1
Validation of the LBN paradigm in rats.
Investigation of its effects on maternal care, sonic
calls and offspring’s neuronal morphology of the
amygdala.
Abstract
Early life adversity (ELA) has been implied to cause hormonal, functional and behavioral changes in responses to stress. Across species, maternal behavior is considered critical for the development of the infants and the formation of the brain. Previous literature has linked changes in maternal care with effects on the offspring. In the current study we validated the LBN paradigm - that focuses on providing the litter with an impoverished environment as a way of producing stress - by revealing alterations in the body, adrenal and thymus weights and studied its effects on maternal behavior and emission of audible vocalizations in rats. Interestingly, it was revealed that the LBN paradigm caused the dams to provide more attention and care to the pups, engaging more with pup-directed behaviors, however the behavior of the LBN dams was significantly more entropic. The pups admitted to the LBN paradigm emitted more sonic calls which could be a parameter affecting the higher pup-directed behavior of the dams. A brain region highly involved in stress responses is the amygdala, thus the morphology of the neurons was studied in the BLA of P9 rats. The male P9 amygdaloid neurons under study did not reveal any differences in the size of the soma and the length of basal and apical dendrites nor in the complexity of arborization between LBN and controls. However, it is possible that the whole sample size should be first collected in order to be able to draw conclusions. Otherwise, the effect stress has on the neurons of the BLA might be time dependent and might be expressed on later stages of the development of the rat brain.
2
Introduction
A widely accepted theory is that the behavior and the physiology of an organism is formed by getting information from both nature (the genome) and nurture (the experiences one acquires throughout life and the environment in which this organism is developed). During the first weeks of life in rodents the dam-pup interaction is considered to be of high importance as it affects the homeostatic mechanisms of the pups. Furthermore, perturbation of this relationship affects the neuroendocrinology and the function of the brain as well as the behavior of the pups later in life due to the fact that the brain is still developing in these early stages of life (Brunson et al., 2005; Laloux et al., 2012; Maccari et al., 2014). Early life adversity can have a strong effect on the development of an organism and thus its brain and can influence the brain activity and behavior of the animals. More specifically such experiences can cause hormonal, functional and behavioral changes to responses to stress. Stress is an individual’s biological and psychological response to a stressor (Papaioannou et al., 2002; Zimmerberg, Rosenthal and Stark, 2003; Maccari et al., 2014; Kim et al., 2017). Long term effects of such experiences have been researched for many years in both humans and animals. During the early stages of life, the maternal care that an individual receives can affect the stress system. In rodents maternal care, especially in the very early stages of life, can be of high importance for the emotional behavior these animals will develop in adult life as well as for the reactivity to stress. When the maternal care provided to the pups is unaltered there is high possibility these animals will develop a stabilized emotional behavior later in life, while when maternal care is deteriorated it can act as a stressor that leads to the expression of abnormal social and emotional behaviors (McEwen, 2008). High licking-grooming mothers, mothers that spend a lot of time licking and grooming their pups and thus taking good care of them, are characterized as high caring mothers and can lead to the development of neophilic pups that show less emotional reactivity, while low licking-grooming mothers, low care providing mothers, create neophobic pups that show high responsiveness of the Hypothalamic-Pituitary-Adrenal (HPA) axis under stressful situations(Meaney and Aitken, 1985; Negrão et al., 2000; Weaver et al., 2004; Lupien et al., 2009).
There are two main systems whose performance is included in stress responses; the Autonomic Nervous System (ANS) and the Hypothalamic-Pituitary-Adrenal (HPA) axis. The latter in stressful situations leads to the production and release of glucocorticoids in the circulation and especially of corticosterone, also referred to as the stress hormone. Corticosterone acts throughout the HPA axis and regulates the levels of stress by preparing the organism to respond to a stressor but also by terminating the response when the stressor is no longer present. More specifically, the levels of corticosterone in rodents and cortisol in humans can terminate the stress response by acting though inhibitory feedback loops throughout different levels of the HPA axis and brain regions. The brain regions that are triggered because of this response are essential as they can affect the response per se. For instance, the activation of the hippocampus downregulates the cascade while the activation of the amygdala acts as an extra trigger to the reactivity of the HPA axis to the stressor. All these activations occur due to the binding of the glucocorticoids to their receptors that are abundant throughout the axis. Glucocorticoids are of high importance for the maturation of the brain as they can influence the construction of the axons and the dendrites of the neurons which in the long run can affect the natural development and function of the brain. As a result, glucocorticoids can have a long-term effect on the brain regions that regulate their production and secretion (Negrão et al., 2000; Lupien et al., 2009).
3
There are many different paradigms used to elucidate stress in rodents at the early stages of life. Some of them are aiming to lead to stress via the manipulation of the environment. One paradigm that is widely used in the production of stress in rodents is the Limited Bedding and Nesting material (LBN) paradigm. In this paradigm the animals are exposed to an impoverished environment that prevents the dam from building a satisfactory nest for the pups, which subsequently becomes a stressful developing environment for the litter (Gilles, Schultz and Baram, 1996; Ivy et al., 2008; Rice et al., 2008; Maccari et al., 2014; Molet et al., 2014; Chen and Baram, 2016; Walker et al., 2017). The validation of this model is based on changes in the size of the organs and the secretion of the molecules participating in the HPA axis function. Thus, various studies have reported an increase in the peripheral corticosterone levels as well as in the weight of the adrenals (responsible for the production of corticosterone), while the thymus and body weight show a diminution (Kioukia-Fougia et al., 2002; Avishai-Eliner et al., 2008; Ivy et al., 2008; Molet et al., 2014; Derks et al., 2016; Walker et al., 2017). In this study we used the LBN model for the induction of stress in early life stages in rats and for this reason some experiments aimed in the validation of this. The interaction between the dam and the pups begins when the dam approaches the pups and begins to take care of them and to nurse them, during the period of the construction of the nest (Stern, 1997). There are different ways for a dam to nurse her pups. The posture that allows the pups to have the best possible access to the nipples is called arched back nursing in which the dam stands in a way that resembles to an arch (also referred to as high kyphosis). Other postures are the low nursing that resembles the arched back with the only difference that the dams are not standing strongly on their feet; and the side/passive nursing in which the dam is laying on her side or back providing little nipple access to the pups (Myers et al., 1989; Stern, 1997).
The LBN model is able to cause alterations in the quantity and quality of the maternal care provided to the pups. In most of the studies that have been conducted, the dams of this model show abnormal behaviors, such as less licking and grooming of the pups, changes in the time spent in off-nest behaviors as well as the frequency in which the dam abandons the nest and the type of nursingthey chose as the more preferred (Liu et al., 1997; Zimmerberg, Rosenthal and Stark, 2003; Ivy et al., 2008; Guadagno, Wong and Walker, 2018; Gallo et al., 2019). However, the results in most of the studies using the LBN paradigm are contradictory, thus more study needs to be done in order to be able to report a directionality of the change in the behavior.
