Magnetic field effects on the structure and molecular behavior of pigeon iron–sulfur protein

Abstract

Pigeon iron–sulfur (Fe–S) cluster assembly 1 homolog (clISCA1) is a target protein for research into the biomagnetoreception mechanism, as the clCRY4/clISCA1 oligomer, a complex composed of the columnar clISCA1 oligomer and the magnetosensor candidate protein cryptochrome‐4 (clCRY4) oligomer, tends to orient itself along weak magnetic fields, such as geomagnetic fields, under blue light. To obtain insight into the magnetic orientation mechanism of the clCRY4/clISCA1 oligomer, we inspected magnetic field effects on the structure and molecular behavior of clISCA1 by small angle X‐ray scattering analysis. The results indicated that the clISCA1 protomer took the Fe–S cluster‐bound globular form and unbound rod‐like form. The globular clISCA1 protomer assembled to form columnar oligomers, which allowed for the binding of many Fe–S clusters at the interface between clISCA1 protomers. Moreover, the translational diffusion and the columnar oligomerization of clISCA1 were controlled by the external static magnetic field and Fe–S clusters bound to clISCA1. However, the columnar clISCA1 oligomer was not oriented along the external static magnetic field (~1 T) when clCRY4 was not bound to clISCA1. This result indicated that clCRY4 has a function to enhance the magnetic orientational property of clCRY4/clISCA1 oligomer.

1. INTRODUCTION

Many species of birds, mammals, reptiles, amphibians, fish, crustaceans, insects, and plants can detect the geomagnetic field for orientation and navigation. ¹ Cryptochromes (CRYs), highly conserved blue light absorbing flavoproteins containing a flavin adenine dinucleotide (FAD), act as quantum magnetic sensors to detect the Earth's magnetic field ² , ³ in many magnetoreceptive species. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ However, it is not clear what types of proteins act as scaffolds for the magnetoreception reaction of CRY or as CRY receptors to transmit magnetic information to the nervous system.

In 2016, the iron–sulfur (Fe–S) cluster assembly 1 homolog (ISCA1), one of the Fe–S cluster carrier proteins contributing to Fe–S cluster biosynthesis machinery in the mitochondria, ¹¹ was identified to potentially interact with CRY. ¹² The 3D structure of the CRY4/ISCA1 oligomer of Columba livia (clCRY4/clISCA1 oligomer) was determined by the combination of cryo‐electron microscopy (cryo‐EM) analysis at 22–25 Å resolution and computational modeling using the crystal structures of the dmCRY of Drosophila melanogaster (PDB ID: 4GU5, sequence identity to clCRY4: 37.35%) and the ecISCA of Escherichia coli (PDB ID: 1R94, sequence identity to clISCA1: 37.25%) as templates. ¹² In that study, the clCRY4/clISCA1 oligomer's structure was characterized by an outer layer double‐helix formed by clCRY4 and a columnar core oligomer formed by clISCA1 (Figure 1). The columnar clISCA1 oligomer and the clCRY4/clISCA1 oligomer observed by cryo‐EM were ~24 nm long, but size exclusion chromatography (SEC) indicated that the number of assemblies of clISCA1 is variable. Interestingly, 45% of the clCRY4/clISCA1 oligomer in the cryo‐EM sample appeared to be parallel to a weak static magnetic field (40 μT) like the geomagnetic field (25–65 μT) ¹³ under blue light, ¹² which is a very rare phenomenon for proteins. This magnetic orientational property improved with increasing magnetic field strength, as 55% of the clCRY4/clISCA1 oligomer paralleled to a static magnetic field of 1 mT. ¹²

FIGURE 1.

Model of the cryptochrome‐4 (clCRY4)/the iron–sulfur cluster assembly 1 homolog (clISCA1) complex of Columba livia proposed by the previous study ¹²

Most proteins are diamagnetic substances, where peptide bonds and aromatic residues are the main sources of a protein's diamagnetic anisotropy. ¹⁴ Additionally, the anisotropic oligomerization of protein molecules, such as columnarization and crystallization, is required for their magnetic orientation, ¹⁵ since the diamagnetic energy of protein caused by the diamagnetic anisotropy must exceed the thermal energy (larger than kT) of disorienting Brownian motion. ¹⁶ For example, the magnetic orientations of collagen fibrils ¹⁷ , ¹⁸ and fibrin polymers ¹⁹ require their huge anisotropic (fibriform or columnar) polymers composed of ~10 ⁷ protein molecules and magnetic fields stronger than 1 T. On the other hand, a 24 nm clCRY4/clISCA1 oligomer containing a columnar clISCA1 oligomer core composed of ~20 clISCA1 monomers tends to be orientated by weak magnetic fields (40 μT – 1 mT) under blue light. ¹² This magnetic orientation mechanism of clCRY4/clISCA1 oligomer and its biological function are one of the most important controversial topics in recent research of biomagnetoreception. ²⁰ , ²¹ , ²² , ²³ , ²⁴

However, effect of the magnetic field on the structure and molecular behavior of clISCA1 has not been elucidated, making it challenging to discern the magnetic orientation mechanism of the clCRY4/clISCA1 oligomer and the function of clISCA1 on biomagnetoreception. ²⁰ , ²¹ , ²² , ²³ , ²⁴ In this article, we report the magnetic field effects on the structure and molecular behavior of clISCA1 clarified by small‐angle X‐ray scattering (SAXS) analyses coupled with the SEC‐SAXS and an Nd–Fe–B permanent magnetic circuit device. ²⁵ , ²⁶ Here, a strong magnetic field (~1 T) was utilized to clearly detect the magnetic field effects on clISCA1 in the same manner as the magnetic orientation studies of some other biomacromolecules. ¹⁷ , ¹⁸ , ¹⁹ , ²⁵ , ²⁶

