A New Measurement System for UV Resonance Raman Spectra of

A new type of ultraviolet resonance Raman (UVRR) measurement system suitable ... Phone: +81-564-55-7340. .... The new system for UVRR spectrometer was...
7 downloads 0 Views 146KB Size
Download PDF
J. Phys. Chem. B 2000, 104, 10765-10774

10765

A New Measurement System for UV Resonance Raman Spectra of Large Proteins and Its Application to Cytochrome c Oxidase† M. Aki,‡ T. Ogura,§ K. Shinzawa-Itoh,|| S. Yoshikawa,|| and T. Kitagawa*,‡,⊥ School of Mathematical and Physical Science, Graduate UniVersity for AdVanced Studies, Myodaiji, Okazaki 444-8585, Japan, Graduate School of Arts and Sciences, The UniVersity of Tokyo, Komaba, Meguro-ku, Tokyo 153-8902, Japan, CREST, Japan Science and Technology, and Department of Life Science Himeji Institute of Technology, Koto, Kamigoricho, Akogun, Hyogo 678-1297, Japan, and Institute for Molecular Science, Okazaki National Research Institutes, Myodaiji, Okazaki 444-8585, Japan ReceiVed: January 28, 2000; In Final Form: May 4, 2000

A new type of ultraviolet resonance Raman (UVRR) measurement system suitable to a limited amount of large protein samples is proposed and the results from its application to bovine cytochrome c oxidase (CcO) is presented. To minimize the sample damage caused by high-flux UV laser illumination and to reject visible fluorescence from the sample, frequency-doubling of a mode-locked Ar+ ion laser and a solar blind multichannel detector were employed, respectively. A new spinning cell was designed so that the sample solution could be stirred during spinning of the cell. Combination of all these devices resulted in successful observation of high quality UVRR spectra of CcO excited at 244 nm. The RR bands of tryptophan and tyrosine residues dominated the observed spectra, while an extra band appeared at 1656 cm-1. The frequency of the extra band as well as those of all other bands were unaltered by the redox change of metal centers and ligand binding to heme a3. Deprotonation of a tyrosine residue(s) with a low pKa value was detected for the resting state at pH 9.1. Examination of all possible assignments led us to conclude that the extra band arose from the linoleoyl side chain of phospholipids and its intensity suggested the presence of 21 linoleoyl groups per CcO molecule.

Introduction Cytochrome c oxidase (CcO, EC 1.9.3.1) is a terminal enzyme of respiratory chain, having redox-active two iron centers (heme a and heme a3) and two copper centers (CuA and CuB), and catalyzes the four electron reduction of dioxygen to water at the heme a3-CuB binuclear site.1 The O2 reduction is coupled with proton translocation across the mitochondrial inner membrane against the concentration gradient of protons2 to generate electrochemical potential for synthesis of ATP from ADP. The ratio of the number of protons to that of electrons transferred were determined to be 4 to 4 in one catalytic cycle,3 but this does not mean that every electron transfer is accompanied by a proton transfer. It is under debate at what stage of the multistep reaction the protons are translocated and how they are coupled.4,5 The reaction mechanism of this enzyme has been studied extensively using time-resolved absorption,6-9 cyrogenic absorption,10,11 EPR,12 and resonance Raman (RR) spectroscopies,13-15 while the X-ray crystallographic structure has been solved with mammalian and bacterial CcO.16,17 The recent progress on the studies of reduction mechanism of O2 has been summarized in review papers.18,19 One of important problems to be solved is the determination of proton carriers and the structural changes of protein moiety caused by the redox and coordination changes at the metal sites. Possible pathways of the vector protons to be transported and the chemical protons †

Part of the special issue “Thomas Spiro Festschrift”. * To whom correspondence should be addressed. Fax: +81-564-554639. Phone: +81-564-55-7340. E-mail: [email protected]. ‡ School of Mathematical and Physical Science. § Graduate School of Arts and Sciences. || Department of Life Science Himeji Institute of Technology. ⊥ Institute for Molecular Science.

to be converted to water have been proposed on the basis of the site-directed mutagenesis20,21 and the observed differences between the X-ray structures of fully oxidized and fully reduced states.22 However, spectroscopic analysis is indispensable for elucidating a detailed mechanism of proton pumping. Since carboxylic-acid side chains are likely candidates of proton carrier, IR spectroscopy plays a critical role in distinguishing between protonation/deprotonation states of carboxylate side chains of protein and porphyrin.23,24 Ultraviolet resonance Raman (UVRR) spectroscopy is also expected to serve as a powerful tool for exploring protein structural changes, particularly on aromatic amino acid residues such as Tyr-244 which is covalently bound to His240 at the ortho position of phenoxy ring and expected to serve as an acid/base catalysis for dioxygen reduction, but so far there has been no UVRR paper of CcO reported. There have been continuing efforts in past 15 years for improved instrumentation of UVRR spectroscopy with excitation around 200-250 nm. In the initial stage, the near-IR emission of a Nd:YAG laser was mixed with visible dye-laser output25 or subjected to H2 Raman shifting26 after generation of higher order harmonics to get a Raman excitation source in the UV region. Because of low repetition rates of lasers (10-30 Hz), pulse energy as high as 0.1 mJ was needed to yield Raman spectra with sufficiently high signal-to-noise (S/N) ratios. When this probe beam is focused to a diameter of 50 µm, its average energy flux would be 5 J/(cm2pulse) and then its pulse energy would correspond to an energy flux of 5 × 108 W/(cm2pulse). Accordingly, damages of samples by illumination of the probe light and some anomaly in Raman intensity by saturation effects were always problematic.27 In the latter case, the authentic intensity of each Raman band was not properly reproduced in

10.1021/jp000357p CCC: $19.00 © 2000 American Chemical Society Published on Web 06/29/2000

