pH-dependent Equilibrium between Long Lived Near-UV Intermediates of Photoactive Yellow Protein*

The long lived intermediate (signaling state) of photoactive yellow protein (PYPM), which is formed in the photocycle, was characterized at various pHs. PYPM at neutral pH was in equilibrium between two spectroscopically distinct states. Absorption maxima of the acidic form (PYPMacid) and alkaline form (PYPMalkali) were located at 367 and 356 nm, respectively. Equilibrium was represented by the Henderson-Hasselbalch equation, in which apparent pKa was 6.4. Content of α- and/or β-structure of PYPMacid was significantly greater than PYPMalkali as demonstrated by the molar ellipticity at 222 nm. In addition, changes in amide I and II modes of β-structure in the difference Fourier transform infrared spectra for formation of PYPMacid was smaller than that of PYPMalkali. The vibrational mode at 1747 cm-1 of protonated Glu-46 was found as a small band for PYPMacid but not for PYPMalkali, suggesting that Glu-46 remains partially protonated in PYPMacid, whereas it is fully deprotonated in PYPMalkali. Small angle x-ray scattering measurements demonstrated that the radius of gyration of PYPMacid was 15.7 Å, whereas for PYPMalkali it was 16.2 Å. These results indicate that PYPMacid assumes a more ordered and compact structure than PYPMalkali. Binding of citrate shifts this equilibrium toward PYPMalkali. UV-visible absorption spectra and difference infrared spectra of the long lived intermediate formed from E46Q mutant was consistent with those of PYPMacid, indicating that the mutation shifts this equilibrium toward PYPMacid. Alterations in the nature of PYPM by pH, citrate, and mutation of Glu-46 are consistently explained by the shift of the equilibrium between PYPMacid and PYPMalkali.

Glu-46 (see Fig. 1), resulting in the yellow color. Photon absorption by the chromophore initiates the photocycle of PYP (7)(8)(9)(10)(11). The primary photochemical event of PYP is the trans-cis isomerization of the chromophore (12,13), like bacterial retinal proteins. Several intermediates are then formed, followed by a return to the dark state in 100 ms under physiological conditions. During the photocycle, photon energy stored in the twisted chromophore is released to the protein moiety to form the putative signaling state (called PYP M , I 2 , or pB). Because of a largely blue-shifted absorption spectrum with respect to the dark state, as well as the long lifetime (ϳ100 ms), PYP M is easily detectable by UV-visible spectroscopy. This spectral blue shift is considered to be caused by protonation of the phenolic oxygen of the chromophore (5,14).
Infrared spectroscopy demonstrated that PYP M is formed from its precursor (PYP L , I 1 , or pR) in two steps (15,16). First, a proton at Glu-46 is transferred to the chromophore, but the global conformational change of the protein moiety does not take place at this stage, as shown by a small absorbance change of the amide mode. This intermediate (pBЈ) has a protonated chromophore, but the chromophore binding site is considered to be preserved. Second, a global conformational change takes place to form PYP M . As the lifetime of pBЈ (ϳ1 ms) is not markedly different from pR (ϳ100 s) and the absorption spectra of PYP M and pBЈ are similar, isolation of pBЈ has been difficult in steady-state measurement. Recently, the presence of pBЈ between PYP L and PYP M has been established by time-resolved UV-visible spectroscopy and Raman spectroscopy (17)(18)(19).
Although the signaling pathway of the PYP system has not yet been identified, detailed characterization of long lived intermediates (signaling state) is essential for understanding the signal transduction mechanism of PYP. We have demonstrated extensive conformational change upon formation of PYP M . The dimension of PYP M is significantly larger than the dark state as shown by ϳ1 Å increase of R g (20,21) and PYP M is in a partially unfolded state (22). The N-terminal cap and the central part undergo a substantial structural change (21,23); however, the cause of this large conformational change has not yet been determined.
Accumulated evidence suggests that properties of PYP M are highly pH-sensitive. The extent of conformational change evaluated by the difference FTIR spectra under acidic conditions (24,25) is significantly smaller than at neutral pH (15,23). The decay rate constant of PYP M is reduced by acidification, and the relationship between rate constant and pH agrees with the Henderson-Hasselbalch equation with a pK a of 6.4 (26). Proton uptake per PYP M formation is also represented by the Henderson-Hasselbalch equation with a pK a of 6.6 (27).
