Cryptochromes: Photochemical and structural insight into magnetoreception

Abstract

Cryptochromes (CRYs) function as blue light photoreceptors in diverse physiological processes in nearly all kingdoms of life. Over the past several decades, they have emerged as the most likely candidates for light‐dependent magnetoreception in animals, however, a long history of conflicts between in vitro photochemistry and in vivo behavioral data complicate validation of CRYs as a magnetosensor. In this review, we highlight the origins of conflicts regarding CRY photochemistry and signal transduction, and identify recent data that provides clarity on potential mechanisms of signal transduction in magnetoreception. The review primarily focuses on examining differences in photochemistry and signal transduction in plant and animal CRYs, and identifies potential modes of convergent evolution within these independent lineages that may identify conserved signaling pathways.

1. INTRODUCTION

Diverse species have evolved mechanisms to sense and adapt to seasonal changes in temperature, food availability, and day length. For many animals, a central aspect of this adaptation is seasonal migration. Despite keen insight into seasonal behavior patterns, a molecular mechanism gating migration and navigation of the Earth's magnetic field remains elusive. Two models have been proposed for candidate magnetosensors, these include ferromagnetic crystals of magnetite, ¹ , ² , ³ and a radical‐pair based model originally theorized by Klaus Schulten. ⁴ , ⁵ , ⁶ The central tenet of the radical‐pair hypothesis was that a photochemical reaction can exist, whereby generation of a caged radical pair can be sensitive to the Earth's magnetic field. Specifically, provided the proper constraints on the radical pair, ⁴ , ⁵ , ⁶ a magnetic field can alter the relative populations of singlet and triplet spin states to bias the population of signaling components. At that time, no protein was known that contained such a mechanism.

Twenty five years later the first Cryptochrome (CRY) protein was discovered, ⁷ and later determined to employ a radical pair based mechanism consistent with Schulten's model. ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ CRY proteins are part of the CRY/Photolyase (PHOT) family (CPF) and have evolved distinctly in both plants and animals, where they can be classified according to sequence homology. ¹⁵ In plants, two CRY isoforms (CRY1 and CRY2) regulate diverse blue‐light responses including, hypocotyl elongation, ⁷ programed cell death, ¹⁶ photoperiodic control of flowering, ¹⁷ stomatal opening, ¹⁸ , ¹⁹ circadian timing, ²⁰ , ²¹ among many others. ²² Animal CRYs can be divided into three lineages termed Type I (insects), Type II (Mammals), and Type IV CRYs (birds, fish, turtles). Type I and II are regulators of the circadian clock in light‐dependent and light‐independent mechanisms respectively, ¹⁰ , ¹¹ , ¹⁵ and Type IV proteins are largely of unknown function, but demonstrate expression patterns and photochemistry ideal for magnetoreception. ¹¹ , ¹⁴ , ²³ To date, all phylogenetic lineages of CRYs have demonstrated evidence of magnetic field effects on organism physiology, here we briefly summarize these findings.

Despite not being in a migratory organism, plant CRY proteins were the first to demonstrate a magnetic field effect on photoproduct yield, establishing these proteins as putative models for magnetic field effects on signal transduction. ²⁴ Magnetic field effects on Arabidopsis thaliana CRY1 (AtCRY1) photochemistry was supported by functional assays, where in a series of elegant experiments conducted by Ahmad and coworkers, they demonstrated an effect of external magnetic fields on both AtCRY1 phosphorylation and downstream effects on hypocotyl elongation in vivo. ²⁵ , ²⁶ Notably, magnetic field effects were dependent upon the field being applied during the dark‐phase of light–dark cycles. ²⁶ These results were the first to unify direct examination of magnetic field effects on fundamental chemistry, structural transitions, and downstream signal transduction in a single CRY protein. However, these results remain controversial in part due to difficulty in replicating the experiments in some labs. ²⁷

In animals, magnetic field effects have been observed in Type I, Type II, and Type IV CRYs. Although not a migratory model, the ease of genetic manipulation in fruit flies has provided great insight into modulation of CRY‐based signaling with external magnetic fields. Photochemical experiments of a Type I CRY from drosophila (dCRY) have verified magnetic field effects on photochemical yield. ²⁸ Further, CRY‐dependent magnetic field effects on circadian timing, ²⁹ neuronal firing rates, ³⁰ locomotor activity, ²⁹ and navigation of a T‐maze ³¹ have been demonstrated in fruit flies. These results are synergistic with genetic studies in the migratory monarch butterfly, where both Type I and Type II CRYs have been argued to be magnetosensors. ³² , ³³ Recent genetic studies in the monarch butterfly indicate that the photoactive Type I CRY, and not the mammalian‐like Type II CRY, is required for light‐dependent magnetic field effects on behavior. ³³ The combined results present a powerful argument indicating that Type I CRYs retain elements capable of generating a magnetic‐field dependent response, however, Type I CRYs are only present in some insects, and most migratory species contain only Type II and Type IV CRYs. Thus, in most animals Type II and/or Type IV CRYs must retain analogous properties.