10766 J. Phys. Chem. B, Vol. 104, No. 46, 2000 the observed Raman spectra of proteins, because the number of ground-state molecules is varied by the laser intensity and a rate of electronic relaxation. This problem was evaded to a certain extent by the use of an Excimer laser-based UV light source, in which dye-laser output pumped by an Excimer laser with a repetition rate as high as 100-200 Hz was frequencydoubled.28 More recently, applications of an intra-cavity frequency-doubled Ar+ or Kr+ ion laser29 and the frequency doubling of Ti:sapphire laser output pumped by a Q-switched kHz Nd:YLF laser30 to UVRR spectroscopy seem to be satisfactory but unfortunately they are very expensive. With regard to a spectrometer, diffraction gratings were used in the second order to obtain higher diffractivity and reasonably high resolution. However, visible fluorescence from the sample, if present, hits the same position of the detector as the UV light does via the first-order diffraction of the grating and thus results in serious interference to the UVRR spectra. To avoid this, three methods were proposed including the use of a solar-blind detector with CsTe photocathode,31 the use of a prism monochromator as a filter for both visible and Rayleigh emission,32 and the combination of the first and second-order diffraction in the first and second stage dispersion of a double spectrometer.33 Here we propose a new practical system for UVRR measurements of large proteins and apply it to bovine heart CcO which presumably stays as larger aggregates than dimer with an apparent molecular weight of Mr ) 420 kDa, since it has been identified as a dimer in the crystal and the intermolecular interactions seem to be strong. Experimental Procedures Purification of Cytochrome c Oxidase. Cytochrome c oxidase (CcO) was isolated from bovine heart according to Yoshikawa et al.34 and was dissolved in 65 mM sodium phosphate buffer, pH 6.8 containing 0.2% (w/v) D-decyl-β-Dmaltopyranoside (Anatrace Co. Ltd.). The enzyme was stored on ice and used within 7 days for spectroscopic measurements. The concentration of the enzyme was determined spectrophotometrically with the assumption of mM ) 605 - 630 ) 23.3 (in terms of heme A for the fully reduced form); it was 70.8 µM unless otherwise stated. To obtain the spectrum of the fully reduced state (a2+a32+), the enzyme was kept in an airtight spinning cell, equilibrated with N2 gas at 20 °C for 5 min, and reduced by sodium dithionite (final concentration, 1 mM). Then, the cell was filled with higher N2 gas pressure (150 kPa) and left standing at 20 °C for 20 min before the start of measurements. The carbon monoxide complex (a2+a32+CO) was formed by equilibrating the a2+a32+ enzyme solution with CO gas for 5 min in the airtight spinning cell and finally the CO gas pressure was fixed to 150 kPa. The pH (or pD) was determined with a Beckman Φ720 pH meter. The pD values were used without correction from the reading. Determination of Enzymatic Activity. The enzymatic activities of CcO used for Raman measurements were determined at 20 °C from the rate of cytochrome c (Cyt-c) oxidation in 100 mM sodium phosphate buffer containing 0.1% (w/v) D-decyl-β-D-maltopyranoside at pH 6.0. The concentration of CcO was adjusted so that 40-70% oxidation of Cyt-c proceeds in 5 min after the beginning of the reaction (practically, ∼1 nM of CcO and 10 µM of Cyt-c). The reduced Cyt-c was obtained by addition of a small amount of sodium dithionite followed by 5 times repetitions of filtering with a membrane filter and dissolving into 10 mM phosphate buffer at pH 7.4. The concentration of Cyt-c was determined spectrophotometrically with the assumption of ∆mM ) Red - Ox ) 18.5 at 550

Aki et al. nm. The catalytic reaction was pursued by the absorbance at 550 nm (Aobs) as a function of time, and its difference from the absorbance of completely oxidized form (Aox), which was obtained by addition of a small amount of potassium ferricyanide to the solution, was regarded to represent the amount of remaining reduced form. The linear part of the Aobs vs time curve was used for the calculations. Purification of Phosphatidylcholines. Lyophilized phosphatidylcholines including dilinoleoyl-L-R-phosphatidylcholine, β-arachidonoyl-γ-stearoyl-L-R-phosphatidylcholine, and dioleoyl-L-R-phosphatidylcholine (Sigma) were used without further purification (only white powders were used). The aqueous solutions of phosphatidylcholines (PC) were prepared by dispersing the lyophilized PC (ca. 100 mg) into 65 mL of 80 mM sodium phosphate buffer, pH 6.8, containing 1% D-decyl-β-D-maltopyranoside by the use of ultrasonication on an ice bath for 20-30 min. The solution had been bubbled with N2 gas at 20 °C for 90 min before incorporation of PC, and the solution was subjected to ultrasonic treatment and left standing at 20 °C for 3 h at least until the white colloidal solution became colorless. The amount of solubilized PC was quantified spectrophotometrically at 790 nm with the Fiske-SubbaRow reagent,35,36 after the solvent exchange to 100 mM Tris-HCl buffer, pH 8.0, containing 1% D-decyl-β-D-maltopyranoside. The solutions of PC were used within 3 days for spectroscopic measurements. Since it is known that the unsaturated fatty acid side chains are decomposed under aerobic laser irradiation, and as a result, the intensity of cis CdC stretching band decreases during the measurements,37 their Raman spectra were measured under anaerobic conditions, while the absorption spectra were obtained under aerobic conditions. In practice, N2 gas was introduced into the PC solutions in the airtight spinning cell at 20 °C for 5 min and a small amount of sodium dithionite was incorporated to remove oxygen completely from the solution (a final concentration, 1 mM). The solution was left to stand at 20 °C under N2 atmosphere (150 kPa) for 30 min before the start of Raman measurement. Materials. Horse heart cytochrome c (Sigma, type IV), N-acetyl-L-tyrosinamide (Sigma) and N-acetyl-L-tryptophanamide (Sigma) were used without further purification. Butyl acetate of spectral grade, trichloroethylene and sodium dithionite were purchased from Nacalai Tesque. The spectral grade methanol and cyclohexane (Dojin Kagaku) were used as purchased. System Setup. The new system for UVRR spectrometer was designed so as to achieve (1) high throughput, (2) low possibility of sample damage, (3) high sensitivity, and (4) no interference by visible fluorescence. To satisfy these requirements, we adopted the following ideas: (1) the use of a single spectrograph for a high throughput, (2) the use of quasi CW excitation light based on a mode-locked laser to reduce a peak power, and (3) the use of a solar blind multichannel UV detector for high efficiency in accumulation of Raman signals and to reject visible light emitted from the sample, if any. Figure 1 illustrates the setup of UVRR spectrometer constructed in this study. An Ar+ ion laser (Spectra Physics, model 2045) was mode-locked with a mode-locker (model 451) at ∼82 MHz with an average pulse width of 100 ps. Since the ordinary mode-locker is designed to the 514.5 nm line, the gain of the 488.0 nm line is too high to be mode-locked. Therefore, the transmittance of output coupler was raised to 30%T with a custom-made item (CVI Laser Corp.) instead of the use of the ordinary 20%T coupler. To evade the damage of mirrors, the optical path was purged with a gentle flow of N2 gas (70 mL/

