, η² = 0.0002) on the startle response on Day 1 (without light) of the LES paradigm (Fig. 3A). No two-way interaction (F (3, 88) = 0.09579, p = 0.9622, η² = 0.003) or main effects of sex (F (3, 88) = 1.113, p = 0.3484, η² = 0.037) or rearing (F (1, 88) = 0.8902, p = 0.3480, η² = 0.010) were found on the startle response on Day 2 (with light) of the LES paradigm (Fig. 3B). However, there was a two-way interaction (F (3, 81) = 2.874, p = 0.0412, η² = 0.096) and a main effect of rearing (F (3, 81) = 3.962, p = 0.0109, η² = 0.128) on LES, measured as an increased % change in startle from baseline when in a brightly lit chamber (Fig. 3). Post hoc tests revealed that PMS-reared males (p = 0.0308, d = 0.998387) and LMS-reared males (p = 0.0097, d = 0.882489) showed lower LES compared to CON, while EMS-reared females showed greater LES (p = 0.0047, d = 1.95925) as compared to their CON counterparts (Fig. 3C).

Fig. 3.

Effects of rearing and sex on situational anxiety, tested using the light-enhanced startle paradigm. (A) There were no effects of sex and rearing on anxiety-like behavior on Day 1 (without light) of the LES paradigm. (B) There were no effects of sex and rearing on anxiety-like behavior on Day 2 (with light) of the LES paradigm. (C) PMS-reared and LMS-reared males showed a lower %change in LES as compared to their CON counterparts. EMS-reared females showed a greater %change in LES as compared to their CON counterparts; n = 9-12. Mean ± SEM with individual data points is shown. (*p< 0.05, **p < 0.01).

3.3. Postnatal timing of maternal separation differentially alters social behavior in a sex-specific manner

Examination of the social interaction behavior data revealed that there was no sex × rearing interaction on approach duration (F (3, 32) = 0.5015, p = 0.6839,η² = 0.045; Fig. 4A) but there was a main effect of rearing (F (1, 32) = 5.269, p = 0.0284, η² = 0.141; Fig. 4A) and sex (F (3, 32) = 15.45, p < 0.0001,η² = 0.592; Fig. 4A). Dunnett's multiple comparisons revealed that at P40, LMS-reared males showed increased approach as compared to CON (p = 0.0002, d = 5.924586; Fig. 4A), as did EMS-reared females (p = 0.0084, d = 2.457719; Fig. 4A), and LMS-reared females (p = 0.0003, d = 3.227598; Fig. 4A). There was no two-way interaction on avoidance duration (F (3, 31) = 2.490, p = 0.0787,η² = 0.194) and no main effect of sex (F (1, 31) = 1.935, p = 0.1742, η² = 0.059), but there was a main effect of rearing (F (3, 31) = 13.90, p < 0.0001, η² = 0.574; Fig. 4B). A priori multiple comparisons revealed that LMS-reared males showed increased avoidance (p = 0.0077, d = 2.525022; Fig. 4B), as well as both EMS-reared females (p = 0.0075, d = 1.885425; Fig. 4B), and LMS-reared females (p = 0.0045, d = 2.540787; Fig. 4B). A main effect of rearing was also observed for nose-to-nose duration (F (1, 30) = 18.79, p = 0.0002, η² = 0.385; Fig. 4C), with LMS-reared males displaying more nose-to-nose interaction than CON (p = 0.0042, d = 2.719944). No two-way interaction between rearing x sex (F (3, 30) = 1.898, p = 0.1513, η² = 0.15951658) nor main effect of sex (F (3, 30) = 2.472, p = 0.0809, η² = 0.198) was found for nose-to-nose duration. There was a rearing × sex interaction (F (3, 28) = 3.053, p = 0.0448, η² = 0.246) and a main effect of rearing (F (1, 28) = 13.75, p = 0.0009, η² = 0.329) on nose-to-anogenital duration. However, there was no main effect of sex (F (3, 28) = 1.106, p = 0.3631, η² = 0.106) and Dunnett's post hoc multiple comparisons yielded no significant results. No rearing × sex interaction (F (3, 30) = 1.308, p = 0.2899, η² = 0.116) and no main effect of rearing (F (3, 30) = 0.8881, p = 0.4585, η² = 0.082) was found on locomotion. There was a main effect of sex (F (1, 30) = 11.78, p = 0.0018, η² = 0.282), with females exhibiting overall higher locomotion than males.

Fig. 4.

