Expression patterns of cryptochrome genes in avian retina suggest involvement of Cry4 in light-dependent magnetoreception

Abstract

The light-dependent magnetic compass of birds provides orientation information about the spatial alignment of the geomagnetic field. It is proposed to be located in the avian retina, and be mediated by a light-induced, biochemical radical-pair mechanism involving cryptochromes as putative receptor molecules. At the same time, cryptochromes are known for their role in the negative feedback loop in the circadian clock. We measured gene expression of Cry1, Cry2 and Cry4 in the retina, muscle and brain of zebra finches over the circadian day to assess whether they showed any circadian rhythmicity. We hypothesized that retinal cryptochromes involved in magnetoreception should be expressed at a constant level over the circadian day, because birds use a light-dependent magnetic compass for orientation not only during migration, but also for spatial orientation tasks in their daily life. Cryptochromes serving in circadian tasks, on the other hand, are expected to be expressed in a rhythmic (circadian) pattern. Cry1 and Cry2 displayed a daily variation in the retina as expected for circadian clock genes, while Cry4 expressed at constant levels over time. We conclude that Cry4 is the most likely candidate magnetoreceptor of the light-dependent magnetic compass in birds.

1. Introduction

Organisms from a broad variety of taxa are able to detect the magnetic field of the Earth and use it for compass information to determine their movement directions over short and long distances [1–4]. The magnetic compass of various invertebrates, amphibians, birds and some mammals has been shown to be dependent on light, i.e. to be functional only under specific intensities and wavelengths of light [5–12]. Magnetoreception of the light-dependent magnetic compass has been suggested to be based on a radical-pair process involving light-sensitive magnetoreceptors [13–17]. Upon photon excitation, an electron is transferred from a donor to an acceptor molecule, resulting in a radical pair in which the two unpaired electrons exist in either a singlet or triplet state. The interconversion between the singlet and triplet intermediates can be affected by an external magnetic field, whereby the ratio of singlet and triplet products depends on the orientation of the radical pair to the magnetic field [4,14,15,17,18]. For a radical-pair process to function, the receptor molecules need to be able to form long-lived, spin-correlated radical pairs upon light excitation that the magnetic field can act upon. Furthermore, they should ideally be distributed in a spherical array to allow for the comparison between receptors aligned at different angles to the magnetic field [14,15].

In birds, the receptors of the light-dependent magnetic compass have been proposed to be located in the avian retina, as the eye is optimized for photon collection and its semi-spherical shape fulfils the requirement for the radical pairs to be aligned at different angles to the magnetic field [14–16]. Theoretically, they can be located in any retinal structure, like photoreceptors or ganglion cells [17]. The magnetoreceptors do not necessarily need to be fixed to any membranes, if the polarization of light reaching the magnetoreceptors is taken into account [19]. We recently showed that overhead polarized light modulated magnetic compass orientation in zebra finches that were trained to magnetic compass cues in a cross maze [20]. This indicates that polarized light directly affected the avian light-dependent magnetic compass, suggesting that the magnetoreceptors are selective for polarized light, which in turn relaxes the constraint that they need to be rotationally restricted [19,20].

Cryptochromes have been proposed as the putative magnetoreceptor molecules, because they are the only known vertebrate photopigments to form long-lived, spin-correlated radical pairs upon light excitation [14]. Cryptochromes are flavoproteins and share a high sequence homology with photolyases, which repair damaged DNA using light energy [21–23]. Cryptochromes are known as developmental factors in plants and circadian clock proteins in animals, where they are involved in regulating internal 24 h cyclic rhythms in behaviour, metabolism and physiology which are entrained daily with input from the environmental light/dark cycle [21–23]. Critical for a role in light-dependent magnetoreception is that the cryptochrome binds the cofactor FAD (flavin adenine dinucleotide), which transforms into an excited state upon excitation by light and forms a radical pair with a yet unknown partner [15,17,24].

