Intergenerational Sex-Specific Transmission of Maternal Social Experience

The social environment is a major determinant of individual stress response and lifetime health. The present study shows that (1) social enrichment has a significant impact on neuroplasticity and behaviour particularly in females; and (2) social enrichment in females can be transmitted to their unexposed female descendants. Two generations (F0 and F1) of male and female rats raised in standard and social housing conditions were examined for neurohormonal and molecular alterations along with changes in four behavioural modalities. In addition to higher cortical neuronal density and cortical thickness, social experience in mothers reduced hypothalamic-pituitary-adrenal (HPA) axis activity in F0 rats and their F1 non-social housing offspring. Only F0 social mothers and their F1 non-social daughters displayed improved novelty-seeking exploratory behaviour and reduced anxiety-related behaviour whereas their motor and cognitive performance remained unchanged. Also, cortical and mRNA measurements in the F1 generation were affected by social experience intergenerationally via the female lineage (mother-to-daughter). These findings indicate that social experience promotes cortical neuroplasticity, neurohormonal and behavioural outcomes, and these changes can be transmitted to the F1 non-social offspring in a sexually dimorphic manner. Thus, a socially stimulating environment may form new biobehavioural phenotypes not only in exposed individuals, but also in their intergenerationally programmed descendants.

Housing Condition. The same size of polycarbonate cage was used for the social and standard housing conditions. F0 animals assigned to the social housing condition were housed and raised in groups of twelve within social housing units (86 cm × 86 cm × 41 cm). Because the present study was designed to focus on the impact of social interaction, no additional enrichment stimuli, such as objects, toys or additional nesting materials, were provided in the social housing condition. Also, food location and type were not changed. Animals were constantly living in the same units until completion of the experiment. Rats assigned to standard housing conditions were housed and raised in non-sibling groups of 2 or 3 in standard size cages (86 cm × 86 cm × 41 cm). Experiment 1 varies the number of cage mates (standard, n = 2-3 vs. social, n = 12) while keeping the size of the environments constant. In Experiment 2, however, all F1 animals (n = 9-10 per group) were housed and raised in groups of 2-3 within standard housing units (86 cm × 86 cm × 41 cm). Aspen wood mixed with shredded paper bedding material was used in both standard and social environments, and changed once per week. Animals were removed from their environments when bedding material was changed.
Blood Samples and Corticosterone Assays. The procedure for blood sampling was previously reported by 10 . Blood samples (0.5-0.7 mL) were taken one day prior to behavioural assessments. Briefly, rats were placed in a restraint tube and blood samples were obtained by a tail notch with a scalpel blade. Blood samples were collected within the first 1-2 min in the tube to ensure circulating corticosterone (CORT) levels did not increase in response to the brief stress of the procedure. Blood samples were collected in heparinized tubes. Plasma was obtained by centrifugation at 8,000 rpm for 7 min. The plasma samples were stored at −20 °C until analyzed for CORT concentration using commercial radioimmunoassay kits (Salimetrics, UK). All samples were collected in the morning hours between 9:00 and 11:00 h. No behavioural testing was performed on blood sampling day. Three animals from Experiment 1 and one animal from Experiment 2 were excluded from CORT analysis due to some technical issues and/or insufficient blood samples. has been developed for the assessment of novelty-seeking exploratory behaviour in Experiment 1 (n = 42) and Experiment 2 (n = 39). The task consisted of a corridor field made of black Plexiglas measuring 165 cm by 165 cm by 35 cm in height. A segregated internal wall (30 cm in height) was placed in parallel with the external wall making a corridor (15 cm width) around the arena where the animals were able to freely move. Also, four small passages inserted on the internal wall of the apparatus allowed animals to expand their search outside the corridor and to the center of the task. The floor of the task was divided into three zones: (1) corridor zone comprising the zone between external and internal walls, (2) open zone comprising the area within the task excluding corridor and the central zones; and (3) central zone comprising the middle area of the arena (15 cm by 15 cm).
