Estrous cycle effects on behavior of C57BL/6J and BALB/cByJ female mice: implications for phenotyping strategies

Authors


*W. Krezel, Institut de Génétique et de Biologie Moléculaire et Cellulaire (IGBMC), Institut Clinique de la Souris (ICS), BP10142, 67404 Illkirch Cedex, France. E-mail: krezel@igbmc.u-strasbg.fr

Abstract

Systematic behavioral phenotyping of genetically modified mice is a powerful method with which to identify the molecular factors implicated in control of animal behavior, with potential relevance for research into neuropsychiatric disorders. A number of such disorders display sex differences, yet the use of female mice in phenotyping strategies has been a rare practice because of the potential variability related to the estrous cycle. We have now investigated the behavioral effects of the estrous cycle in a battery of behavioral tests in C57BL/6J and BALB/cByJ inbred strains of mice. Whereas the performance of BALB/cByJ female mice varied significantly depending on the phase of the estrous cycle in the open field, tail flick and tail suspension tests, the behavior of C57BL/6J females, with the exception of the tail suspension performance, remained stable across all four phases of the estrous cycle in all of the tests including open field, rotarod, startle reflex and pre-pulse inhibition, tail flick and hot plate. We also found that irrespective of the estrous cycle, the behavior of C57BL/6J females was different from that of BALB/cByJ groups in all of the behavioral paradigms. Such strain differences were previously reported in male comparisons, suggesting that the same inter-group differences can be revealed by studying female or male mice. In addition, strain differences were evident even for behaviors that were susceptible to estrous cycle modulations, although their detection might necessitate the constitution of large experimental groups.

Introduction

Genetic manipulations in mice including transgenesis and targeted gene mutagenesis represent powerful tools in biomedical research. The utility of such genetically modified mice for research into neurological and psychiatric diseases is frequently determined by behavioral analysis. Therefore the choice of behavioral tests and their organization into test batteries has recently attracted considerable interest (Crawley & Paylor 1997; Crawley 1999; McIlwain et al. 2001; Rogers et al. 1997, 1999; van der Staay & Steckler 2001; Voikar et al. 2001). Even though the need for the use of female mice in behavioral screens has been emphasized (Crawley & Paylor 1997; Voikar et al. 2001), it is still a rare practice. However, the use of females is justified because a number of neuropsychiatric diseases, or some of their specific symptoms, display sex differences. For instance, a higher prevalence of stress-related disorders such as certain types of anxiety or depression have been observed in women (Frackiewicz et al. 2000; Sandford et al. 2000), whereas cognitive symptoms in schizophrenia or in attention deficit hyperactivity disorder are more pronounced in males (Gershon 2002; Halbreich & Kahn 2003). Sex differences are also observed in neurological disorders, for which the best example could be late-onset alzheimer disease, which is diagnosed more frequently in women than men (Hofman et al. 1991). The sex differences revealed in behavioral screens of female mice also constitute an additional experimental condition, which may facilitate our understanding of mechanisms of animal behavior.

