A Charge-reversal Point Mutation Completely Depletes Flavin Chromophore from European Robin Cryptochrome 4a Protein

Cryptochrome4a(Cry4a)isamagneticallysensitiveproteinthatcouldenablenight- migratory birds to sense the geomagnetic field for navigation. The key to the protein magnetic sensitivity is the flavin adenine dinucleotide (FAD) cofactor, which initiates the electron transfer within the protein leading to a spin-correlated radical pair. De- spite its importance, the mechanism of FAD binding in avian Cry4a proteins remains unclear. Here we show that point mutagenesis of positively charged arginine residue at position 356 to negatively charged glutamic acid completely depletes FAD binding from European robin (Erithacus rubecula) Cry4a. The result indicates that electrostatic in- teractions constitute the primary driving force that enables FAD binding in European robin Cry4a. The finding provides new structural insight into the molecular basis of FAD binding in Cry4 and advances our understanding on the biophysical underpinnings of bird magnetoreception. 3 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint Table of Contents Graphic 4 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint Cryptochrome (Cry) proteins are flavoproteins that are involved in diverse biological pro- cesses including photoreception, circadian regulation, and potentially magnetoreception.1–7 The diverse biological functions are often associated with the falvin adedine dinucleotide (FAD) cofactor that can be non-covelently bound by some Cry proteins. Essentially, FAD is a blue light absorbing chromophore that can trigger the electron transfer process, lead- ing to a likely biological signaling cascade related to the protein.8–11Animal cryptochromes are classified into Type I, II and IV. Type I Cry proteins, such asDrosophila melanogaster Cry (DmCry), are blue-light photoreceptors through its FAD binding capacity.12,13 Type II Crys function as circadian clock regulators in a light-independent manner and do not bind FAD,14–17rather they utilize the primary and secondary binding pockets as protein-protein interaction sites for the circadian proteins CLOCK and PER.18–20Some studies suggest that FAD binding can be obtained through over-saturation of Type II Crys with FAD.21 Type IV Crys, exemplified by avian Cry4a, bind FAD16,22–24non-covalently and tightly. Substantial theoretical and experimental evidence supports that avian Cry4a undergoes photo-induced electron transfer between FAD and a conserved tetrad of tryptophans, a hallmark of radical- pair-based magnetic sensingin vitro.8,16,23,25–28 Despite the importance of FAD for the light-dependent functioning of the protein, the molecular mechanism of FAD binding in Cry proteins is only partially defined. InDmCry, single-point mutations hardly deplete FAD binding and double point mutations only par- tially deplete FAD binding by around 58%.17 Structural analysis suggests that FAD binding in DmCry is mediated through a cooperative effect of electrostatic interactions and steric modulation of the binding pocket accessibility.29 However, so far, there has been no killer experiment to completely deplete FAD from a Cry protein. Uncovering the FAD binding mechanism in avian Cry4 is therefore critical for elucidating the biophysical basis of magnetic sensitivity as it could possibly be manipulated directly through affecting FAD binding. In this study, we use molecular dynamics (MD) simulations to identify key amino acids involved in FAD binding in the European robin (Erithacus rubecula) Cry4a (ErCry4a). Fur- 5 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint thermore, we generate a recombinantErCry4a mutant protein based on the computational results, and experimentally assess FAD binding ability using light absorption spectroscopy and mass spectrometry (MS). Finally, we determine the conformational difference between ErCry4a wild type (WT) and mutant using native electrospray ionization mass spectrometry MS and ion mobility spectrometry MS, and propose a model of FAD binding mechanism in ErCry4a. We first analyzed interactions betweenErCry4a protein and FAD using Protein-Ligand Interaction Profiler (PLIP), a web tool for identification of non-covalent interactions between biologicalmacromoleculesandtheirligands. 30 Theresultsuggeststhat11aminoacidresidues in ErCry4a could interact with FAD (Figure 1A). The 11 amino acid redisues are S245, T246, T247, Q287, W350, H353, R356, F379, D385, D387, and N334. Strikingly, R356 is located parallel to both isoalloxazine and ribitol moieties of FAD, and interacts with FAD via diverse interaction including electrostatic-, cation-π, and hydrogen-bonding interactions. In contrast, S245, T246 and T247 appear to only form a weak hydrogen bonding network with FAD phosphate groups. 