Maternal separation in rats induces neurobiological and behavioral changes on the maternal side

The time after parturition is a sensitive period for mothers where they are prone to develop psychopathological symptoms. Studies investigating dams after separation from their pups (maternal separation, MS) showed that MS induces alterations similar to postpartum depression. This study aims to give further details on affected behavior and neurobiology of dams after MS. MS in rats from postnatal day 2–20 over four hours daily was performed. Upon reunion, maternal behavior, and ultrasonic vocalization (USV) of dams were measured. On the day of weaning, dams were tested for anxiety-like behavior in the elevated-plus-maze and marble burying test. Then Morc1 mRNA in the medial prefrontal cortex and Nr3c1 encoding the glucocorticoid receptor mRNA in the hippocampus were measured using real-time PCR to examine possible neurobiological correlates in psychopathology and social behavior. GABA and glutamate serum levels were analyzed by high-performance liquid chromatography as peripheral markers for stress-induced psychopathology. MS in dams increased maternal care towards pups even though both groups show high levels of maternal behavior even in late lactation. Furthermore, the emission of 50-kHz and 22-kHz USVs increased significantly. No differences in anxiety-like behavior were detected. MS further reduced Morc1 but not Nr3c1 expression. Serum GABA but not glutamate levels were significantly increased in separated dams. This study reinforces the benefit of investigating dams after MS for studying postpartum stress. Subclinical markers mainly connected to depression, namely Morc1 and GABA, proved to be useful allowing for earlier detection of symptoms of critical postpartum stress.


Results
Behavioral results. Maternal care and USV detection. Repeated measurements for parameters over the 4 testing days showed significant results between the groups: F (1/15) = 286.25; P = 0.000.
Anxiety-like behavior. EPM did not reveal significant differences between groups by time spent in the open arms (control: t (10) = 1.478; P = 0.17). The mean number of marbles fully covered did not differ significantly between groups (control: t (10) = 0.158; P = 0.877). Test results are given in the supplementary www.nature.com/scientificreports/ Neurobiological alterations. Dams exposed to chronic separation from pups for four hours daily demonstrated significantly higher Morc1 delta CT values compared to controls: F (1, 15.5) = 8.19; P < 0.01 measured in the mPFC region ( Fig. 2A). Values are given in supplementary Table 3. No significant correlations were found with Morc1 mRNA levels for the parameters with significant behavioral results (Morc1 compared to LG: r = − 0.04; Arched-back nursing: r = 0.34; PD20: 22-kHz: r = 0.14; PD20: 50-kHz r = 0.02). Interestingly, Morc1 levels showed significant correlations with the parameters of self-grooming, first pup retrieval and complete pup retrieval at the last day measured (Morc1 and Self-grooming at PD20: r = − 0.594, P = 0.04; Morc1 and first pup retrieval at PD20: r = − 0.814, P = 0.01; Morc1 and complete pup retrieval at PD20: r = − 0.922, P = 0.01). However, after correcting for multiple testing these values did not reach significance. Furthermore, no significant correlations with Morc1 mRNA levels and anxiety-like measures were detected (Morc1 to marbles fully buried: r = − 0.33 and EPM time spent in open arms: r = − 0.127).

Discussion
Upon reunion, MS lead to overall increased maternal care and an increased number of calls emitted. It might be possible that increased maternal behavior is a manifestation of anxiety leading to exaggerated maternal care similar to overcompensation seen in humans. However, as we did not observe altered anxiety but increased 50-kHz emission, the rewarding aspect upon reunion seems to be more likely. Further studies are needed to support this claim. Testing for changes in anxiety behavior did not show significant differences between groups. However, anxiety behavior might be influenced by estrus as dams tested in the proestrus generally spent more time in the open 35 . Moreover, most studies using the long MS paradigm did not find any alterations in anxietylike behavior in dams 36,37 suggesting no direct effect of MS on the anxiety state of dams.
Neurobiological measurements revealed less Morc1 mRNA in the mPFC after MS exposure but no differences in hippocampal Nr3c1 mRNA expression. Regarding serum level analysis, significantly higher GABA levels were found in dams subjected to MS compared to controls whereas glutamate levels did not differ.