One of the major reasons for the exclusion of female mice from behavioral testing is the hormonal fluctuation that occurs during the estrous cycle; this could potentially affect animal behavior, thus complicating the data interpretation. A number of studies evaluating the effects of the estrous cycle on animal behavior have focused on rats, whereas relatively few studies have been performed in mice. Furthermore, differences in experimental designs and conditions, as well as age of animals and strains complicate any gross inter-laboratory comparisons. In addition, such studies focused frequently on selected phases of the estrous cycle, e.g. estrous and/or proestrous (Gray & Cooney 1982; Romeo et al. 2003; Ryan & Maier 1988; Sternberg et al. 1994) or estrous, proestrous and metestrous (Mogil et al. 2000) rather than on all four phases (Palanza et al. 2001; Plappert et al. 2005). To evaluate to what extent estrous cycle is actually a confounding factor in behavioral studies, and therefore whether it needs always to be taken into account when carrying out behavioral screens, we have now investigated the effects of the estrous cycle on performance of C57BL/6J and BALB/cByJ females in a battery of behavioral tests. Whereas both of these strains are widely used in behavioral and pharmacological experimentation, the C57BL/6J strain is also used for genetic manipulations and has been a frequent choice as the genetic background for genetically modified animals. Since, we recently showed in cross-laboratory strain comparisons that male C57BL/6J and BALB/cByJ mice differ significantly in a number of behavioral parameters (Hoelter et al. 2004), we also wanted to investigate to what extent these differences exist among female mice, and to evaluate whether the effects of the estrous cycle could confound strain differences in animal behavior. The behavioral test battery used in this study was designed to investigate a wide range of brain functions and their pathology including motor co-ordination, pain sensitivity, anxiety, depression and sensory-motor gating processing. With this objective we chose robust behavioral paradigms that are well established in behavioral experimentation and that we (Hoelter et al. 2004) and others (Crawley & Paylor 1997; Crawley 1999; McIlwain et al. 2001; Rogers et al. 1997, 1999; van der Staay & Steckler 2001; Voikar et al. 2001) have recently used in behavioral test batteries to study strain differences or effects of genetic manipulations in mice.

Materials and methods

Mice and monitoring of the estrous cycle

Before the behavioral studies 20 C57BL/6J female mice were purchased from Charles River at the age of 5 weeks and were used exclusively for analysis of the estrous cycle. No BALB/cByJ mice were used in this phase of the experiments. All mice were housed in a 12 : 12 h light : dark cycle (0700 to 1900 h) in ventilated cages, type ‘MICE’ (Charles River, Lyon, France) in groups of five mice/cage with free access to food and water. Starting from the 8th week of age the rate of spontaneous estrous cycle was evaluated on 14 consecutive days. To this end every day between 1000 and 1100 h a vaginal smear was collected by lavage of the vagina with saline solution using a flame-polished glass pipette. To identify the phase of the estrous cycle vaginal smears were processed for Grunewald–Giemsa staining and analyzed by light microscopy (Leica, Germany). For simplification, the estrous cycle was classified into four principle phases: proestrous, estrous, metestrous and diestrous (Fig. 1), according to Allen (Allen 1922) and as described with precision of sub-phase details by Thung et al. (1956). Briefly, proestrous smears contained only nucleated and cornified cells, estrous smears contained exclusively cornified cells, metestrous smears contained large numbers of leukocytes in clumps or dispersed in smeary mucous with nucleated epithelial cells and some cornified cells, whereas during diestrous, although leukocytes were the predominant type of cells, they were less abundant in the smears than during metestrous and they were accompanied by few epithelial nucleated and cornified cells and pleomorphic cells with large dark stained nuclei.

Figure 1.

Identification of the oestrous cycle phases. The examples of Grunewald–Giemsa preparations of vaginal smear washes from mice in (a) proestrous, (b) estrous, (c) metestrous and (d) diestrous phases of the estrous cycle.

All behavioral analyses were carried out on a total of 100 C57BL/6J and 100 BALB/cByJ virgin female mice, which were purchased from Charles River at the age of 5 weeks in four experimental cohorts and were tested starting from the 8th week of age. The housing conditions were as described (see above) and all behavioral tests were performed between 0900 and 1400 h. To ensure the presence of different phases of the estrous cycle on the day of testing, we induced the estrous cycle using the Whitten effect. To this end we introduced a BALB/cByJ or C57BL/6J male mouse aged between 12 and 20 weeks into a grid-separated compartment in the cage of BALB/cByJ or C57BL/6J female mice, respectively, for 24 h, providing it with food and water. For semi-randomly assigned sub-groups such induction was ended by removal of the male mouse 36 or 72 h, or 1 week before the test. The whole cohort of female mice was tested within two consecutive days and with a 1-week interval between each behavioral test (Fig. 2). The vaginal smear was collected from each mouse after each test session to evaluate the estrous cycle. The efficiency of induction of the estrous cycle was evaluated by calculating the per cent of mice presenting each phase of the estrous cycle in the first cohort of 30 C57BL/6J and 30 BALB/cByJ females used for behavioral tests. The data concerning such a representative group of animals were presented for estrous cycles analysed immediately after each of the three consecutive test sessions, opening the test battery, i.e. open field on week 1, rotarod on week 2 and acoustic startle response and pre-pulse inhibition on week 3 (Table 1). Mice for which the interpretation of the phases of the cycle was ambiguous were not included in the behavioral data analysis. The total number of mice used for behavioral tests and for which estrous cycle was clearly defined is presented in Table 2.

