Sex Differences in Spontaneous Behavior and Cognition in Mice Using an Automated Behavior Monitoring System

Abstract

Isolation of sex differences as a key characteristic underlying neurobehavioral differentiation is an essential component of studies in neuroscience. The current study sought to address this concern by observing behavioral differences using an automated home cage system for neurobehavioral assessment, a method rapidly increasing in use due to advances in technology and advantages such as reduced handling stress and cross-lab variability. Sex differences in C57BL/6 mice arose for motor activity and circadian-linked behavior, with females being more active compared to males, and males having a stronger anticipatory increase in activity leading up to the onset of the light phase compared to females. These activity differences were observed not only across the lifespan, but also in different genetic background mouse strains across different testing sites showing the generalizability and robustness of these observed effects. Activity differences were also observed in performance on a spatial learning and reversal task with females making more responses and receiving a corresponding elevation in reward pellets. Notably, there were no sex differences in learning nor achieved accuracy, suggesting these observed effects were predominantly in activity. The outcomes of this study align with previous reports showcasing differences in activity between males and females. The comparison across strains and testing sites showed robust and reproducible differences in behavior between female and male mice that are relevant to consider when designing behavioral studies. Furthermore, the observed sex differences in performance on the learning and reversal procedure raise concern for interpretation of behavior differences between sexes due to the attribution of these differences to motor activity rather than cognition.

1. Introduction

Sex differences in mammalian species are understudied in neuroscience (Beery & Zucker, 2011; Garcia-Sifuentes & Maney, 2021; Mamlouk, Dorris, Barrett & Meitzena, 2020) despite these differences being essential for advancing understanding of the impacts of various external and internal stimuli, both on neural function and behavior (Cahill, 2006; Clayton, 2016; Shansky & Woolley, 2016). While the field has seen a much-needed increase in the inclusion of both sexes (Woitowich, Beery & Woodruff, 2020), sex is still often overlooked as a biological variable resulting in a lack of understanding of key behavioral differences observed in commonly employed rodent models. It is crucial to characterize differences between sexes to judge the expected impact on variability and interpretation of studies including mixed-sex groups. Important differences between sexes exist for locomotor activity, learning, behavioral flexibility, and others (Borbélyová, Janišová, Mysliveéek & Riljak, 2019; Gargiulo et al., 2022; Jonasson, 2005; Mihalick, Langlois & Krienke, 2000; Mishima, Higashitani, Teraoka & Yoshioka, 1986; Stevanovic et al., 2022). These observed behavioral phenotypes may also be associated with differing physiological pathways which may interact with, and therefore cloud, unique impacts of sex.

In traditional behavioral assays in rodents, motor behavior is shown to differ between sexes. For example, females display elevated activity in open field, longer latency and higher required speed to fall during rotarod testing, and baseline differences in hindlimb and forelimb extension for postural maintenance on an inclined plane (Blizard, Lippman & Chen, 1975; Field, Whishaw & Pellis, 2000; Hernandez, Truckenbrod, Campos, Williams & Burke, 2020; Knight, 2021). However, it remains unclear whether observed motoric differences are a common differentiating characteristic, or a byproduct of other processes, such as increased stress during task performance, differences in exploratory behavior, or changes in hormonal state, especially in females throughout the estrus cycle, among others. Sex-dependent performance in motor tasks is shown to be associated with stress (Sturman et al., 2018) and hormonal state (Beatty, 1979; Blizard et al., 1975; Miller et al., 2021), lending credence to these mechanisms influencing these observed effects and thus producing complexity in interpretation of task performance regarding sex or other sex-linked challenges.