Magnetic field effects mediated by animal Type II and Type IV CRYs remain controversial. Both Type II and IV CRYs are prevalent in migratory organisms where the action spectra of light and magnetic‐field dependent reorientation is consistent with CRY chemistry, ³⁴ , ³⁵ , ³⁶ and CRY proteins are prevalent in the retinas of migratory birds which would be necessary to initiate a light sensitive magnetic response. ²³ , ³⁷ , ³⁸ Further, Type II CRYs rescue fruit‐fly navigation in T‐maze assays, ³⁹ and regulate magnetic‐field and light‐dependent reorientation in cockroaches. ⁴⁰ Currently, downstream signaling of CRY4 proteins is largely unknown, however the photochemical properties of CRY4 proteins are ideally suited for a light‐dependent magnetosensor. ¹⁴ Validation of Type II and Type IV CRYs as magnetosensors is currently hampered by two experimental constraints. First, Type II CRYs do not bind the photoactive cofactor FAD in vitro, ⁴¹ , ⁴² , ⁴³ and are not known to take part in light‐dependent signaling. ⁴⁴ , ⁴⁵ Second, downstream signaling factors for CRY4 proteins are poorly understood, and limited to putative protein–protein interaction partners identified in proteomics assays. ⁴⁶ , ⁴⁷

Despite significant evidence of magnetic‐field dependent responses in CRY proteins, validation of a CRY‐based magnetosensor remains elusive due to extensive conflicts between in vitro photochemical experiments, and in vivo behavioral assays, as well as difficulties in performing genetic studies in migratory organisms. Further, the phylogenetic lineages are differentiated by their photochemical properties, protein–protein interaction targets, and mechanisms of signal transduction (Table 1). These conflicts are exacerbated by a history of debate regarding the mechanisms of signal transduction in CRY proteins, specifically in regards to the ground and signaling state of the FAD cofactor, and how these elements contribute to the biological response. Importantly, how this chemistry gates downstream signaling in a magnetosensor model has not been widely explored. The goal of this review is to present the existing data in the context of known CRY photochemistry and signal transduction, and to provide insight based on recent studies that may guide synergistic modes of magnetoreception.

TABLE 1.

Representative CRY proteins and signaling

	Representative proteins	N5 interacting residue	FAD ground state (in vitro/vivo)	FAD signaling state	Accessible RPs
Type I CRYs	dCRY DpCRY1	Cys416 Cys402	FADox/FADox	FAD•−	M1
Type II CRYs	mCRY1 mCRY2	Asn393 Asn411	a	a	a
Type IV CRYS	ClCRY4 ZfCRY4	Asn391 Asn390	FADox/unknown FADox/unknown	Unknown Unknown	M1, M2 M1, M2
Plant CRY1	AtCRY1	Asp396	FADox/FADox	FADH•	M1
Plant CRY2	AtCRY2	Asp393	FADox/FADox	FADH•	M1
6–4 photolyase	Dm64	Asn403	FADox/FADH−	NR	NR
CPD photolyase	EcCPD	Asn379	FADH•/FADH−	NR	NR

^a Type II CRYs do not bind FAD in vitro. NR, not relevant.

2. CRY PHOTOCHEMISTRY AND THE RADICAL PAIR MODEL

Decades of research into the photochemical mechanism of CPF proteins indicate that CRYs are ideally positioned to function as a light‐driven magnetosensor via the radical pair hypothesis. CRY photochemistry is defined by electron transfer between a highly conserved chain of Trp residues termed the Trp‐Triad, and an FAD cofactor within the active site of most CRYs and all PHOTs (Figure 1). In vitro characterization of CPF proteins, confirms the Trp‐Triad facilitates interconversion between four FAD redox states: oxidized FAD (FAD^ox), a one‐electron reduced anionic semiquinone (FAD^•−), a one‐electron reduced neutral semiquinone (FADH^•), and a two‐electron reduced hydroquinone (FADH⁻), where the photochemical action spectra is defined by the absorption spectra of the respective FAD‐redox states (Figure 1(d)).

FIGURE 1.

CRY Photochemistry (a) UV/Blue light photoexcitation of FAD^ox (Spectra: black curve in d) abstracts an electron from a Trp (b: Trp_1 in c) in the Trp‐triad (b: Trp_1–3 in c) to generate the singlet radical pair FAD^•− (Spectra: red curve in d) and TrpH^•+. The Trp radical is propagates through Trp‐triad and is quenched by solvent. In Type IV and plant CRYs FAD^•− is protonated to FADH^• (Spectra: blue curve in d) to stabilize the radical species. Singlet‐triplet interconversion in FAD^•−/FADH^•− and TrpH^•+ may be sensitive to external magnetic fields (M1). Type IV CRYs undergo further photoreduction through the Trp‐triad to form FADH⁻ (Spectra: teal curve in d). FADH⁻ is oxidized by O₂ generating an alternative radical pair (M2). (b) The photocycle of Type I CRYs only accesses the FAD^•− state, in plant CRYs Asp396 (AtCRY1 numbering) transfers a proton to FAD^•− allowing access to FADH^• without further photoreduction. (c) The Trp‐triad is extended by additional Trp or Tyr residues in some CRYs. In ClCRY4 this includes a fourth Trp (Trp_4) and a Tyr (Tyr_5). (d) Spectral profiles of FAD bound to CRYs (inset: redox cycle of FAD)

