The aim of this study was to reassess the suitability of Cry1 as the primary magnetoreceptor in the retina of birds. Our objectives were to identify the distribution and the cellular localization of the Cry1 proteins in the retinas of zebra finches and to test whether the expression of Cry1 was wavelength dependent. Our findings confirm earlier reports from migratory songbirds and chickens that Cry1 is expressed in the outer segments of UV cones throughout the avian retina. However, we could not substantiate earlier reports of wavelength-dependent effects on Cry1 expression.
Cry1 expression in outer segments of UV cones
Our observation that Cry1 was expressed in all SWS1-labelled UV cones in the retinas of all zebra finches examined confirms earlier reports of Cry1a in UV cones of European robins and V cones of chickens 25–27. The occurrence of Cry1 proteins in only the UV cones, and no other retinal tissue, in zebra finches strongly suggests that our Cry1 antibody exclusively labelled the C-terminal of the Cry1a isoform of the protein (cf. 25–27. Cry1b, in contrast, has been found in the retinal ganglion cells, displaced ganglion cells, and inner segments of some unspecified photoreceptors in several species of migratory songbirds and pigeons (Bolte et al. 2016; Niessner et al. 2016; see also Niessner and Winklhofer 2017).
The localisation of Cry1 near the tip of the outer segments of the UV cones is supported by the close proximity to the UV opsins labelled with SWS1. The SWS1 antibody, like opsin antibodies in general, is known to only label the opsins once they are fixed in the membrane of the discs of the outer segment of the cones, but not while being transported from the nucleus in the inner segment to the base of the outer segment, where they are integrated in the membrane of the discs of the outer segment (reviewed by 59). Our findings only partly agree with the study by Niessner and colleagues (2011), which reported Cry1a expression along the full length of the outer segments of V cones in chickens, and not only at the tip of the outer segments. This difference may originate either from species-specific results, or because the antibodies, despite being designed against a fairly similar region, are not identical. The antiserum used by Niessner and colleagues (2011) was made against a 20 aa peptide in the C-terminal region of the Cry1a sequence, while the antibody we used in this study was a commercial polyclonal anti-Cry1 designed against 12 aa peptide, in the same region of the mouse Cry1 sequence and fully homologous to their 20 aa peptide. This leaves an 8 aa peptide that is recognized by Niessner et al.’s antibody but not ours. Regardless of such differences, the localisation of the Cry1 protein far from the nucleus in the inner segment argues for a function not directly involved in the negative feedback loop of the circadian clock, one of which could be magnetoreception.
According to the original radical-pair model, the putative magnetoreceptors should ideally be fixed along at least one molecular axis and be evenly distributed across a hemisphere, like the avian retina, to allow for the comparison of reaction yields of radical pairs with different alignments relative to the magnetic field 4,8,11 [note, however, that some of these requirements are not necessarily needed 49,60–62]. These requirements are met by the localization of Cry1 in the UV cones of the zebra finches, assuming that the Cry1 proteins are aligned roughly along the same axis in the cone outer segments. UV/V cones are in theory ideal candidate locations for cryptochrome-based magnetoreceptors, as pointed out earlier 25,63. Their transparent oil droplets do not filter out UV light 64,65, thus light in the UV and blue spectrum can reach the FAD in the cryptochromes, which show a high absorption of light in this wavelength range (reviewed by 2,36,37). Also, UV/V cones are the least abundant photoreceptors in the avian retina (max. 10%, depending on species; e.g., 65–67), which would minimize interference of light-dependent magnetoreception with vision (but see below).
Distribution of Cry1 across retina
We found an up to nine times higher density of Cry1/SWS1 cones in the fovea of the zebra finch retina compared to the periphery, which agrees well with the general density distributions of cones in retinas of passerine birds, which usually peak in one or two foveas and decrease towards the periphery of the retina 68–71. The fovea is an important area in the visual field with a high density of cones and none or few rods 72. This area of improved visual acuity is used for various visual tasks that require high-resolution vision, like food detection or obstacle avoidance 73. The higher density of Cry1-positive UV cones in the fovea of the zebra finch retina would suggest that, if they were magnetoreceptors, magnetic compass information to be perceived at a higher spatial resolution when viewed through the fovea. This may improve the detection of the magnetic field but could also pose a possible caveat in that the magnetic modulation pattern could interfere with important visual tasks. This will largely depend on how the signals from the Cry1 and UV receptors are processed [see 63, and discussion on the localisation of Cry1 below].