On the other hand, we have reported that PYP M substantially alters its nature, depending on the buffer system (20). Absorption spectra as well as the lifetime of PYP M at acidic pH are affected by multivalent organic ions, such as citrate. Mutation of amino acids nearby the chromophore also affects properties of PYP M . The absorption maximum of E46Q M is 11-nm red-shifted (28). The protein conformational change for formation of E46Q M is significantly smaller than wild type (16,29), suggesting that the photocycle of E46Q may be different from that of wild type. To study whether there is a simple and consistent explanation for the variety of signaling states or whether each phenomenon has an individual specific reason, the spectroscopic properties of PYP M were studied in detail under various conditions using UV-visible spectroscopy, CD spectroscopy, FTIR spectroscopy, and small angle x-ray scattering measurements. The results are explained by the equilibrium between two states of PYP M .

MATERIALS AND METHODS
Sample Preparation-Wild type PYP and E46Q were prepared as described previously (30). For UV-visible and CD spectroscopy, PYP or mutant was suspended in 10 mM TAPS buffer (pH 8.5-8.0), 10 mM MOPS buffer (pH 8.0 -6.5), or 10 mM MES buffer (pH 6.5-4.0) containing 200 mM NaCl (Dojindo, Kumamoto, Japan). To investigate the effect of binding of citrate to PYP M , 50 mM (see Fig. 6) or 1 M (see Fig. 7) sodium citrate was added instead of NaCl (sample concentration ϳ0.15 mg/ml). For SAXS measurements, samples at pH 5.0 and 8.0 were concentrated to 2-10 mg/ml with an ultrafiltration membrane (Centricon YM10, Millipore, Billerica, MA). For FTIR measurement, PYP and E46Q were desalted, lyophilized, and resuspended in 100 mM MES buffer (pH 5.0) or TAPS buffer (pH 8.0) containing 200 mM NaCl at a concentration of 150 mg/ml. UV-visible Spectroscopy-Absorption spectra in the dark and in the photosteady state were obtained by a multichannel CCD/fiber optic spectroscopy system (S2000 system, Ocean Optics, Dunedin, FL) (23). For steady-state measurements, the sample was irradiated with a yellow light obtained from a 150-W cold light source (HL150, HOYA-Schott, Tokyo, Japan) and a Y43 cutoff filter (Ͼ410 nm, Asahi Techno Glass, Chiba, Japan). For time-resolved measurements, the sample was excited with a yellow flash obtained with a short arc xenon flash lamp (SA200, Nissin Electronic, Tokyo, Japan) and a Y43 filter. Absorption spectra of PYP M were calculated by subtracting dark spectra from photosteady state spectra (31). At pH Ͼ 6, contribution of the small amount of PYP L was also subtracted, as reported previously (31).
CD Spectroscopy-Far-UV CD spectra of dark state and photosteady state were measured using a J-725 circular dichroism spectropolarimeter (JASCO, Tokyo, Japan). Temperature was maintained at 5°C with a PYC-347WI Peltier device (JASCO). The sample cell, with a 1-mm light path length, was irradiated at an angle of 30°using a HOYA-Schott HL150 light source and an Y43 cutoff filter. To protect the detector, the band pass filter (214FS10-25, 210 nm Ͻ Ͻ 233 nm, T max ϭ 14%, Andover Corporation, Salem, NH) was set in front of the detector window. It was confirmed that the artifact on CD signal caused by bandbass filter was negligible in this setup (23).
The photosteady state mixture was composed of PYP, PYP L , and PYP M . The CD spectrum of PYP M was calculated assuming that the CD spectrum of PYP L is indistinguishable from that of PYP because FTIR measurements demonstrate that global structural change takes place at PYP M (32). The fraction of PYP M in the photosteady state mixture was estimated by absorption spectroscopy using an S2000 system under the same irradiation conditions as CD spectroscopy.
FTIR Spectroscopy-Difference FTIR spectra between photosteady state and dark state were measured using an FTS6000 Fourier transform infrared spectrometer (23). The sample was put into a CaF 2 cell with a 7.2 m light path length (23). Irradiation light at 436 nm (FWHM ϭ 10 nm) was obtained by an optical interference filter (43161, Edmund Scientific, Barrington, NJ) and a Hoya-Schott HL150 light source.