Stress can have an impact in various brain regions both on a molecular and a morphological-physiological level. One of these brain regions is well known to be the amygdala as it is a region rich in both glucocorticoid (GRs) and mineralocorticoid receptors (MRs). Both GRs and MRs are receptors showing strong affinity for glucocorticoids including corticosterone. It is highly reported in literature that these receptors show changes in their amounts in this particular brain region and mostly in the basolateral amygdala (BLA) and the Central amygdala (CeA) as an outcome of stress induction (Caudal, Jay and Godsil, 2014; Han, Ding and Shi, 2014; Arnett et al., 2015; Prusator and Greenwood-Van Meerveld, 2015). Furthermore, early life experiences seem to affect the development of amygdala connectivity to other brain regions and the emotional outbursts that follow the stress response seem to emerge from amygdala’s response. The function of the HPA axis affects the development of the amygdala because the glucocorticoids that are produced in stressful situations have an impact on amygdala’s reactivity and for that reason enhanced amygdala reactivity might be considered
4
a neural marker of ELS (VanTieghem and Tottenham, 2018). Apart from the molecular level, the amygdala is further affected by stress also on the physiological level. Morphological and functional differences which are dependent from sex have been detected after the application of stress inducing paradigms in amygdala studies in the past (Guadagno, Wong and Walker, 2018). These differences include dendritic hypertrophy in the pyramidal and stellate neurons, increased spine density, increased dendritic arborization at the end of the LBN for both early stages in life but also later on in adulthood compared to control animals (Guadagno, Wong and Walker, 2018; Leem, Yoon and Jo, 2020). However, there is contradictory data in the literature as there is data indicating that the spine density of the BLA neurons remains unaffected by ELA while arborization seems to be enhanced but in early stages of life it might remain unaltered and the effect might be time dependent (Guadagno, Wong and Walker, 2018; Patel et al., 2018). For this reason, it is important to study the effects ELA can have in the morphology of this region, as it is the main region generating fear which is a response highly affected by stress later in life and a brain region highly studied (Zhang and Rosenkranz, 2013).
Rodents have the ability to communicate with each other via the emission of ultrasonic vocalizations (USVs) and sonic vocalizations. In this attempt they produce various different types of vocalizations that are combined in many different ways and it is believed to show a pattern for each type of context. In mice pups the USVs have frequencies ranging from 30-90 kHz while in rat pups the range is from 30-50 kHz. These signals deriving from the pups are believed to affect the behavior of the dam but there is contradictory evidence on the way it is affected (Ehret and Bernecker, 1986; Branchi, Santucci and Alleva, 2006; Heun-Johnson and Levitt, 2016). These vocalizations are produced when the pups are in an awake and active state. The USVs show an ontogenic profile. More specifically, they are abundant during the first post-natal days (P) and in rats they reach their peak on day 10 (Branchi, Santucci and Alleva, 2006). The vocalizations can be affected by the application of paradigms and conditions to the animals leading to an altered behavior of the dam towards her pups (Zimmerberg, Rosenthal and Stark, 2003). There is evidence that stressful early life experiences can cause alterations in the USVs emitted by the pups. The quality and quantity of the emitted squeaks also differs between different strains of animals, with the Sprague-Dawley being the ones that produce the least when compared to other strains, however little is known about the vocalizations produced during the very first days of life (Schwarting and Wöhr, 2018).
In the present study, data concerning the validation of the LBN model that was applied to pups from P2-8 were gathered. During these days the maternal behavior, that was distinct in 12 different units, was scored and analyzed thoroughly in order to determine alterations caused by the provoked stress during the application of the paradigm regarding the duration of the behavior, the consistency as well as the probability of one behavior to lead to another. Simultaneously, the cages were recorded with the aim to analyze the USVs and the sonic vocalizations emitted by the pups. Changes caused by ELA in the morphology of the neurons in the amygdala, and more specifically in the BLA were also examined.
5
Materials and Methods
Animals
Sprague Dawley rats were used in this study in two different batches. In the first batch male adult rats were put in the same cage with two female primiparous adult rats. Once pregnancy was evident the animals were separated, and the female rats were put in new cages alone. The dams of the first batch did not give birth on the same day and for this reason the pups from different litters were not gathered and randomly assigned to the dams but on the contrary they remained with their actual mother throughout the procedure. The other batch of animals consisted of time pregnant animals that were ordered from Janvier and it was considered a better version of the first batch in order to avoid the obstacles faced on the first one (the dams of the first batch did not give birth on the same day and the pups were not mixed). On the 21st
day of their pregnancy all the dams of the second batch (time-pregnant dams) gave birth with a few hours difference from one delivery to another. The date of birth is declared P0. On P2 the pups were mixed and each dam received randomly 10 pups to take care of (1:1 male:female ratio in order to avoid a genetic effect and to standardize the size of the nest. It was our goal to have as similar groups as possible so that differences in groups will not consist a parameter in our experiments). Each cage consisted of 1 dam and 10 pups. The rats were reared in the IWO (Institute Wetenschappen Optralmologie) SPF animal facility, Netherlands Institute for Neuroscience, Amsterdam, they were kept at 22-24 ᵒC, humidity 50-65% and 12 hour reversed light/dark cycle with ad libitum access to chow and water. All experimental procedures were pre-approved by the Centrale Commissie Dierproeven of the Netherlands (AVD801002016695) and by the welfare body of the Netherlands institute for Neuroscience (IVD, protocol number NIN18.16.02)
Animal manipulations
The experimental paradigm used in this study was the LBN. In this paradigm the experimental group, which contains both the dam and the pups she is assigned with, are reared in an impoverished environment in the cage. This impoverished environment contains half paper towel and bedding material that is not capable of covering the whole floor of the cage. The floor also is covered by an aluminium mesh platform placed at ~2,5 cm above the floor. Routine plastic rat cages were used for both the experimental and the control groups. The control group consisted of the dams and the pups that were reared in cages with the standard amount of bedding and one entire paper towel as nesting material (as suggested by Molet (2014)). All the cages of the second batch of animals were kept in a quiet room that contained only the animals used in this study, while the cages of the first batch were in the same room with other experimental animals used in different studies. All the animals in the cages remained undisturbed from P2 to P8. In this time period (P2-8) maternal care and USV recordings were being obtained. In order for the dams from both batches to get acclimatized and habituated to the equipment used for the recording of the squeaks (microphones) and the behaviors (infrared cameras) the set up was present in the stable before the dams gave birth, so that our results would not be biased by such a condition. In this study there were used 8 dams and 80 pups (full sample size). In later analyses and experiments the number of animals used differed.
