Cryptochromes in Mammals and Birds: Clock or Magnetic Compass?

Abstract

Species throughout the animal kingdom use the Earth’s magnetic field (MF) to navigate using either or both of two mechanisms. The first relies on magnetite crystals in tissue where their magnetic moments align with the MF to transduce a signal transmitted to the central nervous system. The second and the subject of this paper involves cryptochrome (CRY) proteins located in cone photoreceptors distributed across the retina, studied most extensively in birds. According to the “Radical Pair Mechanism” (RPM), blue/UV light excites CRY’s flavin cofactor (FAD) to generate radical pairs whose singlet-to-triplet interconversion rate is modulated by an external MF. The signaling product of the RPM produces an impression of the field across the retinal surface. In birds, the resulting signal on the optic nerve is transmitted along the thalamofugal pathway to the primary visual cortex, which projects to brain regions concerned with image processing, memory, and executive function. The net result is a bird’s orientation to the MF’s inclination: its vector angle relative to the Earth’s surface. The quality of ambient light (e.g., polarization) provides additional input to the compass. In birds, the Type IV CRY isoform appears pivotal to the compass, given its positioning within retinal cones; a cytosolic location therein indicating no role in the circadian clock; relatively steady diurnal levels (unlike Type II CRY’s cycling); and a full complement of FAD (essential for photosensitivity). The evidence indicates that mammalian Type II CRY isoforms play a light-independent role in the cellular molecular clock without a photoreceptive function.

Introduction

Animals that seasonally migrate and/or forage locally are guided by various factors, visual, auditory, and olfactory, as well as by the geomagnetic field (1–6). We focus here on magnetoreception described in amphibians, reptiles, and fish, but studied best in birds (7, 8), and to a lesser degree in mammals, including humans (see Supplemental Fig. S1; all Supplemental material is available at https://doi.org/10.6084/m9.figshare.13574777.v1).

Two mechanisms underlie magnetoreception. One involves biomineralized magnetite crystals associated with peripheral afferents that transduce signals to the brain where the magnetic field’s (MF) intensity, spatial gradient, and vector heading are processed into a navigable map (9–11). The other is a compass mechanism based on blue-light-sensitive proteins called cryptochromes, located in retinal photoreceptors. This paper describes the scientific developments undergirding our understanding of the light-based compass and asks whether evidence exists to infer its presence in mammals, including humans. [Note: Topics worthy of further research are referenced in bold italics.]

The Light-Based Compass

Cryptochrome (CRY) proteins are ubiquitous across the plant and animal kingdoms (12, 13), with each isoform tailored through evolutionary adaptation to its host organism. Depending on species and tissue, they may function as light-independent core proteins in the circadian clock or as photoreceptors. In the latter case, light excitation converts CRY into a magnetic field sensor that initiates a signaling sequence allowing an animal to orient to the geomagnetic field or to synchronize an organism’s clock to daily light:dark (LD) cycles.

In birds, CRY proteins positioned within retinal cones respond to a magnetic field (MF) in the presence of blue light by generating spin-correlated radical pairs, creating a signaling state transmitted through the retina and along the optic nerve to the brain. The ultimate result is a migrating bird’s flight orientation guided (at least in part) by the MF’s inclination (its vector angle to earth). The initial sequence is called the “radical pair mechanism” (RPM) and is the “prime-mover” of the magnetic inclination compass (14, 15).

The Cryptochrome (CRY) Molecule

General Description

Mammalian CRY genes, first identified in human DNA libraries (hCRY) (16, 17) evolved from a photolyase molecule first identified in a plant, Arabidopsis thaliana (18), with photolyase repairing DNA through a blue light-induced mechanism (19). Animal CRYs (FIGURE 1) are classified according to structure and function (26). Type I CRYs, found in insects such as Drosophila melanogaster (fruit fly) and Danaus plexippus (Monarch butterfly), transmit photoreceptive input to their circadian clock. Type II CRYs, found in vertebrates including birds and mammals, participate in the negative arm of the cellular circadian clock’s transcription-translation feedback loop (12, 20). Some species, such as the Monarch butterfly and zebrafish, have both (27, 28). Type IV CRY has been identified in birds and zebrafish, with its function(s) under investigation, as described in this paper (21, 22, 26, 27, 29–31).

FIGURE 1.

