Role of Mouse Cryptochrome Blue-Light Photoreceptor in Circadian Photoresponses

See allHide authors and affiliations

Science  20 Nov 1998:
Vol. 282, Issue 5393, pp. 1490-1494
DOI: 10.1126/science.282.5393.1490


Cryptochromes are photoactive pigments in the eye that have been proposed to function as circadian photopigments. Mice lacking the cryptochrome 2 blue-light photoreceptor gene (mCry2) were tested for circadian clock-related functions. The mutant mice had a lower sensitivity to acute light induction of mPer1 in the suprachiasmatic nucleus (SCN) but exhibited normal circadian oscillations of mPer1 and mCry1 messenger RNA in the SCN. Behaviorally, the mutants had an intrinsic circadian period about 1 hour longer than normal and exhibited high-amplitude phase shifts in response to light pulses administered at circadian time 17. These data are consistent with the hypothesis that CRY2 protein modulates circadian responses in mice and suggest that cryptochromes have a role in circadian photoreception in mammals.

Circadian rhythms are oscillations in the biochemical, physiological, and behavioral functions of organisms with a ∼24-hour periodicity (1). Circadian rhythms are synchronized with light-dark cycles, but the molecular basis of this “photoentrainment” is not known (2). Indeed, there is no consensus on the nature of the circadian photoreceptor. Three classes of pigments have been considered as candidates: opsin/retinal-based photopigments (3), tetrapyrrole-based heme pigments (4), and pterin/flavin-containing cryptochrome blue-light photoreceptors (5, 6). Cryptochromes were first identified in plants as structural homologs of the DNA repair enzyme DNA photolyase (7), but they lack DNA repair activity (8) and are involved in mediating growth (9), flowering time (10), and phototropism (11) in response to blue light. Recently, two human and mouse homologs of the plant cryptochromes were discovered (5, 12). Cryptochromes 1 and 2 (CRY1 and CRY2, respectively) lack DNA repair activity (5) and are expressed in the mouse retina (6), and mCry1 exhibits circadian oscillations of expression in the SCN (6) wherein the central pacemaker of the body resides. These observations led to the proposal that cryptochromes were likely to be photopigments for circadian photoentrainment (6).

To test this hypothesis, we created a mouse strain that lacks the predominant form of cryptochrome found in the mouse retina, CRY2, and analyzed its circadian behavior using biochemical and behavioral tests. We generated Cry2 +/− heterozygous mutant mice by established methods (13) using the targeting construct shown in Fig. 1A. Interbreeding of heterozygotes yielded progeny of wild type:Cry2 +/−:Cry2 −/−at a ratio of 1:2:1 (Fig. 1B). The mutant mice were physically and behaviorally normal. We used mice that were at least 6 weeks old and age and sex matched. To determine if CRY2 is a circadian photoreceptor, we tested the light response of the mutant mice using biochemical and behavioral assays. Because the mutant mice still express the CRY1 photoreceptor, we expected a reduction in, rather than elimination of, the circadian photoresponse reactions.

Figure 1

Generation of Cry2 mutant mice. (A) mCry2 targeting scheme. On top is a linear representation of the structural map of CRY2 indicating the approximate locations of binding sites for methenyl tetrahydrofolate (MTHF) and flavin adenine dinucleotide (FAD) as represented in the genomic clone. The top line shows a partial restriction map of the gene, indicating the 1.1 kb that was deleted and replaced by the neomycin (Neo) gene. After homologous recombination (targeting DNA, middle) diagnostic Southern (DNA) hybridization of the recombinant allele in ES cells and mice detected the presence of an Xba I site introduced by the targeting vector that leads to a change of fragment length from 3.8 kb in the wild-type allele to 2.5 kb in the recombinant allele (targeted mCry2 gene, bottom). Primers from the 3′ end of the Neo gene and the 3′ end to the 1.5-kb arm of homology yield a PCR product of 1.6 kb from the recombinant gene. Bg, Bgl I; H, Hind III; N, Not I; X, Xba I. (B) Southern blot analysis of mouse genomic DNA. About 10 μg of mouse genomic (tail) DNA was digested with Xba I and subjected to electrophoresis through a 0.8% agarose gel, which was then probed with 32P-labeled DNA made from the mCry2 genomic clone in (A).