6
Maternal Behavior
From P2 to P8 the maternal behaviour was being recorded with the help of infrared video cameras that were installed in the room were the animals were reared. The maternal behaviour was then scored with BORIS digital software (Friard and Gamba, 2016). The behaviours that were included in this study were: the licking and grooming of the pups, the type of nursing provided to the pups (arched back, low nursing and side/passive nursing), moving around the nest as well as outside of the nest, building of the nest, carrying the pups, rearing, eating, self-grooming and tail-chasing. These behaviours were being tracked 3 times per day in a reversed light/dark cycle environment. The first recording was taking place at 10:00, the second at 15:00 and the last one at 20:00. In this way, we managed to have 2 recordings throughout the dark cycle when the rodents are more active as they are nocturnal animals and 1 in the light cycle when they are less energetic.
Mathematical analysis of the Maternal Behaviour
Data were collected both in dark cycle and light cycle for 8 cohorts. The first 4 cohorts were mated and breed in our labs, the rest 4 consisted of time pregnant females that arrived in our lab on their 13th day of gestation. Maternal behaviors observed in these animals were
compared between the control and the LBN group and were furthermore analyzed depending on the cycle and the directionality (towards the pups or towards the self). Moreover, we compared the mean of the duration of one single episode of one behavior in order to assess how fragmented each behavior is in each condition; and we estimated the predictability of the dam’s transition from one type of behavior to another by calculating the Shannon entropy index (with the methods suggested in Molet (2016)) and producing heat maps in Python programming software.
Measurement of Sonic Vocalizations
For the detection of the audible vocalizations emitted from the pups we recorded the sounds produced in 2 control and 2 LBN cages out of the total 8 cages. For this reason, 2 microphones were placed above these cages. For the detection of the squeaks the Matlab toolbox Deepsqueak(Kevin R Coffey (2020).DeepSqueak (https://github.com/DrCoffey/DeepSqueak), GitHub. Retrieved July 9, 2020) was used from which we used the long rat USV detection network. The automatic detection provided by DeepSqueak itself did not seem sufficiently reliable and for this reason our lab developed a new script aiming to train DeepSqueak with some examples of squeaks that were manually detected by the experimenters. The neuronal network aimed to achieve an automatic way to detect the squeaks sufficiently and reliably pick them. The new script was tested and the squeaks it picked were compared to squeaks that were manually detected. Once the system was trained by the experimenters to detect automatically the squeaks, it was used as a more accurate mean of automatization. For avoiding the emissions of sonic calls that could be emerging from a different cage than the one being recorded we developed another script that could distinct which squeaks were loud enough to be coming from the cage of interest and which not, helping us produce less biased results. The squeaks that we analyzed were deriving from recordings that were created from P2 to P8 thrice a day and more specifically at the same sessions that the scoring of the maternal care was taking place (10:00, 15:00 and 20:00). The recordings were acquired with the help of Avisoft recording equipment.
7
On P9 each animal that was led to get sacrificed was put in a smaller cage in order to get moved from the stable to the biotechnicum room. Then individually they were moved to the biotechnicum room were their weight was calculated with the help of a digital scale. Afterwards, the animals were sacrificed. The animals of the first batch were anesthetized in ice cold water and then they were decapitated while the animals of the second batch were sacrificed via instant decapitation without the administration of anaesthesia and their brains, adrenals, thymuses as well as 1 ml of blood were isolated. The previously mentioned organs were weighed. The brains were weighed only for the second batch of animals. In addition, blood samples were only collected from the second batch of animals. All the dissections were kept in saline. The adrenals were detected with the use of the stereoscope and were carefully removed. The fat around the adrenals was removed as accurately as possible but in doing so in some samples the capsules got detached. For that reason, the weights from the adrenals without the capsule were not included in our dataset during the analysis.
Blood processing
The blood that was collected from the pups during the sacrifices was kept in ice-cold EDTA-coated tubes (Sarstedt, Etten- leur, The Netherlands). These tubes were kept in ice until the all the sacrifices were completed. The tubes were then centrifuged at 13,000 rpm for 15 minutes and the plasma that was distinguished after the centrifugation was isolated in new, clean Eppendorf tubes and stored at -20 °C refrigerator until the ELISA procedure took place. Elisa
After the plasma is isolated, an ELISA assay takes place in order to calculate the corticosterone levels in the peripheral blood of the P9 pups (only of the second batch). For this reason, the IBL Corticosterone ELISA Kit is used (sensitivity< 1.631 nmoI/L ). Since the experimental pups were 20, in the holder 20 microtiter wells were occupied for the samples (some samples were run in duplo to verify the consistency of our results and to reject the possibility of having accidentally performed a mistake during the procedure). On each well 20 μL were dispensed of each standard, control as well as some samples. Then 200 μL of Enzyme Conjugate are added in each of the wells in use and then we mix them thoroughly for 10 seconds and leave them for one hour in room temperature. After one hour the wells are rinsed thrice with diluted Wash Solution. Each well receives 400 μL of the above-mentioned solution. Afterwards, 100 μL of Substrate Solution is added to every used well and then it is left to incubate for 15 minutes at room temperature so that the enzymatic reaction will be able to take place. To stop the reaction, 50 μL of Stop Solution is added on each well. Immediately after the end of the reaction information about the OD has been acquired with the use of a microtiter plate reader. In order to be able to evaluate the OD information, the standard curve of the experiment must be designed where the mean absorbance from each standard solution is plotted (standards contained non-mercury preservative). We used competitive ELISA; thus, the OD levels are inversely proportional to the levels of corticosterone.
8 Golgi Cox
Preparation of the Golgi-Cox solution
The Golgi-Cox solution consists of four main components: 5% Potassium bichromate dissolved in warm Aqua Dest (A.D.), 5% Mercuric chloride dissolved in warm A.D., 5% Potassium chromate dissolved in A.D. (room temperature) and A.D. Two solutions were needed. The first one contained 5 volumes of Potassium bichromate mixed with 5 volumes of mercuric chloride solutions as mentioned previously, dissolved in warm A.D. The second solution contained 4 volumes of Potassium chromate mixed with 10 volumes of A.D. Once these solutions were produced, the first one was transferred slowly in the second one while being stirred. The final solution deriving from this last mixture was stored in a dark bottle for at least 5 days and was kept away from the sunlight, as it is photosensitive. After 5 days a sediment appeared in the bottom of the bottle. The solution was transferred to a new, clean dark bottle without the sediment which in this case was mercurichromate.
Embedding
The Golgi Cox experiment per se consists of three distinct parts: the embedding, the sectioning and the staining. The right hemispheres of the animals were submitted to this procedure. Once the hemispheres were isolated from the P9 rats, they were incubated in the Golgi-Cox solution immediately. There they remained for fourteen days and after that period a cascade of solution changing took place. More specifically, on day fourteen the brains were rinsed with A.D. four times for 5 minutes each time and then they were dehydrated with 70% ethanol where they remained overnight. After 24 hours the brains were dehydrated with 96% ethanol overnight. The next day the brains were dehydrated in absolute ethanol for 8 hours and then they were left overnight in ethanol/ether 1:2 solution (16 hours incubation). The brains got stored in celloidin 4% and 6%, each time for 24 hours and then in 12% where they remained for 48 hours. The brains were then transferred in paper boxes which got immersed in chloroform for 16 hours (overnight) sharp and then were kept in 70% ethanol in fridge in 4ᵒC.
