Regulation of Mammalian Circadian Behavior by Non-rod, Non-cone, Ocular Photoreceptors

See allHide authors and affiliations

Science  16 Apr 1999:
Vol. 284, Issue 5413, pp. 502-504
DOI: 10.1126/science.284.5413.502


Circadian rhythms of mammals are entrained by light to follow the daily solar cycle (photoentrainment). To determine whether retinal rods and cones are required for this response, the effects of light on the regulation of circadian wheel-running behavior were examined in mice lacking these photoreceptors. Mice without cones (cl) or without both rods and cones (rdta/cl) showed unattenuated phase-shifting responses to light. Removal of the eyes abolishes this behavior. Thus, neither rods nor cones are required for photoentrainment, and the murine eye contains additional photoreceptors that regulate the circadian clock.

Mammals use light both to generate a visual image of their environment and to provide time-of-day information. Internal circadian time is synchronized (entrained) with the solar day by light-induced resetting (photoentrainment) mechanisms, which correct for deviations in the period and phase of the endogenous clock (1). Eye loss in both human and nonhuman mammals abolishes photoentrainment, demonstrating that the eyes provide the primary source of light information to the clock (2). However, the retinal projections that convey light information to the visual and circadian centers of the brain are quite distinct (3), and visual blindness due to partial loss of rod and cone photoreceptors is not necessarily associated with an attenuation of circadian responses to light (2). Collectively, these findings have led to speculation that (i) the mammalian circadian system can maintain normal photosensitivity with only small numbers of rods or cones; and (ii) the eye contains unrecognized photoreceptors that mediate, or help mediate, the effects of light on the circadian system (4). In the absence of an experimental model completely lacking rods and cones, distinguishing between these alternatives has been problematic.

Two mouse models have been used previously to examine the impact of rod photoreceptor loss on circadian physiology: (i) mice homozygous for retinal degeneration (rd/rd) gradually lose all rod photoreceptors but retain normal circadian responses to light (5); and (ii) transgenic mice (rdta) undergo specific ablation of rod photoreceptors during early development (6) and are also circadian photosensitive. The responses of rdta mice were about twice as great as those of wild-type and rd/rd mice of the same genetic background (7). Loss of the eyes in both rd/rd andrdta mice abolished the effects of light on the circadian system. Collectively, these results showed that rods are not required for circadian photoentrainment and that the photoreceptors mediating these responses are ocular.

Both rd/rd and rdta mice sustain a secondary degeneration of cone photoreceptors. However, limited numbers of cones remain into old age (6, 8), making them strong candidates for the regulation of temporal physiology. The murine retina contains two populations of cones, sensitive in the green [maximum wavelength (λmax) = 508 nm] (9) and ultraviolet (UV) (λmax = 359 nm) (10). Both cone classes have been implicated in photoentrainment by action spectrum studies (11). Moreover, the identification of a fully functional “green” photopigment (λmax = 534 nm) within the eyes of the blind mole rat (Spalax ehrenbergi) provides indirect evidence for the involvement of cones in circadian regulation (12).

To determine the impact of cone photoreceptor loss on photoentrainment, we used mice in which cone photoreceptors were ablated by the introduction of a synthetic transgene (cl) (13). This construct consists of a portion of the human red cone opsin promoter, attached to an attenuated diphtheria toxin gene (14). The retinas of these mice have normal numbers of rods and a substantially reduced number of UV cones (>95% lost) and appear to lack green cones (<1% remain in some retinas) (13). Our molecular (Fig. 1) (15) and immunocytochemical analysis (16) of the cl retina confirms these findings. Despite this massive insult to cone photoreceptors, cl mice showed unattenuated circadian responses to monochromatic 509-nm light (Fig. 2A). Bilateral enucleation abolished the ability of cl mice to entrain to a 12 hour light:12 hour dark cycle and to phase shift their circadian locomotor rhythm in response to a light pulse (17). In view of the loss of green cones, these data suggest that green cone photoreceptors are not required for photoentrainment. Moreover, the insensitivity of UV cones to 509-nm light (10) suggests that a non-cone photoreceptor is involved in this process. As rods remain unaffected incl mice, under these circumstances, rods might mediate photoentrainment. Although previous studies with both rd/rdand the rdta mouse models indicate that rod photoreceptors are not required for circadian photoentrainment (5,7), our results might reflect redundancy of photoreceptor inputs to the clock, with both rod and cone photoreceptors providing photic input to the circadian system. Hence, the absence of either cell type might be compensated for by the presence of the other. To resolve this issue, we generated mice that carry lesions to both rod and cone photoreceptors by introduction of the cl transgene into mice heterozygous for the rodless (rdta) transgene. Immunocytochemical and mRNA analyses ofrdta/cl mouse retinas (15) indicate that both rod and green cone photoreceptors and their associated photopigments are eliminated from the retinas of these mice (Figs. 1 and3). Despite the absence of rods and green-sensitive cones, rdta/cl mice show unattenuated circadian phase shifts in response to a 15-min monochromatic light (509 nm) pulse of varying irradiance (Fig. 2B).