SAXS Measurements-Small angle x-ray scattering measurements were carried out at BL-10C (Photon Factory, Tsukuba, Japan (20,21)). The exposure time was 5 min. Three to five sets of independent measurements were averaged to improve the signal to noise ratio. Temperature of the cell was maintained at 5°C. R g and I(0) were obtained by Guinier approximation in the small angle region (33,34), where Q ϭ 4sin/ is the amplitude of the scattering vector, I(Q) is the scattering intensity at Q, 2 is the scattering angle, is the wavelength of the x-ray (1.488 Å), I(0) is the scattering intensity at Q ϭ 0, and R g is the radius of gyration. The sample was irradiated with yellow light obtained using a 1000-W tungsten-halogen lamp (HILUX-HR, Tokyo Master, Japan) and Y43 cutoff filter. The scattering profiles of photoproducts were calculated using scattering patterns of the photosteady state and  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 dark state based on the fraction estimated by absorption spectroscopy.

pH-dependent Equilibrium between PYP Intermediates
pH Titration-Absorbance at 395 nm (see Fig. 2c) or the ellipticity at 222 nm (see Fig. 3b) of PYP M was plotted against pH. Data were fitted according to Henderson-Hasselbalch equation, where k 0 and k 1 are a base value and amplitude of the titration curve, respectively. Fig. 2a shows the absorption spectra of PYP suspended in the buffer at pH 4.0 -8.5 in the dark. The absorption spectrum of PYP is independent of pH in this pH region, although it is bleached at pH Ͻ 4.1 (pK a ϭ 3) (35,36). Then samples at various pH values were irradiated by continuous yellow light (Ͼ410 nm), and the absorption spectra of the photosteady state were recorded.

pH-dependent Spectral Change of PYP M -
As the photosteady state is in equilibrium between PYP, PYP L , and PYP M (31,37), the absorption spectrum of PYP M at each pH was calculated by subtracting the absorption spectra of dark state and PYP L from that of the photosteady state, so that the contributions of PYP and PYP L were cancelled (31). Calculated absorption spectra of PYP M at pH 8.5-4.0 are shown in Fig. 2b. The PYP M spectrum at pH 8.5 had a broad shape with the shoulder at around 330 nm ( max ϭ 356 nm). Upon acidification, the spectrum was red-shifted ( max ϭ 367 nm), and the shoulder disappeared. The pH-dependent spectral change of PYP M at pH 8.5-4.0 is virtually a two-state transition with a clear isosbestic point at 362 nm. The absorbance change at 395 nm was plotted against pH ( Fig. 2c). Absorbance at 395 nm was increased from pH 8.5 to 5.0, but decreased at pH Ͻ 5, suggesting that a part of PYP M is denatured at low pH. The absorbance change was fitted by the Henderson-Hasselbalch equation (Equation 2). The pK a was estimated to be 6.4, which is close to that for the rate constant (6.4) (26) or proton uptake (6.6) (27). The two states of PYP M observed at pH 8.5 and at pH 5.0 are hereafter called PYP M alkali and PYP M acid , respectively. Comparison of Secondary Structure Using CD Spectroscopy-Differences in the secondary structure between PYP M alkali and PYP M acid were evaluated using CD spectra in the far-UV region. The CD spectra of PYP at pH 8.0 and 5.0 agreed with each other in the dark state (Fig. 3a, dashed lines), indicating that the secondary structure of the dark state PYP is pH-independent in this pH region. The CD spectra of the photosteady state were then measured, and pure CD spectra of PYP M were calculated using the fraction of the dark state PYP estimated by absorption spectroscopy under the same irradiation conditions. Typical CD spectra for PYP M at pH values 8.0, 6.5, and 5.0 are shown in Fig. 3a. The significant difference indicates that the secondary structure of PYP M alkali is different from that of PYP M acid . The spectrum at pH 6.5 is in between those at pH 8.0 and 5.0, indicating that it is a 1:1 mixture of PYP M acid FIGURE 2. UV-visible absorption spectra for PYP and PYP M at various pH. a, absorption spectra dark state PYP at pH 8.5-4.0 with an interval of 0.5. Spectra were measured at 5°C. b, absorption spectra of PYP M at pH 8.5-4.0. pH values are indicated in the figure. c, absorbance at 395 nm for PYP M was plotted against pH and fitted using the Henderson-Hasselbalch equation.