Sectioning
Brains that follow the Golgi-Cox procedure need to be sectioned with the use of a vibratome. The sections were 200 μm thick. The extra celloidin around the brain was removed, however the brain is still covered with celloidin and has now gained a pyramidal shape. It was then fixed in the vibratome with a 15 degrees angle towards the knife (blade) and immersed in 70% ethanol. The slices that were produced were then kept in 70% ethanol in a well plate and were covered in order to avoid exposure to direct light.
Staining
Once the brain slices were produced they were introduced immediately to the staining procedure. All the steps of the staining procedure took place underneath a fume hood. First step is to rinse the brain slices with A.D. for 5 minutes. Then the slices are moved in 16% ammonia solution for 30 minutes, which helps the Golgi-Cox impregnation to develop covered with a foil to avoid exposure to direct light. This precaution concerning the exposure to light was applied in all the steps. Then the slices are washed in A.D. for 2 minutes and then for 7
9
minutes in 1% sodium thiosulfate, which fixes the staining in the slices. After that, the slices are washed twice in A.D. for 5 minutes each time. Afterwards, the slices are dehydrated in 70% and then in 96% ethanol, each time for 5 minutes. The slices are then washed twice for 10 minutes in absolute ethanol (100%). Last step of the staining procedure is to incubate the slices twice for 10 minutes in Xylene. After the staining procedure reaches its end, the slices are mounted in glass slides by using Entellan and are then covered by glass cover slips. The Golgi stained brain slices were kept in the dark and in room temperature.
Once the Golgi procedure was completed, the staining was observed with the help of a microscope. The level of the BLA was determined with the help of the P9 rat brain atlas (Atlas of the developing rat brain in stereotaxic coordinates P9 by Khazipov, Zaynutdinova, Ogievetsky, Mitrukhina, Manent and Represa).
Image acquisition and Sholl Analysis
Digital images were obtained with a camera that was connected to the
Nikon H600L
microscope by which the morphology of the neurons became visible. With magnification set to 40x three to nine neurons were analyzed per animal depending on the clarity of the neurons and the capability to fully distinct each one of them, appearing to be isolated the one neuron from the other.
The neurons were manually traced with the Sholl analysis through the ImageJ (Fiji) (software Schindelin J, Arganda-Carreras I, Frise E, et al. Fiji: an open-source platform for biological-image analysis. Nat Methods. 2012;9(7):676-682. Published 2012 Jun 28. doi:10.1038/nmeth.2019). The plug in used to do the tracing of the neuron is the Simple Neurite Tracer plug-in. The step size of the concentric circles used in these Sholl analyses is 15 microns as defined in previous studies but also in an internal pilot study of our group. While doing the tracing of the neurons and the Sholl analysis the experimenter was blind to the paradigm that was applied to the experimental animal whose brain was being studied. The final sample size of the neurons equals 70 from 12 different subjects belonging to the second batch of animals (6 control and 6 LBN). 36 neurons belonged to the control group and 34 to the LBN group. All the brains used in this procedure derived from male subjects.
Statistical Analysis
The collected data were analyzed for normality and homoscedasticity. When the data were normally distributed T-tests and ANOVAS were performed otherwise non-parametric Mann-Whitney U-tests were used. Statistical analysis took place in JASP computer software (JASP Team (2020). JASP (Version 0.12.2) [Computer software]). For all the results, Bayesian statistics were applied to define the strength of significance or insignificance. Bayesian statistics were used in order to define whether the non-significance of the effect was deriving from an absence of evidence or it consisted an evidence of absence of the effect. According to Keysers, Gazzola and Wagenmakers (2020) when the Bayes Factor (BF10) is more than 10
this indicates that we have strong evidence that the effect exists, when 3<BF10<10 there is
moderate evidence for the effect, when 1/3<BF10<3 the evidence is insufficient to draw
conclusions, while when BF10<1/3 there is moderate evidence for absence of an effect. The
first batch of animals was used as a pilot study and the second one was mostly considered the experimental one. For this reason, for the first batch two-tailed T-tests were run, but once
10
the directionality of the effect was determined, on the second batch of animals one-tailed T-tests were used according to the directionality that appeared in the first-pilot study. All graphs were produced in R-studio programming software (Reference: R Core Team (2013). R: A language and environment for statistical;computing. R Foundation for Statistical Computing, Vienna, Austria. URL http://www.R-project.org/.), with only one exception; the heat maps depicting the entropy were produced in Python (Van Rossum, G., & Drake Jr, F. L. (1995). Python reference manual. Centrum voor Wiskunde en Informatica Amsterdam).
Results
In the current study, the Limited Bedding and Nesting Material protocol was applied to newborn rats in order to induce stress at the early stages of life. For this reason, the maternal behavior was scored in order to observe whether the paradigm affects the care provided by the mother to the pups. The induction of stress was validated by measuring the peripheral corticosterone levels as well as observing an effect in the development of important organs participating in the HPA axis response to stress such as the brain, the adrenals and the thymus. Once the paradigm was validated, it was of high importance to study whether there was an effect on the morphology of the neurons in the basolateral amygdala (BLA) which is a region highly affected by stress as prementioned in the introduction. For this reason, brain slices were produced with the use of the vibratome, stained with the Golgi staining technique and observed in the microscope from the -3.00mm to -4.00mm bregma level (the levels are corresponding to the P9 rat brain atlas). The morphology of the neurons was analyzed with the help of ImageJ (Fiji) software and the Sholl analysis. From these two we were able to gain information on the complexity of the arborizations and the length of the soma and the dendrites, apical and basal.
Analysis of the maternal care
In the current study the maternal behavior was scored from P2 to P8 thrice a day. Here the two batches of animals are both used for the analysis. There were 4 dams from the first batch and 4 dams from the second batch adding up to a total of 8 dams used for the scoring of the maternal behavior. All the 12 behaviors that were being scored [licking/grooming (LG), arched back (AN), side/passive nursing (SN), low nursing (LN), moving off nest (O), moving around the nest (M), building of the nest (N), carrying the pups (C), rearing (R), self-grooming (S), tail-chasing (T) and eating (E)]. The behaviors were generalized and further categorized in pup-directed, non-pup-directed as well as maternal associated (these are behaviors that appear during the peri-partum period and during lactation and in our case, we included tail-chasing, moving around the nest and building of the nest). As shown in Table 1, most of the behaviors showed a statistically significant difference between the control and the LBN group. The only behaviors that did not differ significantly were arched-back nursing, side/passive nursing, self-grooming and rearing. These behaviors also show a low Bayesian Factor (BF10<0.2 or BF10<1), indicating that there is either moderate evidence or anecdotal evidence
of absence of the effect. What is more, the pup-directed (p-value=0.001) and non-pup-directed (p-value=0.001) behaviors differ significantly between the two groups. The Bayesian factor for the pup-directed effect implies that there is moderate evidence supporting this result (BF10=7.097), while the non-pup-directed effect shows a BF a little higher than 10
11
Table 1 two tailed Mann_whitney U tests Table 2 Bayesian Mann-Whitney U test
d) *** e) *** f) *** a) ***
Figure 1 Plots of the maternal behavior clustered in pup-directed and non-pup-directed for both batches. a) Total duration of the
pup-directed behaviors, comparison between control and LBN group. b) Total duration of the non-pup-directed behaviors. c) Duration of the pup-directed behaviors from P2-P8. d) Total duration of the non-pup-directed behaviors from P2-P8. e) Total duration of the pup-directed behaviors in the 3 different sessions. f) Total duration of the non-pup-directed behaviors in the 3 different sessions. Asterisks indicate significance of the effect.
