Blue light–excited LOV1 and LOV2 domains cooperatively regulate the kinase activity of full-length phototropin2 from Arabidopsis

Phototropin2 (phot2) is a blue-light (BL) receptor that regulates BL-dependent activities for efficient photosynthesis in plants. phot2 comprises two BL-receiving light-oxygen-voltage–sensing domains (LOV1 and LOV2) and a kinase domain. BL-excited LOV2 is thought to be primarily responsible for the BL-dependent activation of the kinase. However, the molecular mechanisms by which small BL-induced conformational changes in the LOV2 domain are transmitted to the kinase remain unclear. Here, we used full-length wild-type and mutant phot2 proteins from Arabidopsis to study their molecular properties in the dark and under BL irradiation. Phosphorylation assays and absorption measurements indicated that the LOV1 domain assists the thermal relaxation of BL-excited LOV2 and vice versa. Using small-angle X-ray scattering and electron microscopy, we observed that phot2 forms a dimer and has a rod shape with a maximum length of 188 Å and a radius of gyration of 44 Å. Under BL, phot2 displayed large conformational changes that bent the rod shape. By superimposing the crystal structures of the LOV1 dimer, LOV2, and a homology model of the kinase to the observed changes, we inferred that the BL-dependent change consisted of positional shifts of both LOV2 and the kinase relative to LOV1. Furthermore, phot2 mutants lacking the photocycle in LOV1 or LOV2 still exhibited conformational changes under BL, suggesting that LOV1 and LOV2 cooperatively contribute to the conformational changes that activate the kinase. These results suggest that BL-activated LOV1 contributes to the kinase activity of phot2. We discuss the possible intramolecular interactions and signaling mechanisms in phot2.