It seems that MS in rat dams during the vulnerable time of postpartum can be seen analogous to the vulnerable timeframe of pups. Accordingly, MS affects both, dams and pups and thus, it is not possible to distinguish between effects induced by the separation alone or by the altered interaction between dams and pups. The postpartum period is characterized by several physiological changes that make women prone to develop hormonal and neurochemical imbalances triggering psychiatric disorders and respective symptoms 38 . www.nature.com/scientificreports/ As highlighted in other studies, we demonstrated that separated rat dams showed altered behavior towards the pups after experiencing separation stress. In this study, the separated dams increased the maternal care towards their pups upon reunion whereas controls demonstrated significantly higher levels of self-maintenance. Thus, the chronic postpartum stress and subsequent induction of maternal care might compensate and buffer the chronic stress of forced separation not only on the maternal side but also buffer stress-related responses on pups with maternal care as a possible mediating factor. Of note, our MS protocol proceeds very similarly to the protocol of other groups 4,7 . However, several differences in time and length of separations were detected 39 . Contrary to findings of other groups 40 , the dams were still maternal until the last day of testing. One interesting topic in the context of high levels of maternal behavior might be mesolimbic dopamine signaling as an up-and downregulation of this system has been shown to increase or decrease the responsiveness of maternal behavior during lactation, respectively 41 . Increased mesolimbic dopamine signaling in late lactation would be in line with increased positive-related 50-kHz USV upon reunion until weaning. This might illustrate that the reunion with pups is more rewarding after a long time of absence. Surprisingly, the 22-kHz frequencies at PD20 upon reunion were also higher in separated dams implying that separation has also led to negative mood-alterations compared to controls. Dams might be overstrained by their pups in a stress-induced mood, therefore increase the number of negatively associated affective calls as well.
On the other hand, the emitted USVs could also serve as a communication tool towards pups 42 meaning that the increased number of calls emitted by dams upon reunion could be a response to the increased number of calls emitted by the pubs upon reunion. This could imply an affective dimension that the separation procedure is not severely limited by adaptation of dams to separation protocol but rather induces more positive and negativerelated USVs in the long-term as detected in a non-invasive way of recording.
Of note, we have seen an increase in the number of 50-kHz frequencies over time which might imply the later tests were conducted with dams, the more positive emission of USVs was done by the dam upon reunion. Therefore, a constant positive affect after reunion with pups is noticeable, which is surprising given the fact that with the aging of pups, the tendency to show maternal care decreases over time. This might indicate a disruption on the behavioral level. However, the reason for the increase in maternal care in the context of the MS paradigm needs further elucidation. PD20 as the final day of testing showed significant results on both frequency ranges, but not on the first day of testing (PD2). This can show possible alterations in the affective state that are induced not immediately but rather through the repeated, chronic stress of separation. Kalinichev and colleagues also www.nature.com/scientificreports/ demonstrated that rat dams were more likely to emit USVs after experiencing the long MS protocol compared to controls 43 underlying our results that rat dams were also affected by separation on mood level. The here found increase in 50-kHz calls is also in line with other groups 18 . However, one limitation of the study is that more precise analysis tools are needed to distinguish between maternal and pup USV when both are recorded simultaneously. Gene expression downregulation of Morc1 in separated dams compared to controls was found. Downregulation of Morc1 either by knock-out variants 25 or hypermethylation of Morc1 DNA 22,23 has been associated with depressive symptoms in other studies.
As there were no significant differences in anxiety-like behavior between groups but significant differences in Morc1 expression it might be possible that reduced Morc1 expression could serve as a subclinical marker indicating changes before pathological alterations manifest in behavior. This would be in line with our previous study investigating healthy young adults with subclinical depressive symptoms 23 . The chronically stressed dams did not directly demonstrate depressive-like behavior while increasing care towards pups. Correlating Morc1 expression and maternal behavior revealed negative associations between Morc1 and self-grooming, first pup retrieval, and complete pup retrieval at the last day measured which did not reach significance after correcting for multiple testing. Again, this might reinforce the assumption of a subclinical pathological picture induced by MS and Morc1 as a potential regulator. Thus, the possible role of Morc1 in subclinical or clinical depressive symptoms needs further investigation. However, Morc1 might still be suitable for detecting chronic stress-induced neurobiological alterations.