UVRR Spectra of Cytochrome c Oxidase

Figure 1. Schematic diagram of the new ultraviolet resonance Raman measurement system constructed in this study. L1-L3, lenses; BBO, a β-barium borate crystal; PB, Pellin-Broca prism; S, sample; OL, a microscope objective lens; LF, liquid filter; PS, a quartz polarization scrambler; AL, a quartz achromatic lens; ICCD, an intensified charge coupled device; BS, a beam stopper.

min). The average output power of the 488.0 nm line was 1W. The fundamental was focused by a lens (L1, f ) 50 mm) into a BBO (β-BaB2O4) crystal (12 × 2 × 2 mm3), and the transmitted light was collimated. The second harmonic at 244.0 nm with an average power of ∼1 mW was separated from the fundamental with a Pellin-Broca prism (PB) and focused into a diameter of 50 µm at a sample (S) by a lens (L3, f ) 100 mm). The energy flux of this light corresponds to ∼1.5 × 10-7 J/cm2pulse. The fundamental and second harmonic were always passed through an acrylic pipe and the BBO crystal was placed in an acrylic box. This was practically important to prevent the power fluctuation of the 244.0 nm line. Raman scattering was collected by a UV microscope lens (OL, Zeiss, Ultrafluar 10X, NA ) 0.20, WD ) 7.4 mm) at right angle and focused onto the entrance slit of the spectrograph (SPEX 1269, f ) 1260 mm, F ) 9) by an achromatic lens (AL, f ) 200 mm). Since the view field of a microscope objective lens is much smaller than that of an ordinary spherical lens, its use to collect Raman scattering is expected to reduce the incorporation of the light arising from scattering at unnecessary parts of the sample such as the bottom of the cell. In fact, the use of the microscope objective lens greatly lowered the level of stray light compared with the case using an ordinary collective lens, and this was substantial for Raman measurements of a giant molecule like CcO that usually provides strong Rayleigh scattering tail. A polarization scrambler (PS) was placed between the objective and achromatic lenses. A liquid filter (LF) could be placed behind the objective, if necessary, to reject the stray light due to Rayleigh scattering.38 In this experiment a 10 mm-thick rectangular cuvette containing butyl acetate/methanol mixed solution (typically butyl acetate: methanol ) 1:4 (v/v)) was used (The concentration of butyl acetate should be adjusted by a trial and error method for individual experiments so as to obtain Raman spectra with the highest S/N ratio, since the optimum concentration depends on the sample). Raman scattering was dispersed with a blazed-holographic grating with 3600 groove/mm blazed at 250 nm and was focused onto the CsTe photocathode of an Intensified Charge Coupled Device (ICCD, Princeton Instruments, model ICCD-1024MG-E/1). The CsTe photocathode has sensitivity only between 200 and 300 nm, and this is essential to reject visible fluorescence. The CCD has 1024 × 256 pixels, but the central 700 × 256 pixels were practically active due to the limited size of the microchannel plate (MCP) intensifier. The CCD was cooled to -29 °C. New Spinning Cell. Difficulty in application of UVRR spectroscopy to important protein samples is that continuous

J. Phys. Chem. B, Vol. 104, No. 46, 2000 10767

Figure 2. Schematic diagram of the new spinning cell holder contructed in this study. The materials of the cell holder and the glass cell are stainless steel and Suprasil with high chemical purity and excellent homogeneity in all axes, respectively.

illumination of the probe light to the sample for long time may denature the protein, whereas the amount of sample is limited. Even when a spinning cell is used, the molecules on the cell surface, which are mainly illuminated by laser light, are little exchanged with molecules inside. To overcome this problem, a new spinning cell with a function of effective stirring during spinning of the cell (diameter ) 5 mm, sample volume ) 50500 µL, 600 rpm) was designed as shown in Figure 2. The idea is to let a small handmade hemi-disk magnet (diameter ) 1.5 mm, thickness ) 2.0 mm) stand still during spinning of the cell by a strong outside magnet. The inside magnet is hold at the bottom of the cell at the opposite side of the laser incidence. This works as an obstructor for the sample flow when the cell is spun and as a result, the solution is effectively stirred. The inside small magnet was prepared by cutting a commercial magnet bar. The cell holder consists of two parts; The upper part is to be screwed to the shaft of the motor at one hand and to be combined with the lower part by three screws at the other, while the lower part holds the glass cell in its center hole of the cell holder. Since the two parts are combined through contact of surfaces and the glass cell is hold by two O-rings, the spin axis of the glass cell is retained stable without swinging. Accordingly, the replacement of the sample does not require realignment of the optics. The use of this spinning cell enabled the observation of high quality UVRR spectra of CcO without any light-induced damage. Raman Measurements. The sample volume put into the spinning cell was 500 µL for the oxidized form but 250 µL for the reduced and reduced CO-bound forms. Temperature of sample was kept at ∼15 °C by flushing with cooled N2 gas against the spinning cell. The average power of the 244 nm line at the sample was reduced to ∼220 µW (∼0.35 × 10-7 J/cm2pulse) by a neutral density filter to minimize the sample damage. A UVRR spectrum presented here is a sum of spectra obtained through ∼120 exposures, each exposure accumulating the data for 15s60 s. The sample was replaced with fresh one every 10 min. Integrity of the sample after exposure to the UV laser light was examined by the enzymatic activity as well as careful comparison of visible absorption spectra measured before and after the UVRR measurements and also of the UVRR spectra obtained upon the first and last exposures. If some spectral changes were recognized, the Raman spectrum was discarded. For the adopted spectra, the enzymatic activity was 13 ( 1 and 15 ( 1 s-1µM-1 for the CcO sample before and

10768 J. Phys. Chem. B, Vol. 104, No. 46, 2000

Figure 3. Effects of stirring for the solution under spinning of the cell. Spectra A and C were obtained with the stirring magnet bar, whereas spectra A′ and C′ were obtained in its absence for oxidized CcO; pH 6.1 for A and A′ and pH 9.1 for C and C′. Traces B and D indicate difference spectra; B ) A - A′ and D ) C - C′. Experimental conditions: laser power, 200 µW; spectral slit width, 12. 8 cm-1; accumulation time, 10 min; buffer, 65 mM sodium phosphate (pH 6.1) or sodium borate (pH 9.1) containing 0.2% (w/v) D-decyl-β-Dmaltopyranoside. Trace E ()C - A) depicts the difference spectra between pH 9.1 minus pH 6.1, for which the individual raw spectra were obtained with accumulation time of 30 min.

