Determination of the estrous cycle phases
We compared the behaviors between males and females and also between females in different phases of the estrous cycle, as it is well known that both human and rodent females alter their behavior depending on the estrous cycle phase [
21‐
24].
For all female mice, estrous cycle phase was determined by vaginal smear cytology analyses during the 4 days prior to handling. Briefly, we rinsed the vagina with 150–200 μL sterile water. The smear was placed on a glass slide (FRC-01, Matsunami Glass industries, Osaka, Japan). After drying, 50 μL of Giemsa stain solution (Merck, Tokyo, Japan) was applied to the smear, which was left to stand for 10–20 min and then washed with distilled water. After drying, the smear was observed under a light microscope (Nikon Diaphot 300, Nikon Corporation, Tokyo, Japan), and we then classified it as either proestrus, estrus, diestrus [
25], or ‘not determined’ (nd) depending on the results of the analysis. To control for any behavioral effects of this procedure between sexes, males were also treated in the same way, with sterile water applied under the scrotum.
Behavioral tests
Two-way active avoidance test
Initially, we examined whether any sex differences occurred in active avoidance. In this test, 14 male and 16 female (3 proestrus, 6 estrus, and 7 diestrus) mice were used. The mice were required to learn the association between an auditory cue and a nociceptive foot-shock stimulus, and to then avoid the foot-shock by perceiving the auditory cue, across trials. The experimental procedure was based on a previous study [
26]. Briefly, mice were placed in 1 of 2 adjacent compartments, separated by a partition, in a shuttle box (height = 185 mm, width = 300 mm, depth = 115 mm; Passive Avoidance System, Bio-Medica, Osaka, Japan). Constant luminance was maintained in both compartments. Immediately after the partition was removed, the mice could move freely between the two compartments. One minute after placement, the auditory cue was presented for 5 s in the shock compartment. The final 2 s of cue presentation were accompanied by the foot-shock. This procedure was repeated for 100 trials. The inter-trial interval was set at 20 ± 3 s. The active avoidance rate was defined as the number of entries into the safe compartment across 100 trials.
To ensure that the foot-shock did not disrupt their behavior, the current intensity of foot-shock for males and females was set at their pain threshold, which was determined in a pilot study (males = 0.11 ± 0.005 mA, females = 0.09 ± 0.003 mA). The threshold was determined as the average of individual thresholds within the group, and these were measured as the minimum current that induced a jumping response when the intensity was gradually increased from 0.089 mA by manually adjusting the current controller.
Passive avoidance test
In the active avoidance test, animals faced a threatening context in which their mobility (the active movement between compartments) constituted an adaptive behavior to avoiding a threat. Next, we performed the passive avoidance test to test whether any sex differences existed in a converse situation where the subject’s immobility (staying in one compartment) would be more adaptive. Therefore, if a difference in the avoidance pattern between the sexes was observed in the active avoidance test, we predicted that the passive avoidance test would show the opposite result.
In the passive avoidance test, 11 male and 18 female (3 proestrus, 8 estrus, 6 diestrus, and 1 nd) mice were examined. While the experimental procedure was based on a previous study [
27], the foot-shock current intensity was set at the pain threshold for both males and females similar to the active avoidance test (see above). The same shuttle box was used as in the active avoidance test, except that one compartment was darkened while normal illumination was maintained in the other.
This test comprised of training and test sessions. In the training session, mice learned the association between a dark compartment and foot-shock that would enable them to anticipate the upcoming foot-shock. The two compartments were initially separated by a partition. After a mouse was placed in the light compartment, the partition was removed. When the mouse entered the dark compartment, the partition was closed again. Ten seconds after entry, a foot-shock was applied for 2 s. Ten seconds later, the partition was opened and the mouse returned to the light compartment. Twenty-four hours after the training session, the test session was initiated. The partition was removed at the beginning of the session, and a mouse was then placed in the light compartment. The mouse could then move freely between the two compartments, and its behavior was recorded over 700 s. Since the latency to enter and the number of entries into the dark compartment should each reflect avoidance, both were measured from the recorded video. The latency ceiling was fixed at 700 s.