Figure 2.

The flow scheme for behavioral test battery. The behavioral tests were carried out at 1-week intervals and were initiated at the 8th week of age for all mice. Before each test, animals from each strain were randomly assigned to three groups and were exposed to a grid-separated male for 24 h to induce the estrous cycle. The male was removed 36 h, 72 h, or 1 week before testing. Vaginal smears were collected immediately after each test session for evaluation of estrous cycle.

Table 1.  The efficiency of induction of the estrous cycle in mice (Whitten effect)
StrainPost-inductionProestrousEstrousMetestrousDiestrous
  1. The percentage of mice displaying different stages of the estrous cycle are shown for each strain and for each delay after removing the inductor male. Values were calculated from 30 C57BL/6J and 30 BALB/cByJ females for which the estrous cycle was evaluated three times, immediately after: the open field, rotarod and acoustic startle response and pre-pulse inhibition test sessions. n, the number of samples with clearly identified estrous cycle, which were used for calculation of the estrous cycle.

C57BL/6J36 h (= 26)47221813
72 h (= 27)4483513
1 week (= 27)19211842
BALB/cByJ36 h (= 25)2726298
72 h (= 28)1439434
1 week (= 27)8155027
Table 2.  Number of mice used for behavioral testing for which the EC was clearly identified
StainCycleOpen fieldRotarodPre-pulse inhibitionNoci-ceptionTail suspension
C57BL/6Jproestrous1914152414
estrous3524331719
metestrous2026232325
diestrous1620152732
BALB/cByJproestrous1315172318
estrous2427251626
metestrous3236333229
diestrous1413171316

Behavioral procedures

The order of selected behavioral tests (Fig. 2) was carefully defined from the least to the most stressful, taking into consideration data in the literature (McIlwain et al. 2001) and our previous cross-laboratory validation performed as a part of the EUMORPHIA (European Mouse Research for Public Health and Industrial Application, www.eumorphia.org) program (Hoelter et al. 2004). All procedures were carried out in accordance with the European Communities Council Directive of 24 November 1986 (86/609/EEC).

Rotarod test

The test was carried on the semi-automatic rotarod apparatus (Panlab, Barcelona, Spain, www.panlab-sl.com). Mice were given three trials separated by a 5-min to 10-min interval, during which the rotation speed accelerated from 4 to 40 r.p.m. in 5 min. The average latency to fall from the rotarod was used as a measure of motor co-ordination.

The open field test

Mice were tested in automated open-fields (44.3 ×44.3 × 16.8 cm) made of PVC with transparent walls and a black floor, and covered with transparent PVC (Panlab, Barcelona, Spain). The open field arenas were virtually divided into central and peripheral regions. The periphery was 8 cm large and the central zone represents about 40% of the total open field surface. The open fields were placed in a room homogeneously illuminated at 150 Lux. Each mouse was placed in the periphery of the open field and allowed to explore the apparatus freely for 30 min, with the experimenter out of the animal’s sight. The distance travelled and the time spent in the central and peripheral regions were recorded over the test session. The per cent time spent in the center was used as an index of emotionality/anxiety.

Tail suspension test

Mice were tested in the tail suspension apparatus (MED Associates Inc., St Albans, VT; www.med-associates.com). Following an initial period of vigorous struggling after being suspended by the tail on the movement detector, animals gradually abated into immobility. The duration of immobility has been inferred as an index of behavioral despair. Immobility time was scored throughout the 6 min duration of the test.