2. RESULTS AND DISCUSSION

2.1. Structural properties of clISCA1

The SEC‐SAXS elution profile derived from forward X‐ray scattering I(0) and UV absorbance at 280 nm (A ₂₈₀) of clISCA1 showed a broad peak at elution numbers 240–340, corresponding to ~1.7 ml of the elution volume (Figure 2a). Here, the OLIGOMER program ²⁷ calculated the volume fractions of eluted components in the clISAC1 solution so that the discrepancies (chi‐squared values) were minimized between the experimental scattering curves I(q)s obtained from the SAXS measurement and the theoretical scattering curves I _m (q)s calculated from model structures of clISCA1 (see Sections 4.2 and 4.3 and Figure S1a–d). The I _m (q)s showed a good fit for I(q)s at elution numbers 270–330 (Figure 2b); the chi‐squared values between I _m (q)s and I(q)s were <1.0. The result indicated that the clISCA1 solution contains monomer, dimer, tetramer, and larger columnar oligomers. These components consisted of two types of protomers (Figure 2c): Type‐A and Type‐B (Figure S1b). The Type‐A protomer of clISCA1 has a similar structure to the globular protomers in clISCA1 homologs (Figure 3), such as chains A and B of PDB ID: 2D2A (RMSD for Cα atoms: 1.8 ± 0.1 Å), ²⁸ chains A and B of PDB ID: 1R94 (RMSD for Cα atoms: 1.8 ± 0.1 Å), ²⁹ and chains A and C of PDB ID: 1X0G (RMSD for Cα atoms: 2.0 ± 0.1 Å). ³⁰ The Type‐B protomer of clISCA1 has a similar structure to the rod‐like protomer in the clISCA1 homolog, such as chains B and D of PDB ID: 1X0G (RMSD for Cα atoms: 1.9 ± 0.2 Å). The Type‐A monomer was eluted over the entire elution peak (nos. 270–330) as shown in Figure 2c. The oligomers composed of Type‐A protomers were primarily eluted in the first half of the elution peak (nos. 270–284 for the Type‐A dimer and nos. 270–302 for Type‐A tetramers or larger oligomers). The Type‐B dimer was eluted over a wide range of the elution peak (nos. 270–274 and nos. 278–324). The Type‐A/Type‐B complex and Type‐B oligomers larger than a dimer were not eluted, suggesting that the Type‐B protomer inhibits the oligomerization of clISCA1. The average volume fractions of eluted components at elution numbers 270–330 were 63% for the Type‐A monomer, 15% for the Type‐B monomer, 5% for the Type‐A dimer, 12% for the Type‐B dimer and 5% for the Type‐A tetramer or more. Thus, the volume fraction of the Type‐A oligomer (dimer or more) contained at elution numbers 270–330 was calculated to be 10%.

FIGURE 2.

SEC‐SAXS analysis of clISCA1. (a) SEC‐SAXS elution profile of clISCA1. Vertical dashed lines indicate each elution fraction corresponding to the I(q)s in (b). (b) The experimental I(q)s at elution numbers 270–330 (dots) of (a) and the theoretical Im(q)s estimated by oligomer analysis (lines). (c) Volume fractions of the eluted components at elution numbers 270–330 estimated by the oligomer analysis of SEC‐SAXS data

FIGURE 3.

Sequence alignment of clISCA1 and its homologs. The chain IDs are shown in parentheses. Sequence homologies are highlighted in red; sequence identities are shown as white letters on a red background box. The locations of the secondary structures are schematically shown above and below the sequences. The α‐helices and β‐strands are shown as arrows and coils, respectively. Strict beta turns are shown as TT letters, and Cys residues are indicated by blue stars. 1X0G forms two types of protomer structures, Type‐A and Type‐B (see the main text)

The UV–vis absorption spectra of clISCA1 were acquired simultaneously with the SEC‐SAXS measurement (Figure 4a). The absorption peaks at 330 and 420 nm that reflected the existence of the Fe–S cluster ¹² disappeared when Fe–S clusters were removed by dialysis after the addition of an iron chelating agent (0.1 M EDTA). These absorption peaks were obviously higher at elution numbers 270–290 than at 295–330. For example, the difference in the UV absorption spectra obtained by subtracting the spectrum of the Fe–S cluster unbound form of clISCA1 (apo‐clISCA1) from spectra of elution numbers 270–330 indicates that the peak heights at 330 and 420 nm of elution number 270 were 4.4 and 2.7 times higher than those at elution number 305 (Figure 4b), respectively. This result suggested that the clISCA1 components in elution number 270 have a higher avidity to Fe–S clusters than those in elution number 305 by 2.7 times or more. Taken together with the result of oligomer analysis (Figure 2c), the UV–vis absorption spectra indicated that the clISCA1 oligomers composed of the Type‐A protomer have a higher avidity to Fe–S clusters than those of the Type‐A monomer, the Type‐B monomer, and the Type‐B dimer. Therefore, the avidity of clICSA1 for the Fe–S cluster is improved by the oligomerization of the Type‐A protomer, accompanying the increase of the magnetic susceptibility of its oligomer due to the magnetism of Fe atoms.

FIGURE 4.