To assess different locomotor aspects of free exploration, two variations of the CFT were used: (1) plain CFT without a central object, and (2) CFT with a central object. All rats were individually allowed to freely explore the environment starting from one corner of the CFT arena facing the apparatus external wall. The same corner Figure 1. Experimental design. F0 and F1 generations (males and females) were exposed to either standard and social housing conditions. Experiment 1(F0): Male and female pups gathered from 5 different litters were randomly assigned to four experimental groups (n = 12/group) in two housing conditions at postnatal day 21. While standard animals were kept for 96 days in standard housing in groups of 2-3, social animals were raised in groups of 12 for the same period of time. On days 97-101 behavioural assessments were completed. Five rats per group were maintained for pairing, and 7 rats were used for tissue collection. Mating: Five standard males and 5 social females, and also 5 social males and 5 standard females were paired for mating. Experiment 2(F1): Forty pups from F0 dams who were previously raised in either standard or social housing conditions were randomly selected for Experiment 2 (F1). Based on their mothers' housing conditions, the F1 offspring were then split into males and females (n = 9-11/group) and housed in groups of 2-3 for 93 days. All F1 animals were euthanized for histological assessments after behavioural testing was completed on days 94-98. was used for all rats during a test period. The experimenter left the room immediately after placing the rat in the task. All test sessions lasted 8 min, during which performance was recorded under dim illumination by a ceiling-mounted camera (CCTV Auto tracking PTZ; SONY, Tokyo, Japan) and analysed by a computer tracking system (SINA motiongraph, V.II, 2011, Tabriz, Iran) through analysis of the time spent in each zone, and path speed. After each animal, the apparatus was cleaned with 70% alcohol.

Spatial Learning and Memory. Morris Water Task (MWT):
The hidden platform version of the MWT (151 cm diameter) was used to assess spatial performance, learning and memory 11 . Animals (Experiment 1: n = 28; Experiment 2: n = 28) were tested by a one-day testing protocol (10 trials per animal). The maximum duration of each swim trial was 60 s, and the location of the hidden platform (quadrant 3) remained constant from trial to trial. Movements of the animals including time spent to find the hidden platform (latency), swim length, swim speed and thigmotaxis (wall-hugging behaviour) were recorded and analyzed by an image-computerized tracking system (HVS Image 2020, UK). A no-platform probe trial (30 seconds) was also performed three hours after the completion of the single-session hidden platform testing as a measurement of reference memory. The percentage of time that the animals spent in each quadrant of the task was recorded.
Sensorimotor Balance. Balance Beam Task (BBT): In the present study, the BBT was employed to test sensorimotor integration (coordination and balance; from 12 with modifications). Animals of both experiments (Experiment 1: n = 40; Experiment 2: n = 32) were positioned on one end of a wooden round bar (2 cm wide, 125 cm long, 50 cm high) and their home cage was located at the other end of the bar. A foam pad was placed underneath to cushion falling animals. The animals were tested for three trials on the bar, and the latency to traverse the bar and number of times the hind feet slipped off the bar were recorded.
Anxiety-Related Behaviour. Elevated Plus Maze (EPM): Anxiety-related behaviour was assessed in the EPM 13 under dim illumination. The apparatus made of black Plexiglas consisted of two open and two closed arms (all 55 × 11 cm) and was elevated 83 cm above the floor. The open arms had no side or end walls, but the closed arms had side and end walls (35 cm high). Rats in Experiment 1 (n = 32) and 2 (n = 30) were placed individually in the central square facing either the left or right open arm, and were allowed to explore the maze for 5 min. The experimenter left the room immediately after placing the rat on the maze, and the behaviour of the animals in the maze was transmitted by a camera (CCTV Auto tracking PTZ; SONY, Tokyo, Japan) and analysed by a computer tracking system (SINA Motiongraph, V.II, 2011, Tabriz, Iran) via analysis of the standard measures (time spent in open and closed arms) of the EPM. Path speed and path length (distance traveled) in the maze were also analysed. In order to minimize olfactory cues, the apparatus was cleaned with 70% alcohol after testing each animal.
Histological Assessment. Animals in Experiment 1 (F0 rats; n = 7/group) were euthanized and intracardially perfused 11 . Brains including olfactory bulb and cerebellum were removed and weighed. Brains were then fixed for coronal sectioning (40 µm) and cresyl violet staining. From the coronal sections, measurements included neuronal density (Gray Value Index; GVI) and cortical thickness in both hemispheres.
Neuronal Density and Cytoarchitectonics. Neuronal density analysis (quantitative cytoarchitectonics) was performed using ImageJ 1.47b (http://imagej.nih.gov/ij; NIH, USA). Three regions of interest (ROI) were determined between primary and secondary motor cortices (M1 and M2) and the secondary somatosensory cortical region of each hemisphere. Both left and right ROIs included the same cortical regions and mainly all six cortical layers. An absolute GVI within the ROIs was separately measured for each region 14 .