In contrary to Morc1, our results on Nr3c1 mRNA expression in the hippocampus to analyze possible acute stress reaction adaptations in the central nervous system did not reveal significant differences between groups. This has also been shown in another study with mice dams after MS 44 . Nevertheless, contradictory results with higher and lower Nr3c1 mRNA expression have been reported in dams experiencing MS 12,45 . As Nr3c1 expression is a possible acute stress marker 46 , and is highly influenced by the maternal care the mother expresses upon reunion with pups 47 , upregulation driven by environmental factors likely compensate neuronal adaptation of respective receptor expression.
We examined GABA und glutamate serum levels as potential peripheral markers for depression 48 and other stress-related disorders. Studies with depressed women showed that measurements of plasma GABA levels correlated with aggressiveness levels 34 . Moreover, as there are several markers associated with stress, this study aimed at covering both central and peripheral markers to better characterize Morc1′s possible role in the context of stress and depression and to facilitate translation of obtained results. Thus, finding significantly higher GABA levels after MS reinforces the pathological effects of MS on dams described above. However, other studies demonstrate that the severity of depression as a chronic stress-related disorder and its symptoms are associated with glutamate levels in the blood 49,50 . As stress-related disorders have an impact on the blood-brain barrier, peripheral GABA and glutamate levels can serve as a non-invasive peripheral marker to enlighten peripheral stress-induced alterations.
The fact that some behavioral and biological pathology markers were detected (as increased maternal care and USV together with increased GABA serum levels as well as decrease Morc1 mRNA) but no difference in others was shown (such as Nr3c1 mRNA and glutamate serum levels together with no altered anxiety) might indicate that long term MS exposure induces subclinical depressive changes but by the time of measurement only alters some markers and not all at once.
In conclusion, analyzing maternally separated dams on the behavioral and neurobiological level can provide important information on the vulnerable time of postpartum. Investigating the maternal side might also prove necessary when analyzing the consequences of MS on pups. As maternal behavior is significantly increased after MS, this alteration might have effects on chronic stress coping of the pups. Analyzing USVs of dams upon reunion with pups proves to be an easy-to-apply and reliable tool to characterize affective states of dams. Combining behavioral and affective analyses can expand our knowledge on several chronic stress-related disorders in animal models and humans.
Besides the sensitive neonatal period, it seems that the sensitive postpartum time is vulnerable to chronic stress-induced alterations in Morc1 expression and GABA serum levels as well. Further steps to characterize the psychopathological effects of MS on dams should include more behavioral analysis on depressive-like behavior.

Materials and methods
Animals. All experiments were conducted in accordance with the principles of Germany's Animal Welfare Act after approval by the animal ethics committee of the Landesamt für Natur, Umwelt und Verbraucherschutz (LANUV) in Northrhine-Westfalia. 32 timed-pregnant Sprague-Dawley rats (Charles River Laboratories, Sulzfeld, Germany) arrived at gestational days 13 to 15. Pregnant dams were housed separately in plastic cages at controlled room temperatures (22 °C) and humidity conditions (22 ± 2 °C and 55 ± 25%) with standard lighting (12 h light and 12 h of dark cycle, lights on at 11.00 pm). Dams were randomly assigned to MS or control group by a noninvolved colleague. The day of birth was considered PD0. At PD2 pups were sexed and culled to ten pups (if possible, pups were culled to five males and five female pups).

Experimental design.
In total, 32 rat dams were tested in two cohorts. In cohort 1 (N = 16) 8 rats dams underwent MS (experimental group) and 8 dams served as control group. All dams in cohort 1 were analyzed for brain and serum alterations at PD21 but did not undergo behavioral testing. In cohort 2 (N = 16) 8 rats dams underwent MS (experimental group) and 8 dams served as control group. All dams in cohort 2 underwent behavioral testing at PD21 and were further analyzed for brain and serum alterations. www.nature.com/scientificreports/ All dams were anesthetized with an intraperitoneal injection of ketamine and xylazine (ratio of 10:1) and sacrificed by decapitation for serum and brain extraction.
