Factors that determine the unusually low reduction ... - Springer Link

J Biol Inorg Chem (2001) 6: 708±716 DOI 10.1007/s007750100249

O R I GI N A L A R T IC L E

John S. Vrettos á Michael J. Rei¯er á Olaf Kievit K.V. Lakshmi á Julio C. de Paula á Gary W. Brudvig

Factors that determine the unusually low reduction potential of cytochrome c550 in cyanobacterial photosystem II Received: 19 December 2000 / Accepted: 12 April 2001 / Published online: 9 June 2001 Ó SBIC 2001

Abstract A new puri®cation protocol for cytochrome c550 (cyt c550) from His-tagged Synechocystis PCC 6803 photosystem II (PSII) was developed which allows the protein to be isolated in high yield and purity. Electron paramagnetic resonance spectroscopy of cyt c550, both free in solution and in intact PSII preparations, yields identical spectra with g values at 1.50, 2.23, and 2.87, which are characteristic for a ferric low-spin bis-histidine coordinated heme. The resonance Raman spectrum of the isolated protein exhibits features characteristic of bis-histidine axial ligation of the iron and a slight ru‚ing of the heme macrocycle. Together, these results indicate that the heme structure is not very di€erent from most c-type cytochromes, and thus the structure of the heme does not account for its unusually low reduction potential. A direct electrochemical measurement of the reduction potential was performed using square wave voltammetry on a pyrolytic graphite edge electrode, yielding E1/2=±108 mV (vs. NHE) with a peak separation of 5 mV. This value is 150 mV more positive than that previously measured by redox titrations. Because the behavior of the protein in the electrochemistry experiments is indicative of adsorption to the electrode surface, we surmise that binding of the protein to the electrode excludes solvent water from the heme-binding site. We conclude that the degree of solvent exposure makes a signi®cant contribution to the heme reduction potential. Similarly, the binding of cyt c550 to PSII may also reduce the solvent exposure of the heme, and so the direct electrochemical value of the reduction potential may be relevant to the protein in its native state. J.S. Vrettos á M.J. Rei¯er á O. Kievit á K.V. Lakshmi G.W. Brudvig (&) Department of Chemistry, Yale University, New Haven, CT 06520-8107, USA E-mail: [email protected] Tel.: +1-203-4325202 Fax: +1-203-4326144 J.C. de Paula Department of Chemistry, Haverford College, Haverford, PA 19041-1392, USA

Keywords Cytochrome c550 á EPR spectroscopy á Photosystem II á Protein electrochemistry á Resonance Raman spectroscopy

Introduction Cytochromes play important roles in photosynthesis [1, 2]. One such cytochrome with a structure and function that remain elusive is cytochrome c550 (cyt c550), also known as cyt c549 or low-potential (LP) cyt c. Cyt c550 was ®rst isolated and characterized from Anacystis nidulans in the 1960s [3]. Not found in higher plants, it has been identi®ed in several algae, as well as in cyanobacteria including Synechocystis PCC 6803 and Synechococcus vulcanus. Cyt c550 and a 12 kDa protein in cyanobacteria are extrinsic membrane-bound proteins analogous to the 17 and 23 kDa extrinsic proteins of photosystem II (PSII) of higher plants. They have similar in¯uences on some of the cofactors in PSII and bind in similar regions of the lumenal membrane surface, but their functions are very di€erent [4, 5]. The reduction potential of cyt c550 is unusually low for a c-type heme; hence its label as LP cyt c. A midpoint potential of ±260 mV was found by redox titrations, using 2-anthraquinonesulfonic acid to reduce the protein in 0.1 M phosphate bu€er at pH 7.0, for cyt c550 puri®ed from both A. nidulans [3] and Synechocystis PCC 6803 [6]. Cyt c550 from Microcystis aeruginosa and Aphanizomenon ¯os-aquae have been found to be reducible by dithionite (E1/2=±1.12 V), but not by ascorbate (E1/2=0.058 V). The reduced form of cyt c550 is known to auto-oxidize under aerobic conditions [7]. Such a low reduction potential is well below the range normally expected for a mono-heme c-type cytochrome. Cytochromes c of various classes typically have reduction potentials of around 0 mV or higher [8]. Several roles have been suggested for cyt c550. According to Krogmann and Smith [9, 10], the function of cyt c550 may be related to anaerobic disposal of electrons from carbohydrate reserves, or fermentation to sustain