Cryptochromes are widely distributed in the majority of animal taxa, including all vertebrate classes, and can be classified in the following main functional groups [21–23,25,26]: (1) (insect or invertebrate) type I cryptochromes are found exclusively in invertebrates and have retained light sensitivity and act as the primary photoreceptors that entrain the circadian clock; (2) (vertebrate) type II cryptochromes are found in vertebrates and some insects and do not appear to be directly sensitive to light, but instead act as part of the negative feedback loop in the circadian oscillator by inhibiting CLOCK/BMAL1-driven transcription and thereby repressing their own expression; (3) type IV cryptochromes are found in fish, frogs and birds, and have retained the capability to harvest light, like the type I cryptochromes, and could thus act as the circadian photoreceptors or alternatively be involved in light-dependent magnetoreception [27,28]. In birds, four members of the cryptochrome gene family have been identified in the retinas of migratory and non-migratory birds (reviewed by [29]): three members of animal type II cryptochromes, Cry1, with the two isoforms, Cry1a and Cry1b, and Cry2, and one type IV cryptochrome, Cry4 (for references, see below).

Cry1 is widely found in the retinas of birds. Cry1a is expressed in the cytoplasm of UV/V cones in a light-dependent manner consistent with magnetic compass orientation [30,31]. Cry1b has been localized in the cytoplasm of ganglion cells, displaced ganglion cells and inner segments of photoreceptors [32–35]. With their cytoplasmic localization and topographic distribution both Cry1 s would theoretically fulfil the requirements of the radical-pair model [14,15]. In nocturnally migrating songbirds, night-time expression of Cry1 in the retina was reported to be upregulated during the migration period, which was argued to be a strong indication for an involvement of Cry1 in avian magnetoreception [32,35–37] (but see [34]). Such upregulation was only observed when the birds showed migratory restlessness during the night, and not when they were resting [36], suggesting a role of Cry1 in either night activity or magnetoreception.

However, Cry1 expression usually follows clear circadian oscillations with low expression levels during the dark phase and a steady increase during the light phase, culminating with peak expression in the middle to the end of the circadian day, followed by a relatively rapid decrease in expression during the circadian night [33,38–42]. In the retina of chickens, this rhythmicity was shown to persist under constant light or constant dark conditions for at least 1-2 days, suggesting that the expression of Cry1 is not directly light sensitive, but that it is an integral part of the circadian clock [33,43,44]. This circadian behaviour argues against an involvement of Cry1 in magnetoreception, as animals are expected to use their magnetic compass for orientation not only during long-distance migration, but in their daily life, as shown in a growing number of studies with non-migratory birds [12,20,45,46]. Furthermore, recent evidence suggests that vertebrate type II cryptochromes do not bind FAD in vivo, thus they may not be able to directly harvest light to generate a radical pair [26]. There is the possibility, however, that a yet unknown antenna pigment is involved in light harvesting [9], whereby Cry1 could nonetheless be involved in avian magnetoreception [26]. Cry1 might be involved in optimizing retinal processes for vision at night. It is well established that avian retinas have an autonomous clock (reviewed by [47–49]), which regulates numerous processes of retinal physiology, like e.g. changes in the sensitivity of the retina to light, shift of rod–cone dominancy between day and night, or the timing of disk shedding [50,51]. Nocturnally migrating birds have been shown to invert neuronal activity in the visual centres in the brain during migration [52], thus they likely also adjust their visual system to this temporary nocturnality. An upregulation of the expression of genes involved in magnetoreception during nights when nocturnally migrating songbirds actively migrate is also possible.

Cry2 has been found to be expressed in the retinas of several bird species [32,38–43,53]. Compared to mammalian Cry2, which shows a clear circadian rhythmicity [54,55], however, the circadian nature of Cry2 in birds is less clear. Some studies have reported no rhythmicity [38,44], whereas others have found support for a tendency [39] or clear rhythmicity [40,42]. Its mRNA expression in the inner nuclear layer, the photoreceptor layer and to a limited degree in the ganglion cell layer [43] would argue for a role in primary photo- or magnetoreception. However, because of its nuclear, rather than cytoplasmic, localization the involvement of Cry2 in sensory tasks has been largely excluded [32,53].