To obtain serum probes, trunk blood was collected immediately after decapitation. We allowed the blood to clot for 20 min at room temperature. Thereafter, it was centrifuged at room temperature at 1100 g for 20 min, serum was extracted and stored at − 80 °C. Dissected brains were stored at − 80 °C for later preparation of the hippocampus and mPFC. The estrus cycle was determined according to the protocol of Marcondes and colleagues 51 to control for cycle state influences at PD21. All dams were in the same cycling stage represented by a wide-open vagina with swollen, moist, pink tissue and by mostly nucleated and some cornified epithelial cells in the vaginal smear. This stage might either reflect proestrus 51 or postpartum estrus 52 . Concerning the molecular analysis, the number of included animals constitutes as follows. For serum analysis, two animal samples (one animal of each group) were excluded due to technical errors and one sample per group for glutamate and GABA each. Therewith, the serum of 14 control dams and 15 MS dams was included for glutamate level analysis and serum of 15 control dams and 14 MS dams for GABA level analysis. Moreover, only the animals from cohort 1 were examined for Nr3c1 mRNA levels. For three mPFC samples, the amount of extracted mRNA was not sufficient for Morc1 analysis. Therefore, they were excluded.
The sample size was chosen using G*Power statistical power analysis to ensure adequate power to detect a specific effect 53 .
Maternal separation. MS dams from cohort 1 and 2 (n = 16, experimental group) were separated from their pups for four hours daily from PD2-20 4 . They were placed in a separate adjacent cage (with free access to food and water ad libitum) in the same room with the separated pups. After the separation, both dams and pups were returned to their home cages.
In controls from cohort 1 and 2 (n = 16, control group), the dam and pups were only separated for weighing for a few minutes every 4th day.
Maternal behavior. Maternal behavior assessment was conducted during the dark phase at around 4PM.
Animals could habituate to the experimental room for 20-30 min before testing. On PD 2, 6, 14, and 20 motherpup interactions of separated and control dams were videotaped for 15 min after reunion with pups. Two highresolution SONY cameras were focusing on two sides of the maternal cage (one camera from the front and another from the side in approximately 20 cm distance). Maternal behavior parameters included: nursing (the dam arched-back over pups enabling access to dams' nipples 11 , licking and grooming of pups (either head or anogenital region of pups), retrieval time of the first pup, and time to retrieve the complete litter. Furthermore, measuring time carrying pups from outside to inside the nest and time of regrouping within the nest was conducted. Parameters for self-maintenance comprised time dams spent self-grooming and rearing (with at least one paw off the ground). Maternal behavior was analyzed manually by the experimenter blinded to condition. Besides, all videos were analyzed by an independent party for subjective accuracy.

Ultrasonic vocalizations (USVs).
USVs have proven to reflect rodents' affective state in a non-invasive and accurate way 54 . While positive stimuli elicit shorter 50-kHz frequency ranges (33-90 kHz), negative and aversive stimuli are associated with 22-kHz frequency spans (15-32.9 kHz) 19 . To assess relative changes, each dams' USVs were counted 15 min after reunion, respectively. Real-time recording with Batlogger M (Elekon AG, Lucerne, Switzerland) was positioned rectangular above the cage in approximately 10 cm height on PD 2, 6, 14, and 21 (Fig. 2). The minimum frequency was set at 15 kHz and the maximum frequency at 155 kHz with Volume = 2 and automatic setup trigger. Recorded USVs were analyzed using BatExplorer with FFT size of 1024 and overlap of 80% (Elekon AG, Lucerne, Switzerland). To distinguish between dams' and pups' USVs, nonfrequency-modulated calls were regarded maternal whereas pups demonstrated atypical and U-shaped calls 18 . Only maternal calls were counted.
Anxiety-like behavior. To assess alterations in anxiety-like behavior after MS, EPM, and marble burying 55 were performed on the day of weaning (PD21). The EPM consisted of a plus-shaped maze approximately 50 cm in height. Two open arms (50 cm × 10 cm) and two closed arms (50 cm × 10 cm) faced in opposite directions. In a five-minute trial time, the time spent in the open and closed arms were scored and the frequency of crossings within the open or close spaces was measured (dams started in the enclosed arms). EPM was conducted during the dark phase at around 4PM followed by the marble burying test.
For the marble burying, 20 marbles were positioned symmetrically on a standard Macrolon IV cage on smoothened bedding. On three fourth of the cage area, four rows of five marbles were positioned. After 15 min, the number of marbles buried by dams (100% invisible) was counted. The number of buried marbles serves as an indirect measure for anxiety-like behavior 55 .