Figure 1

Effect of transgenic ablation on the expression of photoreceptor genes in cl andrdta/cl mice. Northern blot (A throughC) and RT-PCR (D through F) detection of mRNA-encoding green cone opsin (A and D), UV cone opsin (B and E), and rod opsin (C and F) in wild-type and transgenic retinas (15). Introduction of the cl transgene rendered green cone opsin mRNA undetectable by Northern blot (A) in eithercl or rdta/cl genotypes. RT-PCR techniques also failed to amplify a band visible on an ethidium bromide–stained agarose gel in either genotype (D). The effect of the cltransgene on UV cones was less marked, with UV cone opsin mRNA detectable in both cl and rdta/cl mice by Northern blotting (B) and RT-PCR techniques (E). Rod photoreceptors were unaffected by the cl transgene. By contrast, therdta/cl retina contained no rod opsin transcript (C and F); bp, base pairs.

Figure 2

Irradiance-dependent phase shifts of circadian locomotor activity (17). (A) Phase shifts of locomotor activity in cl mice. Phase shifts (mean ± SEM) of wild-type and cl mice, after exposure to a defined irradiance, 15-min monochromatic light (509 nm) pulse delivered at CT16 (n = 6 to 15 animals per genotype at each irradiance). There were no significant differences between cl or wild-type mice at irradiances that produce either saturating or subsaturating phase shifts [two-way analysis of variance (ANOVA): P > 0.05]. (B) Phase shifts of locomotor activity in rdta/cltransgenic mice. Phase shifts (mean ± SEM) of wild-type andrdta/cl mice, after exposure to a 15-min monochromatic light (509 nm) pulse delivered at CT16 (n = 5 to 7 animals per genotype at each irradiance). Both genotypes showed an irradiance-dependent increase in the amplitude of phase shifts. However, at an irradiance of 5.7 μW/cm2, phase shifts were significantly enhanced in rdta/cl mice, compared with wild-type mice (two-way ANOVA: P < 0.001; post hoc Student–Newman-Keuls tests comparing genotypes at each irradiance: *,P < 0.05). For further discussion, see (23).

Figure 3

Histological analysis of serial sections from wild-type (A through C) andrdta/cl (D through F) retinas (15). Immunocytochemical staining failed to identify rod or cone photoreceptors in the retinas of rdta/cl mice. Tissue was fixed with Bouins (75% picric acid, 25% formalin, and 5% acetic acid) for 24 hours and paraffin-embedded, and 8-μm sections were treated with antibodies recognizing rod (A and D), rod and green cone (B and E), and UV cone (C and F) photoreceptors. Visualization was accomplished with ABC methods (Vectastain Elite, Vector Labs). GCL, ganglion cell layer; IPL, inner plexiform layer; INL, inner nuclear layer; ONL, outer nuclear layer; IS, inner segments; OS, outer segments; and RPE, retinal pigment epithelium. Scale bar, 40 μm.

These results demonstrate that the mammalian eye contains non-rod, non-cone photoreceptors capable of regulating circadian behavioral responses to light. Published data suggest strongly that these receptors use a vitamin A–based photopigment (11, 18). Nonetheless, their molecular basis has been the subject of considerable recent speculation (19–22). The rdta/cl model provides an opportunity to address this issue by determining the spectral sensitivity of these uncharacterized photoreceptors.

  • * Present address: School of Biological Sciences, University of Surrey, Guildford, Surrey GU2 5XH, UK.

  • To whom correspondence should be addressed. E-mail: r.foster{at}


View Abstract

Navigate This Article