Genetic and behavioral differences among five inbred mouse strains commonly used in the production of transgenic and knockout mice


G. Bothe, Taconic Albany, One University Place, Rensselaer, NY 12144, USA. E-mail:


Five strains of mice commonly used in transgenic and knockout production were compared with regard to genetic background and behavior. These strains were: C57BL/6J, C57BL/6NTac, 129P3/J (formerly 129/J), 129S6/SvEvTac (formerly 129/SvEvTac) and FVB/NTac. Genotypes for 342 microsatellite markers and performance in three behavioral tests (rotorod, open field activity and habituation, and contextual and cued fear conditioning) were determined. C57BL/6J and C57BL/6NTac were found to be true substrains; there were only 12 microsatellite differences between them. Given the data on the genetic background, one might predict that the two C57BL/6 substrains should be very similar behaviorally. Indeed, there were no significant behavioral differences between C57BL/6J and C57BL/6NTac. Contrary to literature reports on other 129 strains, 129S6/SvEvTac often performed similarly to C57BL/6 strains, except that it was less active. FVB/NTac showed impaired rotorod learning and cued fear conditioning. Therefore, both 129S6/SvEvTac and C57BL/6 are recommended as background strains for targeted mutations when researchers want to evaluate their mice in any of these three behavior tests. However, any transgene on the FVB/NTac background should be transferred to B6. Habituation to the open field was analyzed using the parameters: total distance, center distance, velocity and vertical activity. Contrary to earlier studies, we found that all strains habituated to the open field in at least two of these parameters (center distance and velocity).

During the past decade, knockout and transgenic mice have offered great insight into basic biological mechanisms. More and more of these strains are continually being distributed from centralized repositories (Mutant Mouse Regional Resource Centers [MMRRC,], European Mouse Mutant Archive [EMMA,], Jackson Laboratories [] and Taconic Farms []). When describing the phenotype of these knockout and transgenic mice, researchers frequently include behavioral analyses. However, behavior is strongly influenced by genetic background, which makes it difficult to interpret the results of these behavioral findings. In gene targeting experiments, 129 strains are almost exclusively used, whereas C57BL/6 is often used for backcrossing at a later stage of the experiment. For pronuclear microinjection, FVB and C57BL/6 are frequently used. Transgenic mice made in FVB are often backcrossed to C57BL/6. Thus, behavioral characterization of the 129, C57BL/6 and FVB strains is vital for interpretation of the behavior of knockout and transgenic mice. Furthermore, as there are a number of different suppliers of these strains, it is imperative to determine, via genetic microsatellite analysis, their relatedness. Data on genetic polymorphisms among substrains are scarce for the 129 strain, and do not exist for the different C57BL/6 substrains. Thus, evidence of genetic divergence among substrains is important for interpretation of behavioral analysis of transgenic and knockout mice.

We therefore set out to characterize the genetic background and behavior of five inbred strains of mice (129S6/SvEvTac (formerly 129/SvEvTac, Festing et al. 1999), 129P3/J (formerly 129/J), C57BL/6NTac, C57BL/6J and FVB/NTac). A set of microsatellite markers was analyzed for size differences. For the behavioral characterization, we chose three tests that can be run at relatively high throughput, are at least partly automated and cover a wide variety of behavior parameters. We used the rotorod for testing motor performance and motor learning (Crawley 2000; Dunham & Miya 1957), an open field activity monitor for measuring exploratory behavior (Bolivar et al. 2000) and an automated monitor to measure contextual and cued fear conditioning behavior (Bolivar et al. 2001).

In this study, genetic and behavioral data were directly compared for the first time, the use of multiple parameters of open field behavior was introduced into inbred strain comparisons, and it was shown (contrary to earlier literature reports) that all mouse strains tested habituate to an open field, given enough exposures. We also reported on the performance of C57BL/6NTac and FVB/NTac, which had not been tested previously. Together our genetic and behavioral investigations provide a unique opportunity to compare the relationship between genes and behavior in the inbred mouse strains most frequently used in knockout and transgenic research.