6 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint Figure 1: Computational simulation suggests key amino acids that affect FAD binding. (A) Illustration of interactions between amino acid residues and the flavin ade- nine dinucleotide (FAD) cofactor inside European robin (Erithacus rubecula) Cry4a protein (ErCry4a). A total of 11 amino acid residues inErCry4a potentially interacts with FAD through hydrogen bond, hydrophobic interaction and cation-π interactions. (B) Predicted changes in the binding free energy upon mutation to alanine for the 11 amino acid residues that interact with the FAD. For the higher∆∆H values, the mutations are predicted to more significantly disrupt FAD binding. (C) Predicted binding affinity for FAD inErCry4 wild type (WT) and R356E mutant. The absolute value of∆H in R356E is 16.2 kcal/mol smaller than that of WT, indicating a weaker FAD binding affinity in R356E compared to that in WT. It should be stressed that the absolute∆H values cannot be used to directly determine the FAD binding affinity, but these calculations reliably predict changes in binding affinities. (D) Structural comparison ofErCry4a andDrosophila melanogastercryptochrome (DmCry). The β-sheet on top of FAD inDmCry is absent inErCry4a, indicating a negli- gible steric constraints effect on FAD binding inErCry4a. To quantify the interactions between the 11 residues and FAD, we computed binding free energy of X-to-alanine mutants, where X refers to one of the 11 residues. Alanine scanning mutagenesis is ideal for testing how much a particular residue contributes to binding because alanine is small and chemically neutral.31,32 The computational alanine scanning result shows that the R356 residue stands out of the 11 amino acid residues surrounding FAD due to the high binding free energy (10 kcal/mol), indicating that R356 is the most energetically significant contributor to FAD binding (Figure 1B). We hypothesize that the strong interaction between R356 and FAD is primarily driven by 7 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint electrostatic forces, as arginine has a positive charge while FAD is negatively charged. To test the hypothesis, we introduced a charge-reversal mutation, replacing R356 with the negatively charged glutamic acid (R356E). Computational simulations suggest that the FAD binding affinity of R356E is 16.2 kcal/mol lower than that of the wild type, indicating substantially weaker FAD binding in the mutant (Figure 1C). Futhermore, structural analysis reveals that a β-sheet critical for FAD binding inDmCry is absent inErCry4a (Figure 1D), suggesting that steric constraints on FAD binding inErCry4a are negligible. Figure 2: A single-point mutant ErCry4a R356E completely deplete FAD from ErCry4. (A)Both R356Eand WTproteins elutefromthe anionexchangecolumn atsimilar solvent conductivity during gradient elution process of anion exchange chromatography. (B) Sodium dodecyl sulfate – polyacrylamide gel electrophoresis Coomassie Blue Dye staining suggests that both proteins are comparably pure. (C) UV-visible absorption spectrum shows that the ErCry4 R356E mutant does not absorb blue light between 350 nm and 500 nm, indicating the absense of FAD (the blue-light chromophore) binding. ErCry4 wild type (WT) is used as a positive control showing characteristic spectral features of FAD binding. Following thein silicoresults, we generated theErCry4a R356E mutant in the wet-lab through site-directed mutagenesis and recombinant protein expression. During the gradient elution process of anion exchange chromatography, R356E mutant eluted from the anion exchange column at a solvent conductivity similar to that of WT protein (Figure 2A), sug- gesting that R356E mutagenesis maintains a similar surface charge pattern as in the WT 8 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint protein. The purity of both protein samples is also similar, as shown by the single band on the Commassie blue staining image of a protein gel (Figure 2B). Most importantly, the UV–vis spectrum ofErCry4a R356E mutant appears flat (near-to-null) in the absorbing range between 300 nm and 500 nm, indicating no blue-light absorption of R356E mutant and thus no FAD binding (Figure 2C). As a positive control, the WT protein shows vibra- tional fine structure around 350 nm and 450 nm, which are characteristic features typically associated with FAD binding in cryptochrome proteins. Furthermore, mass spectrometry was employed to determine the protein sequence and cofactor binding status ofErCry4a WT and R356E mutant. Under denaturing conditions, both proteins display broad charge state distributions, ranging from∼ 18+ to 46+. The broad charge state distributions suggest that the proteins are completely unfolded and thus FAD was removed from the protein. This allowed