RESUMO
Depression is highly prevalent in Alzheimer Disease (AD); however, there is paucity of studies that focus specifically on the assessment of depression-relevant phenotypes in AD mouse models. Conditional doxycycline-dependent transgenic mouse models reproducing amyloidosis (TetOffAPPsi) and/or tau (TetOffTauP301L) pathology starting at middle age (6 months) were used in this study. As AD patients can experience depressive symptoms relatively early in disease, testing was conducted at early, pre-pathology stages of Aß and/or tau accumulation (starting from 45 days of transgenes expression). Tau-related differences were detected in the Novelty Suppressed Feeding task (NSF), whereas APP-related differences were observed predominantly in measures of the Open Field (OF) and Forced Swim tasks (FST). Effects of combined production of Aß and tau were detected in immobility during the 1st half of the Tail Suspension task (TST). These data demonstrate that results from different tasks are difficult to reconcile using task/variable-centered interpretations in which a single task/variable is assigned an ad-hoc meaning relevant to depression. An alternative, concept-oriented, approach is based on multiple variables/tests, with an understanding of their possible inter-dependence and utilization of statistical approaches that handle correlated data sets. The existence of strong correlations within and between some of the tasks supported utilization of factor analyses (FA). FA explained a similar amount of variability across the genotypes (â¼80%) and identified two factors stable across genotypes and representing motor activity and anxiety measures in OF. In contrast, variables related to FST, TST, and NSFT did not demonstrate a structure of factor loadings that would support the existence of a single integral factor of "depressive state" measured by these tasks. In addition, factor loadings varied between genotypes, indicating that genotype-specific between-task correlations need to be considered for interpretations of findings in any single task. In general, this study demonstrates that utilization of multiple tasks to characterize behavioral phenotypes, an approach that is finally gaining more widespread adoption, requires a step of data integration across different behavioral tests for appropriate interpretations.