Depriving Mice of Sleep also Deprives of Food

Abstract

Both sleep-wake behavior and circadian rhythms are tightly coupled to energy metabolism and food intake. Altered feeding times in mice are known to entrain clock gene rhythms in the brain and liver, and sleep-deprived humans tend to eat more and gain weight. Previous observations in mice showing that sleep deprivation (SD) changes clock gene expression might thus relate to altered food intake, and not to the loss of sleep per se. Whether SD affects food intake in the mouse and how this might affect clock gene expression is, however, unknown. We therefore quantified (i) the cortical expression of the clock genes Per1, Per2, Dbp, and Cry1 in mice that had access to food or not during a 6 h SD, and (ii) food intake during baseline, SD, and recovery sleep. We found that food deprivation did not modify the SD-incurred clock gene changes in the cortex. Moreover, we discovered that although food intake during SD did not differ from the baseline, mice lost weight and increased food intake during subsequent recovery. We conclude that SD is associated with food deprivation and that the resulting energy deficit might contribute to the effects of SD that are commonly interpreted as a response to sleep loss.

Keywords: body composition; circadian rhythms; clock genes; energy expenditure; fat; food intake; lean body mass; sleep deprivation.

Figure 1

Effects of food availability on cortical clock gene expression. Having access to food (‘Food’, left) or not (‘No Food’, right, grey background) during a sleep deprivation (SD) did not affect the SD-induced changes in the cortical expression Per1, Per2, Cry1, Dbp, and Homer1a. Values for each transcript were first expressed relative to the expression of three housekeeping genes (Eef1a, Gapdh, and Rsp9) within each sample and subsequently to the mean of the two control groups (0 level, dark-grey bars) that each accompanied the SD (mint bars) of the two food conditions. Number of mice for each group in parentheses (error bars reflect 1 SEM). Food did not change the SD response (two-way ANOVA factor ‘Food’ p > 0.57, factor ‘SD’ p < 0.0021; interaction p > 0.17 for each of the five transcripts). Asterisks designate a significant SD effect within each food condition (two-sided post hoc t-tests; p < 0.05). mRNA levels of housekeeping genes were not affected by either SD or food access (not shown). All tissues were collected at ZT6.

Figure 2

Effects of sleep deprivation on food intake. (A) Hourly values of food intake accumulated over 24 h and expressed as % of average daily food intake during the last 5 baseline days (Days −4 to 0; 100% = 4.15 g) preceding the sleep-deprivation (SD) day (Day 1). Food intake during SD (mint symbols and area) did not differ from baseline (grey symbols, 5-day averages). Horizontal lines in graph indicate % of total daily food eaten at consecutive 6 h intervals during baseline. Lower graph: SD-baseline difference in accumulated food-intake curves depicted in the upper graph. Gray area indicates the 12 h dark period. Asterisks designate hours with significant differences in food intake from baseline (paired two-sided t-tests; p < 0.05; n = 11–17; see Methods for n/day). (B) Daily food intake for the 10 days before and the 5 days after the SD. White symbols (Days −9 to −5) depict values for 5 additional baseline days recorded in cohort 1 (n = 6). Please note that these extra baseline days were not included in the analyses. Horizontal dashed line reflects mean over Days −4 to 0 (=100% in (A)). SD affected food intake (Days 1 to 5 vs. Days −4 to 0; two-way ANOVA factor ‘SD’ p = 0.004, ‘Day’ p = 0.89, interaction p = 0.78). Asterisks mark days with significant increase in food intake compared with baseline (average Days −4 to 0; paired two-sided t-tests; p < 0.05; n = 11–17; see Methods for n/day). (C) Recovery-baseline differences in accumulated hourly food-intake values for Days 2 to 5. As on Day 1, the increase in food intake on Day 2 was mainly due to extra food eaten during the 2nd half of the light phase, whereas for Days 3 to 5, extra food intake occurred at the end of the dark phase. Asterisks as in (A). Error bars mark ±1 SEM.

Figure 3

The sleep deprivation (SD) evoked dissociation between food intake and time spent awake might underlie increased food intake during recovery. (A) The time course of hourly values of food intake in baseline (mint symbols; mean over Days −4 to 0; same data as in Figure 2A), closely follows that of time spent awake (orange symbols). (B) Mean hourly values of wakefulness and food intake strongly correlated, with only small residuals that did not systematically vary with time of day (lower graph in (A)). (B’) Equation of the linear regression in panel B used to predict expected food intake in panels (C,D) (middle) based on time spent awake. (C) Given the equations in (B’), the expected food intake can be calculated for Day 1 including the SD (upper graph). SD induced a disassociation between food intake and time spent awake and animals accrued a deficit of 1.89 g of food intake during the SD that reduced to 0.86 g by the end of Day 1 (lower graph). (D) During Days 2 to 5 after SD, the relationship between waking and eating was highly correlated as in baseline (left panel, compare with (B)), and wake-derived expected food intake (given equation (B’)) predicted remarkably well the observed data (average over the 4 days, middle panel). The 0.86 g food deficit estimated for Day 1 (see (C)), was maintained during the following 4 days and stabilized at −0.71 g (right panel; non-linear regression, exponential saturating function fitted to 12 h values).

Figure 4

Effect of sleep deprivation (SD) on body weight and body composition. (A) SD caused a 4.8% weight loss in mice with ad libitum access to food (mint bar; p = 1.3 × 10⁻⁵, paired two-sided t-test, n = 9). Body weight measured after the 6 h SD (grey bar) ending at ZT6 was 1.3 ± 0.1 g lower compared with that reached at this time of day on two preceding baseline days (bsl; grey bar; average of the 2 days). (B) Body composition was measured in a separate cohort following the exact same protocol as in A. SD similarly reduced whole body weight in this cohort (−1.5 ± 0.1 g, 5.2%; p = 9.6 × 10⁻⁵, paired two-sided t-test, n = 6). Lean body mass decreased to the same extent (−1.3 ± 0.1 g, 5.2%; p = 3.1 × 10⁻⁴). Fat mass was reduced as well but not significantly so (−0.04 ± 0.04 g, 3.0%, p = 0.33). Details as in (A). * marks significant SD vs. bsl effects. Note different scale for grams of fat mass. Error bars mark 1 SEM.