Cortical Thickness. Four points (medial, central, lateral, and ventrolateral) from each brain were selected 14 based on coordinates by Paxinos and Watson 15 . The most rostral section measured was located at ~3.20 mm anterior to bregma and the most caudal section at ~−6.30 mm posterior to bregma. For each point, a vector was considered from the tangent of the outer edge to the inner edge of the cortex. ImageJ software 1.47b (http://imagej.nih.gov/ij; NIH, USA) was used to record up to eight measurements of cortical thickness from each coronal section, four from each hemisphere. Cortical thickness in the present experiment was a mean measure of both hemispheres in three consecutive slices.
Golgi-Cox Staining. Animals in Experiment 2 (F1 rats; n = 5-6/group) were also transcardially perfused after all behavioural testing was completed. All brains were extracted from the skull and weighed. Brains were then preserved in Golgi-Cox solution (30 ml) for 14 days in a dark location, and were transferred to 30% sucrose solution for 15 days 16,17 with modification. Sectioning was performed on a vibratome at 200 µm, and the slices were mounted on gelatin-coated slides. All slices were stained according to published procedure 16 . The procedure allowed staining of entire dendrites of individual neurons. For examination, dendrites had to be completely stained and visible, free of stain precipitation and not affected by blood vessels. Neurons in the ROIs were drawn with a camera lucida. Prefrontal individual pyramidal cells (layers II-III) from the Golgi-Cox stained slices were selected and traced for analysis using a camera lucida mounted on a microscope at 200 × magnification. Neuromorphological measurements included apical and basilar Sholl analysis of dendritic length that intersect concentric circles spaced 20 μm apart on the concentric ring procedure of Sholl, dendritic branch and spine density or the number of spine protrusions on a 50-μm segment of dendrite traced at 1,000 × magnification 16,17 . In the present experiment, a total of 12 neurons (6 per hemisphere) were traced in each animal.
In Situ Hybridization. Animals (F1 rats; n = 4-5/group) were euthanized with an overdose of sodium pentobarbital at the end of Experiment 2 to rapidly remove the brains. Brains were frozen on dry ice and stored at −80 °C until further processed for in situ hybridization. The prefrontal cortex (PFC) was isolated from 1 mm thick frozen coronal sections using a blunted 16-gauge stainless steel needle based on the coordinates (between ~4.70-2.70 mm from bregma) for adult rat brains. All in situ hybridization procedures were carried out as previously described in detail 18 with modifications) and all sections were run under identical experimental conditions. Briefly, brains were sectioned on a cryostat (15 µm). The fixed, air-dried sections were incubated overnight with two 33P-labelled 48-mer oligonucleotide probes in hybridization buffer and the excess and unbound probe was washed off. Brain sections were exposed to BioMax MR-1 X-ray films for 4 weeks. Levels of brain-derived neurotrophic factor (BDNF) mRNA were analyzed and masked by optical densitometry of autoradiographic films using a computerized image analysis system (MCID, Canada) and ImageJ 1.47b (http://imagej.nih.gov/ij; NIH, USA).
BDNF Protein Analysis. The enzyme-linked immunosorbent assay (ELISA) was used to quantify BDNF protein levels in the PFC 18 . Briefly, the tissues were homogenized in 100x (w/v) ice-cold homogenization buffer containing a protease-inhibitor cocktail and further diluted 1:9 in this buffer to a total dilution of 1000x 18 . ELISA was performed on the homogenate using the BDNF Emax Immuno Assay Systems (Promega KK, Tokyo, Japan).

Statistical analysis.
In all statistical analyses (SPSS 16.0, SPSS Inc., USA), a p-value of less than 0.05 was chosen as the significance level. Effects of main factors (Group; 4 levels, Sex; 2 levels, Generation; 2 levels, Litter; 5 levels) were analyzed for body weight (31-32 days), brain weight, plasma CORT, behavioural measures (time spent in each zone in CFT, latency, swim length and speed, thigmotaxis, and probe function in the MWT, time to cross the beam and foot slips on the BBT, and time spent in open and closed arms accompanied by path length and speed on the EPM) and neuromorphological indices (GVI, cortical thickness, dendritic complexity) along with molecular alterations (BDNF mRNA and proteins) by repeated measures, one-and two-way ANOVAs. Also a two-level Group effect (F0: standard vs. social or F1: standard [from non-social mothers] vs. standard [from social mothers]) was analysed where no significant difference between males and females was observed. Post-hoc Tukey test was used to adjust for multiple comparisons where needed. Familywise error was considered prior to the multiple post-hoc analyses if necessary. All data are presented as mean ± standard error.

