12620
J. Phys. Chem. B 2005, 109, 12620-12626
Light-Induced Structural Changes in the Active Site of the BLUF Domain in AppA by Raman Spectroscopy Masashi Unno,*,† Ryota Sano,† Shinji Masuda,‡,§ Taka-aki Ono,‡ and Seigo Yamauchi† Institute of Multidisciplinary Research for AdVanced Materials, Tohoku UniVersity, Sendai 980-8577, Japan, and Laboratory for Photo-Biology (1), RIKEN Photodynamics Research Center, The Institute of Physical and Chemical Research, 519-1399 Aoba, Aramaki, Aoba, Sendai 980-0845, Japan ReceiVed: May 2, 2005
The flavin-adenine-dinucleotide-binding BLUF domain constitutes a new class of blue-light receptors, and the N-terminal domain of AppA is a representative of this family. AppA functions as a transcriptional antirepressor, controlling the photosynthesis gene expression in the purple bacterium Rhodobacter sphaeroides. Upon light absorption, AppA undergoes a photocycle with a signaling state, which exhibits an approximately 10 nm red shift in the UV-vis absorption spectrum. We have characterized light-dependent changes in the active site of an AppA BLUF domain by Raman spectroscopy. The present study has found that altered chromophore-protein interactions, including a hydrogen bond at the C4dO position and structural changes around the N10-ribityl side chain, are key events in this activation process. These structural alterations are proposed to be responsible for the transmission of the light signal in the BLUF domain. This is the first report on a signaling-state Raman spectrum of a blue-light photoreceptor with a flavin chromophore.
Introduction Biological photoreceptors often use trans/cis photoisomerization of their chromophore molecules for light-signal transduction. They include retinal in rhodopsins, phytochromobilin in phytochromes, and 4-hydroxycinnamic acid in photoactive yellow protein.1 In these proteins, a chromophore with altered structure triggers a series of protein structural changes that ultimately lead to a signaling state of the transducer protein. In contrast, the signaling state of the LOV (light, oxygen, or voltage) domain of phototoropins, which is the most characterized family of flavin-containing photoreceptors, accompanies a transient adduct formation between the C4a atom of the isoalloxazine ring (Figure 1) and the sulfur atom of a nearby conserved cysteine residue.2 Presumably, this is because lightinduced conformational changes, such as trans/cis isomerization, are not feasible for a flavin molecule. Recently, a novel flavin-binding domain functioning as a bluelight receptor was defined in several proteins, and this is the so-called BLUF (blue-light-using flavin adenine dinucleotide (FAD)) domain.3 To date, five BLUF proteins have been characterized, which are AppA in the purple bacterium Rhodobacter sphaeroides,4 PAC from Euglena gracilis,5 Slr1694 from Synechocystis sp. PCC6803,6-8 YcgF from Escherichia coli,9 and Tll0078 from Thermosynechococcus elongatus BP-1.10 AppA is a representative of BLUF proteins and functions as a light-dependent transcriptional antirepressor, controlling the photosynthesis gene expression by changing in association with a repressor PpsR for photosynthetic genes.4 Upon light absorption, the BLUF proteins show characteristic red-shifted absorption (∼10 nm) in the UV-vis spectrum of the FAD chro* Author to whom correspondence should be addressed. Phone: +8122-217-5618. Fax: +81-22-217-5616. E-mail:
[email protected]. † Tohoku University. ‡ The Institute of Physical and Chemical Research. § Present address: Graduate School of Bioscience and Biotechnology, Tokyo Institute of Technology, Yokohama 226-8501, Japan.
Figure 1. Structure of the isoalloxazine ring.