assessing the subtle mass difference at- tributable solely to amino acid substitution. Deconvolution of the denatured spectra yields a mass difference of∼ 29 Da between WT and R356E, which matches the theoretical mass difference between arginine and glutamic acid being 27.08 Da (Figure 3A). Consistently, a minor mass shift was also observed in native MS measurements on FAD-free WT protein and the R356E mutant protein (Figure 3B, the right panel). Together, the MS results validate that arginine has been successfully replaced with glutamic acid in the R356E mutant. Under native-like condition, ErCry4a WT and mutant rotein charge state distributions narrows between 12+ and 16+, dominated by 13+ and 14+, consistent with folded conformations. The R356E mutant is detected exclusively without FAD, whereas the majority of WT (94%) binds FAD, indicated by a 786 Da mass difference (Figure 3A). It is interesting to note that, a dimer species is observed in both denaturing (Figure 3A) and native (Figure 3B) measure- ments. The protein dimers are very likely formed through the covalently disulfide bonds due to the absence of reducing reagent in the buffer solvent. The observation is consistent with our previous study onErCry4a dimer.33 9 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint Figure 3: Mass spectrometry measurements validates thatErCry4a R356E mu- tant does not bind FAD.(A) The denaturing MS detected a mass difference of∼ 29 Da mass between WT and R356E mutant under the denaturing condition. The observed mass difference matches the theoretical mass difference between arginine and glutamic acid residues (27 Da), confirming the successful amino acid substition. The stars in the left panel mark denaturing protein dimers in the m/z range of 2400 and 3400. (B) Native MS detected a mass difference of 786 Da between WT and R356E under the native condition. The mass difference is consistent with the theoretical molecular weight of FAD. The dimer population is highlighted by a magnification factor of five at the m/z range of 5600 and 6800. (C) Ion mobility spectrometry - mass spectrometry (IMS-MS) shows that the collision cross section of R356E mutant is slightly left-shifted compared to FAD-bound WT protein, indicating a slighted more compacted conformation of R356E compared to the WT protein with FAD- bound. 10 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint Ion mobility spectrometry (IMS) measurements provide further insight into conforma- tional properties of the two proteins. The FAD-bound WT protein exhibits a slightly larger collision cross section (CCS) value (Figure 3C) compared to the R356E mutant protein. For example, the CCS value at 14+ charge state is 3624.7 Ų for the FAD-bound WT protein and 3531.0 Ų for the R356E mutant protein. The measured CCS difference (ΔCCS) between the two protein is 2.7%. However, the expected CCS difference due to arginine substitution by glutamic acid is approximately∼ 0.9%, which was calculated asCCS ∝ m2/3.34 The ob- served 2.7% shift therefore implies a conformational effect attributable to cofactor-dependent packing rather than the amino acid residue mass alone. Protein conformation ofErCry4a R356E mutant is probably more compacted than that of the WT protein. In summary, our study clearly shows that site-specific electrostatic repulsion is sufficient to completely block FAD binding inErCry4a. To our knowledge, this is the first report of a single-point mutation that completely depletes FAD binding in any Cry protein. Previous work onDmCry required dual mutations (R298E + Q311E) and the mutations only partially reducedFADoccupancyby58%. 17 FADbindinginCryproteinshasbeenproposedtodepend on two factors, i) non-covalent interactions between FAD and the binding pocket and ii) the steric availability of the binding pocket itself.29 In Type I Cry, strong non-covalent interactions and a conformationally closed binding pocket make FAD depletion difficult, typically requiring multiple mutations—one to weaken the non-covalent interactions and another to open the pocket.17,29 Specifically, in the DmCry double mutant, R298E was used to remove theβ-sheet lid and Q311E was used to remove the pincer holding FAD in place.29 In the present study, we discovered that FAD binding is primarily mediated by electrostatic attraction, while steric constraints play a negligible role inErCry4a (Figure 4). It is perhaps suprising that one could achieve complete FAD depletion by a single-point mutation. However, an evolutionary divergence in the FAD-binding mechanism is likely to have occurred between Type VI Cry, exemplified by ErCry4a, and Type I Cry, exemplified by DmCry, due to differences in their protein structures. The apparent absence of a canonical 11 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint β-sheet