Sex differences in spatial and reinforcement processing have been shown inconsistently in rodent models. In some studies, males tend to learn the location of a hidden platform in a Morris water maze task more readily than their female counterparts (Kavaliers et al., 1996; Perrot-Sinal, Kostenuik, Ossenkopp & Kavaliers, 1996). However, this effect is inconsistent and could be partially attributed to differences in task parameters or the determination of behavioral endpoints (Ge, Qi, Qiao, Wang & Zhou, 2013; Roof & Stein, 1999). Applying spatial and reinforcement processing to tasks assessing behavioral flexibility, such as spatial alternation or reversal procedures, allows these differences to become a bit clearer. Gargiulo and colleagues (2022) reported that female Sprague-Dawley rats tested in an operant strategy shifting procedure displayed higher incidence of both omission errors during reversal and perseverative errors during the extra-dimensional shift as compared to males (Gargiulo et al., 2022), lending further support for a sex difference in behavioral flexibility. Performance in discrimination and reversal procedures appears to be partially mediated by stress, with acute stress impairing discrimination learning in male and female rats (Gargiulo et al., 2022) and chronic stress increasing perseverative errors in female ovariectomized rats (McLaughlin et al., 2009). Perseverative behavior in female rats observed by Gargiulo and colleagues (2022) may be attributed to reduced willingness to explore the novel reward location due to higher novelty stress, a phenomenon that has been shown previously in female mice who displayed reduced exploration of a novel object following handling-induced stress (Stevanovic et al., 2022). These outcomes further strengthen the argument that internal and external processes differentially mediate behavior between sexes resulting in unclear interpretation of differences and the impacts those differences have in behavioral studies.

As mentioned, one established potential mediator for observed differences between males and females is acute and/or chronic stress. Acute stress can be induced by handling animals either before or during task performance (Balcombe, Bernard, & Sandusky, 2004) which may have important implications for interpreting baseline sex differences. Male and female mice differ in novelty preference depending on the magnitude of handling, whereas female mice have lower novelty preference than males after minimal handling but similar preference after chronic handling (Stevanovic et al., 2022). Further, not only do sex differences in neurobehavioral assessments arise to greater magnitude depending on handling strategy (Gouveia & Hurst, 2019), but behavior in males and females can also be differentially altered (Sensini et al., 2020). Due to these handling-induced differences, the purpose of the current study was to assess baseline sex-dependent behavioral phenotypes in adult mice in the absence of daily handling utilizing an automated home-cage monitoring system across multiple days (Grieco et al., 2021). Behaviors assayed included animals’ spontaneous behavior, spatial learning, and reversal learning. The generalizability of the observed effects in spontaneous behavior was determined by comparing behavior in various murine models across various ages during early- to mid-adulthood.

2. Methods

2.1. Animals

Animals included 32 (16 Female) mice purchased from either Taconic Laboratories (N = 16, 8F; Albany, NY) or Jackson Laboratories (N = 16, 8F; Bar Harbor, ME) at 6wks of age. Upon arrival at the National Institute of Environmental Health Sciences (NIEHS), animals were given one week of habituation. Behavioral testing began either when animals were 8wks of age (C57BL/6J; Jackson) or when animals were approximately 11wks of age (C57BL/6NTac; Taconic). Prior to behavioral testing, animals were housed in an AAALAC-approved animal facility under a 12:12 light/dark cycle (lights on at 6am). Animals were group housed in ventilated cages on standard wood-chip bedding in sets of 3-5 conspecifics upon arrival and then singly housed in static cages on Alpha-dri^® bedding (Shepherd Specialty Papers, Inc.) within the behavioral testing room beginning several days prior to the start of testing. Animals had ad libitum access to standard NIH-31 rodent chow (Ziegler Bros, Inc.), except for those that underwent CognitonWall^™ testing, see below, and ad libitum access to reverse osmosis (RO) water.

All procedures were in accordance with NIEHS/NIH Humane Care and Use of Animals in Research protocols.

2.2. PhenoTyper Procedure

For behavioral testing, animals were singly housed in opaque Noldus PhenoTyper boxes (3000 series, dimensions of L=30 x W = 30 x H = 35 cm; Noldus IT, Wageningen, The Netherlands; Figure 1A). At the NIEHS facility, PhenoTypers were situated in Caron^® chambers allowing for regulation of temperature, humidity, illumination, and sound attenuation. Each PhenoTyper had Alpha-dri^® bedding, an opaque white shelter in the corner of the chamber closest to the entrance, a nestlet as nesting material placed within the shelter, a water bottle with RO water on the wall across from the shelter—near the entrance, and a food hopper with either NIH-31 feed on the back wall. Testing was facilitated by EthoVision XT16 software running a set of experimental protocols as described previously (Logan et al., 2019, Loos et al., 2014).