Examination of the in vitro CRY photocycles has identified two radical pairs, termed M1 and M2, that meet some or all of the criteria of the RP‐hypothesis (Figure 1(a)). The M1 and M2 mechanisms are differentiated by three factors: (1) The nature of the radical pair, (2) The ground state of the FAD chromophore, and (3) The role of light in generation of the radical pair. The M1 mechanism is produced through the forward photochemical cycle and requires an FAD^ox ground state, where photoreduction of FAD generates a [FAD^•− TrpH^•+] radical pair. The M1 mechanism is supported by quantum mechanical calculations that confirmed M1 demonstrates characteristics consistent with the RP‐hypothesis. ⁴ , ⁴⁸ Further, in vitro ultrafast spectroscopy confirmed that millitesla magnetic fields alter the photochemical yield of FAD^•−. ²⁵ , ²⁸ The M2 mechanism is dependent on flavin reoxidation and requires formation of FADH⁻ through one of two methods: (1) Light‐dependent photoreduction of FADH^•, or (2) Light independent reduction through cellular processes. Subsequent reactions between FADH⁻ and oxygen then generate a semiquinone‐superoxide [FADH^• O₂ ^•−] radical pair (M2). The M2 mechanism is supported by in vivo assays that indicate organisms reorient under red‐light conditions, ⁴⁹ , ⁵⁰ and in vivo studies in birds and plants indicating that magnetic‐field dependent responses can occur when the magnetic field is only present in the dark. ²⁶ , ⁵¹ Notably, quantum mechanical calculations suggest that the spin dynamics of the superoxide radical are incompatible with the RP‐hypothesis ⁵² and in vivo data exists that contradicts a dark‐dependent radical pair. ⁵³ , ⁵⁴

The RP‐model was complicated by initial studies of CRY proteins that questioned a Trp‐triad based model, where mutations that abrogated photochemistry in vitro had no effect on in vivo function. ¹³ , ⁵⁵ , ⁵⁶ , ⁵⁷ These studies instead posited that CRYs signal via a photolyase‐like mechanism, where the ground state of the FAD cofactor is FADH⁻ and light induces electron transfer from FADH⁻ to either redox active amino acids, or an unknown electron acceptor (Z) to generate a [FADH^• − Z^•] radical pair, which initiates signaling. ⁵⁸ , ⁵⁹ Although, such a mechanism remains a possible alternative in the context of magnetosensing, recent discoveries have mostly reconciled conflicts between in vitro and in vivo data regarding the Trp‐Triad. Specifically, in several systems the Trp‐triad chain can be extended to include a fourth, or even fifth redox active amino acid that impacts photochemical yield (Figure 1(c)). ⁶⁰ In several cases Tyr residues substitute for Trp species, enabling a long‐lived terminal radical, ¹³ , ⁶¹ , ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ⁶⁶ , ⁶⁷ or enhancement of photochemical yield. ¹⁴ In addition, alternatives to the canonical Trp‐Triad pathway have been identified that can likely substitute in vivo. ⁶⁸ Further, in plants, ATP or other nucleotides can bind to a cleft near the FAD binding pocket to enhance photochemical yield, even in the presence of Trp‐triad variants. ⁵⁷ , ⁶⁹ CRY photochemistry can also be modified by protein–protein interaction partners, where in plants BIC2 alters the CRY2 structure to block Trp‐triad mediated photoreduction. ⁷⁰ These observations mirror studies in some PHOTs, where binding DNA or other elements near the DNA binding cleft can tune FAD reduction potentials to alter photochemical outcomes. ⁷¹ These recent studies indicate that in a cellular context, CRY photochemistry can be modified or tuned, and alternative photochemistry or radical‐pairs may exist, and need to be considered in identifying a magnetoreception mechanism. Despite these alternative reaction pathways, general consensus CRY photocycles have been identified in CRY lineages.

Characterization of AtCRY1 and AtCRY2 in vitro and in vivo indicate that signal transduction stems from a photocycle consisting of photoreduction to FAD^•− followed by proton transfer from a conserved Aspartate (AtCRY1: D396; AtCRY2: D393) to stabilize a FADH^• signaling state ⁶⁴ , ⁷² , ⁷³ , ⁷⁴ (Figure 1(b); Table 1). The consensus photocycle stems from a combination of solution biophysics, and in vivo correlations between CRY action spectra and known photochemistry. First, action spectra for CRY function mirror the FAD^ox spectra indicating an FAD^ox ground state and FAD^•− or FADH^• signaling states. ⁷⁴ , ⁷⁵ Second, UV–vis and IR studies verify that light‐induced conformational changes persist in protein variants only capable of generating FAD^•−, but in WT proteins subsequent proton transfer to form FADH^• stabilizes the signaling state in vitro and in vivo. ⁷² , ⁷³ , ⁷⁶ , ⁷⁷ Third, SEC experiments confirm that photoreduction to FADH^• is sufficient for CRY‐CIB protein complex formation and CRY oligomerization. ⁷⁰ Both protein complexes readily form in optogenetic tools upon blue‐light exposure, where subsequent illumination with green light to attempt to populate the FADH⁻ state results in no change in function. ⁷⁸ , ⁷⁹ These latter results are consistent with in vitro and in vivo characterization of the quantum yields of photoreduction to generate FADH^• and FADH⁻ states, where quantum yield calculations indicate that AtCRY1 does not form FADH⁻, and CRY2 does so with very low yield. ⁸⁰ The combined results indicate that AtCRY photocycles are arrested at the FADH^• state, without further access to FADH⁻. As a result, in vitro and in vivo experiments indicate that M1 is likely the only accessible radical pair. These observations conflict with in vivo studies, which indicate that a magnetic field is only required during the dark‐phase of the photocycle, which would require formation of FADH⁻ (M2, Figure 1(a)). ²⁵