lid inErCry4a raises broader questions about how Cry proteins have structurally diversified to support species-specific physiological functions. Figure 4: A model of FAD binding mechanism inErCry4a protein: electrostatic attractions secure FAD non-covalently binding in Cryptochrome 4a.The modelled structure of ErCry4a WT and R356E mutant are shown in yellow and blue, respectively. The inset shows the interaction interface between the FAD and the R356E residue. Single-point mutations are generally preferred in functional studies, as they introduce minimal perturbations to protein structure compared to dual or multiple mutations. In the present study, the single-point mutation preserved the overall charge profile ofErCry4a WT protein, with only a slightly more compacted conformation observed in the mutant. The FAD-free R356E mutant ofErCry4a will serve as an ideal negative control to investigate the structural and functional role of FAD in cryptochrome signaling and protein-protein interactions.35 In a much broader context, flavin-bound proteins have recently been employed as photosensitizers in photocatalytic proximity labeling, a technique that uses light-activated catalysts to tag nearby proteins or molecules for identification.36 A flavin-free Cry4a mutant can also provides a valuable negative control in this emerging application. In this regard, 12 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint our study establishes a framework for generating flavin-free cryptochrome apoproteins. In conclusion, this study provides new molecular insight into cofactor stabilization in avian Cry4a, revealing that electrostatic interactions, rather than steric constraints, domi- nate FAD retention. The R356E mutant serves as a powerful negative control for functional studies of ErCry4a, enabling precise dissection of FAD-dependent signaling pathways in magnetoreception and in emerging photocatalytic proximity labeling applications. Future work should investigate whether the charge-reversal single-point mutagenesis strategy simi- larly disrupts FAD binding in Cry4a proteins from other bird species, and how FAD binding influences Cry4a conformational dynamics and protein–protein interactions,9–11which are both critical for understanding its role in light-dependent magnetic sensing.7

A homology model ofErCry4a was built based on theClCry4 template (PDB ID: 6PU0)16 in Swiss model workspace. The resultingErCry4a homology model was solvated in a water box with a padding distance of 12 Å. Hydrogens were added to theErCry4a model, while the topology was created using pdb2gmx from the gromacs software.37,38 The protonation state of ErCry4a was estimated at pH 7 using PDB2PQR39 fromm the APBS biomolec- ular solvation software suite website.40 FAD was parametrized using GAFF2,41 while the ErCry4a protein was simulated using the ff99SB-ILDN forcefield.3 Specifically, simulations were carried out using a two femtoseconds timestep with the leap-frog integrator. Energy minimization prior to the molecular dynamics simulation was carried out using the steep- est decent method. The temperature of the system was set to 300 K using the modified Berendsen thermostat and pressure was controlled at 1 bar using the Parinello-Rahman barostat.42,43 Long range electrostatic interactions were modeled using the Particle mesh Ewald summation approach.44 Constraints were set on hydrogen bond distances using the LINCS algorithm.45 Starting velocities were generated with a random seed. The simulation 13 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint protocol included: i) energy minimization for 20,000 steps; ii) equilibration with restraints on the protein in the constant-temperature ensemble for 10 picoseconds; iii) equilibration in the constant-pressure ensemble for 100 picoseconds; iiii) production simulation in the constant-pressure ensemble for 50 nanoseconds. The residues that potentially influence FAD binding affinity were scanned using the protein ligand interaction profiler webserver.30 The in silico alanine scanning technique was subsequently used to quantify the contribution of the residues to the FAD binding affinity. Constant-pressure simulations were carried out following the equilibration and 20 frames were prepared as the input for the alanine scan analysis. Mutations of the FAD-surrounding amino acid residues to alanine were carried out using gmx_MMPBSA.46 FAD binding affinities forErCry4a wild type and mutant proteins were calculated using the Poisson–Boltzmann method implemented ingmx_MMPBSA.46 The analysis was carried out in Prism using the Welch’st-test to find significance and differences in binding affinities. ErCry4a R356E mutant DNA was generated through a polymerase chain reaction (PCR) using the Q5 site-directed mutagenesis kit (New England Biolabs, Ipswich, MA, USA). The PCR primer sequences employed were 5’-TCACCTGGCTGAACACGCCGTCG-3’ and 5’- TGGATCCAGCCTTCCTGG-3’. The DNA sequence was confirmed via Sanger sequencing (LGC Genomics, Berlin, Germany). TheErCry4a R356E mutant and the WT protein were expressed and purified as previously reported.23,25 The UV-visible absorption spectra were measured using the Cary 60 UV-Vis Spectrophotometer (Agilent, Santa Clara, CA, US). 