Schematic of the cryptochrome (CRY) molecule CRY in animals is nominally a 60- to 70-kDa molecule with a highly conserved photolyase homology region (PHR) of ∼500 amino acids. The carboxy (COOH) end of the molecule has a “tail” of highly variable length and is referred to here as the COOH-terminal tail (CTT); other commonly used terms include COOH-terminal extension and carboxy-terminus. In each species/cellular context, the CTT has evolved to participate in specific functions (e.g., control CRY shuttling between cytosol and nucleus, sequestering a protein for its degradation, controlling CRY’s degradation). Depending on their functional niche, CTTs vary in length from <30 to >125 amino acids (20). In some animals, certain CRY isoforms in specific tissues (e.g., retina) are photoreceptive to near-UV/blue light due to its PHR’s high binding affinity to its chromophore cofactor, flavin adenine dinucleotide (FAD); animal CRYs with low FAD binding affinity are not believed to be photoreceptive (21, 22). Flavin adenine dinucleotide (FAD) molecule: the isoalloxazine ring is shown in its oxidized (FAD_OX or ground) state. FAD confers photosensitivity to animal cryptochromes with an absorption peak in its oxidized form, considered the ground state in vertebrates, of ∼450 nm (blue-violet) (16, 22–25). After blue light excitation, it experiences a redox photocycle that includes reduced and neutral radical states, ultimately returning to the oxidized state. (This FAD graphic is licensed under the Creative Commons Attribution-Share Alike 3.0 Unported license.)

The Inclination (Magnetic) Compass

Though the biophysical basis for the RPM in birds was first proposed in 2000 (14) and developed since (32–34), its empirical basis can be traced to studies published in the early 1970s. In a landmark study of European robins, Wiltschko and Wiltschko (35) manipulated the ambient MF vectors, flipping the horizontal or vertical components or both. They reported that the birds oriented to the field’s inclination, rather than to its horizontal polarity (FIGURE 2). By the mid-1990s, the light dependency of the inclination compass in birds was well established (36). Further studies identified associated factors affecting orientation, including the ambient light’s wavelength, intensity (37), and polarization (38–40).

FIGURE 2.

Inclination compass versus polarity compass A representation of the experiment by Wiltschko and Wiltschko (35) that demonstrated the inclination compass in the European robin. Flipping the magnetic field’s horizontal (H) and vertical (V) components either separately or together provides a diagnostic tool for discriminating the inclination compass from a polarity compass. Bkg, background.

From Eye to Brain to Behavioral Response

Prime mover.

CRY is the only protein in vertebrates known to produce radical pairs upon photoexcitation (21, 30, 31). The CRY photocycle in birds begins with FAD in its oxidized state (FAD_OX) absorbing light in the UV-blue band, generating radical pairs as either singlets (S, opposite electron spins), which can readily recombine, or triplets (T, same electron spins), which cannot. S-T interconversion continues until they return to their ground state, interact with other molecules, or undergo spin relaxation. S-T interconversion affects the level and ratio of their respective spin states, which represents the signaling state (FIGURE 3 and see Supplemental Fig. S2 for absorption spectra through the FAD photocycle).

FIGURE 3.

The ambient magnetic field modulates light-based radical formation in a bird’s retinal cryptochrome Radical production is initiated when blue-green light strikes the flavin adenine dinucleotide cofactor (FAD) within the cryptochrome molecules, which are situated in the retina’s cone cells. By way of an electron transfer through a triad of tryptophan (TRP) molecules, the photo-excited FAD is reduced, which results in spin-correlated radical pairs. Pairs with opposite spin, called “singlets” (left side of the teeter-totter), can readily recombine returning FAD to its ground state, whereas pairs of like spin, called “triplets” (right), cannot recombine immediately and must go through an intermediate stage before FAD returns to its ground state. The ambient magnetic field modulates the balance of the teeter-totter, thereby affecting the signaling state of the cryptochrome molecule.

What ties photo-induced radical pair generation to the inclination compass is the dependence of S-T interconversion probability on the external MF. With no external MF present, S-T interconversion results from the hyperfine interaction wherein the spin of at least one electron of the pair interacts with the MF moment originating from its atomic nucleus. Some elements, such as carbon and oxygen, are devoid of a magnetic moment, while others, such as hydrogen and nitrogen, have magnetic moments and can form a radical pair with one of the former. With an external MF applied, the triplet radicals split into three energy states (Zeeman splitting), thereby modulating the probability of S-T interconversion (FIGURE 4) (42). Most recently, Ikeya and Woodward (43) bolstered the RPM, demonstrating magnetic field-dependent autofluorescence from natural flavin-based molecules (but not CRY) in HeLa cells under blue light.

FIGURE 4.

Effect of an external magnetic field on radical pair states An external magnetic field causes a triplet radical with a hyperfine interaction to split into 3 states. Singlet-triplet (S-T) interconversions for all 3 triplet states occur so long as the applied field is below the energy level of the hyperfine interaction (see text). As the field increases, the energy gap between the S and T₊₁ states progressively widens, and widens with the T_-1 state on either side of their crossover (open square). Beyond the hyperfine interaction, only S-to-T₀ interconversions take place, owing to the nonmagnetic character of the T₀ state. Graphic modeled on McLauchlan (41). [Note: The 3 triplet states are T₊₁, in which the spins align with the applied field, T_-1, in which the spins oppose the applied field, and T₀, in which the spins are antiparallel but in phase. Because T₀ is nonmagnetic due to its spins canceling each other, its energy remains constant as the applied field is increased.] Inset: free radical yield versus an external static magnetic field. The graph illustrates the yield of a radical signal as a function of the external magnetic field relative to baseline with no field applied. For relatively “low” fields, an increase is seen before yield tails off. The curve peaks when the field equals the hyperfine interaction. (The yield curve is drawn after a figure in an unpublished tutorial by Hore entitled “The Radical Pair Mechanism,” 2003.)