In animal CRYs elucidating photochemical mechanisms is even more complex. Type I, II, and IV CRYs can be differentiated by their role in light‐dependent (Type I and Type IV) or light independent pathways (Type II). ⁴⁵ Currently, Type I CRYs are the best characterized in animals and demonstrate a photocycle unique among CPF members. Specifically, Type I CRYs exist in an FAD^ox ground state, but do not generate FADH^•, rather blue‐light illumination leads to FAD^•‐ without further protonation or photochemical reduction (Table 1). ¹² , ⁸¹ Analysis of redox potentials of Type I CRYs confirm they should exist in the FAD^ox state under cellular conditions, and limited proteolysis studies confirm that FAD^•− formation is both necessary and sufficient for downstream signaling. ⁸² , ⁸³ , ⁸⁴ Thus, Type I CRYs should function through an M1 mechanism in magnetoreception.

Type II and IV CRYs diverge in terms of photochemical signaling. Most notably, Type II CRYs do not purify with bound FAD, and are not known to partake in light‐dependent functions (Table 1). Arguments for a role in magnetoreception largely are dependent upon immunohistochemistry data suggesting light‐induced conformational changes in retina, ⁸⁵ , ⁸⁶ however these results have been contradicted by other groups. ⁸⁷ Further, Type II CRYs have been reconstituted with FAD in vitro, ⁴³ , ⁶² however they require biologically irrelevant concentrations of FAD, and studies have confirmed the binding affinity of Type II CRYs is insufficient to lead to functional FAD binding in vivo. ⁴¹

In contrast, Type IV CRYs readily bind FAD and undergo photochemistry. For, Type IV CRYs, the ground state in vivo is unknown, however purified proteins exist in the FAD^ox state and generate FADH^• upon blue‐light exposure with a characteristic red‐shift in the long wavelength peaks. ¹¹ , ¹⁴ The red‐shift positions Type IV CRYs to be sensitive to red‐light in a manner consistent with behavioral studies in birds. ⁴⁹ , ⁵⁰ Further exposure to blue‐ or red‐light leads to efficient formation of the fully reduced FADH⁻ state that is slowly reoxidized in the presence of oxygen. ¹¹ , ¹⁴ In this manner, Type IV CRYs efficiently generate the FADH⁻ state, and thus appear to be the only CRYs capable of employing both an M1 and M2 mechanism (Table 1).

Contradictions between in vivo behavioral assays and in vitro photochemistry for M1 and M2 mechanisms has led to controversy within the magnetosensor community. These conflicts are exacerbated by a history of debate regarding how in vitro photochemistry dictates signal transduction in CRY proteins. The primary questions revolve around what structural determinants differentiate light‐dependent CRYs (Plant, Type I, and Type IV) from light‐independent CRYs, and what factors select for specific modes of photochemistry in vitro and in vivo. Recent progress in CRY structure and function has shed light on both factors and identify aspects of convergent evolution leading to common modes of signal transduction across the independent lineages.

3. CRY STRUCTURE AND FUNCTION

CPF members can be divided into two functional regions termed the Photolyase Homology Region (PHR) and in CRYs, a C‐terminal tail (CTT) (Figure 2(a)). PHRs are well studied, and are structurally conserved in all CRYs, where they retain two pockets termed the primary and secondary pockets. The PHR is composed of an N‐terminal αβ‐domain that is fused to an α‐helical C‐terminal domain through a long flexible linker. Both core domains contain cofactor binding sites. The primary cofactor pocket is found within the α‐helical domain, which recognizes FAD in all photoactive members of the CPF family, but is empty in Type II animal CRYs. ⁴³ , ⁴⁵ , ⁸⁸ The FAD cofactor binds in a U‐shape configuration placing the adenine moiety in close proximity to the flavin isoalloxazine ring (Figures 1(c) and 2(b)). The secondary pocket is contained within the αβ‐domain and binds antennae pigments that enhance light‐dependent DNA repair in photolyases. ⁸⁹ , ⁹⁰ In CRYs, the secondary pocket is typically empty (Figure 2(a), (c)), contains amino acid substitutions that block antennae pigment recognition, ⁹¹ , ⁹² and is largely of unknown function in most CRY lineages. Structural studies of mammalian type II CRYs have identified the secondary pocket as a protein–protein interaction site essential to repression of CLOCK‐BMAL mediated transcription. ⁹² , ⁹³

FIGURE 2.