pH-dependent Equilibrium between PYP Intermediates
and PYP M alkali . The loss of ellipticity at 222 nm using light is 36% for PYP M alkali (pH 8.0) and 9% for PYP M acid (pH 5.0). In Fig. 3b, the lightinduced ellipticity change at 222 nm (⌬[] 222 nm ) is plotted against pH. From pH 8.5 to 5.0, ⌬[] 222 nm gradually decreased, indicating that the loss of secondary structure is reduced. However, it increased at pH Ͻ 5, as demonstrated by the absorption spectroscopy (Fig. 2c), also indicating partial denaturation of PYP M . The apparent pK a for the ellipticity change was 6.4, which agrees with the change in the absorption spectrum. Thus changes in the absorption spectrum of PYP M correspond well with the secondary structural change. Figs. 2c and 3b show that the fraction of PYP M acid is maximal at pH 5.0, and the equilibration shift is almost saturated at pH 8.0. PYP M acid and PYP M alkali were further characterized using PYP solutions at pH 5.0 and pH 8.0, respectively.
Comparison of Secondary Structure Using FTIR Spectroscopy-The difference between PYP M acid and PYP M alkali was studied by FTIR spectroscopy using PYP solution at pH 5.0 or pH 8.0. Spectra were measured in the dark and under continuous illumination, then difference FTIR spectra between photosteady state and dark state were calculated. Vibrational bands between 1500 and 900 cm Ϫ1 , where chromophore bands appear, were insensitive to pH for the dark state (negative bands); however, the positive bands at 1004 and 1193 cm Ϫ1 for PYP M alkali were slightly down-shifted in PYP M acid . This demonstrates a structural difference in the chromophore between PYP M acid and PYP M alkali that would cause an 11-nm difference in the absorption maximum, whereas changes in the protein environment may also contribute.
Global conformational change induced by light is evaluated by the absorbance change in the amide I (ϳ1650 cm Ϫ1 ) and amide II (ϳ1550 cm Ϫ1 ) mode regions. The negative band at 1646 cm Ϫ1 at pH 8.0 was reduced and split into two bands at 1644 and 1658 cm Ϫ1 at pH 5.0. Second derivative spectra (Fig. 4b) show that a 1646 cm Ϫ1 band at pH 8.0 is composed of 1658, 1645, and 1635 cm Ϫ1 bands, with the 1645 cm Ϫ1 band being the largest of them. At pH 5.0, part of the 1645 cm Ϫ1 mode was shifted to 1658 cm Ϫ1 , resulting in a split peak. A similar shift was observed in N terminus-truncated PYP (23). FTIR spectra and their second derivative spectra clearly show that the 1529 cm Ϫ1 band, a typical amide II mode of a ␤-structure, was reduced by acidification. Together with the CD spectra results, structural change of both ␣-helices and ␤-sheets at pH 5.0 is smaller than at pH 8.0. It should be noted that the positive 1747 cm Ϫ1 band, which indicates protonation of Glu-46, is observed for PYP M acid . The shoulder of this band suggests that Glu-46 is flexible and in two states in PYP M acid . As the peak area of the negative band at 1736 cm Ϫ1 is 5.3 times larger than that of the positive band at 1747 cm Ϫ1 , it is likely that Glu-46 is partially protonated in PYP M acid . Solution Structure of PYP M acid and PYP M alkali -The tertiary structures of PYP M alkali and PYP M acid were characterized by small angle x-ray scattering measurements. The scattering profile was measured in the dark and under illumination at pH 8.0 and 5.0. The scattering profiles for PYP M in each preparation (2-10 mg/ml) at pH 8.0 and 5.0 were calculated by subtracting scattering profiles of the dark state from that of the photosteady state, in which the amount of dark state was estimated by UV-visible spectroscopy. Fig. 5 shows Guinier plots (ln I(Q) versus Q 2 ) for dark state PYP at pH 5.0 (a) and 8.0 (b), plus PYP M acid (a) and PYP M alkali (b). All Guinier plots were fitted in the linear region (Q 2 Ͻ 0.022), and the square of radius of gyration (R g 2 ) was obtained from the slope. In all samples, illumination of PYP made the slope of Guinier plots steeper, indicating that R g is increased upon PYP M formation. However, the increase in the slope for PYP M alkali (Fig. 5b) was clearly larger than for PYP M acid (Fig. 5a). R g 2 values estimated at various concentrations were plotted against concentration (Fig. 5c), and the intrinsic values of R g 2 were obtained by extrapolation to a concentration of zero. R g values thus obtained are shown in Table 1.
