- Top of page
- Materials and methods
The behavioral characterization of rodent strains in different studies and laboratories can provide unreplicable results even when genotypes are kept constant and environmental control is maximized. In the present study, the influence of common laboratory environmental variables and their interaction with genotype on the results of behavioral tests of anxiety/emotionality were investigated. To this end, the inbred rat strains Lewis (LEW) and spontaneously hypertensive rats (SHR), which are known to differ for numerous emotionality-related behaviors, were tested in the open field (OF), elevated plus maze (EPM) and black/white box (BWB), while three environmental factors were systematically controlled and analyzed: (1) the experimenter handling the animal (familiar or unfamiliar); (2) the position of the home cage (top or bottom shelf of the rack) and (3) the behavioral state of the animal immediately before the test (arousal or rest). Experimenter familiarity did not alter the behavior of rats in the OF. Cage position, on the other hand, influenced the behavior in the OF and BWB, with rats housed in top cages appearing less anxious like than those housed in the bottom. In the BWB (but not in the OF), these effects were genotype dependent. Finally, the behavioral state of the animals prior to testing altered the results of the EPM in a strain-dependent manner, with some anxiety-related genotypic differences being found only among rats that were aroused in their home cages. This study showed that common variations in the laboratory environment interact with genotype in behavioral tests of anxiety/emotionality. Recognizing and understanding such variations can help in the design of more effective experiments.
During the past decade, massive and systematic behavioral testing of multiple strains of laboratory rodents became a key step in neuroscience research, particularly in the field of behavior genetics. When populations of animals developed through inbreeding, selective breeding, transgenesis or gene targeting started to be characterized, it became evident that behavioral data from different laboratories were often not reproducible. Such discrepancies could result not only from differences in the genetic background of the animals but also from variations in the testing and/or laboratory environments (Chesler et al. 2002; Van der Staay & Steckler 2002; Wahlsten et al. 2003; Würbel 2002). A seminal work by Crabbe et al. (1999) has shown that major efforts to standardize both the test situation and the rearing environment do not guarantee the consistency of results across laboratories. For example, the mouse strains A/J and 129/SvEvTac, when compared to each other, displayed either higher or lower locomotor response to cocaine, depending on the laboratory where they were tested, in spite of environmental factors having been rigorously controlled. On the one hand, these data revealed that subtle, uncontrollable environmental variations between and within laboratories can influence behavior genetic studies. On the other hand, the effects of such an environmental background remained largely undefined.
Identifying and describing environmental factors that are potentially relevant to a large number of studies and laboratories are of foremost importance. This knowledge will contribute to improve the consistency of results, the interpretation of genotypic differences and the design of behavioral experiments. This problem is certainly not restricted to genetic studies and is not limited to mice, because behavioral testing is also essential in pharmacology, for example, and numerous studies using inbred and transgenic strains are carried out with rats.
The aim of the present study was thus to investigate the effects of some laboratory environmental variants which are commonly present in many, if not all, rodent behavioral studies through the use of a genetic model that has been well validated for the study of emotionality/anxiety. To this end, the inbred rat strains Lewis (LEW) and spontaneously hypertensive rats (SHR) were tested in three behavioral models widely used in the study of anxiety, namely the open field (OF), elevated plus maze (EPM) and black/white box (BWB). The LEW and SHR strains were chosen, because they have previously shown contrasting anxiety-related behaviors in all the aforementioned tests, with the former avoiding the aversive stimulus of each test more markedly than the latter (Ramos et al. 1997, 1998, 2002). Moreover, such strain differences seem reasonably robust, as they were confirmed in different substrains, laboratories and countries (Ramos et al. 2002) and they were shown to resist to variations in the test situation, such as day/night testing in the BWB, first/second trial in the EPM (unpublished) and strong/dim light in the OF (Ramos et al. 2002).
