Effects of Exposure to Intermittent versus Continuous Red Light on Human Circadian Rhythms, Melatonin Suppression, and Pupillary Constriction

Classical visual photoreceptors (i.e., rods and cones) are required for sight, but not for non-visual light responses including entrainment of circadian rhythms, inhibition of pineal gland synthesis of the hormone melatonin, and the pupillary light reflex [1]–[3]. In the absence of rod and cone function, these light responses are mediated by intrinsically photosensitive retinal ganglion cells (ipRGCs) that contain the photopigment melanopsin [4], [5]. Melanopsin-containing ipRGCs project directly to the suprachiasmatic nucleus (SCN) in the hypothalamus to regulate light resetting of circadian rhythms and melatonin suppression, and to the pretectal area to mediate light-induced pupillary constriction [6], [7]. The ipRGCs respond preferentially to short-wavelength light in the blue portion of the visual spectrum (λ_max = ∼480 nm) [8], [9], hence non-visual light responses are most sensitive to blue light in blind humans or sightless mice with intact function of the inner retina [3], [4], [10], [11].

Melanopsin cells respond directly to light but can also be activated extrinsically by light stimulation of rod and cone photoreceptors in the outer retina [8], [9], [12]. Consequently, non-visual light responses remain intact even when melanopsin is rendered dysfunctional [13], [14]. In melanopsin null mice, exposure to light during the subjective night induces circadian phase resetting and inhibits mRNA levels for arylalkylamine N-acetyltransferase [5], the rate-limiting enzyme of the melatonin biosynthetic pathway. Circadian phase resetting, melatonin suppression, and pupillary light responses are only eliminated when rod-cone and melanopsin signaling pathways are disrupted simultaneously [4], [5]. These studies demonstrate that classical visual photoreceptors can drive non-visual light responses, but the role of cone photoreceptors remains controversial. Mice that lack rod and melanopsin function, but have intact cones, do not show reliable circadian photic entrainment [15], [16]; however, phase shift responses to short-duration light stimuli (1 or 5 min) are attenuated in mice that lack functional mid-wavelength-sensitive cones (M-cones) [17]. In humans, the sensitivity of melatonin suppression to 555-nm green light, which stimulates the three-cone photopic system maximally, decreases gradually over time during exposure to continuous light [18]. Similarly, the magnitude of pupillary constriction decreases gradually during exposure to green light [10], and there is a short-wavelength shift in spectral responses over time [10], [19]. Together, these studies suggest that cone photoreceptors contribute to non-visual light responses at the beginning of light exposure, but their role diminishes relative to melanopsin with increasing duration of light.

In principle, the contribution of cones to non-visual light responses could be enhanced by giving intermittent dark pulses that allow cone photoreceptors the opportunity to dark adapt between light pulses. As shown for circadian phase resetting in mice [15], and the pupillary light reflex in humans [10], exposure to alternating long-wavelength light and darkness can increase some non-visual light responses by activating cone photoreceptors repeatedly. In the present study, we therefore tested the hypothesis that human volunteers exposed to intermittent red light and darkness would show larger circadian phase shifts, and stronger suppression of melatonin, relative to individuals exposed to continuous red light.

To assess the contribution of cone photoreceptors to circadian responses, participants were exposed to 6 h of continuous red light or alternating red light and darkness (631 nm, 13 log photons cm⁻² s⁻¹; n = 8 in each group) during the early biological night, as part of a 6-day laboratory study (Fig. 1A–B). An additional group of participants was exposed to bright white light (2,500 lux; n = 8), which served as a positive control for evaluating the relative magnitude of non-visual responses to red light. Circadian rhythms of melatonin, cortisol, core body temperature, forehead skin temperature, and heart rate were assessed on the day before and after the light exposure procedure (Fig. 2). There was a main effect of lighting condition on phase resetting, such that bright white light induced phase shifts that were greater in magnitude compared to either of the red light conditions (F_2,20.2 = 8.76, P = 0.002; post-hoc P<0.003 for both comparisons). There was no difference in circadian responses to continuous versus intermittent red light (post-hoc P = 0.69) assessed across different physiologic measures (Table 1), however, and the average phase shift response to red light was close to an hour in both conditions (Fig. 3A). By comparison, bright white light elicited an average phase shift of −2.64 h, equivalent to resetting the circadian system about 40° of longitude in the westward direction. Although the magnitude of phase shift responses did not differ significantly across physiologic measures (F_4,84.0 = 1.12; P = 0.35), there were large inter-individual differences (Wald Z test of between-subject variance; z = 2.41, P = 0.016) in phase shift responses (Fig. 3A). Based on the average phase shift response across all available circadian rhythm markers, 6 out of 16 participants exposed to red light showed phase delay shifts greater than an hour (continuous, n = 4; intermittent, n = 2). While most subjects showed little or no circadian response to red light, 2 participants showed phase resetting responses that were similar in magnitude to those who were exposed to bright white light (Fig. 3B–C), demonstrating that red light can elicit robust circadian responses in some individuals.