UV–vis absorption spectra of clISCA1 during the elution from the SEC corresponding to elution numbers 270–330 from SEC‐SAXS. (a) UV–vis absorption spectra normalized by the absorption at 280 nm. A black dashed line indicates the spectrum of apo‐clISCA1 solution after the addition of 0.1 M EDTA and dialysis against 20 mM Tris–HCl buffer (pH 8) containing 0.15 M NaCl and 10 mM 3‐mercapto‐1,2‐propanediol to remove Fe atoms. (b) Difference in UV absorption spectra normalized by the absorption at 280 nm. These spectra were obtained by subtracting the spectrum of apo‐clISCA1 (dashed line in (a)) from spectra of elution numbers 270–330

2.2. Magnetic response of clISCA1

To inspect the possibility of the magnetic orientation of clISCA1 oligomer alone, the anisotropies of SAXS data collected at sample position numbers 4 and 5 in the magnetic circuit (Figure 5a,b) ²⁵ , ²⁶ were analyzed with the anisotropic masks for 2D‐SAXS data processing; the mask [h] and the mask [v] were used to extract the scattering data of the horizontal zone and the vertical zone around the beam center of the 2D‐SAXS data (see Section 4.4 and Figure S2). The extracted 2D‐SAXS data were converted to 1D‐SAXS data I(q)s by the circle average (Figure 6a). As shown in Figure 6a, I(q)s at q < 0.1 Å⁻¹ obtained using the mask [h] and the mask [v] were very similar for both sample position numbers 4 and 5 (chi‐squared values were < 0.7), despite that a static magnetic field of about 1 T was applied to the clISCA1 solutions for 60 min. Thus, X‐ray scattering was isotropic at sample position numbers 4 and 5. This result indicated that the clISCA1 oligomer did not orient along the external static magnetic fields.

FIGURE 5.

The periodic Nd–Fe–B permanent magnetic circuit used for the SAXS measurement. (a) Appearance of the magnetic circuit. y‐axis indicates the optical axis of the incident X‐ray beam. (b) The periodicity of magnetic flux density (B) and its gradient (ΔB) in the magnetic circuit, and the schematic graph of the translational force Ftrans for the particles in the sample solution induced by the magnetic field gradient along the x‐axis. Blue arrows show the direction of the magnetic field, which is parallel to the z‐axis at sample position numbers 4 and 6 and to the y‐axis (vertical direction to the paper) at numbers 3, 5, and 7. Continuous and broken arrows colored in red show the directions of Ftrans for the diamagnetic and ferromagnetic particles, respectively

FIGURE 6.

Magnetic response of clISCA1. (a) Anisotropic analysis of I(q)s of 29.3 mg ml⁻¹ of clISCA1 solution. I(q)s obtained using the mask [h] and the mask [v] (Figure S2) are shown by continuous lines and open circles, respectively. I(q)s obtained without applying the external magnetic field (0 min) are colored in black. I(q)s at 60 min of the retention time at sample position numbers 4 and 5 in the magnetic circuit are colored in blue and red, respectively. The enlarged view shows the q < 0.1 Å−1 region. (b and c) I(q)s of 29.3 mg ml⁻¹ clISCA1 solution obtained at sample position numbers 4(b) and 5(c) using the mask [n] shown in Figure S2. The insets show the enlarged views of the q < 0.1 Å−1 region. (d and e) P(r) functions obtained from I(q)s of (b) and (c), respectively. The insets show Rg, Rc, and Dmax of the clISCA1 particles evaluated from I(q)s and P(r)s at sample position numbers 4(b and d) and 5(c and e)

Conversely, the scattering intensities at q < 0.01 Å⁻¹ of I(q)s obtained using the normal mask [n] (Figure S2) increased with the retention time of the clISCA1 solutions at sample position numbers 4 and 5 in the magnetic circuit (Figure 6b,c). Since this change in I(q) was not due to protein aggregation by radiation damage (Figure S3a,b), the result indicated that the oligomerization of clISCA1 molecules was magnetically induced. I(q)s were subsequently converted to the distance distribution functions (P(r)s), which reflect the average shape of the particle in the sample solution in real space (see Section 4.5). The profile of P(r) was changed from a nearly bell‐shaped peak (0 min) to a tailed peak (60 min) according to retention time of the clISCA1 solution in the magnetic circuit (Figure 6d,e), suggesting that the particles in the clISCA1 solution changed from a nearly globular shape to an anisotropic (columnar) shape. The average radius of gyration (R _g ), evaluated by Guinier analysis, increased from 47 to 55 Å at position 4 (Figure 6d) and from 47 to 58 Å at position 5 (Figure 6e). The maximum diameter (D _max) of the particles in the clISCA1 solution, which was evaluated from the real space distance r at P(r) = 0, increased from 158 to 191 Å at position 4 (Figure 6d) and from 158 to 212 Å at position 5 (Figure 6e). Conversely, the average cross‐sectional radius of gyration (R _c ) during 60 min of retention time in the magnetic circuit was almost constant (18 ± 1 Å) at positions 4 and 5 (Figure 6d,e). These results indicated that the cross‐section diameter of the columnar clISCA1 oligomer does not change, but the columnar oligomer's long axis elongates when the clISCA1 solution is held in the magnetic circuit. Since the Type‐B protomer assembles only up to a dimer (Figure 2c), and the Fe–S clusters bind to the columnar oligomer formed by the Type‐A protomers (Figure 4a), the elongation of the clISCA1 oligomer through the application of an external static magnetic field should be due to the association of the Type‐A protomers.