Figure 1:

A) Representative image of the Noldus PhenoTyper system used to assess spontaneous activity and cognition in mice. PhenoTypers were housed inside a Caron^® chamber for regulation of temperature and humidity and facilitation of sound attenuation. Animals were singly housed in PhenoTypers with animal behavior monitored by an overhead camera and acquired using EthoVision XT16 Software. B) Schematic representation of the CognitionWall procedure. The CognitionWall is placed in the back corner of the chamber and has a pellet dispenser port at the rear. During the discrimination learning phase, mice are trained to run through the left-most port five times for food reinforcement. After two days, the protocol switches to the reversal learning phase, where animals must enter through the right-most hole five times to obtain food reinforcement. CognitonWall schematic adapted from Remmelink, Smit, et al., 2016 under creative commons under Cold Spring Harbor Press.

During Phase 1, spontaneous behavior was assessed in sixteen 8-wk old C57BL/6J mice. Mice were individually housed in PhenoTyper boxes and had free access to rodent chow and water. Activity, movement kinetics, and sleep patterns were observed across a testing period of 60hrs, allowing determination of animals’ activity over a prolonged duration of time. For Phase 1, animals were housed in the PhenoTyper approximately two hours before the onset of the dark phase with behavior monitoring beginning at the onset of the dark phase. Behavioral measures were assessed across both dark and light phases and changes in behavior were observed during the transition periods between these phases (see Loos et al., 2014).

In Phase 2, discrimination and reversal learning were assessed in sixteen 11-wk old C57BL/6NTac mice. During Phase 2, free access to NIH-31 feed was revoked (the food hopper entrance was blocked by an opaque white sliding door) and animals were trained to perform a spatial learning and reversal procedure for access to food reinforcement using a CognitionWall^™ (Figure 1B; Remmelink, Aartsma-Rus, et al., 2016; Remmelink, Smit, et al., 2016). The CognitionWall is an opaque three-hole wall that is placed in the back-rear corner adjacent to the previous food hopper entrance. In the rear of the wall is a port for access to a pellet dispenser that dispenses a single 20mg sucrose pellet (Dustless Precision Rodent Pellets, Bio-Serv, CAT#: F0071) as the food reinforcer when response criteria are met. During the first two days of Phase 2, discrimination learning (DL) was assessed in animals by training them to enter the left-most hole of the wall five times for food reinforcement (FR5). For the last two days, reversal learning (RL) was assessed. Here, the animal was required to enter the right-most hole (FR5) for reinforcement. During both the discrimination and reversal phases, the amount of time, number of entries, number of errors, and number of pellets received to reach over 85% accuracy (taken at 87%) were examined. Accuracy was also assessed every 50 entries to observe animals’ acquisition during both discrimination and reversal phases.

2.3. Muli-Lab Comparison of Strain and Age

In order to assess the generalizability as well as the reproducibility of findings observed in C57BL/6 mice, behavior was compared to data of separate cohorts from Vrije Universiteit Amsterdam (Amsterdam, The Netherlands) which were processed and provided by collaborators at Sylics (Bilthoven, The Netherlands). A total of 140 (N = 55F) C57BL/6J, 53 (N = 31F) Grlb(spa), and 59 (N = 33F) SOD1(G93A) mice were assessed across various ages during early to mid-adulthood. Additional detail for these strains is contained in the Supplement.

2.2. Data Analysis

Data were collected using EthoVision XT16 software and uploaded to the AHCODA^™ data analysis platform (Sylics Bioinformatics^®, The Netherlands; Koopmans, Smit, Verhage, & Loos, 2017) for pre-processing and determination of behavior endpoints. Binned data (1hr) were compiled, analyzed, and visualized in R (version 4.2.2) using the packages tidyverse (version 2.0.0), nlme (3.1-162), and cowplot (1.1.1).

During spontaneous behavior, activity duration, locomotor activity (speed, distance, movement pauses, etc.), and time spent in and out of the shelter were compiled into 1hr epochs with the mean taken for each day (light phase) and night (dark phase), 12hrs/each. Effects of sex on activity were assessed during the light phase of day 2 and the dark phase of night 3. Assumptions of normality were checked using Shapiro-Wilks test. For normally distributed data, effects of sex were assessed using two sample t-test. For non-normally distributed data, effects of sex were assessed using Kruskal Wallis. To determine whether effects of sex were specific to the dark phase on night 3, the main effect of sex, main effect of night, and the interaction between sex and night was assessed using one-way ANOVA—assumption of normality was tested using Shapiro-Wilks test. For animals assessed across the lifespan, effects of sex and age were determined using one-way ANOVA, as these data were not repeated measures. During the discrimination and reversal procedure, accuracy every 50 entries, entrances and errors to criterion, the number of pellets received, and maximum accuracy were compiled. Accuracy per 50 entries as well as the number of entries required to reach over 85% accuracy were measured to determine the progression of training between sexes. For non-repeated data, assumptions of normality were assessed using Shapiro-Wilks test and differences between sexes were assessed either using two sample t-test or Kruskal Wallis depending on whether the assumption of normality was valid. For repeated data, a linear mixed model with subject set as the random effect was used to assess the main effect of sex, the main effect of trial—binned by every 50 trials—and the interaction between sex and trial on acquired accuracy. Trial was modeled using a quadratic term to account for the curvilinear increase in accuracy over trials.