Fear-conditioning test
To test sex difference in associative learning, and how it may contribute to any observed differences in avoidance, we performed a fear-conditioning test based on a protocol [
28].
Twelve male and 31 female (8 proestrus, 9 estrus, 10 diestrus, 4 nd) mice were used. All stimulus presentation was computer-controlled (Image FZC, O’Hara & Co., Tokyo, Japan). This test comprised of four sessions: training, contextual fear-conditioning (CXT), pre-auditory-cued fear-conditioning (pre-AUD), and auditory-cued fear-conditioning (AUD). The training session was carried out on day 1, and the remaining three sessions were conducted the next day.
In the training session, mice learned the association between a foot-shock and an accompanying cue. Mice were placed individually in an operant chamber (width = 120 mm, height and depth = 110 mm; O’Hara & Co.). After placement, contextual (a mixture of implicit cues, such as odor, field-view, and sound, in the chamber) cues were presented for 30 s. The last 2 s of auditory (10 kHz, 75 dB) cue presentation were accompanied by a 0.75 mA foot-shock delivered from stainless steel bars on the floor. After the foot-shock, the mice were returned to their home cages. After the training session, it was expected that mice could anticipate the upcoming foot-shock whenever either cue was presented.
Twenty-four hours after the training session, mice were subjected to the CXT session. They were re-exposed to the same chamber and the same contextual cues, as in the training session. They spent 180 s in this chamber with neither an auditory cue nor a foot-shock.
Two hours later, the same mice were subjected to the pre-AUD and AUD sessions. They were placed in a novel chamber, which formed a triangular prism (side = 110 mm, height = 120 mm, O’Hara & Co.), for 360 s. To mask the olfactory cues, the chamber was cleaned with sodium hypochlorite before and after each use. The mice were re-exposed, 180 s after placement, to the same auditory cue as in the training session (AUD). To ensure that the mice were subjected to the AUD session without generalized contextual fear carried over from the preceding session, we also measured their response in the first 180 s (pre-AUD).
We measured freezing as a conditioned response in each mouse, as it is recognized as a behavioral index of associative learning. Behavior in the chamber was recorded at one frame per second, and a freezing response was defined as any instance when a difference of pixel intensities between two successive frames was less than 30% (Image FZC).
Light/dark and light/light test
In the active and passive avoidance tests, the perception of potential threats prompts the subsequent avoidance. We expected that if the perception of potential threats was influenced by a sex difference in both types of avoidance, the difference would be diminished in the absence of the perceived threat. Therefore, we carried out the light/dark (L/D) and light/light (L/L) tests as additional contexts. In the L/D and L/L tests, mice did not encounter the threat or the threat-predicting cue, respectively. The sessions and the measured behavior were the same as in the passive avoidance test (see above).
In the L/D test, 12 mice (6 male and 6 female) were used. The chamber was the same as in the passive avoidance test, except that the foot-shock was never applied. In the L/L test, 36 mice (6 male and 30 female) were used. The chamber was the same as in the active avoidance test, except that the auditory stimulus was never presented.
Tail suspension test
We tested whether a sex difference arises in the escape component using the tail suspension test (TST) and the forced swim test (FST). In both tests, mice faced an imminent threat, rather than a potential one.
The TST was used to examine 12 male and 17 female (5 proestrus, 6 estrus, 4 diestrus, 2 nd) mice. The experimental procedure was based on previous studies 29–31]. Briefly, mice were suspended by their tails from a fixture 35 cm above the floor for 360 s. Their behavior was recorded and analyzed by motion analysis software (Image FZC).
Since immobility indicates how individuals attempt to escape behavior from an imminent threat [
13,
14,
32‐
34], we measured the immobility of the trunk, excluding slight movements of the limbs and tail, and calculated the total time spent immobile (immobility time) for data analysis. This procedure was conducted daily on 2 consecutive days.
Forced swim test
The FST was used to examine 12 male and 18 female (4 proestrus, 4 estrus, 7 diestrus, and 3 nd) mice. The experimental procedure was based on previous studies [
29‐
31]. For 360 s, mice were placed in a glass beaker (diameter = 135 mm, height = 200 mm; Hario Glass Co., Tokyo, Japan), which was filled to a height of 130 mm with water at 25 ± 1 °C. Behavior in the pool was recorded and analyzed using Image FZC software. Immobility was measured as passive floating with slight movements of the limbs and tail, and the immobility time was calculated. This procedure was conducted daily on 2 consecutive days.