Materials and methods


Mice of the strains 129S6/SvEvTac, C57BL/6NTac and FVB/NTac were obtained from the production colony at Taconic Farms (Germantown, NY) at eight (129S6/SvEvTac & C57BL/6NTac) or four (FVB/NTac) weeks of age, and maintained at the behavioral testing facility. 129P3/J and C57BL/6J mice were obtained from The Jackson Laboratory (Bar Harbor, ME), quarantined for six weeks at the Taconic quarantine facility, and transferred to the behavior testing facility. Standard mouse cages (28 × 17 × 12 cm L×W×H) were used to house four mice per cage. Mice were kept on a 12/12 light/dark cycle, and experiments were performed between two and 10 hours after lights-on (06:00). Care was taken to distribute experiments and subjects evenly throughout the day, and to test all subgroups both in morning and afternoon experiments. Mice were tested between 10 and 16 weeks of age, starting with rotorod testing. A total of 143 mice were tested (12–17 per sex per strain). All animals were handled and tested by the same technician.

Sizing of microsatellite markers

DNA samples were obtained from suppliers for each of the strains tested in our behavioral assays. Whole-genome marker scans were carried out on two mouse genomic DNA (gDNA) samples from each strain tested using a microsatellite marker set of 412 dinucleotide repeat markers developed at the Whitehead Institute/MIT Genome Center, and purchased from Research Genetics, Inc. (Huntsville, AL). This set covered all autosomes and the X chromosome. One DNA primer of each pair used to amplify microsatellite markers using PCR was fluorescently labeled with either HEX, 6-FAM or TET fluorophore. PCR amplifications were performed using a standard cycling protocol (95 °C/2 min; 40 × [94 °C/30 seconds, 55 °C/30 seconds, 72 °C/45 seconds]; 72 °C/2 min), Qiagen Taq polymerase (Qiagen, Valencia, CA), TaqStart Antibody (BD ClonTech, Palo Alto, CA; 14:1 molar ratio antibody:DNA polymerase, respectively), and 50 ng gDNA template. Platinum Genotype Tsp DNA polymerase (Invitrogen Life Technologies, Carlsbad, CA) was occasionally substituted for Qiagen Taq in instances where non-template nucleotide addition hindered accurate allele size determination.

Allele sizes for amplified microsatellite markers were determined using an Applied Biosystems (Foster City, CA) Model 373 DNA Analyzer with accompanying analysis software. Diluted aliquots of PCR products were multiplexed post-PCR (six per sample) and allowed to air dry overnight in the dark. Labeled PCR products were then mixed with TAMRA-labeled in-lane molecular weight standards (Applied Biosystems) and formamide, denatured at 92 °C for 2 min, then loaded into wells of 8% ‘Long Ranger’ polyacrylamide, 8.33 M urea gels. Allele sizes (in base pairs) of amplified markers were determined after gel electrophoresis using ABI Prism GeneScan (v. 2.1.1) and GenoTyper (v. 2.1) software. Quality of results was judged for each marker. Markers that performed poorly with regard to artifacts, signal strength, ‘stutter’ (n-2×m artifacts), and sample-to-sample consistency of signal strength were repeated once. Markers that failed in the repeat reaction were not included in the analysis.


