Unanticipated daytime melatonin secretion on a simulated night shift schedule generates a distinctive 24-h melatonin rhythm with antiphasic daytime and nighttime peaks

Abstract

The daily rhythm of plasma melatonin concentrations is typically unimodal, with one broad peak during the circadian night and near-undetectable levels during the circadian day. Light at night acutely suppresses melatonin secretion and phase shifts its endogenous circadian rhythm. In contrast, exposure to darkness during the circadian day has not generally been reported to increase circulating melatonin concentrations acutely. Here, in a highly-controlled simulated night shift protocol with 12-h inverted behavioral/environmental cycles, we unexpectedly found that circulating melatonin levels were significantly increased during daytime sleep (p < .0001). This resulted in a secondary melatonin peak during the circadian day in addition to the primary peak during the circadian night, when sleep occurred during the circadian day following an overnight shift. This distinctive diurnal melatonin rhythm with antiphasic peaks could not be readily anticipated from the behavioral/environmental factors in the protocol (e.g., light exposure, posture, diet, activity) or from current mathematical model simulations of circadian pacemaker output. The observation, therefore, challenges our current understanding of underlying physiological mechanisms that regulate melatonin secretion. Interestingly, the increase in melatonin concentration observed during daytime sleep was positively correlated with the change in timing of melatonin nighttime peak (p = .002), but not with the degree of light-induced melatonin suppression during nighttime wakefulness (p = .92). Both the increase in daytime melatonin concentrations and the change in the timing of the nighttime peak became larger after repeated exposure to simulated night shifts (p = .002 and p = .006, respectively). Furthermore, we found that melatonin secretion during daytime sleep was positively associated with an increase in 24-h glucose and insulin levels during the night shift protocol (p = .014 and p = .027, respectively). Future studies are needed to elucidate the key factor(s) driving the unexpected daytime melatonin secretion and the melatonin rhythm with antiphasic peaks during shifted sleep/wake schedules, the underlying mechanisms of their relationship with glucose metabolism, and the relevance for diabetes risk among shift workers.

Keywords: circadian pacemaker; glucose metabolism; melatonin; night shift.

Figure 1.

The day shift protocol (Top) and night shift protocol (Bottom), as part of the randomized, cross-over design. 24-h melatonin profiles were assessed on days 5 and 7 in the day shift protocol and across days 5/6 and 7/8 in the night shift protocol (orange and purple dash lines, as test day 1 and test day 3, respectively). In the night shift protocol, melatonin profiles under dim light on day 4 were also assessed (brown dotted line). Light levels indicated are in the horizontal angle of gaze. Green boxes represent meals (wide) and snacks (narrow).

Figure 2.

Melatonin profiles on test day 1 and test day 3 in day shift (black) and night shift (red) protocols. (A) Representative melatonin pattern with antiphasic daytime and nighttime peaks from one participant. (B) Melatonin profiles of group average. Gray bar and red bar above the x-axis represent sleep opportunity during the day shift and the night shift, respectively. Re-plotted melatonin profiles are presented with dashed line. The insets on the top right in (B) show the average melatonin levels during the 8-h daytime windows that were compared between the two protocols. P values, statistical significance for the effect of shift (i.e., simulated day shift vs. night shift). Data are represented as mean ± SEM.

Figure 3.

Individual melatonin profiles in response to simulated night shift conditions. Each individual melatonin profile is labeled with different colors and symbols and kept the same between test day 1 (left) and test day 3 (right) in day shift and night shift protocols. (A) Re-plotted melatonin profiles are presented with dashed line. The symbols above the line profiles mark the peak timing of the corresponding individual melatonin profiles. The size of each symbol represents the magnitude of the peak. Dark gray bar, sleep opportunity in the darkness (0 lux). Light gray bar, wakefulness under dim light condition (4 lux). (B) Rayleigh plots showing the changes in the parameters of melatonin rhythm under night shift protocol as compared to the day shift protocol. The radial length and phase angle of each symbol represents the magnitudes of melatonin suppression (note: shorter radial length represents stronger suppression) and the change in peak timing, respectively, and the black arrow the aggregate phase vector. Note the blue solid circle represents MSW14, in which a nocturnal melatonin peak was absent in test day 1, and the daytime peak at ~2PM was identified as the primary peak. It was excluded from the subsequent analysis involving change in timing of nighttime melatonin peak.

Figure 4.

Melatonin increase during daytime sleep was positively correlated with change in time of nocturnal melatonin peak (left), but not with suppression of the nocturnal melatonin peak during wakefulness (right) during the simulated night shifts. Each individual is labeled with different colors and symbols. A linear regression line is shown in black with 95% confidence interval in grey. P values, statistical significance for adjusted association of melatonin increase during daytime sleep with change in peak time and suppression of peak.

Figure 5.

Relationships between changes in melatonin rhythm and changes in 24-h glucose (top) and insulin (bottom) AUC during simulated night shifts compared to day shifts. (A) Melatonin increase during daytime sleep was positively correlated 24-h glucose and insulin AUC. (B) Change in nocturnal melatonin peak timing was positively correlated with 24-h glucose AUC, but not with insulin AUC. (C) Suppression of nocturnal melatonin peak had no significant correlation with either 24-h glucose or insulin AUC. Each individual is labeled with different colors and symbols. Linear regression line is shown in black with 95% confidence interval in grey. P values, statistical significance for adjusted association between the changes in melatonin rhythm parameters and glucose/insulin measures.

Figure 6.

Summary of the relationships among changes in daytime vs. nighttime melatonin, and changes in 24-h glucose and insulin levels in response to simulated night shift protocol as compared to the simulated day shift protocol in the current study.