Phototropin2 (phot2) is a blue-light (BL) receptor that regulates BL-dependent activities for efficient photosynthesis in plants. phot2 comprises two BL-receiving light-oxygenvoltage-sensing domains (LOV1 and LOV2
) and a kinase domain. BL-excited LOV2 is thought to be primarily responsible for the BL-dependent activation of the kinase. However, the molecular mechanisms by which small BL-induced conformational changes in the LOV2 domain are transmitted to the kinase remain unclear. Here, we used full-length wild-type and mutant phot2 proteins from Arabidopsis to study their molecular properties in the dark and under BL irradiation. Phosphorylation assays and absorption measurements indicated that the LOV1 domain assists the thermal relaxation of BL-excited LOV2 and vice versa. Using small-angle X-ray scattering and electron microscopy, we observed that phot2 forms a dimer and has a rod shape with a maximum length of 188 Å and a radius of gyration of 44 Å . Under BL, phot2 displayed large conformational changes that bent the rod shape. By superimposing the crystal structures of the LOV1 dimer, LOV2, and a homology model of the kinase to the observed changes, we inferred that the BL-dependent change consisted of positional shifts of both LOV2 and the kinase relative to LOV1. Furthermore, phot2 mutants lacking the photocycle in LOV1 or LOV2 still exhibited conformational changes under BL, suggesting that LOV1 and LOV2 cooperatively contribute to the conformational changes that activate the kinase. These results suggest that BL-activated LOV1 contributes to the kinase activity of phot2. We discuss the possible intramolecular interactions and signaling mechanisms in phot2.
Phototropin (phot) 2 is a blue-light (BL) receptor protein diversely present in green algae to land plants and is involved in the adaptation of organisms for efficient photosynthesis (1). In Chlamydomonas reinhardtii (Cr), phot mediates feedback regulation of photosynthesis through gene expression (2). In land plants, phototropism (3,4), chloroplast movement (5,6), stomata opening (7), and leaf expansion (8) are known as photomediated responses. Most higher plants have two isoforms of phot, designated as phot1 and phot2. phot1 and phot2 are activated by BL redundantly to regulate stomata opening, phototropism, and the accumulation of chloroplasts. The avoidance response of chloroplasts under high-fluence irradiation is mediated by only phot2 (5). phot1 and phot2 have different sensitivities to BL intensity; phot1 works in a broad range of BL intensities, whereas phot2 acts as a photoreceptor activated under high-fluence irradiation (4).
Phot proteins comprise ϳ1000 amino acid residues and two flavin mononucleotide (FMN) molecules and fold into three functional domains: two light-oxygen-voltage domains (LOV1 and LOV2) in the N-terminal half and one serine/threonine kinase domain (STK) in the C-terminal half (1) (Fig. 1A). LOV, which belongs to the Per-Arnt-Sim superfamily (9, 10), noncovalently binds one FMN molecule in a pocket formed by a characteristic tertiary structure named the ␣/␤-scaffold (11,12). The STK is classified into group VIII of the AGC protein kinase family and is expected to have a tertiary structure similar to protein kinase A (13).
Upon blue-light excitation, each LOV domain in the dark state (D 450 state) undergoes a characteristic photocycle, and then the FMN transiently forms a covalent bond with a wellconserved cysteine residue in LOV (S 390 state) (14 -16). In phot2, the cysteine residues are Cys-170 in LOV1 and Cys-426 in LOV2. This local structural change induces a reorganization of hydrogen bonds around the chromophore (17). The adduct form thermally reverts to the D 450 state with a time constant from seconds to minutes in the dark (16). The phosphorylation activity of STK is enhanced by the small conformational changes of LOV2 in the S 390 state (18). Therefore, LOV2 acts as the major regulator and forms a minimum functional unit with STK (LOV2-STK) (19,20). Regarding the regulation of STK activity by BL-activated LOV2, unfolding and conformational changes of the J␣-helix, which associates with the C-terminal lateral region of the ␣/␤-scaffold of LOV2, is critical for the intramolecular signaling (21). BL-activated phot displays autophosphorylation (19) and further phosphorylates substrate proteins, such as ATP-binding cassette subfamily B19 (22), phytochrome kinase substrate 4 (23), and blue-light signaling 1 (24), in the downstream part of the signaling pathway. Therefore, phot is a converter of BL stimuli to phosphorylation signals that activate signal transduction cascades, which induce the BL responses of plants.
Our structural study on LOV2-STK of Arabidopsis thaliana (At) phot2 (25) as a minimum functional unit revealed that LOV2 and STK are in a tandem arrangement and display rearrangements under BL irradiation. Similar conformational changes were observed in Crphot (26) and LOV2-STK of Atphot1 (27). In addition, we identified certain amino acid residues that are critical for regulating the STK activity in the AЈ␣/A␤ gap of the N-terminal region of LOV2 and in the linker region connecting LOV2 and STK in Atphot1 (28,29).
The analyses of spectroscopy and kinase assays for LOV2-STKs of both Atphot1 and Atphot2 demonstrated that the fluence-dependent sensitivity of phot correlated with the recovery speed from S 390 to D 450 ; faster recovery resulted in a lower sensitivity, whereas slower recovery resulted in higher sensitivity (30). In Crphot, the truncation of LOV1 induced an acceleration of the recovery of LOV2, suggesting that LOV1 worked as an enhancer for light perception (26). In phot proteins of land plants, LOV1 also contributes to the regulation of LOV2 and the kinase activity (19,31). However, details of the molecular mechanism and intramolecular interactions remain unclear.
In this study, we established a large-scale preparation system to provide full-length Atphot2, which comprised all residues in the amino acid sequence, and characterized the molecular properties of the wild-type protein and its mutants in terms of their kinase activities, dark recovery kinetics, and molecular structures. We report the results of biochemical, spectroscopic, and structural studies using small-angle X-ray scattering (SAXS) and electron microscopy (EM) and discuss the regulatory mechanism of Atphot2.

Preparation of phot2
Wild-type Atphot2 (WT) was overexpressed and prepared at a purity greater than 95% as examined by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) (Fig.  S1A). In addition, three mutants (C170A, C426A, and D720N) were prepared for biochemical assays. The C170A (C426A) mutant can bind one FMN molecule to the pocket of the ␣/␤-scaffold of LOV1 (LOV2) but has lost the ability to form the S 390 state in the photoreaction cycles of LOV1 (LOV2). D720N mutation results in the complete loss of the kinase activity of STK.
By size-exclusion column chromatography (SEC), the apparent molecular weight of the WT was estimated as ϳ220,000, indicating a dimeric form of WT in solution (Fig. S1B). The dark-adapted WT (D 450 state) had absorption peaks at 474, 446, and 371 nm and a shoulder around 410 nm consistent with those of the LOV1-LOV2 fragment (19). The absorbance values at 273 and 446 nm in the dark state indicated that each subunit was associated with two FMN molecules. Under BL irradiation, the WT showed absorption changes of the LOV domains ( Fig.  1B) as observed for the LOV1-LOV2 fragment (16).