mophore. In the case of AppA, the red-shifted state appears on time scales of 90 and 570 ps.11 Similarly, this process completes within 10 ns for Tll0078.10 The light-induced state has been assumed to be a signaling state, which reverses to the dark state relatively slowly with a period of seconds to minutes.4-6,9,10 The signaling state of the BLUF proteins was attributed to altered π-π stacking interactions between the isoalloxazine ring and a conserved tyrosine residue (Tyr21) on the basis of a NMR analysis using wild-type AppA and some mutants.12 A Fourier transform infrared (FTIR) spectroscopic study suggested that deprotonation of the FAD cofactor at N3 is responsible for the signaling state of AppA, because the observed light-induced FTIR difference spectrum showed some similarities with the spectral changes upon deprotonation of free FAD.13 However, a series of FTIR studies on Slr1694 and AppA have suggested the rearrangement of a hydrogen bond, so that the C4dO carbonyl group of the isoalloxazine ring becomes more strongly hydrogen-bonded with nearby residues in the signaling state.6-8,14 These studies also found that the IR bands for both the dark and signaling states exhibit sensitivity to deuteration of exchangeable protons, suggesting that the deprotonation of FAD in the signaling state is unlikely. Vibrational spectroscopy, such as Raman scattering and IR absorption methods, is a potential and sensitive technique for functional and structural elucidation of biological molecules. For chromoproteins, an advantage of Raman spectroscopy over the IR method is its selectivity; i.e., all of the protein components contribute to IR spectra, while a vibrational spectrum of the
10.1021/jp0522664 CCC: $30.25 © 2005 American Chemical Society Published on Web 06/01/2005
Raman Study of the BLUF Domain in AppA
J. Phys. Chem. B, Vol. 109, No. 25, 2005 12621
chromophore can be selectively obtained with a minimum overlap from the protein moiety. Here, we report the first Raman investigation of a BLUF domain in AppA. We also present the results of normal mode calculations of flavin molecules using density functional theory (DFT). These studies allow us to assign most of the observed Raman bands of AppA and offer spectroscopic evidence that the light-induced changes in chromophore-protein interactions are responsible for the light perception mechanism. Furthermore, the present experimental and computational studies also have an important implication for the analysis of vibrational spectra of a flavin molecule. With the information presented here, vibrational spectroscopy provides a unique approach for studying dynamic processes in flavin-containing systems in general. Materials and Methods Sample Preparations. Expression of the BLUF domain of AppA (AppA126), protein purification, and reconstitution of the FAD chromophore were performed as described previously.14 Uniformly 15N- or 13C-labeled AppA126 and 15N- or 13C-labeled apoprotein reconstituted with unlabeled FAD are denoted as 15N, 13C, 15N(apo)/14N(FAD), and 13C(apo)/12C(FAD), respectively. The protein samples were dissolved in H2O (D2O) buffered with 50 mM Tris-HCl (DCl) and 1 mM NaCl at pH (pD) 8.0. The protein concentration was 0.6 mM. FAD was dissolved in 50 mM Tris-HCl buffer with 1 mM NaCl at pH 8.0 or 50 mM KCl/12 mM NaOH buffer at pH 12. FAD (sodium salt, purity 95%) was obtained from Sigma (St. Louis, MO) and was used without further purification. Raman Spectroscopy. Raman spectra were recorded at room temperature using a liquid-nitrogen-cooled CCD detector (Instrument S. A.) after a Triax190 spectrometer (Instrument S. A.) removed the excitation light and a Spex 500M spectrometer dispersed the scattered light as described previously.15,16 The entrance slit width of 0.3 mm used corresponds to the spectral resolution of 5 cm-1 in our spectral region. Samples were excited with the 647.1 nm line available from a krypton ion laser (BeamLok 2065, Spectra-Physics Lasers). The laser power at the sample was 170 mW. The measurements were made on samples contained in a 3 × 1.5 × 48 mm3 quartz cuvette. For the measurement of the signaling state, the samples were illuminated for 0.5 s with a continuous light at 441.6 nm (1.3 mW), which is obtained from a helium-cadmium laser (IK5651R-G, Kimmon Electric), before collecting Raman signals for 60 s. This cycle was repeated more than 120 times to improve the signal-to-noise ratio of the spectra. No change in spectroscopic properties was detected after the Raman measurements. DFT Calculations. Prior to calculation of the vibrational frequencies as well as excited states, all molecules were geometry-optimized in their ground state on the B3LYP/631G** level of theory using the Gaussian03 program.17 The calculated frequencies were scaled using a factor 0.9613. The computational models used for the excited states are configuration interaction singles (CIS)18 and time-dependent density functional theory (TD-DFT).19 The 6-31G* basis set was used for calculating excited states. In the TD-DFT method, we applied the hybrid B3LYP exchange-correlation functional. Results and Discussion Raman Spectra of AppA126. Figure 2 shows the Raman spectra of AppA126 in the dark (trace a) and signaling (trace b) states. The spectra were obtained under a nonresonance condition (λex ) 647.1 nm) to avoid fluorescence from the
Figure 2. Raman and difference spectra of AppA126 and FAD. The spectra were obtained at 647.1 nm excitation. (a) Dark and (b) light states of AppA126 in H2O buffer, (c) dark and (d) light states of AppA126 in D2O buffer, (e) protonated FAD at pH 8, and (f) deprotonated FAD at pH 12. Traces g, h, and i are the difference spectra of (b - a) × 3, (d - c) × 3, and (f - e) × 0.75, respectively. The notation introduced by Bowman and Spiro20 is designated in the figure.