Acoustic startle reflex and pre-pulse inhibition

Startle reflex reactivity and pre-pulse inhibition were assessed in the same session using eight startle devices (SRLAB, San Diego Instruments, San Diego, CA). The session was initiated with a 5-min acclimation period followed by 10 different trial types: acoustic startle pulse alone (white noise, 120 dB/40 milliseconds, ST120); eight different pre-pulse trials in which either 20-millisecond long 70, 80, 85, or 90 dB stimuli were presented alone or preceding the pulse by 50 milliseconds, and finally one trial (BN) in which only the background noise was presented to measure the baseline movement in the Plexiglas cylinder. The test session began with five presentations of the startle pulse alone that were excluded from statistical analysis. Each acoustic or BN trial was presented 10 times in random order. The average Inter-Trial Interval (ITI) was 15 seconds (10–20 seconds). The startle response was recorded every millisecond for 65 milliseconds after the onset of startle.

Tail flick test

For this test, mice were immobilized and placed in the apparatus consisting of a shutter-controlled lamp as a heat source (Bioseb, Paris, France, www.bioseb.com). Three consecutive trials with an interval of about 1 min were performed at three different sites on the tail. For each trial, the tail of the animal was placed in the bundle of the heating light and the time taken by the animal to flick its tail was recorded (cut-off 20 seconds). The temperature of 65 °C generated by the heat-source was evaluated experimentally using thermometer measurements until stabilization of the temperature.

Hot plate test

Mice were placed individually on a hot plate adjusted to 52 °C (Bioseb, Paris, France, www.bioseb.com) surrounded by a glass cylinder and the latency of the first pain reaction (licking, paw flinches, little leaps) was recorded. The test was ended if the mouse failed to show any sign of pain within 30 seconds to avoid potential injury.

Statistical analysis

A global analysis of data was performed using analysis of variance (anova) with two between-subject factors (strain and estrous cycle) and, if required, one within-subject factor (time or intensity of acoustic stimulus). When a significant effect of estrous cycle (EC) or interaction Strain × EC was observed, post-hoc analyses were performed using a Protected Least Significant Difference (PLSD) Fisher test. For the parameter of the per cent of time spent in the center of the open field, we have also analyzed the effects of the estrous cycle for each strain individually using one-way anova, because large, fivefold inter-strain differences could occlude any estrous cycle effects in one or both strains and lead to lack of significant estrous cycle effect or interaction in the two-way anova model.

Results

Induction and analysis of the estrous cycle

Before the behavioral studies we analyzed the rate and synchrony of spontaneous estrous cycles using a group of 20 C57BL/6J female mice that were not used later for the behavioral analysis. Identification of different phases of the estrous cycle according to vaginal smears (Fig. 1) collected from spontaneously cycling females revealed that mice housed in the same room in groups cycled simultaneously with the exception of one or two mice per cage (data not shown). This variability was increased by the end of 14 days of the testing period. Such synchronization of estrous cycles excluded the possibility of simultaneous testing of females in different phases of the estrous cycle. Hence, in groups of mice used for behavioral analysis we have induced the estrous cycle using the Whitten effect at different time-points before behavioral tests to control the frequency of estrous cycle phases on the day of the test. Such induction appeared to be effective, as illustrated for the first cohorts of 30 C57BL/6J and 30 BALB/cByJ females used for behavioral analysis (Table 1), for which the estrous cycle was evaluated immediately after each of the three first consecutive test sessions, i.e. open field, rotarod and acoustic startle response and pre-pulse inhibition. For the C57BL/6J mice the 36-h latency after removing an inductor male appeared to be the most efficient way to induce proestrous (47% of females), whereas latency of 72 h resulted in a maximum of 48% estrous and 35% metestrous females. One-week after induction there was an increase in the percentage of diestrous females up to 42%. In BALB/cByJ mice also, 36 h after removing the inductor male, proestrous was the predominant phase of the cycle and was identified in 37% of females. Both estrous (39%) and metestrous (43%) were the two most frequent phases 72 h after induction. Finally, 1 week after induction metestrous was the predominant phase (50%), although this latency appeared also the most effective for inducing diestrous phase, which was observed in 27% of females.