c)
**
b)
12
As shown in Figure1(a,b) the LBN group shows higher pup-directed behaviors and lower non-pup-directed behaviors. After observing the qqplots that were produced in JASP, with the help of ANOVAs it was revealed that the duration of these behaviors changes significantly throughout the days (main0020effect of the day pup-directed/day p-value=0.013, between P2 & P7 PTukey=0.003 ; non-pup-directed/day p-value<0.001, between P2 & P7 and P2 &P8 and
P3&P7 and P3&P8 and P4&P7 PTukey<0.05), more specifically as the days pass by the duration
of the behaviors that are significant in each group increase respectively (Figure 1(c,d))(Table 3). The non-pup-directed behaviors show a significant interaction between the group and the postnatal day (p-value=0.013, Tukey post-hoc test of group*day interaction, PTukey<0.001 from
comparing). Since the behaviors vary with the passing of the days, it was of high interest to check whether a similar effect is encountered throughout the sessions that the behaviors were being scored (10:00, 15:00 and 20:00). Indeed, after running the statistical analysis it was revealed that both pup-directed and non-pup-directed behaviors were affected significantly by the time of the day that the behaviors were taking place (pup-directed/session p-value<0.001, between all sessions PTukey<0.01; non-pup-directed/session p-value<0.001, between all
sessions PTukey<0.01). Once again, the non-pup-directed behaviors showed a significant
interaction between the group and the time that the behaviors were occurring (non-pup-directed Group*Time p-value=0.049,post-hoc analysis of Group*Time interaction, PTukey<0.001 from comparing). The non-pup-directed behaviors seem to be preferred mostly
during the dark phase (10:00 & 15:00), while the pup-directed behaviors are at their maximal levels during light phase (last session, 20:00) (Figure1(e,f)).
Single behaviors of the dams were further analyzed. The licking/grooming showed a significant difference between the two groups, with the LBN group always dedicating more time in the licking/grooming of the pups compared to the Control dams (p-value< 0.001). However, this pattern was not affected by the session or the postnatal day. Moreover, the LBN dams were less prone to leave the nest and engaged less with off-nest behaviors when compared to the Control dams value<0.001). The off-nest behavior is affected by the session
(p-value=0.003) and the postnatal day (p-value<0.001). Off-nest behavior is increasing
significantly as the postnatal days pass by (main effect of the day PTukey<0.05 between P2&P7
and P2&P8 and P3&P7 and P4&P7) and during the dark phase (PTukey<0.03 between Time
10&20 and 15&20). The most prominent nursing style mentioned in literature is low nursing. In our case there was a significant difference in the duration the LBN group associated with the low nursing behavior when compared with the control group, with the LBN group dedicating more time performing the low nursing behavior. Once again, this behavior seems to be significantly affected by the session (p-value<0.001) and the postnatal day (p-value<0.001). During the dark phase and after P5 the LBN group increases drastically the engagement with the low nursing behavior compared to the control group. The interaction between the group and session shows significance (p-value<0.001; interaction at 10:00 between groups, PTukey<0.005). Lastly, one pup-directed behavior of interest that was further analyzed was the
carrying of the pups, which was significantly higher in the LBN group, from P4-6 and mostly during the dark phase (significant difference between days, p-value=0.002; interaction between group*day, p-value=0.002, PTukey<0.001 for LBN P2 compared to all the rest of the
days on both the same group and the control; significant difference between sessions,
p-value=0.032).
The mean duration of individual bouts of a single licking/grooming session was significantly shorter in the LBN group compared to the control group (p-value<0.001). This indicates that
13
this behavior was terminated and initialized again more frequently in the LBN group, thus the LBN showed longer LG sessions that were more fragmented (Figure 3).
A ) B) C) D) F) *** *** E) G) H) *** *** *** *** *** ***
14
Figure 2 Maternal behaviors associated by the dam in total, throughout the days and throughout the three different
times of recordings. A) Duration of licking and grooming. B) Duration of off-nest behavior. C) Duration of low nursing. D) Duration of carrying the pups. E) Duration of off-nest behavior from P2-P8. F) Duration of off-nest behavior throughout the sessions. G) Duration of licking and grooming from P2-P8. H) Duration of licking and grooming throughout the sessions. I) Duration of carrying the pups from P2-P8. H) Duration of carrying the pups grooming throughout the sessions.. When p-value<0.001, use of three asterisks high in the graphs to indicate significance. The asterisks next to the title indicate a significance of the individual parameter and not the specific day or time of the interaction.
I)
*** ***
A) B)
C)
D)
Figure 3 Mean duration of individual licking/grooming bouts. A) Mean duration of individual licking/grooming bouts was significantly
shorter in the LBN group, leading to a highly fragmented care provided to the pups. B) Statistical analysis of the Mean duration of individual licking/grooming bouts per group with the use of a t-test. C) Calculation of the Bayesian factor for the Mean duration of individual licking/grooming bouts. D) Analysis with the use of classical ANOVA aiming to reveal possible interaction between the group of animals and the postnatal day. Asterisk indicate significance.
15
ANOVA
Batch Analysis Pup Directed Non-Pup Directed
p-value BF10 p-value BF10 1 Group 0.008 <0.001 Day 0.06 0.998 0.022 1 interaction 0.346 0.327 0.103 0.902 Group 0.003 <0.001 Time <0.001 0.291 <0.001 1 Interaction 0.443 0.077 0.165 0.614 2 Group 0.022 0.004 Day 0.109 1.301 <0.001 1 interaction 0.554 5.749e-4 0.149 0.682 Group 0.01 0.003 Time <0.001 2.824 <0.001 1 Interaction 0.444 2.49 0.177 0.584 Both Group <0.001 <0.001 Day <0.001 1 <0.001 1 interaction 0.407 0.229 0.049 0.212 Group <0.001 <0.001 Time 0.013 0.276 <0.001 0.873 Interaction 0.295 1 0.013 1
Table 3 ANOVA analysis for the first and second batch of animals regarding the pup-directed and the
non-pup-directed behaviors. The first BF refers to Group+Day or Group+Time respectively, while the second one indicates Group+Time+Group*Time and respectively for the day.