20 µM protein was used for MS measurements. Prior to the native MS measurements, protein samples were buffer-exchanged to 150 mM ammonium acetate (pH 8) using ZebaTM Micro Spin desalting columns with a molecular weight exclusion limit of 40 kDa (Ther- moFisher Scientific). Prior to the denaturing MS measurements, proteins were incubated in pH 2 150 mM ammonium acetate solution for 5 minutes. The MS and ion mobility spectom- etry–MS (IMS-MS) data were acquired on a timsTOF Ultra1 instrument (Bruker Daltonics) in positive ion mode using a CaptiveSpray 2 nanoESI source. To generate an ion source, the 14 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint capillary voltage was set to 1.1 kV and dry gas was pumped at the speed of 3L · min−1. The general temperature of the ion source was 80°C. For the ion optics (native transmission), the parameters were as follows: Deflection 0 Delta = 105 V, Funnel 0 RF = 250 Vpp, Multipole 0 RF = 200 V, Deflection 1 Delta = 80 V, Funnel 1 RF = 400 Vpp, isCID Energy = 100-125 eV, Funnel 2 RF = 600 Vpp, Multipole RF = 1200 Vpp, Collision energy: 10 eV, Collision RF: 2000 Vpp, Ion Energy = 10 eV, Low Mass = 500 m/z, Tranfer time = 120µs, Pre-pulse storage = 120µs. The trapped ion mobility spectrometry in/out tunnel pressures used were 2.4/0.7 mbar (ΔP 1.7 mbar). The exact parameters were: Accumulation = 20 – 30 ms;1/K0 ramp= 100 ms; 1/K0 window = 0.6 V · s · cm−1; scan ramp rate = 1.1 kV · s−1. DC potentials: DC potentials: ∆t1 = −20 V, ∆t2 = −120 V, ∆t4 = 100 V, ∆t5 = 0 V, ∆t6 = 20 − 40 V. m/z and 1/K0 were calibrated with ESI-L low-concentration tune mix (Agilent). The collision cross section (CCS) in N2 were calculated from the Mason–Schamp equation using the manufacturer CCS values. The MS and IMS-MS spectra were exported using Compass DataAnalysis software (Bruker Daltonics, v6.1). The deconvoluted spectra were obtains using UniDec47 with the following parameters: m/z ratio range: 1500-5000 Th, charge range: 10-50, mass range: 30-150 kDa, sample mass set to every 5 Da, peak FWHM: 2 Th, peak shape: Gaussian. AUTHOR INFORMATION Corresponding Author Jingjing Xu- University of Southern Denmark, Campusvej 55, 5230 Odense M, Den- mark; orcid.org/0000-0003-0311-5711; Email: [email protected] Authors Emil Sjulstok Rasmussen-Department of Neuroscience, University of Texas South- western Medical Center, 6000 Harry Hines Blvd. Dallas, TX 75390-9111, USA;orcid.org/0000- 15 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint 0003-2938-7248 Francis Berthias -Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark; orcid.org/0000-0001-5178- 8967 Jessica Schmidt -Institute of Biology and Environmental Sciences, Carl von Ossi- etzky Universität Oldenburg, Carl-von-Ossietzky-Straße 9-11, 26129 Oldenburg, Germany; orcid.org/0000-0002-8393-4532 Henrik Mouritsen - Carl von Ossietzky Universität Oldenburg, Carl-von-Ossietzky- Straße 9-11, 26129 Oldenburg, Germany; orcid.org/0000-0001-7082-4839 Ole N. Jensen - Department of Biochemistry and Molecular Biology, University of Southern Denmark, Campusvej 55, 5230 Odense M, Denmark; orcid.org/0000-0003-1862- 8528 IliaA.Solov’yov-Carl-von-Ossietzky-Straße 9-11, 26129 Oldenburg, Germany; orcid.org/0000- 0002-8626-145X Author Contributions J.X. and E.S. contributed equally to this work. Notes The authors declare no conflicts of interest.

This work was supported by Lundbeck Foundation (Postdoc fellowship grant number: R402- 2022-1372, awardedtoJ.X.grantnumber: R303-2018-3175, awardedtoE.S.R);NovoNordisk Foundation (INTEGRA, grant number NNF20OC0061575 to O.N.J.); European Research Council (under the European Union’s Horizon 2020 research and innovation programme, grant agreement no. 810002, Synergy Grant: ‘QuantumBirds’, awarded to H.M.); Volkswa- gen Foundation (Lichtenberg Professorship to I.A.S.), the Deutsche Forschungsgemeinschaft 16 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint (SFB 1372: Magnetoreception and Navigation in Vertebrates, no. 395940726, and EXC- 3051: NaviSense, no. 533653176 to H.M. and I.A.S); TRR386/1-2023 Hyperpolarization in molecular systems HYP*MOL, no 514664767 to I.A.S.), the Ministry for Science and Culture of Lower Saxony (Simulations Meet Experiments on the Nanoscale: Opening up the Quantum World to Artificial Intelligence (SMART) and Dynamik auf der Nanoskala: Vonkohärenten Elementarprozessen zur Funktionalität (DyNano)).