In the magnetic circuit (Figure 5a,b), the ferromagnetic, ferrimagnetic, paramagnetic, and antiferromagnetic particles at sample position numbers 3–5 move toward 4. ²⁵ , ²⁶ Conversely, the diamagnetic particles at sample position numbers 4–6 move toward 5. This was because the magnetic flux density B along the x‐axis in the magnetic circuit was at its maximum at position 4 and minimum at position 5 (Figure 5b), and the particles having negative and positive susceptibility move toward lower B and higher B, respectively (see Section 4.4). Therefore, this result indicated that clISCA1 molecules gathered at position 5 are in the diamagnetic state, whereas those gathered at 4 are in other magnetic states. These magnetic states and the translational force (F _trans) along the x‐axis in the magnetic circuit [proportionality expression (3)] for clISCA1 should be changed by the 2Fe–2S clusters bound to clISCA1, as the magnetic susceptibility of Fe (1.59 × 10⁵ in CGS units) ³¹ is much greater than that of protein molecule (e.g., −0.826 × 10⁻⁶ for bovine serum albumin in CGS units). ³² Moreover, the 2Fe–2S cluster becomes a paramagnetic state (spin quantum number S = 1/2) when it is in the reduced state [2Fe–2S]¹⁺ (containing Fe(III) and Fe(II)) and a diamagnetic state (S = 0) when it is in the oxidized state [2Fe–2S]²⁺ (containing two Fe(III)s). ³³ Thus, the [2Fe–2S]¹⁺‐bound clISCA1 would gather at position 4, and the [2Fe–2S]²⁺‐bound clISCA1 would gather at position 5. It was concluded that the external static magnetic field could control the translational diffusion of clISCA1 and identify the redox state of 2Fe–2S clusters bound to clISCA1. The magnetically altered translational diffusion induces the molecular condensation of Type‐A clISCA1 and the conformation‐selective (Type‐A) self‐oligomerization.

2.3. Fe–S cluster binding mechanism of clISCA1

Figure 7 schematically summarizes the suggested functional mechanism of clISCA1. When ISCA1 interacts with the ISCA1 homolog, ISCA2, ISCA1 acts as a 4Fe–4S cluster carrier protein in the mitochondrial Fe–S cluster biosynthesis machinery. ³⁴ In this machinery, the conformational changeability of the clISCA1 protomer between Type‐A and Type‐B accompanied by the change in affinity for Fe–S clusters may be required for the catch‐and‐release of Fe–S cluster upon 4Fe–4S cluster transportation. Conversely, clISCA1 retains 2Fe–2S clusters when it oligomerizes itself or interacts with clCRY4. ¹² , ²¹ The 2Fe–2S cluster or 4Fe–4S cluster binding site on protein molecules is generally formed by side chains of four Cys residues. ³⁵ , ³⁶ Since one clISCA1 molecule has only three Cys (Cys60, Cys124, and Cys126) as shown in Figure 3, the Fe–S cluster binding site formation on clISCA1 needs the assembly of more than two clISCA1 molecules. Therefore, the Type‐A selective self‐oligomerization to generate a columnar oligomer should accompany the gathering of four Cys residues at least at the interface between ISCA1 molecules.

FIGURE 7.

Functional mechanism of clISCA1. Green arrows show the Fe–S cluster transportation process proposed in the mitochondrial Fe–S cluster biosynthesis machinery. ^{11 , 34} Red arrows show the clISCA1 self‐oligomerization and the clCRY4/clISCA1 complexation process, which may contribute to the magnetoreception of the pigeon. (a) Magnetic self‐oligomerization of Type‐A clISCA1. (b) The hypothetical solenoid model of the clCRY4/Type‐A clISCA1 oligomer. The white coil shows a conceptual schematic drawing of the solenoid. Gray arrows are the external magnetic field lines. CRYR, CRY receptor candidate; ISCA2, iron–sulfur cluster assembly 2 homolog; MFE, magnetic field effect; NFU1, iron–sulfur cluster scaffold; NUBPL, nucleotide‐binding protein‐like. The structure of ISCA2 bound form of clISCA1 (TypeX) is unknown

Moreover, the concentrations of clISCA1 molecule and Fe atom in the elution peak of SEC‐SAXS (Figure 2a) were evaluated using the UV–vis absorption measurement and the inductively coupled plasma‐mass spectrometry (ICP‐MS) (see Section 4.6), respectively. The result indicated that the elution peak (nos. 270–330) of SEC‐SAXS contained 2.7 mg  ml⁻¹ (181 μM) clISCA1 molecules and 0.9 μg ml⁻¹ (16 μM) Fe atoms. This means that the molar ratio between clISCA1 molecules and Fe atoms was 100:9. Since the volume fraction of the 2Fe–2S cluster bindable Type‐A oligomer (dimer or more) in the elution peak at nos. 270–330 was 10% (see Section 2.1), the molar ratio between clISCA1 monomers in the Type‐A oligomer and Fe atoms was evaluated to be 10:9. This result suggests that approximately one 2Fe–2S cluster binds per one clISCA1 dimer in the Type‐A oligomer (Figure 7a).

The formation of 2Fe–2S cluster binding sites between clISCA1 molecules in its oligomer form should also allow spatially regular arrangement of many 2Fe–2S clusters along the columnar oligomer's long axis (Figure 7a), which may improve the dipole magnetic moment and the magnetic anisotropy of the columnar oligomer. Moreover, a previous study indicated that the interaction between clCRY4 and clISCA1 is nearly eliminated when the Fe–S cluster binding ability of clISCA1 is lost by the C60A/C124A/C126A mutation. ¹² As the release of Fe–S clusters occurs cooperatively with the conformational change of clISCA1 from Type‐A to Type‐B (Figures 2c and 4a,b), it can be considered that the clCRY4/clISCA1 interaction needs the oligomerization of the 2Fe–2S cluster‐bound form of clISCA1, namely, the Type‐A protomer.