after the Raman measurements, respectively. Visible absorption spectra were measured with a Hitachi 220S spectrophotometer. Raman shifts were calibrated with cyclohexane and cyclohexane/trichloroethylene mixed solution (cyclohexane:trichloroethylene ) 4:1 (V/V), whose Raman spectra excited with a visible laser were calibrated with indene. The wavenumber coverage by the detector was approximately 560 cm-1 and thus the wavenumber resolution was ∼0.8 cm-1/pixel. The spectra were intensity-corrected by the use of a D2 standard lamp (Hamamatsu, model C1518 and model H4141SV). Regarding the capability of spectrometer it is noted that the new system has made it possible to observe RR spectrum of rhodamin 6G solutions with sufficiently high S/N ratios (not shown), although the measurements of RR spectra of its solutions in the visible region has been practically impossible due to strong fluorescence. This is owed to the use of the solar-blind detector and partly due to the use of the 250 nm-blazed grating. Results The performance of the new spinning cell with a stirring function was examined with UVRR measurements of CcO. Figure 3 shows the 244.0 nm excited RR spectra of oxidized CcO at pH 6.1 (A and A′) and pH 9.1 (C and C′), for which spectra A and C and spectra A′ and C′ were measured with and without the stirring magnet bar in the spinning cell, respectively. Their difference spectra B ()A - A′) and D ()C - C′) reflect effects of the new spinning cell. At neutral pH, only intensity reduction of the band at 1656 cm-1 in the absence of the stirring magnet bar is conspicuous, but at pH 9.1 an additional differential pattern is seen. The assignment of the 1656 cm-1 band will be discussed in detail later. These spectra were obtained from 10 min accumulation but, besides them, the

Aki et al. spectra for the sample at pH 6.1 and 9.1 were obtained with 30 min accumulation for the sample under stirring conditions and their difference was calculated as depicted by trace E. It is noted that the photosensitive band at 1656 cm-1 is canceled completely but a pH dependent spectral change is observed around 1600 cm-1. This difference peak demonstrates the decrease of tyrosine (1620 cm-1) and the corresponding increase of tyrosinate (1600 cm-1) at alkaline pH,33 although the 1600 cm-1 band is not so intense as that of an ordinary tyrosinate, which will be discussed in a separate paper. The difference peak was not observed for the reduced and reduced CO-bond forms, indicating deprotonation of a tyrosine residue near heme a3. The UVRR spectra of CcO excited at 244 nm are presented in Figure 4, where spectra of the fully oxidized form (A), the fully reduced form (B) and the fully reduced CO-bound form (C) and their differences (E and F) are compared with the spectrum of the aqueous solution of N-acetyl-L-tyrosinamide and N-acetyl-L-tryptophan-amide mixture (D). The mole ratio of tryptophan (Trp) to tyrosine (Tyr) in the mixed solution was adjusted to be the same as that of the enzyme, that is, 57/72.39 The Raman bands due to vibrations of Tyr, Trp, and phenylalanine are labeled by Y, W, and F, respectively. RR bands of Tyr are seen at 1620 (Y8a), 1206 (Y7a), 1174 (Y9a), and 850825 (Y1 and 2*Y16a tyrosine doublet) and those of Trp are seen at 1620 (W1), 1580 (W2), 1554 (W3), 1493 (W4), 1460 (W5), 1360-1340 (W7; tryptophan doublet), 1256 (W9), 1235 (W10), 1149 (W12), 1127 (W13), 1009 (W16), 876 (W17), and 758 (W18), while their vibrational modes have been elucidated by Harada and Takeuchi.40 A shoulder at a lower frequency side of W16 and the band at 1029 cm-1 (F18a) are assigned to the vibrations of phenylalanine. Although it is known that tyrosinate gives a strong Raman band at 1598 cm-1 upon excitation at 244 nm,33 such a peak is not observed in any of these spectra. One would notice at a glance that the UVRR spectra of CcO are extremely close to that of the amino acid solution, meaning that the most of Raman bands arise from Trp and Tyr residues upon excitation at 244 nm. Neither the difference between the spectra of fully oxidized and fully reduced forms (E ) A - B) nor the difference between the spectra of reduced CO-bound and fully reduced forms (F ) C - B) give any distinctive features in this frequency region, indicating that the redox changes of the four metal centers and ligand binding to heme a3 cause little structural changes regarding Tyr and Trp residues. However, we note that an extra Raman band is observed at 1656 cm-1 in addition of the ordinary Tyr and Trp Raman bands. This frequency is close to that of amide I band. If it were so, it should give an appreciable frequency shift upon deuteration of the amide group. Accordingly, the enzyme was subjected to deuteration by five times repetitions of centrifugal concentration and dissolution in D2O. Figure 5 shows the UVRR spectra of the fully oxidized form in D2O (A) and H2O (B), and their difference (C). Spectrum (A) was observed 30 h after the last procedure of solvent exchange. The Y8b band of Tyr (deuteration sensitive component of split Y8 band, 1587 cm-1 in D2O)40 and W18 band of Trp clearly exhibited deuteration shifts to low frequencies. Other bands also show weak changes upon deuteration. Nevertheless, the band at 1656 cm-1 gave no features in the difference spectrum (C), indicating that this band is not shifted by deuteration of proteins and thus is not assigned to the amide-I mode. An A-type heme has appreciable UV absorption in the 200300 nm region. Therefore, Raman excitation in this wavelength region may yield resonance-enhanced Raman bands of heme.41

UVRR Spectra of Cytochrome c Oxidase

J. Phys. Chem. B, Vol. 104, No. 46, 2000 10769

Figure 4. Ultraviolet resonance Raman spectra of bovine heart cytochrome c oxidase: (A) fully oxidized form, (B) fully reduced form, (C) reduced CO-bound form. (D) UV Raman spectrum of an aqueous solution of N-acetyl-L-tyrosinamide mixed with N-acetyl-L-tryptophan-amide in which the mole ratio of Tyr to Trp was adjusted to be 72 to 57. (E) Difference, spectrum A - spectrum B; (F) difference, spectrum C - spectrum B. Experimental conditions: laser power 200 µW for cytochrome c oxidase, 500 µW for an amino acid mixed solution; spectral slit width, 12.8 cm-1; accumulation times, 60 min for cytochrome c oxidase, 20 min for an amino acid solution; buffer, 50 mM sodium phosphate, pH 6.8 for an amino acid mixture.