3. Results

3.1. Spontaneous Behavior

8wk-old C57BL/6J mice were tested in a 60hr spontaneous behavior protocol. Data across this period were binned into 1hr epochs with activity duration, distance moved, and time spent within the shelter depicted to represent the alteration in behavior across the dark and light phases (Figure 2). Average activity was expectedly higher in animals during the dark phases compared to the light phases. Rest activities (such as shelter entrances or time spent in the shelter) were higher in the light phases compared to the dark phases.

Figure 2:

Representative graphs for activity over the 60hrs of spontaneous behavior in C57BL/6J mice. Animals are generally more active during the dark phases, shaded sections, than during the light phases. On average, females tended to be more active in the dark phases as compared to males, seen here both for activity duration and for distance moved. Animals spend more time in the shelter during the light phases compared to the dark phases and this was not sex dependent. Data are group means ± SEM. Colored markers in the top panel represent sexes for all three panels.

An overall trend emerged with higher activity in females compared to males. To assess the magnitude of these effects, the average hourly activity each light and dark phase (i.e., the average hourly activity each night and each day) was isolated and differences between sexes were determined by two sample t-test. Behavior during the dark phases of nights 1 and 2 is associated with habituation to the new home-cage environment, so only night 3 was isolated. Importantly, sex differences were observed across all three observed nights (Supp. Table 1). Figure 3 depicts significant outcomes observed during the dark phase, as measured on night 3, with notable differences between sexes observed across almost all isolated behavioral endpoints (no differences were observed for time spent in the shelter, H(1) = 2.824, p = 0.093, not graphed). Overall, females tended to have higher activity as compared to males during the dark phase. Further, the higher activity observed in female mice remained across nights (Supp. Table 1) and did not appear to be night-dependent, given a lack of significant interaction between sex and night for any measure.

Figure 3:

Mean activity differences during spontaneous behavior for the dark phase of night 3. Females tended to be more active than males, seen by higher activity in females in almost all endpoints (except shelter time, not depicted). These differences were also observed during both the nights 1 and 2. Data are means ± SEM. Points represent individual subject data. * p < 0.05; ** p < 0.01, *** p < 0.001.

Table 1 shows the outcomes of spontaneous behavior during the light phase as measured on day 2, as day 1 was for habituation. Animals are generally less active during the light phase resulting in higher variability and fewer differences being observed between the sexes. However, some notable differences in activity remained, including movement acceleration and velocity and both the duration of shelter visit bouts and the number of shelter entrances.

Table 1:

Outcomes of sex on spontaneous behavior during the light phase, Day 2

Measure/Hour	Male	Female	T	df	p
Maximum Acceleration (cm/s2)	1.68±0.14	2.30±0.18	2.655	14	0.019
Duration of Movement Bout (s)	1.05±0.03	1.04±0.03	−0.326	14	0.749
Distance Traveled in Bout (cm)	9.35±0.57	10.62±0.6	1.535	14	0.147
Activity Duration (s)	161.36±13.48	186.29±25.25	0.871	14	0.398
Time in Shelter (s)***	3163.96±124.61	3259±47.36	0.011	1	0.916
Shelter Visit Bout Length	14.33±0.72	17.84±0.69	3.513	14	0.003
Shelter Entrances	3.73±0.7	6.98±0.75	3.16	14	0.007
Distance Moved (cm)***	755.41±67.29	931.51±178.71	0.397	1	0.529
Movement Velocity (cm/s)	8.33±0.67	11.44±1.03	2.53	14	0.024

Groups are 8 animals/sex. Data for males and females for each measure are means ± SEM. Effects of sex were assessed using two sample t-test or Kruskal-Wallis***, with either the T or H statistic reported with corresponding degrees of freedom. Highlighted are significant outcomes of sex.