The frequency at which S-T interconversions occur (the Larmor frequency) is proportional to the applied field at roughly 28 kHz per microtesla (µT) (42). The energy associated with the hyperfine interaction can be expressed in terms of an equivalent MF. When the energy associated with the applied field exceeds the hyperfine interaction, S-T interconversion probability diminishes since interconversion is prohibited between the singlet state and two of the three triplet states (41). The result is a low field effect (LFE) wherein the radical yield increases at low fields on the order of several mT, and as the field increases, the yield tails off (FIGURE 4, inset). Maeda et al. (44) demonstrated the LFE empirically in Arabidopsis thaliana CRY and Escherichia coli photolyase. Wideband (∼2 kHz to <10 MHz) radiofrequency exposures superimposed on the background MF can interfere with orientation in birds (45, 46), and such disruption is diagnostic of an inclination compass.

The generation of spin-correlated radical pairs by blue-light-excited CRY molecules is the prime-moving biophysical interaction initiating the signal from retina to brain resulting in magnetic compass-guided orientation. However, this singular interaction, though necessary, is not sufficient to itself result in a behavioral response. Along the retina-to-brain pathway, a chain of factors, each necessary and collectively sufficient, processes the signal that results in a successful behavioral outcome.

Magnetic field effect across the retina.

The angle of the magnetic field vector relative to the axis of a radical pair is one determinant of the pair’s product yield. Thus the CRY molecules deployed within cone photoreceptors require a measure of positional stability, without which each cone would have randomized magnetic field-CRY interactions resulting in a random distribution of reaction product across the retina. The membranous disks within the cones’ outer segments appear attractive as a docking site for orderly deployment of CRY molecules (42, 47–49). For example, Worster et al. (49) recently proposed a model in which CRY molecules are in two adjacent cones (perhaps in both members of a double cone) but are positioned as mirror images. The magnetic field signal detection would operate as a type of differential amplifier, in which the response to incoming light would be bucked, that is, would be the same in both cells, with the ratio of reaction product from the respective cells creating a light-independent signal (49). This is just one possibility, and the exact details of field-CRY interactions in a cellular context remain unresolved.

Unlike the mammalian retina in which cones are centrally located (50), avian cones “tile the retina as highly ordered mosaics (51).” Thus the ambient magnetic field intersects the retinal surface, as illustrated in FIGURE 5, such that gradients of the field’s orientation relative to a radical pair’s axis produces patterns of reaction product across the entire retina. Furthermore, head scanning behavior (which would modulate such patterns) is reportedly “used to locate the reference direction provided by the geomagnetic field (53).”

FIGURE 5.

Intersection of ambient magnetic field with retinal surface The retina is represented as a hemisphere with a sagittal view of the center plane shown. $\overrightarrow{B}$ is the ambient magnetic field vector; Φ is the field’s angle with respect to the hemisphere’s radial vector, $\overrightarrow{r}$ (Note: the magnetic field’s inclination is the angle of $\overrightarrow{B}$ relative to the ground plane, not to $\overrightarrow{r}$); and Θ is the acute angle between the magnetic field vector and the tangential plane to the retinal surface (red lines). Moving clockwise from +90° to −90°, Θ changes, which varies the angle of the field with respect to the cryptochrome molecules in the retina’s outer layer, thus affecting radical yield (14, 52). Similarly, the intersection angle of field to retinal surface tangent varies around each transverse cross section. Every combination of the factors illustrated plus the field’s magnitude, ambient lighting conditions, orientation of the bird’s head, etc. affect radical yield, thereby creating a unique “image” of the magnetic field across the retinal surface.

“Image” formation across the retina.

In a landmark paper, Ritz et al. (14) applied quantum mechanics to model a retina with a simplified set of CRY molecules (just one of the pair sensitive to the field vector’s angle) distributed across its surface. They calculated quantum yields as functions of field magnitude, the angle of the field vector relative to the retinal surface (Θ in FIGURE 5), and the radical pair decay rates. They demonstrated increased rates of radical yields at fields of 0.05 mT (approximately the geomagnetic field in temperate zones) or less across a 10-fold range of decay rates. Importantly, for a bird with a line of sight parallel to the vector of a 0.05-mT field, the map of reaction yields projected onto the center of the retina displayed as a fuzzy dark gray round disk with a lighter gray surround. The disk moved across the retinal surface as the angle between the bird and the field changed and as the bird’s directional heading was rotated through 360°. With this model, magnetic field-induced excitation of flavin-based molecules deployed within cones across the avian retina produces a discernible pattern analogous to how visible objects register their patterns via opsin transduction in the photoreceptors. It is thus plausible that a bird perceives the magnetic field in a visual sense, however undefined. Similar results of retinal simulations have been reported by others (54).