After scoring the maternal care with BORIS software, apart from analyzing each behavior per se, we were able to analyze the quality of the maternal care and create heat maps in Python indicating the entropy. Entropy declares the predictability of the transition from one behavior to another.
The following heat maps graphically represent the predictability for transitioning from one behavior to another. The probability of a behavior to follow another is depicted along a color scale. The bluer the color the more unpredictable the transition, while the redder the color the more predictable the transition. The transitions between the two groups show differences between them, indicating that the paradigm affects the predictability of transitioning to another behavior and the behavior per se. What is more, the Control group seems to have either high or low probabilities of the behavioral sequences, while the LBN group shows mostly a high or mild unpredictability (Figure 4).
16
Table 4 Statistical analysis with the use of Classical ANOVA and Bayesian statistics. A) Calculation of the
significance of day, time and condition in the entropy rates as well as the interaction between day and time, day and condition, time and condition and day, time and condition together. B) Calculation of the bays factor of the entropy grouped by group.
CTR
LBN
Figure 4 Ηeat maps. On the left, heat map for the control group. On the right heat map for the experimental LBN group representative of one single observation. Blue color is close to 0 indicating high unpredictability. Red color close to 1 indicating high predictability. Green color indicates moderate mid-rate of predictability (as much predictable as unpredictable). All measurements apply to the respective dams of each group.
A)
17
After performing the statistical analysis, it was revealed that some of the parameters under examination were significantly affecting the outcome. The average entropy rate for the overall maternal behavior was calculated and as depicted in Figure 5 the index was significantly higher for the LBN group indicating that the maternal care of dams in the LBN group are characterized by a higher unpredictability (p-value<0.001). The BF for the entropy index is very high indicating extremely strong evidence of the effect (BF10= 21196.102 as shown in
table 4). What is more, a significant change in the entropy rate appeared to be caused by the change of the sessions (p-value<0.001). The dams appeared to have a higher entropic behavior during the dark phase (10:00 and 15:00) and a less entropic routine during the light phase (20:00) in both groups (Table 4; Figure 6).
Figure 6 Graphical representation of the difference between the LBN and control group throughout the three
different time sessions (10:00, 15:00 and 20:00) from P2-8.
Figure 5 Comparison of the Shannon entropy index between the CTR and LBN group. The LBN entropy rates are
significantly higher on average that the CTR ones. Asterisk indicates significance of the effect.
18 Sonic calls
To gain more information on the effect that the LBN paradigm has on pups and dam we analyzed the sonic vocalizations.. First, the sonic calls for the first batch of animals were analyzed with the use of two-tailed T-tests. This priority aimed to define the directionality of the difference between the two groups so that in the second batch of animals a one tail T-test would be used. When comparing the number of sonic calls detected in the first batch, the LBN group shows a significantly augmented number when compared to the control group (p-value < 0.001; Figure 7A). The BF is also high (BF10=25.879) indicating strong evidence of the
existence of the effect (Table 5C). The emission of the squeaks when running the statistics seems to be unaffected by the session and the postnatal day (the session has a BF=0.003 and the postnatal day BF=0.003 indicating absence of the effect). However, when observing the graphs occurring from the analyses, there is a clear trend in the emission of the squeaks, with the first days being more abundant in squeaks (all the Bays Factors for the days are higher than 0.33 indicating anecdotal evidence of absence meaning that with a higher sample size these results could change) (Figure 7B). With the use of ANOVAs, a significant interaction between the group and the session was revealed (interaction Group*Session, p-value=0.029; comparisons between CTR,10&LBN,10; CTR,15&LBN,15 show PTukey<0.05). The LBN group
does not show a change in the pattern of emitting the calls, while the control group shows a clear, progressive increase as the day goes by (lower at 10:00 and highest emission at 20:00) (Figure 7C).
The second batch was analyzed with a one-tail T-test in favor of the LBN (CTR<LBN), and once again the number of the sonic calls was significantly higher in the LBN group
(p-value<0.001; Figure 7D; Table 5D). Bayesian statistics showed an incredibly high Bayesian
factor (BF10=147,942; Table 5F) strengthening the reliability of our data indicating strong
evidence of the effect. In this batch the session seems to have a significant effect on the emission of the calls (p-value<0.001; Table 5D). More specifically, the squeaks are at a low level in the beginning of the day and then progressively increase, reaching the highest levels in the last session (20:00) during the light phase in both control and LBN group, but with the LBN group always showing a much higher amount of calls (Figure 7F).
Β) Α)
19 B) C) E) A) *** D) *** F) *
Figure 7 Graphical representations of the sonic calls. First illustrated the effect of the paradigm on the first batch of animals, below
respectively for the second batch. A) Sum of the sonic calls throughout the days and sessions for the first batch. B) Number of sonic calls of the first batch on each day from P2-8. C) Number of sonic calls throughout the three sessions of recording for the first batch. D) Sum of the sonic calls throughout the days and sessions for the second batch. E) Number of sonic calls of the second batch on each day from P2-8. F) Number of sonic calls throughout the three sessions of recording for the second batch. Asterisk indicates significance.
D)
E)
Table 5 Statistical analysis with the help of classical ANOVA. A) Estimation of the effect of the group and the session on the number of calls
of the first batch. B) Estimation of the effect of the group and the age on the number of calls of the first batch. C) Calculation of the BF10 for the first batch. D) Estimation of the effect of the group and the session on the number of calls of the second batch. E) Estimation of the effect of the group and the age on the number of calls of the second batch. F) Calculation of the BF10 for the second batch.
F)
20 Readouts on P9
When the pups reached the 9th day of their life, they were sacrificed in order to collect their
brains, adrenals, thymuses and a sample of their blood. Before sacrificing the animals they were weighed. Brains and corticosterone were only measured in the second batch of animals. Their weights were also measured on P2 to evaluate that there were no differences between the two groups before applying the different conditions (LBN paradigm from P2-P8). Indeed, the two groups did not show a difference in the weights at P2 (p-value=0.059, BF10=1.341
indicating anecdotal evidence of an effect and p-value=0.713, BF10=0.329 strong evidence of
absence respectively for the 1st and 2nd batch). On P9 the LBN animals weighed significantly
less than the control group in both batches (first batch value<0.001; second batch
p-value<0.001). Since the difference in the body weight of the animals of the two groups differed
significantly the rest of the weights were normalized on 100g of body weight. The adrenals of the P9 pups were significantly bigger in the LBN group compared to the control group (adrenal weight first batch p-value<0.001; second batch p-value=0.005). The weight of the thymuses was significantly lower in the LBN animals of both batches when compared to the control group (thymus weight first batch p-value<0.001; second batch p-value=0.005) (Tables 6A, C respectively). On the second batch the brain weights were also calculated. The LBN group showed significantly higher brain weights than the control group (p-value=0.007). Lastly, the corticosterone levels that were calculated from the blood sample that was collected during the sacrifices, did not show a significant difference between the two groups (Table 6C). The Bayesian factor for this parameter indicated that there is not enough evidence to draw conclusions (BF10=0.978), thus a bigger sample size might be needed in order for this
parameter to reach significance (Table 6D) (Figure 8).