The formyl CdO, the vinyl CdC, and the farnesyl CdC stretching modes of heme A are possible candidates for the 1656 cm-1 band. In the RR spectra of CcO excited near the Soret band, the formyl CdO stretching mode was observed at 1676 and 1650 cm-1 for the a33+ and a3+ hemes of the fully oxidized form, respectively, and at 1665 and 1610 cm-1 for a32+ and a2+ hemes of the fully reduced form, respectively.42 Thus, their frequencies should be different between the fully oxidized and fully reduced forms. However, in the redox difference RR spectrum excited at 244 nm (Figure 4E), there is no differential type peak, meaning that the frequency of the 1656 cm-1 band remains unaltered by the redox change of hemes. The vinyl Cd C stretching of heme A of the fully reduced form is observed at 1625 cm-1 in the Soret-excited RR spectra.42 This frequency is far from that of the 1656 cm-1 band. Heme A has a farnesyl side chain at position 2. Figure 6 shows the UV RR spectrum of trans,trans-farnesol, which can be a model compound of the farnesyl side chain of heme A. The farnesyl CdC stretching modes are observed at 1643 and 1670 cm-1 upon excitation at 244 nm. The presence of two bands might be due to the terminal and internal CdC bonds or the consequence of vibrational interactions among three CdC bonds. In the case of CcO, the 1656 cm-1 band is a single band whose position is distinct from that of the farnesyl CdC stretching modes of model compound. These results indicate that the formyl CdO, the vinyl CdC, and the farnesyl CdC stretching modes of heme A scarcely contribute to the 1656 cm-1 band in Figure 4. On the other hand, according to X-ray crystallographic analysis,22 many phospholipids are contained as intrinsic constituents of CcO molecule. The cis CdC stretching modes of phospho-

lipids with unsaturated acyl chains appear around 1650 cm-1, while trans CdC stretching modes appear at appreciably higher frequencies.43 In the case of octopus rhodopsin, the cis CdC stretching Raman band of lipid acyl chains was indeed intensityenhanced upon excitation at 244 nm.37 Phospholipids of CcO are highly likely to have unsaturated fatty acyl groups. However, the resolution of the X-ray crystallographic analysis is not high enough for identification of the unsaturated points of the fatty acyl groups. The absorption and vibrational spectra of the acyl chains depend on the number of CdC bonds in a chain as well as the cis/trans configurations. Accordingly, resonance Raman spectra of three phosphatidyl cholines with different numbers of CdC bonds in a single acyl chain were examined, that is, oleic acid side chain (contains one double bond), linoleic acid side chain (contains two double bonds), and arachidonic acid side chain (contains four double bonds). In practice, dioleoyl(18:1,cis-9)-L-R-phosphatidylcholine, dilineoyl(C18:2,cis-9,12)-L-R-phosphatidylcholine, and β-arachidonoyl(C20:4,cis-5,8,11,14)-γ-stearoyl(C18:0)-L-R-phosphatidylcholine, which are commercially available and well characterized, were examined. The absorption spectra of these phosphatidylcholines are shown in Figure 7, where spectra in the 210-360 nm region of linoleoyl type (A), arachidonoyl type (B), and oleoyl type (C) are displayed together with their structures. The linoleoyl type (A) exhibits a prominent peak at 233 nm and a broad weak peak around 276 nm. The arachidonoyl and oleoyl types have no remarkable peak in the wavelength region longer than 220 nm. The UVRR spectra of the three kinds of phospholipids excited at 244 nm are shown in Figure 8, where spectra (A - C) were obtained under anaerobic conditions by adding a small amount

10770 J. Phys. Chem. B, Vol. 104, No. 46, 2000

Aki et al.

Figure 5. Ultraviolet resonance Raman spectra of fully oxidized cytochrome c oxidase in (A) D2O at pD ) 6.7 and in (B) H2O at pH ) 6.8 and their difference spectrum (C) ()A - B). Experimental conditions: laser power 300 µW; spectral slit width, 12.8 cm-1; accumulation times, 30 min.

Figure 6. Ultraviolet resonance Raman spectra of trans,trans-farnesol. Experimental conditions: laser power 160 µW; spectral slit width, 12.8 cm-1; accumulation times, 25 min; sample concentration, 6.0 mM; buffer, 80 mM sodium phosphate (pH 6.8) containing 1% (W/V) D-decyl-βD-maltopyranoside.

of sodium dithionite. To normalize the intensity of three RR spectra, the measurements were carried out for the phospholipid solutions whose solvents were the same Trp/Tyr mixed solution (Trp/Tyr ) 57/72) as used for Figure 4D, and the intensity of the 1620 cm-1 band of Trp and Tyr was used as an internal intensity standard. Then, the contributions from the amino acids were subtracted by the use of Figure 4D. The intensities of the

resulting spectra of three pure phospholipids are scaled by the number of double bonds in the unsaturated fatty acid side chain(s). The cis CdC stretching mode of linoleoyl type was observed at 1656 cm-1, at the same frequency as that of the extra Raman band of CcO (Figure 4A, B, and C). The corresponding band of arachidonoyl type was observed at 1658 cm-1 but its intensity was much weaker. The oleoyl type provided no discernible

UVRR Spectra of Cytochrome c Oxidase

J. Phys. Chem. B, Vol. 104, No. 46, 2000 10771 of the largest molar absorption coefficient (in terms of the Cd C double bond) at the Raman excitation wavelength (244 nm). It is interesting to see how much the phospholipids are contained in the CcO molecule of the present preparation. To quantitatively evaluate it from Raman intensity of the 1656 cm-1 band in Figure 4A-C, we carried out the titration experiments with the linoleoyl phosphatidylcholine to the Trp/Tyr mixed solution, in which the ratio of Trp/Tyr was adjusted to be 57/ 72, the same ratio as that in bovine CcO. The results are shown in Figure 9, where the 244 nm excited RR spectra of the solution containing different amounts of linoleoyl phosphatidylcholine and the same amount of Trp and Tyr residues are depicted. Only the intensity of the 1656 cm-1 band is altered in the unaltered spectra of the Trp and Tyr contributions. The intensities of the 1656 cm-1 band (I1656) relative to that of Y8a and W1 band at 1620 cm-1 (I1620) are plotted against the total number of CdC double bonds of phosphatidylcholine in the inset. It is clear that I1656/I1620 exhibits a linear dependence on the number of CdC bonds. Since the I1656/ I1620 value observed for CcO was 0.6, it corresponds to the presence of 42 CdC bonds per CcO molecule. This suggests that the 1656 cm-1 band of CcO mainly arises from the cis CdC stretching mode of 21 linoleoyl side chains. If the phospholipids are solely cardiolipin with the linoleoyl side chains, it would correspond to five cardiolipin molecules per CcO. Discussion

Figure 7. Absorption spectra of (A) dilinoleoyl-L-R-phosphatidylcholine, (B) β-arachidonoyl-γ-stearoyl-L-R-phosphatidylcholine, (C) and dioleoyl-L-R-phosphatidylcholine. Experimental conditions: light path length, 1 cm; sample concentrations, 0.1 mM; buffer, 80 mM sodium phosphate (pH 6.8) containing 1% (W/V) D-decyl-β-Dmaltopyranoside.