At both pH values, y intersections of Guinier plots were not altered by illumination (Fig. 5, a and b), indicating that neither aggregation nor binding of solute ions to PYP M (20) take place. Therefore, the difference in R g between the samples at pH 8.0 and pH 5 (20). Here the effect of citrate on equilibrium of PYP M was investigated. PYP M spectra at pH 8.0 and 5.0 in the presence of 50 mM citrate were similarly calculated from the photosteady state spectra (Fig. 6a). Both agreed with the absorption spectrum of PYP M alkali at pH 8.0 in the absence of citrate. Under the same conditions, CD spectra of PYP M in the presence of citrate were measured (Fig.  6b). Although CD spectra of PYP in the dark were not affected by citrate,  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 spectra of PYP M at pH 5.0 in the presence of citrate agreed with that for PYP M alkali . In addition, R g values also agreed with each other (20), indicating there is no difference in protein structure. These results indicate that binding of citrate shifts the equilibrium between PYP M acid and PYP M alkali toward PYP M alkali . In other words, pK a of the equilibrium shifts to less than 5 because of the binding of citrate to PYP M .

pH-dependent Equilibrium between PYP Intermediates
The shift in equilibrium caused by the binding of citrate to PYP M was observed by transient spectroscopy at a millisecond time scale (Fig. 7). PYP suspended in 1 M citrate buffer (pH 5.0) was excited by a yellow flash, then transient difference spectra were measured. Just after the flash, an absorbance increase at 360 nm was observed, indicating formation of PYP M acid . It was blue-shifted over time, and finally the positive band of the difference spectra had a maximum at 355 nm and shoulder at 330 nm. This spectral shape agreed with that for PYP M alkali . Absorbance change at 383 nm was plotted against the time after flash (inset), fitted to an exponential curve, and the time constant of this shift was estimated to be 300 ms. In 1 M acetate buffer (pH 5.0), PYP M acid was

pH-dependent Equilibrium between PYP Intermediates
formed just after the flash, but little spectral shift was observed in this time scale (data not shown).  (16,29). Absorption maximum of E46Q M at pH 7 is 11-nm red-shifted from PYP M alkali (28), whereas PYP M alkali is dominant at pH 7 for wild type. These characteristics of E46Q M are similar to those of PYP M acid . To examine the similarity between PYP M acid and E46Q M , spectroscopic properties of E46Q M were compared with PYP M acid (Fig. 8). Absorption spectrum of E46Q M at pH 6.5 was calculated from the absorption spectrum of the photosteady state (Fig. 8a). At pH Ͼ 6.5, difference spectra between dark state and photosteady state became significantly smaller because of the short lifetime of E46Q M . Absorption spectrum of E46Q M at pH 6.5 agreed with that at pH 7.0 obtained from the transient difference spectra 10 ms after the flash (28). At a higher pH, the blue shift of E46Q M was not observed, but a significant amount of E46Q L was formed (37). 4 Absorption spectra of E46Q M at pH 6.5 agreed with that of PYP M acid at pH 5.0. At pH 4.0, the spectrum of E46Q M was blue-shifted. This agreed with spectra of PYP M at pH 2.0 (acid denatured state of PYP M ) and pH 8.0 (PYP M alkali ) (Fig. 8a). As it is unlikely that E46Q M is converted into its alkaline form by acidification, E46Q M is readily denatured at pH 4.0. pK a of phenolic oxygen of the chromophore of E46Q is higher than wild type, suggesting that E46Q is less resistant to acidification than wild type. The apparent pK a of E46Q M was estimated to be 5.2 (data not shown).