Kinase activity and phosphorylation sites of phot2
The kinase activity of the purified WT was examined by an autophosphorylation assay with and without BL irradiation (Fig. 1C). The WT showed phosphorylation activity in the dark as observed both in vivo and in vitro (3,19), and the kinase activity was significantly enhanced under BL. As expected, the D720A mutant displayed little phosphorylation. The C170A mutant showed BL-accelerated autophosphorylation activity as observed in the WT, whereas the C426A mutant displayed little autophosphorylation under BL.
In a gel-shift assay, the WT, just after purification, had an apparent molecular weight larger than those of the WT incubated with -protein phosphatase (PPase; New England Biolabs) and the D720N mutant (Fig. 1D). This result implied that the purified WT would be phosphorylated during overexpression and purification. In fact, 25 phosphorylation sites were identified in the previous study of the in vivo autophosphorylation of Atphot2 (32).
Autophosphorylation sites in the N-terminal half of PPasetreated phot2 in the dark and under BL irradiation were identified by mass spectrometry (MS) (Fig. 1E). No phosphorylation sites were detected in the N-terminal half of the D720N mutant used as a reference. The PPase treatment dephosphorylated the WT except at Ser-284 located near the C terminus of LOV1. After incubation in the presence of ATP in the dark, PPasetreated WT was phosphorylated at Ser-22. After BL irradiation for 1 h, an additional phosphorylation was identified at Ser-106, suggesting a BL-dependent conformational change around the N-terminal region of LOV1.

Photoconversion to S 390 and dark reversion of C170A/D720N, C426A/D720N, and C170A/C426A/D720N mutants
In addition to the three single mutants described in the previous sections, two double mutants (C170A/D720N and C426A/D720N) and one triple mutant (C170A/C426A/ D720N) were overexpressed and purified to study the influence of the photocycle of each LOV domain on the BL responses of phot2 (Fig. S1C). These mutants lack the kinase activity and are free from heterogeneity regarding phosphorylation, making them suitable for physicochemical measurements requiring homogeneity of specimens.
The conversion from D 450 to S 390 and the dark reversion of S 390 to D 450 were measured by ultraviolet (UV)-visible absorp-LOV1 and -2 cooperatively activate Arabidopsis phototropin2 tion spectroscopy. The mutants were measured using SAXS as described below; therefore, they were suspended in the same buffer used in the SAXS experiments. In phot, the formation of S 390 in LOV2 is essential (19), and the dark-reversion speed of S 390 to D 450 is one of the key factors for the activation of the kinase function (30). Both LOVs in the D720N mutant were almost completely converted to their S 390 state under BL fluence at more than 450 mol m Ϫ2 s Ϫ1 ( Fig. 2A). BL irradiation at a fluence of 450 mol m Ϫ2 s Ϫ1 converted more than 90% of C170A/D720N and C426A/D720N mutants to the S 390 states of LOV2 and LOV1, respectively (Fig. 2, B and C). The triple mutant, prepared as a reference for the other three mutants, displayed little change in its absorption spectrum under BL irradiation (Fig. 2D).

LOV1 and -2 cooperatively activate Arabidopsis phototropin2
single-exponential curve with a 1 ⁄ 2 value of 168 s, which was approximately 3 times longer than that of LOV2 in the D720N mutant. These findings suggested a significant influence of LOV1 on the dark reversion of LOV2. In contrast, the dark reversion of LOV1 in the C426A/D720N mutant, which was approximated by a single-exponential curve with a 1 ⁄ 2 value of 68 s, was faster than that of the D720N mutant. Mutual influences on the photocycles between LOV1 and LOV2 were suggested by the prolongation of the S 390 lifetime of LOV2 in the C170A mutant and the acceleration of the dark reversion of LOV1 in the C426A mutant.