We first quantified induction of mPer1 mRNA expression in the SCN by acute light pulses. mPer1 is a candidate circadian clock gene (14) that is expressed in the SCN with a robust circadian rhythmicity in 12-hour light:12-hour dark (LD12:12) cycles, reaching a maximum at zeitgeber time 4 (ZT4) and a minimum at ZT20 (by convention ZT0 is the time at which the light is turned on). Importantly, mPer1 is one of the immediate-early genes in the SCN that can be induced with acute light pulses at night when mRNA levels are low (15). To test whether CRY2 protein functions as a circadian photopigment, we exposed wild-type and Cry2 null mice to two subsaturating light doses at ZT18 and quantified mPer1 induction in the SCN by in situ hybridization (Fig. 2A). At both light doses, the induction of mPer1 in the SCN ofCry2 −/− mice was reduced by 50 to 60% relative to the wild type. This reduction was roughly proportional to the reduction of total cryptochrome mRNA levels in the retina of the mutant mice (mCry2 mRNA represents about 70% of the total) and suggests that Cry2 −/− mice are deficient in photic induction of mPer1.

Figure 2

(A) ReducedmPer1 induction by acute light exposure in the SCN ofCry2 −/− mutant mice.Cry2 +/+ and Cry2 −/−mice kept under a LD12:12 cycle were exposed to a light pulse (2 μmol m−2 s−1 or 50 μmol m−2 s−1 for 30 min) at ZT18, returned to darkness for 30 min, and then decapitated. Coronal brain sections (18 μm thick) were sampled from the rostral to the caudal end (images left to right) of the SCN. The sections were hybridized to a35S-labeled probe (24) consisting of nucleotides 539 to 1481 of mPer1 cDNA (GenBank accession numberAB002108). Autoradiograms were digitized with a COFU 4815 charge-coupled device camera, and NIH Image 1.6 software was used for quantitation. All the mPer1 signals from the rostral to the caudal regions of the SCN were summed (18 to 24 sections). The transcript levels are expressed relative to that inCry2 +/+ mice after 50-μmol m−2 s−1 light pulse. The differences between the wild-type and mutant mice were significant (P < 0.01) at both light fluences. The bars represent SEM (n= 4 to 5). The photographs to the right of each column are from representative autoradiograms of two sets of experiments. For clarity, only every other section is shown. (B) Circadian expression of mPer1 and mCry1 in the SCN ofCry2 +/+ and Cry2 −/−mice. Animals maintained under a LD12:12 cycle were kept in DD for 54 hours and then time points were taken for in situ hybridization every 4 hours starting at CT18 on the basis of the LD12:12 phasing. Coronal brain sections encompassing the SCN were subjected to in situ hybridization with 35S-labeledmCry1 (24) or mPer1 probes. Quantitative analyses were performed as in (A), and the values are expressed relative to the highest wild-type figures. Data points from a representative experiment are shown. Symbols: open circles,mPer1 in wild type (wt); closed circles,mPer1 in Cry2 −/−; open triangles,mCry1 in wt; closed triangles, mCry1 inCry2 −/−. Note the phase delay, but normal amplitude, in the Cry2 −/− mice relative to theCry2 +/+ animals. The phase difference between the two is consistent with the longer free-running period in theCry2 −/− mice (see Fig. 3).

Although the most likely explanation for defective induction ofmPer1 in Cry2 −/− animals is a reduction in the amount of the photoactive pigment mediating the response, the data are also consistent with an alteration of the circadian clock itself. That is, the reduction could be due to a change in the circadian regulation of mPer1 in the SCN, rather than simply an alteration at the photoreceptor level. To determine if there were alterations in endogenous rhythmic expression of clock-related genes in the SCN of the Cry2 −/− mice, we measured the mRNA levels of mPer1 and of mCry1. Both these genes exhibit robust circadian oscillations in wild-type mice (6, 15), and they represent two different rhythmic outputs of the clock in the SCN. There were no differences in the amplitudes of the circadian rhythms of expression of eithermPer1 or mCry1 in the SCNs of wild-type andCry2 mutant mice (Fig. 2B), suggesting that the basic oscillations of the circadian clock in the SCN are normal inCry2 −/− mice. The phases of mCry1and mPer1 mRNA rhythms in the mutant mice, however, were shifted by about 4 hours relative to the wild type on the third cycle of constant darkness, which is consistent with a lengthening of the period by about 1 hour in the mutant.

We used locomotor activity (wheel-running) as a behavioral output to examine the mutant mice for the three basic circadian properties: intrinsic period, persistence of rhythmicity under constant conditions, and ability to synchronize with light/dark cycle (photoentrainability). There were no obvious differences in periods or amplitudes of activity among wild-type, Cry2 +/−, andCry2 −/− animals in LD12:12 cycles (Fig. 3, A to C). Similarly, there were no significant differences among Cry2 genotypes in the phase angle of entrainment to LD12:12 as determined from the initial free-run in constant darkness (DD). However, as might be expected from a partial disruption of entrainment function, theCry2 −/− mice had significantly greater variance (wild type, 595 min; Cry2 +/−, 1150 min; Cry2 −/−, 2801 min; F(9,11) = 4.70; P < 0.01) in the free-run phase than did wild-type mice. The variance of heterozygotes was intermediate, but not significantly different from wild type. These data suggest thatCry2 −/− mice are capable of normal entrainment to bright-light cycles, but the increased variance suggests that they do so with less precision than wild-type animals (16).

Figure 3

Effect of Cry2 genotype on circadian rhythm in mice. (A to D) Effect on free-running period of locomoter activity. Representative free-running activities of (A) wild-type, (B) Cry2 +/−, and (C)Cry2 −/− animals are presented. Mice maintained in LD12:12 cycles were placed in DD conditions on the day indicated by the line. Panel (D) shows the calculated free-running periods for each animal (open circles) and the group averages (closed circles) with the SEM for each group. (E to H) Effect on phase shifts in response to saturating light pulses. (E) Wild-type, (F) Cry2 +/−, and (G)Cry2 −/− mice were maintained in DD conditions for at least 10 days before and after a 6-hour light pulse was given at CT17 on the day indicated by an arrow. The average value for the phase shifts between CT16.5 and CT18.5 in each group of mice (n = 10, 10, and 6 for Cry2 +/+,Cry2 +/−, and Cry2 −/−animals, respectively) is shown in (H) along with the SEM.

In contrast to its modest effect on photoentrainment, theCry2 −/− genotype had a drastic and paradoxical effect on the free-running period in DD (Fig. 3, A to D). TheCry2 −/− mice had significantly (P < 0.05) longer free-running periods than eitherCry2 +/− or wild-type mice (Fig. 3D). One homozygous Cry2 null animal became arrhythmic on transfer to DD but exhibited a persistent rhythm after exposure to a saturating light pulse (17). Thus, the Cry2 null mutation appears not to affect the persistent rhythmic behavior under constant conditions but clearly causes an increase in the length of the intrinsic period (18). Lengthened circadian period may result from an alteration in some input process to the circadian pacemaker, such as an alteration mimicking constant light (Aschoff's rule) (19), or from an alteration in the pacemaker mechanism itself. To examine these two possibilities, we determined phase shifts of the activity rhythms in response to light pulses as a measure of the circadian clock's responsiveness to light. The phases of the light pulses were determined according to the phase and period of the activity rhythm for each individual mouse, so that pulses were given at a predetermined circadian time (CT12 being defined as the onset of activity). The animals were exposed to saturating light pulses of 6-hour duration (17) at CT 17.5 + 1 with at least 10 days between pulses to determine the responses to light in the phase delay and near the “breakpoint” (or maximum) region of the mouse phase response curve to light (20).

The resulting data show that the response to saturating light pulses near the breakpoint (∼CT18) was significantly different among the three genotypes (Fig. 3, E to H). The magnitude of phase shifting was highest in the Cry2 −/− mice and lowest in the wild-type animals (21). A high-amplitude phase shift with light pulses given near the breakpoint is indicative of “type 0” phase resetting (22) and can occur when the strength of the stimulus is enhanced or amplified, or when the amplitude of the pacemaker is reduced. Because the mPer1 and mCry1circadian mRNA rhythms in the mutant animals are normal (Fig. 2B), we suggest that the lack of CRY2 enhances the strength of the inducing signal.

The combination of reduced photic sensitivity (Fig. 2A) and increased phase shifts in response to light pulses (Fig. 3, E to H) is not without precedent. Increased phase shifts combined with reduced light sensitivities have been reported in old golden hamsters (23), and old mice show a lengthening of the circadian period in constant darkness and a reduced light sensitivity (24). Whether these phenomena are caused by alterations in the circadian photoreceptor remains to be determined. The marine dinoflagellate Gonylaux polyhedra contains two circadian photosensors, and complex interactions occur between these two photoreceptive pathways (25). InArabidopsis, Cry2 mutants antagonize the phytochrome B pathway that regulates flowering time (10). The increased phase shifts in the null animals implies that, in mice, CRY2 may antagonize phase-shifting responses mediated by CRY1 or another photoreceptor.

A second unexpected effect of the Cry2 mutation was the lengthening of the free-running period. If cryptochromes function solely as photoreceptors, then they would not be expected to affect circadian rhythms in constant darkness. However, inArabidopsis, photoreceptor mutants also have lengthened circadian periods (26). Although the “dark effect” of CRY2 seems counterintuitive, it is not surprising. The prototype of this class of proteins, DNA photolyase, in addition to repairing DNA by phototransduction, performs a dark function by interacting with the excision repair system when light is absent (27). Conceivably, CRY2 may directly interact with components of the pacemaker mechanism.

Recent evidence from mouse (28) and Drosophila(29) indicates that the molecular components of the clock are localized in the nucleus. If CRY2 functions at the interface of signal input and the clock mechanism, it should also be nuclear. Nuclear localization of cryptochromes is inferred by the presence of a bipartite nuclear localization sequence Pro-Lys-Arg-Lys-X13-Lys-Arg-Ala-Arg (where X13 represents 13 nonconserved amino acids) in mouse and human hCRY2 (5) and by the finding that hCRY2 interacts with the nuclear serine-threonine phosphatase 5 (30). To examine the subcellular localization of CRY2, we constructed a hCRY2-GFP (green fluorescent protein) fusion gene, transfected HeLa cells, and determined the localization of the fusion protein by fluorescence microscopy. CRY2 was found almost exclusively in the nucleus (Fig. 4), consistent with a possible role in circadian photoreception.

Figure 4

Nuclear accumulation of CRY2-GFP fusion protein after transient transfection. HeLa cells were seeded at a density of 2 × 106 cells per 100-mm dish and transfected with 5 μg of either pEGFP (vector control) orpHCRY2-GFP construct by using LiipofectAMINE (Life Technologies). The cells were analyzed 4 hours after transfection with an Olympus fluorescence microscope. Magnification about ×400.

In conjunction with data from other organisms, our data suggest that cryptochromes are likely to be circadian photoreceptors. First, aNeurospora strain defective in riboflavin biosynthesis has a severely compromised circadian photoentrainment potential, suggesting that the Neurospora circadian photoreceptor, like cryptochrome, is a flavoprotein (31). Second, inDrosophila, the action spectrum for phase shifts reveals a peak in the 420- to 480-nm range (32), which is consistent with the absorption maximum (420 to 430 nm) of Drosophilacryptochrome (33). Furthermore, Drosophiladefective in the visual photoreception-phototransduction system because of either β-carotene deficiency (34), mutations in rhodopsin encoding genes (33), or mutations in genes involved in visual signal transduction (35, 36) are nearly normal in their photoentrainment, suggesting that the pigment for circadian entrainment in Drosophila may be cryptochrome and not an opsin. Third, rd/rd (retinal degeneration) mice lose all their rods and virtually all their cones upon aging (2,3); although there is no detectable opsin in the retina of 5-month-old animals (37) and their electroretinograms are flat (2), these blind mice exhibit normal circadian photoresponses (37, 38). The action spectrum for phase shifts in aged rd/rd mice has a peak at 480 nm (39), which is consistent with a role for cryptochromes, and significantly, the retinal ganglion cell and inner nuclear layers (where mCry1 and mCry2 genes are expressed) (6) are intact (38).

Finally, our results are consistent with a dual circadian photoreceptor system in mice, with CRY2 functioning as one of the photopigments. The role of CRY1 or other noncryptochrome molecules as circadian photopigments remains to be determined.

  • * To whom correspondence should be addressed. E-mail: bsiler.biochem{at}


View Abstract

Navigate This Article