Results
Social experience reduced HPA axis activity. A summary of plasma CORT levels for both generations (n = 10-12/group) is shown in Fig. 2A,B. There was a significant effect of Group (standard-two levels vs. social-two levels; F 3,43 = 14.11, p ≤ 0.02; One-Way ANOVA) linked to reduced CORT levels in social groups (399.59 ± 53 ng/ml vs. 259.28 ± 54 ng/ml) indicating that both male and female F0 animals raised in the social housing condition had lowered HPA axis activity, thus reduced level of circulating CORT (all p ≤ 0.05, Post-hoc; Fig. 2A). One-Way ANOVA did not reveal significant differences between litters (p ≥ 0.76).
Despite of being reared in standard housing conditions, F1 animals born to socially-housed mothers showed diminished levels of circulating CORT when compared with animals of the same generation whose mothers were raised in the standard housing condition (269.22 ± 51 ng/ml vs. 429.67 ± 48 ng/ml). A significant effect of Group (standard-two levels vs. standard-two levels; F 3,37 = 13.84, p ≤ 0.01; One-way ANOVA) was found suggesting that HPA axis activity in F1 animals born to F0 social mothers was significantly impacted by the ancestral experiences (all p ≥ 0.05; Post-hoc; Fig. 2B). There was no significant difference between generations (p ≥ 0.37) or litters (p ≥ 0.59).
Social experience reduced body weight only in F0 females. There was a significant main effect of Group (F 3,46 = 6.37, p ≤ 0.04; One-way ANOVA) and Sex (F 1,46 = 7.09, p ≤ 0.04; One-way ANOVA) in the F0 animals (n = 12/group) where only females in the social housing group showed lower body weight compared with other groups (411.5 g vs. 467.41 g, 451.5 g and 465.83 g; all p ≤ 0.05, Post-hoc, Fig. 2C). Furthermore, a significant effect of Litter (F 5,44 = 13.11, p ≤ 0.046; One-way ANOVA, Table 1) was found through which litter 2 displayed lower body weight than other litters (all p ≤ 0.05, Post-hoc). Also, no effect of Group (p ≥ 0.84), Litter (p ≥ 0.89) and Sex (p ≥ 0.61) was found in the F1 rats in terms of body weight. Inter-generational comparisons did not show differences between F0 and F1 animals (p ≥ 0.08), in spite of slightly reduced body weight in the F1 generation (Fig. 2D).
Social experience generated sexually dimorphic novelty-seeking behaviours in F0 and F1 generations. An illustration of the corridor field task (CFT) along with paths taken by F0 rats are shown in  Social experience for 96 days significantly reduced plasma CORT levels in F0 rats (n = 12/group). F1 offspring that did not experience social interaction during postnatal development inherited the reduced level of HPA axis activity from their social mother. No significant difference was found between F0 social males and females. (C,D) Effects of social experience on body weight in F0 and F1 generations. Only F0 social females (n = 12) showed lower body weight when compared with other groups. (E&F) Brain weight and housing conditions in F0 and F1 generations. Despite a slight increase in brain weight, the F0 socially-housed animals were not significantly different when compared with standard rats. Asterisks indicate significant differences: *p ≤ 0.05; One-Way ANOVA. Error bars show ± SEM. [females] 0.126 m/s). There was, however, no significant difference between groups in terms of path speed (data not shown).
