The fallacy of behavioral phenotyping without standardisation

Abstract

Behavioral phenotyping of mutant mice is a new and challenging task for the behavioral neuroscientist. Therefore, standardisation of the experimental conditions is required to permit comparisons between the results of experiments within and between laboratories. Once mutation-induced behavioral changes have been identified, phenotyping of mouse mutants should be performed along a systematic trajectory, which allows for an in-depth characterisation of the mutant under investigation.Keywords: Behavior, behavioral phenotyping, knockout, mouse, mutant, standardization, transgenic The primary aims of behavioral phenotyping of mouse mutants are to gain insight into the behavioral effects of experimentally induced mutations and to elucidate the possible biological significance of a gene product for the modulation of behavior. Therefore, a battery of tests is needed which covers a broad range of the behavioral repertoire of the mouse. The main objectives are (i) to construct, develop and evaluate theories about the relationship between a mutation and behavior and (ii) to develop models with high validity for the prediction of the effects of a manipulation in (human) disease.Both objectives demand that the behavioral paradigms used be validated. This requires experimental techniques which permit assessment of (i) how well a measure of a particular process, trait or state, reflects theoretical assumptions (construct validity), (ii) whether a manipulation predicts performance in the condition to be modelled (predictive validity) and (iii) to study the degree of similarities between the measure of the model and the disorder it simulates (face validity; Willner 1991). Depending on whether a model or a 9 theory is under scrutiny, the different types of validity will be of different importance.Validated tests reflect the standard by which effects of experimental manipulations must be judged. One often overlooked aspect, however, is that the nature and/or magnitude of a behavioral effect caused by a mutation can differ across tests thought to measure the same process, trait or state, i.e. which are believed to possess concurrent validity, when in fact they do not.Unfortunately, only very limited information is available about the concurrent validity of many tests used to phenotype mouse mutants. This problem gains even more relevance by the finding that results obtained in ostensibly the same task may differ, both across and within laboratories. These differences may, for example, be due to the history of the animal, time of testing, level of illumination or small details in the construction of the apparatus (Trullas & Skolnick 1993;Crawley et al. 1997;Rampon et al. 2000). In the worst case, we are faced with different results for reasons that remain undetected. Even in situations where conditions were equated as rigorously as possible across laboratories, laboratory-specific effects were noted (Crabbe et al. 1999). This clearly shows that reproducibility is an issue in current behavioral phenotyping of mouse mutants. Many inve...

Different data from different labs: Lessons from studies of gene–environment interaction

“…However, widely accepted, standard procedures for testing mouse behaviors are not currently available (see Würbel, 2000Würbel, , 2002Wahlsten, 2001;van der Staay and Steckler, 2002). Instead, behavioral assessment protocols tend to be unique to each laboratory testing mice.…”

Functional genomics of neural and behavioral plasticity

“…In most behavioral genetic experiments environmental conditions are kept constant (for a discussion of standardization see Würbel, 2002;van der Staay and Steckler, 2002). However, it has become increasingly clear that complex and nonlinear G ϫ E interactions exist (Pigliucci, 2001;Sokolowski & Wahlsten, 2001;Wade, 2001).…”

Animal Models of Cognitive Disorders

“…Therefore, it is of utmost importance to define a given psychological process and to break down these criteria into components testable in animals (van der Staay and Steckler, 2002).…”

“…Therefore, a natural environment must be considered to be as artificial as every other environmental setting. Moreover, it has been argued that such 'natural' environments are defined less precisely and hence are less amenable to standardization and comparison (Van der Staay and Steckler, 2002). To have a valid animal model (of cognitive disorders), however, it is of utmost relevance to have tasks at hand that, irrespective of the degree of ecological relevance, allow a clear-cut dissociation between the behaviour of interest (e.g.…”

“…These objectives demand validation of the model (Sarter and Bruno, 2001;van der Staay and Steckler, 2001;van der Staay and Steckler, 2002;Willner, 1991). The focus of validation will depend on the type of model (1) screening models, (2) more complex or simulation models, and (3) assay or mechanistic models (Figure XV-1.1) (Ellenbroek and Cools, 2000;Willner, 1991) (see also Chapters III and XVII-1).…”