Cry4 has been described in the retinas of several bird species [27–29,39,42]. It has been located in the cytoplasm of the visual pigment layer, the inner nuclear layer and the ganglion cell layer in chickens [27,28,39]. By contrast to the type II cryptochromes, Cry4 has a high affinity to FAD [42,55], and does not inhibit CLOCK:BMAL1 transactivation, thus it is not part of the negative feedback loop of the circadian clock [56]. Most important for a potential role as a magnetoreceptor is that retinal Cry4 can undergo structural changes in the carboxyl-terminal region in a light-dependent manner consistent with light-dependent magnetic compass orientation in birds [27,28,56]. A recent study found that Cry4 expression follows no circadian rhythmicity, and might, therefore, be a likely candidate as an magnetoreceptor of the light-dependent magnetic compass [42].

Nevertheless, there is no consensus to date over which of the cryptochromes, if any, is involved in avian light-dependent magnetoreception [17,24,57]. In invertebrates, various types of behavioural responses to magnetic fields have been reported to be mediated by both insect type I and vertebrate type II cryptochromes [58–62]. In birds, however, direct evidence for the involvement of specific cryptochromes in avian magnetoreception is still missing. In this study, we focused on the expression patterns of three cryptochrome genes, Cry1, Cry2 and Cry4, in the zebra finch retina over the circadian day, measured by real-time quantitative PCR (qPCR), to evaluate their possible involvement as primary receptor molecules in avian light-dependent magnetoreception. Samples from brain and muscular tissue were used as controls. Zebra finches are small, non-migratory passerines that occur naturally throughout most of the Australian territory (reviewed by [63,64]). They have been shown in spatial orientation experiments to use their magnetic compass at any time of day and irrespective of time of year [12,20,46]. We, therefore, hypothesized that cryptochrome genes in the retina that are involved in magnetoreception should be expressed at a constant level over circadian time, whereas cryptochromes not involved in magnetoreception, but instead involved in circadian tasks, should be expressed in a rhythmic (circadian) pattern.

2. Material and methods

2.1. Animals

The study species was the zebra finch (Taenopygia guttata Reichenbach 1862). The experimental animals belonged to a permanent captive breeding colony at Stensoffa Field Station, near Lund (Sweden). The birds were held indoors under a constant 12 L : 12 D cycle, without input from the natural light, and, therefore, not changing with season. The birds belonged to a heterogeneous group of sexually mature individuals, including both males and females.

2.2. Tissue collection

The samples were collected at Stensoffa Field Station during the summer and autumn of 2015 and 2016 from 39 adult zebra finches (25 males and 14 females) at three time points (Zeitgeber Time 2 (ZT2, 15 birds), ZT8 (12 birds) and ZT14 (12 birds), ±30 min; ZT0 = lights on) (electronic supplementary material, table S1). Birds were sacrificed by cervical dislocation followed by decapitation. We collected whole retinas without sclera, muscular tissue from the extrinsic eye muscles and 3 mm³ sections of brain tissue from the dorsal side of both brain (cerebrum) hemispheres. The samples were dissected, placed in RNAlater (QIAGEN, Hilden, Germany) at room temperature within 5–7 min post-mortem, and transported to Lund University where they remained at room temperature until RNA extraction within the following 48 h. From each individual bird, three data points were obtained, corresponding to the three tissues examined at each time point (ZT). All laboratory work was carried out in the Molecular Ecology and Evolution Lab (MEEL), Biology Department, Lund University, Sweden.