Retina to optic nerve: analog to digital conversion.

Unlike all-or-none action potentials transmitted along most axons in the body, the signals from the retina’s outer photoreceptor- to the inner ganglion-layer are controlled by graded potentials in both the photoreceptors and bipolar cells. These cells have specialized structures anchored at their base referred to as ribbons, appearing at a thickened basilar membrane narrow and rectangular (ribbon-like) in edge view and platelike and horseshoe-shaped in side view. The cells are densely populated with glutamate-rich vesicles near the presynaptic membrane, with the ribbons storing a readily available reservoir of tethered vesicles. Ca²⁺ voltage-gated channels regulate vesicular exocytosis into the synaptic cleft with the glutamate reclaimed through subsequent endocytosis in a continuing cycle, thus maintaining a tonic state of responsiveness to ambient and changing visual conditions (intensity, spectral mix, shape, motion, etc.) (55, 56). The photoreceptors form ribbon synapses with bipolar and horizontal cells. The horizontal cells coordinate signaling among photoreceptors and between the photoreceptors and the bipolar cells. Bipolar cells communicate through ribbon synapses with both ganglion and amacrine cells with the latter providing feedback to the bipolar cells and modulating signals to the retinal ganglion cells (RGC) (55, 56). The RGC form the optic nerve transmitting signals via all-or-none action potentials to visual centers in the brain and beyond. Thus analog signals into the RGC mediated by graded potentials within photoreceptor and bipolar cells (with horizontal and amacrine cell feedback pathways) convert to digital (all-or-none) signals on the optic nerve. This system, also found in the cochlea and conserved across all vertebrates (56–60), permits flexibility within the retina in such matters as adapting to changing lighting conditions, adjusting to contrast in the visual field, and tracking moving objects (61). Likewise, as with the physically visible world, the magnetic field “images” impressed on the photoreceptor layer also benefit from analog signal conditioning within the retina.

Signal processing in brain.

Avian and mammalian visual systems, though homologous to one another (62–65), diverged from their common ancestor (stem amniotes) ∼300 million yr ago (64). In avian species, the major share of projections from RGC follow the tectofugal pathway (66–69) leading first to the optic tectum in the midbrain (equivalent to the mammalian superior colliculus) and connecting sequentially with the nucleus rotundus and the entopallium in the forebrain and on to higher brain centers concerned with visual processing leading to motor responses and memory (70–72). The tectofugal pathway processes motion, spatial resolution, spectral content, and brightness (73–76). The thalamofugal pathway in avian species consists of projections from the retina through the lateral geniculate nucleus to the visual Wulst (German for bulge). Although the Wulst is homologous in structure to the mammalian visual cortex (but without a laminar architecture) (62, 63, 65), it plays a subordinate role as lesions to the Wulst have minor effects on vision (67). Conversely, the thalamofugal is the major mammalian visual pathway passing from the RGC layer through the lateral geniculate nucleus to the visual cortex (67, 77, 78).

Although the thalamofugal pathway is not the dominant pathway in birds for seeing the physical world, it is now recognized as the pathway through which magnetic compass signals pass from the RGC layer to the Wulst (71). The signaling factors between the photoreceptor and RGC layers that differentiate the pathways in the optic nerve initiated by opsin-mediated vision of the physical world from CRY-mediated retinal impressions of the magnetic field are yet to be determined. Nonetheless, connections relevant to migratory activity operating via the thalamofugal pathway were established starting with the initial observation that at night, five regions near or within the visual Wulst were positive for the immediate expression gene ZENK but only in the migratory (garden warblers and European robins) but not in the nonmigratory (zebra finch and canary) species tested and only at night (when most songbirds migrate). The collective region named “cluster N” for “night activation” was suggested as the brain region involved in magnetic compass orientation in nighttime migrants (167). Subsequent tracer and behavioral studies revealed that cluster N was innervated by the thalamofugal pathway (79) and that manipulation of the local magnetic field produced appropriate migratory orientation unless cluster N was lesioned (80). Sectioning the ophthalmic branch of the trigeminal nerve, believed to convey magnetic map signals through a magnetite mechanism (8), was ineffective in altering compass behavior (80). Conceivably, the use of the less dominant thalamofugal pathway (in birds) enables magnetic compass signaling and processing to be segregated from the physical world’s visual signals in the tectofugal pathway. How these bifurcated pathways complement one another with respect to magnetoreception deserves further study.

