Extraocular Circadian Phototransduction in Humans

See allHide authors and affiliations

Science  16 Jan 1998:
Vol. 279, Issue 5349, pp. 396-399
DOI: 10.1126/science.279.5349.396


Physiological and behavioral rhythms are governed by an endogenous circadian clock. The response of the human circadian clock to extraocular light exposure was monitored by measurement of body temperature and melatonin concentrations throughout the circadian cycle before and after light pulses presented to the popliteal region (behind the knee). A systematic relation was found between the timing of the light pulse and the magnitude and direction of phase shifts, resulting in the generation of a phase response curve. These findings challenge the belief that mammals are incapable of extraretinal circadian phototransduction and have implications for the development of more effective treatments for sleep and circadian rhythm disorders.

Circadian rhythms are endogenously generated oscillations of about 24 hours that provide temporal structure to a wide range of behavioral and physiological functions. Because the endogenous clock tends to run at a period close to but not exactly 24 hours, a daily adjustment, usually by the natural light-dark cycle, is required to synchronize or entrain circadian rhythms to the external environment. Many vertebrate and nonvertebrate species have multiple photoreceptor systems through which circadian entrainment may be achieved (1-3). In the house sparrow, for example, three discrete input pathways for light to act on the circadian system have been identified (4). Similarly, a number of fish, amphibian, and reptile species have extraocular and extrapineal pathways for circadian light transduction (5).

The photoreceptors responsible for entraining the mammalian biological clock may not be the same cells that mediate vision (6). Mice homozygous for the autosomal recessive allelerd (retinally degenerate), which have no electrophysiological or behavioral visual responses to light, can be entrained to a light-dark cycle (7). Likewise, bright light suppresses melatonin output in some totally blind humans, despite the fact that they have no conscious light perception and no pupillary light reflex (8). Such findings support the hypothesis that all vertebrates, including mammals, have specialized nonvisual photoreceptors that mediate circadian responses to the light-dark cycle. It is generally assumed, however, that nonvisual circadian photoreceptors in mammals reside within the retina, and that mammals do not have the capacity for extraocular circadian photoreception (1, 2, 9). This conclusion is based on studies showing a failure of several rodent species to entrain to a light-dark cycle or to respond to pulses of light with shifts in circadian phase after complete optic enucleation (10). In addition, Czeisler and co-workers found an absence of light-induced melatonin suppression during ocular shielding in two individuals who did show suppression when light fell on their eyes (8). A decade earlier, Wehr and co-workers reported a lack of clinical response in seasonal affective disorder when patients' skin (face, neck, arms, legs) was exposed to a bright light stimulus while their eyes were shielded (11). However, in that study, no physiological measures of light response, such as melatonin secretion or temperature phase response, were obtained.

Here we present results that demonstrate that the human circadian response to light can be mediated through an extraocular route. A total of 33 phase-shifting trials were carried out in 15 healthy individuals (mean age, 35.7 years; range, 22 to 67 years; 13 males, 2 females) (12). Each laboratory session lasted for four consecutive days and nights, during which the participants were assigned randomly to either a control or an active condition. Successive laboratory visits were separated by at least 10 days. During the active sessions (phase delay, n = 13; phase advance, n= 11), light was presented at various times relative to the baseline circadian phase, in order to examine phase response throughout the circadian cycle. The extraocular light stimulus consisted of a 3-hour pulse of light presented to the popliteal region, the area directly behind the knee joint (13). The stimulus was presented in ambient light of less than 20 lux. Throughout their stay in the laboratory, when not sleeping and not involved in the experimental light manipulation, participants were in constant illumination of less than 50 lux.

On the night before (night 1 in the laboratory) and the nights after the light stimulus (nights 3 and 4) participants were required to remain in bed (and were allowed to sleep) from 2400 until noon the following day. On the light exposure night (night 2 in the laboratory), sleep was necessarily displaced to accommodate presentation of the 3-hour light pulse. With the exception of this interval, the participants were in bed from 2400 until noon on night 2 as well. Sleep was not permitted during the light exposure interval, and continuous electroencephalogram and video monitoring of participants throughout the exposure interval ensured compliance.