709

an organism during prolonged dark and anaerobic conditions. Cyt c550 shares regions of sequence homology with the high-potential cyt c553, which functions as an electron donor to photosystem I (PSI) in cyanobacteria and eukaryotic algae, and cyt c6, which can act as an electron carrier between the cyt b6f complex and PSI [6, 9, 10]. According to Shen and Inoue [11], cyt c550 can accept electrons from ferredoxin II in the presence of dithionite, and is proposed to remove excess electrons in anaerobically grown cells. Shen and co-workers [12, 13, 14] have also found that cyt c550 stabilizes the oxygenevolving complex of PSII in a manner analogous to the extrinsic polypeptides found in higher plants. Indeed, the 3.8 AÊ resolution crystal structure of PSII from Synechococcus elongatus shows that cyt c550 binds in the vicinity of the oxygen-evolving complex [15]. In the current study, we explore the factors contributing to the unusually low reduction potential of cyt c550. Puri®ed cyt c550 was prepared by high-salt elution from His-tagged PSII bound to a Ni-NTA matrix. This new method gives a high yield of puri®ed cyt c550. Electron paramagnetic resonance (EPR) and resonance Raman spectroscopy results are characteristic of a bishistidine ligation of the heme and indicate that the heme structure is not very di€erent from other c-type cytochromes. The reduction potential of cyt c550 was found to be signi®cantly higher when the protein was adsorbed onto a graphite electrode surface versus free in solution. An increase in the reduction potential for cyt c550, compared to that measured by redox titrations [3, 6], may also occur when the protein binds to the periplasmic surface of PSII, and so the direct electrochemical measurement may be relevant to the biological structure and function of cyt c550.

Materials and methods Puri®cation of cyt c550 His-tagged PSII was prepared from late log-phase cultures of an engineered strain of Synechocystis PCC 6803 [16]. A detergentsolubilized extract was then prepared after the method of Tang and Diner [17], except that solubilization was made directly from broken cells without the additional step of preparing thylakoid membranes. After pelleting the unsolubilized material, the extract was incubated with Ni2+-NTA agarose (Qiagen) and unbound material was eluted with wash bu€er (50 mM MES, pH 6.0, 20 mM CaCl2, 5 mM MgCl2, 5 mM imidazole, 0.1% w/v b-dodecylmaltoside, 25% w/v glycerol). The column was then exchanged into wash bu€er lacking detergent and the cyt c550 and other extrinsic proteins were eluted using a high-salt bu€er (50 mM MES, pH 6.0, 1 M CaCl2, 5 mM MgCl2, 25% w/v glycerol). The eluate was subsequently concentrated by ultra®ltration (Centricon-3, Amicon) and exchanged into a low-salt bu€er (20 mM MES, pH 6.0, 10 mM NaCl). Cyt c550 was puri®ed by FPLC, using a MonoQ HR 5/5 column (10 mM MES, pH 6.0, 10 mM NaCl) and a 10±500 mM NaCl gradient. The puri®ed cyt c550 was ®nally concentrated to 5±10 mg/mL using a Centricon-3 concentrator. Samples were analyzed by SDS-PAGE on 15% acrylamide gels with 5.5 M urea according the procedure of Chua [18]. Bands were visualized with Coomassie blue. Densitometry of stained gels was performed on a BioRad 1650 densitometer. Optical measurements were performed using a Perkin-Elmer

Lambda 3B UV-visible spectrophotometer. For electrochemistry, the samples were in bu€ers containing 10 mM potassium phosphate (J.T. Baker) or 3-(N-morpholino)propanesulfonic acid (MOPS) (Sigma), pH 7.0, with either 10 mM or 0.1 M KCl as an electrolyte. Protein concentrations were determined by UV-visible spectroscopy using 549=25 mM±1 cm±1 [6]. Resonance Raman spectroscopy Resonance Raman spectra were obtained with a SPEX Raman 500 spectrometer equipped with a liquid nitrogen cooled CCD detector (ISA CCD-2000) using a Liconix 2040N He-Cd laser operating at 442 nm. The incident laser power was 34 mW. Rayleigh scatter from the 442 nm laser line was rejected by an Omega Filters sharpedge ®lter. The detection geometry was 90° using a quartz cuvette. The spectra were obtained at room temperature on samples in 10 mM phosphate bu€er at pH 7. The sample integrity was veri®ed by measurement of the UV-vis absorption spectrum after data collection. A baseline correction was done using Origin (version 6.0, Microcal Software). All features in the spectra were reproducible. EPR spectroscopy EPR measurements were performed on a Varian E-9 spectrometer (operating at a frequency of 9.28 GHz) equipped with a TE102 cavity and a helium-¯ow cryostat (ESR 900, Oxford Instruments). His-tagged Synechocystis PCC 6803 PS II and cyt c550 samples were oxidized with 100 lM ferricyanide prior to loading into 4 mm quartz EPR tubes (Wilmad Glass). Selective reduction of cyt b559 in His-tagged Synechocystis PSII was achieved by the addition of ascorbate from a 120 mM stock solution to a ®nal concentration of 4 mM. All spectra were acquired at 7 K with a modulation of 20 G and microwave power of 5 mW. Typically, 4±6 spectra were averaged for each reported spectrum. The g values were calibrated by using the YDá EPR signal in PSII as an internal reference. Electrochemistry Cyclic voltammograms (CVs) were run at 100 mV/s and square wave voltammograms (SWVs) were run at a pulse height of 50 mV and a frequency of 100 Hz, as previously described [19]. Unless otherwise stated, all reduction potentials are reported vs. NHE [20]. All SWV data included both a baseline and straight-line subtraction [21]. For the control experiments, horse heart cyt c (type VI) was obtained from Sigma. To run the electrochemistry experiments, the cell was ®rst ¯ushed with N2 for 15±30 min, after which the standard calomel electrode (SCE) and pyrolytic graphite edge electrode (PGE) were added, followed by the protein sample. Two procedures were used for the addition of protein. In the ®rst one, 20 lL of sample was added to the cell and experiments were conducted immediately, taking CVs and SWVs over time. In the second, the PGE electrode was immersed in the sample for 30 min, shaken o€, and then inserted into the cell, after which 20 lL of bu€er was added. During both the preparations and the experiments, a continuous ¯ow of N2 was maintained over the sample. After letting the sample equilibrate for 10 min, a trial CV was run to check for trace amounts of O2. Following this, the CV and/or SWV experiments were performed.