sample. The main features of the AppA126 spectra resemble closely those of free FAD (traces e and f),21 indicating that most of the observed bands in the AppA126 spectra can be ascribed to vibrational modes for the FAD chromophore. This selectivity is an advantage of the Raman technique for exploring the structure of the active site. In fact, a recent FTIR study on isotopically labeled AppA126 demonstrates that most of the observed IR signals originate from a protein moiety.14 The broad Raman band around 1670 cm-1 is absent in free FAD and is ascribed to the amide I mode.22 Analogously, the AppA spectra involve a broad band in the 1220-1260 cm-1 region, which can be characterized by overlapping of a broad amide III band22 and two FAD bands around 1230 and 1260 cm-1. Figure 2 also shows the spectra of dark (trace c) and signaling (trace d) states of AppA126 in D2O buffer. Several Raman bands change in their intensities and frequencies upon deuteration. The prominent downshifts of the bands around 1140-1260 cm-1 as well as an increased intensity of the shoulder at ∼1705 cm-1 are mainly ascribed to deuteration of the N3-H proton.21,22 However, the D2O effects on the amide II and III modes22 account for the increased and decreased intensities of the broad bands around 1460 and 1240 cm-1, respectively. Figure 2 also illustrates that the Raman spectra of AppA126 exhibit small but distinct changes between the dark and the
12622 J. Phys. Chem. B, Vol. 109, No. 25, 2005 signaling states. These differences are clearly observable in the light-dark difference spectra (traces g and h). The amplitude of the light-induced spectral changes is more pronounced for AppA126 in the D2O (trace h) than that in the H2O buffer (trace g), because a slower decay of the signaling state in the D2O buffer14 allows higher population of the signaling state during the measurements. The fraction of the signaling state was estimated from the intensity loss of the Raman band around 1705 cm-1 (see below) of the dark state to be about 70% and 80% for the H2O and D2O samples, respectively. The observed spectral changes accompanying the signaling-state formation are summarized in the following four points. (i) A shoulder at 1706 cm-1 downshifts by ∼20 cm-1. This is more evident for AppA126 in D2O, where a band at 1702 cm-1 exhibits a -16 cm-1 shift. (ii) The bands at 1628, 1581, and 1547 cm-1 are enhanced in intensity. (iii) The band at 1405 cm-1 upshifts by 3 cm-1. (iv) The Raman bands below 1400 cm-1 change in their intensities and/or frequencies. A possible interpretation of these light-induced spectral changes is the deprotonation of FAD at the N3 position.13 This idea, however, can be ruled out because of the following two reasons. First, the deuteration of exchangeable protons affects the Raman spectra of AppA126 in both the dark (trace a f c) and the signaling (trace b f d) states. This result suggests that the FAD chromophore is protonated in the signaling state as well as in the dark state. Second, trace i shows the difference spectrum obtained by subtracting the FAD spectrum at pH 8 (trace e) from that at pH 12 (trace f) for H2O solutions. Although this difference spectrum is caused by the deprotonation at N3, the observed spectral features clearly differ from the lightinduced changes in AppA126 (trace g). On the basis of these observations, we conclude that the formation of the signaling state does not accompany deprotonation of FAD chromophore at the N3 position. Because large structural changes are not feasible for the isoalloxazine ring, the alteration of chromophore-protein interactions is the most likely reason for the light-dependent spectral changes in AppA126. Assignments of FAD Raman Bands of AppA126. To explore the nature of the light-dependent changes in the FAD chromophore of AppA, firm assignments of the observed Raman bands are crucial. To this end, we examine the effects of isotope labeling on the spectra as shown in Figure 3, which illustrates the Raman spectra of the dark state AppA126 at 1260-1760 cm-1. Traces a and b are the spectra for 15N(apo)/14N(FAD) and 13C(apo)/12C(FAD) AppA126, respectively, where the Raman bands for FAD are not affected by the labeling. Traces c and d are the spectra of uniformly 15N- and 13C-labeled samples, respectively. Therefore, the spectral alterations of traces a f c and b f d are primarily ascribed to the 15N and 13C substitutions of FAD, respectively.24 Table 1 summarizes the observed frequencies and the isotope shifts for dark-state AppA126. In addition to the isotopic labeling studies, we performed normal coordinate calculations using DFT for the assignments. In the calculations, lumiflavin was employed as a model of the FAD chromophore with no (model 1) or five water molecules (model 2) to mimic possible hydrogen bonds with the amino acid residues and/or water molecules. Because of the lack of a crystal structure of BLUF domains, water molecules were used as possible hydrogen-bonding partners. The components were optimized to yield the structure illustrated in Figure 4 and Table 1S of the Supporting Information. The computed vibrational frequencies and band assignments (see below) are given in Table 1. Most of the assignment is consistent with a
Unno et al.