(1) Ahmad, M.; Cashmore, A. R. HY4 gene of A. thaliana encodes a protein with charac- teristics of a blue-light photoreceptor.Nature 1993, 366, 162–166. (2) Sancar, A. Regulation of the mammalian circadian clock by cryptochrome.Journal of Biological Chemistry2004, 279, 34079–34082. (3) Liedvogel, M.; Mouritsen, H. Cryptochromes—a potential magnetoreceptor: what do we know and what do we want to know?Journal of the Royal Society Interface2010, 7, S147–S162. (4) Ritz, T.; Adem, S.; Schulten, K. A model for photoreceptor-based magnetoreception in birds. Biophysical journal2000, 78, 707–718. (5) Bartölke, R.; Behrmann, H.; Görtemaker, K.; Yee, C.; Xu, J.; Behrmann, E.; Koch, K.- W. The secrets of cryptochromes: photoreceptors, clock proteins, and magnetic sensors. Neuroforum 2021, 27, 151–157. (6) Solov’yov, I. A.; Domratcheva, T.; Schulten, K. Separation of photo-induced radical pair in cryptochrome to a functionally critical distance.Scientific Reports 2014, 4, 3845. 17 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint (7) Hore, P. J.; Mouritsen, H. The radical-pair mechanism of magnetoreception.Annual review of biophysics2016, 45, 299–344. (8) Timmer, D.; Frederiksen, A.; Lunemann, D. C.; Thomas, A. R.; Xu, J.; Bartolke, R.; Schmidt, J.; Kubar, T.; De Sio, A.; Solov’yov, I. A.; others Tracking the electron transfer cascade in European robin cryptochrome 4 mutants.Journal of the American Chemical Society2023, 145, 11566–11578. (9) Wu, H.; Scholten, A.; Einwich, A.; Mouritsen, H.; Koch, K.-W. Protein-protein inter- action of the putative magnetoreceptor cryptochrome 4 expressed in the avian retina. Scientific reports2020, 10, 7364. (10) Görtemaker, K.; Yee, C.; Bartölke, R.; Behrmann, H.; Voß, J.-O.; Schmidt, J.; Xu, J.; Solovyeva, V.; Leberecht, B.; Behrmann, E.; others Direct interaction of avian cryp- tochrome 4 with a cone specific G-protein.Cells 2022, 11, 2043. (11) Yee, C.; Bartölke, R.; Görtemaker, K.; Schmidt, J.; Leberecht, B.; Mouritsen, H.; Koch, K.-W. Comparison of retinol binding protein 1 with cone specific G-protein as putative effector molecules in cryptochrome signalling. Scientific Reports 2024, 14, 28326. (12) Haug, M. F.; Gesemann, M.; Lazović, V.; Neuhauss, S. C. Eumetazoan cryptochrome phylogeny and evolution.Genome biology and evolution2015, 7, 601–619. (13) Ozturk, N. Phylogenetic and functional classification of the photolyase/cryptochrome family.Photochemistry and photobiology2017, 93, 104–111. (14) Buhr, E. D.; Takahashi, J. S. Molecular components of the Mammalian circadian clock. Circadian clocks2013, 3–27. (15) Thresher, R. J.; Vitaterna, M. H.; Miyamoto, Y.; Kazantsev, A.; Hsu, D. S.; Pe- tit, C.; Selby, C. P.; Dawut, L.; Smithies, O.; Takahashi, J. S.; others Role of mouse 18 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint cryptochrome blue-light photoreceptor in circadian photoresponses.Science 1998, 282, 1490–1494. (16) Zoltowski, B. D.; Chelliah, Y.; Wickramaratne, A.; Jarocha, L.; Karki, N.; Xu, W.; Mouritsen, H.; Hore, P. J.; Hibbs, R. E.; Green, C. B.; others Chemical and struc- tural analysis of a photoactive vertebrate cryptochrome from pigeon.Proceedings of the National Academy of Sciences2019, 116, 19449–19457. (17) Kutta, R.; Archipowa, N.; Johannissen, L.; Jones, A.; Scrutton, N. Vertebrate cryp- tochromes are vestigial flavoproteins. Sci Rep 7: 44906. 