F1 rats (n = 9-11/group) displayed a similar profile of novelty-seeking behaviour in the CFT as did the F0 generation. No central object: Samples of path trajectories taken by F1 animals are shown in Fig. 3F. Significant effects of Group (F 3,34 = 8.51, p ≤ 0.04; Repeated-measures ANOVA) and Sex (F 1,34 = 13.09, p ≤ 0.04; Repeated-measures ANOVA) as well as an effect of Zone (F 2,34 = 33.59, p ≤ 0.001; Repeated-measures ANOVA) but not Litter were found. The interactions between factors were not significant (all p ≥ 0.05; Two-way ANOVA), except for Sex × Litter (p ≤ 0.03) and Sex × Litter × Generation (p ≤ 0.05). A one-way ANOVA also indicated a significant difference between males and females whose mothers were raised in social housing conditions ( Fig. 3G) suggesting that daughters of socially-reared mothers explored the arena more than their male siblings. Moreover, when compared with females whose fathers were raised in the social housing condition, female rats from socially-housed mothers explored more open and central zones of the CFT (Corridor: 369 ± 24 s vs. 243 ± 23 s, F 1,19 = 26.14, p ≤ 0.03, Open: 72 ± 19 s vs. 176 ± 21 s, F 1,19 = 13.86, p ≤ 0.02; Central: 39 ± 5 s vs. 61 ± 3 s, F 1,19 = 11.52, p ≤ 0.04; One-way ANOVA). Thus, socially-housed fathers failed to transfer the new behavioural phenotype to their daughters. No significant difference was found between male and female rats whose fathers were raised in the social housing condition in terms of time spent in each zone (all p ≥ 0.05). Central object: A significant main effect of Group (F 3,34 = 19.33, p ≤ 0.03; Repeated-measures ANOVA) and Sex (F 1,34 = 14.63, p ≤ 0.02; Repeated-measures ANOVA) as well as Zones (F 2,34 = 27.43, p ≤ 0.01; Repeated-measures ANOVA) but not Litter was found with a central object in the CFT. ANOVA showed that females whose mothers were raised in a social environment spent significantly less time in the corridor (241 ± 23 s vs. 169 ± 23 s, F 1,19 = 21.39, p ≤ 0.03; One-way ANOVA), and more time in the open zone (133 ± 21 s vs. 177 ± 25 s, F 1,19 = 11.46, p ≤ 0.04; One-way ANOVA) than their male siblings in spite of being raised in the standard housing condition. No difference was found between groups in terms of the time spent in the central zone (106 ± 14 s vs. 134 ± 13 s, p ≥ 0.33; Fig. 3H,I). A comparison between female groups whose fathers or mothers were socially-housed showed that the latter spent significantly less time in the corridor (261 ± 25 s vs. 169 ± 23 s, [females] 0.124 m/s; data not shown). In summary, social experience alone appears to affect the novelty-seeking exploratory behaviour in the CFT in a sexually dimorphic manner. Occupancy plots of paths taken by rats (averaged for 6-8 animals/group) during 8-min exploration in the central zone of the CFT as an indicator for novelty-seeking behaviour is shown in Fig. 3J. Also, Fig. 3K profiles an insignificant Litter effect in F0 and F1 rats in terms of the time spent in different zones of the CFT (all p ≥ 0.05).
Spatial navigation was not affected by social experience. Both generations have shown comparable profiles of spatial navigation in the MWT. Latency measurements over 10 trials of testing decreased across groups suggesting that all rats, regardless of their generations and sex were able to acquire and retrieve the spatial information in a similar rate. Repeated-measures ANOVA showed a significant effect of Trial (F0: F 9,25 = 11.37, p ≤ 0.04; F1: F 9,25 = 17.86, p ≤ 0.03) but no effect of Group (all p ≥ 0.05). Group by Trial effect in both generations was also significant (all p ≥ 0.05). No difference was found in swim length and speed among groups and  Social experience had no effect on sensorimotor integration. The BBT was used to test motor coordination and balance. The F0 social groups in Experiment 1 crossed the beam faster than F0 standard groups, however, this trend did not achieve significance (all p ≥ 0.05). No group or sex differences were found in the number of foot slips (all p ≥ 0.05; data not shown).
Social experience reduced anxiety-related behaviours. Rats were tested in the elevated plus maze (EPM; Fig. 3L) to assess possible anxiolytic effects of housing conditions. Reduced levels of anxiety-related behaviours were found in F0 social females compared to social males as indicated by    Fig. 3N). Furthermore, path length and speed in the EPM were not affected by housing condition and/or parental experience in F1 rats (all p ≥ 0.05).

Social experience in the F0 generation increased neuronal density and cortical thickness.
Neuronal density: The regions of interest (ROI) for gray value index (GVI) 14 in the F0 generation accompanied by a summary of housing-related changes in cortical neuronal density in F0 rats (n = 7/group) is shown in Fig. 4A Fig. 4D). Thus, social experience during development may particularly improve cortical cytoarchitectonics in female rats.

Maternal social experience in F1 females increased PFC dendritic complexity and spine density.