The rotorod test was the first test to be performed in all animals. An Accurotor Rotorod (Accuscan, Columbus, OH) was used, which is comprised of a rod of 3 cm diameter that is separated into four compartments by acrylic discs. Mice were placed on the rod facing forward, and the rod was started after all four mice had been placed on it. The rod was accelerated from 0 to 15 r.p.m. during the first minute and then run at constant speed (15 r.p.m) for two more minutes. Each mouse was tested three times per day (early morning, late morning and afternoon) on three consecutive days. The latency to fall was determined by infrared beams. A ‘passive rotation’ was defined as a mouse clinging to the rotorod for more than one full rotation, and thus staying on the rod without balancing. In the experiments reported here, no mouse was able to perform passive rotations. This is in contrast to other studies (McFadyen et al. 2003) and may be due to the relatively low speed of rotation we used. To make the comparison to previously published results easier, we calculated a ‘motor learning index’ as D3/(D1 + D3), where D1 is the average latency to fall on day 1, and D3 is the average latency to fall on day 3 (Cook et al. 2002). This learning score normalizes for first-day performance.

Activity and habituation to an open field

Five days following the rotorod test, mice were tested for open field activity. Animals were tested following the procedure outlined in Bolivar et al. (2000). Briefly, each mouse was weighed, placed in a holding cage for five minutes, and then placed in the center of a dark (less than 0.1 lux) Digiscan activity monitor (Accuscan, Columbus, OH) for a five-minute testing period. Mice were tested on four consecutive days. The following variables were examined directly from the data collected by the system: total distance traveled, centre distance, number of vertical beam breaks, movement time. Later, velocity was calculated as total distance/movement time, and percentage center distance as 100 × center distance/total distance. To ease comparison to earlier publications, an ‘activity index’ or activity change ratio was calculated for each animal and each activity measure as Dn/(D1 + Dn), where D1 is the distance traveled, percentage or velocity, on day 1, and Dn is the corresponding day (n = 2, 3 or 4) value (Cook et al. 2002). This activity change ratio was originally developed by Nadel (1968) to measure intrasession habituation. Thus, values were normalized for initial activity and could be compared more easily among strains.

Contextual and cued fear conditioning

Because of its aversive nature, contextual and cued fear conditioning was the last test performed on the mice (4–6 days after the open field test). The procedures we used have been described in detail previously (Bolivar et al. 2001). In that study, several control groups were used (e.g. presentation of the tone (CS) and shock (US) in isolation) to determine the effect of US-CS pairing on performance. This protocol has been used in several publications since then (Bolivar et al. 2002; Bolivar et al. 2003; Cook et al. 2002), and therefore was followed here as well. The San Diego Instruments Freeze Monitor (San Diego Instruments, San Diego, CA) was used. The freeze monitor system delivers an electrical foot shock of 0.5 mA (as measured by the manufacturer) as US and a tone (90 dB) as CS. Movements were recorded by infrared beams spaced 2 cm apart, and a per-minute average of beam breaks was calculated. This activity measure was used throughout the experiment to describe baseline (naïve) activity, activity in altered environment, and contextual and cued session performance. After each test, each Freeze Monitor was cleaned with 70% isopropanol to reduce residual stress odors. Each Freeze Monitor was placed in an isolating cabinet, which was equipped with a small fan producing a background noise. The light level in the box was low, i.e. less than 0.1 lux.

Day 1

Animals were placed into the freeze monitor after one-hour acclimatization to the testing room and left for 150 seconds to explore the test chamber. Activity during this time was considered baseline activity. For conditioning, three pairings of tone (3000 Hz) and foot shock were delivered (pairings were spaced 150 seconds apart). The tone (CS) had a duration of 5 seconds, and the foot shock (US) followed immediately after the CS (0.0 seconds interval, 0.5 seconds duration). All mice showed a strong reaction to foot shock and reduced their activity significantly.

Day 2

On the next day, mice were returned to the test chamber and monitored for five min without any tone or shock being delivered. The average activity during this period was used as a measure of fear conditioning to context, and was compared to the baseline activity on Day 1. Two hours later, the test chamber was altered by replacing the grid floor with sheet plastic, putting striped panels on the walls, adding orange oil in a cup outside the test chamber, and switching off the fan of the isolation cabinet. Each mouse was placed into this altered context and activity was measured. After 180 seconds, 240 seconds and 300 seconds, the conditioning tone was delivered without foot shock. The average activity during this 3-min period was calculated, divided by the activity in the altered environment (prior to the presentation of the conditioning tone), and used as a measure of cued conditioning.

