Depriving Mice of Sleep also Deprives of Food

Round 1

Reviewer 1 Report

The manuscript entitled Depriving mice of sleep also deprives of food is presented for peer review. Food access is one of the main zeitgebers of the circadian system. Works in rodent models shown contradictory data. Authors correctly mentioned that studies in mice and rats show that feeding-related adjustments were not immediate and it could take up to a couple of days until variables such as wakefulness and corticosterone production become entrained and peak prior to the new time of food availability. The authors questioned whether access to food during sleep deprivation affects the SD-induced changes in clock-gene expression in the cerebral cortex. Next, they assessed whether, like humans, mice eat more when kept awake. 

Manuscript is well-written and structured. I have several suggestions. 

1. How do you control sleep deprivation?
2. Please provide the choice of core clock genes. I suggest to discuss it. BMAL1 but not CLOCK is also changes in circadian manner.
3. I doubt on that SD protocol. Is that better to shift schedule 18:6 to 12:12 then apply SD?
4. Does SD result in stress. If so, please provide corticosterone ELISA data evidencing that. Also, leptin and ghrelin data would be helpful.
5. Sleep is regulated by both circadian and homeostatic drivers. Please discuss that in your paper.

Author Response

Thank you for reviewing our manuscript, your positive assessment, and the comments.

1. How do you control sleep deprivation?

Sleep deprivation (SD) was performed by direct observation of the mice by trained experimenters. The details of this method were described in the Methods (Lines 363-369) and it is an established method in a number of labs, including ours, to challenge the sleep homeostat. EEG recordings show that it is highly efficient at preventing sleep (see e.g. 2^nd results paragraph in Diessler, Jan et al. PLoS Biology 2018; doi: 10.1371/journal.pbio.2005750). For the analyses of food intake and body weight the animals served as their own controls (SD vs. baseline). For the effects on gene expression, undisturbed, non-SD cohorts served as controls. We have explained this now more explicitly in the Methods (Lines 410-412 and 430-431).

 

2. Please provide the choice of core clock genes. I suggest to discuss it. BMAL1 but not CLOCK is also changes in circadian manner.

Hor, Yueng et al. (PNAS 2019; doi: 10.1073/pnas.1910590116doi: 10.1073/pnas.1910590116) showed that SD affected the expression of almost all known clock genes but not all do so immediately at the end of SD. We selected Per1, Per2, Cry1, and Dbp because their expression to change at that time point (see doi: 10.1186/1471-2202-8-87 and 10.5665/sleep.244010.5665/sleep.2440). We have now explained this rational in the Methods section (Lines 401-404).

 

3. I doubt on that SD protocol. Is that better to shift schedule 18:6 to 12:12 then apply SD?

We opted for the standard SD protocol used in the field in which mice remain stably entrained to a LD12:12 light-dark cycle throughout the experiment. Changing the light-dark cycle would have affected the circadian system rendering the interpretation of the SD results difficult.

 

4. Does SD result in stress. If so, please provide corticosterone ELISA data evidencing that. Also, leptin and ghrelin data would be helpful.

Acute SD indeed represents a mild form of stress (Meerlo et al. Sleep Med Rev 2008; doi: 10.1016/j.smrv.2007.07.007) as was explained in the discussion (Lines 305-306). For the levels of corticosterone reached after SD, using the same SD protocol, we refer to Mongrain et al. SLEEP 2010 (doi: 10.1093/sleep/33.9.1147). Given the unexpected results we report in our study, it would indeed warrant follow-up studies to quantify leptin and ghrelin levels. However, this clearly falls outside the scope of this publication and instead we relied on reports by others in the rat for the discussion (Lines 283-285). We added a section on leptin, ghrelin, food intake, and SD in the Discussion (Lines 285-294).

 

5. Sleep is regulated by both circadian and homeostatic drivers. Please discuss that in your paper.

We have included a statement mentioning the homeostatic aspect of sleep regulation in the Introduction (Lines 40-41).

Reviewer 2 Report

This paper is well-written and provides interesting information on the association between sleep deprivation, consequent changes in food consumption, and clock gene / clock-controlled gene expression in mice.

Results from this study underline principal differences in the effects of circadian rhythm interference by the third factor, sleep deprivation in this study, on the expression of core clock and clock-controlled genes and food consumption between nocturnal laboratory animals vs. diurnal species such as humans. As consequence, there are fundamentally different weight changes between these species under the condition of deprived sleep.

This study is important also to show the need for human field studies since studies from the animal models show fundamentally different results as far as the relationships between sleep-wake cycle, food consumption, and circadian rhythms are concerned.

Minor comments:

Authors may wish to consider adding the following works while discussing effects of food deprivation in humans: DOI: 10.3945/ajcn.2009.28523 ; doi: 10.1002/1098-108x(200011)28:3<272::aid-eat4>3.0.co;2-q. and experimental animals: DOI: 10.1152/ajpregu.00734.2002.

Also, authors may wish to consider for discussion that in the nocturnal rodents it was found that daytime feeding restriction was associated with a reduced sleep drive (doi: 10.1093/sleep/zsz157), a vice versa scenario to what is found in the current work.

Author Response

Thank you for taking the time to review our manuscript, the positive comments, and pointing us to these useful references.

We now cited 3 of suggested references in the manuscript (Lines 50, 289, and 349) but felt we could not readily integrate the reference regarding the effects of food deprivation on eating behaviour in subjects suffering from eating disorders (doi:10.1002/1098-108x(200011)28:3<272::aid-eat4>3.0.co;2-q.).