A family history of social experience significantly affected neuronal morphology in offspring via the maternal lineage (Fig. 4E-K). Dendritic branching: Fig. 4F illustrates changes in dendritic branching in F1 rats (n = 5-6/group) raised in standard housing.  Fig. 4J). No difference was found between groups and sexes in basilar spine density (all p ≥ 0.05, Fig. 4K). In summary, social experience in mothers alters neuronal morphology more likely in their female F1 offspring.

Maternal social experience in F1 females increased BDNF mRNA and protein expression in the medial prefrontal cortex (mPFC). BDNF mRNA expression:
A representative autoradiograph of cortical BDNF mRNA expression is shown in Fig. 4L. In situ hybridization revealed that BDNF mRNA in F1 animals born to either standard or social mothers was abundantly and differentially expressed in the mPFC leading to a significant effect of Group (F 3,15 = 17.34, p ≤ 0.03; One-way ANOVA) and Sex (F 1,15 = 24.08, p = 0.02; One-way ANOVA) suggesting that F0 mothers' social experience had a noticeable impact on BDNF mRNA expression of F1 non-social females only (n = 5; 19.71 ± 1.32 vs. 14.43 ± 1.33, 11.88 ± 1.36 and 13.9 ± 1.34; all p ≤ 0.05, Post-hoc, Fig. 4M). No effect of Hemisphere and Litter was found (all p ≥ 0.05). BDNF protein expression: One-way ANOVA indicated no differences between right and left PFC in terms of BDNF protein expression (p = 0.69). However, BDNF protein expression in male and female rats (n = 5/group) born to social mothers was significantly increased when compared with animals whose fathers were socially reared (143.64 ± 13.46 ng/g vs. 101.60 ± 13.75 ng/g; all p ≤ 0.05, Post-hoc, following an effect of Group: F 3,15 = 9.37, p = 0.04; Fig. 4N). No other significant effect was found. Thus, enhanced BDNF mRNA and protein expression in mPFC in F1 females provide further evidence that F0 social mothers transferred their social experience at the level of molecular changes only to their female offspring.

Discussion
The brain is flexible to accommodate a wide range of environmental stimuli, and even responds to experiences intergenerationally transmitted through the ancestors. The present study expands these findings by showing that (1) social experience has a significant impact on neuroanatomy and behaviour particularly in females; and (2) consequences of social experience in females can be transmitted to their unexposed female descendants. Specifically, social stimulation in the F0 generation reduced HPA axis activity in both males and females, which was accompanied by increased cortical thickness and neuronal density. Behaviourally, social stimulation led to more exploratory activity in F0 females only when compared with their social male siblings and with non-socialized counterparts. However, spatial and motor functions were not affected by social experience. In the offspring, F1 non-social females born to F0 social mothers displayed behavioural, anatomical, and molecular alterations that followed those changes seen in their social mothers: non-social females showed increased dendritic complexity and higher spine density along with enhanced BDNF mRNA and protein expression in the mPFC. While both males and females showed decreased HPA axis activity, only females displayed increased locomotion and reduced anxiety-related behaviours. Our findings confirm that social experience alone generates beneficial responses in a rat model that are transmitted to the F1 generation in a sexually dimorphic manner. The present findings demonstrate that social experiences represent an integral component of enriched environment (EE) therapeutic strategies. By isolating a single element of the EE, social stimulation, from the other constituents (e.g. cognitive, sensory and motor stimulation), we challenge the main conclusion drawn from Rosenzweig's report 1 . Our findings indicate that both neuronal density and cortical thickness in F0 animals, particularly females, were remarkably impacted by mere social experiences. Interestingly, the cortical alterations seen in F0 rats were transmitted to F1 non-social female animals in terms of higher dendritic complexities and enhanced expression of cortical BDNF mRNA and protein. The observations in social animals and their non-social descendants support the influential impact of social experience on brain structure and function in both generations.
A number of possibilities should be considered as the main source of discrepancy between the present results and the findings reported by Rosenzweig's laboratory 1 . First, Rosenzweig's study used a single-sex design with only males, while the present study design monitored sex-dependent differences in response to social interactions. Second and more importantly, Rosenzweig's work did not reveal an effect of mere social grouping, perhaps because animals were exposed to social interaction for only 30 days, albeit from postnatal day 25, whereas presently animals did experience social life for more than 90 days. This procedural contrast highlights the salience of social stimulation as a function of time, i.e., its duration in neurobehavioural plasticity. Therefore, the present findings were able to profile a long-term developmental trajectory of social experience-dependent neuroplasticity and behaviour in females that has been absent in previous studies.
