- Top of page
- Materials and Methods
Although genetic background is acknowledged as a potentially important determinant of mutant phenotypes, publications on genetically modified mice far outnumber those on progenitor strains. We have recently reported major differences in basal anxiety levels (elevated plus-maze & light/dark exploration) among three strains (C57BL/6JOlaHsd, 129/SvEv and 129S2/SvHsd) employed as progenitor stock in European laboratories (Rodgers et al. in press). Furthermore, the phenotypes of these inbred strains differed significantly from that of an outbred strain (Swiss-Webster) commonly used in behavioural pharmacology. In view of these findings, the present study assessed possible differences in the anxiolytic efficacy of chlordiazepoxide (0, 7.5 & 15.0 mg/kg, IP) in three of these strains (Swiss-Webster (SW), C57BL/6JOlaHsd (C57) & 129S2/SvHsd (129)). Experimentally naive mice were exposed to the elevated plus-maze, sessions were videotaped and behaviour analysed using ethological software. The performance of control subjects confirmed significant strain differences in basal levels of activity (SW > C57 > 129) and anxiety-related behaviours (129 = SW > C57), with hypolocomotion dominating the 129 profile. SW mice displayed an anxioselective response to both doses of chlordiazepoxide (CDP), with significant reductions in open arm avoidance and risk assessment observed in the absence of any change in general activity. In direct contrast, the lower dose of CDP (7.5 mg/kg) was without effect in either inbred strain, whereas treatment with 15.0 mg/kg induced a profile indicative of muscle relaxation/mild sedation in C57 mice and virtually abolished all behavioural activity in 129 mice. Although the absence of an anxiolytic response to CDP in C57 mice may be attributed to their low basal anxiety levels, the profile of 129 mice strongly suggests an abnormality in benzodiazepine/GABAA receptor function. The implications of these findings for research on mutant mice are discussed.
In this context, current research in behavioural genomics faces a number of conceptual and methodological problems, not least of which are the interpretative difficulties posed by genetic background (Crawley 1996; Crawley & Paylor 1997; Gerlai 1996, 2001; Gingrich & Hen 2000). While gene targeting is typically conducted on embryonic stem cells from one mouse strain (usually a 129 substrain), another (usually C57BL/6 J) is employed for blastocyst donation and subsequent breeding programmes (Crawley 2000; Gerlai 2001). As such, the phenotypic effect of a targeted mutation may be confounded by polymorphic alleles contributed by the two specific progenitor strains employed. Further genetic differences between the parental strains may also affect the phenotype of interest and, if linked to the targeted gene (i.e. flanking alleles), may cosegregate to themselves influence the mutant phenotype (Gerlai 2001; Homanics et al. 1999; Quinlan et al. 2000). As the genes linked to the mutation will be 129-derived in knock-out animals, but C57-derived in wild-type controls, it may be impossible to ascribe the mutant phenotype to an effect of the targeted mutation as opposed to an effect of the cosegregating genes. Such difficulties are of particular concern when the phenotype of the mutant closely resembles that of the 129 parent (Gerlai 1996, 2001) and, indeed, several examples exist where behavioural phenotypes originally attributed to targeted mutations have subsequently been found to reflect characteristics inherent to the 129 parental strain (e.g. Kelly et al. 1998; Le Roy et al. 2000). It is, therefore, essential not only to fully characterise the behavioural phenotypes of common background strains (Crawley 1996; Gerlai 1996, 2001; Rogers et al. 1999; Tarantino et al. 2000), but also to directly compare these phenotypes (behavioural & pharmacological) with those of the knockouts and their wild type controls (Contet et al. 2001; Homanics et al. 1999 Quinlan et al. 2000).