Molecular structure and BL-induced conformational changes of the D720N mutant
The low-resolution structure and BL-induced structural change of phot2 in solution were studied by SAXS for the D720N mutant because the heterogeneity of the phosphorylated WT and, probably, the phosphatase-treated WT was unsuitable for SAXS measurements. For each specimen solution, SAXS profiles were recorded in the dark (dark1), under BL at more than 450 mol m Ϫ2 s Ϫ1 (light1), and again in the dark (dark2).
The scattering intensities of light1 increased in S Ͻ 0.007 Å Ϫ1 , whereas the profiles in the other regions remained almost the same (Fig. 3A). The intensity change in light1 at S ϭ 0.005 Å Ϫ1 corresponded to ϳ8% of the intensity of dark1. The profiles of dark2 resembled those of dark1, indicating that the BLdependent SAXS changes were almost photoreversible (Fig. 3A, inset). These photoreversible SAXS changes were observed in the concentration range measured thus far.
The Guinier analysis (Fig. 3B) and the concentration dependences of the radius of gyration (R g ) and zero-angle scattering intensities (Fig. 3C) indicated that the specimen solutions were almost monodispersive (Fig. 4C). A dimeric association of the D720N mutant was confirmed from the apparent molecular weight estimated from the zero-angle scattering intensity. This finding was consistent with the result of SEC (Fig. S1B). The R g values of dark1 (44.3 Ϯ 1.4 Å) and dark2 (44.9 Ϯ 1.4 Å) were comparable and slightly larger than that of light1 (43.7 Ϯ 1.4 Å) (Fig. 3C). The R g values were likely photoreversible as expected from the profiles. The distance distribution functions P(r) of dark1, light1, and dark2, with shoulders at r Ͼ 70 Å, reflected their dimeric structures (Fig. 3D). The maximum dimensions were almost the same among dark1 (188 Ϯ 2 Å), light1 (186 Ϯ 2 Å), and dark2 (187 Ϯ 2 Å).
The molecular shapes of dark1, light1, and dark2 were restored by the program GASBOR. The model shapes were multiply restored for each condition, and then the most probable shape was selected after classification of the restored molecular shapes using multivariate analyses (Fig. 4, A and B). The molecular shapes of dark1 and dark2 appeared as elongated Z-shapes with approximate dimensions of 185 Å long and 80 Å wide in the Z-shape view and 55 Å thick in the rod-shape view. The molecular model of light1 appeared with a Z-shape but had a bend around the center in the rod -shape view, indicating BL-induced conformational changes. The bend induced movements of both edges of the rod shape by ϳ30 Å.
Furthermore, the molecular shapes of the negatively stained C170A/C426A/D720N mutant were visualized by transmission electron microscopy. Although negative staining might induce conformational changes of C170A/C426A/D720N, molecular images and dimensions consistent with the SAXS models were observed (Fig. 4C), indicating that the model shapes predicted from the SAXS profiles would approximate the molecular shape of the D720N mutant.

BL-induced SAXS changes in the C170A/D720N and C426A/D720N mutants
Next, to understand which BL-activated LOV domain was responsible for the BL-induced conformational changes in the D720N mutant, SAXS of the C170A/D720N, C426A/D720N, and C170A/C426A/D720N mutants was measured in the dark and under BL irradiation (Fig. 4E). The three mutants were almost monodispersive and displayed profiles similar to that of D720N mutant in the dark (Fig. 3D). The magnitudes of the The inset shows difference SAXS profiles between dark1 and light1 (magenta dots) and between light1 and dark2 (cyan dots). B, Guinier plots, which show the dependence of the logarithm of the scattering intensity on the square of the scattering vector, for dark1, light1, and dark2 of the D720N mutant. For clarity, each plot is shifted along the ordinate. The concentrations from the top to bottom in each state were 1.2, 1.0, and 0.5 mg ml Ϫ1 , respectively. The least-squares fitting method was applied for the data in the region indicated by the arrows to calculate the I(0) and R g assuming the Guinier approximation. The high-resolution edges of the region (S 2 ϭ 15 ϫ 10 Ϫ6 Å 2 ) satisfy the criteria for applying the approximation (SR g Ͻ (2) Ϫ1 ). C, the concentration dependences of C/I(0, C) (upper panel) and R g (C) 2 (lower panel) values of dark1, light1, and dark2. For the C/I(0, C) data, the standard deviations are smaller than the sizes of the symbols. D, P(r) functions of the D720N mutant in the dark1, light1, and dark2. a.u., arbitrary units.

LOV1 and -2 cooperatively activate Arabidopsis phototropin2
differences in the SAXS profiles (dark1 Ϫ light1 and light1 Ϫ dark2) of both the C170A/D720N and C426A/D720N mutants were smaller than those of D720N mutant, even when taking into account the incomplete conversion of each LOV in C170A/D720N and C426A/D720N mutants (ϳ90%). In contrast, the C170A/C426A/D720N mutant showed little change in SAXS and the absorption spectra (Fig. 2D). Therefore, both BL-excited LOV1 and LOV2 would induce the conformational changes detectable by SAXS in the C426A/D720N and C170A/ D720N mutants, respectively. However, the magnitudes in the SAXS differences in the double mutants were smaller than those of the D720N mutant.