Statistical analyses

Statview and JMP software (SAS Institute, Cary, NC) was used for statistical analyses. Effects and interactions were calculated by repeated-measure anova, and by one-way anova where appropriate. For strain comparisons, learning or activity indices were calculated (see above) and t-tests were performed. The Bonferroni correction was used to adjust the cutoff P-value for the significance of multiple comparisons. A cutoff value of P = 0.005 was used for the rejection of the null hypothesis when five strains were compared, as there are 10 possible comparisons of five strains.


Genetic differences: microsatellite markers

Of 412 microsatellites that were in the original marker set, 342 amplified reliably and were chosen for further analysis. The five mouse strains could be placed into three groups based on the number of discrepant markers: FVB/NTac, the two 129 strains and the two C57BL/6 substrains (Fig. 1). The two 129 strains differed in 33 markers, and 13 of these differences were 6 bp or larger. This is consistent with genetic contamination by crosses with other strains, as has been previously reported (Simpson et al. 1997). The C57BL/6 strains differed in only 12 microsatellites, and there were no differences larger than 4 bp. This is consistent with them being true substrains that have differed by point mutations only. Microsatellite mutations occur by ‘slippage’ of the DNA polymerase during DNA replication, and occur at a high rate of 10-3-10-5 (Ellegren 2000; Webster et al. 2002; Yue et al. 2002). Assuming that the two strains were separated for 148 generations (Taconic Farms 2003), and that 25% of mutations are fixed (brother/sister mating scheme), a microsatellite mutation rate of 4.6×10−4 per generation was calculated, which is within the range reported in the literature.

Figure 1.

 Genetic differences between mouse strains shown as number of discrepant microsatellite markers. A total of 342 markers were tested. The two C57BL/6 strains: C57BL/6J and C57BL/6NTac have the smallest differences. See text for additional discussions. Abbreviations: 129P3/S6 = 129P3/J vs. 129S6/SvEvTac, B6J/B6Tac = C57BL/6J vs. C57BL/6NTac, B6Tac/129S6 = C57BL/6NTac vs. 129S6/SvEvTac, FVB/B6Tac = FVB/NTac vs. C57BL/6NTac, 129S6/FVB = 129S6/SvEvTac vs. FVB/NTac.

Based on the genetic differences, we asked whether there would be similar differences in behavior, and tested the same inbred strains in three behavior experiments: rotorod, open field behavior and fear conditioning.


The initial performance on the rotorod, i.e. the average latency to fall on day 1, differed widely among individual mice, but did not appear to be greatly influenced by strain (Fig. 2, first bar of each group). There was a main effect of sex: females showed a longer latency than males (F133,1 = 25.5, P < 0.0001; average latency of males, 17 seconds; females, 28 seconds). However, no main effect of strain and no strain × sex interaction were observed. As female mice on average have a smaller weight than males, we were interested in a possible effect of weight on rotorod performance (Brown et al. 2002; Cook et al. 2002; McFadyen et al. 2003). Inclusion of weight as an additional anova parameter did not improve the model (data not shown). Also, ‘passive rotations’, i.e. mice clinging to the rod and being carried around passively, did not occur. Due to the slow rotation speed, mice were unable to complete a passive rotation and fell off the rod.

Figure 2.

 Latency to fall on the rotorod. Shown are means ± SEM for day 1 through day 3, collapsed over sex.