Reviewer 3 Report

Sleep and metabolism are tightly linked with circadian clocks, but how sleep disruption affects circadian rhythms and feeding intake remains unclear.  In this manuscript, the authors first used qPCR to validate clock-gene expression in mice at the transcriptive level, they found a handful of clock genes were upregulated significantly in the cerebral cortex during sleep deprivation (SD), but it’s not corelated to food availability. Next the authors measured daily food intake and found that mice did not consume more food during SD but ate more during the recovery days. Further analysis revealed that food intake was dissociated from awake time by sleep deprivation. Finally, the authors showed sleep deprivation in mice causes energy deficit and weight loss.   

Overall, this is a carefully executed study that contains some novel and intriguing findings to follow up. The topic is timely. The paper is clearly organized and well-written. In summary, this paper represents an important contribution to our understanding of how sleep modulates food intake and energy metabolism. This paper will appeal to both biological rhythms community and scientists interested in metabolism. I think it is a strong paper and only have a few suggestions (mainly in the nature of expanding discussion) to improve it prior to publication.

Minor comments:       

1. The SD protocol used in this study is not very clearly described, will this “gentle handling” method disrupt normal food intake in mice? For example, will such “gentle handling” method also disrupt food intake when mice were awake during the dark phase?
2. Despite of the interesting findings in this paper, the underlying mechanisms are not addressed sufficiently. I would like to see some potential working hypothesis presented in the discussion.
3. Unlike mice, humans consume more food when sleep was deprived and gain weight during SD (as mentioned by authors). It would be nice to see more discussion about what may cause this difference.
4. Equation in Figure 3 B’ should be moved to the materials and methods section.
5. In Figure 3, error bars were missing.
6. In Figure 4, what do the error bars stand for (stand error or stand deviation)?

 

Author Response

Many thanks for reviewing the manuscript, this positive assessment, and the suggestions to improve the manuscript.

Minor comments:       

1. The SD protocol used in this study is not very clearly described, will this “gentle handling” method disrupt normal food intake in mice?

We have tried to improve the description of the ‘’gentle handling’’ method (Lines 365-369). Animals remained in their home cage with ad libitum access to food and were not disturbed unless they showed signs of sleep. Moreover, they did consume the same amount of food as at this time during baseline. We therefore think that the SD protocol allows free access to food. However, because this method is considered a mild stressor, factors like the presence of the experimenters might have prevented mice from eating more than they did. The possibility of stress limiting food intake was mentioned in the Discussion (Lines 305-307).

For example, will such “gentle handling” method also disrupt food intake when mice were awake during the dark phase?

This is an interesting suggestion that we have not tested. We did perform a SD in the first 3h of the dark-period (ZT12-15) when mice normally eat the most and wakefulness reaches highest levels (Vassalli & Franken, PNAS 2017; doi: 10.1073/pnas.1700983114). This intervention did not do much in terms of increasing time-spent-awake (and not many interventions were needed to keep animals awake), but it did increase corticosterone levels (as were Per2 and Homer1a expression levels in the brain). So also at this time-of-day, the mere presence of the experimenters might be sufficient to interfere with feeding behaviors (and possibly other behaviors conditional on the mouse ‘feeling safe’). We now have mentioned this follow-up experiment in the Discussion (Lines 307-311).

 

2. Despite of the interesting findings in this paper, the underlying mechanisms are not addressed sufficiently. I would like to see some potential working hypothesis presented in the discussion.

3. Unlike mice, humans consume more food when sleep was deprived and gain weight during SD (as mentioned by authors). It would be nice to see more discussion about what may cause this difference.

We merged these 2 comments as they relate to the same issue.

The working hypothesis concerning the SD-induced changes in clock-gene expression was explained in the Discussion (paragraph starting at Line 312). This aspect was however not the main focus of that experiment, which was instead aimed at excluding food intake as a contributing factor. We therefore feel this aspect was sufficiently covered by referring to our previous work and reviews on this topic (Lines 327-328).

The working hypothesis concerning the observed weight loss during SD and the subsequent eating more during recovery was extensively covered in the section ‘Sleep deprivation or food deprivation?’ (starting at Line 235). We concluded that the effects must be attributed to the higher energy expenditure associated with the animal being awake (compared to being asleep) that was not met by increased energy intake. This let us to suggest that while animals ate as much as during baseline, they should have eaten more, an interpretation that was supported by the analysis presented in Figure 3.

As to the mechanism that preventing mice from eating more we can only speculate. We offered that stress associated with the SD protocol might play a role but feel that given the data presented, we cannot go much further with this. We now mentioned this possibility more explicitly in the Discussion (Lines 306-307), which might also be relevant in explaining the species difference observed in the response to this intervention (Lines 341-342). We also expanded on the signals associated with hunger and the interaction between SD and the activity of AgRP-positive neurons in the arcuate nucleus might be of interest in this context (Lines 285-294). 

 

4. Equation in Figure 3 B’ should be moved to the materials and methods section.

We prefer to keep this equation as a graphical element as it is key to understanding the analysis presented in this figure; the equation follows from the results in Panel A and then determines ‘Expected food intake’ values in Panels C and D. We think that keeping it in the figure will help the reader follow the analyses more easily then to refer to the Methods. If not acceptable, we suggest to move this equation to the Results section.

 

5. In Figure 3, error bars were missing.

This was intentional as the correlation analyses could be performed at the group level only, as was explained in the Results section (Lines 147-149). For food-intake estimates of the variance were provided in Fig.2. Moreover, it will be difficult to estimate the variance of the ‘Expected food intake’ values and of the residuals.

 

6. In Figure 4, what do the error bars stand for (stand error or stand deviation)?

Standard errors of the mean. We now have added this information also to this figure legend (Line 219).

Round 2

Reviewer 1 Report

Authors did the great job and addressed all my issues. Thank you!