Prior to the onset of each light phase there was an observed upward or downward shift in activity, depending on whether behavior was active or resting (Figure 4). This “anticipation of light” alteration in activity is indicative of a change in behavior driven by the light/dark cycle (Loos et al., 2014). Data across the last five hours during night 3 were isolated and a change score was calculated comparing the last two hours of this period to the first three hours (Figure 4). The assumption of normality was assessed using Shapiro-Wilks and differences between sexes were assessed either using two-sample t-test or Kruskal-Wallis. These data are represented in Table 2.

Figure 4:

Diagram depicting the alteration in behavior at the end of the dark phase on night 3, prior to the lights coming on. This shift, depicted in the shaded region, is the “anticipation of light” (AOL) response occurring during the last five hours of this phase. The anticipation of light response is determined by taking the average of hours 8-10 ($T_0$) and the average of hours 11-12 ($T_1$) and calculating a change score between them as:

$$AOL=\frac{T_1-T_0}{T_1+T_0}$$

Table 2:

Outcomes of sex on anticipation of light during the dark phase, Night 3

Measure/Hour	Male	Female	T	df	p
Maximum Acceleration (cm/s2)	−0.056±0.054	−0.047±0.041	0.138	14	0.892
Duration of Movement Bout (s)	0.008±0.018	0.016±0.012	0.330	14	0.747
Distance Traveled in Bout (cm)	−0.006±0.026	0.014±0.020	0.605	14	0.555
Activity Duration (s)	0.418±0.095	−0.181±0.123	−3.857	14	0.002
Shelter Visit Bout Length	0.187±0.080	−0.050±0.086	−2.016	14	0.063
Time in Shelter (s)	−0.263±0.075	0.259±0.134	3.398	14	0.004
Shelter Entrances***	0.416±0.138	−0.125±0.143	5.338	1	0.021
Distance Moved (cm)	0.354±0.088	−0.210±0.115	−3.895	14	0.002
Movement Velocity (cm/s)***	−0.075±0.053	−0.018±0.045	0.706	1	0.401

Groups are 8 animals/sex. Data for males and females for each measure are means ± SEM. Effects of sex were assessed using two sample t-test or Kruskal-Wallis***, with either the T or H statistic reported with corresponding degrees of freedom. Highlighted are significant outcomes of sex.

Differences between sexes were observed for changes in active behavior (activity duration and distance moved) and rest (time in shelter and shelter entrances) with females tending to have a less pronounced shift in activity compared to males who tended to have a notable increase in activity near the end of the dark phase.

While we observed clear differences between sexes on activity in these 8wk old

C57BL/6J mice, it was of concern whether this effect was generalizable outside of these young adult mice as this would decrease the validity of these endpoints as predictive measures of baseline neurobehavioral phenotype. We addressed this concern by comparing activity duration during the dark phase of night 3 across various ages among three mouse models assessed in a separate laboratory in Europe (Figure 5). There was a main effect of sex for both the C57BL/6J (F(1, 130) = 56.718, p<0.01) and SOD1 (F(1,53) = 10.74, p < 0.01) mice. While there was also a main effect of age for both strains (C57: F(4, 130) = 2.545, p = 0.043; SOD1: F(2, 53) = 7.546, p = 0.001), the effect of sex did not interact with age for either strain (C57: F(4, 130) = 0.773, p = 0.545; SOD1: F(2, 53) = 0.322, p = 0.726) suggesting that there was an overall sex-difference across ages and that this effect did not depend on the age of assessment. On the other hand, Grlb(spa) mice showed no main effect of sex (F(1,45) = 0.027, p = 0.871) or interaction between sex and age (F(3,45) = 1.924, p = 0.139) suggesting a loss of this key phenotype in this strain.

Figure 5:

Age-dependent effects on activity duration across three murine models, C57BL/6J, SOD1(G93A), and Grlb(spa). A significant main effect of sex (p < 0.01) was observed across ages both for C57BL/6J mice and for the SOD1 mutants. There was no trend of sex difference for the Grlb(spa) mutants. The elevated activity duration across ages in female C57BL/6J animals was like that observed in 8wk old C57BL/6J mice of the current study. The lack of differences in the Grlb(spa) mutants may be associated with either a lack of sex-differentiation or sex-dependent differences in motoric capacity related to the loss of glycine receptors in the CNS in this mutant strain. Data are means ± SEM. Colored marker in the left panel represent sex for all panels.