There was significant motor learning in all strains tested (Fig. 2). Learning scores (D3/(D1 + D3)) were between 0.559 and 0.651, and the probability of the null hypothesis was 0.0043 or lower for each strain analyzed separately (data not shown). There was a significant influence of strain on motor learning (F4,133 = 4.1, P = 0.004, Table 1), but no influence of sex and no strain × sex interaction. The strain with the highest learning score was C57BL/6NTac (score 0.651); the strain with the lowest learning score was FVB/NTac (0.559). To analyze whether strain differences were significant, we calculated t-tests and applied the Bonferroni correction to the strain comparisons (Table 1). FVB mice differed significantly from two other strains, 129S6/SvEvTac (0.636, P = 0.004) and C57BL/6NTac (P = 0.0005). There was no significant difference between the two B6 strains, C57BL/6J and C57BL/6NTac.

Table 1.  Strain differences in test performance
  1. Behavioral performance index values were determined and compared by Bonferroni t-tests as described in Materials and methods. For each pair of strains, only those tests are listed that differ significantly. RRod: motor learning on the rotorod; VACT: vertical activity in the open field on day 1; TD-1: total distance traveled in the open field on day 1; TD-h: habituation of total distance traveled in the open field; % CD1: % of distance traveled in the center on day 1; CD-h: habituation of percentage center distance; VELO: velocity on day 1; CON: context test in fear conditioning; CUE: cued test in fear conditioning. There were no strain differences in the habituation of velocity.

 TD-1TD-h TD-h TD-hTD-1 
  CD-h   CD-h  
  CUE    CON 
   TD-1 TD-1 TD-1 
C57BL/6J       VACT
C57BL/6NTac      RRodVACT

Open field behavior

A comparison of the open field behavior of the five mouse strains is shown in Figs 3 and 4, and significant strain differences are summarized in Table 1. To allow better comparisons to other papers (Bolivar et al. 2000; Cook et al. 2002), total distance, and not total beam breaks, was analyzed in detail. Total beam breaks mirrored total distance quite closely (data not shown).

Figure 3.

Activity in the open field on day 1 (mean ± SEM). Left: Total distance (cm) traveled in the open field on day 1. Right: Vertical activity (number of beam breaks). Males (M) and females (F) of the five strains are shown separately.

Figure 4.

Habituation to the open field (mean ± SEM). Top: total distance (TOTDIST) on days 1–4; middle: center distance (% CTRDIST); bottom: velocity (VELO), on day 1 through 4. See Table 1 for strain comparisons.

Open field activity on day 1

Overall, there was a main effect of strain (F4,132 = 40.4, P < 0.0001), a main effect of sex (F1,132 = 12.0, P = 0.0007) and a strain × sex interaction (F4,132 = 9.1, P < 0.0001) for total distance traveled on the first day of testing. Generally, FVB mice were the most active, and 129S6/SvEvTac were the least active (see Fig. 3).

All strains showed a preference for the periphery of the open field and were less active in the center (Fig. 4, first bar of each group). The center activity, measured as percentage of total distance traveled in the center, was significantly influenced by strain (F4,132 = 6.3, P = 0.0001), sex (F1,132 = 8.8, P = 0.004), and by the strain × sex interaction (F4,132 = 3.2, P = 0.02). FVB/NTac mice were most active in the center of the field, whereas 129S6/SvEvTac were least active (see Table 1 for strain differences).

The velocity of movement was calculated as the quotient of total distance moved by total time the animal was moving (not resting). This measure showed significant strain differences (F4,132 = 58.5, P < 0.0001). There was no main effect of sex on velocity, but there was a significant strain × sex interaction (F4,132 = 4.8, P = 0.001). Velocity had the smallest within-strain variability of all parameters analyzed (cf. error bars in Fig. 4). FVB/NTac mice moved most quickly, whereas the two 129 strains were the slowest (see Table 1 for strain differences).