CRY Structure. (a) Structure of PHR domain containing an N‐terminal α/β domain (blue) connected by a long flexible linker (white), and a C‐terminal α helical domain (salmon). The presence of a CTT (black) differentiates CRYs from PHOTs. Solvent accessible surface area for the secondary pocket is shown in gray mesh. PDB ID: 6PU0. (b) Functionally divergent motifs, protrusion motif (teal), phosphate binding motif (purple) and C‐terminal lid, facilitate flavin binding and gate CTT (salmon) recognition. PDB ID: 4GU5. (c) The Ser‐Loop differentiates CRY subtypes (blue: mouse CRY2 [mCRY2]; cyan: dCRY) and gates protein–protein interactions in Type II CRYs with the secondary pocket. The Ser‐loop contacts the α15‐α16 helices (yellow) potentially enabling cross talk between the FAD binding pocket and secondary pocket. PDB IDs: 4GU5 (dCRY) and 4I6E (mCRY2)

Although CRY structures are homologous, several key regions differentiate members of the CPF family. First, all CRYs contain a species and isoform specific variable C‐terminal tail (CTT). Second, significant sequence and structural variations exist in disordered loop regions near the primary and secondary pocket. Specifically, functional divergence is observed in members of the CRY superfamily based on three‐loop regions near the primary pocket. These include a Phosphate‐Binding‐Motif (PBM) that recognizes a phosphate moiety in some crystal structures, a loop region near the FAD binding pocket termed the C‐terminal lid due to its ability to restrain the CTT of dCRY in its bound conformation, and a protrusion motif that can expand the size of the DNA‐binding cleft to facilitate CTT recognition (Figure 2(b)). In addition, there are distinct structural variances near the secondary binding pocket that have proven important for recognition of protein–protein interaction partners (Figure 2(c)). This peripheral site termed the Ser‐loop, based on mammalian CRY structures, is important for regulating CRY‐PER and CRY‐CLOCK interactions in an isoform specific manner. ⁹³ , ⁹⁴ In all cases, these variable regions appear to gate competitive protein–protein interaction networks in response to changing conditions. Currently, the cellular and chemical triggers gating selection of protein–protein interaction targets is a subject of immense interest. Below, we highlight the current state of understanding of structural factors selecting for different modes of CRY photochemistry, and conserved mechanisms of signal transduction in CRY protein–protein interactions.

4. MOLECULAR DETERMINANTS OF FLAVIN BINDING

Comparisons of Type I and Type II animal CRY structures provides mechanisms gating the ability to bind FAD. Crystal structures of a Type II animal CRY that were soaked with 1 mM FAD indicate that despite retaining FAD binding residues, differences within the disordered loops of the species specific PBM, C‐terminal lid, and protrusion motif result in non‐biologically relevant FAD binding affinity. ⁴³ Most notable are differences in the protrusion motif and C‐terminal lid, which are shorter in Type II CRYs resulting in an “open” conformation of the FAD‐binding pocket (Figure 3(a)). Examination of residues contacting FAD reveal minimal variations that could impact FAD binding, with the primary deviation residing at N419 in dCRY (Ser396 in mCRY1), which interacts with the adenine ring of FAD, and R298 (dCRY) within the C‐terminal lid that interacts with the pyrophosphate group of FAD through a bridging water molecule (Figure 3(b), (c)). Based on these observations, researchers have targeted these sites for mutation in both Type I and Type II CRYs with minimal success in disrupting flavin binding in Type I CRYs or restoring FAD binding in Type II CRYs. ⁴¹

FIGURE 3.

Flavin binding in CRYs. (a) Differences within the protrusion motif and C‐terminal lid differentiate dCRY (white), mCRY2 (orange), AtCRY1 (green), and ClCRY4 (yellow). An insertion in the protrusion motif in dCRY results in a “closed” conformation in the FAD pocket, compared to mCRY (arrow). The PBM (not shown) is absent in both mCRY and ClCRY4 structures consistent with an open conformation. PDB IDs: 4GU5 (dCRY), 6PU0 (ClCRY4), 1U3C (AtCRY1), and 4I6E (mCRY2). (b) FAD binding is affected by a pair of amine residues (Q311 and N419 dCRY numbering) that stabilize flavin in animal CRYs. Q311 is conserved in dCRY and ClCRY4 and hydrogen bonds to an FAD phosphate group. N419 hydrogen bonds with the adenosine moiety. The residue diverges in both mCRY2 and plant CRYs. R237 forms a salt bridge with the phosphate in dCRY. An equivalent interaction is not possible in Type II and IV CRYs (His) or AtCRY (Phe). Important residues are labeled dCRY/mCRY2/ClCRY4/AtCRY1. (c) Sequence alignment of residues important for FAD binding and tuning FAD chemistry. Colors differentiate residues conserved between CRYs

Analysis of structures within the CPF indicate that neither variations in active site residues, nor differences between “open” and “closed” conformations are likely to be sufficient to explain the lack of FAD binding in Type II CRYs. Residue identities at the position equivalent to N419 are variable in flavin binding CRYs and PHOTs, where plant CRYs contain a Gly residue incapable of interacting with the adenine ring (Figure 3(b), (c)). In this manner, an H‐bonding residue at this site is neither necessary or sufficient for FAD binding. Further, recent structures of a Type IV CRY call into question the role of “open” and “closed” conformations. ¹⁴ Dark state structures of ClCRY4 reveal FAD bound in an “open” conformation, where ClCRY4 does not contain the extended protrusion motif and density for the PBM is absent due to its dynamic “open” conformation (Figure 3(a)). Thus, at this juncture a structural explanation for the lack of FAD binding in Type II CRYs remains elusive, and likely resides in complex dynamics of the variable loop regions near the FAD binding pocket that are masked by the static nature of crystal structures. ⁴²