Raman band. The band intensity of cis CdC stretching mode of linoleoyl type is strongest among the three species, because

Performance of the New System. The UVRR spectra of CcO displayed in Figure 4 demonstrate the successful applicability of the present system to large proteins. Since CcO is thought to stay as a dimer or larger oligomers, the molecular weight of the unit would be larger than 420 kDa. Nevertheless, the RR spectra could be observed until 700 cm-1 or lower. This is owed to the use of a liquid filter and microscope objective lens for collection of scattered light. The high S/N ratios obtained are

Figure 8. Ultraviolet resonance Raman spectra of (A) dilinoleoyl-L-R-phosphatidylcholine, (Β) β-arachidonoyl-γ-stearoyl-L-R-phosphatidylcholine, and (C) dioleoyl-L-R-phosphatidylcholine. Experimental conditions: laser power, 140 µW; spectral slit width, 12.8 cm-1; accumulation times, 15 min for (A), 20 min for (Β) and (C); concentrations of phosphatidylcholine, 0.6 mM for (A), 1.1 mM for (Β), 1.7 mM for (C); the solution contains the same amount of Tyr anf Trp as used for spectrum D in Figure 3 and the intensity of amino acids were used as internal intensity standard. Buffer, 80 mM sodium phosphate (pH 6.8) containing 1% (W/V) D-decyl-β-D-maltopyranoside. The spectrum of amino acids were subtracted from the observed spectra after the intensity calibration.

10772 J. Phys. Chem. B, Vol. 104, No. 46, 2000

Aki et al.

Figure 9. Ultraviolet resonance Raman spectroscopic titration of dilinoleoyl-L-R-phosphatidylcholine for the aqueous solution of N-acetyl-Ltyrosinamide and N-acetyl-L-tryptophan-amide in which the mole ratio of Trp to Tyr was adjusted to be 57 to 72 as that in CcO. Experimental conditions: laser power 140 µW; spectral slit width, 12.8 cm-1; accumulation times, 15 min; buffer, 80 mM sodium phosphate (pH 6.8) containing 1% (W/V) D-decyl-β-D-maltopyranoside. The intensity of Raman spectra was scaled with the band at 1620 cm-1 (Y8a and W1) as an intensity standard. The inset shows titration plot of relative intensity (I1656/I1620) against the number of cis CdC bonds. The solid line denotes the best fit line passing through the experimental points marked by circles with r ) 0.99.

partly due to the elimination of visible fluorescence from the light entering into a detector head. On the other hand, an effect of the new spinning cell was clearly observed for spectra A and A′ in Figure 3. It is known that the cis CdC stretching band of phospholipids are easily photooxidized in the presence of oxygen. Although this may also occur in the new spinning cell, there is a clear difference between the ordinary and new spinning cells. The positive and negative peaks at 1615 and 1600 cm-1, respectively, in spectrum D in Figure 3 may mean that tyrosinate is more easily photooxidized to tyrosine radical27 than neutral tyrosine, although further study is necessary for it. It is noted, however, that the difference peak in spectrum E in Figure 3 would not be caused by a photoreaction, because there is no peak at 1656 cm-1, which is the most photosensitive band in the present spectrum. Thus, it would suggest that some tyrosine residues are deprotonated at pH 9.1 while identification of these residues remains to be determined. Structural Information on CcO from UVRR Spectra. The UVRR spectrum of CcO excited at 244 nm (Figure 4) is dominated by the RR bands of Trp and Tyr residues. The bovine CcO contains 72 Tyr residues and 57 Trp residues.39 The band at 1620 cm-1 is expected to have the contributions from both Y8a and W1 modes under this excitation wavelength. Tyr-244 is covalently bound to His-240 at ortho position of phenoxy ring,22 and its vibrational spectrum should be different from those of other Tyr residues. In fact, the RR spectrum of orthoimidazolic-para-cresol, a model compound of Tyr-244, was different from that of para-cresol, and its Y8a bands of neutral and alkaline forms were weaker than those of ordinary cresol (unpublished results). Since the pKa value of Tyr-244 is deduced to be lower than those of other Tyr residues by 1.5,44 it is likely that only Tyr-244 is deprotonated at pH 9.1. In this case, the difference peak in Figure 3E reflects the spectrum of Tyr-244. At neutral pH, on the other hand, since the contribution of Tyr-

244 might be smaller than 1/72 of ordinary Y8a intensity of Tyr, it might not be distinguished from a noise under the present S/N ratios of Figure 4(A-C). The RR spectrum of Tyr-244 could be revealed in near future upon further improvement of the measurement system. The relative intensities of Y7a to Y9a bands and Y1 to 2*Y16 bands of Tyr are known to be sensitive to hydrogen bonding of the phenol OH group.45,46 Y7a is much weaker in the non H-bonding case. The relative intensities of CcO in spectra A, B, and C in Figure 4 are almost the same as that of the aqueous N-acetyl-L-tyrosine-amide solution (Figure 4D), suggesting that the phenolic hydroxy groups of most Tyr are hydrogen bonded as a proton donor in hydrophobic environments.46 The bands of Trp around 1360-1340 cm-1 are referred as the Fermi doublet, relative intensity of which reflects the environments around Trp residues.47 The I1360/I1340 intensity ratios and the intensities of W3 band of spectra A, B, and C in Figure 4 are larger than those for the aqueous N-acetyl-L-tryptophan-amide solution (Figure 4D), suggesting that the environments around Trp residues of CcO are mostly hydrophobic. Structural Change Associated with Redox and Ligation Changes. The redox difference spectrum (E ) A - B in Figure 4) exhibits a small positive peak at 1656 cm-1, indicating that the cis CdC stretching band of the fully reduced form is weaker than that of the fully oxidized form, while the peak position remains unaltered. This change in the intensity of cis CdC stretching mode of linoleoyl type presumably reflects a change in a torsion angle around the CH2-CH2 single bond adjacent to the CHdCH double bond, since the presence of weak coupling between the vibrations of the polymethylene chains at both ends and central CH ) CH-CH2CH ) CH group has been noted.48 Since the 1656 cm-1 band was demonstrated to arise from phospholipids containing linoleoyl side chains, the intensity change would suggest that some conformational change