Open field

The open field test has been performed on behaviorally naive females, as it was the first test in the test battery. Two-way anova on overall locomotor activity revealed a significant effect of Strain (F[1,165] = 237, < 0.001), whereas the effect of the phase of the estrous cycle or the interaction Strain × EC were not significant. Indeed, irrespective of the phase of the estrous cycle, the C57BL/6J females displayed significantly more locomotor activity than their BALB/cByJ counterparts during the 30 min of the test in the open field (Fig. 3a).

Figure 3.

Open field performance of female mice during different phases of estrous cycle. The distance covered by C57BL/6J and BALB/cByJ mice was represented for the entire 30 min of the test (a, left panel) or as a function of time (a, middle and right panel for each respective strain). The central activity was expressed as per cent of time spent in the center of the apparatus during 30 min of the test (b, left panel) or during 5-min epochs (b, middle and right panels for each respective strain). **< 0.01 and *< 0.05.

Analysis of the percentage of time spent in the center of the open field revealed that although the global estrous cycle effect or Strain × EC interaction was not significant there was a significant interaction of Strain × Time indicating that depending on the strain the exploration of the center evolved differently across the test session (F[5,825] = 29, < 0.001). Indeed, C57BL/6J female mice spent increasingly more time in the center of the open field throughout the testing period, whereas BALB/cByJ mice showed the opposite tendency and by the end of the test they explored the center less or at least at the same rate as at the beginning of the session (Fig. 3b, compare middle and right panel). These data suggest that C57BL/6J mice became less anxious with time, while the BALB/cByJ strain increased its anxiety or alternatively habituated to the arena and thus reduced its motor activity as testing progressed. If considering the mean performance of each group during 30 min, we also observed a significant effect of Strain (F[1,165] = 158, < 0.001). Indeed, irrespective of the estrous cycle the percentage of time spent by BALB/cByJ females in the center of the open field (2.3 ± 0.3%) was significantly lower than that of C57BL/6J mice (12.3 ± 0.6%), suggesting increased anxiety in BALB/cByJ females (Fig. 3b left panel). Considering that such a significant strain difference could occlude any estrous cycle effects or potential Strain × EC interactions, we also analysed each strain individually using one-way anova and identified a significant effect of the phase of estrous cycle on the percentage of time spent in the center of the arena in BALB/cByJ strain (F[3,79] = 3, < 0.05). Such an effect might be related to low anxiety levels in proestrous females (Fig. 3b, left panel), which spent 4.1 ± 1% of their time in the central zone, which was significantly more than 1.5 ± 0.4% for estrous (< 0.01) or 2.1 ± 0.4% for metestrous (< 0.05) females as revealed by Fisher PLSD post-hoc analysis.

Rotarod

The performance on the accelerated rotarod was not affected by the estrous cycle, but a significant difference was found between the two strains (F[1,167] = 16.9, < 0.001). BALB/cByJ mice performed significantly worse than C57BL/6J, displaying shorter latency of 115 ± 3.5 seconds to fall from the rotarod, compared to 139.3 ± 4.3 seconds for C57BL/6J mice (Fig. 4).

Figure 4.