Discussion
The biochemical, spectroscopic, and structural studies on purified full-length phot2 and its mutants suggested a contribution of LOV1, as well as LOV2, to the kinase activity and BL-dependent conformational changes of Atphot2. Here, we discuss the implications of the results for the illustration of the molecular organization and intramolecular signaling to induce cellular signaling cascades in the BL response by phot2.

Domain organization and BL-induced conformational changes in Atphot2
The WT and mutants exist as dimers in solution as suggested by SEC, SAXS, and EM (Figs. S1B and 4, B and C). The molecular shape of dark1 (and dark2) of the D720N mutant was approximated by arranging the molecular models of the LOV1 dimer and monomeric LOV2-STK proposed in our previous studies (Fig. 5A). Considering the fact that LOV1 forms a dimer both in the crystal (33) and in solution (34), the LOV1 dimer probably acts as a dimerization site and occupies the central part of the molecular shape. Then the crystal structure of phot2 LOV2 (35) and a homology model of STK (25) were fitted to the edge lobes of the molecular shapes of dark1 and light1.
Under BL, the LOV1 dimer exhibited little conformational change when viewed at a low resolution (34). In addition, LOV2-STK exhibits BL-induced rearrangement of LOV2 relative to STK. Therefore, the global changes in the molecular shape under BL would be mainly attributed to the BL-induced rearrangement of LOV2 and STK observed in Atphot2 LOV2-STK (25) (Fig. 5A). Most of the phosphorylation sites identified in the N-terminal half of Atphot2 (32) are far from the active site of STK in the model of light1; therefore, those sites would be phosphorylated by other phot2 dimers. In contrast, Ser-22, which is located in the edge of the long N-terminal tail composed of ϳ100 amino acid residues (Fig. 1A), might reach the active-site cleft of STK to be self-phosphorylated.
To ensure the homogeneity of specimens in physicochemical measurements, we studied kinase-inactivated mutants. However, phosphorylation may still have certain effects on BL-induced structural changes in wild-type phot2. For investigating the influences of phosphorylation on conformational changes, purification techniques to separate wild-type phot2 at different phosphorylation levels will be necessary in the future.
In crude extract of wild-type phot2 under BL, more than 25 phosphorylation sites were found (32). In contrast, in the in Figure 4. A, the SAXS profiles of the D720N mutant in dark1 (red dots), light1 (blue dots), and dark2 (green dots) states are compared with those calculated from their GASBOR models (black). For clarity, profiles plotted in semilogarithmic form are shifted appropriately along the ordinate. B, the molecular shapes of dark1 (pink mesh), light1 (cyan mesh), and dark2 (green mesh) are illustrated as density maps of dummy residues in 4 ϫ 4 ϫ 4-Å 3 cubes after averaging a set of the most probable models. In the lower panel, the 2-fold symmetry axis is indicated on the model views rotated by 90°from the upper. C, typical images of C170A/ C246A/D720N mutant viewed by EM. D, the SAXS profiles of the C170A/D720N, C426A/D720N, and C170A/C426A/D720N mutants in dark1 (red dots), light1 (blue dots), and dark2 (green dots). E, differences in the SAXS profiles of the three mutants between dark1 and light1 (magenta dots) and between light1 and dark2 (cyan dots).

LOV1 and -2 cooperatively activate Arabidopsis phototropin2
vitro assay (Fig. 1E), PPase-treated WT was phosphorylated at a few sites under BL. This difference suggests that other factors accelerate phosphorylation, for instance a molecular complex with the partners of a kinase cascade in vivo. In the dark, phot2 was dephosphorylated by any phosphatase in vivo (36). Although the SAXS experiments were conducted for D720N mutant without phosphorylation, the structure and the BL-induced conformational changes would be similar to those of phot2 with a small number of phosphorylated residues in vivo. Fig. 5B compares the expected molecular structures of Atphot1, Atphot2, and Crphot (25)(26)(27). Monomeric Crphot is thought to be located near the origin of the molecular evolution of phot (37). In land plants, phot2 is thought to be a prototype of phot and would be an origin of the diverse evolution to phot1 (38). Although Atphot2 forms a dimer by the association of LOV1, both LOV1 and LOV2 are involved in dimer formation in Atphot1. Therefore, the arrangement and association of LOV1 and LOV2 domains might become more complex in the course of evolution, in the order of Crphot, Atphot2, and Atphot1, and would be crucial for the functions of phot1 and phot2. A phot2 mutant with the LOV1 and LOV2 swapped from phot1 does not induce a normal dark positioning, which is a phot2-specific response in Arabidopsis (39). Based on the differences in the association modes between phot1 and phot2 dimers, the swapping may cause changes in the conformation and arrangement of LOV1 and LOV2, making it difficult for the swapped mutant to recognize the substrates, which is necessary to induce the dark positioning.