2.4. Insight into the magnetic orientation mechanism of clCRY4/clISCA1 oligomer

The mechanism of magnetic orientation of the clCRY4/clISCA1 oligomer is in the midst of debate. ²⁰ , ²¹ , ²² , ²³ , ²⁴ For example, the magnetic moment (m _Fe) of a complex of 40 aligned Fe atoms, which may exist in a 24 nm clCRY4/clISCA1 oligomer, was evaluated to be 2.1 × 10⁻²¹ T J⁻¹ in the best case scenario. ²⁴ However, the interaction energy (m _Fe • B _Earth) between m _Fe and the strength of the geomagnetic field B _Earth (25–65 μT) is in the range of 5.3 × 10⁻²⁶ J–1.4 × 10⁻²⁵ J. ²⁴ This interaction energy is five orders of magnitude lower than the thermal energy per degree of freedom in the complex (k _B • T = 4.0 × 10⁻²¹ J, where k _B is the Boltzmann constant) at 293 K; m _Fe • B _Earth/k _B • T is in the range of 1.3 × 10⁻⁵–3.4 × 10⁻⁵. Thus, the magnetic orientation of the clCRY4/clISCA1 oligomer cannot be explained only by the effect of the coordination of the 40 Fe spins. On the other hand, theoretical calculations using the rigid cylindrical model resembling the columnar ISCA1 oligomer suggested that the spin‐mechanical interaction at the atomic scale gave rise to a high blocking temperature that allows a good alignment of the protein's magnetic moment with the geomagnetic field at room temperature. ²⁰ Regardless of which conclusion is correct, it is a fact that the clISCA1 oligomer alone does not orient itself along external static magnetic field lines (Figure 6a). This result suggests that clCRY4 may have an unknown function for enhancing the magnetic orientational property of the clCRY4/clISCA1 oligomer.

The magnetic orientation mechanism of the clCRY4/clISCA1 oligomer could be explained by the oligomer having an iron‐core inserted solenoid‐like structure (Figure 7b). In Figure 7b, the Fe–S clusters arranged along the long axis of the clISCA1 oligomer may act as an iron core for the helically surrounding clCRY4. If the blue light absorption of a FAD ² in clCRY4 causes the helicoidal electron transfer around the iron core of the clCRY4/clISCA1 oligomer, the magnetic moment of the clCRY4/clISCA1 oligomer (m _ol) will be increased in proportion to the helical electric current (I) and the number of the identical turns of clCRY4 (N) as m _ol = I •A•N. Here, A is the vector area of the solenoid, and N increases with elongation of clCRY4/clISCA1 oligomer. The helical electric current of the clCRY4 oligomers may also enhance the attractive interaction between the solenoid of clCRY4 and iron‐core of clISCA1 like a solenoid switch. Moreover, the induced‐fit effect of clISCA1 due to the interaction with clCRY4 may restrict the conformational change from Type‐A to Type‐B (Figure 2c) and the conformational disorder of clISCA1 (Figure S1a), which may improve the rigidity of the clISCA1 columnar oligomer. This solenoid‐like structure may be able to improve the dipole magnetic moment of the clCRY4/clISCA1 oligomer, explaining its magnetic orientation mechanism under blue light.

3. CONCLUSION

The clISCA1 protomer takes the Fe–S cluster‐bound globular form (Type‐A) and unbound rod‐like form (Type‐B). The self‐oligomerization of clISCA1, the clCRY4/clISCA1 interaction, and the magnetic orientation of the clCRY4/clISCA1 complex all require the Type‐A protomer of clISCA1. Moreover, we revealed that the application of the external static magnetic field regulates the translational diffusion of clISCA1 and promotes the Type‐A selective self‐oligomerization of clISCA1. The relationship between the magnetoreception and the magnetic orientation of the clCRY4/clISCA1 oligomer is still unknown. ²⁰ , ²¹ , ²² , ²³ , ²⁴ However, here it was clarified that clCRY4 enhances the magnetic orientational property of clCRY4/clISCA1 oligomer when clISCA1 acts as a scaffold for the clCRY4's photochemical reaction. From the obtained results, the solenoid‐like structure model was proposed to explain the magnetic orientation mechanism of the clCRY4/clISCA1 oligomer under blue light.

The magnetic property of clISCA1 clarified in this study has the potential to be applied to the development of novel technology for molecular manipulation. For example, the oligomerization and spatial arrangement of a clISCA1 fusion protein may be manipulated by an external static magnetic field of about 1 T or more, which may be useful for controlling the local protein concentration in the body and for protein crystallization. Furthermore, it was suggested that the [2Fe–2S]¹⁺ and [2Fe–2S]²⁺ bound forms of clISCA1 oligomer could be selected or identified magnetically by the difference in those translational diffusions. This magnetic selectivity of the redox state of Fe–S cluster could be used to control the Fe–S cluster‐dependent biochemical reactions, such as the maintenance of DNA integrity, regulation of gene expression and protein translation, energy production, antiviral response, and so forth. ³⁷ , ³⁸

It must be noted here that the affinity between CRY and ISCA1 is species‐dependent, even for magnetoreceptive species. For example, it is yet unknown whether the interaction between CRY4 (erCRY4) and ISCA1 (erISCA1) exists in Erithacus rubecula (European robin), whereas dmCRY interacts with D. melanogaster's ISCA1 and erISCA1. ³⁹ Moreover, the photochemistry of erCRY4 is more magnetically sensitive than those of clCRY4 and chicken CRY4, although these animals are magnetoreceptive species. ⁴⁰ For some species, the possibility of other magnetoreception mechanisms remains, such as mechanisms in which ISCA1 homologs assist CRY's function or mechanisms that do not utilize ISCA1.