Figure 3. High-frequency Raman spectra of dark-state AppA126. The spectra for (a) 15N(apo)/14N(FAD), (b) 13C(apo)/12C(FAD), (c) 15N, and (d) 13C samples are shown.
previous work,25 except for the methyl deformation modes. Figure 5 illustrates atomic displacements for important normal modes. The weak but distinct band at 1706 cm-1 with a large 12C/ 13C isotopic shift (-38 cm-1) is assigned to the carbonyl C4d O stretching vibration ν10 of the isoalloxazine ring based on the observed and calculated frequency and the isotopic shift. This assignment is consistent with that from the reported FTIR study on AppA12614 as well as the Raman study on free FAD.26 Note that the downshift of the ν10 band by deuteration (Figure 2) was reproduced in the calculation with water molecules (model 2), although the calculated shift (-20 cm-1) is larger than the observed shift (-4 cm-1).27 The bands at 1628, 1581, 1547, and 1500 cm-1, exhibiting moderate and large sensitivities to 15N and 13C substitutions, are assignable to CdC and/or Cd N stretching vibrations of the isoalloxazine ring (ν12-ν16). The ν13 and ν14 modes contain larger contributions from the CdN stretching motions than the ν12 and ν16 modes, accounting for larger 15N-induced shifts for the 1581 (ν13) and 1547 cm-1 (ν14) bands than those for the 1628 (ν12) and 1500 cm-1 (ν16) bands. The ν20 and ν21 modes are mainly allocated to the isoalloxazine ring I and II motions including N10-methyl deformation and CsC stretching vibrations, and the large downshifts of 12-34 cm-1 are expected by 13C substitution. Therefore, the band observed at 1405 cm-1 can be ascribed to an overlap of the ν20 and ν21 modes, because of its frequency and ca. -30 cm-1 shifts by 13C substitution. As the calculations indicate a larger 13C downshift in ν20 than ν21, the 1381 and 1371 cm-1 bands for the uniformly 13C-labeled AppA126 (trace d) may be assigned to ν20 and ν21, respectively. The DFT calculations demonstrate that most of the normal modes below 1400 cm-1 are due to vibrations of all three rings of the isoalloxazine moiety. However, the present calculations provide little information on these modes, because a preliminary DFT calculation indicates that a side chain at the N10 position of lumiflavin (R ) CH3 in Figure 1) significantly affects these lower-frequency modes. Further DFT calculations using riboflavin (R ) C5H11O4 in
Raman Study of the BLUF Domain in AppA
J. Phys. Chem. B, Vol. 109, No. 25, 2005 12623
TABLE 1: Observed and Calculated Vibrational Frequency (cm-1) of Dark-State AppA126 and Its Models νobsa
νcalb (model 1)
νcalb (model 2)
1706 (-4,-2,-38)
1744 (0,-1,-45)
1708 (-20,-4,-32)
1628 (+2,-1,-55)
1616 (0,-2,-54)
1616 (0,-2,-54)
assignmentc
band
I
ν10
νC4dO, νC2dO, δN3-H
ν12
νCC(ring I, 8b)
1581 (0,-7,-47)
1574 (0,-11,-45)
1569 (0,-10,-44)
II
ν13
νCN, νCC(ring I, 8b)
1547 (-1,-9,-45)
1536 (0,-7,-37) 1523 (0,-19,-38)
1536 (-1,-8,-45) 1519 (-3,-24,-35)
III
ν14 ν15
νCC(ring I, 8a) νCN, νCC(ring I, 8a)
1500 (0,-4,-50)
1478 (0,-3,-53)
1480 (0,-3,-55)
IV
ν16
νCC(ring I, 19b)
∼1452
1457 (0,-4,+1)
1456 (0,-4,-15)
V
ν17
Me deformation
1405 (0,-5,-24∼-34)
1415 (0,-4,-20) 1394 (0,-2,-27)
1418 (0,-3,-12) 1405 (-1,-2,-34)
VI
ν20 ν21
νCC(ring I, 19a), Me defomation νCC(ring I, 19a), Me defomation
1349 (+1,-4,-41)
1329 (-2,-6,-43) 1321 (0,-8,-31)
1347 (-4,-6,-40) 1321 (-1,-5,-36)
VII
ν26 ν27
νCC, νCN(ring I, II, III) νCC, νCN(ring I, II, III)
a Experimentally observed values. The numbers in the parentheses are the isotopic shifts of N3-D - na, 15N - na, and 13C - na, respectively, where na is the natural abundance. b Calculated vibrational frequencies for models 1 and 2. The numbers in the parentheses are the isotopic shifts of N3-D - na, 15N - na, and 13C - na, respectively. c The observed Raman bands are assigned to the calculated normal modes of lumiflavin. Approximate descriptions of the calculated modes are described. For ring I vibrations, the corresponding vibrational modes of benzene are indicated.