Role of Glu-46 on Global Conformational Change-Mutation
Difference FTIR spectrum between E46Q M and E46Q was recorded at pH 6.0 and then compared with that between PYP M acid and PYP recorded at pH 5.0 (Fig. 8b). The intensity of difference spectra in the amide I and amide II regions shows protein structural change. The shape and intensity of 1608 -1575 cm Ϫ1 band for E46Q M were comparable with those of PYP M acid . PYP at pH 5.0 shows two bands at 1658 and 1644 cm Ϫ1 . The negative band of E46Q at 1646 cm Ϫ1 was not clearly separated, but the shoulder at ϳ1655 cm Ϫ1 would correspond to 1658 cm Ϫ1 band of PYP. The negative band of E46Q at 1646 cm Ϫ1 is larger than PYP, whereas the positive band of E46Q M at 1625 cm Ϫ1 was smaller than PYP M acid . Consequently, the difference absorbance between peak (1625 cm Ϫ1 ) and valley (1646 cm Ϫ1 ) for E46Q was 89% of PYP. This indicates that the protein structure of E46Q M is comparable with that of PYP M acid . Therefore, absence of negative charge at position 46 would prevent photoactivated PYP from large structural change (16,29) by shifting the PYP M acid -PYP M alkali equilibrium toward PYP M acid state. In other wavelength regions, overall shape of difference spectra were similar, but significant differences were found. The 1747/1736 cm Ϫ1 band of Glu-46 was not observed for E46Q. The 1163 cm Ϫ1 and 1302 cm Ϫ1 bands of dark state PYP shift to 1158 cm Ϫ1 and 1308 cm Ϫ1 , respectively. The former was observed in the Raman spectroscopy as the shift from 1161 cm Ϫ1 to 1154 cm Ϫ1 (38), which would mainly cause a red shift of the absorption spectrum.

DISCUSSION
The present results clearly demonstrate that the characteristics of PYP M are altered by pH. The pH-dependent changes in absorption spectra and CD spectra were represented with the Henderson-Hasselbalch equation, for both of which the apparent pK a was 6.4. Together with the results of FTIR and SAXS, the pH-dependent change in the nature of PYP M is explained by a pH-dependent equilibrium between two types of PYP M (PYP M alkali and PYP M acid ). This is consistent with the previous suggestion by NMR that PYP M (pB) is in an equilibrium between a well ordered state and a partially unfolded state (39). They are different in secondary and tertiary structures as well as the chromophore-protein interaction. The pH-dependent changes in decay rate constant of PYP M (26) and proton uptake (27) are explained by this equilibrium.
The characteristics of E46Q M are comparable with that of PYP M acid , suggesting that replacement of Glu-46 by Gln shifts this equilibrium toward PYP M acid . This seems to suggest the protonation of Glu-46 converts PYP M alkali into PYP M acid and vise versa. In fact, a small positive band at 1747 cm Ϫ1 was observed in difference FTIR spectrum between PYP M acid and PYP. This band is assigned to Glu-46 in protonated form, but the peak area of the positive band is 19% of the negative band. As the area is not intensified even at pH 4.0 (data not shown), Glu-46 is not fully protonated in PYP M acid . Therefore it is unlikely that the protonation state of Glu-46 solely switches between PYP M alkali and PYP M acid , although its negative charge is required for a global conformational change (16,29).
The apparent pK a value of 6.4 in this equilibrium is comparable with that of the imidazole group of the histidine side chain, suggesting the involvement of histidine. The equilibrium shifts toward PYP M acid at low   FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 salt concentration (pK a ϭ 6.7, data not shown). As electrostatic interaction is enhanced at low salt concentration, it is likely that PYP M acid is stabilized by the basic amino acid residue, which is charged at acidic pH. PYP has two histidine residues (His-3 and His-108). Although His-3 is fully exposed to the solvent, His-108 is buried between the ␤-scaffold and the N-terminal cap. Because the structure of the N-terminal cap changes upon PYP M formation at neutral pH, the protonation state of His-108 possibly correlates with the structure of PYP M (25,27).

pH-dependent Equilibrium between PYP Intermediates
Citrate blue shifts the absorption spectra of PYP M and increases R g at pH 5 (20). Here the citrate-bound form of PYP M was characterized in detail, demonstrating that the absorption spectrum, the amount of secondary structure and the R g value of the citrate-bound state of PYP M (PYP M Ј) were in full agreement with those of PYP M alkali . Therefore, the primary effect of binding of citrate is a shift of the equilibrium toward PYP M alkali . NMR analysis of the photoproduct of E46Q has shown that the structural change is limited to ␣3 and the chromophore (29). Assuming that the structure of E46Q M is comparable with PYP M acid , the main structural change for formation of PYP M acid is the conformational change of ␣3. It should be noted that citrate binds to PYP M acid to shift the equilibrium, indicating that the citrate binding site is active in PYP M acid . This finding is consistent with our previous speculation that Arg-52 located in between ␣3 and ␣4 is flipped and forms the citrate binding site (20).