4. MATERIALS AND METHODS

4.1. Materials

The clISCA1 gene was cloned into a pCold‐ProS2 DNA vector (Takara Bio) and transformed into E. coli BL21 Star (DE3) cells. A thrombin site and a factor Xa site between an HRV3C protease site and the clISCA1 sequence in the pCold‐ProS2 DNA vector were replaced with a GGGS linker. Cells were incubated overnight in lysogeny broth at 37°C. When the optical density at 600 nm reached 0.8, expression of clISCA1 was induced for 24 h at 15°C by adding 1 mM isopropyl β‐d‐1‐thiogalactopyranoside. The cells were lysed in ice‐cold 20 mM Tris–HCl buffer (pH 8.0) containing 0.15 M NaCl and 10 mM 3‐mercapto‐1,2‐propanediol by sonication (SMT UH‐150 sonifier with a 5 mm tip) for 20 min with a 90% pulse. The soluble fraction was obtained by centrifugation at 12,000 rpm for 30 min. The expressed clISCA1 was purified to homogeneity using the Ni Sepharose 6 Fast Flow resin (GE Healthcare, NJ, USA) in the anaerobic chamber ANX‐3 (Hirasawa, Japan) to prevent the oxidation of clISCA1. The His‐tag and ProS2‐tag at the N‐terminus of clISCA1 were subsequently removed by HRV3C protease digestion in the anaerobic chamber for 48 h at 20°C. After HRV3C protease digestion, the GPGGGGS peptide sequence instead of Met1 is located at the N‐terminus of clISCA1 (Figure 3).

The clISCA1 molecule after HRV3C protease digestion was purified by SEC using a Superose 6 Increase 10/300 gl column (GE Healthcare). The purified clISCA1 was stored in 20 mM Tris–HCl buffer (pH 8.0) containing 0.15 M NaCl and 10 mM 3‐mercapto‐1,2‐propanediol at −30°C. Prior to SAXS experiments, the sample was purified again using the same buffer and column.

4.2. Modeling and characterization of clISCA1

The amino acid sequences of clISCA1 and the top three clISCA1 homologs in the PDB [2D2A (SufA from E. coli, sequence identity: 40.0%), 1R94 (ISCA from E. coli, sequence identity: 37.7%), and 1X0G (ISCA from Thermosynechococcus elongatus BP‐1, sequence identity: 29.3%)] were compared to analyze their basic structural characteristics (Figure 3). The number of amino acid residues and theoretical molecular weight of clISCA1 were 138 aa and 14,710 Da, respectively, smaller than those of 2D2A (145 aa and 16,052 Da) and larger than those of 1R94 (118 aa and 12,825 Da) and 1X0G (112 aa and 12,355 Da). clISCA1 has a long N‐terminal region similar to 2D2A. The disorder probabilities of clISCA1 and its homologs were also evaluated by the program SPOT‐Disoder2 (https://sparks‐lab.org/server/spot‐disorder2/) ⁴¹ using the amino acid sequence information (Figure S1a). The intrinsically disordered regions predicted by SPOT‐Disoder2 were G1–Q28 and G131–I138 for clISCA1, M‐6–P26 and G131–V138 for 2D2A, L139–H150 for 1R94, and V142–S145 for 1X0G, indicating that clISCA1 has long flexible regions most similar to 2D2A.

As intrinsically disordered regions generally prevent protein crystallization, and the crystallization and the structure determination of clISCA1 by X‐ray crystallography have not been successful, we built model structures of clISCA1 molecule using the Rosetta program (http://robetta.bakerlab.org/). ⁴² , ⁴³ The models built by the following procedure are summarized in Figure S1b. The arbitrariness of the model building was eliminated by Rosetta Ab Initio modeling, which can generate a model structure from only the amino acid sequence information. The generated clISCA1 model was the globular protomer (Type‐A), similar to chains A and B of 2D2A, chains A and B of 1R94, and chains A and C of 1X0G. The biological unit of 1X0G contains globular protomers (chains A and C) and rod‐like protomers (chains B and D) with the same sequences, suggesting that the ISCA1‐like protein may exhibit a structural polymorphism. As the high structural flexibility of clISCA1 might also allow it to form Type‐A and other structures, the rod‐like clISCA1 protomer model (Type‐B) was also generated by Rosetta comparative modeling with the conformations of chains B and D of 1X0G as template. The conformations of the Type‐A and Type‐B protomer models of clISCA1 were validated via Ramachandran analysis with the Rampage program ⁴⁴ and the Coot program, ⁴⁵ as shown in Figure S1c,d. The Type‐A protomer model had 94.1% of residues in preferred regions, 3.7% of residues in allowed regions, and 2.2% of residues in outlier regions. The Type‐B protomer model had 95.6% of residues in preferred regions, 3.7% of residues in allowed regions, and 0.7% of residues in outlier regions. Finally, the oligomer models of clISCA1 were constructed based on the crystal packing patterns of 1R94 for the Type‐A oligomer and of 1X0G for the Type‐B oligomer and the Type‐A/Type‐B complex (Figure S1b). More specifically, the oligomer models from a dimer to octamer composed of the Type‐A protomer were constructed by imitating the crystal packing pattern of 1R94 (left column in Figure S1b), as clISCA1 forms a columnar oligomer and the crystal packing of 1R94 contains a columnar molecular arrangement similar to that predicted by previous studies. ¹² , ²¹ The oligomer models of a dimer and tetramer composed of the Type‐B protomer model were constructed by imitating the crystal packing pattern of 1X0G (middle column in Figure S1b), as 1X0G contains the rod‐like protomer (chains B and D) similar to Type‐B. The Type‐A/Type‐B complex models composed of both Type‐A and Type‐B protomers were also constructed by imitating the crystal packing pattern of 1X0G (right column in Figure S1b). A total of 12 clISCA1 models were constructed (Figure S1b). Models larger than an octamer were excluded, as those models were difficult to construct due to their structural diversity and flexibility. The form factors [theoretical SAXS curves I _m (q)s] of all models were calculated by the program FFMAKER (https://www.embl‐hamburg.de/biosaxs/manuals/ffmaker.html), which was required for the oligomer analysis.