Several biological and environmental variables are known to influence behavioral tests. They are therefore equated and/or listed in most publications in the field. Among them, one could mention age (Imhof et al. 1993), sex (Imhof et al. 1993; Johnston & File 1991), time of testing (Gentsch et al. 1982), illumination level (Cardenas et al. 2001), floor surface (Morato & Castrechini 1989), transportation to the test room (Morato & Brandão 1997), previous experience in the apparatus (Bertoglio & Carobrez 2000; File et al. 1992; Treit et al. 1993), handling (Henderson 1970) and group housing (Maisonnette et al. 1993). Other factors, however, in spite of being potentially relevant, have rarely been considered and/or investigated. Herein, LEW and SHR rats were tested using the same protocols and apparatuses that were used in previous studies, but three environmental factors that had never been considered for these strains and tests have had their variations systematically controlled and analyzed.
The first of these factors was the experimenter performing the test, which has been shown to affect the output of nociception tests (Chesler et al. 2002). In the present work, this factor varied between one subject who was familiar to the animals and had large experience with all practices of the laboratory and another who had never entered the housing facilities. The second variable was the position of the cages inside the housing room, which has been shown to influence sensitivity to d-amphetamine in rats (Exner & Clark 1993). Like in many laboratories, this factor could vary between the highest (top) and the lowest (bottom) of five-rack shelves. Among other things, the illumination level was very different between these two housing situations.
Finally, the behavior of the animal immediately before testing was monitored and analyzed. Arousal has been proposed to be one of the two dimensions of human affective experience and, according to Heller's model of brain activity and affect, anxiety could result from the association between unpleasant affect and high arousal or activation (Schmidtke & Heller 2004). Behavioral tests of anxiety are typically short in time but are carried out during sessions that last several hours and involve many different animals. Even if these individuals do not to differ in the psychological trait being inferred, it can be expected that their behavioral (e.g. sleeping, eating and playing) and hence their arousal status at the moment preceding the test will vary. Yet, as pointed out by Lister (1990), most models of anxiety provide measures of state anxiety, which is thought to be a temporary and not a permanent feature of the individual. We have thus hypothesized that the ‘anxious state’ inferred during a specific test can be influenced by the arousal state of the animal immediately before the trial. We approached this problem in the EPM by testing rats that had been in a state of either ‘rest’ or ‘arousal’ in their home cages for at least 5 min continuously before the test.
- Top of page
- Materials and methods
The results of the present work showed that some common environmental factors which are expected to vary within and between laboratories devoted to behavioral testing can alter not only the baseline levels of anxiety-related behaviors but also the manifestation of genotypic differences between rodent strains. The location of the animals' home cage from weaning to the day of test was shown to influence fearfulness in the OF test regardless of strain or sex. In the BWB, on the other hand, this same variable affected fear-related behaviors in a genotype-dependent manner, in such a way that some strain differences appeared only in rats housed in top shelves. The behavioral state of the animal observed immediately before testing has also been shown to affect the output of the EPM test, with some anxiety-related differences appearing between strains only for rats that were aroused before testing. The familiarity of the animals with the experimenter, however, was found not to influence any of the strain's behavior in the OF test.
Previous studies had shown that LEW rats of both sexes, when compared with their SHR counterparts, avoid more the aversive stimuli of different tests such as the central area of the OF, the open arms of the EPM and the white compartment of the BWB (Berton et al. 1998; Ramos et al. 1997, 1998, 2002). The analysis of an intercross between these strains showed that some of the behavioral contrasts between them were highly (>50%) heritable (Ramos et al. 1998). In spite of the apparent robustness of this genetic model, some interstudy variability has been observed. For instance, strain differences in the percentage of open-arm entries, one of the main indices of anxiety from the EPM, were sometimes but not always significant, depending on the study (Ramos et al. 1997, 1998, 2002; Vendruscolo et al. 2003). We have therefore hypothesized that environmental factors may not only alter the baseline behavioral scores of LEW and SHR rats but also interact with genotype. However, when we varied environmental aspects of the testing situation itself such as day/night in the BWB, previous experience in the EPM (unpublished) and illumination in the OF (Ramos et al. 2002), the strain differences remained unchanged. These findings suggested that factors acting before the test might be more relevant for strain–environment interactions, an idea that is partially supported by the results of the present study.