Functional roles of LOV1 in phot2
Currently, it is believed that for phot2 LOV2 is primarily responsible for the BL-induced conformational changes and that LOV1 only acts as a dimerization site and has little influence on the BL-dependent conformational changes that activate STK. From the present phosphorylation assay for C170A and C426A mutants (Fig. 1E), phosphorylation activity is likely dominated by LOV2. However, influences of LOV1 on the photocycle of LOV2 and the kinase activity have been reported. For instance, the C426A mutant of Atphot2 is reported to have kinase activity in vitro (19) and in vivo (31,40). LOV2-STK of Atphot1 displays a reduction in the fluence-dependent response compared with the full-length protein in vivo (20). In Crphot, LOV1 contributes to increase the sensitivity of the fluence-dependent response (26). The dark reversion of LOV depends on the size of the LOV-containing fragment (16,19).
The present study of the purified full-length phot2 revealed that, at the molecular level, BL-induced small conformational changes in LOV1 would influence the photocycle of LOV2 and vice versa (Fig. 2E). Therefore, LOV1 and probably its N-and C-terminal regions would play a role to produce the long-lived S 390 of LOV2 that is advantageous to sustainably activate STK. In addition, the BL-dependent SAXS changes in C170A/D720N and C426A/D720A mutants were smaller than those of the D720N mutant. This result suggested that BL-induced LOV1 and LOV2 cooperatively contribute to BL-induced SAXS changes in the D720N mutant (Fig. 4E). , and the homology model of STK (N-domain (red) and C-domain (green)) in the molecular shapes (white mesh) of dark1 and light1. The homology model of STK was built from the crystal structure of cAMP-dependent protein kinase in the unliganded state (Protein Data Bank code 1J3H) as a template. In this arrangement, STK in one subunit could not easily contact the other subunit for intersubunit phosphorylation. B, comparison of putative models of Crphot, the Atphot1 dimer expected from the SAXS model for dimeric phot1 LOV2-STK, and the Atphot2 dimer from this study. In the models, the J␣ regions are colored purple. The arrows indicate the side chains of the active-site aspartate residues illustrated as a space-filling model colored purple.

LOV1 and -2 cooperatively activate Arabidopsis phototropin2
The BL-dependent phosphorylation of Ser-106 in PPasetreated WT (Fig. 1E) may provide a clue to understand how the small conformational changes in LOV1 propagate to LOV2. As the N-and C-terminal regions in the ␣/␤-scaffold of LOV1 are in close location, the conformational changes around Ser-106 near the N terminus of LOV1 may be transmitted to the linker region connecting the N and C termini of LOV2. In the case of phot1 LOV2-STK, the N-and C-terminal regions are critical for transmitting small conformational changes in LOV2 to STK (27,28). In the next step to understand the molecular mechanism, it will be necessary to perform mutation analyses for the N-and C-terminal regions of LOV1.
Each overexpression vector was transformed into Escherichia coli Rosetta2 (DE3) cells. Overexpression of the Atphot2 protein was induced by adding isopropyl 1-thio-␤-D-galactopyranoside at a final concentration of 0.05 mM to the culture in the dark. Cells were harvested under a dim red light and stored at Ϫ80°C until purification.

Purification
Purification was performed under a dim red light. Cells were freeze-thawed and suspended in a buffer containing 0.5 M NaCl, 10% (w/v) glycerol, and 20 mM HEPES-NaOH (pH 7.5). A protease inhibitor mixture (Roche Applied Science), Triton X-100 (final concentration of 1% (w/v)), and lysozyme (final concentration of 1 mg ml Ϫ1 ) were then added to the suspension. After incubation on ice for 1 h, cells were lysed by sonication. The crude extract was centrifuged (100,000 ϫ g, 30 min, 4°C), and its supernatant was loaded onto a nickel-affinity column (His-Trap, GE Healthcare). Partially purified protein, which was eluted with a buffer containing 0.5 M imidazole, was purified further by SEC using a HiLoad 26/600 Superdex 200 prep grade (GE Healthcare) column equilibrated by a buffer containing 100 mM NaCl, 10% (w/v) glycerol, and 20 mM Tris-HCl (pH 7.8). Finally, fractions containing Atphot2 were loaded onto an anion-exchange column (Mono Q HR 5/5, GE Healthcare) and purified with a NaCl gradient. Fractions containing Atphot2 were concentrated using ultrafiltration (Amicon Ultra, Merck Millipore). The purity of the specimens was examined using SDS-PAGE after each column chromatography. The apparent molecular weight of Atphot2 was measured by SEC using a HiLoad 16/60 Superdex 200 prep grade (GE Healthcare) column equilibrated by a buffer containing 100 mM NaCl, 10% (w/v) glycerol, and 20 mM Tris-HCl (pH 7.8). For SAXS measurements and EM, Atphot2 was purified further by additional SEC using a buffer containing 20 mM Tris-HCl (pH 7.8), 500 mM NaCl, and 10% (w/v) glycerol.