5. MOLECULAR DETERMINANTS OF CRY PHOTOCHEMISTRY

Alteration of the CRY photochemical outcomes differentiating CRY lineages and PHOTs, appears to depend on two chemical factors: (1) Modulation of the FAD and FADH^• reduction potentials, ⁷¹ and (2) Alteration of proton transfer necessary to generate FADH^•. ⁶⁴ , ⁷³ , ⁷⁶ , ⁷⁷ Structural and photochemical studies reveal a primary site adjacent to the N5 position of the isoalloxazine ring, which differentiates Type I (Cys), Type IV (Asn), PHOTs (Asn), and plant CRYs (Asp) (Figure 4(a), Table 1). Specifically, potentiometric analysis of CPF proteins reveals that the residue identity modulates the FAD reduction potential to select for FAD^ox in Type I and plant CRYs, as well as the FADH⁻ state necessary for DNA repair in photolyases. ⁷¹ , ⁹⁵ , ⁹⁶ Further, in plants Asp residues are responsible for proton transfer to FAD^•− to form FADH^• and differentiate plant CRY chemistry from Type I animal CRYs ⁶⁴ , ⁷³ , ⁷⁶ , ⁷⁷ (Figure 1(b)).

FIGURE 4.

Signal Transduction. (a) CRY photochemistry is tuned by the residue identity opposite the N5 position and modification of solvent access through L405 (dCRY). In dCRY signal propagation is initiated by FAD^•−, which alters H‐bonding to the CTT through H378. Residue labels are dCRY/ClCRY4/AtCRY1. (b) Changes in H378 H‐bonding propagates to the DNA binding groove and CTT. F534 binds to the photolyase DNA lesion site, which is occupied by ATP (green: AtCRY1) or glycerol (yellow: ClCRY4) in structures lacking the CTT. The site is bracketed by four residues highly conserved in animal CRYs, with only Type I CRYs deviating (N382 vs. H357). Residue numbering are for ClCRY4 (yellow). dCRY (white) and mCRY1 (wheat) conserve the binding pocket. (c) Isoform selective regulators (TH301) of mCRY1 bind to the primary pocket and DNA‐lesion site. Recognition of small molecules mirrors interactions in photoactive Type I and IV CRYs (b). (d) In Type II CRYs, PER (blue) wraps around the entirety of mCRY1 contacting the primary pocket, coiled‐coil helix (salmon), and Ser‐loop (light blue). Distance restraints suggest CLOCK (wheat) binds to the secondary pocket, ⁹³ and NMR indicates BMAL interacts with the coiled‐coil helix. ¹¹³ (e) Both BIC (Plant; light orange) and PER (Type II; blue) wrap around the CRY protein contacting both the primary, secondary pockets, and Ser‐loop (Green: AtCRY1; light blue: mCRY1). PDB IDs: 4GU5 (dCRY), 6PU0 (ClCRY4), and 1U3C (AtCRY1), 6K8K (AtCRY2‐BIC), 6KX8 (mCRY1‐TH301), 5T5X (mCRY1), and 6OF7 (mCRY1‐PER2)

Mutational studies in Type I dCRY reveal that C416D or C416N variants alone, are insufficient to select for FADH^•. Both C416D and C416N variants are capable of generating some FADH^•, but with low yield. ⁸⁴ Structural investigations indicate that solvent access to the N5 position is also necessary for tuning photochemical outcomes. Robust generation of the FADH^• state requires both L405E and C416N variants, where introduction of the L405E variant found in plant CRYs and Type IV CRYs leads to selection of the neutral semiquinone ⁸⁴ (Figure 4(a)). The ability to tune photochemical outcomes in CRY systems has afforded researchers a valuable tool instrumental in deciphering the ground and signaling states gating signal transduction in CRY proteins.

6. COUPLING CRY PHOTOCHEMISTRY TO SIGNAL TRANSDUCTION

To resolve conflicts between in vitro photochemistry and in vivo phenotypes, researchers elucidated mechanisms of signal transduction that couple FAD chemistry to conformational changes and protein–protein interactions. In all CRY lineages, signal transduction hinges upon conformational changes within the CTT. Focus on the CTT was based on several lines of evidence. First, overexpression of plant CRY CTTs are sufficient to generate a constitutively active response. ⁹⁷ , ⁹⁸ Second, in drosophila, truncating the CTT resulted in a constitutively active protein leading to degradation of Timeless (TIM) and CRY. ⁹⁹ Third, limited proteolysis studies demonstrated light‐dependent release of the CTT from the core PHR domain in Type I CRYs, ⁸² and light‐dependent stabilization of the CTT in the Type IV CRY from Columba livia (ClCRY4). ¹¹ , ¹⁴ , ⁴⁷ Lastly, even in light‐independent Type II CRYs, deletion of exon 11 within the CTT results in lengthening of circadian period. ¹⁰⁰ To date, only structures of dCRY have been solved with an intact CTT, ⁹¹ , ¹⁰¹ thus most of our insight into light‐regulated conformational changes within the CTT can be extrapolated from studies of dCRY and synergistic in vitro and in vivo studies of other CRY lineages.