The differences that might be caused in these parameters due to the sex of the pups seem to be of high interest. For this reason, ANOVAs were run in order to see the group and sex effect on these parameters as well as the interaction of the two. On the first batch none significant differences occurred due to the sex of the animals. However, on the second batch of animals, a statistically significant interaction between the group and the sex of the animals concerning the body weight (p-value=0.043, PTukey=0.006 between CTR males and LBN males) as well as
the brain weight (p-value=0.042, PTukey=0.001 when comparing CTR,females&LBN,females;
CTR,females&LBN,males; and CTR, males& LBN males) was observed.
Table 6 Statistical analysis with the use of T-test and Bayesian T-tests. A) Calculation of the p-values in the parameters
studied in the first batch. B) Bayesian analysis of the parameters studied in the first batch C) Calculation of the p-values in the parameters studied in the second batch. B) Bayesian analysis of the parameters studied in the second batch.
A) B)
21
Table 7 Classic ANOVA for the estimation of the effect the group and sex have on the body weight (left) and brain weight (right) at P9 as
well as the interaction of the two.
Figure 8 Graphical representation of the alterations caused to the parameters under study. A) Mean of the P2 body weights in the first batch.
B) Mean of the P9 body weights in the first batch. C)Mean of the adrenal weights of the first batch at P9. D) Mean of the thymus weights of the first batch at P9. E) Mean of the body weights of the second batch at P2. F) Mean of the body weights of the second batch at P9. G) Mean of the adrenal weights of the second batch at P9. H) Mean of the thymus weights of the second batch at P9. I) Mean of the brain weights of the second batch at P9. J) Mean of the corticosterone levels of the second batch at P9. Asterisks indicate significance. Blue dots represent males and pink dots females.
A) B) C) D) E) F) G) H J) I)
***
***
***
***
***
***
**
22 Morphological analysis of the Amygdaloid Neurons
Once the brains were collected and weighed as prementioned, they were directly put in the Golgi-Cox procedure in order for the neurons to be stained throughout the brain (this part concerned only the second batch of animals). This aimed to unveil morphological differences in the neurons that might be caused by ELA. In this study we traced the neurons on FIJI computer software. The neurons that were analyzed in the current study belonged to the basolateral amygdala of the P9 brains and more specifically to the BLA. In total 70 neurons were used from 12 different subjects (6 Control 6 LBN), 34 neurons from the control group and 36 from the LBN group. We compared the size of the soma and the length of the dendrites, apical and basal, between the control and LBN group. Additionally, the sum and average of the intersecting radii as well as the ramification index (the number of dendrite intersections for each circle is measured and the highest value is divided by the number of primary dendrites, consisting a measure to evaluate arborization) were compared between the two groups. The statistical analysis showed that there were no statistically significant results in the current study regarding the prementioned parameters. The groups did not show any difference in any of the parameters under investigation. The Bayesian factors were very low, indicating that there is moderate evidence for the absence of the effect or anecdotal evidence of the effect (the latter was observed in the statistical analysis concerning the average number of intersections of the basal dendrites and the ramification index, indicating that a bigger sample size might be needed in order for the results to reach significance)(Table 8, Figure 9, 10). In Figure 11, representative pictures of the tracing produced in FIJI.
Table 8 Statistical analysis of the neurons. A) Sum and Mean of intersections as well as the ramification index of the
whole neurons. B) Bayesian analysis of the whole neurons. C) Sum and Mean of intersections as well as the ramification index including only the apical dendrites. D) Bayesian analysis of the apical dendrites. E) Sum and Mean of intersections as well as the ramification index of the basal dendrites. F) Bayesian analysis of the basal dendrites.
A) B)
C) D)
23
B) C)
D) E)
Figure 9 Graphical representation of the neuronal data of the Control and LBN group. A) Mean length of the soma of the Control and
LBN group. B) Mean length of the apical dendrites of the Control and LBN group. C) Mean length of the primal apical dendrite of each neuron from the Control and LBN group. D) Mean length of the basal dendrites of the Control and LBN group. E) Mean length of the primal basal dendrite of each neuron from the Control and LBN group. All lengths are expressed in pixels. No significant effect on any of the parameters.
24
Figure 10 Graphical representation of the neuronal data of the Control and LBN group. A) Sum of the intersections
of the Control and LBN group. B) Average number of intersections per group. C) The ramification index of the neurons as a measurement of the arborization. None significant effect.
A) B) C)
25
Figure 11 Tracings of amygdaloid neurons imported in FIJI. Traced with the Simple
Neurite Tracer plug-in. The soma, apical and basal dendrites are traced, and their length is then further analyzed with the use of Sholl analysis. The cross in the picture on top is placed on the starting point of the analysis.
Soma Basal dendrites Apical dendrites Axon Basal dendrites Apical dendrites Basal dendrites
26
Discussion
ELA is mentioned to have an effect on both the dam and her pups, affecting the behavior of the dam in these early stages of maternity which can exert an effect on the development of the pups. This study concentrated on the behavior of the dam towards the pups, the sonic calls that the pups were emitting throughout the days and the sessions, the alterations caused until P9 on the body, adrenal, thymus and brain weight and lastly the morphological development of the neurons in the amygdala, which is a region highly affected by stress since it is abundant in GRs and MRs and is mentioned in previous studies to be highly affected by ELA but also to play an active role in the expression of stress.
Alterations in the duration of the maternal behavior between the two groups
The maternal behavior was scored from P2-P8 (when the LBN paradigm was applied) three times a day, 10:00, 15:00 and 20:00. The maternal behaviors were further categorized into three larger groups that concerned the directionality of the behavior. They were categorized in pup-directed, non-pup-directed and maternal associated (the latter included moving around the nest, nest building and tail chasing). Afterwards, they were statistically analyzed and represented in graphs. The pup-directed and non-pup-directed behavior’s duration showed a significant difference between the two groups under examination (LBN and Control). More specifically, the LBN group seemed to engage mostly with the pup-directed behaviors, while the Control group was engaging mostly with the non-pup-directed behaviors. This preference in maternal behavior was significantly affected by date and time. With the passing of the days the preference got stronger reaching the highest level of duration at P8 which is also the last day that the observations took place. The pup-directed behaviors were more prominent during the light phase session (20:00), while the non-pup-directed behaviors were more prominent during the dark phase, where the animals are more awake since they are nocturnal. This indicates that the LBN dams were providing their pups with more care in comparison to the control dams.