Kinase assay
Kinase assays were performed as reported previously (41). An Atphot2 solution was incubated for 15 min at 20°C in a kinase reaction buffer containing 100 mM NaCl, 10% (w/v) glycerol, 10 mM MgCl 2 , 10 M [␥-32 P]ATP (3.7 kBq l Ϫ1 ), and 20 mM Tris-HCl (pH 7.8). For blue-light excitation, a blue-lightemitting diode panel with an emission maximum at 475 nm (ISL-150X150-88, CCS Inc., Japan) was used. The reaction was stopped by boiling for 3 min after addition of SDS sample buffer. After SDS-PAGE of the specimens, the polyacrylamide gel was stained with Coomassie Brilliant Blue and dried. Phosphorylated bands were visualized using an imaging plate (Fujifilm) and a scanner (Storm, GE Healthcare). The signal intensities were measured using ImageJ software.

Mass spectrometry
In-gel digestions were performed as described previously (42). An LTQ-Orbitrap XL (Thermo Fisher Scientific) coupled with an EASY-nLC 1000 (Thermo Fisher Scientific) was used for nano-LC-tandem mass spectrometry (MS/MS) analyses as described previously (43). Mass Navigator version 1.3 (Mitsui Knowledge Industry) with the default parameters for the LTQ-Orbitrap XL was used to create peak lists on the basis of the recorded fragmentation spectra as described previously (44). Peptides were identified by using automated database searching with Mascot version 2.3.02 (Matrix Science) in The Arabidopsis Information Resource database (TAIR10_pep_20101214), which contains protein sequence information of His-tagged Atphot2 with a precursor mass tolerance of 3 ppm, a fragment ion mass tolerance of 0.8 Da, and strict trypsin specificity, allowing for up to two missed cleavages. Carbamidomethylation of Cys was set as a fixed modification, and oxidation of Met and phosphorylation of Ser, Thr, and Tyr were allowed as variable modifications. For quantitative analysis, MS peaks were detected using an in-house 2DICAL software package (45) that identified all MS/MS spectra using Mascot software, adjusted the retention times of each LC-MS data point utilizing the similarity index of the mass spectrum pattern, and grouped peaks from different samples that were derived from the same peptides in the direction of acquisition time. The peak intensities of the MS chromatograms were used as quantitative values.

Measurement of UV-visible absorption spectra
Absorption spectra and the time course of absorption changes in the dark reversion from S 390 to D 450 were measured at 293 K using a U-3100 spectrophotometer (Hitachi High-Technologies, Japan). The sample chamber was equipped with a 0305 thermostat controller (Hitachi High-Technologies) and was modified for measurements under continuous illumination by a BL-emitting diode with an emission maximum at 450 nm LOV1 and -2 cooperatively activate Arabidopsis phototropin2 (LUXEON star, Lumileds Lighting) (40). The procedures for measurements under irradiation of more than 200 mol m Ϫ2 s Ϫ1 were reported previously (27). The specimens were dissolved in 20 mM Tris-HCl, 500 mM NaCl, 1 mM EDTA, and 10% (w/v) glycerol (pH 7.8). The reversion kinetics from S 390 to D 450 in the dark were monitored by the changes in absorption at 450 nm.