We previously studied conformational change upon the formation of PYP M alkali by CD spectroscopy (23). Results indicated that the difference maximum in CD spectrum was 222 nm, indicating that a structural change in ␣-helices is mainly involved. However, this change cannot be explained by the short ␣-helices like ␣1 and ␣2 located in the N-terminal cap segment. Thus an unfolding of the long helices was strongly suggested. Therefore, the difference between PYP M acid and PYP M alkali is considered to be the unfolding of ␣3 and/or ␣5. Glu-46 and Thr-50 are located in ␣3, and Tyr-42 is located between ␤2 and ␣3. They are hydrogen-bonded with the phenolic oxygen of the chromophore. It should be noted that Glu-46 is in the central part of ␣3. As negative charge at position 46 is required for the conformational change, further structural change in ␣3 would take place in addition to that in ␣5. FTIR spectra demonstrate that the amide II mode of the ␤-sheet (1530 cm Ϫ1 ) is largely changed upon formation of PYP M alkali , indicating that the conformation of the ␤-sheet is also changed.
At neutral pH, two types of PYP M (I 2 /I 2 Ј or pBЈ/pB) are sequentially formed (17,18). The latter has a lifetime of ϳ100 ms and is accumulated by steady-state illumination. The former has a lifetime close to PYP L (ϳ1 ms) but the absorption spectrum is close to the latter. The characteristics of pBЈ are a relatively small protein conformational change (15,16) and slightly red-shifted absorption spectrum (17). In detailed analysis of pH dependence of the PYP photocycle, data were explained without an equilibrium between pB and pBЈ (17). However, recent double flash experiments detected a species, which has photoreversal kinetics similar to I 2 over ϳ100-ms time scale, and the authors suggested an equilibrium between I 2 and I 2 Ј (18). As this experiment was carried out at pH 6, at which 70% of PYP M is in the acidic form, this shows that the photoreversal kinetics of PYP M acid agree with that of I 2 (pBЈ). Absorbance change in amide mode of transient difference FTIR spectrum 450 s after excitation (15) is comparable with that of the PYP M acid /PYP spectrum. Therefore, it is reasonable to conclude that pBЈ (I 2 ) and PYP M acid are the same species and that pBЈ is trapped at acidic pH. Recently, the crystal structures of photocycle intermediates formed from PYP and E46Q were analyzed by time-resolved crystallography (40 -42). They showed the presence of two blue-shifted intermediates of PYP (pB 1 and pB 2 ) whose decay time constants are ϳ10 and ϳ100 ms (42). These values are consistent with the decay time constants of pBЈ (PYP M acid , ϳ1 ms) and pB (PYP M alkali , ϳ100 ms), respectively. To examine whether or not the structures of pB 1 and pB 2 correspond to PYP M acid and PYP M alkali , respectively, the SAXS profiles of pB 1 and pB 2 were calculated from their crystal structures (PDB codes, 1TS0 and 1TS6, respectively) (42) using CRYSOL software (43). Guinier plots of calculated SAXS profile gave the equal values of R g for pB 1 and pB 2 , indicating that pB 1 and pB 2 cannot be distinguished by SAXS experiments. Because experimental R g values for PYP M acid and PYP M alkali are significantly different (Table 1), pB 2 would not correspond to PYP M alkali . PYP M acid state in solution would be the mixture of pB 1 and pB 2 , and the large structural change for formation of PYP M alkali is restricted by crystal packing. For E46Q, three late intermediates were found (IL1, IL2, and IL3) (41). IL1 and IL2 show the similar change in electron density observed in PYP. However, no difference in R g values was resulted from their crystal structures. Together with our results that E46Q M is comparable with PYP M acid , blue-shifted intermediates formed in the crystal are qualitatively the same as PYP M acid . The present data provide detailed characteristics of pBЈ (I 2 or PYP M acid ), which is in the stage just before the global conformational change takes place in the photocycle. These observations are of importance in elucidating the mechanism of light-induced global conformational change of PYP.