2.3. RNA extraction and cDNA synthesis

RNA was extracted from samples stabilized in RNAlater with the RNeasy Mini Kit (QIAGEN) according to the manufacturer's instructions. The amount of starting tissue ranged between 0.01–0.02 g, weighed at the time of dissection. The tissue was disrupted and homogenized using a TissueLyser system (QIAGEN), using 5 mm diameter stainless steel beads at 30 Hz for 2 min. RNA concentration and purity was determined with a Nanodrop 2000 (Thermo Fisher, Hvidovre, Denmark), with 260/280 absorbance ratio above 1.7 to be acceptable. Integrity of the RNA was determined with a Bioanalyzer 2100 and the RNA Nano Kit (Agilent, Santa Clara, USA). Only samples with a RIN (RNA Integrity Number, [65]) greater than 7 were accepted und used in further analyses. cDNA was amplified from the extracted RNA with the RETROScript—Reverse Transcription for RT-PCR kit (Ambion-Life Technologies, Austin, USA), following the manufacturer's instructions for two-step RT-PCR. Nuclease-free water was used as a control for non-reverse transcriptase experiments (No RT).

2.4. Primer design and qPCR

Specific primers for the qPCR were designed using Geneious 8 (www.geneious.com, [66]). Criteria for the primer design were melting temperatures between 58° and 60°, primer length of 20–25 nt, guanine–cytosine content between 40 and 60% and resulting amplicons of 120 base pairs for the cryptochromes. All primers were designed to span exon–exon boundaries to avoid amplification of genomic DNA contaminating the sample. Two sets of primer pairs were designed for Cry1 (there are no reports of Cry1 isoforms in zebra finches), Cry2 and Cry4, and one set of primer pairs was designed for two different reference genes (previously optimized for use with avian species, specifically the zebra finch): HPRT (hypo-xanthine-guanine phosphoribosyl-transferase 1) and PGK1 (phospho-glycerate kinase 1) [67] (electronic supplementary material, table S2).

The qPCRs were carried out with the Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen, Carlsbad, USA), following the manufacturer's instruction for a 20 µl final volume. All experiments were run in a MX3000P qPCR detection system and the MXPro software (Stratagene, San Diego, USA). The cycling conditions were: 50°C, 2 min step followed by a denaturation step at 95°C, 2 min step. Then 40 cycles of 95°C for 15 s and 60°C for 30 s, followed by standard melting curve analysis (95°C, 1 min; 55°C, 30 s; 95°C, 30 s). No-RT products (reaction without reverse transcriptase) from the cDNA step were used as negative controls. Specificity of the primers was confirmed by BLAST against the zebra finch genome. The primers were successfully tested in the different tissue samples of the zebra finch, evaluating that each primer should amplify a single product, reflected as a single peak in the melting curve analysis. Also, the efficiency of the amplification stayed between 97% and 103%, determined by measuring the quantification cycle (Cq value) 10-fold dilution over five orders of magnitude. Only one of the two sets of primers for Cry1, Cry2 and Cry4 was used subsequently.

The qPCR data were analysed as described by [68]. Normalized relative quantities (NRQ) of each biological replicate were calculated using the methods by [69] and [70] for multiple reference genes and biological replicates. NRQ were log2 transformed prior to statistical analysis. One-way ANOVA tests and post hoc Tukey tests (Rstudio. www.rstudio.com, Boston, USA) were used to test for differences in expression levels between time points or between tissues. Data coming from the same individual (different tissues from the same bird) was also analysed using individual as a random effect to take the non-independence among data points, and the results were the same. ANOVA results are shown for all tests because all the assumptions were met. NRQ linear values were used for illustration purposes.

3. Results

An initial comparison between the expression pattern of the different cryptochromes in the left and right retinas of individual zebra finches did not reveal any significant differences between the eyes (p > 0.100, one-way ANOVA). Therefore, we subsequently analysed RNA from either just one retina or a mix of both retinas of a bird. See electronic supplementary material, table S3 for detailed statistics on all tests performed.

Cry-specific expression levels varied significantly between the different tissues (figure 1): expression levels of any of the three cryptochromes were significantly higher in the muscle tissue compared to the other tissues (one-way ANOVAs: Cry1, p < 0.001; Cry2, p < 0.001; Cry4, p < 0.001). Expression levels in the retina were similar to those in the brain for Cry1 and Cry2 (p > 0.200, Tukey, figure 1a,b), whereas Cry4 was expressed at higher levels in the retina compared to the brain tissue (p < 0.001, Tukey).