The study of emotionality in mice has been dominated by tests involving spontaneous exploration of novel environments, e.g. the open field, light/dark exploration & elevated plus-maze paradigms (Belzung & Griebel 2001; Holmes 2001; Rodgers 1997, 2001). Over the past 50 years, such tests have provided unequivocal evidence for major phenotypic differences among commonly used inbred mouse strains (Agmo et al. 1999; Anisman et al. 2001; Beuzen & Belzung 1995; Crawley & Davis 1982; Griebel et al. 2000; Lepicard et al. 2000; Mathis et al. 1994; Ohl et al. 2001; Oliverio et al. 1973; Peeler & Nowakowski 1987; Sudak & Maas 1964; Thompson 1953; Trullas & Skolnick 1993; Van Gaalen & Steckler 2000). Fortunately, most of these studies have included the C57BL/6J strain in their comparative profiling, with the general consensus that these animals are more active and less anxious relative to most other strains (for review: Crawley 2000). More recently, however, research has begun to focus on the comparative behavioural phenotypes of C57BL/6J and 129 mice. Despite the existence of multiple 129 substrains (Simpson et al. 1997), it has been widely reported that these animals are much less active, and arguably, more anxious than their C57BL/6J counterparts (Bolivar et al. 2000; Carter et al. 2001; Contet et al. 2001; Cook et al. 2001; Crabbe et al. 1999; Holmes et al. 2002; McIlwain et al. 2001; Rogers et al. 1999; Tarantino et al. 2000; Voikar et al. 2001). These consistent findings raise some difficult issues, particularly in view of reports that high fear/anxiety is a commonly observed phenotype in knockout mice (Belzung & Griebel 2001; Clement et al. 2002; Holmes 2001; Nelson & Young 1998).
In our own laboratory, we have recently confirmed major phenotypic differences between C57BL/6JOlaHsd, 129/SvEv and 129S2/SvHsd mice in two commonly used models of anxiety, the elevated plus-maze and light/dark exploration tests (Rodgers et al. in press). As reported by others, the two 129 substrains were less active and displayed higher levels of anxiety-related behaviour (as assessed by conventional and ethological measures), findings that clearly have important implications for the interpretation of mutant phenotypes in these tests. Of equal importance, however, all three inbred strains studied were found to differ behaviourally from the outbred Swiss-Webster strain. Thus, while C57 mice displayed lower levels of activity and anxiety in both paradigms, the profile of 129 mice (both substrains) was characterised by profound hypolocomotion associated with some behavioural differences consistent with a higher level of anxiety. The significance of phenotypic comparisons with the Swiss-Webster strain lies in the fact that outbred mice (and Swiss-Webster in particular) have been widely used in the validation of exploration-based animal models of anxiety and are routinely employed in contemporary research on the behavioural pharmacology of anxiety (Rodgers 2001; Van Gaalen & Steckler 2000). It is, therefore, pertinent to note that the field of behavioural genomics is increasingly turning to pharmacological strategies for more refined hypothesis testing. Recent examples of the potential value of this approach include the demonstration of altered benzodiazepine sensitivity in mice with targeted disruption of the serotonin (5-HT)1A receptor (Olivier et al. 2001; Sibille et al. 2000) and various subunits of the GABAA receptor complex (Crestani et al. 1999; Low et al. 2000; Mohler et al. 2002; Rudolph et al. 1999). However, the pharmacological sensitivity of any behavioural test (and anxiety models are no exception) is highly dependent upon behavioural baselines, with ‘ceiling’ and ‘floor’ effects of major concern (Rodgers 1997, 2001).
In view of the above evidence for major strain differences in basal anxiety levels, and the considerable literature on strain differences in sensitivity to benzodiazepine receptor ligands (e.g. Belzung et al. 2000; Crawley & Davis 1982; Desforges et al. 1989; Griebel et al. 1993, 2000; Kopp et al. 1999; Lepicard et al. 2000; Mathis et al. 1994; Nutt & Lister 1988; Ohl et al. 2001; Weizman et al. 1999), the present study compares the behavioural effects of chlordiazepoxide in Swiss-Webster, C57BL/6JOlaHsd and 129S2/SvHsd mice exposed to the elevated plus-maze. In addition to this primary aim, the results should also clarify whether the commonly observed behavioural phenotype of 129 mice in exploration-based tests truly reflects high anxiety or merely hypolocomotion. An ethological scoring technique was used to provide more detailed behavioural profiles of drug action than is possible using only the conventional measures of open arm avoidance.
A preliminary report of this work has been presented to the 2002 Annual Meeting of the British Association for Psychopharmacology.