3.2. Discrimination and Reversal

Discrimination learning, reinforcement processing, and reversal were assessed in 11wk-old C57BL/6NTac mice across 4 days using a CognitionWall^™. Data during DL were compiled to assess animal’s learning of the location of the reward. Data during RL were compiled to assess animals’ ability to transition behavior away from the previously reinforcing port to the new port which is a measure of perseveration and reinforcement processing. All statistical outputs are reported in Supplementary Table 2. Figure 6 shows average accuracy across the first 1250 entries, in 50 entry bins, during DL (Fig. 6A) and RL (Fig. 6B). Animals’ accuracy during DL reliably increased from chance, about 33%, to above 75% after 1250 entries. This increase was not sex-dependent (Supp. Table 2; “Accuracy per 50 Entries-DL”), showcasing a lack of effect of sex on the increase in accuracy during the discrimination phase.

Figure 6:

Outcomes during Discrimination (DL) and Reversal (RL) Learning in C57BL/6NTac mice. A&B) Accuracy across the first 1250 entries. During DL, animals’ accuracy increased from chance (33%) to over 75% showcasing steady improvement in task performance. This effect was not dependent on sex. During RL, animals responding reliably shifted when the reinforcing hole switched, seen by the increase in accuracy from 0% to over 75% within the first 1250 entries. This transition was sex-dependent with males requiring fewer trials to reach 50% accuracy than females. C&D) The total number of pellets received during DL and RL. During DL, females received more pellets than males. This trend did not extend over to RL. E&F) The total number of entries during DL and RL. During DL, total entries into the ports was higher in females than males. This effect was absent during RL. Data are means ± SEM. Colored markers in A represent sexes for A and B. * p < 0.05; *** p < 0.001.

During RL, accuracy reliably shifted away from the previously reinforcing port to the newly reinforcing port, with this transition taking approximately 1000 entries. While there was no main effect of sex during RL, there was an interaction between entry number and sex showing that the transition to the newly reinforcing alternative was sex-dependent (Supp. Table 2; “Accuracy per 50 Entries-RL”) with males tending to require fewer entries to reach 50% accuracy than females. Importantly, there were no sex-dependent differences in either the number of entries it took for animals to achieve over 85% accuracy nor in the maximum achieved accuracy by the end of the transition. This suggests that, while males perseverated slightly less on the previously reinforcing hole, both sexes were similar in the time it took to fully transfer behavior to the newly reinforcing alternative.

Figure 6 also shows the number of pellets received during DL and RL between sexes. While there was no significant difference in the transition of accuracy across entries between sexes (Fig. 6A), there was a significant difference in the total number of pellets received during DL (Fig. 6C) with females receiving more pellets than males. This did not carry over to RL (Fig. 6D) where there was no significant effect of sex on the number of pellets received.

Figure 6 shows the total number of entries animals made during DL and RL. During DL, females made more entries into each of the ports compared to males (Fig. 6E). However, there were no differences between sexes during RL (Fig. 6F). The higher number of entries made by females during DL may correspond to their elevated number of pellets received due to females being more active, and therefore having increased probability of encountering the reinforcer as compared to males.

4. Discussion

The lack of reported sex differences within neuroscience and the larger field of biological medicine despite clear evidence that differences exist is of concern. Characterization of sex differences is essential for advancing understanding of external and internal insults on long-term health outcomes (Beery & Zucker, 2011; Clayton, 2016; Woitowich et al., 2020; Zucker, Prendergast, Beery, 2022). Often female subjects are underrepresented in behavioral studies resulting in complexities when sex differences interact with experimental manipulations. The current study provides further support for the presence of key sex differences in various behavioral modalities in C57BL/6 mice that were also seen across several murine strains and across various ages. Outcomes of this study may have important implications in our understanding of neurobiology and behavior. Differences were observed predominantly in measures of activity with females tending to exhibit higher basal activity than males, with this difference remaining both across the lifespan and between strains. These effects were not only a characteristic of their spontaneous behavior, but also arose in performance on tasks assessing cognition. These results should be taken into consideration for the design and interpretation of behavioral results across studies.