Real-time quantitative polymerase chain reaction (rtPCR) of mPFC and hippocampus. The mPFC was dissected according to Watson's The Rat Brain in Stereotaxic Coordinates at bregma 2.20 mm 56 . For the hippocampus, the whole hippocampus was dissected as defined by Watson's rat brain atlas 56 . Isolation of RNA from the hippocampus and mPFC samples were obtained using NucleoSpin TriPrep (Macherey-Nagel, Düren, Germany) with slight modifications. For each sample, 40 μl of RNase-free water was added to obtain RNA. The concentration and quality of RNA were measured using 1 μl of the sample in NanoDrop ND-1000 Spectrophotometer (PEQLAB Biotechnologie, Erlangen, Germany).
To quantify mRNA levels, RNA was first reverse transcribed to cDNA using the High-Capacity RNA-to-cDNA Kit (Thermofisher Scientific, Darmstadt, Germany) according to the manufacturer's protocol. Then, 60 ng of High-performance liquid chromatography (HPLC) for serum GABA and glutamate. HPLC grade acetonitrile was obtained from VWR Chemicals (Langenfeld, Germany). Phtalaldehyd, 2-mercaptoethanol, glutamate, and GABA were ordered from SIGMA (Sigma-Aldrich, Taufkirchen, Germany). Furthermore, 1 M sodium hydroxide (Waldeck, Münster, Germany), methanol (Carl Roth, Karlsruhe, Germany), perchloric acid (Sigma-Aldrich, Steinheim, Germany), phosphor acid (J.T. Baker, Deventer, Netherlands) and 1 M hydrochloric acid and sodium bicarbonate from stock were used. All reagents were of the highest purity available, and solely Milli-Q deionized water was used. Dilution series for GABA and glutamate standards were created for every new mobile phase setup before samples were assessed. For analytic identification and quantification, GABA and glutamate in serum were identified by their characteristic retention times as determined by standard injections of GABA and glutamate. The calibration curve of standard solutions was obtained from a blank, and at least four standard solutions, and the correlation coefficients of the curves were not less than 0.98. For sample preparation, 50 μl serum samples were deproteinized using an equal volume of 0.1 M perchloric acid to extract free amino acids. Serum samples and perchloric acid were centrifuged (15 min, 10,000 rpm, 4 °C) and put on ice. Then, 30 μl of supernatant was loaded into the vials. For the derivatization procedure, O-phtalaldehyde was synthesized. Samples of each serum probe were conducted in duplicates.
The serum glutamate and GABA protocol of Lee 59 was adapted with slight modifications: Two mobile phases with 5% (pH 3.54) and 28% acetonitrile (pH 3.14) were set up separately using phosphoric acid to adjust pH (reverse phase chromatography).
Sample peak areas were measured via Labsolutions integrator system (LCsolution Version 1.24 SP1 Integration Time Program) and compared with the calibration curve standard to quantify respective concentrations. Data analysis parameters for area integration for all probes were determined with the width of 2000s, the slope of 10-30 μV/min, the drift of 0 μV/min, the doubling time (T.DBL) of 1000 min, the minimum areas/height of 1000 counts. Moreover, the integration areas were calculated with horizontal baseline, and negative peaks were rejected.

Statistical analysis.
For maternal care, a second independent rater reexamined all video material. Intraclass correlation is an established statistical method to assess interrater differences of video material in subjective mother-infant relationships 60 . Only parameters showing an excellent inter-rater agreement 61 with r = > 0.75 were used for further analysis. Thus, the parameter carrying of pups (r = 0.59 and Cronbach's Alpha = 0.61) was excluded.
Each maternal care parameter was analyzed separately. For the analysis, 2 treatments (separation or control) × 4 time points (PD2, PD6, PD14, and PD20) were analyzed in a two-way, repeated measures Analysis of Variance (ANOVA) for all parameters 36 and significance was set at p < 0.00625 to control for False Discovery Rate (FDR) when analyzing multiple comparisons (P-value < 0.05 divided by 2 treatments with 4 time points). Only treatment effects, but no time point or treatment × time point interaction effects were found, and thus group differences were further analyzed by student's t-test 62 . Analogous to maternal care videos, USVs were analyzed separately for the two frequency ranges using the repeated measures ANOVA applying FDR, followed by student's t-test. Anxiety behavior was also assessed using student's t-test.
Data were correlated using Pearson's correlation coefficient. Brain and serum data were analyzed using a One-Way ANOVA with group as an independent variable (separation or control) and GABA and glutamate serum concentrations, delta CT values of Morc1 and Nr3c1 as dependent variables. Statistical analysis was performed using SPSS (IBM SPSS Statistics 25). Significance levels were set at p-value < 0.05.