- Top of page
- Materials and Methods
As noted by many authors (Crawley 1996; Gerlai 1996, 2001), and eloquently emphasised by Holmes et al. (2002), background gene effects are an important consideration in transgenic and knockout studies both as a determinant of baseline performance and as a potential false positive influence on mutant phenotypes. However, widely reported strain differences in responsivity to benzodiazepine receptor ligands (e.g. Belzung et al. 2000; Crawley & Davis 1982; Desforges et al. 1989; Griebel et al. 1993, 2000; Kopp et al. 1999; Lepicard et al. 2000; Mathis et al. 1994; Nutt & Lister 1988; Ohl et al. 2001; Weizman et al. 1999) would strongly suggest that background genes may also have an important influence on the pharmacological phenotypes of mutant animals (Homanics et al. 1999; Quinlan et al. 2000). In view of recently reported differences in benzodiazepine sensitivity in mice with targeted mutations of 5-HT1A receptors and various subunits of the GABAA receptor complex (e.g. Crestani et al. 1999; Sibille et al. 2000), the present study was designed to investigate possible differences in responsivity to the anxiolytic effects of CDP in C57BL/6JOlaHsd and 129S2/SvHsd mice. These strains, frequently used in Europe as progenitor stock, are known to differ in basal plus-maze anxiety levels (Contet et al. 2001; Rodgers et al. in press). The inclusion of the outbred Swiss-Webster comparator was based on their extensive use in studies on the behavioural pharmacology of anxiety (Van Gaalen & Steckler 2000) and, importantly, a significantly different plus-maze phenotype relative to the two progenitor strains under investigation (Rodgers et al. in press).
Taken in isolation, these data might suggest that the 129 mice were simply less active than their C57 counterparts. However, as extreme fear or anxiety can suppress general activity/exploration (e.g. Blanchard et al. 1993), it is possible that the hypoactive profile of 129 mice reflects a greater anxiety response to the threats posed by the plus-maze. Although no differences were found for percentage open arm entries (reflecting the significant suppression of both open and closed arm entries), 129 mice did spend substantially (albeit non-significantly) less time than C57 mice on the open arms. Furthermore, consistent with the large amount of time spent on the centre platform, and with previous findings (Rodgers et al. in press; Rogers et al. 1999), they displayed very much higher levels of active risk assessment (total stretched attend postures) and higher protected levels of both this behaviour and head-dipping. Thus, relative to the C57BL/6JOlaHsd phenotype, the plus-maze profile of the 129S2/SvHsd strain is characterised by hypolocomotion, avoidance of both the open and (particularly) the enclosed arms relative to the centre platform, and high levels of risk assessment (predominantly from the centre platform towards open and enclosed arms). Although this behaviour pattern would not be inconsistent with a reluctance of 129 mice to actively explore a potentially dangerous novel environment (i.e. higher anxiety), perhaps a more parsimonious interpretation is that it simply reflects the observed strain difference in basal activity levels. Pharmacological analysis therefore assumes central importance in helping to distinguish between these alternate interpretations of the 129 profile (see below).
Consistent with previous observations (Rodgers et al. in press), the plus-maze phenotypes of the two inbred strains were also found to differ significantly from that of Swiss-Webster mice. More specifically, C57 animals displayed significantly higher scores than SW for all measures of open arm activity (i.e. open arm entries, percentage open arm entries and percentage open arm time) as well as lower scores for active risk assessment (total SAP) and the protected forms of head-dipping and SAP. This profile is considered to reflect a lower basal level of anxiety in C57 mice since, despite a lower locomotor activity score (closed arm entries) than SW (1) no strain difference was observed on total arm entries and (2) C57 mice actually showed significantly higher levels of open arm entries, rearing and head-dipping. In contrast, 129 animals differed significantly from SW only on measures of general activity, with lower levels of total arm entries, closed arm entries and rearing. Although these mice also had a higher score than SW for percentage open arm entries, this finding is considered spurious in that it clearly reflects the significant reduction in closed arm entries rather than an increase in open arm entries. Overall, present findings support the conclusion that, whereas C57 mice are less active but also less anxious than SW mice (see also Griebel et al. 2000; Rodgers et al. in press; Van Gaalen & Steckler 2000), the major difference between the 129 and SW strains is best described in terms of the profound hypoactive profile of the former.