Behavior in the current study was performed in an automated home-cage environment to allow for testing of animals’ natural repertoire of behavior in the absence of daily interference and handling. It is well established that home-cage disruption and animal handling have marked impacts on animal behavior and physiological response (Balcombe, Barnard, & Sandusky, 2004; Gärtner et al., 1980; Neely et al., 2018) that may cloud interpretations of behavior differences conferred not only between sexes, but also with genetic knockouts, exposures, lesions, and other challenges. To address this, C57BL/6 mice in the current study were assayed in young adulthood to determine whether key sex differences in spontaneous behavior exist, what those differences are, and the implications of those differences in tasks of cognition. Sex dependent differences in basal activity were observed in 8wk old mice. Compared to males, female mice tended to have elevated movement kinetics, including movement acceleration and velocity, and activity, including distance moved, bout travel distance, activity duration, and the duration of movement during movement bouts. This elevated activity in females echoes that reported previously, with increased ambulatory and locomotor behavior reported in females tested in traditional studies of motor activity (Blizard et al., 1975; Borbélyová et al., 2019; Knight et al., 2021; Krause et al., 2021; Stevanovic et al., 2022). Interestingly, these effects were not specific to C57BL/6J mice, to age, or to testing site. Activity differences remained throughout the lifespan in C57BL/6J mice tested at a different site and were also observed in genetic SOD1 mutants in a similar pattern. Of note, 8wk old C57BL/6J mice were tested at a site in the US, whereas the other strains (including older C57BL/6J mice) were tested in a facility in Europe, highlighting the reproducibility of these findings across laboratories.

Sex dependent behavioral differences were not unique to the dark phase. During the light phase, males and females differed in basic movement kinetics (such as higher movement acceleration and velocity in females). Further, males and females tended to differ in some rest behaviors, such as the number of entrances, which may suggest some difference in sleep behavior between sexes. However, given the lack of difference in the amount of time spent in the shelter, this is likely not the case. This aligns with what has been seen previously in assays of sleep in rodent models, with reports in mice that show no trend of elevated wakefulness in females during the light phase, nor elevated non-REM or REM sleep, but an elevation in wakefulness in females during the dark phase (Paul et al., 2006).

To expound on the observed alterations in sleep behavior in the current study, previous reports have shown differences in such behaviors between sexes in rodents (Krizo & Mintz, 2015; Iwahana, Karatsoreos, Shibata, & Silver, 2008; Kuljis et al., 2013; Paul et al., 2006). While data of the current study show that sleep duration, approximated by the amount of time animals spent in the shelter, did not differ between males and females, sleep duration is only one measure of sleep behavior. The transition of behavior between the end of the dark phase and the beginning of the light phase, the “anticipation of light”, has also been shown to be associated with circadian behavior as it is associated with cues of light onset (Loos et al., 2014). Males and females varied in measures of anticipatory light behavior, with differences arising for activity duration, time spent in the shelter, number of shelter entrances, and distance moved. Importantly, these measures were opposing between males and females. While males tended to have an increase in activity leading up to the onset of the light phase, seen for activity duration, shelter entrances, and distance moved, in females these measures declined near the end of the dark phase. On the other hand, rest behaviors in males tended to decline, with males spending less time in the shelter at the end of the dark phase, while they increased in females, with females tending to spend more time in the shelter. Taken together, these findings suggest that males become more active at the end of the night prior to sleep, whereas females become less active. This difference is important as it suggests fundamental sex-linked differences in circadian cues that may have implications in interpretation of behavior in murine sleep models.

Various factors may contribute to the robust activity differences seen between male and female mice of this study. One important mediator are the gonadal hormones, with evidence for the role of gonadal hormones on motoric differentiation arising from rodent models of Parkinson’s disease, a motoric disorder typified by deterioration of dopaminergic cell bodies in the substantia nigra (Antzoulatos, Jakowec, Petzinger, & Wood, 2010; Gillies, Murray, Dexter, & McArthur, 2004; Krause et al., 2021; Smith & Dahodwala, 2014). Further, gonadal hormones are shown to interact with circadian processes, seen by a loss of sex-differentiation in non-REM sleep patterns following gonadectomy in mice (Paul et al., 2006). Another key component that could be driving this effect is differences in neurotransmission, namely in the monoamines, choline, or cannabinoids. Evidence to support this comes in sex-dependent alterations in motoric behaviors in other mouse models following an acute or chronic drug challenge with nicotine, psychomotor stimulants, or THC (Caldarone, King, & Picciotto, 2008; Ohia-Nwoko, Haile, & Kosten, 2017; Van Swearingen, Walker, & Kuhn, 2013; Wiley, 2003). This complex interplay between hormonal signaling and neurotransmission may be an important factor dictating basal activity levels between sexes that likely also contribute largely to other behavioral modalities.