Vertical movements (rearing and leaning) are not contained in the distance parameters and form a separate class of movements. The vertical activity is depicted in Fig. 3. Three groups of responses are apparent: the 129 strains have low vertical activity, B6 strains have medium activity and the FVB strain has high activity. In an anova, there were significant strain differences in vertical activity (Fig. 3, F4,133 = 119.6, P < 0.0001), but sex and the strain × sex interaction had no influence on vertical activity. There were no significant differences between 129P3/J and 129S6/SvEvTac, and between C57BL/6J and C57BL/6NTac. All other strain pairs were significantly different (see Table 1).

Habituation in the open field

Habituation is defined as any change in response due to repeated exposures. Crusio and Schwegler (1987) were the first to use a similar definition when they defined habituation as any change in activity from Day 1 to Day 2. It has been shown previously (Bolivar et al. 2000), that the total distance traveled can increase, decrease or remain unchanged when mice are exposed repeatedly to the same open field. As can be seen in Fig. 4, there were differences in habituation among the five strains tested. A repeated-measures anova shows a significant habituation effect on total distance traveled (F3,134 = 6.1, P = 0.0006) and a significant habituation × strain interaction (F12,355 = 5.0, P < 0.0001), whereas there was no habituation × sex, or habituation × sex × strain interaction. We then calculated the activity indices (activity change or habituation scores) to compare the five strains; an activity index of 0.5 indicates unchanged activity (Nadel 1968). The total distance index for day 4 was less than 0.5 in C57BL/6J (P = 0.006) and C57BL/6NTac (P = 0.0002), but not significantly different from 0.5 in 129S6/SvEvTac, 129P3/J and FVB/NTac. If the lack of habituation in these three strains would be indicative of poor memory, one would expect other measures of open field activity to be unchanged as well. We therefore analyzed the change in percentage center distance and velocity.

The distance traveled in the center of the open field, expressed as percent of total distance, decreased strongly from day 1 to day 4 (Fig. 4). This change was highly significant (F3,135 = 23.3, P < 0.0001), and there was a significant interaction with strain (F12,357 = 2.3, P = 0.009, see Table 1 for strain comparisons), but not with sex, or with strain × sex. When strains were analyzed separately, only 129P3/J showed no change in percentage center distance between day 1 and day 4.

Interestingly, the velocity of movement, calculated as the total distance/time animal was moving, increased in all strains (Fig. 4). The effect of habituation on velocity was highly significant (F3,133 = 28.7, P < 0.0001) and all strains showed increased velocity when analyzed separately (t ≥ 3.0, P ≤ 0.005). However, there was no influence of strain, sex or strain × sex interaction on the habituation of velocity.

Finally, we examined whether the three groups detected in the genetic background screening: 129 strains, B6 substrains and FVB, could also be found in the open field response. 129 strains showed the lowest vertical activity, B6 a medium one and FVB/NTac was most active. In the horizontal activity measure, total distance, 129P3/J was more similar to the B6 substrains. The habituation behavior of 129P3/J was quite different from 129S6/SvEvTac, whereas it resembled FVB/NTac most closely. Thus, the two B6 substrains did not differ significantly in any parameter tested, the 129 strains showed some differences and FVB was quite different from the other strains.

Contextual and cued fear conditioning

The results of the contextual and cued fear conditioning experiments are shown in Figs 5 and 6. For analysis and strain comparisons, context and cued learning were calculated as ‘context activity during test trial/baseline activity’ and ‘activity after test tone in altered environment/activity before test tone in altered environment’, respectively. Mice of all five strains were able to learn (defined as a suppression of activity) both the context and the cued parts of the fear-conditioning paradigm (P < 0.0001 for all subjects, P = 0.001 when strains were tested separately). Context learning was significantly influenced by strain (F4,133 = 13.8, P < 0.0001), but there was no main effect of sex or strain × sex interaction. Strains differed significantly from each other (see Fig. 5 and Table 1). Particularly, 129S6/SvEvTac reacted more strongly to context than any other strain except 129P3/J, and FVB/NTac reacted more weakly than 129S6/SvEvTac, 129P3/J and C57BL/6J.

Figure 5.