Effects of estrous cycle on locomotor co-ordination. The mean latency to fall from the accelerated rotarod is shown for C57BL/6J and BALB/cByJ females

Acoustic startle reflex and pre-pulse inhibition

The estrous cycle did not affect startle reactivity to acoustic stimuli for either the lower intensities used as pre-pulses or the startling pulse (F[3,170) = 1.3, NS, and F[3,170) = 2.2 NS, for pre-pulses and ST120 respectively; Fig. 5a). However, we found that irrespective of the phase of the estrous cycle, the startle reactivity threshold was significantly different between the two strains (Strain × Stimulus intensity interaction: F[4,680] = 28, < 0.001; Fig. 5a). In C57BL/6J mice the minimal reaction was detected at 80 dB whereas in BALB/cByJ it was detected at 85 dB (< 0.01 vs. baseline activity level, respectively). For pre-pulse intensities below 90 dB, the reactions of the mice ranged between 1 and 10% of the main startle response to the 120-dB pulse depending on the strain and the phase of the estrous cycle. However, for the 90-dB pre-pulse the startle reaction reached around 20%, especially in metestrous C57BL/6J females. There was also a significant strain difference in the startle response to a 120-dB pulse (F[1,170] = 242, < 0.001), with C57BL/6J mice having the lowest score (Fig. 5a).

Figure 5.

Effects of estrous cycle on startle reflex and pre-pulse inhibition of acoustic startle. (a) Startle reactivity to background noise, or to 70, 80, 85, 90 dB acoustic stimulation, and startle reflex to 120-dB stimulus were presented for C57BL/6J and BALB/cByJ females. (b) The per cent of pre-pulse inhibition of the startle response was presented as a function of the pre-pulse intensity. BN, white noise; P, acoustic pulse intensity; ST, acoustic startle to 120 dB; PP, pre-pulse intensity.

Considering the relatively high startle reaction for a 90-dB pre-pulse, this intensity was excluded from pre-pulse inhibition analysis. Pre-pulse inhibition of the startle reflex remained comparable across different phases of the estrous cycle for all pre-pulse intensities (F[3,170) = 1.4, NS) (Fig. 5b), but the global PPI performances were significantly different between the two strains (C57BL/6J = 53.5 ± 1.5, BALB/cByJ = 45.8 ± 1.3; F[1,170] = 7.5, < 0.01). The magnitude of pre-pulse inhibition also evolved differently between the two strains as a function of the pre-pulse intensities (F[2,340] = 8.7, < 0.001). In the C57BL/6J mouse strain, the pre-pulse inhibition level reached a plateau starting from 80 dB, while in the BALB/cByJ mice it increased as a function of pre-pulse intensity and reached the highest level at 85 dB (Fig. 5b).

Tail flick and hot plate performance

Nociception was tested in C57BL/6J and BALB/cByJ females using tail flick and hot plate paradigms. The estrous cycle affected the performance of the two strains differently only in the tail flick test as illustrated by significant interaction for Strain × EC (F[3,167] = 11.5, < 0.001). The post-hoc analysis revealed that BALB/cByJ mice removed their tail more rapidly during proestrous compared to all the other phases of the cycle (< 0.001, Fig. 6a), suggesting that there is increased pain sensitivity during proestrous in this strain. Finally, irrespective of estrous cycle phase the two strains were significantly different in the tail flick (F[1,167] = 22.3, < 0.001) or in the hot plate (F[1,167] = 45.8, < 0.001) paradigms. In particular, C57BL/6J displayed significantly lower pain thresholds (4.4 ± 0.2 and 10.2 ± 0.2 seconds) than BALB/cByJ females (5.3 ± 0.1 and 13.6 ± 0.5 seconds) in the tail flick and hot plate tests, respectively.

Figure 6.

The effect of the estrous cycle on nociception. The performance in the tail flick (a) and hot-plate tests (b) is shown for C57BL/6J and BALB/cByJ mice. ***< 0.001, significantly different from other phases of the estrous cycle of the same strain.