The loop from sensory inputs of all types to a motor response is routed through a cortical region, the caudolateral nidopallium (NCL) responsible for executive functioning, and is homologous to the mammalian prefrontal cortex (PFC) (63, 71). As summarized by Güntürkün (63), like the PFC, the NCL “receives afferents from secondary and tertiary sensory areas of all modalities and projects back onto them. In addition, the NCL projects to most parts of the somatic and limbic striatum, as well as to motor output structures. Thus identical to PFC, the avian NCL is a convergence zone between the ascending sensory and the descending motor systems” (63).

Of immediate relevance is that in lateral-eyed birds, such as pigeons and passerines (warblers, robins, etc.), the foveal field of view (lateral to the bird) projects in a monocular fashion to the thalamofugal pathway, whereas the dorsotemporal field (frontal) projects in a binocular fashion to the tectofugal pathway. Importantly, their respective projections to the NCL have limited overlap (81), such that processing of the physically visible world is segregated to a degree from (or does not interfere with) processing of the incoming message from the magnetic compass. The anatomic and functional pathways in the avian central nervous system (CNS) leading from sensory input to motor output are described in excellent detail by Mouritsen et al. (71).

Signal-to-noise issue.

In recent years, the study of bioelectromagnetics has been challenged to explain how low-level biological effects from exposure to electromagnetic fields, whether from power lines, home appliances, or cell phones and base stations, are at all possible (82–86). The major challenge has been to rationalize how such exposures can overcome thermal, shot, and 1/f noise within tissue, not to mention ongoing biological background noise from tissue activity. For example, in the brain, in situ signals of at least a few tenths of a volt per meter (V/m) have been necessary to observe effects, and in many cases, much larger fields were required (87–91).

The geomagnetic field (∼0.05 mT) is relatively static (fluctuates very slowly) and thus with a near zero rate of change with time, the electric field induced in tissue is essentially nil in a stationary bird. A larger induced field results from cutting lines of magnetic flux in flight, but the induced electric field is insufficient to produce an effect in brain tissue [<1 mV/m at a velocity of ∼18 m/s (∼40 miles/h)]. The question then is how such a minute amount of energy triggers signaling from retinal cones to the CNS that can guide a bird’s positional orientation. The answer derives from the fact that for transduction of signal from CRY, noise is not the obstacle. As McLauchlan and Stiener (92) stated, “Here it might be stressed that the interactions which cause singlet-triplet interconversion are weak ones in the sense that their energies are far below the mean thermal energy at room temperature in size; they are effective entirely through their influence on the kinetics, not the thermodynamics, of systems.” CRY excitation by light and its downstream effects on flight orientation exemplifies amplification in biological systems leading from an infinitesimal input to a behavioral output.

Resolving CRY’s Role in Mammals

The first reports of human CRY (16, 17) stimulated research into CRY as the long-sought mammalian circadian photoreceptor predicated on two factors: first, studies of mice with retinal degeneration (rd mice) (93–96) as well as studies of human subjects with various degrees of blindness (97–99) gave strong indications of a “nonimage forming” (NIF) branch of the visual system complementing the classical image-forming branch (100–103). Second, studies in Drosophila confirmed its CRY isoform (dCRY) as a light-sensitive molecule feeding photic input to its core circadian oscillator (104–108), suggesting a like photosensitivity in mammalian CRY.

The hypothesis of CRY as a circadian photoreceptor was furthered by subsequent studies in which exposure of Cry-mutant mice to light led to disrupted phase shifts in both locomotor activity (109–112) and per rhythms in the suprachiasmatic nucleus; per codes for a key factor in a cell’s molecular clock (110, 112).

However, two developments derailed the hypothesis. First, further study in mice favored its role as a light-independent nuclear protein functioning in the circadian clock feedback loop (113, 114), with Okamura et al. (114) concluding that “mCRY1 and mCRY2 are dispensable for light-induced phase shifting of the biological clock,” and “…are indispensable components of the core oscillator” (underline added for emphasis).

Second, conclusive evidence accumulated that melanopsin within intrinsically photosensitive retinal ganglion cells was the photoreceptor in the NIF visual pathway leading to LD-driven circadian entrainment (115–117). Further information concerning melanopsin is available in various reviews (118–123).

We now know that the core oscillators in Drosophila and mammals are negative feedback loops in which heterodimerized proteins, CRY-PER in mammals and TIM-PER in Drosophila (Supplemental Fig. S3) translocate from the cytoplasm to the nucleus where they inhibit their own transcription factors, CLOCK-BMAL1 in mammals and CLOCK-CYC in Drosophila. In addition, each has additional positive and negative feedback loops and other controls (phosphorylation, acetylation, and ubiquitylation, which tag molecules for degradation, etc.) to provide fine tuning to the daily cycle. Thus, whereas dCRY serves primarily as a photoreceptor in Drosophila to drive its core oscillator, mammalian CRY is an integral part of the core oscillator but not itself responsive to light. This difference largely derives from dCRY having stoichiometric amounts of FAD, a necessary constituent for its light sensitivity (25). By contrast, mammalian CRYs have substoichiometric quantities of FAD, which obviates any significant light-induced signal that could be delivered to a clock (124, 125). Excellent reviews of molecular clocks are available (126–129).