After this significant observation, it was of high interest to take a closer look into separate behaviors concerned the most prominent way of a dam to express her affection and care towards her pups. The behaviors chosen for the prementioned reason were licking and grooming, off-nest behavior, low-nursing and carrying of the pups. In most of the behavior’s duration (one-on-one) there was a significant difference between the LBN and control group. The only behaviors that indicated insignificance were arched back nursing, side/passive nursing, self-grooming and rearing, consisting the 1/3 of the total behaviors under examination. The LBN group was dedicating more time in licking/grooming, lowly nursing and carrying the pups, indicating that indeed the care that was provided to the pups was higher in the experimental group when compared to the control. What is more, the off-nest behavior was significantly shorter in the LBN group, indicating that the control dams were associating less with their pups. This effect has been mentioned before in previous studies that applied various paradigms in order to induce ELA, including the LBN paradigm, however there are studies showing the exact opposite effect in the behaviors of the dam (Molet et al., 2016). According to a review on these various studies by Macrì & Würbel, 2006 there is a logical explanation that might support this effect, leaning on the fact that under stressful conditions the dams might become more protective and attentive against the adversity they are facing, thus providing more care to protect their pups; while when the conditions are stable the dams might be less attentive and more relaxed, thus spending less time taking care of the pups and
27
engaging with more off-nest and non-pup-directed behaviors, as observed in our study. In the literature there still exists contradictory evidence regarding the alteration of the maternal behavior due to ELA paradigms, thus it is of high importance to exert more research on this domain and try to unravel the source causing the production of contradicting results.
Higher unpredictability of the maternal behaviors in the LBN group
To further understand the alterations in the maternal behavior, the entropy index was calculated. For this reason, we used a matrix containing the behaviors scored and assessed the transitions from one behavior to another leading to the production of a value indicating the possibility of transitioning from one behavior “A” to a behavior “B”. The two groups showed a different transition from one behavior to another and the entropy rates indicated a higher unpredictability in the behavior the LBN dam decides to follow, while the control group shows higher predictability and more moderate unpredictability. These results are in accordance with the findings in the study performed by J Molet et al., 2016. In this study the LBN paradigm was applied from P2-9 and the analysis the LBN group showed a significantly higher unpredictability when compared to the control group. Their results concerning the mean duration of individual licking/grooming bouts are also in accordance with our results. Thus, licking grooming seems to be longer in duration in the LBN group but it is also more fragmented. This creates a greater suspicion that this paradigm and stress in early stages of maternity can have an effect in the behavior the dam chooses to associate with which can affect the development of her infants in the long run.
Higher Audible vocalizations/Sonic calls in the LBN group
To obtain more insight on the response of the mother towards her pups, it was decided to record the sonic calls from the cages. Sonic calls consist a means of communication in rodents and are mentioned to be possibly able to lead to harm avoidance caused by the mother (e.g. prevents the dam from stepping on the pup) as well as it can bring the mother to the infants by capturing the mother’s attention (Stern, 1997). The two batches of animals were analyzed separately. The first batch was used as the “pilot” to determine the directionality of the expected outcome that should be encountered also in the second batch analysis since the conditions were the same. In both batches, the number of sonic calls emitted from the LBN cages is significantly higher than the control respective one. In the first batch there is no significant effect on the emission deriving from the time or the day of the recording, however the graphs show a trend in the emission that could be of interest for future studies and there is a significant interaction between the group and the session, with the squeaks being more prominent during the light phase meaning that these variables are not independent from eachother. On the contrary, the second batch shows a significant effect of the session in the production of the squeaks. During the last session the number of squeaks reaches its maximal level in both groups. The second batch also shows a trend in the days (p-value=0.06 and BF10=3,493e) which might be able to reach significance with a higher sample size.
These enhanced squeaks occurring in the LBN group mostly during the last session (20:00, light phase) when combined with the outcomes in the maternal behavior they could lead to a correlation between the emission of the squeaks and the maternal behavior the dam chooses to engage with. These results could indicate that the dam responds to the augmented amount of calls deriving from her pups and runs to their aid, thus explaining also why the LBN dams are providing more care to their pups. However, we cannot conclude such a phenomenon in the current study due to the little number of the cages being recorded, but it could be an
28
observation worth conducting more studies on in order to fully estimate it. There is evidence in the literature indicating that such a correlation between the squeaks and the maternal behavior might be possible, however most of the studies are performed in mice, creating a clear limitation when trying to compare results and reach conclusions. In a previous study by Heun-Johnson & Levitt, 2016, that was conducted in mice raised in LBN conditions from P2-9, it was shown that the emissions of the USVs as well as those of the sonic calls were significantly increased in the LBN group and this difference was also accompanied by longer on-nest periods (similar to our pup-directed distinction) that the dam dedicated to the LBN group when compared to the control group. In this study they related the emission of the squeaks with the number of times the dam entered the nest (immediately when she was entering the nest the pups were generating audible vocalizations) and indicated that the squeaks might signal adversity and maltreatment of the pups (stepping on the pups, scratching them etc.). When the dam is off the nest the pups seem to emit less squeaks. These different explanations concerning the squeaks throughout the literature point towards the need of making a comparison of the time the squeaks are emitted and the behavior the mother engages with at that specific moment in order to determine whether the squeaks are due to aversity or due to higher demand in attention and care. The current study cannot formally exclude the possibility that some squeaks might be generated by the dam, however it strongly supports the fact that some -if not all- the squeaks are coming from the pups, since in the analyses that took place we could see several squeaks occurring at the same time, thus this is clear evidence that the majority of the squeaks are emitted by the pups.
Validating the LBN model with measurements at P9
On P9 the pups were sacrificed, and some measurements were taken in order to validate the successful application of the LBN paradigm. In previous studies the validation of the model included a decrease in the body weight of the LBN animals, a decrease in the weight of the thymuses, a hypertrophy of the adrenals that are producing and releasing corticosterone in the periphery and an increase in the peripheral corticosterone levels (Kioukia-Fougia et al., 2002; Avishai-Eliner et al., 2008; Ivy et al., 2008; Molet et al., 2014; Derks et al., 2016; Walker
et al., 2017). Our results are in accordance with these previously mentioned results allowing
us to have strong suspicions that the paradigm was successfully implemented. However, we did not observe a significant difference in the corticosterone levels and the Bayesian statistics supported that there is no strong evidence, which indicates that a greater sample size might be needed in order to be able to observe the effect. We also reported a significant difference in the brain weights. The same results were observed in a study by Guadagno et al., 2018 were there was also absence of a statistically significant change in the corticosterone levels of the P9 pups that were submitted to the LBN paradigm from P1-9, indicating that high corticosterone levels might not be needed in such early life stages in order for stress to exert an overall effect on the development of the subject, in contrast to the literature on adult subjects. Lastly, the LBN brains weighed significantly more than the control brains. This parameter was calculated as an extra measurement in the validation of the model and could be suggested as a future parameter to be under examination. The BF for this parameter suggests moderate evidence of the effect (the LBN brains weigh more than the Control ones).