2017. (18) Nangle, S. N.; Rosensweig, C.; Koike, N.; Tei, H.; Takahashi, J. S.; Green, C. B.; Zheng, N. Molecular assembly of the period-cryptochrome circadian transcriptional repressor complex.Elife 2014, 3, e03674. (19) Michael, A. K.; Fribourgh, J. L.; Chelliah, Y.; Sandate, C. R.; Hura, G. L.; Schneidman- Duhovny, D.; Tripathi, S. M.; Takahashi, J. S.; Partch, C. L. Formation of a repressive complex in the mammalian circadian clock is mediated by the secondary pocket of CRY1.Pnas 2017, 114, 1560–1565. (20) Rosensweig, C.; Reynolds, K. A.; Gao, P.; Laothamatas, I.; Shan, Y.; Ranganathan, R.; Takahashi, J. S.; Green, C. B. An evolutionary hotspot defines functional differences between CRYPTOCHROMES.Nature Communications2018, 9, 1138. (21) Xing, W.; Busino, L.; Hinds, T. R.; Marionni, S. T.; Saifee, N. H.; Bush, M. F.; Pagano, N., Michele Zheng SCFFBXL3 ubiquitin ligase targets cryptochromes at their cofactor pocket.Nature 2013, 496, 64–68. (22) Qin, S.; Yin, H.; Yang, C.; Dou, Y.; Liu, Z.; Zhang, P.; Yu, H.; Huang, Y.; Feng, J.; Hao, J.; others A magnetic protein biocompass.Nature materials2016, 15, 217–226. 19 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint (23) Xu, J.; Jarocha, L. E.; Zollitsch, T.; Konowalczyk, M.; Henbest, K. B.; Richert, S.; Golesworthy, M. J.; Schmidt, J.; Déjean, V.; Sowood, D. J.; others Magnetic sensitivity of cryptochrome 4 from a migratory songbird.Nature 2021, 594, 535–540. (24) Ozturk, N.; Selby, C. P.; Song, S.-H.; Ye, R.; Tan, C.; Kao, Y.-T.; Zhong, D.; Sancar, A. Comparative photochemistry of animal type 1 and type 4 cryptochromes.Biochemistry 2009, 48, 8585–8593. (25) Gravell, J.; Pitcher, T.; Henbest, K.; Schmidt, J.; Buffet, M.; Moise, G.; Gehrckens, A.; Cubbin, D.; Stuhec, A.; Antill, L.; others Spectroscopic characterisation of radical pair photochemistry in non-migratory avian cryptochromes: magnetic field effects in GgCry4a. Journal of the American Chemical Society2025, (26) Wong, S. Y.; Wei, Y.; Mouritsen, H.; Solov’yov, I. A.; Hore, P. Cryptochrome magne- toreception: four tryptophans could be better than three.Journal of the Royal Society Interface 2021, 18, 20210601. (27) Frederiksen, A.; Langebrake, C.; Hanic, M.; Manthey, G.; Mouritsen, H.; Liedvogel, M.; Solov’yov, I. A. Mutational study of the tryptophan tetrad important for electron trans- fer in European robin cryptochrome 4a.ACS omega2023, 8, 26425–26436. (28) Hochstoeger, T.; Al Said, T.; Maestre, D.; Walter, F.; Vilceanu, A.; Pedron, M.; Cush- ion, T. D.; Snider, W.; Nimpf, S.; Nordmann, G. C.; others The biophysical, molecular, and anatomical landscape of pigeon CRY4: A candidate light-based quantal magne- tosensor. Science advances2020, 6, eabb9110. (29) Sjulstok, E.; Solov’yov, I. A. Structural explanations of flavin adenine dinucleotide binding in Drosophila melanogaster cryptochrome.The Journal of Physical Chemistry Letters 2020, 11, 3866–3870. (30) Adasme, M.; Linnemann, K.; Bolz, S.; Kaiser, F.; Salentin, S.; Haupt, V.; PLIP, M. S. Expanding the scope of the protein–ligand interaction profiler to DNA and RNA., 2021, 20 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint 49, pp.W530-W534. DOI: https://doi. org/10.1093/nar/gkab294. PMID: https://www. ncbi. nlm. nih. gov/pubmed/339502142021, (31) Martins, S. A.; Perez, M. A.; Moreira, I. S.; Sousa, S. F.; Ramos, M.; Fernandes, P. Computational alanine scanning mutagenesis: MM-PBSA vs TI.Journal of chemical theory and computation2013, 9, 1311–1319. (32) Liu, X.; Peng, L.; Zhou, Y.; Zhang, Y.; Zhang, J. Z. Computational alanine scanning with interaction entropy for protein–ligand binding free energies.Journal of chemical theory and computation2018, 14, 1772–1780. (33) Hanic, M.; Antill, L. M.; Gehrckens, A. S.; Schmidt, J.; Gortemaker, K.; Bartolke, R.; El-Baba, T. J.; Xu, J.; Koch, K.