Body core temperature was recorded continuously (14). In a subset of sessions (n = 18), hourly saliva samples were also collected for melatonin assay (15). The nadir of the temperature rhythm and the dim light melatonin onset (DLMO) were used to evaluate circadian phase before and after presentation of the light pulse (16). The magnitude of the phase shift achieved in each trial was determined by comparing participants' baseline circadian phase (during the first 24 hours in the laboratory) with the phase determined during the final 24 hours in the laboratory.

Examples from single individuals of the phase shifts achieved as a result of light presented before (producing a delay) and after (producing an advance) the minimum of the circadian temperature rhythm (T min) are shown in Figs. 1 and 2. For all active sessions, there was a systematic relation between the timing of the light pulse and the magnitude and direction of the phase shift, resulting in a classic phase response curve (Fig.3A) . Paired ttests revealed that shifts in both the delay and advance directions were statistically significant (mean delay = 1.43 hours,P = 0.0001; mean advance = 0.58 hours,P = 0.024). Six of the seven participants who underwent both active and control conditions showed a larger shift in the active condition when compared with their own control condition (mean difference = 1.29 hours, P = 0.011). It should be noted, however, that the phase of light presentation was not matched for individual participants under active and control conditions.

Figure 1

Example of a delay in circadian phase in response to a 3-hour bright light presentation to the popliteal region. Light was presented on one occasion between 0100 and 0400 on night 2 in the laboratory (black bar) while the participant (a 29-year-old male) remained awake and seated in a dimly lit room (ambient illumination <20 lux). The circadian phase was determined by fitting a complex cosine curve (dotted line) to the raw body core temperature data (solid line). Resulting phase estimates are indicated by vertical lines. The baseline (night 1) circadian phase (A) occurred at 0404; the circadian phase after light presentation (B) (last 24 hours in the laboratory) occurred at 0708. The phase angle between the midpoint of the light stimulus and the fitted body temperature minimum at baseline was 1.57 hours. The resulting phase delay was 3.06 hours.

Figure 2

Example of an advance in circadian phase in response to a 3-hour bright light presentation to the popliteal region. Light was presented on one occasion between 0600 and 0900 after night 2 in the laboratory (black bar) while the participant (a 44-year-old male) remained awake and seated in a dimly lit room (ambient illumination <20 lux). The circadian phase was determined by fitting a complex cosine curve (dotted line) to the raw body core temperature data (solid line). Resulting phase estimates are indicated by vertical lines. The baseline (night 1) circadian phase (A) occurred at 0713; the circadian phase after light presentation (B) (last 24 hours in the laboratory) occurred at 0453. The phase angle between the midpoint of the light stimulus and the fitted body temperature minimum at baseline was 0.28 hour. The resulting phase advance was 2.34 hours.

Figure 3

Response of the endogenous circadian pacemaker, as measured by body core temperature (A) and by DLMO (B), to a single 3-hour presentation of bright light to the popliteal region. Each point represents the phase shift observed (advances are designated by positive numbers and delays by negative numbers on the y axis) in response to bright light presented at a given time relative to the phase of body core temperature at baseline. “Timing of light relative toT min” (x axis) refers to the interval between the midpoint of light presentation and the fitted temperature minimum. The magnitude of the observed phase shifts varied systematically as a function of this relation. The strong correlation between the two phase markers used (ρ = 0.704, P = 0.009) strongly suggests that the extraocular light stimulus directly influenced the endogenous circadian clock and not simply the output variables. (C) The response of the circadian clock, as measured by body temperature, to the no-light control condition. All no-light presentations occurred before T min; therefore, only that portion of the x axis is shown.