Effects of rearing and sex on social interaction behavior. (A) Approach duration: LMS-reared males and females as well as EMS-reared females spent more time approaching a conspecific; n = 5. (B) Avoidance duration: LMS-reared males and females as well as EMS-reared females spent longer moving away from the conspecific; n = 5. (C) Nose-to-nose duration: LMS-reared males spent more time establishing nose-to-nose contact with the conspecific; n = 5. (D) Nose-to-anogenital duration: a sex × rearing interaction and main effect of rearing was found, but no significant pairwise differences; n = 5. (E) Distance traveled: females exhibited higher locomotion than males; n = 5. Mean ± SEM with individual data points is shown. (**p< 0.01, ***p < 0.001).

3.4. Postnatal timing of maternal separation differentially alters spatial memory in a sex-specific manner

Spatial memory was assessed using the win-shift task. The number of days to reach criterion was unaffected by rearing (F (3, 43) = 2.066, p = 0.118, η² = 0.107) or sex (F (1, 43) = 2.860, p = 0.098, η² = 0.049), and there was no rearing × sex interaction (F (3,43) = 2.001, p = 0.128, η² = 0.103; Fig. 5A). Within-phase errors (re-entries into correct arms after the reward was already received) were unaffected by rearing (F (3, 39) = 0.869, p = 0.465, η² = 0.048), but a main effect of sex showed that within-phase errors were higher in females (F (1, 39) = 7.239, p = 0.010, η² = 0.133; Fig. 5B). No rearing × sex interaction was noted for within-phase errors (F(3,39) = 1.909; p = 0.144, η² = 0.105). A main effect of rearing (F (3, 37) = 4.879, p = 0.0059, η² = 0.261) on between-phase errors (entries into arms that were baited in Phase 1 but unbaited in Phase 2) was associated with more errors committed by females exposed to PMS (p = 0.023, d = 2.806884; Fig. 5C) and by males exposed to EMS (p = 0.032, d = 0.974398; Fig. 5C), compared to their CON counterparts. No main effect of sex (F (1, 37) = 0.104, p = 0.247, η² = 0.002) or rearing × sex interaction (F (3, 37) = 1.456, p = 0.242, η² = 0.008) on between-phase errors was observed.

Fig. 5.

Effects of rearing and sex on spatial memory, tested using the delayed win shift task. (A) Number of days to reach criterion; n = 6-8. (B) Within-phase errors: females committed more errors than males; n = 5-7. (C) Between-phase errors: EMS-reared males and PMS-reared females displayed greater errors during the task as compared to their CON counterparts; n = 5-7. Mean ± SEM with individual data points is shown. (*p < 0.05).

4. Discussion

As true with all models for studying early life adversity, the MS paradigm lends itself to inter-lab variability. Implementation of MS can vary in duration and timing, separation of pups from each other versus keeping the litter together, and location of the dam relative to the pups (Brenhouse and Bath, 2019). Our findings demonstrate that the behavioral and developmental consequences of MS depend on both the timing of exposure and sex, supporting the concept that early life adversity exerts window-specific effects across multiple outcome domains. By directly comparing early (P2–P10), late (P11–P20), and prolonged (P2–P20) MS within the same experimental framework, we show that discrete postnatal periods confer differential vulnerability for pubertal maturation, anxiety-like behavior, social interaction, and spatial memory. We note that our MS paradigm also involves separating each pup from its littermates, which contributes isolation stress to the experienced adversity. Taken together, our findings align with extensive human and preclinical literature indicating that the developmental timing of adversity is a key determinant of long-term behavioral outcomes (Callaghan and Tottenham, 2016; Lehmann and Feldon, 2000; McLaughlin et al., 2021; Wang et al., 2020), and extend this work by demonstrating that even within the postnatal period, early and late separation windows produce distinct, sex-dependent behavioral phenotypes.

Consistent with our predictions and prior reports, females exposed to early or prolonged MS exhibited accelerated pubertal initiation, whereas males showed no alteration in pubertal timing. These findings replicate and extend previous work demonstrating that MS can advance pubertal onset in females (Belsky et al., 2015; Granata et al., 2024; Grassi-Oliveira et al., 2016), while also highlighting that rearing effects on pubertal timing may be constrained to specific postnatal windows. Hypothalamic circuits regulating stress and reproduction are highly sensitive to early-life disruption in a sex-dependent manner (Bangasser and Valentino, 2014; Shansky and Woolley, 2016), and direct comparisons between early and late neonatal periods have provided evidence that the hypothalamus-pituitary-adrenal HPA axis is particularly sensitive to early postnatal stress exposure (McCormick et al., 2001). Impacts of early, but not late, MS on pubertal timing, therefore, may reflect a sensitive period for neuronal populations in the medial preoptic area and arcuate nucleus that control the initiation of puberty; these regions have previously been shown to exhibit early vulnerability (Barelle et al., 2025; Eck et al., 2022).