Effect of monochromatic light on Cry1 expression
One of the key indications for an involvement of Cry1 in avian magnetoreception was based on the observation that Cry1 could only be detected in chicken retinas after exposure to UV to yellow light, but not after exposure to red light or darkness 26,27. Exposure of zebra finches to 461 nm (blue), 521 nm (green) or 633 nm (red) light for one hour prior to dissection did not result in any differences in expression. We detected the Cry1 protein in the retinas of zebra finches exposed to any of the light conditions, irrespective of the spectrum and the location in the retina (centre or periphery; Fig 5). This came as a surprise, since our primary antibody was designed to detect the same region (C-terminal of Cry1), and was therefore almost identical to the one used by Niessner and colleagues 26,27. In both cases, the same antibody against SWS1 was used, in similar concentrations. Niessner and colleagues argued that their Cry1a antibody detected the Cry1a protein only after it had undergone a conformational change in the C-terminal upon activation by UV to yellow light, since they detected Cry1a only after exposing the birds to UV to yellow light prior to dissection, but not after exposure to darkness or red light 26,27. However, the assumption that Cry1 undergoes a conformational change was based on evidence from plant (Arabidopsis) and type I (Drosophila) cryptochromes, which are both known to be directly light sensitive 74,75. We are not aware of any studies showing that vertebrate type II cryptochromes undergo a conformational change in the C-terminal as a result of light activation. On the contrary, they are suggested to be vestigial flavoproteins which do not stably bind FAD and use the C-terminal for interactions with other clock proteins instead 40–42,76. Assuming that the Cry1 in the outer segments of the UV/V cones are located too far away from the nucleus to be directly involved in the circadian clock, the C-terminal should be detectable to antibodies under any light condition. We observed that the signal from the immunolabelled cells in some regions of our whole mounts was masked by the presence of remnants of the pigment epithelium. Despite being removed to the most extent and bleached to avoid darkening of the preparation, the pigment epithelium interfered with the fluorescent signal of the marked cells, making it almost not differentiable from the background, and therefore easy to be misinterpreted as non-labelled tissue. Own immunostainings of retinas of robins and chickens will have to show whether this may be the explanation for the discrepancies between studies, or whether differences between the antibodies or bird species are responsible.
Localization of Cry1 in the avian UV/V cones suggests unique function
Based on the cellular location of the Cry1 proteins at the base of the outer segments of the UV cones, Cry1 could well be the thought-after candidate magnetoreceptor of the light-dependent magnetic compass. However, the high density of the Cry1-containing UV cones in the central retina is not necessarily in support of a role in magnetoreception, even though it does not exclude this possibility. The molecular and functional properties of Cry1 also argue against its involvement in light-dependent, radical-pair-based magnetoreception. Nevertheless, the possibility remains that Cry1 might be involved in signal transduction further downstream in the signalling cascade.
If Cry1 is not involved in light-dependent magnetoreception in birds, why is it expressed in the outer segment of all UV cones across the entire retina of the zebra finch, but in none of the other photoreceptors or retinal cells? Together with the reports of Cry1a expression in UV cones of European robins and V cones of chickens 25–27 our findings suggest that Cry1 likely has a very specific function which is unique to cones expressing the SWS1 pigment and which is not required in any of the other cone or rod photoreceptors, unless other cryptochromes are expressed in those photoreceptors instead. However, to date there is no convincing evidence that either Cry1b or Cry2 are expressed in avian photoreceptors. Cry1b has been reported in ganglion cells and a few displaced ganglion cells in retinas of pigeons (Columba livia), European robins and Northern wheatears (Oenanthe oenanthe) 56,57, and possibly in the inner segment of photoreceptors, but this latter observation was only made by one of the groups 56 and could not be substantiated by the other group 57. Thus, all evidence points towards a very specific function of Cry1 in only UV/V cones. Interestingly, Cry1 has also been found to be expressed in short-wavelength sensitive SWS1 (S1) cones in representatives of some groups of mammals (Canidae, Mustelidae and Ursidae within Carnivora, Hominidae, some Cercopithecidae, and possibly Lemuridae and Callitrichidae) 77. It might suggest that the expression of Cry1 is a more widespread feature of SWS1-expressing cones, common to birds and mammals, and possibly vertebrates in general.
SWS1-expressing photoreceptors are unique in that they absorb light in the UV to V spectrum 78–80, which is the visible light spectrum with the highest energy and has been shown to damage the retina of vertebrates 81. The vertebrate ancestral state of SWS1 opsins is suggested to be UV sensitive, but was independently replaced by V sensitivity in various lineages (reviewed by 78,82). These include birds which likely possessed an ancestral V-sensitive pigment with certain lineages secondarily regaining UV sensitivity 82. Since both the UV and V cones of birds contain transparent oil droplets which contain no carotenoids and do not filter short-wavelength light 69, Cry1 located in these cones could possibly be involved in UV/V-light protection. However, in mammals UV-light below 400 nm is often absorbed by the cornea and lens 83, and there does not appear to be any relationship between Cry1 expression and the degree of UV light below 400 nm reaching the SWS1 cones in different groups of mammal 57.
Other possible functions of the Cry1 proteins in the outer segments of UV/V cones in birds could be linked to the growing body of evidence that cryptochrome proteins, independent of their role as circadian clock regulators, are directly involved in various metabolic processes, like e.g., glucocorticoid signalling 84, modulation of the cAMP pathway 85, and DNA damage response (reviewed by 86).