Results The isolation of cyt c550 from His-tagged PSII by NiNTA metal-anity chromatography is a novel technique that results in a high yield of pure protein. Shown in Fig. 1 is the densitometry scan of cyt c550 analyzed by

710

Fig. 1 Densitometry scan of a sample of puri®ed cyt c550 analyzed by SDS-PAGE. The positions of molecular weight standards are indicated, with their masses given in kDa. The apparent molecular mass of cyt c550 is 22 kDa

Fig. 3 EPR spectra of cyt b559 and cyt c550: a in a His-tagged Synechocystis PCC 6803 PS II sample (4.3 mg Chl/mL) oxidized with 100 lM ferricyanide; b in the same sample reduced with 4 mM ascorbate. c EPR spectrum of 126 lM cyt c550 puri®ed from Histagged Synechocystis PCC 6803 PS II. Also shown below these spectra is the magni®ed view of the high-®eld region highlighting the gXX turning point. The g2 region in the PSII spectra has been deleted to remove interference from the dark-stable tyrosine D EPR signal

Fig. 2 UV-visible absorption spectrum of 4.4 lM puri®ed cyt c550: oxidized (solid line) and 2 mM sodium dithionite reduced (dashed line). The arrow indicates the excitation wavelength of the laser illumination used for the resonance Raman spectra

SDS-PAGE. The apparent molecular mass of the protein, determined by a standard curve constructed from the positions of the molecular weight standards, is 22 kDa. There are no detectable contaminants in the sample. This is further demonstrated by the UV-vis spectrum of the oxidized and reduced forms of the protein (Fig. 2), which shows no contributions from other chromophores. Shown in Fig. 3 are EPR spectra of cyt b559 and cyt c550. As can be seen in Fig. 3a, His-tagged Synechocystis PCC 6803 PSII oxidized with ferricyanide displays overlapping EPR signals from both cyt b559 and cyt c550. Reduction of cyt b559 in the same His-tagged Synechocystis PSII sample by ascorbate yields EPR signals arising primarily from cyt c550 (Fig. 3b). Shown in Fig. 3c is an EPR spectrum of cyt c550 puri®ed from Histagged Synechocystis PCC 6803 PSII.

The high-frequency region of the resonance Raman spectrum of cyt c550 is shown in Fig. 4. It is typical of c-type cytochromes and consistent with a six-coordinate low-spin (6c,ls) ferric heme with bis-His axial ligation. A comparison of the vibrational modes of cyt c550 with those of the bis(N-methylimidazole) complex of ferric protoporphyrin IX dimethyl ester [Fe(III)PPIX-DME] [22] and the imidazole adduct of ferric microperoxidase8 [MP8(III)] [23] is presented in Table 1. The vibrational modes are assigned on the basis of the normal coordinate analysis of isotopically labeled cyt c [24]. The low-frequency region of the Raman spectrum (Fig. 5) is also typical for ferric (6c,ls) hemes [24]. These modes arise mostly from vibrations of the peripheral substituents, the pyrrole rings, and the Fe-N stretches. No substantial di€erences were observed between cyt c550 and other hemes besides a somewhat higher value of m9, a Fe-N(pyrrole) stretch. The complicated multiplet of peaks between 300 and 450 cm±1, characteristic of hemes, is readily apparent. Figure 6 shows the results from the SWV of cyt c550 in MOPS bu€er with 10 mM KCl electrolyte. A potential of E1/2=±108 mV with DEP=5 mV was measured.

711

Fig. 4 High-frequency region of the resonance Raman spectrum of 500 lM cyt c550 in 10 mM phosphate bu€er, pH 7.0. The peaks labeled S are from the solvent. Instrumental settings: excitation wavelength, 442 nm; incident laser power, 34 mW; integration time, 60 s; 180 scans

The conditions for the SWV measurements were determined by control experiments done with cyt c, using the well-developed procedures that have been reported for CV and SWV of cyt c [21, 25, 26, 27, 28]. Our results for cyt c agreed well with previous SWV results obtained for cyt c [28]. Typical literature values for the reduction potential of horse heart cyt c are 12±15 mV vs. SCE on surface-modi®ed gold electrodes [21, 25, 26, 27, 28]; using SWV, we measured a reduction potential of 15 mV vs. SCE. