CRYPTOCHROME mediates light-dependent magnetosensitivity in Drosophila

The ability of an animal to detect geomagnetic fields has substantial biological relevance as it is used by many invertebrate and vertebrate species for orientation and navigation purposes, including homing, building activity and long-distance migration^2,4. Three general modes of magnetoreception have been proposed⁵. One mode is electromagnetic induction by the Earth’s magnetic field as may occur in electrosensitive marine fish, although there is scant evidence supporting such sensing. The two other modes, for which experimental evidence does exist, are a magnetite-based process^6–8 and chemical-based reactions^9,10 that are modulated by magnetic fields. One chemical model of magnetoreception proposes that magnetic information is transmitted to the nervous system through the light-induced product of magnetically sensitive radical-pair reactions in specialized photoreceptors³.

CRYs are flavoproteins that have been postulated to generate magnetosensitive radical pairs that could provide a photoinduced electron transfer reaction for the detection of magnetic fields³. CRY proteins are best known for their roles in the regulation of circadian clocks^11,12 and can be categorized into two groups based on current phylogenetic and functional relationships^13,14. Drosophila-like CRYs are sensitive to light in the UV-A/blue range¹⁵ and function primarily as photoreceptors that synchronize (entrain) circadian clocks. Vertebrate-like CRYs, which have also been found in every non-drosophilid insect so far examined¹⁴, do not appear to be directly light sensitive. Instead, vertebrate-like CRY proteins are potent repressors of the CLOCK and CYCLE/BMAL1 transcription factors, which as heterodimers, drive the intracellular transcriptional feedback loop of the circadian clock mechanism in all animals studied.

Although there is good behavioural evidence for the involvement of short-wavelength photoreceptors in the detection of a geomagnetic field^5,16–18, an essential link between CRY and magnetoreception has not been established in any animal. Drosophila is ideally suited to investigate a role for CRY as a magnetoreceptor, because they only have the light-sensitive CRY¹⁴, whose action spectrum peaks in the UV-A range (350 – 400 nm) with a plateau in the near blue (430 – 450 nm)^19,20. Importantly, flies that lack CRY²¹ or harbor the severely hypomorphic cry^b mutation^22,23 can be used to evaluate the role of CRY in magnetosensitive responses.

We initiated our studies by developing a novel behavioural assay for magnetosensitivity in Drosophila (Fig. 1a). In this illuminated apparatus, flies experience a magnetic field generated by an electric coil system and display their magnetosensitivity in a binary-choice T-maze. The two-coil system is ideal for behavioural studies of magnetosensitivity, because it produces a magnetic field on one side of the T-maze, while producing no field on the opposite side. This design eliminates non-magnetic differences such as heat generated by the electric coils between sides during test sessions²⁴. Flies were tested either for their response to the magnetic field in the naïve state (naïve group) or following a training session pairing the field with sucrose reward (trained group).

Wild-type Canton-S, white-eyed w;Canton-S, Oregon-R-S, and Berlin-K strains all developed a learned preference for a magnetic field (Fig. 1b). The trained groups in the two Canton-S lines showed the greatest response to the field (P=0.002, one-way ANOVA) and were the only ones to show a naïve avoidance of the field (P<0.0001, one-sample t-test). Thus, Drosophila consistently show magnetosensitivity that varies in magnitude in a strain-dependent manner. The similarity of behavioural responses between red-eyed, wild-type Canton-S flies and white-eyed w;Canton-S flies shows that eye color does not substantially alter behavioural responses to the magnetic field.

Because wild-type Canton-S flies showed the most robust trained and naïve responses of the strains tested, we used them to determine whether the magnetic responses we observed were light-dependent. We assayed naïve and trained Canton-S flies under different long-wavelength pass filters that transmitted wavelengths of light at > 500 nm, > 420 nm, or > 400 nm (Fig. 2a). In contrast to flies assayed under full-spectrum light (Fig. 1b and Fig. 2a), flies did not exhibit either naïve or trained responses to the field when wavelengths <420 nm were blocked (Fig 2b). Because the filter which blocked light <420 nm, also caused a 13% decrease in total irradiance (Fig. 2c, red line), we examined whether the filter-induced lack of behavioural responses to the magnetic field was secondary to the decrement in irradiance. When Canton-S flies were studied under full-spectrum light, with a total irradiance level lower than that imposed by the filter (Fig. 2c, blue line), the flies still showed significant naïve (P=0.0005, one-sample t-test) and trained responses to the magnetic field (Fig. 2d). Thus, the filter-induced loss of behavioural responses to the magnetic field is due to the loss of short wavelength light.

Behavioural responses to the magnet were partially restored when 400–420 nm light was included (Fig. 2b), which is consistent with the action spectrum of Drosophila CRY tailing into the near blue¹⁹, and, as expected, the trained response was weaker than under full-spectrum light (full spectrum vs. > 400 nm, P<0.001, Student’s t-test). This wavelength-dependent effect of the magnetic field on behaviour suggests that Drosophila has a photoreceptor-based magnetosensitive system. Moreover, because the response to the magnetic field requires UV-A/blue light (<420 nm) (Fig. 2e), these data are consistent with the hypothesis that CRY can function as a magnetoreceptor in Drosophila.

We next used CRY-deficient cry⁰ mutant flies to directly examine whether CRY is required for magnetosensitive behaviour. We tested two of the newly generated cry⁰ fly lines, because in cry⁰ flies, the entire cry coding sequence has been replaced with mini-white⁺ by homologous recombination, ensuring that, unlike in the more commonly used CRY-defective cry^b flies, there is no possibility of residual CRY activity²¹. In addition, the three cry⁰ fly lines (cry⁰¹ through cry⁰³) were backcrossed independently into a w¹¹¹⁸ background²¹. Thus, we were able to use the appropriate w¹¹¹⁸ control flies to test the contribution of the cry gene in magnetosensitive behaviour.

Control w¹¹¹⁸ flies exhibited a clear naïve preference for, rather than avoidance of, the magnetic field (Fig. 3a). The difference in the direction of the naïve response to the magnetic field between Canton-S flies and the w¹¹¹⁸ line re-emphasizes the importance of controlling for genetic background for studies of magnetosensitivity in flies. Nonetheless, like Canton-S flies, the naïve response of w¹¹¹⁸ flies to the magnetic field was light dependent; the naïve preference for the magnetic field was abolished in the absence of UV-A/blue light (<420 nm) (Fig. 3a).

Homozygous cry⁰² flies lacking CRY did not show a naïve response to the magnet under full-spectrum light, in contrast to the significant naïve responses manifested by both w¹¹¹⁸ and heterozygous cry⁰²/+ flies (Fig. 3b). Training control w¹¹¹⁸ flies to prefer the magnetic field under full-spectrum light significantly enhanced their naïve preference for the field (Fig. 3c). In contrast, homozygous cry⁰¹ flies did not show either a naïve preference for the field (like cry⁰² flies) or an enhanced preference for the field after training (Fig. 3c). The loss of the response to the magnetic field in the CRY-deficient flies resembled the behaviour when w¹¹¹⁸ flies are deprived of UV-A/blue light (Fig. 3a), which is consistent with CRY being the relevant light sensor. These data using two cry null strains strongly suggest that both naïve and trained responses to the magnetic field in Drosophila require CRY function.

The CRY-defective cry^b mutant flies are also unable to respond to the magnetic field; cry^b is a chemically-induced missense mutation that renders CRY^B essentially non-functional^22,23. Because the genetic background of cry^b mutant flies is not well-defined, we compared behavioural responses to the magnetic field between homozygous cry^b flies and heterozygous cry^b/Canton-S flies. Whereas homozygous cry^b flies did not show either naïve or trained responses to the magnetic field under full-spectrum light, heterozygous cry^b/Canton-S flies showed significant naïve (P=0.0004, one-sample t-test) and trained responses (Fig. 3d); the trained response in the heterozygotes was less than that of wild-type Canton-S flies (Fig. 1b) and likely results from differences in genetic background.

To rule out non-cry mutations as the reason for the lack of magnetic responses in cry^b mutants, we showed that the cry^b mutation fails to complement the cry⁰¹ null mutation. Transheterozygous cry^b/cry⁰¹ flies did not show significant naïve or trained responses to the magnet, while heterozygous cry⁰¹/Canton-S and cry^b/Canton-S flies did (naïve response, P=0.006, one-sample t-test; Fig. 3d). Taken together, these data indicate that the cry locus is necessary for light-dependent magnetosensitivity in Drosophila. Furthermore, the lack of a trained response in both cry⁰¹ and cry^b mutant flies is consistent with CRY being an essential component of the magnetosensitive sensory input pathway and perhaps the magnetoreceptor itself.

Because light-activated CRY interacts with the critical circadian clock protein TIMELESS to reset the circadian clock mechanism²⁵, we examined whether an intact circadian system is necessary for the CRY-dependent magnetosensitive responses in wild-type Canton-S flies. Circadian arrhythmicity was induced by constant light (LL) which disrupts circadian clock function in CRY-containing cells by causing the constant degradation of not only CRY, but also TIMELESS and then PERIOD²⁵. We subsequently tested behavioural responses to the magnetic field after at least five days in LL when the flies were shown to exhibit arrhythmic locomotor behaviour (Fig. 4a), disrupted PERIOD abundance rhythms (Fig. 4b), and to express constant low levels of CRY (Fig. 4c). Strikingly, these arrhythmic flies continued to show significant naïve (P=0.004, one-sample t-test) and trained responses to the magnetic field (Fig. 4d). Thus, the continuous activation of CRY by light does not disrupt its ability to sense the magnet, and an intact circadian system is not required for the magnetoreception mechanism to operate.

There are two other published reports of magnetosensitivity in Drosophila^26,27. One describes behavioural evidence that male wild-type Oregon-R flies exhibit a light-dependent magnetic compass response in a radial maze whereas female flies did not respond to the magnet²⁷. Additionally, male flies responded in opposite directions when tested under either 365 nm or 500 nm light. In our studies, both male and female flies showed a magnetic response. Regardless of experimental differences, both the previous study²⁷ and ours demonstrate that fruit flies can respond to a magnetic field in a wavelength-dependent manner.

Our results extend substantially the presence of a light-dependent magnetic sense in Drosophila by showing the necessity of CRY. We cannot distinguish unequivocally whether fly CRY functions as the actual magnetoreceptor or is an essential component downstream of the receptor. CRY is necessary for both the naïve and trained responses to the magnetic field which is consistent with the notion that CRY is in the input pathway of magnetic sensing. In addition, the continued behavioural responses to the magnet in LL, in which the known CRY signalling components are being constantly degraded and the circadian clock is rendered non-functional, is also consistent with an input function. The most compelling evidence supporting a magnetoreceptor role for CRY is that the CRY-dependent behavioural responses to the magnetic field require UV-A/blue light, which matches the action spectrum of Drosophila CRY^19,20.

Our behavioural assay for magnetosensitivity does not currently have a pure directional component, and therefore it is difficult to directly relate our findings to the use of geomagnetic fields for animal orientation and navigation. Nevertheless, it is likely that the response we have identified is the prototype for CRY’s involvement in chemical-based magnetic sensing. Thus, our findings open new avenues of investigation into the cellular and molecular basis of chemical-based magnetic sensing in animals. The powerful genetics of Drosophila will facilitate an understanding of the precise mechanism of action of CRY in magnetosensitivity, such as the actual involvement of magnetosensitive radical pairs produced by photoinduced electron transfer reactions²⁸. Our data further show that the biological functions of Drosophila CRY extend beyond those in circadian clocks.