In dCRY the CTT binds within the photolyase DNA‐binding cleft, where F534 in the CTT helix inserts into the pocket occupied by DNA lesions in PHOTs (Figure 4(b)). ⁹¹ Binding of the CTT to the DNA‐binding cleft is facilitated by rearrangements mediated by variable loops within the PBM, C‐terminal lid, and the protrusion motif (Figure 2(b)). Structural studies suggest a similar mode of CTT binding is likely conserved across the CRY superfamily. First, ClCRY4 structures contain a truncated CTT which orients toward the DNA binding cleft lined by the PBM, C‐terminal lid, and protrusion motif. ¹⁴ Similarly, drug‐discovery efforts in Type II CRYs indicate that elements within the CTT interact with the C‐terminal lid and PBM to gate isoform selectivity in CRY‐regulator compounds (Figure 4(c)). ⁸⁸ , ¹⁰² Second, in plant CRYs lacking the CTT, ATP binds to the DNA binding cleft (Figure 4(b)), and light induced autophosphorylation of CTT residues is coupled to ATP recognition at this site. ¹⁰³ In vivo, ATP stabilizes the open conformation where ATP and the CTT compete for binding. ⁷³ , ¹⁰³ Notably, the residues lining the DNA‐lesion site are highly conserved in animal CRYs and interact with small molecules in most CRY structures that have been solved to date (Figure 4(b), (c)). Thus, despite demonstrating differences in sequence identify, and differences in photochemical pathways, signal transduction mechanisms may be conserved in the CRY superfamily.

In dCRY a mechanism coupling CTT conformational changes to FAD chemistry has been elucidated through solution biochemistry, computational, and structural approaches that have recently been validated in vivo. First, limited proteolysis studies, and time‐resolved small‐angle x‐ray scattering (TrSAXS) indicate that chemical or photochemical reduction to form the FAD^•− state is sufficient for CTT release, and CTT conformational changes are coupled to structural transitions within the PBM. ⁸² , ¹⁰⁴ Second, computational studies predicted that CTT release was dependent on formation of FAD^•−, where introduction of the negative charge altered H‐bonding interactions to H378 that propagated to the CTT (Figure 4(a), (b)). ¹⁰⁵ These computational results were supported by mutations in vivo, but contradicted by TrSAXS experiments in vitro, where mutations to H378 did not significantly impact time‐dependent conformational changes, rather stabilized the ground state conformation. ¹⁰⁴ Conflicts between these two studies have recently been resolved by exploiting CRY variants, which tune photochemical outcomes. Using EPR and spin labeled CTTs, Crane and coworkers demonstrated that CRY proteins competent for FAD^•− formation underwent CTT release following blue‐light exposure, but the conformational change was abrogated in L405E:C416N that only form FADH^•. ⁸⁴ Thus, although no light‐state structures of dCRY exist, the cumulative data indicates light‐dependent formation of FAD^•− results in alteration of H‐bonding through H378, release of the CTT from the PHR core, and local ordering of the PBM.

Combining the signal transduction mechanism with in vivo analysis of signaling, provides insight into how CRY photochemistry dictates selection of protein–protein interactions targets. CTT conformational changes expose a protein‐interaction surface, which recruits TIM and Jetlag (JET) to induce TIM degradation and phase resetting of the circadian clock. ⁹¹ , ¹⁰⁶ , ¹⁰⁷ , ¹⁰⁸ Subsequent to TIM degradation, CRY undergoes light‐induced degradation through Ramshackle/BRWD3. ¹⁰⁹ Although both TIM binding and CRY degradation are dependent on CTT release, recent studies indicate that they may employ different downstream mechanisms or protein–protein interactions surfaces. Namely, H378R/K variants stabilize dCRY against light‐dependent degradation but retain activity against TIM. ⁸⁴ Currently, no structures of dCRY protein complexes exist, however some insight may be garnered from structural and biophysical characterization of Type II CRYs, where the CTT gates competitive interactions at the primary and secondary pocket to gate selection of CRY protein–protein interactions. ¹¹⁰

7. CTT DYNAMICS IN OTHER PHOTOACTIVE CRY LINEAGES

How CTT dynamics is coupled to primary photochemistry in Plant CRYs and Type IV animal CRYs remains an open question. Existing studies suggest that these lineages likely conserve aspects of the dCRY mechanism albeit with modifications to enable species specific downstream signaling. Both proteins diverge from Type I CRYs at the level of primary chemistry, whereby the stable signaling state of the FAD cofactor is believed to be FADH^•. However, formation of FADH^• proceeds through the FAD^•− state, ⁸ , ¹⁰ , ¹⁴ , ⁷⁴ , ¹¹¹ thus, it is possible, that transient formation of FAD^•− can facilitate CTT release. Consistent with such a mechanism, mutations that select for FAD^•− retain light induced conformational changes in plant CRYs, however, only in the presence of ATP which locks them in the open conformation following excitation. ⁷³ Further, plant CRYs are phosphorylated within the CTT following photoexcitation, ²⁶ , ¹⁰³ and it is believed this contributes to prolonging the active conformation of plant CRYs. ⁷³