Being handled by an unfamiliar experimenter before the OF test was not enough to change the behavioral profile of the two strains. Crabbe et al. (1999) suggested that the experimenter was one of the causes of variable results when mouse strains were compared at three different laboratories. Accordingly, Chesler et al. (2002) found that the experimenter was an important cause of interstudy variability in nociception data. In the present study, differently from these previous reports, only two experimenters were compared and all efforts were made to standardize their procedures, movements and clothing. Had all these factors not been controlled or had more experimenters been compared, then a significant effect might have been observed. The limited number of experimenters used herein impedes strong conclusions at this point, but the present data suggest that the control of all pretest procedures and of some experimenter traits and habits may help to minimize data variability that arises due to different staff members having to perform the same tests in a given laboratory.
The home-cage position in the housing rack (top or bottom) was shown to affect the results of two behavioral tests. In the OF, rats housed on top shelves displayed more central locomotion than those housed in the bottom, with this effect being similar across strains and sexes. Rodents tend to avoid exploring the central area of a novel large arena. Yet, such an avoidance can be reversed by anxiolytic drugs (Ramos & Mormède 1998; Vendruscolo et al. 2003) and is therefore thought to be anxiety related. The present results suggest thus that rats housed in top shelves are less fearful in this test that is performed under dim-light conditions. In the BWB, which has high luminosity as its main aversive component, a similar influence of cage position was observed. This effect, however, was strain and sex specific. Interestingly, strain differences in the time spent in the white compartment, a highly contrasting measure for rats from top cages, were not significant for male rats housed in the bottom. Thus, any study not controlling such an environmental variable and having a bias on cage location (e.g. most rats housed in bottom shelves) might totally miss important genotypic differences that could be otherwise revealed.
Overall, the present results suggest that being housed in either higher or lower cages, the rats exhibit changes in their emotional reactivity, which is in agreement with one previous report on responses to d-amphetamine (Exner & Clark 1993). The control of cage location has also been the object of concern in one study on carcinogenicity (Herzberg & Lagakos 1992). But, to our knowledge, this is the first study to demonstrate the effect of cage location on strain comparisons in behavioral tests. It is not possible to determine what environmental factor(s) was (were) responsible for such a phenomenon herein, but we can suggest that luminosity played a central role, because it varied between 240 (top) and 12 lx (bottom). Nevertheless, other factors not controlled herein could be expected to vary between the two housing situations such as odor, air quality and visual stimuli (other than light).
The EPM is a model of anxiety which sometimes produces unreplicable results (Hogg 1996). The present data showed that the arousal state of the animals immediately before this test can contribute to decrease replicability. For the two most important indices of anxiety (i.e. time spent and percentage of entries in the open arms), LEW and SHR male rats differed from each other only when tested after being aroused. The present results corroborate our hypothesis that the arousal status of an animal prior to testing can influence its ‘state anxiety’. Moreover, the observation that most genetic differences in experimental anxiety were found only in aroused rats is compatible with Heller's proposition that human anxiety, differently from depression, is expressed in a situation of high arousal or activation (Schmidtke & Heller 2004). One can suppose that the arousal status before testing is very likely to vary among individual animals within and between studies and that such a variation is not necessarily random. If an individual's state of arousal is affected by the cage order at which it is tested, for example, then one could expect that cage order would influence the output of a test. Such an influence has in fact been observed in nociception studies (Chesler et al. 2002) and, in the present study, it has been shown to act on measures of the OF and BWB. In general (data not shown), the first rat to be tested in a given cage was more anxious–like and less active than rats tested afterwards.
In summary, the results of this study showed that at least two different laboratory environmental factors have relevant influence on the results of behavioral tests and that some of this influence is genotype dependent. These findings are relevant for studies that aim at comparing animals with different genotypes but also have a potential impact on other types of neurobiological experiments. Further studies should investigate the causes of such environmental effects and their impact on gene expression and use this knowledge to propose changes in pretesting procedures. Meanwhile, an effort to randomize these effects would be advisable. For example, balancing strains/treatments across different housing situations (Herzberg & Lagakos 1992) and/or manipulating each cage a few minutes before starting the tests, in such a way that all rats get initially aroused, might help to increase the reproducibility of results in the OF, BWB and EPM tests.