The currently observed strain differences in basal plus-maze performance have important implications not only for the interpretation of the behavioural phenotypes of mutant animals but also their pharmacological phenotypes. The latter point is dramatically illustrated through the highly contrasting effects of CDP in the three strains. As expected (e.g. Cao & Rodgers 1997; Holmes & Rodgers 1999), Swiss-Webster mice displayed a significant and behaviourally selective anxiolytic response to benzodiazepine treatment. At both doses examined (7.5 and 15.0 mg/kg), CDP increased total and open arm entries, percentage open arm entries and percentage open arm time. Importantly, these effects were observed in the absence of a significant effect on the principal index of plus-maze locomotion (i.e. closed arm entries). Furthermore, characteristic of a range of benzodiazepine receptor ligands in both the mouse (Cao & Rodgers 1997; Cole & Rodgers 1993; Cole et al. 1995; Holmes & Rodgers 1999Johnson & Rodgers 1996) and rat (e.g. Griebel et al. 1996) plus-maze, CDP enhanced total head-dipping (7.5 mg/kg), decreased percent protected head-dipping (both doses) and dose-dependently reduced risk assessment (total SAP). In marked contrast to this profile, CDP failed to selectively reduce indices of anxiety in either C57BL/6JOlaHsd or 129S2SvHsd mice. This absence of anxiolytic response to CDP in the two background strains was not due to an inappropriate dose range, as 7.5 mg/kg failed to exert any behavioural effects while 15.0 mg/kg produced significant changes in some measures.
In C57 mice, 15 mg/kg CDP failed to alter the principal indices of open arm avoidance (open arm entries, percentage open arm entries and percentage open arm time), a profile most parsimoniously attributed to the low basal level of anxiety of this strain under present test conditions. A similar ‘floor’ effect has been encountered in many other studies in which C57 mice have failed to show (or have only weakly shown) an anxiolytic response to CDP and/or diazepam (e.g. Agmo et al. 1999; Belzung et al. 2000; Griebel et al. 1993, 2000; Kopp et al. 1999; Lepicard et al. 2000; Ohl et al. 2001). Nevertheless, it is pertinent to note that the C57BL/6 J strain is also relatively insensitive to the convulsant effects of the benzodiazepine receptor inverse agonists, FG7142 and methyl-β-carboline-3-carboxylate (e.g. Mathis et al. 1994; Nutt & Lister 1988) and, unlike BALB/c mice, fails to display an anxiolytic-like response to the benzodiazpeine receptor antagonist, flumazenil (Belzung et al. 2000). This differentiation in pharmacological phenotype, particularly between C57BL/6J and BALB/c mice, may be due to significant differences in benzodiazepine receptor density (whole brain; Robertson 1979; amygdala, Hode et al. 2000). However, it is important to note that, in the present study, C57 mice were not completely insensitive to the behavioural effects of CDP. Although the reductions in risk assessment (total SAP) and the percentage protected form of head-dipping seen at 15 mg/kg would not be inconsistent with a mild anxiolytic action, this dose also significantly increased grooming and reduced both rearing and total head dips. As none of the latter actions is typically associated with benzodiazepine-induced anxiolysis in the plus-maze, the overall pattern of behavioural change seen at this dose level is more congruent with a muscle relaxant or borderline sedative action.