Cognitive performance was assessed in C57L/6NTac mice to determine differences between sexes on spatial discrimination learning and reversal. This was an essential component of this assay, as it is of concern whether the motor differences observed in animals during spontaneous behavior could have implications on cognitive task performance. During the discrimination phase, females entered through each of the ports more frequently and received a higher number of pellets despite displaying no significant difference in achieved accuracy. At first glance, the increase in reward pellets received suggests that females performed better than males during the training phase. However, because overall accuracy was not affected this did not appear to be the case. It is more likely that this discrepancy is associated with higher activity in females confounding traditional measures of task performance, such as reinforcement count and accuracy. During the reversal learning phase, there were no differences between sexes on either entries or pellets received despite males acquiring higher accuracy faster than females. This suggests that females perseverate more on the previously reinforcing alternative than males, a behavior that has been shown previously (Chen et al., 2021; Gargiulo et al., 2022). However, the lack of difference in the number of pellets or the number of entries during the reversal phase suggests either 1) a reduction in activity in females after the first two days, which suggests that novelty-induced stress caused by the new home-cage environment contributes to the differences observed during the discrimination phase, or 2) an attenuation associated with the task parameters—e.g., switching to the newly reinforcing alternative reduces the differential in activity between males and females. Evidence exists to support both conclusions (Chen et al., 2021; Domonkos et al., 2017; Borbélyová et al., 2019; Luine, Gomez, Beck, & Bowman, 2017; McLaughlin et al., 2009), suggesting that males and females employ different learning strategies to overcome cognitive challenges. One limitation of the current study is that a trial-by-trial design was not utilized, wherein there is a limit to the number of entries that can occur, and the number of pellets that can be received, which may also contribute to the observed sex-difference. This is because performance in this task is dependent both on accuracy, with more accurate entries resulting in more pellets being received, as well as overall activity, such that more entries will also result in a higher rate of pellets being received regardless of accuracy. This highlights a difference between traditional operant assays where animals spent a very limited amount of time in the apparatus vs. the automated home cage approach where they live in the apparatus. This implicates an important difference in strategy being utilized between closed and open economy designs that may also play a role in design and interpretation of procedures used to assess sex, and other behavioral differences.

5. Conclusion and Limitations

The current study aims to fill gaps in understanding the importance of comparing behavioral differences between sexes in the absence of transient effects of stress induced by handling during task performance by observing behavior in the animals’ home-cage environment. Handling impacts animal behavior (Balcombe et al., 2004; Neely, 2018), with behavioral differences arising due to handling stress being sex-dependent (Stevanovic et al., 2022). Because of this, it is important to assess baseline behavioral phenotypes between sexes in these rodent models to provide a framework for these differences and how they may interact with measures of behavior. While the behavior assessments conducted in this study were without daily handling for task acquisition, one important limitation here is the introduction of stress as these animals were singly housed during the observation period. While this singly housed design was made to account for traditional assays testing animals individually, the prolonged nature of this may confer some differences in behavior output, as seen previously (Guo et al., 2004; Liu et al., 2019), that should be accounted for in future assessments. Despite this, task performance was sex-dependent which could be a confounding factor in attempting to assess the neurobehavioral impacts of drugs, contaminants, lesions, or genetic knockouts. Further, the results of this study are akin to those reported previously, highlighting the reproducibility and generalizability of these differences and the importance of these differences in enhancing understanding of behavioral data in neuroscience. In conclusion, this study uses home-cage observations to show robust and reproducible differences in behavior between female and male mice that are not only important to consider when designing behavioral studies, but that also provide a unique perspective for the interpretation of behavioral differences with aspirations to make more translatable approaches to human disease models.

Highlights.

• Adult female C57BL/6 mice have higher baseline motor activity than males at night in PhenoTyper

• Males have a higher anticipatory change in activity leading up to the light phase compared to females

• Baseline activity differences seen in C57BL/6 mice were also seen in mice across different ages and in different genetic mutants

• Female mice are more active than males during cognition testing in the PhenoTyper resulting in altered task peformance