-W.; Mouritsen, H.; others Dimerization of European robin cryptochrome 4a.The Journal of Physical Chemistry B2023, 127, 6251–6264. (34) Rolland, A. D.; Biberic, L. S.; Prell, J. S. Investigation of charge-state-dependent com- paction of protein ions with native ion mobility–mass spectrometry and theory.Journal of the American Society for Mass Spectrometry2022, 33, 369–381. (35) Görtemaker, K.; Yee, C.; Bartölke, R.; Behrmann, H.; Voß, J.; Schmidt, J.; Xu, J.; Solovyeva, V.; Leberecht, B.; Behrmann, E.; others Direct interaction of avian cryp- tochrome 4 with a cone specific G-protein. Cells. 2022. V. 11. (36) Knutson, S. D.; Buksh, B. F.; Huth, S. W.; Morgan, D. C.; MacMillan, D. W. Current advances in photocatalytic proximity labeling.Cell Chemical Biology2024, 31, 1145– 1161. (37) Berendsen, H. J.; van der Spoel, D.; van Drunen, R. GROMACS: A message-passing parallel molecular dynamics implementation.Computer physics communications1995, 91, 43–56. 21 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint (38) Abraham, M. J.; Murtola, T.; Schulz, R.; Páll, S.; Smith, J. C.; Hess, B.; Lindahl, E. GROMACS: High performance molecular simulations through multi-level parallelism from laptops to supercomputers.SoftwareX 2015, 1, 19–25. (39) Dolinsky, T. J.; Nielsen, J. E.; McCammon, J. A.; Baker, N. A. PDB2PQR: an auto- mated pipeline for the setup of Poisson–Boltzmann electrostatics calculations.Nucleic acids research2004, 32, W665–W667. (40) Unni, S.; Huang, Y.; Hanson, R. M.; Tobias, M.; Krishnan, S.; Li, W. W.; Nielsen, J. E.; Baker, N. A. Web servers and services for electrostatics calculations with APBS and PDB2PQR. Journal of computational chemistry2011, 32, 1488–1491. (41) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. Development and testing of a general amber force field.Journal of computational chemistry2004, 25, 1157–1174. (42) Parrinello, M.; Rahman, A. Polymorphic transitions in single crystals: A new molecular dynamics method.Journal of Applied physics1981, 52, 7182–7190. (43) Aoki, K.; Yonezawa, F. Constant-pressure molecular-dynamics simulations of the crystal-smectic transition in systems of soft parallel spherocylinders.Physical Review A 1992, 46, 6541. (44) Essmann, U.; Perera, L.; Berkowitz, M. L.; Darden, T.; Lee, H.; Pedersen, L. G. A smooth particle mesh Ewald method.The Journal of chemical physics1995, 103, 8577– 8593. (45) Hess, B. P-LINCS: A parallel linear constraint solver for molecular simulation.Journal of chemical theory and computation2008, 4, 116–122. (46) Valdés-Tresanco, M. S.; Valdés-Tresanco, M. E.; Valiente, P. A.; Moreno, E. 22 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint gmx_MMPBSA: a new tool to perform end-state free energy calculations with GRO- MACS. Journal of chemical theory and computation2021, 17, 6281–6291. (47) Marty, M. T.; Baldwin, A. J.; Marklund, E. G.; Hochberg, G. K.; Benesch, J. L.; Robinson, C. V. Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensembles.Analytical chemistry2015, 87, 4370–4376. 23 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint 24 .CC-BY-NC-ND 4.0 International licenseavailable under a (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made The copyright holder for this preprintthis version posted November 25, 2025. ; https://doi.org/10.1101/2025.11.24.690116doi: bioRxiv preprint