UVRR Spectra of Cytochrome c Oxidase of the polymethylene chains of the phospholipid takes place in coupling with the redox change of the metal centers. It is reported that some cardiolipin molecules cannot be removed from the enzyme preparation without loss of the enzyme activity.49 There is a suggestion that some cardiolipin in CcO may be functioning in the internal electron transfer between heme a and heme a3-CuB site.50 If the cardiolipins have linoleoyl side chains, the conformational change in the acyl chains of phospholipids discovered in the present study could have a crucial role in controlling the internal electron transfer. On the other hand, it is noted that the Tyr and Trp bands exhibit neither a frequency shift nor an intensity change upon the redox change. This result indicates that the redox-coupled conformational changes of Tyr and Trp residues and environmental changes around Tyr and Trp residues are not detectable in the UVRR spectra excited at 244 nm. According to the X-ray crystallographic analysis of CcO, the redox-coupled conformation changes are seen only in the segment from Gly-49 to Asn55 of subunit I.22 This segment has only one aromatic residue, that is, Tyr-54, whose phenolic hydroxy group keeps hydrogen bonds with one of the propionate groups of heme a and also with the peptide bond between Ser-441 and Tyr-440 in both the fully oxidized and fully reduced forms. Accordingly, little change of UVRR spectra between the fully reduced and fully oxidized forms is consistent with the X-ray results. The ligation difference spectrum (F ) C - B in Figure 4) did not give any peaks for the 1656 cm-1 mode as well as for the Tyr and Trp modes. This suggests that no conformational changes of linoleoyl side chains of phospholipid molecules as well as Tyr and Trp residues of the protein are involved in the process of ligand binding to heme a32+. This is compatible to the conclusion of X-ray crystallographic analysis22 that the fully reduced form shows the conformation identical to that of the fully reduced CO-bound form. Acknowledgment. This study was supported by Grants-inAid for Scientific Research (C) to T.O. (08680730) and for Priority Areas to T.K. (Molecular Biometallics, 08249106) and T.O. (Biological Machinery, 11169210) from the Ministry of Education, Science, Sports and Culture, Japan. References and Notes (1) (a) Wikstro¨m, M.; Krab, K.; Saraste, M. Cytochrome Oxidase, A Synthesis; Academic: New York, 1981. (b) Malmstro¨m, B. G. Chem. ReV. 1990, 1247-1260 (2) Wiksto¨rm, M. Nature (London) 1984, 308, 558-560. (3) Sone, N.; Hinkle, P. C. J. Biol. Chem. 1982, 257, 12600-12604. (4) Michel, H. Proc. Natl. Acad. Sci. U.S.A. 1998, 95, 12819-12824. (5) Verkhorsky, M. I.; Jasaitis, A.; Verkhorskaya, M. L.; Morgan, J. E.; Wikstro¨m, M. Nature (London) 1999, 400, 480-482. (6) Gibson, Q. H.: Greenwood, C. Biochem. J. 1963, 86, 541-554. (7) Orii, Y. Ann. N. Y. Acad. Sci. 1988, 550, 105-117. (8) (a) Oliveberg, M.; Brzezinski, P.; Malmstro¨m, B. G. Biochim. Biophys. Acta 1989, 977, 322-328. (b) Oliveberg, M.; Malmstro¨m, B. G. Biochemistry 1992, 31, 3560-3563. (9) Sucheta, A.; Georgiadis, K. E.; Einarsdottir, O. Biochemistry 1997, 36, 554-565. (10) Chance, B.; Saronio, C.; Leigh, J. S. Jr. J. Biol. Chem. 1975, 250, 9226-9237. (11) Clore, G. M.; Andreasson, L. E.; Karlsson, B.; Aasa, R.; Malmstro¨m, B. G. Biochem. J. 1980, 185, 139-154. (12) (a) Witt, S. N.; Chan, S. I. J. Biol. Chem. 1987, 262, 1446-1448. (b) Blair, D. F.; Witt, S. N.; Chan, S. I. J. Am. Chem. Soc. 1985, 107, 7389-7399. (13) (a) Han, S.; Ching, Y.-C.; Rousseau, D. L. Proc. Natl. Acad. Sci. U.S.A. 1990, 87, 8408-8412. (b) Han, S.; Ching, Y.-C.; Rousseau, D. L. Nature (London) 1990, 348, 89-90. (14) (a) Varotsis, C.; Woodruff, W. H.; Babcock, G. T. J. Am. Chem. Soc. 1990, 111, 6439-6440; 1990, 112, 1297. (b) Varotsis, C.; Woodruff,