Tail suspension

The performance of female mice in the tail suspension test was analysed to evaluate despair-related behavior and the findings are presented in Fig. 7. A global analysis of data revealed a significant effect of estrous cycle (F[3,171 = 3.2, < 0.05) and a significant interaction for Strain × EC (F[3,171 = 3, < 0.05), suggesting that the effects of the estrous cycle were different depending on the strain. The C57BL/6J females displayed the longest immobilization of 197 ± 8 seconds during the metestrous phase, which was significantly longer than 169 ± 10 seconds of immobility during proestrous (< 0.05) or 164 ± 8 seconds during diestrous (< 0.01) phases. Interestingly, BALB/cByJ females exhibited the longest immobilization of 236 ± 9 seconds during the estrous phase, which was significantly longer than 196 ± 9 seconds during proestrous (= 0.01), 199 ± 8 seconds during diestrous (< 0.05) or 200 ± 8 seconds during metestrous phase (< 0.01). Despite marked effects of the estrous cycle on the immobility time in both strains, we found that irrespective of the phase of the estrous cycle, BALB/cByJ females displayed significantly more despair behaviors than C57BL/6J females because they had longer immobility scores: 177.3 ± 4.5 for C57BL/6J and 209.6 ± 5.5 for BALB/cByJ (F[1,163] = 30.6, < 0.001).

Figure 7.

Estrous cycle effects on despair behavior in the tail suspension test. The total immobility during the 6-min test is shown for C57BL/6J and BALB/cByJ mice. **< 0.01; *< 0.05.

Discussion

We have used a battery of behavioral tests to evaluate and compare the effects of the estrous cycle on a number of brain functions in C57BL/6J and BALB/cByJ mice. Before such tests, we found that the majority of females housed in the same room displayed the same cycle phases, which excluded the possibility of comparing the behavioral effects of different phases of the cycle during the same testing sessions. To obtain homogeneous representation of different phases of the estrous cycle for each experimental session, we have taken advantage of the Whitten effect to induce estrous cycle at different latencies before testing. Induction carried out 36 h, 72 h and 1 week before testing appeared to be a relatively reliable method for modulating the estrous cycle in C57BL/6J and BALB/cByJ females and allowed the generation of cohorts that were representative of different phases of the cycle.

The current behavioral analysis revealed that estrous cycle effects were most robust on despair-like behaviors because both strains showed significant effects of the estrous cycle on the immobility time in the tail suspension test. Interestingly, the cycle affected each of the strains differently. The longest immobility, suggestive of increased despair behavior was displayed by metestrous females in C57BL/6J strain, whereas in BALB/cByJ strain, such an increase was observed during estrous when compared to other phases of the cycle. The origin of such strain-specific estrous effects on despair-like behaviors is not known.

The estrous cycle also affected anxiety-sensitive measures, although such effects were not as robust as for despair-like behaviors because they concerned only the BALB/cByJ strain. An increase of exploration of the center of the open field, suggestive of reduced anxiety, was observed during the proestrous phase in BALB/cByJ females as compared to estrous or metestrous phases. Reduced anxiety during the proestrous phase could also be suggested by a strong tendency, although not significant, for a high number of entries and short latency for the first entrance to the center of the open field, which were displayed by proestrous as compared to estrous and metestrous BALB/cByJ females (data not shown). The lack of any estrous cycle effect on anxiety-sensitive measures in C57BL/6J mice agrees with previous studies in this strain using standard and modified open field paradigms (Palanza et al. 2001; Romeo et al. 2003). Thus an anxiolytic nature of the proestrous phase might be strain specific, although this needs to be further confirmed with other anxiety-sensitive tests. Interestingly, the anxiolytic effect of proestrous has also been reported in rats (Frye et al. 2000; Marcondes et al. 2001; Mora et al. 1996). The significant effect of the estrous cycle was also evident in other behavioral parameters. Thus, for BALB/cByJ females, the analgesic effect of the proestrous phase was illustrated by short latency to remove the tail from the heat source in the tail flick test of nociception, as compared with performance in other phases of the cycle. These data are in agreement with some of the rat data (Frye et al. 1993), although opposite observations (Martinez-Gomez et al. 1994; Molina et al. 1994) were also reported and may originate from the methodological variability between studies (Mogil et al. 2000). In contrast to the tail flick paradigm, no estrous cycle effect was observed in the hot plate test, which is also used to evaluate thermal nociception.

Interestingly, behaviors in several tasks were insensitive to the effects of the estrous cycle in both strains, including motor co-ordination on the rotarod, hot plate performance, acoustic startle reflex and pre-pulse inhibition. The latter findings are consistent with a recent report by Plappert et al. (2005). Although opposite findings were reported in rat studies (Koch 1998) any direct interspecies comparisons are difficult because of differences in stimulus and/or sensitivity settings.

We have also found that irrespective of the estrous cycle effects, the global performance of C57BL/6J females was different from that of BALB/cByJ female mice. Thus, C57BL/6J females appeared more active in the open field test and covered longer distances than their BALB/cByJ counterparts. In addition, BALB/cByJ females displayed increased measures of anxiety in this test compared to the C57BL/6J group. Interestingly, studies of male mice carried out in our experimental settings (Hölter et al. 2004) and other laboratories (Belzung & Berton 1997; Hölter et al. 2004) provided evidence that the BALB/cByJ males are more anxious than their C57BL/6J counterparts. This similarity between strain comparisons of female and male mice strongly suggests, that strain differences in female behavior do not result from variability related to the estrous cycle, but reflect true strain differences. Strain differences were also observed in the task for motor co-ordination. The low performance of BALB/cByJ females in the rotarod test was reminiscent of data from male comparisons (Brooks et al. 2004; Hölter et al. 2004; McFadyen et al. 2003). Similarly, the higher threshold of pain detection in the hot plate and tail flick tests displayed by BALB/cByJ in comparison with C57BL/6J females, was also reported from male studies (Hölter et al. 2004; Lariviere et al. 2002). Strain differences were also evident in the startle response, which was significantly higher in BALB/cByJ female mice than in C57BL/6J mice. An analogous difference between C57BL/6J and BALB/cByJ mice was reported from our laboratory (Aubert et al. 2006; Hölter et al. 2004) and others (Brooks et al. 2004; Hölter et al. 2004). In contrast, BALB/cByJ females displayed overall poor pre-pulse inhibition performances as compared to C57BL/6J females. Again, such strain difference was previously reported in BALB/cByJ and C57BL/6J male mice although less marked which may reflect gender-dependent strain differences in sensory motor gating. Interestingly, the despair behaviors in the tail suspension test, despite robust effects of the estrous cycle, were more pronounced in BALB/cByJ than in C57BL/6J female mice. In fact, similar data on despair behaviors were obtained in male studies in the tail suspension test (our unpublished data).

In conclusion, behavioral phenotyping of female mice can be carried out without the risk of running into confounding effects of the estrous cycle, although the choice of behavioral paradigms and/or genetic background is critical for such an approach. In particular, our observations suggest that, with the exception of the tail suspension test and the tail flick paradigm, C57BL/6J female mice were not susceptible to estrous cycle phase in any of the measured parameters in the open field, rotarod, startle reflex and PPI and hot plate paradigms. In contrast to C57BL/6J mice, the BALB/cByJ females were more prone to behavioral modulations by estrous cycle, although their performance in several tests remained insensitive to it. In addition, our findings, based on strain comparisons between C57BL/6J and BALB/cByJ females, suggest that behavioral phenotyping performed blind to estrous cycle phase is susceptible to reveal differences in those behaviors that are subject to modulations by estrous cycle. Such an approach might necessitate the monitoring of the estrous cycle or constitution of larger experimental cohorts to homogenize distribution of different estrous phases, which should prevent false-negative (e.g. the tail suspension performance of C57BL/6J mice is not different from BALB/cByJ females in most of the estrous cycle phases, despite a clear overall strain difference) or false-positive results (e.g. dominant effect of metestrous phase in tail suspension test in C57BL/6J strain).

Acknowledgments

We thank Prof. Pierre Chambon for his support and comments on the manuscript. This work was financially supported by the Institut National de la Santé et de la Recherche Médicale (INSERM), the Centre National de la Recherche Scientifique (CNRS), the Hôpital Universitaire de Strasbourg, the Collége de France, the Institut Universitaire de France and the Fondation Fyssen.

Ancillary