Figure 1.

Normalized relative quantities of cryptochrome mRNA in retina, muscle, and brain tissue of the zebra finch. (a), Cry1; (b) Cry2; (c) Cry4. Boxplots show min and max values (whiskers), first and third quartiles (box limits), and median (box inner line) of the normalized relative quantities of mRNA. Dots in the centre of the boxplot represent mean values of the normalized relative quantities of mRNA. Significance levels are given as follows: *p < 0.05, **p < 0.01, ***p < 0.001. (Online version in colour.)

The expression patterns of Cry1, Cry2 and Cry4 over the circadian day varied between genes and tissues (figure 2). Cry1 showed a daily variation in the retina, with increasing expression over the circadian day (p < 0.050, one-way ANOVA), with significant difference between ZT2 and ZT14 (p < 0.050, Tukey; figure 2a). In contrast, Cry1 expression did not differ over time in the muscle tissue (p = 0.628, one-way ANOVA; figure 2d) or in the brain (p = 0.182, one-way ANOVA; figure 2g).

Figure 2.

Normalized relative quantities of cryptochrome mRNA of Cry1, Cry2 and Cry4 genes in retina, muscle, and brain tissues of the zebra finch at Zeitgeber Times ZT2, ZT8 and ZT14. (a–c) Expression in the retina; (d–f) expression in the muscle; (g–i) expression in the brain. Lights on: ZT0-ZT12, lights off: ZT12-ZT24. figure 1 for further information. (Online version in colour.)

Cry2 showed a daily variation in the retina and muscle tissues, but contrary to Cry1, the expression pattern seems to decrease over the circadian day. The variation in the retina is significant (p < 0.005, one-way ANOVA), with higher expression at ZT2 (which was significantly higher than the expression at ZT8, p < 0.005, Tukey; figure 2b), A similar pattern is observed in muscle (p < 0.050, one-way ANOVA), being significantly higher at ZT2 (compared to ZT8, p < 0.050, Tukey; figure 2e). The expression of Cry2 in brain tissue (figure 2h) did not vary between time points (p = 0.557, one-way ANOVA).

The expression pattern of Cry4 differed from the expression of both Cry1 and Cry2. In the retina, Cry4 expression was constant over time (p = 0.836, one-way ANOVA; figure 2c). Expression in muscle or brain tissue did not show any variation either (p = 0.057, one-way ANOVA figure 2f; p = 0.554, one-way ANOVA; figure 2i, respectively).

4. Discussion

In this study, we focused on the circadian expression patterns of three cryptochrome genes in the zebra finch retina to evaluate their possible involvement as primary receptor molecules in avian light-dependent magnetoreception. In previous studies, it has been shown that the use of a magnetic compass for orientation is not restricted to the migratory season or a specific time of day and that it is functional throughout the circadian day and throughout the year [12,20,46,71]. We, therefore, hypothesized that the expression of putative receptor genes involved in light-dependent magnetoreception in the zebra finch retina should be independent of circadian rhythmicity.

4.1. Cry expression in the retina

We found a clear difference between the circadian expression patterns of the two vertebrate type II cryptochromes, Cry1 and Cry2, in the zebra finch retina on one hand, and the expression pattern of the type IV cryptochrome, Cry4, on the other hand (figure 2). Cry1 and Cry2 displayed a clear daily variation, which suggests the involvement of these cryptochromes in circadian activity, whereas the expression of Cry4 was constant over circadian time, which agrees with an involvement in light-dependent magnetoreception.