J. Phys. Chem. B, Vol. 104, No. 46, 2000 10773 W. H., Babcock, G. T. J. Biol. Chem. 1990, 265, 11131-11136. (c) Varotsis, C.; Zhang, Y.; Appelman, E. H.; Babcock, G. T. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 237-241. (15) (a) Ogura, T.; Kitagawa, T. J. Am. Chem. Soc. 1987, 109, 21772179. (b) Ogura, T.; Yoshikawa, S.; Kitagawa, T. Biochemistry 1989, 28, 8022-8027. (c) Ogura, T.; Takahashi, S.; Shinzawa-Itoh, K.; Yoshikawa, S.; Kitagawa, T. J. Am. Chem. Soc. 1990, 112, 5630-5631. (d) Ogura, T.; Takahashi, S.; Shinzawa-Itoh, K.; Yoshikawa, S.; Kitagawa, T. J. Biol. Chem. 1990, 265, 14721-14723. (e) Ogura, T.; Hirota, S.; Proshlyakov, D. A.; Shinzawa-Itoh, K.; Yoshikawa, S.; Kitagawa, T. Bull. Chem. Soc. Jpn. 1991, 64, 2901-2907. (f) Ogura, T.; Takahashi, S.; Hirota, S.; Shinzawa-Itoh, K.; Yoshikawa, S.; Appelman, E. H.; Kitagawa, T. J. Am. Chem. Soc. 1993, 115, 8527-8536. (g) Ogura, T.; Takahashi, S.; ShinzawaItoh, K.; Yoshikawa, S.; Kitagawa, T. J. Am. Chem. Soc. 1996, 118, 54435449. (16) Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Yamaguchi, H.; Shinzawa-Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S. Science 1995, 269, 1069-1074. (17) Iwata, S.; Ostermeier, C.; Ludwig, B.; Michel, H. Nature (London) 1995, 376, 660-669. (18) Kitagawa, T.; Ogura, T. Prog. Inorg. Chem. 1997, 45, 431-479. (19) Ferguson-Miller, S.; Babcock, G. T. Chem. ReV. 1996, 96, 28892907. (20) Thomas, J. W.; Lemieux, L. J.; Alben, J. O.; Gennis, R. B. Biochemistry 1993, 32, 11173-11180 (21) Fetter, J. R.; Qian, J.; Shapleigh, J.; Thomas, J. W.; Garcia-Horsman, A.; Schmidt, E.; Hosler, J.; Babcock, G. T.; Gennis,. B.; Ferguson-Miller, S. Proc. Natl. Acad. Sci. U.S.A. 1995, 92, 1604-1608. (22) Yoshikawa, S.; Shinzawa-Itoh, K.; Nakashima, R.; Yaono, R.; Yamashita, E.; Inoue, N.; Yao, M.; Frei, M. J.; Libeu, C. P.; Mizushima, T.; Yamaguchi, H.; Tomizaki, T.; Tsukihara, T. Science 1998, 280, 17231729. (23) Hellwig, P.; Behr, J.; Ostermeier, C.; Richter, O.-M. H.; Pfitzner, U.; Odenwald, A.; Ludwig, B.; Michel, H.; Mantele, W. Biochemistry 1998, 37, 7390-7399. (24) Iwase, T.; Varotsis, C.; Shinzawa-Itoh, K.; Yoshikawa, S.; Kitagawa, T. J. Am. Chem. Soc. 1999, 121, 1415-1416. (25) Asher, S. A.; Johnson, C. R.; Murtaugh, J. ReV. Sci. Instrum. 1983, 54, 1657-1662. (26) Rava, R. P.; Spiro, T. G. J. Phys. Chem. 1985, 89, 1856-1861. (27) Johnson, C. R.; Ludwig, M.; Asher, S. A. J. Am. Chem. Soc. 1986, 108, 905-912. (28) Jones, C. M.; Devito, V. L.; Harmon, P. A.; Asher, S. A. Appl. Spectrosc. 1987, 41, 1268-1275. (29) Asher, S. A.; Bormett, R. W.; Chen, X. G.; Lemmon, D. H.; Cho, N.; Peterson, P.; Arrigoni, M.; Spinelli, L.; Cannon, J. Appl. Spectrosc. 1993, 47, 628-633. (30) Zhao, X.; Tengroth, C.; Chen, R.; Simpson, W. R.; Spiro, T. G. J. Raman Spectrosc. 1999, 30, 773-776. (31) (a) Ziegler, L. D.; Hudson, B. H. J. Chem. Phys. 1981, 74, 982. (b) Kaminaka, S.; Ogura, T.; Kitagawa, T. J. Am. Chem. Soc. 1990, 112, 23-27. (32) Hashimoto, S.; Ikeda, T.; Takeuchi, H.; Harada, I. Appl. Spectrosc. 1993, 47, 1283-1285. (33) Kaminaka, S.; Kitagawa, T. Appl. Spectrosc. 1992, 46, 1804-1808. (34) Yoshikawa, S.; Choc, M. G.; O’Toole, M. C.; Caughey, W. S. J. Biol. Chem. 1977, 252, 5498-5508. (35) (a) King, E. J. Biochem. J. 1932, 26, 292. (b) Allen, R. J. L. Biochem. J. 1940, 34, 858. (36) Bartlett, G. R. J. Biol. Chem. 1959, 234, 466-467. (37) Hashimoto, S.; Takeuchi, H.; Nakagawa, M.; Tsuda, M. FEBS Lett. 1996, 398, 239-242. (38) Kleimeyer, J. A.; Fister, J. C., III; Zimmerman, J.; Harris, J. M. Appl. Spectrosc. 1996, 50, 1597-1602. (39) Tsukihara, T.; Aoyama, H.; Yamashita, E.; Tomizaki, T.; Yamaguchi, H.; Shinzawa-Itoh, K.; Nakashima, R.; Yaono, R.; Yoshikawa, S. Science 1996, 272, 1136-1144 and Brookhaven Protein Data Bank, #10CC. (40) Harada, I.; Takeuchi, H. In Spectroscopy of Biological Systems; Clark, R. J. H.; Hester, R. H., Eds.; John Wiley and Sons: Chichester, 1986; pp 113-175. (41) DeVito, V. L.; Cai, M.-Z.; Asher, S. A.; Kehres, L. A.; Smith, K. M. J. Phys. Chem. 1992, 96, 6917-6922. (42) (a) Callahan, P. M.; Babcock, G. T. Biochemistry 1983, 22, 452461. (b) Babcock, G. T.; Callahan, P. M. Biochemistry 1983, 22, 23142319. (43) Lin-Vien, D.; Colthup, N. B.; Fateley, W. G.; Grasselli, J. G. In The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules; Academic Press: New York, 1991; pp 73-75. (44) McCauley, K. M.; Vrtis, J. M.; Dupout, J.; van der Donk, W. A. J. Am. Chem. Soc. 2000, 122, 2403-2404. (45) Siamwiza, M. N.; Lord, R. C.; Chen, M. C.; Takamatsu, T.; Harada, I.; Matsuura, H.; Shimanouchi, T. Biochemistry 1975, 14, 4870-4876.

10774 J. Phys. Chem. B, Vol. 104, No. 46, 2000 (46) Hashimoto, S.; Yabusaki, T.; Takeuchi, H.; Harada, I. Biospectroscopy 1995, 1, 375-385. (47) (a) Miura, T.; Takeuchi, H.; Harada, I. Biochemistry 1988, 27, 8894. (b) Harada, I.; Yamagishi, T.; Uchida, K.; Takeuchi, H. J. Am. Chem. Soc. 1990, 112, 2443-2445.

Aki et al. (48) Koyama, Y.; Ikeda, K. Chem. Phys. Lipids 1980, 26, 149172. (49) Robinson, N. C. Biochemistry 1982, 21, 184-188. (50) Mahapatro, S. N.; Robinson, N. C. Biochemistry 1990, 29, 764770.