Assessing REM Sleep in Mice Using Video Data.pdf

SLEEP, Vol. 35, No. 3, 2012 433 Assessing REM Sleep in Mice Using Video Data—McShane et al scored, whereas with 4-second epochs, there are 21,600 ep- ochs. Thus, the necessity for surgery, time to recover from sur- gery, and scoring of large numbers of epochs adds expense and makes studies of sleep in mice very labor intensive. We seek an alternative high-throughput strategy that will obviate the need for EEG/EMG recording. This strategy can be used (1) in studies that evaluate changes in mRNA, protein, etc. in response to sleep, wake, and sleep deprivation and (2) to screen the large panel of knockout mice that have already been created.3 Two approaches to high-throughput phenotyping have al- ready been proposed.4,5 One approach is based on determining inactivity either by electronic beam splits or by video analy- sis; any duration of inactivity that lasts 40 seconds or more is considered sleep. This approach, termed the 40-second Rule, has been validated by comparison with manual scores based on EEG/EMG recordings in both young4 and old6 C57BL/6J mice. The other strategy is based on piezoelectric detection of mouse movements by pressure sensors in the floor of the mouse cage; analysis of the data recorded by such sensors reveals patterns that are characteristic of sleep and wakefulness.5 Although these methods are quite accurate to determine wake and sleep, they cannot distinguish non-rapid eye move- ment sleep (NREM) from rapid eye movement sleep (REM). Nonetheless, our video recordings show a subtle signal for REM sleep. In particular, the area and aspect ratio of the mouse, respectively, increased and decreased when the mouse went from NREM to REM sleep (see Figure 1). This, we believe, is related to the mouse becoming more atonic in REM sleep. The goal of this study was, therefore, to determine wheth- er we could develop an algorithm to identify REM vs NREM INTRODUCTION A major focus of current research in mice is to elucidate gene produc