During early infancy, stress exposure can also alter maturation of extra-hypothalamic circuits that regulate pubertal timing and emotional behavior, thereby increasing vulnerability to later-emerging affective dysregulation (Lupien et al., 2009; Maniam et al., 2014; Plotsky and Meaney, 1993; Shanks et al., 2000). While baseline startle (Day 1) did not differ across groups, we observed that females, but not males, exposed to early MS exhibited enhanced LES in adolescence. In contrast, males exhibited a blunted LES response, particularly following late MS. While reduced LES is often interpreted as diminished anxiety-like responding, blunted startle reactivity may also indicate altered threat processing of aversive contextual cues (Grillon and Baas, 2003). LES relies on the bed nucleus of the stria terminalis (BNST; Toufexis et al., 2005), which exhibits sex-specific activity patterns during fear expression (Hon et al., 2025) and social interaction (Perkins et al., 2017). The BNST undergoes developmental maturation in the early postnatal period (Carter, 2017; Chung et al., 2000; del Abril et al., 1987), rendering it vulnerable to adversity-induced changes (Dent et al., 2007). While little is known about sex differences in early postnatal development of the BNST (del Abril et al., 1987), our findings suggest that differential impacts of MS across infancy on BNST-dependent behavior reflect sex-specific sensitive periods that require further investigation. Sex-divergent effects of MS may also suggest that early adversity may bias defensive responding differently in males and females, consistent with clinical observations that trauma exposure is more often associated with internalizing symptomatology in females and altered threat processing or fearfulness in males (Gray, 1971; Taylor et al., 2000).

Maternal separation exerted timing-dependent effects on social behavior that may further reflect impacts on emotional regulation circuitry. Late MS selectively altered approach–avoidance behavior in males, whereas females displayed social alterations following either early or late MS. While overall locomotion was unchanged, subjects exposed to MS spent more time moving both towards and away from a strange conspecific, which may be interpreted as heightened approach-avoidance ambivalence or heightened social vigilance (Miller, 1944). These findings add to prior work demonstrating that social behavior is particularly sensitive to early adversity and that sex moderates these effects (Farrell et al., 2016; Holland et al., 2014; Nguyen et al., 2025; Zimmerberg and Sageser, 2011). Similar to anxious potentiation of acoustic startle, social behavior is heavily regulated by the BNST and its connectivity with cortical and limbic regions, making the BNST a likely substrate for the effects of early life adversity (Halladay and Herron, 2023). Our observation that males and females exhibit social behavior changes after early or late MS, with a male-specific requirement for only late exposure, points once again to a possible sex difference in the early developmental trajectory of the BNST. Maternally separated males also demonstrated more nose-to-nose contact with a conspecific, which is a highly investigative social touch that incites sex-specific cortical activity in males and females (Ebbesen et al., 2019). Notably, the broader sensitivity of females across separation windows parallels reports of heightened female vulnerability to affect-related consequences of early-life stress (Bangasser and Valentino, 2014), while the more restricted effects in males suggest window-specific modulation of social behavior. Effects of MS on adolescent social behavior were restricted to short-term exposures (either P2-10 or P11-20), while no effects of prolonged MS were observed. LES enhancement in females was also restricted to a single early window, without an effect of prolonged MS. We propose that prolonged MS exposure outside of a sensitive window may buffer the impacts on some behavioral processes, which further highlights the importance of directly comparing duration and timing of adversity exposure.

Maternal separation during discrete windows also produced timing-specific effects on cognitive performance; males were impaired after early MS exposure while females were only impaired after prolonged MS. These findings contrast with prior reports indicating that later postnatal MS more consistently disrupts cognitive function, particularly in spatial learning tasks such as the Morris water maze (Aisa et al., 2007; Cao et al., 2014). Importantly, these tasks differ substantially in their cognitive demands and experimental structure. Morris water maze performance relies heavily on stress exposure, escape motivation, and repeated spatial navigation under aversive conditions, whereas the win–shift task places greater demands on working memory, trial-to-trial updating, and performance stability under appetitive, low-stress conditions (Olton, 1978; Olton and Samuelson, 1976). Adversity during earlier postnatal development may therefore preferentially impact circuits recruited for consistent performance and flexible updating across delays, rather than reference memory or navigation under stress. Spatial memory in the win-shift task relies on communication between the ventral subiculum region of the hippocampus and the prelimbic prefrontal cortex (Floresco et al., 1997), which is sensitive to neonatal perturbations, as demonstrated by long-term cortical dysfunction after neonatal ventral hippocampus lesions (Meyer and Louilot, 2014; Tseng et al., 2009).