SAXS measurement
SAXS were collected at beamline BL45XU of SPring-8 using a PILATUS 300K-W detector (Dectris, Switzerland) as a series of 20 frames of 1-s exposures. The wavelength of the X-ray was tuned to 1.0000 Å. The camera distance was ϳ2500 mm. We used a single-specimen cell with quartz windows of 0.01-mm thickness and a path length of 3.0 mm. The temperature of the samples was maintained at 293 K.
Prior to the SAXS measurements, the specimen solutions were centrifuged at 15,000 ϫ g for 1 h. Dynamic light scattering measurements confirmed that the supernatants were monodispersive. SAXS of D720N, C170A/D720N, C426A/D720N, and C170A/C426A/D720N mutants was collected in a concentration range from 0.5 to 1.2 mg ml Ϫ1 . For each specimen solution, three data were collected sequentially in the dark, under BL irradiation after a preirradiation of 2 min, and in the dark after 15-min dark reversion. Specimen solutions were irradiated using a light-emitting diode placed at 100-mm distance from the specimen position. The fluence rate at the specimen position was 450 mol m Ϫ2 s Ϫ1 . Little radiation damage of all samples was confirmed by the stabilities in successively recorded SAXS profiles as well as by absorption spectra and SDS-PAGE patterns after the X-ray exposure. SAXS data of hen egg white lysozyme (apparent molecular weight of 14,300; Wako Chemical Industry, Osaka, Japan) was collected as a reference to estimate the apparent molecular weight of the D720N mutant.

SAXS analysis
One-dimensional SAXS profiles were reduced from the twodimensionally recorded SAXS patterns after subtracting the background scattering from a buffer solution. Profiles in a verysmall-angle region were analyzed by the Guinier approximation (46). Scattering intensity I(S, C) at a scattering vector length S from a solution at a protein concentration C was approximated by the forward scattering intensity I(S ϭ 0, C) and R g (C) as follows, where 2 is the scattering angle and is the wavelength of the X-ray. For a dilute solution, the concentration dependences of I(S, C) and R g (C) 2 were approximated as follows, where K is a constant. M w is the apparent molecular weight of the protein. A and B are the second virial coefficient and a parameter reflecting the mode of intermolecular interactions, respectively (46). The sign of B is identical to that of A. Assuming a partial specific volume of 0.74 cm 3 /g for soluble proteins, the apparent molecular weight of a target protein is estimated by comparing the I(S ϭ 0, C ϭ 0) value with that of a reference protein (lysozyme) with a known apparent molecular weight. The distance-distribution function P(r) was calculated using the program GNOM (47).

Molecular shape
The low-resolution molecular models of the D720N mutant in different states were restored as an assembly of small spheres of 3.8 Å, called dummy residues, using the program GASBOR (48), which minimizes the difference between experimental and calculated SAXS profiles under the constraints of maintaining a compact and interconnected arrangement of dummy residues. The difference in the observed (I exp (S)) and model-calculated (I model (S)) scattering profiles was monitored by the 2 value defined as follows, where N is the number of experimental data points. S j is the scattering vector of the jth data point, K is a scale factor, and (S j ) is the statistical error in the experimental scattering profile.
The 2 values depend on the number of dummy residues; therefore, we searched an optimum number of dummy residues giving the smallest 2 by varying the number from 800 to 1000 per subunit with a step of 25. For each number, 20 independent GASBOR calculations, starting from different models, were conducted, and then we found the optimum number (600 dummy residues per subunit). In the SAXS measurements, D720N and the other mutants were dissolved in 500 mM NaCl solution. Then the small electron density contrast between the D720N molecule and the solvent resulted in a smaller number of dummy residues than the number of amino acid residues. GASBOR calculations were performed for 576 different models for each state on a supercomputer system with 576 cores of Intel Xeon CPU X5690 (3.7 GHz per core).
After superimposing the 576 models as regards their moment of inertia, the distribution of dummy residues in each model was digitized in an assembly of 4 ϫ 4 ϫ 4-Å 3 voxels. Each model was then expressed as a point of multidimensional space spanned by the density in voxels. By principal component analysis, the distribution of molecular shapes in a multidimensional space was reduced onto the plane spanned by the first and second principal components. The models were then classified into 10 groups by K-means clustering after principal component analysis. Subsequently, we selected a molecular shape by averaging the models in the class composed of the largest number of models. The details of the analysis will be published LOV1 and -2 cooperatively activate Arabidopsis phototropin2 elsewhere. 3 Finally, the molecular shapes were expressed as density maps of the dummy residues in 4 ϫ 4 ϫ 4-Å 3 cubes. The molecular models were illustrated using PyMOL (49).

Electron microscopy
A 3-l solution of the C170A/C426A/D720N mutant was applied on a glow-discharged continuous carbon grid. The excess solution was removed using a filter paper, and the sample was subsequently stained on the carbon grid with 2% uranyl acetate. The negatively stained sample was observed using a JEM-1011 transmission electron microscope (JEOL, Japan) operated at 100 kV and equipped with an F114 slow-scan charge-coupled device camera (TVIPS, Germany) at a nominal magnification of 50,000ϫ.