4.3. SEC‐SAXS experiments

SEC‐SAXS data were collected at BL‐10C at PF ⁴⁶ using the ACQUITY UPLC H‐Class (Waters, MA), Superose 6 Increase 10/300 gl column, and a temperature‐controlled flow‐through cell that allowed UV–vis measurements directly on the SAXS sample. The wavelength used was 1.5 Å. SAXS data were collected using a PILATUS3 2 M detector (Dectris) at 20°C. For SEC‐SAXS measurement, 0.5 ml of 6.9 mg ml⁻¹ clISCA1 was loaded onto the column. Flow rate was set to 0.4 ml min⁻¹ and decreased to 0.05 ml min⁻¹ when the protein eluted from the column to ensure long enough exposure time and better counting statistics in the obtained data. The SEC‐SAXS images corresponding to 20 s of exposure were collected during elution. The SEC‐SAXS data were processed using the program Serial Analyzer ver. 1.3.0. ⁴⁷ Raw data (2D‐SAXS data) were converted to 1D‐SAXS curves by the circular average with a normal mask [n] (Figure S2) to remove the beam stop and other objectionable parts of the images. Buffer scattering was subtracted from each curve to yield the sample scattering curve I(q) using the program SAngler ver. 2.1.33 ⁴⁸ for all SAXS analyses. The I(q)s of elution numbers 270–330 were obtained by averaging the I(q)s of the range, including the two frames before and after, to improve the statistical accuracy of the data. The UV–vis absorption spectra of clISCA1 were also measured simultaneously with the SEC‐SAXS measurements. The SEC‐SAXS data were deposited in the Small Angle Scattering Biological Data Bank (SASBDB) as SASDLD8, SASDLE8, SASDLF8, SASDLG8, SASDLH8, SASDLJ8, and SASDLK8 for elution numbers 270, 280, 290, 300, 310, 320, and 330, respectively.

4.4. Static SAXS experiments with a magnetic circuit

The magnetic response analyses of clISCA1 by SAXS measurements with a periodic Nd–Fe–B permanent magnetic circuit were conducted at BL‐10C at PF (Figure 5a). ²⁵ , ²⁶ In the magnetic circuit, the magnetic flux B and its gradient ΔB along the x‐axis are proportional to $\sin x$ and $\cos x$, respectively (Figure 5b). ²⁵ , ²⁶ Here, x is the sample position on the x‐axis described in radians. The magnetic moment m of a particle can be expressed by Equation (1).

$$m=\chi ·H,$$ (1)

where χ and H are the magnetic susceptibility and the magnetic field strength, respectively. H is proportional to B as Equation (2).

$$B=\mu ·H,$$ (2)

where μ is the permeability. Since the magnetic translational force F _trans is expressed as a product of m and ΔB, F _trans for a particle in a magnetic circuit can be expressed by the proportionality expression (3).

$$F_{\mathit{\text{trans}}}~m·\mathrm{\Delta }B~\chi ·B·\mathrm{\Delta }B~\chi ·\sin x·\cos x$$ (3)

This expression indicates that the direction of particle movement depends on whether a particle's χ value is positive or negative.

The wavelength used for the static SAXS measurements within the magnetic circuit was 1.55 Å. The clISCA1 solution was prepared at higher protein concentrations (29.3 mg ml⁻¹) than the SEC‐SAXS experiments, which is suitable for the observation of the higher ordered oligomerization of protein molecules. ⁴⁹ The clISCA1 solution was contained in a quartz cell with a 2 mm path length. The time‐resolved measurements were conducted at 20°C with retention times of 0, 5, 10, 30, and 60 min at sample position numbers 4 and 5 in the magnetic circuit (Figure 5b). The X‐ray exposure time in each frame was 1 min with a total X‐ray exposure time for one sample of 5 min. The obtained 2D‐SAXS data were converted to 1D‐SAXS curves by the circular average with three different types of masks ([n], [h], and [v] in Figure S2). After subtracting the buffer scattering, I(q)s obtained from two different sample‐to‐detector distances (0.3 and 3 m) were merged using the program ATSAS 3.0.3. ⁵⁰ The I(q)s values were further converted to P(r)s (Figure 6d,e) to inspect the retention time dependency of clISCA1 structure in the magnetic circuit (see Section 4.5). The SAXS data of clISCA1 obtained with the magnetic circuit were deposited in the SASBDB as SASDL68, SASDL78, SASDL88, SASDL98, SASDLA8, SASDLL8, SASDLB8, and SASDLC8.

4.5. SAXS data analysis

The volume fractions of the components (the clISCA1 protomer and oligomer models generated by the program Rosetta) in the sample solutions were estimated by the program OLIGOMER. ²⁷ The scattering curve I(q) from a mixture of different components is written as:

$$I\left(q\right)=\sum \left\{w_i·I_i\left(q\right)\right\},$$ (4)

where w _i and I _i (q) are the volume fraction and the scattering curve from the ith component. q is the magnitude of the scattering vector defined by the following equation:

$$q=\left(4\pi /\lambda \right)\sin \left(\theta /2\right),$$ (5)

where θ and λ are the scattering angle and the wavelength, respectively. Given the theoretical SAXS curve I _m (q) (form factors) from the model structures of the components (Figure S1b), OLIGOMER found the volume fractions by solving a system of linear equations using the algorithm of nonnegative or unconstrained least‐squares to minimize the discrepancy between I _m (q) and the experimental SAXS curve I(q) by the chi‐squared test.