The observed sex difference in working memory (within-phase errors) was, however, consistent with previous reports that men (Astur et al., 1998; Sandstrom et al., 1998) and male rodents (Halladay and Herron, 2023) commit fewer working memory errors in the water maze. Females have been found, in comparison, to perform better than males on non-spatial working memory tasks, possibly due to a heavier reliance on the hippocampus to perform cognitive tasks in males and a heavier reliance on frontal cortical regions to perform cognitive tasks in females (Gur et al., 2002). Differences in motivational state, reinforcement type, and stress load across tasks likely contribute to divergent findings and underscore the importance of task selection when interpreting the cognitive consequences of early adversity.

Together, these findings demonstrate that developmental timing and sex critically shape the behavioral consequences of MS, and that parsing discrete postnatal windows reveals patterns of vulnerability that would be obscured by prolonged or undifferentiated stress paradigms. By showing that early, late, and prolonged MS differentially impact pubertal development, anxiety-like behavior, social interaction, and cognition, this work underscores the importance of distinguishing sensitive developmental periods when modeling the behavioral consequences of early life adversity.

4.1. Strengths, limitations and future directions

The present study has several notable strengths. First, by directly comparing early (P2–P10), late (P11–P20), and prolonged (P2–P20) MS within the same experimental framework, we were able to isolate developmental timing as a key variable rather than inferring timing effects across independent studies. This approach allows for clearer interpretation of window-specific outcomes. Second, the inclusion of both males and females enabled identification of sex-dependent vulnerabilities, which is particularly relevant given known sex differences in stress responsivity, pubertal timing, and affective behavior. Third, the assessment of multiple behavioral domains such as social interaction, cognitive performance and acoustic startle provides a broader developmental profile of stress-related phenotypes rather than focusing on a single endpoint.

Several limitations should also be considered. Although the study identifies timing-dependent behavioral effects, it does not directly measure the neurobiological mechanisms that mediate these outcomes. In particular, while we discuss potential contributions of HPA axis maturation, limbic circuit development, and pubertal neuroendocrine signaling, these systems were not longitudinally assessed across the separation windows. Additionally, MS paradigms inherently alter both deprivation and post-reunion caregiving dynamics, making it difficult to fully dissociate the effects of sensory loss from subsequent changes in maternal behavior. Finally, behavioral assessments were conducted at discrete developmental time points, limiting conclusions about the trajectory of circuit maturation across adolescence into adulthood.

Future studies will be necessary to determine how developmental timing of stress exposure interacts with maturation of stress-sensitive neural circuits to drive the observed behavioral phenotypes. Longitudinal examination of HPA axis function, corticolimbic circuitry (e.g., amygdala–prefrontal and BNST pathways), and pubertal neuroendocrine signaling across the early and late postnatal windows would clarify whether distinct sensitive periods correspond to specific phases of circuit assembly or receptor expression. Incorporating molecular and circuit-level analyses, such as receptor expression, synaptic maturation markers, immediate-early gene mapping following behavioral testing, or in vivo circuit manipulation would help establish causal links between window-specific stress exposure and later behavioral outcomes. Finally, given the sex-dependent findings, future work should explicitly examine whether stress timing differentially alters developmental trajectories of gonadal hormone signaling and stress circuitry in males and females. Together, such approaches would advance understanding of how early-life adversity becomes biologically embedded in a timing- and sex-specific manner.

Funding

This work was partially funded by the Karen Toffler Charitable Trust.

CRediT authorship contribution statement

Angarika Balakrishnan: Conceptualization, Data curation, Formal analysis, Investigation, Validation, Writing – original draft, Writing – review & editing. Prabarna Ganguly: Investigation, Methodology. Sheta Roy: Data curation, Investigation, Visualization, Writing – original draft. Heather C. Brenhouse: Conceptualization, Data curation, Formal analysis, Funding acquisition, Methodology, Project administration, Resources, Supervision, Visualization, Writing – original draft, Writing – review & editing.

Declaration of competing interest

The authors declare no competing interests.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.