Figure 4. Optimized structures for the active site models of AppA126. The following list gives the description of the models: model 1, lumiflavin; model 2, lumiflavin surrounded by five water molecules; model 3a, lumiflavin and four water molecules without a hydrogen bond at the N1 position; model 3b, lumiflavin and four water molecules without a hydrogen bond at the C2dO position; model 3c, lumiflavin surrounded by two water molecules around the N1 and N5 positions and penta-1,5-diol that hydrogen bonds to the C2dO and C4dO moieties; model 3d, lumiflavin and four water molecules without a hydrogen bond at the C4dO position; model 3e, lumiflavin and four water molecules without a hydrogen bond at the N5 position; models 4a-c, 7,8-dimethyl-10-ethylisoalloxazine with different C10a-N10-C1′-C2′ dihedral angles τ. The values of τ for models 4a, 4b, and 4c are 80.0°, 91.4°, and 100.0°, respectively.
Figure 1) are currently in progress, and the assignments in this region will be described elsewhere. Light-Induced Structural Changes in AppA. A. C4dO Moiety. On the basis of the assignments described above, we now discuss the light-driven structural changes in the active site of AppA126. A noticeable change in the Raman spectrum
upon formation of the signaling state is an approximately 20 cm-1 downshift of the C4dO stretching vibration ν10, which is observed at 1706 and 1702 cm-1 in H2O and D2O, respectively, for dark-state AppA126 (Figure 2). Because a CdO stretching frequency is expected to be sensitive to hydrogen bonding,29 we have performed DFT calculations using lumiflavin models
12624 J. Phys. Chem. B, Vol. 109, No. 25, 2005
Unno et al.
Figure 5. Atomic displacement vectors for some vibrational modes of lumiflavin (model 1).
TABLE 2: Effects of Formation or Disruption of Hydrogen Bonds on Vibrational Frequencies (cm-1) and Absorption Maxima (nm) of Lumiflavin model 2a
model 3aa (N1)
model 3ba (C2dO)
model 3ca (N3sH)
ν10 ν20 ν21
1708 1418 1405
1707(1) 1415(3) 1397(8)
Vibrational Frequencies 1724(-16) 1720(-12) 1418(0) 1417(1) 1405(0) 1405(0)
11A′f21A′ 11A′f31A′ 11A′f51A′
418 339 253
420(-2) 336(3) 253(0)
423(-5) 337(2) 252(1)
Excitation Energiesc 420(-2) 343(-4) 254(-1)
model 3da (C4dO)
model 3ea (N5)
AppAb
1736(-28) 1418(0) 1405(0)
1713(-5) 1419(-1) 1404(1)
1702(-16) 1406(3)
410(8) 335(4) 254(-1)
408(10) 328(9) 253(0)
443(13)d 370(8)d
a Theoretically calculated values. The numbers in the parentheses are the shifts of model 2 - model 3x (x ) a-e). For models 3a-e, a position that does not contain a hydrogen bond is indicated. b Experimentally observed values for AppA126. The numbers in the parentheses are the shifts of the signaling state - dark state. c Calculated by the TD-DFT method. d Reference 14.