Next, we compared melatonin levels during the light exposure session with levels measured in dim light on the previous day (Fig. 4A). Exposure to bright white light significantly reduced the area under the curve of the melatonin rhythm relative to continuous or intermittent red light (F_2,16 = 12.60; P<0.001; post-hoc P<0.001 for both comparisons). Although subjects who were exposed to bright white light showed larger phase shift responses and stronger melatonin suppression than those who were exposed to red light, the degree of melatonin suppression did not associate with the magnitude of circadian phase resetting when assessed across participants in all groups (n = 19, Spearman's rho = −0.23, P = 0.34). A reduction in melatonin AUC of greater than 10% was observed in only 2 participants exposed to red light (crl11, 24.9%; irl12, 36.9%), and the 2 individuals who showed the largest phase shifts (crl30 and irl31; Fig. 3) did not exhibit melatonin suppression (Fig. 4B). By comparison, all participants exhibited pupillary constriction during exposure to red light (Fig. 4C–D). At the beginning of exposure to continuous red light, pupil diameter decreased to about half of the dark-adapted size and then increased gradually over time. During exposure to intermittent red light and darkness, each light pulse elicited a similar response, and pupil diameter dilated toward the dark-adapted state during intervening dark periods. In bright white light, pupillary constriction was strong and sustained, with pupil diameter decreasing to about a third of the size measured in darkness.

Our results show that exposure to alternating red light and darkness in humans did not result in stronger circadian phase resetting or melatonin suppression, as compared to exposure to continuous red light. Hence, contrary to our initial hypothesis, repeatedly activating cone photoreceptors did not enhance circadian or melatonin suppression responses to red light. These findings contrast with results for the pupillary light reflex, which can be enhanced nearly two-fold by exposing the eyes to intermittent light [10]. The present findings are surprising, as it was recently shown that alternating red light and darkness can enhance circadian phase shift responses in transgenic mice carrying human L-cone opsin (Opn1mw^R mice) substituted for the native M-cone opsin [15]. In that study, exposure to 43 min of alternating red light and darkness (1 min on, 2 min off; 644 nm) induced circadian resetting responses that were about an hour greater than exposure to 15 min of continuous light (range, ∼12.5 to 13.5 log photons cm⁻² s⁻¹), even though total illumination time was the same for each condition. In our study, total illumination time in the intermittent light condition was half the amount for subjects exposed to continuous red light. Hence, we cannot rule out the possibility that our results would more closely resemble those obtained in Opn1mw^R mice if we had equated light exposures by total illumination time. It should be noted, however, that phase shift responses were smaller in our subjects compared with Opn1mw^R mice, despite an additional 2 h 45 min of exposure to light, suggesting that results in the mouse model might not translate directly to humans.

To our knowledge, our study is the first to examine the effects of intermittent long-wavelength light on human circadian responses. In prior work, exposure to alternating cycles of bright white light (∼10,000 lux) and dim light (<15 lux) elicited phase shifts similar in magnitude to continuous light [22], even when total illumination time was reduced to 23% of the continuous light exposure [23]. As such, intermittent light was found to be more efficient than continuous light at resetting circadian rhythms when assessed on a per unit time basis. More recently, it was reported that the human circadian system can respond to a series of 2-ms pulses of moderately bright light (473 lux) given over the course of an hour [24]. Therefore, similar to non-visual light responses measured in rodents [25], [26], the circadian system in humans shows a remarkable capacity to temporally integrate light information across intervening periods of darkness. Since we tested only one combination of wavelength, corneal irradiance, and frequency (i.e., 631 nm, 13 log photons cm⁻² s⁻¹, ∼1 min lights on/off), it is possible that other intermittent long-wavelength light stimuli might prove more effective at resetting circadian rhythms and suppressing melatonin.

It is well-established that circadian and melatonin suppression responses are short-wavelength sensitive in the photopic visual range [27]–[30]. In the present study, several subjects showed circadian phase shift responses to red light that were greater than an hour, including 2 individuals who exhibited phase shifts that were comparable in magnitude to responses induced by bright white light. Based on prior work, the red light stimulus we used was outside the range of spectral sensitivity for the intrinsic melanopsin cell response [9], [31], suggesting that visual photoreceptors are capable of resetting the circadian system in humans. These findings are consistent with previous work demonstrating that exposure to 5 hours of red light in the morning over three consecutive days reset the human melatonin rhythm about an hour earlier [32]. More recently it was shown that exposure to low-irradiance 555-nm green light (<12.5 log photons cm⁻² s⁻¹) shifted human circadian rhythms by a greater amount than that expected for a melanopsin-only response [18]. Given that Opn1mw^R mice show significant phase shift responses to long-wavelength red light [15], and mice lacking M-cones show deficits in circadian responses to short-duration light stimuli [17], we hypothesize that our results for red light-induced resetting of circadian rhythms are explained by activation of cone photoreceptors. Rods can elicit non-visual light responses in the photopic visual range in mice [33], however, and transgenic mice with disruption of rod and melanopsin signaling do not reliably entrain to light-dark cycles despite having intact cone function [15], [16]. To assess whether these findings translate to humans, non-visual light responses could be studied in patients with selective loss of either rod or cone photoreceptor function, e.g. individuals with congenital stationary night blindness or congenital achromatopsia.