Cryptochrome’s Potential Role in the Avian Magnetic Compass

In a “proof-of-principle” experiment, Maeda et al. (130) synthesized a molecule composed of carotenoid, porphyrin, and carotenoid groups simulating radical-forming electron transfer in a photosynthetic reaction center. They recorded responses to a MF that were consistent “by analogy” with a flavin-based CRY response to a MF, which “seems to be the first observation of a chemical effect of a magnetic field as weak as ∼50 µT.” They then isolated and purified E. coli photolyase (repairs UV-damaged DNA) and A. thaliana CRY (circadian and photoperiodic functions) (44). Both satisfied RPM requirements exhibiting the requisite spectra, kinetics, and radical yield versus magnetic field curves, including the LFE at ∼1–3 mT, with the investigators hinting at possible magnetosensitivity in humans.

In parallel, several studies zeroed in on avian CRY1a as appearing to possess the characteristics required for an avian magnetic compass. In retinal preparations of European robins and domestic chickens, CRY1a populated the outer segment of UV cones and blue light activated antisera binding to the CTT (48, 131). Another pair of studies (one in a chicken retinal preparation and the other in live robins) suggested the nonradical reduced FAD photoproduct as the signaling state (168, 169). However, this was inconsistent with other reports of the reduced or neutral FAD radical serving as the intracellular signal (44, 54, 125, 130, 132). Finally, a study of retinal preparations reported 15 of 90 mammalian species (48 families; 16 orders) positive for CRY1a in UV cones (133). Nevertheless, an RPM-based magnetic compass in mammals remains speculative.

To establish a molecular basis for the divergent functions of Type I (photosensitive) and Type II CRY (not photosensitive), Kutta et al. (22) conducted spectroscopic and crystallographic analyses combined with modeling simulations to characterize their respective FAD content and FAD/CRY dissociation constants. UV absorbance spectra of purified recombinant CRYs revealed no FAD in either hCRY1 or hCRY2 but a full complement in dCRY. The absence of FAD in hCRY spectra was supported by low binding affinities for FAD measured for each hCRY isoform. Simulations of the FAD/binding pocket energetics revealed that FAD’s conformation in mCRY (mouse CRY used as a surrogate for hCRY) resulted in a greater distance to its binding pocket, compared with dCRY, correlating with the former’s lower FAD-binding affinity. Analyses of comparative amino acid sequences in mCRY and dCRY involved with FAD binding, together with specific mutations in selected residues, provided support for these differences. Importantly, poor CRY-FAD affinity also extended to avian Type II CRYs, prompting these investigators to call Type II CRYs “vestigial flavoproteins,” doubting CRY1a in avian cones as a photo-transducer required for a magnetic compass. Rather, Kutta et al. proposed that CRY1a’s “action is likely restricted to signal transduction in the dark via protein interaction cascades.”

Type IV CRY: The Elusive Compass Magnetoreceptor?

CRY1a as the putative avian compass’ magnetoreceptor did not reconcile with Type II CRYs containing substoichiometric FAD (26, 124, 125, 134). However, in parallel, evidence accumulated supporting Type IV CRY (CRY4) as suited to a photo/magnetoreceptor compass.

In the first study of CRY4 in any species, zebrafish zCRY4 was one of seven CRY/photolyase isoforms (27), including a 6-4 photolyase which as expected, repaired DNA damage from UV exposure (in an E. coli system); zCRY4 was ineffective. In an NIH3T3 cellular assay, zCRY4 displayed circadian rhythmicity, but unlike the Type II zCRYs, did not repress CLOCK:BMAL1, evidence that zCRY4 is not likely to participate in the negative circadian feedback loop. However, zCRY4’s sequence similarity to dCRY in one of two phylogenetic analyses suggested zCRY4 as “a likely candidate for the zebrafish circadian photoreceptor” (27).

A study of brain tissue from house sparrows tracked the 24-h expression of cloned clock genes (Clock, Bmal1, Cry1&2, and Per2&3) and Cry4 (29). The nucleotide and protein identity sequences (nis and pis) of the clock genes were high, compared with their human homologs. However, Cry4‘s highest nis (79%) and pis (63%) were with zebrafish zCry4. Although all clock genes exhibited clear circadian expression, sparrow Cry4 remained essentially flat across the entire day.