From a developmental perspective, the changes observed in the present study may be explained by maternal responsiveness to offspring. Maternal care behaviours, such as licking, grooming frequency and lactation history are critical to offspring survival and development 19,20 . Therefore, results can also be potentially confounded by the quality and quantity of maternal behaviours during early development, and these characteristics may even propagate across generations 21 . Previously, studies of experiential effects on parent-offspring relationships have highlighted the elemental role of epigenetic mechanisms 22 , heritable phenotypic influences that may contribute in inducing variations in the parental brain and behaviours, and their subsequent offspring. In the absence of a systematic assessment of maternal behaviours, it is difficult to rule out these influences. How such variations in maternal behaviours in social and non-social mothers and their corresponding maternal care shape sex-dependent phenotypes still awaits further investigation.
Sex-specific morphological effects induced by social stimulation also highlight the critical function of sex steroids and other hormonal correlates involved in neuroanatomical alterations. As previously reported 23 there are fundamental differences between males and females in dendritic organization of the mPFC. Because the mPFC contains both estrogen and progesterone receptors 24 , the sexual dimorphism here can be attributed to changes in gonadal hormones in females influenced by social experience. Moreover, the neuropeptide oxytocin and related biological pathways play a key role in the anatomical and functional response to social interaction. Through oxytocin action, maternal care may be intensified, social bonding enhanced and stress responses may be dampened, thus affecting the offspring's behavioural phenotype 25 . Moreover, oxytocin may mediate sexual dimorphisms in regional neuroplasticity 26 . Thus, along with the anxiolytic actions of oxytocin 27 , social interaction may influence brain anatomy and facilitate novelty-seeking behaviours.
The prefrontal cortex (PFC) plays a key role in the control of social behaviour 28 , including play fighting. Social behaviour especially in adolescent rats is expressed by play fighting, which is vital to the maturation of brain and behaviour 29 . In turn, lesions particularly to the mPFC lead to alterations in social play behaviour in rats 30 . Experimental manipulations of play behaviour in rats result in neuromorphological changes in the PFC 31 , and functional integrity of this region has been shown to be essential for the expression of social behaviour, including play 32 . In addition, play and social interaction are associated with sensory stimulation and motor activity. Thus, social stimulation arguably represents a key aspect of an effective EE intervention that is able to promote brain development and lifelong neuroplasticity.
Our findings reveal that hypothalamic-pituitary-adrenal (HPA) axis response in F0 rats was remarkably affected by social stimulation. Notably, both social males and females transmitted the new neurohormonal phenotype to their F1 descendants. The HPA axis is known as a neurohormonal hallmark of emotionality in animals and humans 33 , and its function was previously shown to be regulated by genetic and epigenetic factors 20 , environmental interventions 7,34 and ancestral influences 35 . The observation that parental social experience can be translated into diminished HPA axis activity and transmitted to subsequent generation supports the concept of developmental programming of neurohormonal responses 36 . Although not fully known yet, the mechanisms likely include epigenetic regulation such as DNA methylation and histone acetylation to alter glucocorticoid receptor expression 37 during fetal 38 , and postnatal development 34 . Thus, a specific profile of central expression of glucocorticoid receptors 39 and mineralocorticoid receptors 40 in socially raised animals may result in new hormonal phenotypes and arguably attenuated stress responses. Responses to novelty in rats are potentially heritable 41 . Furthermore, spontaneous locomotion or social responses in female rats are not influenced by estrus cycle 42,43 , and the estrous cycle in female rats is closely associated with CORT response, a mechanism that may also affect novelty seeking. Although the present strategy for randomization of sample selection and group assignment might diminish confounding effects of the estrous cycle, conclusions about novelty seeking behaviours and also the HPA axis activity in the absence of estrous cycle monitoring should be drawn with caution.
The changes in brain morphology may at least in part be causally linked to BDNF expression 44 . For example, increased corticosterone circulation by gestational aversive experiences has been suggested to be negatively correlated with BDNF expression in the PFC 45 . While social stimulation reduced HPA axis activity, the present observations reveal increased BDNF mRNA expression in the PFC of F1 non-social rats born to F0 social mothers. The findings extend current knowledge by showing that maternal social experience may program BDNF expression in the subsequent generation in a sexually dimorphic manner. Growing evidence suggests that adverse maternal experiences may particularly program BDNF expression and thus alter brain development and mental and emotional health in later life 46 . In general, BDNF expression seems to represent a particularly sensitive measure of socioeconomic and maternal lifestyle 47 . Moreover, paternal social experiences influence BDNF expression and offspring behaviour via the maternal lineage 48 . Through these mechanisms and others, parental experiences are able to influence behaviour and brain of their offspring and potentially even further generations 17,49 .