To our knowledge, there are no existing reports on the anxiolytic efficacy of benzodiazepines in any 129 substrain. In the present study, 15 mg/kg CDP almost completely abolished behavioural activity in 129S2/SvHsd mice, a truly remarkable profile given the total lack of effect observed with the lower dose. Closed arm activity (entries and time) and rearing were completely eliminated from the behavioural profile, while head-dipping and risk assessment (total SAP) were reduced to near-zero levels. Furthermore, although percentage open arm entries were increased to 100%, this change can be directly ascribed to the elimination of closed arm entries and, therefore, a pure mathematical artefact. Detailed observation of videotapes indicated that, with initial placement on the centre platform of the maze (constant orientation = facing an open arm), these mice invariably remained in that area for some time occasionally stretching forward (SAP) towards the facing open arm before backing onto the opposite open arm. This pattern of behavioural change suggests a very strong sedative-like action of CDP (15.0 mg/kg) in 129S2/SvHsd mice, an effect clearly not seen in either Swiss-Webster or C57BL/6JOlaHsd animals. Importantly, the complete absence of an anxiolytic-like response to CDP (7.5–15.0 mg/kg) in 129 mice implies that their basal hypolocomotor profile in the plus-maze is not due to a high basal level of fear/anxiety. Although it remains theoretically possible that anxiety-related behaviour in this strain is dependent upon benzodiazepine-insensitive (possibly serotonergic) mechanisms, present results suggest that 129S2/SvHsd mice may have a major abnormality in GABAA receptor mechanisms. More specifically, it is currently believed that the anxiolytic effects of benzodiazepines involve α2-GABAA receptors (Low et al. 2000), whereas their sedative effects are mediated by α1-GABAA receptors (Rudolph et al. 1999). As such, the behavioural profile of CDP in 129S2/SvHsd mice (i.e. sedation only) indicates that this substrain either has a deficiency or subsensivitity in α2-GABAA receptors or an overexpression or supersensitivity in α1-GABAA receptors. If independently verified by receptor binding studies, and in other 129 substrains, this abnormality would seem to have considerable significance for basic research strategies in behavioural genomics.
In summary, two inbred mouse strains (C57BL/6JOlaHsd & 129S2/SvHsd) commonly used (at least in Europe) as background stock for gene targeting research differ markedly both in basal plus-maze performance and in responsivity to a reference benzodiazepine (CDP). These findings have important implications for the interpretation of mutant phenotypes, particularly those that have been created on a mixed 129 × C57 background. Firstly, as the progenitor strains themselves differ in basal plus-maze phenotype, care must be taken to ensure that the mutant phenotype is not confounded by polymorphic (or cosegregating linked) alleles contributed by the parental strains (Gerlai 1996, 2001; Homanics et al. 1999). This is a particularly important issue in anxiety research where the vast majority of knockout strains displaying alterations in anxiety-like behaviour have been characterised as ‘anxiogenic’ relative to their wild-type controls (Belzung & Griebel 2001; Clement et al. 2002). As C57BL/6JOlaHsd and 129S2/SvHsd mice show major phenotypic differences in tests such as the plus-maze, and since genes linked to the mutation in wild-type controls will be C57-derived (Homanics et al. 1999), a false positive ‘anxiogenic-like’ profile in the knockout (where linked genes will be 129-derived) may be almost unavoidable. Indeed, current findings strongly suggest that the term ‘anxiogenic-like’ may be a serious misrepresentation of certain phenotypes which could more accurately be described in terms of locomotor activity per se. The question must, therefore, arise as to the suitability of using 129-derived mice in exploration-based tests or, at the very least, the suitability of using these tests alone in the ‘emotional’ phenotyping of mutant animals (see also Clement et al. 2002). Secondly, if researchers are to continue using tests such as the plus-maze, it is essential they recognise that test conditions validated for one strain may not be valid for other strains. Present results clearly show that conditions validated for Swiss-Webster mice are not appropriate for studies with C57BL/6JOlaHsd mice. While increasing the aversiveness of the test environment would undoubtedly yield a more anxious behavioural baseline from which to obtain an anxiolytic drug response in C57-derived mice, these even more aversive conditions are unlikely to be optimal for research on 129-derived mice. In view of the mixed genetic background of mutant mice, such reasoning implies the need to screen mouse mutants in two versions of the same test: one validated behaviourally and pharmacologically for C57 mice and the other validated for 129 mice. Finally, studies reporting altered sensitivity (usually decreased, e.g. Low et al. 2000; Sibille et al. 2000) to the anxiolytic effects of benzodiazepines in mutant animals require very cautious evaluation in the light of present findings. Depending on basal performance, wild-type controls may well show an anxiolytic response to CDP or diazepam, but the attenuation/absence of such a response in the mutant may reflect inherent abnormalities in benzodiazepine-GABAA receptor function in the 129 parental strain. In this context, it is pertinent to note the recently reported significance of genetic background for the detection of altered benzodiazepine sensitivity in 5-HT1A receptor knockout mice (Olivier et al. 2001).