Ozturk et al. (124) examined spectroscopic properties of zCRY4 and ggCRY4 (chickens) isolated from an insect cell viral expression system. Both displayed similar absorption spectra but different compared with dCRY. Whereas dCRY exhibited light-induced proteolysis in three different cell lines, zCRY4 remained stable in all three (ggCRY4 was not further addressed). In a comparative summary, Ozturk et al. indicated that zCRY4 did not repress CLOCK:BMAL1 transcription, again reflecting a nonclock function, whereas Type II CRY from the monarch butterfly functioned in the circadian negative feedback loop. They concluded, “…type 4 CRYs are unique among all of the animal CRYs…they neither repress Clock:BMal1 nor undergo photoinduced proteolysis. The full complement of FAD…and the unique expression patterns of these CRYs in zebrafish and chicken tissues that are photosensitive are strong circumstantial evidence for a circadian photoreceptor function of these cryptochromes.” They did not suggest nor rule out a magnetoreceptive role for CRY4.

Japanese investigators reported ggCRY4 in the retina of young chicks with no clear evidence of a diurnal cycle (135, 136). In human embryonic kidney cells, ggCRY4, unlike ggCRY1/2, did not repress CLOCK:BMAL1 and was associated with the cytosolic fraction in lysates (clock CRYs are nuclear-based) (135). Also, in chicks and a yeast expression system, a monoclonal antibody targeting a segment of the CTT (C1-Ab) immunoprecipitated ggCRY4 more strongly in dark than in light (136, 137). These observations led to a proposed model wherein upon blue illumination, the end of the CTT folds in to the ggCRY4 molecule near or at the FAD binding pocket shielding the tail from binding to C1-Ab; in darkness, the CTT detaches from this position exposing itself to C1-Ab, resulting in greater immunoprecipitation compared with the illuminated condition (137). As in Ozturk et al. (124), FAD was recovered at stoichiometric levels from ggCRY4 (137). These findings are in line with ggCRY4 as potentially magnetosensitive.

Ozturk et al. reported that dCRY exposed to blue light experienced enhanced trypsinization, compared with dark conditions. When deprived of its CTT, trypsinization was equally effective in fragmenting the molecule in dark as in light, reflecting the CTT’s “protective” role in Drosophila (138). Thus ggCRY4 and dCRY, with similar physical features and stoichiometric FAD, undergo opposing light-induced conformational changes with dCRY opening its CTT and ggCRY4 folding its CTT into the CRY molecule. These observations reflected the molecules’ evolution along distinct phylogenetic lineages (13, 26) adapted to their respective niches (139).

Günther et al. (21) studied the temporal expression of retinal Cry4 and clock genes in European robins (er) and chickens. erCry4 mRNA (during autumnal migration season) showed little diurnal variation as compared with the clock genes erCry1a and erCry1b, which exhibited pronounced rhythms and erCry2 with only modest cycling. erCry4 expression more than doubled in autumn relative to nonmigratory seasons. Neither erCry1a nor erCry1b varied seasonally, and erCry2 autumnal expression was increased by ∼50% relative to nonmigratory seasons (perhaps a false positive). For chickens (nonmigratory), ggCry4 along with ggCry1a, ggCry1b, and ggCry2 were all generally unaffected by season at postnatal days 1 and 41.

Poly- and monoclonal antibodies specific to CRY4 localized erCRY4 and ggCRY4 to the outer segments of double cones and long wavelength cones, that is, in cytosolic compartments likely associated with the internal membrane disks. Günther et al. suspect previous studies reporting CRY4 in multiple retinal layers was due to insufficient antibody specificity. Finally, their structural model of the erCRY4 molecule was consistent with its high FAD-binding affinity, and, combined with its retinal location and lack of a role in the circadian clock, they concluded that CRY4 is plausibly magnetoreceptive (21).

Wu et al. (140) used a yeast-two-hybrid assay to screen for proteins with which photoexcited erCRY4 might interact. Possibilities included membrane-associated long-wavelength sensitive opsin (iodopsin) and a potassium voltage-gated channel protein, as well as retinal binding protein and retinal G protein-coupled receptor, which participate in the trans/cis retinal opsin cycle (140). Further study is needed to elucidate the intracellular targets of photoexcited CRY4.

Like chickens, zebra finches are nonmigratory and were thought initially to lack a magnetic compass (141), but subsequent investigations reported a light-based magnetic sense for short distances (142–144). Pinzon-Rodriguez et al. (31) studied Cry1, Cry2, and Cry4 in zebra finches (zfCry), reporting the expression of all three in the retina, brain, and muscle. zfCry4 expression did not vary in retina, brain, or muscle over the three time points sampled, whereas zfCry1 and zfCry2 showed evidence of cycling (31). They concluded that these findings together with previous evidence point to CRY4 as the prime candidate for the magnetoreceptor involved in compass-based orientation.