that contain five water molecules (model 2) and four water molecules with no hydrogen bond at the N1 (model 3a), C2d O (model 3b), N3sH (model 3c), C2dO (model 3d), or N5 (model 3e) position, respectively (Figure 4).28 As summarized in Table 2, we have found that formation or strengthening a hydrogen bond at either the C2dO (corresponding to the change from model 3b to model 2, model 3b f 2), N3sH (model 3c f 2), or C4dO (model 3d f 2) position downshifts the ν10 frequency, comparable with the observed shift, suggesting that a change in at least one of these hydrogen bonds accounts for the observed downshift of the C4dO stretching vibration. As summarized in Table 1S of the Supporting Information, models 3b-d exhibit the altered C2dO, C2sN3, N3dC4, and/or C4d O bond lengths. Thus, the downshift of the ν10 mode is attributable to geometric changes of the C2(dO)sN3HsC4(dO) moiety of the isoalloxazine ring. A recent FTIR study on Slr1694 found a correlation between the ν10 frequency and the red shift in the UV-vis absorption.8 We therefore applied TD-DFT and CIS methods to calculate the absorption spectra of the models discussed above. The results shown in Table 2 and Table 2S of the Supporting Information indicate that the formation of the hydrogen bond at the C4dO (model 3d f 2) or N5 (model 3e f 2) position induces the red shift in the UV-vis absorption of lumiflavin, whereas the hydrogen bonding at the other positions such as C2dO (model
3b f 2) and N3sH (model 3c f 2) causes the spectral blue shift. Thus, only the change of model 3d f 2 is consistent with the experimental findings of the downshift of the ν10 mode and the red shift in the UV-vis absorption spectrum. These results lead to the conclusion that the light-dependent changes in AppA126 involve formation or strengthening a hydrogen bond at the C4dO position. Upon formation of the signaling state, the CdC and/or Cd N stretching modes at 1550-1650 cm-1 (ν12, ν13, and ν14) increase in intensity as shown in Figure 2. This can be attributable to a change in strength of the hydrogen bond at the C4dO position, because the altered hydrogen bond could affect the polarizability of the CdC and CdN bonds in the isoalloxazine ring, and therefore causes a change in the Raman intensity of the corresponding stretching vibration. Alternatively, changes in an electronic state responsible for the red-shifted UV-vis absorption of flavin may cause the increased intensity of these Raman bands, if the excitation wavelength falls in preresonance with some π-π* electronic transition. Further experiments at different excitation wavelengths may clarify this point. B. N10-Ribityl Moiety. Another notable difference between the dark and signaling states of AppA126 is the upshift of the ν20/ν21 modes around 1405 cm-1 (Figure 2). As illustrated in Figure 5, the ν20/ν21 modes are coupled vibrations between ring I and II motions and N10-methyl deformation motions. Because
Raman Study of the BLUF Domain in AppA
J. Phys. Chem. B, Vol. 109, No. 25, 2005 12625
TABLE 3: Calculated Vibrational Frequencies (cm-1) of Three Conformations of 7,8-Dimethyl-10-ethylisoalloxazine ν10 ν20 ν21 c
model 4aa
model 4bb
model 4cc
1744 1413 1372
1744 1415 1377
1744 1415 1374
a The C10a-N10-C1′-C2′ dihedral angle τ is 80.0°. b τ ) 91.4°. τ ) 100.0°.
these modes little involve the ring III vibration, it is expected that the ν20/ν21 frequencies are insensitive to the hydrogenbonding interactions at the ring III moiety. In fact, the results in Table 2 demonstrate that the formation of a hydrogen bond at either the C2dO, N3sH, C4dO, or N5 position (model 3b-e f 2) little affects the ν20/ν21 modes. However, the hydrogenbonding interaction of the isoalloxazine ring at the N1 position (model 3a f model 2) causes moderate upshifts (3-8 cm-1) of the ν20 and ν21 frequencies. An inspection of the optimized structures for models 2 and 3a-e reveals that the upshifts of the ν20/ν21 modes can be predominantly ascribed to structural changes around the N10-C1′ (CH3) moiety such that the C10aN10-C1′ bond angle (119.7°) in model 2 is slightly larger than that (118.9°) in the structure without a hydrogen bond at the N1 position (model 