Despite these similarities, two factors differentiate Type IV and plant CRYs from dCRY. First, whereas Type II and Type IV CRYs conserve H378, AtCRY1 and AtCRY2 contain residue substitutions at this site (AtCRY1: Asp; AtCRY2: Asn) (Figure 4(a)). Thus, the mechanism coupling FAD reduction to the CTT may differ. Notably, in AtCRY1, photoactivation leads to conformational changes with the αβ domain that has not been observed in other CRYs. ⁷³ , ⁷⁷ , ¹¹² It is intriguing to postulate that such changes may lead to reorganization of the secondary pocket to alter CTT interactions at this site in a manner analogous to light‐independent regulation of Type II CRY:CLOCK interactions. Second, Type IV CRYs can access the FADH⁻ state, which will also lead to introduction of negative charge. In such a case, transitioning between FADH^• and FADH⁻ states may be the relevant aspect of the photocycle for in vivo signal transduction, and could relay conformational changes to the CTT in a manner analogous to dCRY. Such an altered mechanism may explain why, in contrast to Type I CRYs, Type IV CRYs appear to undergo local ordering of the CTT following photoexcitation instead of CTT release. ¹⁴ Currently, more elaborate experimentations in the context of full‐length proteins are needed to decipher the mechanism of signal transduction in these systems.

8. PROTEIN–PROTEIN INTERACTIONS AND BEYOND

Despite detailed knowledge of CRY protein–protein interaction targets and their effect on physiological processes, we have limited access to structures of CRY‐protein complexes, and minimal insight into interaction partners relevant to magnetoreception. Given that protein interactions can modify the CRY photocycle, resolving remaining conflicts between photochemistry and magnetosensing cannot occur without understanding of the signaling components involved. Below, we summarize progress in understanding CRY protein‐interactions to highlight potential synergistic modes of signaling that can guide identification of a magnetosensory pathway.

Currently, the best model for understanding dynamic modes gating CRY protein–protein interactions stem from Type II CRYs that function in a light and FAD‐independent manner. ⁴¹ Biophysical and structural studies of CRY:Period (PER), CRY:CLOCK, CRY:BMAL, and CRY:Fbxl3 indicate dynamic competition between protein–protein interaction targets that are gated by the CTT (primary pocket) and Ser‐loop (secondary pocket). ⁹² , ⁹³ , ⁹⁴ , ¹¹⁰ , ¹¹³ , ¹¹⁴ These studies identify four protein–protein interaction surfaces impacting signal transduction that often overlap between binding partners. These include, the primary (Fbxl3) and secondary (CLOCK) pockets, a CRY coiled‐coil helix at the end of the PHR domain (BMAL), and a sinuous surface wrapping around the entirety of the protein (PER) (Figure 4(d)). Recently, the first structure of a photoactive CRY protein complex was determined, where AtCRY2 binds BIC2 in a manner analogous to that of CRY:PER, and competes with CIB2 and CRY2 oligomerization in protein‐complex formation (Figure 4(e)). ⁷⁰ These results suggest, that the modes of protein–protein interaction and competition between binding partners is likely conserved in CRY lineages and may function as a template for identifying and characterizing protein–protein interactions relevant to magnetosensing.

At this juncture, we have limited to no insight into viable magnetoreceptor complexes. Recent mass‐spec proteomics studies in Type IV CRYs has identified potential putative CRY protein‐interaction targets that may help elucidate the biological role of Type IV CRYs and possible magnetosensor complexes. ⁴⁶ , ⁴⁷ We note one intriguing candidate, GRIP proteins, which function as PDZ containing scaffolding proteins important for regulating AMPA‐type glutamate receptors in retinal ganglion cells. Studies of ClCRY4 indicate that CRY4 colocalizes with GRIP protein in retinal tissue. ⁴⁷ Further, Type I CRYs have also been determined to interact with the PDZ containing scaffolding protein Inactivation No Afterpotential D (INAD) in a light dependent manner in visual photoreceptor cells. ¹¹⁵ Given that Type I CRYs have been validated as essential for magnetoreception in the monarch butterfly, and that CRY4 proteins appear to be uniquely present in most migratory animals, it is intriguing to speculate that light‐regulated interactions with scaffolding proteins important for visual photoreception may be at the root of magnetosensing in animals.

Validating such mechanisms requires keener insight into how signal transduction, and protein–protein interactions are coupled to photochemical reactions generating a radical pair. Resolution of the current debates in the field may require identification of possible new radical pair mechanisms. ⁶³ , ¹¹⁶ We note that several have been recently proposed, including a light‐driven mechanism mirroring photolyase chemistry, and alternative radical pair components through protein–protein or small molecule interactions, ⁴ , ¹⁴ , ⁵¹ potentially including a third radical. ¹¹⁷ The recent advent of a tractable genetic model in the monarch butterfly should greatly expand our understanding of magenetoreception over the next several years.

AUTHOR CONTRIBUTIONS

Nischal Karki: Conceptualization; writing‐original draft; writing‐review & editing. Satyam Vergish: Conceptualization; writing‐original draft; writing‐review & editing.

Karki N, Vergish S, Zoltowski BD. Cryptochromes: Photochemical and structural insight into magnetoreception. Protein Science. 2021;30:1521–1534. 10.1002/pro.4124