Contextual fear conditioning. Mice were exposed to the same context as during the training session, but no shock or tone was applied. Activity is shown as percentage of baseline activity in the training trial (mean ± SEM). A low value is indicative of fear conditioning to context. Two C57BL/6NTac showed an aberrant behavior; they reacted to the context by increased activity.

Figure 6.

Cued fear conditioning. Mice were exposed to an altered context. Baseline activity in this altered context was recorded, and the same sound was displayed as during the training session. Activity after the cue is shown as percentage of activity before cue in altered context (mean ± SEM). A low value is indicative of fear conditioning to cue.

For the cued fear conditioning data, there was a significant main effect of strain (F4,129 = 26.8, P < 0.0001), and also of sex (F1,129 = 9.71, P = 0.002) and a strain × sex interaction (F4,129 = 5.1, P = 0.0008). Bonferroni t-tests showed that 129S6/SvEvTac mice reacted more strongly to the cue than any other strain (P = 0.0002), and that FVB/NTac mice reacted more weakly than C57BL/6J and C57BL/6NTac (P < 0.0001, Fig. 6 and Table 1). There was no difference between the two B6 substrains.


In this paper, we examined the genetic and behavioral diversity of five inbred mouse strains commonly used in transgenic and knockout production. Based on our data, the five strains: 129P3/J, 129S6/SvEvTac, C57BL/6J, C57BL/6NTac and FVB/NTac, can be grouped into three genetic groups: 129, B6 and FVB. The microsatellite data (see for entire dataset) obtained in this study will be useful for accelerated backcrossing experiments and for the quality control of inbred lines. Whereas the genetic divergence of the 129 strains (129P3/J, 129S6/SvEvTac) found in our study supports previously reported data (Simpson et al. 1997), nothing has been reported on the relatedness of substrains of C57BL/6. Our data demonstrate that C57BL/6J and C57BL/6NTac are true substrains, and that neither has been accidentally crossed to other inbred mouse lines. Even after separate breeding for 148 generations, only 12 microsatellite polymorphisms were detected. The two strains differ by only two or four base pairs at these loci, which is consistent with single mutation events during which the DNA polymerase ‘slipped’ by one or two repeat units during replication. Assuming that 25% of all spontaneous mutations are fixed in a brother/sister mating scheme, we calculated a microsatellite mutation rate of 4.6×10−4 per generation, which is within the range reported in the literature (Brinkmann et al. 1998; Ellegren 2000; Yue et al. 2002). Given the microsatellite data, we asked whether there are similar differences in behavior, with the two B6 strains being more similar than other strains.

In the rotorod test, we distinguished between initial performance and motor learning. We found no significant strain differences in initial performance. However, female mice performed better than males overall, which is consistent with previous studies (McFadyen et al. 2003; Tarantino et al. 2000). In contrast to previous studies (Brown et al. 2002; Cook et al. 2002; McFadyen et al. 2003), we found that weight had only a marginal effect on rotorod performances. These apparent inconsistencies may be the result of protocol and apparatus differences across laboratories, as parameters such as the diameter and material covering the rod, speed and rate of acceleration of the rod and the number of trials can all have large effects on rotororod performance (Rustay et al. 2003a; Rustay et al. 2003b). In agreement with previous findings (Cook et al. 2002; McFadyen et al. 2003; Tarantino et al. 2000), all of the inbred strains learned the rotorod task, i.e. they significantly increased latency to fall over trials. However, there were strain differences, with FVB mice displaying less motor learning than other strains. There were no significant differences in motor learning between the two B6 strains.