3a) (Figures 4 and Table 1S of the Supporting Information). To examine the effects of structural changes around the N10-C1′ moiety on the ν20 and ν21 frequencies, we have calculated the vibrational frequencies for models 4a-c, where 7,8-dimethyl-10-ethylisoalloxazine (R ) C(1′)H2-C(2′)H3 in Figure 1) has different C10a-N10-C1′C2′ dihedral angles (Table 3 and Figure 4). The results listed in Table 3 show that the ν20 and ν21 modes are sensitive to the structural changes in the N10-C2H5 moiety, whereas the ν10 mode exhibits no change. Therefore, these results suggest that the formation of the signaling state accompanies structural modifications of the N10-ribityl moiety. This idea is consistent with the finding that the light-dependent spectral changes are also observed in the lower-frequency region (1100-1300 cm-1), where large motions of the ribityl side chain are expected to appear as noted above. Functional Implications. The present study is the first report for the Raman spectra of a BLUF domain and provides unequivocal evidence that the formation of the signaling state of AppA126 does not accompany deprotonation of the FAD chromophore at the N3 position (Figure 2). This result implies that the skeletal structure of the isoalloxazine ring remains unchanged during the photocycle of the BLUF domain. This is clearly different from the best-characterized family of flavincontaining photoreceptors, the LOV domains, where a covalent C4a flavin-cysteinyl adduct is the signaling state.2 Despite a lack of major skeletal alterations in the isoalloxazine ring, we have found that most of the Raman bands of the FAD chromophore in AppA126 change their frequencies and/or intensities upon formation of the signaling state (Figure 2). This indicates the alteration of chromophore-protein interactions during the photocycle. Normal coordinate calculations based on DFT suggest that the altered interactions involve lightinduced changes in the N10-ribityl side chain as well as in the hydrogen bond at the C4dO position, implying that changes in the interactions are not limited to the ring III moiety. A possible interpretation of the results is that the FAD chromophore is capable of adopting in two different configurations and/or positions in the BLUF domain and light illumination results in a switch between the two states. Slight movements of the FAD chromophore within a protein are realized in some flavoenzymes such as p-hydroxylbenzoate hydroxylase.25,30 The change in the
position of the FAD chromophore could also account for the proposed changes in π-π stacking interactions between Tyr21 and the isoalloxazine ring.12 The global nature of the interaction changes is consistent with the recent FTIR studies, where relatively large structural changes are detected in the protein backbone of the BLUF domains of AppA and Slr1694.6-8,14 Therefore, we propose that light-induced changes in interactions between FAD and protein moiety are linked to the conformational changes of BLUF domains, which initiate the transmission of the light signal. Acknowledgment. We are grateful to Dr. Koji Hasegawa for helpful discussion and Mr. Kei Takahashi for help with quantum chemical calculations. This work was supported by grants from the Association for the Progress of New Chemistry (M.U.) and the Ministry of Education, Culture, Science, Sports, and Technology (16570131 to M.U. and 16770046 to S.M.). A part of the computations were performed at the Research Center for Computational Science, Okazaki, Japan. Supporting Information Available: Optimized geometries of the models discussed in the text and results of the excitedstate calculations. This material is available free of charge via the Internet at http://pubs.acs.org. References and Notes (1) van der Horst, M. A.; Hellingwerf, K. J. Acc. Chem. Res. 2004, 37, 13-20. (2) Salomon, M.; Christie, J. M.; Knieb, E.; Lempert, U.; Briggs, W. R. Biochemistry 2000, 39, 9401-9410. (3) Gomelsky, M.; Klug, G. Trends Biochem. Sci. 2002, 27, 497500. (4) Masuda, S.; Bauer, C. E. Cell 2002, 110, 613-23. (5) Iseki, M.; Matsunaga, S.; Murakami, A.; Ohno, K.; Shiga, K.; Yoshida, K.; Sugai, M.; Takahashi, T.; Hori, T.; Watanabe, M. Nature 2002, 415, 1047-51. (6) Masuda, S.; Hasegawa, K.; Ishii, A.; Ono, T. Biochemistry 2004, 43, 5304-5313. (7) Hasegawa, K.; Masuda, S.; Ono, T. Biochemistry 2004, 43, 1497914986. (8) Hasegawa, K.; Masuda, S.; Ono, T. Plant Cell Physiol. 