In the present study, we did not observe a significant melatonin suppression response to long-wavelength red light, which is consistent with prior work demonstrating that only very bright red light exposures (e.g., 630 nm, 18 log photons cm⁻² s⁻¹) suppress melatonin synthesis in humans [27], [34]. Nonetheless, several participants appeared to show circadian phase resetting, and all subjects exhibited pupillary constriction in response to red light. These findings suggest that response thresholds differ across non-visual light responses. Similar to our results, a previous study reported no correlation between the magnitude of phase resetting and the amount of melatonin suppression in response to moderately-bright broad-bandwidth red light [32]. Also, exposure to flashes of broad spectrum white light can reset human circadian rhythms without a reduction in salivary melatonin concentration [24]. Response thresholds for circadian phase shifting and suppression of pineal melatonin also differ substantially in golden hamsters [35], and similar findings have been reported for circadian photic entrainment and pupillary constriction in mice exposed to dim red light [36]. These differences in sensitivity might arise at the level of ipRGCs [37], [38], or could be due to differences in central processing of light information.

An important limitation of the present study is that we did not include a negative control group, and therefore we cannot be sure that the average phase resetting response to red light was greater than the amount of circadian drift in the absence of light exposure. Based on a previous study that used the forced desynchrony protocol to assess circadian period [39], we would expect circadian phase to drift later by ∼20 min over 2 days in most adult males. Since several participants exposed to red light exhibited phase delay responses that exceeded an hour, and 2 individuals displayed phase shift responses that were greater than −2 h (subjects crl30 and irl31; Fig. 3), we consider it unlikely that our results can be attributed solely to circadian drift. To assess accurately the contribution of circadian drift to phase shift responses, it would be necessary to measure circadian period in each subject. Assessing circadian period was beyond the scope of the present study, but the time difference between circadian phase and the sleep cycle (i.e., the phase angle of entrainment) has been shown to correlate with circadian period [40], [41]. It is worth noting, then, that the time difference between the CBT rhythm minimum and wake time was in the normal range in subjects crl30 and irl31 (4.2 h and 3.8 h before wake time, respectively), as compared to phase angle measured across all subjects (mean ± SD = 3.8 h±1.1 h). Hence, it is unlikely that the larger-than-average phase shift responses in subjects crl30 and irl31 can be explained by an unusually long circadian period relative to other subjects, although we did not test this directly.

Another limitation of our work is that we studied a relatively small number of subjects. We had the statistical power to detect a difference of about 1 hour in circadian phase resetting between groups. We therefore cannot rule out the possibility that, with a much larger group of subjects, a small but significant difference would be found between groups exposed to continuous versus intermittent red light. Our findings suggest, however, that the effect size of such a difference would be very small. We found that individual differences in circadian light responses were substantial, but we did not assess whether such differences are stable and reproducible, as participants completed the protocol only once. Aside from potential trait-like differences in circadian sensitivity, other potential sources of variability for resetting responses include small differences in circadian timing of light exposure, and intrinsic variability in the SCN pacemaker and/or rhythms that were assessed. Also, we did not use a mydriatic agent to dilate subjects' pupils, as one of the goals was to assess whether an intermittent light stimulus could potentially be used to enhance circadian responses in a real-world setting where the pupils are free to respond to light. Therefore, we cannot exclude the possibility that our results for circadian resetting and melatonin suppression were affected by differences in retinal irradiance across light exposure groups, as a result of between-group differences in the magnitude of pupillary constriction [42].

In conclusion, this study provides evidence that cone photoreceptors contribute to circadian phase resetting in some individuals. The basis for inter-individual differences in circadian phase resetting remains to be determined, but our results suggest that some people are very sensitive to red light and can respond to stimuli that target classical visual photoreceptors. In such individuals, exposure to even dim light in the late evening hours could potentially delay the circadian clock and the onset of sleep. Future studies should therefore examine whether individual differences in sensitivity of the circadian system to light modulate relative risk for delayed sleep phase disorder and social jet lag.

We thank study volunteers, as well as research staff and students in the Chronobiology and Sleep Laboratory for their assistance in carrying out these studies. Additional support was provided by staff at the SingHealth Investigational Medicine Unit.

Research was supported by the National Medical Research Council, Singapore under NMRC/NIG/1000/2009 (JJG); and the Duke-NUS Signature Research Program funded by the Agency for Science, Technology, and Research, Singapore, and the Ministry of Health, Singapore. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.