In 18 of the 24 active sessions, we assessed the phase response of a second circadian marker, the onset of the endogenous melatonin rhythm under dim light conditions (DLMO) (Fig. 3B). The direction and magnitude of the shifts in DLMO were equivalent to those for temperature. Indeed, there was a significant correlation between the shift in body core temperature and the shift in DLMO (Spearman rank-order correlation: ρ = 0.704; P = 0.009). As with temperature, delay and advance shifts in DLMO were statistically significant (mean delay = 0.92 hours, P = 0.0009; mean advance = 1.17 hours, P = 0.021).

The phase shifts in the active sessions were the consequence of the light administration and not systematically influenced by the experimental procedure itself. In the control condition, participants underwent the identical protocol as in the delay condition, including application of the fiber optic pad and activation of the exhaust fans (13). However, in the control condition, the halogen bulb providing illumination to the optic pad was disconnected. Because in all conditions the light source was not turned on until they were seated and an opaque “skirt” was in place, participants were unaware of whether light was actually being presented during a given session. Comparison of the phase of body temperature at baseline and after the control manipulation revealed no significant shift as a result of exposure to this protocol (mean shift = 0.37 hours,P = 0.103), with five individuals showing delays, three showing advances, and one showing no phase change. There was no relation between the degree or direction of shift and proximity of sham light exposure to T min (Fig. 3C). Repeated measures analysis of variance of the circadian temperature phase across the entire control condition revealed no significant change in phase (P = 0.539), confirming the reliability of temperature as a circadian phase marker. The average intra-individual standard deviation in the control condition was 25.2 min.

Our results challenge the widely held belief that mammals are incapable of extraocular circadian phototransduction. The overall temporal profile, as well as the magnitude of the phase shifts achieved with our extraocular light stimulus, is similar to those reported by investigators who used single-pulse, full-spectrum light stimuli presented to the eyes (17). The light stimulus we used was composed of a relatively narrow bandwidth (455 to 540 nm). Yet, in a pilot study with an identical protocol but with broad-band white light from commercial fluorescent light boxes (18) placed beneath participants' knees, we observed phase delays (no advances were attempted) of similar magnitude to those reported here.

Accurate characterization of the mechanisms of extraocular phototransduction has been difficult, even with respect to the ubiquitous “deep brain photoreceptors” of nonmammalian vertebrates (19). Oren has recently proposed a model in which the circulatory system plays a key role (20). This “humoral phototransduction” hypothesis posits that light of sufficient intensity, falling on a vascular surface, sets in motion a process by which neuroactive gases transported in and regulated by blood-borne photoreceptors (for example, hemoglobin in erythrocytes) stimulate the neural pathways that entrain biological rhythms. In support of the model is evidence that bright light can dissociate neuroactive gases such as carbon monoxide (CO) and nitric oxide (NO) from heme moieties (21); that light exposure can further increase circulating NO concentrations by increasing the activity of NO synthase (22); and that NO can shift circadian phase in a manner similar to light (23). These facts, when integrated with the vasodilating capability of CO and NO (24), constitute a mechanism by which photic cues can be conveyed to the circadian clock. Although the author proposes that humoral phototransduction occurs primarily by means of light falling on retinal vasculature, the hypothesis may apply to extraocular, peripherally mediated circadian phototransduction as well. Whatever the mechanism that underlies extraocular circadian phototransduction, the pathway involved is likely distinct from those currently hypothesized to be responsible for providing the human circadian clock with photic information (6).

Timed bright light exposure is an effective treatment for sleep and circadian rhythm disorders including jet lag, shift work sleep disturbance, age-related insomnia, and advanced- and delayed-sleep phase syndromes (25). The finding that the endogenous clock can be manipulated through an extraocular route could lead to the development of delivery systems and treatment regimens that may increase the effectiveness of this promising nondrug treatment. For example, treatment regimens that use extraocular light exposure may be able to take advantage of more efficient timing schedules. The nature of the phase response curve to light dictates that the largest shifts, both advances and delays, occur at times during which people are usually asleep.


View Abstract

Navigate This Article