2005, 46, 136-146. (9) Rajagopal, S.; Key, J. M.; Purcell, E. B.; Boerema, D. J.; Moffat, K. Photochem. Photobiol. 2005, 80, 542-547. (10) Fukushima, Y.; Okajima, K.; Shibata, Y.; Ikeuchi, M.; Itoh, S. Biochemistry 2005, 44, 5149-5158. (11) Gauden, M.; Yeremenko, S.; Laan, W.; van Stokkum, I. H. M.; Ihalainen, J. A.; van Grondelle, R.; Hellingwerf, K. J.; Kennis, J. T. M. Biochemistry 2005, 44, 3653-3662. (12) Kraft, B. J.; Masuda, S.; Kikuchi, J.; Dragnea, V.; Tollin, G.; Zaleski, J. M.; Bauer, C. E. Biochemistry 2003, 42, 6726-6734. (13) Laan W.; van der Horst M. A.; van Stokkum I. H.; Hellingwerf, K. J. Photochem. Photobiol. 2003, 78, 290-297. (14) Masuda, S.; Hasegawa, K.; Ono, T. Biochemistry 2005, 44, 12151224. (15) Unno, M.; Kumauchi, M.; Sasaki, J.; Tokunaga, F.; Yamauchi, S. J. Phys. Chem. B 2003, 107, 2837-2845. (16) Unno, M.; Kumauchi, M.; Hamada, N.; Tokunaga, F.; Yamauchi, S. J. Biol. Chem. 2004, 279, 23855-23858. (17) Frisch, M. J.; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.; Robb, M. A.; Cheeseman, J. R.; Montgomery, Jr., J. A.; Vreven, T.; Kudin, K. N.; Burant, J. C.; Millam, J. M.; Iyengar, S. S.; Tomasi, J.; Barone, V.; Mennucci, B.; Cossi, M.; Scalmani, G.; Rega, N.; Petersson, G. A.; Nakatsuji, H.; Hada, M.; Ehara, M.; Toyota, K.; Fukuda, R.; Hasegawa, J.; Ishida, M.; Nakajima, T.; Honda, Y.; Kitao, O.; Nakai, H.; Klene, M.; Li, X.; Knox, J. E.; Hratchian, H. P.; Cross, J. B.; Bakken, V.; Adamo, C.; Jaramillo, J.; Gomperts, R.; Stratmann, R. E.; Yazyev, O.; Austin, A. J.; Cammi, R.; Pomelli, C.; Ochterski, J. W.; Ayala, P. Y.; Morokuma, K.; Voth, G. A.; Salvador, P.; Dannenberg, J. J.; Zakrzewski, V. G.; Dapprich, S.; Daniels, A. D.; Strain, M. C.; Farkas, O.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.; Foresman, J. B.; Ortiz, J. V.; Cui, Q.; Baboul, A. G.; Clifford, S.; Cioslowski, J.; Stefanov, B. B.; Liu, G.; Liashenko, A.; Piskorz, P.; Komaromi, I.; Martin, R. L.; Fox, D. J.; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.; Challacombe, M.; Gill, P. M. W.;
12626 J. Phys. Chem. B, Vol. 109, No. 25, 2005 Johnson, B.; Chen, W.; Wong, M. W.; Gonzalez, C.; Pople, J. A. Gaussian 03, revision B.04; Gaussian, Inc.: Wallingford, CT, 2004. (18) Foresman, J. B.; Head-Gordon, M.; Pople, J. A.; Frisch, M. J. J. Phys. Chem. 1992, 96, 135-149. (19) Stratmann, R. E.; Scuseria, G. E.; Frisch, M. J. J. Chem. Phys. 1998, 109, 8218-8224. (20) Bowman, W. D.; Spiro, T. G. Biochemistry 1981, 20, 3313-3318. (21) Kim, M.; Carey, P. R. J. Am. Chem. Soc. 1993, 115, 7015-7016. (22) Harada, I.; Takeuchi, H. In Spectroscopy of Biological Systems; Clark, R. J. H.; Hester, R. E., Eds.; John Wiley & Sons: Chichester, U. K., 1981; Vol. 13, 113-175. (23) There are some variations in the chromophore content upon heterologous overexpression of AppA126 in E. coli.24 This causes a difference in intensity for protein Raman bands such as amide I at ∼1650 cm-1. (24) Laan W.; Bednarz T.; Heberle J.; Hellingwerf K. J. Photochem. Photobiol. Sci. 2004, 3, 1011-1016.
Unno et al. (25) Zheng, Y.; Dong, J.; Palfey, B. A.; Carey, P. R. Biochemistry 1999, 38, 16727-16732. (26) Hazekawa, I.; Nishina, Y.; Sato, K.; Shichiri, M.; Miura, R.; Shiga, K. J. Biochem. 1997, 121, 1147-1154. (27) We have found that the calculated shift highly depends on the model. For example, if we replaced water molecules with acetamide, then the calculated H/D exchange shift became ca. 5 cm-1. Further DFT studies are in progress to achieve a qualitative agreement with experiments. (28) For model 3c, water molecules W2 and W4 are replaced with penta1,5-diol. Note that, in the absence of W3 molecule, water W2 formed a hydrogen bond with the N3-H moiety in the optimized structure. (29) Unno, M.; Kumauchi, M.; Sasaki, J.; Tokunaga, F.; Yamauchi, S. Biochemistry 2002, 41, 5668-5674. (30) Wang, J.; Ortiz-Maldonado, M.; Entsch, B.; Massey, V.; Ballou, D.; Gatti, D. L. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 608-613.