Cry1 expression in the zebra finch retina showed a clear circadian rhythmicity with the lowest levels measured just after lights on and the highest levels just after lights off (figure 2a). This variation is comparable to Cry1 expression in the retinas of other birds, with low expression levels during the dark phase and a steady increase during the light phase, culminating with peak expression in the middle to the end of the circadian day, followed by a relatively rapid decrease in expression during the circadian night [33,36,38–41]. At first sight our results appear to disagree with an earlier study on Cry1 expression in zebra finch retinas [32], which found moderate expression levels during the ‘day’ and almost no expression at ‘night’. However, when the circadian times of the measurements are taken into account, our results agree well with these earlier findings. The ‘day-time’ values in [32] were measured after constant illumination of 12–14 h, just before the lights were switched off in the holding room, i.e. around the birds' end of the day. This corresponds with the circadian time when Cry1 expression in our study and in other species approached maximum levels. The ‘night-time’ values reported by [32] were measured 2–4 h after lights off, thus during those hours of the birds' circadian night, when Cry1 expression in our and other studies had started to decrease towards baseline levels.

The cyclic behaviour of Cry2 that we observed in the zebra finch retina, with higher expression levels at the onset of the light period (figure 2b), agrees with findings in other bird species [39,40,42] (but see [38,44]). Cry2, like Cry1, therefore, likely acts as a core circadian clock protein [22,26], rather than as a magnetoreceptor. This is supported by recent evidence suggesting that vertebrate type II cryptochromes do not bind FAD in vivo and, therefore, are not inherently photosensitive and may not be able to directly generate a radical pair [26]. In addition, the localization of Cry2 in the cell nucleus, where most of the circadian feedback cycle takes place [22], makes an involvement of Cry2 in primary magnetoreception highly unlikely [32,53]. Thus, we conclude that the circadian expression of Cry1 and Cry2 in the zebra finch retina argues against their involvement in avian magnetoreception. Cry1 and Cry2 might still be involved in the signalling pathway of a magnetoreceptor, but not in the primary reception process of the light-dependent magnetic compass in birds [26].

In contrast to the circadian expression pattern of Cry1 and Cry2, we found a constant expression of Cry4 mRNA in the zebra finch retinas across the measured time points (figure 2c). This is consistent with previous reports of Cry4 expression in the retinas of chickens and European robins, where no statistical difference was found between different circadian times, despite of some variation in the measurements [27,39,42]. Such a constant expression, and thereby the availability of receptor proteins throughout the day and night, is consistent with a potential role in magnetoreception, as outlined above. As a type IV cryptochrome, avian Cry4 contains bound FAD and is thereby inherently photosensitive and can form a radical pair upon light activation [26–28,42]. Indeed, retinal Cry4 from birds has recently been shown to undergo structural changes in the carboxyl-terminal region in a light-dependent manner consistent with magnetic compass orientation in birds [27,28,56]. However, considering that a growing number of studies has shown that birds use their magnetic compass not only for orientation during migration, but also for short-distance spatial orientation task in their daily life [12,20,45,46], we argue that any magnetoreceptor should be expressed at equal levels irrespective of time of day and season, thus would not be expected to follow a circadian expression pattern. In conclusion, our data suggest that Cry4 is the most likely candidate for the magnetoreceptor of the avian light-dependent magnetic compass, and that Cry1 and Cry2 are part of the retinal circadian clock in zebra finches.

4.2. Expression of cryptochromes in peripheral tissue

It is well established that the central circadian pacemaker in birds is comprised three oscillators located in the retina, the pineal gland, and in the hypothalamus, which each receives independent photic input (reviewed by [48,72]). These three independent clocks interact with each other, although the degree of interaction varies greatly between species and environmental conditions [37,40]. In addition to these central clocks, clock genes may be rhythmically expressed also in peripheral tissues, like brain regions not belonging to the hypothalamus or the pineal, different organs, muscle and skin [38–40,72,73]. However, the expression patterns of different clock genes in these peripheral tissues vary greatly between tissues and also relative to the central clock, indicating a complex organization in the regulation of the different clocks [37,38,40,41,48,73].

