Sleep Network Deterioration as a Function of Dim-Light-At-Night Exposure Duration in a Mouse Model

Round 1

Reviewer 1 Report

The paper quantifies the impact of dim light at night on a highly integrative set of variables related to sleep-wake regulation (periodicity and fractal measures in locomotor activity, sleep architecture and EEG power density in sleep-specific frequency ranges). As reported by the authors, vigilance state and EEG spectral power data are partly published (3-months DLAN condition, Panagiotou et al., 2020), but the authors add substantial information, mainly by adding further conditions (different DLAN durations, n = 5) as well as by analyzing fractality in locomotor activity.

p. 3, line 123: For the DFA analysis, the authors use the range from 3-8 h. They argue that their data above 8h are not statistically reliable due to restriction of data length and cite Hu et al., 2001. This choice is not clear. 15 days of recording are not sufficient to go beyond 8 hours time windows?) Hu et al. (2004) observed that scale-invariant features cannot be caused by ultradian rhythms at time scales less than 4 hours. Furthermore, SCN lesion in rats abolished the scale-invariant locomotor pattern between 4 and 24 hours (Hu, 2007).

If I understood correctly, a LD control group, tested after one day, one week, one month and three months is lacking. If that’s correct, how do the authors exclude that confounding factors (e.g. habituation, signal quality, distance to surgery) are underlying the modulation in the EEG measures, but also fractal measures in locomotor activity?

The number of generated statistical models might be critical considering the low sample size. Were the factors “treatment”, “time of day”, “light-dark” etc. considered as repeated measures? The authors mention Bonferroni correction but it is not clear what is corrected for: did the authors account for the number of computed models?

Please specify how day-night amplitude was calculated.

"It seems that sleep deprivation did not restore behavior during the subsequent recovery period, as if the mice were less susceptible to the elevated sleep pressure". Statistically, the authors did not observe an interaction between the factor “Day” and “Treatment” which would underground such a differential susceptibility to sleep deprivation? Did the authors compute a ratio between baseline and SD to assess the rebound?

Reference to baseline measures is not obvious in the Figures (e.g. in Figure 5 the text referes to baseline (BL) but the graph does not indicate “BL” (but L1.2)

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The topic of light at night for health and well-being is an important field of research. Animal research may clearly contribute to our knowledge and in this study, the authors try to solve a piece of the puzzle and use relatively new ways of analyzing rest-activity patterns. In this context, it is an important manuscript. I do have some minor concerns and suggestions for improvement. Please find my comments, in chronological order, below.

Abstract: I feel the abstract is written too much from the expectation that light at night is bad for sleep, also in nocturnal animals.  See the points below for a selection of comments.

Line 14: "dramatic sleep changes" sounds pathetic, too much emotion for me in a scientific paper. My proposal would be to use a less suggestive term as 'big' or 'large' or 'huge'.

Lines 23-24: a significantly less healthy rest-activity rhythm - what is a healthy rest-activity rhythm and what is an unhealthy rest-activity rhythm? If this is not defined and if health measures are not collected, I would refrain from using the term 'healthy'. If the authors agree, please replace it with a more descriptional term, otherwise please give arguments why using the term 'healthy' is appropriate.

Line 26-27: the same holds for the term 'deleterious' effects, this contains already an interpretation. Why are the effects deleterious? The following sentence is too vague in my opinion: What are the potential compensatory mechanisms? Are they measured? Or is this a speculation? Please clarify and use arguments based on your data.

Introduction:

Line 61: I am missing somewhere in the manuscript the discussion about possible differences between light at night in nocturnal and diurnal species. In mice, you apply the dim light during their active phase, while in diurnal species it is the rest phase. Maybe here, in line 61, this item can be added? 

Lines 66-73 is a replication of the summary without extra information. For the inexperienced reader, not an expert in the DFA and its implications, it is completely unclear why the authors conclude about a 'less healthy physiology' and that 'brain integrity' is compromised. Please extend a bit on these concepts in the introduction.

Methods:

Line 76: why only male mice? Do you expect a sex difference? Wouldn't it be worthwhile to include female mice as well and why did you not do this?

Lines 117-121: The method of DFA is explained in mathematical terms without translating its interpretation. Please clarify. e.g. 'scale-invariance in the data' and please explain sentences 119-121: This value alpha is the scaling exponent....it signifies the correlation properties in the signal'. What does this mean? Please clarify why you are able to use ranges up to 8 h as a circadian measure? Actually in the discussion, lines 314 - 327, you explain the interpretation of DFA a bit more in detail, but in my opinion this information is needed much earlier in the manuscript to be able to understand the results.

Results:

Lines 180-183: again, related to the earlier request to explain DFA, the result of what the finding means is not explained; please explain what it means that alpha is lower in animals exposed to DLAN, give the interpretation.

Lines 186-: it is for the first time that the abbreviation D1 L1 D2 and L2 are used. Please explain these terms in figure 1. That would be helpful.

Line 191: I would say that if there is no significant decrease, that there is no decrease. Why do you interpret it as a decrease?

Line 192-193 and more in general (also e.g. line 259: The main effects of DLAN are on behaviour during the dark phase, while animals are normally awake. Could it be that the effects during the first DLAN (1d condition) are more a masking effect, a direct acute effect to escape the light?

Line 204: I had some difficulty understanding this sentence, shoudl 'time' maybe be replaced with 'timing'? 

Line 205: the x-axis of S2 is not linear, while the points are plotted as if there is a linear relationship. Is a linear regression explained variance appropriate?

Lines 243-244: "There is a small and gradual increase in SWA up to 1m DLAN during nREM sleep". Line 244 states: "In accordance with the general decrease found in the EEG power density, ...." which decrease is meant?

Discussion:

Lines 356-363: As mentioned in the introduction, a discussion of the possible differences between nocturnal and diurnal animals with respect to the light during the active phase or during the inactive phase would be appropriate. Do you expect the same mechanism, and why?

Lines 395 - : I do like the discussion on the idea that longterm exposure to DLAN affects the network structure of the brain and that compensatory mechanisms do not work any longer. But I am missing the arguments that that is happening here. Could it just be a longterm effect of sleep disturbances, or do we actually see the compensatory mechanisms instead of the disturbance? And again, in the context of nocturnal mice and diurnal humans, is it so simple that we can translate these findings to the detrimental effects of light at night in humans? Please elaborate on this item.

Author Response

Please see the attachment.

Author Response File: Author Response.docx