The main mechanism for developmental programming by environmental interventions involves altered activity of HPA axis 7 . Hypoactivity of the HPA system linked to social stimulation in both sexes and generations may also determine the behavioural phenotype. Here, the behavioural phenotype was characterized by improved novelty-seeking exploratory behaviour and reduced anxiety-related behaviour. Accordingly, our previous observations revealed that increased HPA axis activity is accompanied by impaired novelty seeking in a complex environment 10 . However, the central hypothesis underlying the use of the CFT in the present experiments was that when a rat is placed in the arena, it may involve a strong motivational propensity to stay in the corridor (anxiolytic effect), and also induces an emotional situation through perceived threat within the open area (anxiogenic effect). Therefore, the corridor and open zones in the CFT depict a more dynamic picture of the psychological, mainly motivational correlates of exploratory behaviour than the classic open-field task (OFT). While the OFT seems an anxiogenic task by nature 50 , and open-field activity provides the primary indicator of emotionality 15 , the specific physical characteristics of the CFT appear to impose more motivational inhibitions to the animals, thus providing a measure of motivational levels of exploratory behaviours.
The central object in the CFT was employed to encourage exploration and/or to enhance novelty-seeking behaviour 51 . The F0 mothers who experienced social stimulation as well as their female F1 offspring extended the level of exploration to the central areas, thus growing the radius of exploratory activity and indicating lowered stress response 10 . These data not only confirm that the CFT is ideally suited to differentiate experience-dependent psychomotor profiles, but also replicates the literature of experience-dependent novelty-seeking exploratory behaviours in a new framework. They also confirm that females seem to be more responsive to social stimulation than males.
Experiences, such as social stimulation, may generate robust phenotypes that may even be transmitted to the offspring through physiological programming 36 or epigenetic mechanisms 52 , which then determine stress sensitivity in the F1 generation and beyond. Early experiences in particular, including prenatal stress, malnutrition or exposure to toxicants, seem to be the most salient to result in phenotype programming in the offspring 52 . A stress-sensitive phenotype is usually linked to downregulation in hippocampal and cortical glucocorticoid receptor and neuronal densities 7 . Through these mechanisms ancestral experiences may promote the formation of new behavioural phenotypes in subsequent generations 49,53 via the paternal or maternal lineages 54,55 . Moreover, early-gestation stress translates into adverse consequences for brain development, behaviour, and overall health 37 and is associated with epigenetic inheritance of a stress-sensitive phenotype in second-generation offspring of the male lineage 22 and third-generation offspring in the female lineage 53 . While most research focused on phenotype programming by an early environment, experiences in later life may also produce stress-sensitive phenotypes 56 . The present data show that the pre-conceptional social environment generates a phenotype that may be transmitted to subsequent generations in a sex-dependent manner.

Conclusions
The classic paradigm of EE combines inanimate and social interaction although there is no consensus yet on which EE paradigm is ideal to promote brain development and behaviour 57 . Unlike the classic EE model, our paradigm isolated the impact of social stimulation to show that social experience generates new sex-dependent neurohormonal and behavioural phenotypes. In agreement with Welch and others 8 , we report in rats that mere social experience can change brain structure and behaviour. Moreover, we show that social experience-induced alterations can be transmitted to the non-exposed offspring. The data suggest that females more than males benefit from a socially stimulating environment. The present findings show that social mothers predominantly transfer the newly formed phenotype to their non-social daughters through mechanisms involving altered HPA axis response and BDNF expression. Programming of the experience-and sex-dependent phenotype is arguably generated via epigenetic regulation. Additional novelty and complexity of the environment (e.g. greater levels of sensory stimulation and physical activity) may have further promoted structural and functional improvements as reported in traditional EE models. Nevertheless, the present findings represent a stepping stone for further investigation of stress-associated psychomotor pathologies and related endocrine and epigenetic biomarkers, and their mitigation by social stimulation. Most importantly, a socially stimulating environment may moderate these pathologies not only in the afflicted individuals, but also in their intergenerationally programmed descendants.