In brain and muscle tissue, we found a significant difference in expression levels between time points, indicating circadian rhythmicity, only in Cry2 expressed in the muscle tissue, and a tendency of a rhythmic expression in Cry4 in the muscle. The expression pattern of Cry2 in the muscle (figure 2e) followed the Cry2 expression in the retina, similar to what has previously been found in chickens [39]. The expression of Cry4 in the zebra finch muscle (figure 2f) was slightly elevated towards the end of the light phase, an observation that has previously been made in avian Cry4 in heart tissue [39]. It is currently unclear why cryptochrome genes express rhythmically in some, but not in other, peripheral tissues. The absence of daily rhythmicity might be the result of not having a measurement point for the middle of the dark phase, which could be masking a rhythmicity in some cases. It is also relevant to point out that an absence of daily rhythmicity does not necessarily mean an absence of circadian behaviour in the transcribed protein. In mice, for example, Cry1 genes exhibited a daily variation, while Cry2 remained constant throughout the day; both resulting proteins, however, displayed a clear circadian rhythm, being highly present in the suprachiasmatic nucleus (SCN) during the day, but not during the night [74]. Taken together, our findings provide further evidence for the diversity of expression patterns shown by the different cryptochromes in different tissues.

4.3. Overall expression in different tissues

A comparison of the overall expression of the cryptochrome genes between tissues revealed that the expression levels of all three cryptochrome genes were higher in muscle tissue than in the retina or brain (figure 1), which was rather unexpected. However, there might be a simple explanation related to the type of muscle that we analysed. We used the extraocular muscles of the zebra finch eye, easily obtained during the enucleation of the eye. During the dissection of these muscles, special care was taken to avoid inclusion of tissue from the optic nerve and adjacent fat and lacrimal glands. However, the Harderian gland, which lubricates the nictitating membrane in most members of tetrapods, is embedded between the extraocular muscles on the posterior side of the eyeball, and has been reported to be of considerable size in the house sparrow (Passer domesticus) [75]. In blind subterranean mole-rats (Spalax ehrenbergi) the Harderian gland has been found to express high levels of Cry1 and Cry2 with a clear circadian rhythmicity [76]. We cannot exclude that we collected parts of the Harderian gland with the muscle samples in our zebra finches, which may explain the much higher cryptochrome expression levels in the muscle compared to the retina and brain tissues.

5. Conclusion

Taken together, our results strongly suggest that Cry4 is the putative magnetoreceptor of the light-dependent magnetic compass in birds. It is expressed in the retina of zebra finches at constant levels across the day, independent of a circadian rhythm. Also, as a type IV cryptochrome it has the molecular properties to be directly light sensitive, thus it can initiate a light-dependent reaction as required by the radical-pair mechanism. Cry1 and Cry2 on the other hand are expressed with clear circadian rhythmicity in the zebra finch retina. This indicates that they are more likely involved in the retinal circadian clock, like type II cryptochromes in other vertebrates. However, more studies are needed to explain why migratory birds upregulate Cry1 expression on nights with active migratory restlessness. If the upregulation is connected to magnetic compass orientation, we would expect a similar upregulation also in day migrants during days of active migration, or in zebra finches during active training in our magnetic compass assay [12,20]. It is important to note that future studies investigating cryptochrome expression to study the possible involvement of cryptochromes in animal magnetoreception should be carried out under strictly controlled circadian conditions.

Ethics

Tissue samples were taken following ethical guidelines approved by the Malmö-Lund Animal Ethics Committee (permits M 423-12, and M 24-16).

Data accessibility

Normalized Cq values for all experiments are provided as a supplementary Excel file.

Authors' contributions

A.P.-R., S.B. and R.M. designed the study. A.P.-R. carried out the laboratory work and analysed the data. A.P.-R. drafted a first version of the manuscript. A.P.-R. and R.M. wrote the final version of the paper with input from S.B.

Competing interests

The authors declare no competing interest.

Funding

This work was funded by Vetenskapsrådet (2011-4765 and 2015-04869 to R.M.), Crafoordska Stiftelsen (2010-1001 and 2013-0737 to R.M.) and Colciencias (Grant 568 from Departamento Administrativo de Ciencia, Tecnología e Innovación to A.P.-R.).