Measurement of basal activity in familiar home cage
We checked whether a sex difference arises in the basal activity in the familiar home cage in the absence of any threat. This test was used to examine 8 male and 8 female mice. Each mouse was housed in a separate standard laboratory home cage (width = 203 mm, height = 118 mm, depth = 133 mm) for 3 days. On the 4th day, their activity was measured for 1 min with a camera (1.3 million pixels, viewing angle of 78°, 30 frames/s). The mean velocity, total travel distance, and distance from the center were calculated from the track point on the body by Motion Analyzer (version 1.4.21.0, Keyence Co. Ltd., Osaka, Japan).
Data analysis
In all tests, we recorded 1–3 behavioral measures per test for each mouse, and then calculated the group mean with a 95% confidence interval (CI) for each of the following groups: males, females, and females in each estrous phase. We excluded subjects that exceeded 2 standard deviations of the group mean. For statistical analysis, females in specific estrous phases and all females group were treated as independent groups whenever either was compared with males. We performed Shaffer’s modified sequentially rejective Bonferroni procedure as a multiple comparison whenever we found statistical significance in 1- or 2-factorial analysis of variance (ANOVA). Alpha was set at 0.05. We calculated the effect size [Pearson’s
R on
t test and the multiple comparisons, generalized eta-squared (η
G
2
) on ANOVA] where η
G
2
is a recommended effect size statistic for various experimental designs [
35], including the mixed design used in the present study. These effect size statistics are available indices that are standardized to quantify the practical magnitude of the relationship between independent and dependent variables, independent of the sample size, and thereby, enable results to be compared with other studies [
35]. The magnitude of the effect size was interpreted as either small, medium, or large in accordance with the standard guidelines [
36,
37]. The sex ratio (SR) = male/female * 100 was calculated from the group mean of males and that of females in each test. Statistical analysis was done by customized codes in MATLAB (MathWorks Inc., MA, US).
In the active avoidance test, we calculated the avoidance rate in 5 bins of 20 trials for each mouse, and the group mean for males, females, and females in each estrous phase. We performed a 2-way [sex (males, females) × bin (1st, 2nd, 3rd, 4th, 5th)] ANOVA for the avoidance rate. To compare males with females in each estrous phase, another 2-way [sex (males, proestrus, estrus, diestrus) × bin (1st, 2nd, 3rd, 4th, 5th)] ANOVA was performed.
In the passive avoidance test, we calculated the latency to enter and the number of entries into the dark compartment for each mouse in the test session, and then, the group mean for males, females, and females in each estrous phase. We performed a 2-sample t test (males vs. females) and a 1-way [sex (males, proestrus, estrus, diestrus)] ANOVA for latency. A 2-sample t test (males vs. females) for the number of transitions was also performed.
In the fear-conditioning test, we calculated the percentage of time spent freezing (freezing rate) for each mouse in each session, and then, the group mean for males, females, and females in each estrous phase. We performed a 2-way [sex (males or females) × session (training, CXT, pre-AUD, AUD)] ANOVA for the freezing rate. To compare males with females in each estrous phase, another 2-way [sex (males, proestrus, estrus, diestrus) × session (training, CXT, pre-AUD, AUD)] ANOVA was performed.
As in the passive avoidance test, we performed a 2-sample t test (males vs. females) for latency in the L/D and L/L tests.
In both the TST and FST, we calculated the immobility time for each mouse on both days. The group mean was then calculated for males, females, and females in each estrous phase. We performed 2-way [sex (males or females) × day (1st or 2nd)] ANOVA for immobility time. To compare males with females in each estrous phase, another 2-way [sex (males, proestrus, estrus, diestrus) × day (1st or 2nd)] ANOVA was performed.
For measurement of the basal activity in the familiar home cage, mean velocity, total travel distance, and distance from the center of the cage were measured. After the individual mean and the group mean were calculated, a 2-sample t test (males vs. females) was performed for each motion parameter.