The average radius of gyration (R _g ) and the average cross‐sectional radius of gyration (R _c ) were evaluated from I(q)s with Guinier analysis using the programs ATSAS 3.0.3 ⁵⁰ and SCÅTTER. ⁵¹ I(q) can be approximated at the Guinier region (q • R _g  < 1.3) of the scattering curves as follows ⁵² :

$$I\left(q\right)=I\left(0\right)\mathit{exp}\left(-R_g^2{q}^2/3\right),$$ (6)

where I(0) denotes the intensity at q = 0. For rod‐like particles, such as the Type‐A clISCA1 columnar oligomer, the scattering curves have an approximate form as follows:

$$I\left(q\right)=I_c\left(q\right)/q,$$ (7)

where I _c (q) denotes the scattering curve of the cross‐section of the particle. The Guinier region of I _c (q) can be approximated as follows:

$$I_c\left(q\right)=I_c\left(0\right)\mathit{exp}\left(-R_c^2{q}^2/2\right).$$ (8)

The R _c and I _c (0) values can be evaluated from a linear fit to the cross‐sectional Guinier plot (ln{q • I(q)} vs. q ² plot).

Moreover, I(q)s were further converted to the distance distribution function P(r)s by Fourier transform using the program SCÅTTER ⁵¹ and the following Equation (9).

$$P\left(r\right)=\frac{2}{\pi }{\int }_0^{\infty }\mathit{rqI}\left(q\right)\sin \left(\mathit{rq}\right)\mathit{dq},$$ (9)

where r is the distance in real space. The P(r) function depends on the averaged particle shape and on the intraparticle scattering distribution. In this study, P(r)s for the magnetic response analyses of clISCA1 were calculated from I(q)s at q = 0.016–0.518 Å⁻¹ using the Legendre method in the program SCÅTTER, which is relatively accurate for highly elongated particles. ⁵¹ The maximum diameter (D _max) of the particle was estimated from the P(r) function satisfying the condition P(r) = 0 for r > D _max.

4.6. ICP‐MS analysis

Concentration of Fe atoms in the elution peak (nos. 270–330) of SEC‐SAXS (Figure 2a) was evaluated by ICP‐MS. As a sample for ICP‐MS, an eluate containing 2.7 mg ml⁻¹ clISCA1 (3.0 ml in total) was collected from elution numbers 270–330 of three SEC‐SAXS measurements. The collected elute was concentrated to 9.2 mg ml⁻¹ clISCA1 by a Millipore centrifugation filter tube (Amicon Ultra, 10 kDa). Then, 0.3 ml of a concentrated sample was diluted with 10% trace‐metal grade nitric acid to a final volume of 50 ml and was completely digested in the Ethos EZ microwave unit (Milestone). Digests were analyzed for trace Fe atoms on the Agilent 8,800 ICP‐MS (Agilent Technologies).

CONFLICT OF INTEREST

The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

Shigeki Arai: Conceptualization (lead); data curation (lead); formal analysis (lead); funding acquisition (lead); investigation (lead); methodology (equal); project administration (lead); resources (lead); supervision (lead); validation (lead); visualization (lead); writing – original draft (lead); writing – review and editing (lead). Rumi Shimizu: Investigation (supporting). Motoyasu Adachi: Funding acquisition (supporting); investigation (supporting). Mitsuhiro Hirai: Data curation (supporting); investigation (supporting); methodology (equal).

Supporting information

FIGURE S1 Modeling of clISCA1. (a) Disorder probabilities of clISCA1 and its homologs predicted by the program SPOT‐Disorder2. Closed and open circles indicate the ordered and disordered regions, respectively. (b) clISCA1 models used for oligomer analysis. Ribbon models show the Type‐A protomer constructed by Rosetta Ab Initio modeling and the Type‐B protomer constructed by Rosetta comparative modeling. Circles and rods show the schematic drawings of the Type‐A protomer and the Type‐B protomer, respectively. The numbers in angstroms represent the theoretical gyration radii R _g s of the protomers. (c and d) The Ramachandran plots of the Type‐A protomer model (c) and the Type‐B protomer model (d).

FIGURE S2 Masks for the 2D‐SAXS data processing. The green zone indicates the removal zone to calculate 1D‐SAXS curves. [n] is the normal mask. [h] and [v] are masks to extract the scattering data of the horizontal zone and the vertical zone around the beam center, respectively.

FIGURE S3 Radiation resistivity of clISCA1. (a and b) X‐ray exposure time dependency of I(q)s of a 29.3 mg ml⁻¹ clISCA1 solution (a) and a 6.9 mg ml⁻¹ clISCA1 solution (b). SAXS data were collected at BL‐10C beamline at PF without the magnetic circuit. The clISCA1 solutions were contained in a quartz cell with a 1 mm path length. The wavelength of the incident X‐ray beam was 1.55 Å. The X‐ray exposure time in each frame was 1 min with a total X‐ray exposure time of 5 min. SAXS data of (a) were collected over a wide q‐range (q = 0.006–2.1 Å⁻¹) by merging the scattering data obtained from two different sample‐to‐detector distances (0.3 and 3 m), and those of (b) were collected in the q‐range between 0.006 and 0.2 Å⁻¹. I(q)s were obtained by the subtraction of buffer scattering from 1D‐SAXS data after the circular average of 2D‐SAXS data with a normal mask [n] (Figure S2). In both (a) and (b), I(q)s were almost constant during X‐ray exposure for 5 min; the chi‐squared values between each I(q) were <0.99 for (a) and <0.47 for (b). This result indicates that an X‐ray exposure time for 5 min was short enough to avoid radiation damage to clISCA1. These SAXS data were deposited in the SASBDB as SASDL58, SASDLT7, SASDLU7, SASDLV7, SASDLW7, and SASDLX7 for 29.3 mg ml⁻¹ clISCA1 (a), and SASDLY7, SASDLZ7, SASDL28, SASDL38, and SASDL48 for 6.9 mg ml⁻¹ clISCA1 (b).

Arai S, Shimizu R, Adachi M, Hirai M. Magnetic field effects on the structure and molecular behavior of pigeon iron–sulfur protein. Protein Science. 2022;31(6):e4313. 10.1002/pro.4313