Activity levels in the open field found in this study are generally compatible with published data (Bolivar et al. 2000; Paulus et al. 1999): FVB/NTac mice were highly active and 129S6/SvEvTac were less active than B6 mice. However, FVB/NTac mice were more active than previously reported for another substrain, FVB/NJ (Bolivar et al. 2000), even though the two substrains have been separate for only about 55 generations (Taconic Farms 2003). Further study is needed to investigate whether there is a significant difference between the two substrains, or whether this is due to different experimental conditions. Female 129P3/J mice were also more active than reported previously (Cook et al. 2002), and female C57BL/6J were more active than males. It is interesting to note that those two strains were bred at The Jackson Laboratory and went through six weeks of quarantine before being tested, which may have influenced results. There is always the possibility that environmental differences prior to testing could be at least partially responsible for behavioral differences seen in this study. Although Crabbe et al. (1999) reported that different ‘shipping histories’ of mice produced ‘almost no effects’, the potential for effects due to extended quarantine need further study.

When habituation to the open field during the four test sessions was analyzed, we found that habituation of total distance, percentage of distance traveled in the center and velocity did not covary. Therefore, one might speculate that habituation to a new environment is a complex process. Parallel to a decrease in true exploration, a habitual activity pattern is established, which includes constant or even increased activity on the margins of the open field, while the center is visited less, and the speed of movement during excursions increases as well as the time the animal spends resting. In a natural habitat, the animal would use this habitual activity to check periodically whether new food sources or mates appeared in the territory. The activity within a small area such as the one used in this experiment would constitute only a very small part of the mouse's daily routine. Open spaces would be avoided for fear of predators. This hypothesis would lead to a number of predictions, e.g. that activity should reach a plateau after a number of days, that center activity and its decrease over time should be correlated to measures of anxiety, that path stereotypy should increase over time, etc. Further studies varying these parameters might provide interesting information.

In both contextual and cued fear conditioning, even though all five inbred mouse strains displayed activity suppression, there were inbred strain differences. The strain differences of contextual and cued fear covaried, FVB/NTac reacted least and 129S6/SvEvTac reacted most strongly in both tests. This general pattern has been reported in other studies (Bolivar et al. 2001; Cook et al. 2002). However, there are important differences in the response to tone and context. Mice in this study responded less strongly to both context and cue than in a previous study using the same apparatus (Cook et al. 2002). As we ran the test at a very low light level (less than 0.1 lux) this may have reduced the response, but may also have made differences in the reaction to context more prominent. Thus, we once again find that small protocol differences can have significant effects on interlaboratory reliability, even when the same apparatus and basic protocol are used.

Generally, there were significant differences in performance in all three behavioral tests. FVB/NTac mice learned poorly on the rotorod and in the contextual and cued fear-conditioning paradigm. They habituated to the open field by increasing margin activity and decreasing center activity. 129S6/SvEvTac mice learned well on the rotorod and in the fear paradigm while having low open field activity. C57BL/6J and C57BL/6NTac mice showed good rotorod learning as well, reacted less in the contextual and cued fear test and had medium to high open field activity. Thus, performance in the three tests was not very tightly correlated, but high activity seemed to reduce the context and cued fear responses. This may be due to motivational or general activity level differences.

129S6/SvEvTac and both B6 strains appear to be equally well suited for the behavior screens described in this paper. 129S6/SvEvTac is also a relatively good breeder (Taconic Farms 2003). When a knockout mouse is developed on this 129 strain, it appears advisable to keep a population of knockout mice with pure 129S6/SvEvTac background and to start phenotype testing with this population. However, it would be prudent to start backcrossing to other strains in parallel to these experiments, because some experiments may require a different genetic background. Also, data from two different genetic backgrounds would allow for more general conclusions about the actual effects of the target gene, as well as provide information about any modifier genes.


We would like to thank Jen Earl for help with establishing the behavior tests in our laboratory at Taconic, Dr Lorraine Flaherty (Genomics Institute) for help with the initial grant proposal, and Dr Robert Keefe (Genomics Institute) for sizing microsatellite markers. This work was supported by NIH grant 1 U42 RR14820-01 (‘Mutant Mouse Regional Resource Centers’).