In a study of pigeon CRY (clCRY), Wang et al. (145) procured gene sequence data for a homolog of clCRY2 and two homologs of clCRY1. One of the two clCRY1 homologs aligned more compatibly with other vertebrate CRY4 sequences and was reclassified. The investigators suggest that other CRY4 sequences may also have been misclassified as CRY1 (but did not specify which). The three CRYs were distributed across all 11 pigeon tissue sites tested. When expressed in NIH3T3 cells transfected with Cry-bearing plasmids, clCRY1 was seen primarily within cell nuclei, consistent with a circadian clock function, while clCRY4 was cytosolic, ruling out a clock function. Expression of clCry1 and clCry2 over a 24-h period was rhythmic in various tissues, as expected for clock genes. In the retina, clCry1 expression was rhythmic, while clCry4 did not cycle significantly. Although the investigators hint at a magnetoreceptive capability, they ascribe clCRY4’s function as unknown, also pointing out that CRY4 has not been reported in mammals.

Wang et al. purified clCRY1 and clCRY4 from an insect cell viral expression system, with clCRY1 FAD-deficient, displaying a vacant absorption spectrum (145). In contrast, clCRY4 with a stoichiometric amount of FAD displayed a spectrum typical of the oxidized, photosensitive state. Further analyses revealed a light-intensity dependent photocycle with the FAD cycle including an anionic radical and a neutral radical and a fully reduced form before returning to its oxidized form in the dark. A mutation in the tryptophan (Trp) triad essentially eliminated reduction of oxidized FAD, blocking the photocycle. Finally, using combinations of light or dark, with and without trypsin treatment, Wang et al. demonstrated light-dependent conformational changes in clCRY4.

Zoltowski et al. (146) used crystallography and spectral techniques to assess the molecular basis for the exquisite sensitivity of clCRY4 to very low light levels, assuming their explanation applied also to night-migratory songbird species, even though pigeons themselves do not migrate at night. “Evolutionary conservation” of two specific residues enhanced transfer of electrons in the Trp chain and promoted an increased FAD quantum yield. However, they note that further research is necessary to resolve conflicts between molecular mechanisms and behavioral observations in birds.

Conclusions

The timeline of the interwoven strands of research contributing to our understanding of CRY as a clock protein and a magnetosensitive molecule is shown in Supplemental Fig. S4. The research that ensued after Ritz et al. (14) published their quantum model (in 2000) of how the MF can imprint its image onto the avian retina by way of the RPM, led to studies addressing the hypothesis of Type II CRY in cones as the magnetoreceptive agent. Type IV had not been described until the same year in zebrafish and not until 2006 in an avian species (29). Over time, Type IV displayed comparatively better characteristics suited to an avian magnetoreceptor, as shown in Table 1. The lack of evidence of its presence in humans and other mammals (26, 145), however, would argue against a mammalian inclination compass. Further the lack of cones in the mammalian retina’s periphery, as opposed to their distribution across the retina’s entire surface in avian species, prevents the magnetic field’s full-field impression on the retina as required for a magnetic inclination compass. These developments, together with research in mammals revealing melanopsin’s NIF role as distinct from CRY’s role as a light-independent constituent of the molecular clock, demonstrate the self-correcting nature of scientific inquiry.

Table 1.

Type II CRY compared with Type IV CRY

Attribute	Type II	Type IV
Absorption spectrum	None	Flavin
Stoichiometric FAD	No	Yes
FAD affinity	Low	High
Light-induced conformation change	Yes	Yes
In cones	Yes	Yes
Cellular location	Nuclear	Cytosolic
Repress CLOCK:BMAL1	Yes	No
Circadian rhythm in retina	Yes	No*
Seasonal variation	No#	Yes
Reported in mammals	Yes	Not Yet
Viable magnetoreceptor candidate?	No	Yes

FAD, flavin adenine dinucleotide cofactor. *Cycles in zebrafish retina. #Modest seasonal increase of erCry2 expression but not erCRY1a&b.

Despite the mechanistic differences between the light-based compass and the magnetite-based map, challenges remain to explain how they are integrated within the CNS (147), even though their respective neural pathways have been described (71). Perhaps nowhere has compass/map coordination been documented more than in a set of studies in which birds are physically or virtually displaced along an east-west route (148–152). In general, juveniles participating in their first fall migration rely on the light-based compass as an inbred response, heading south on the trajectory they would take from breeding grounds (where they were hatched) but with their route displaced. Experienced birds, however, having already mapped their wintering grounds in the previous season adjust their flight pattern to compensate for displacement.

Several studies have suggested the existence of a mammalian magnetite-based map, including in poorly sighted species such as bats (153–155) and subterranean ground moles (156–160). Others reported magnetic orientation capabilities in terrestrial mammals (161–163). In humans, two laboratory investigations studied electroencephalograms in volunteers exposed to magnetic fields manipulated in various ways. The first reported no clear evidence of response (164), and the second reported alpha-band changes from field manipulations diagnostic of only magnetite- but not CRY-based detection (165). A third study suggested a human inclination compass, reporting that under blue light, starved men (but not women) oriented toward food when associated with the magnetic field (166). The evidence in this paper would argue otherwise. Only further research will resolve whether humans possess a sixth sense.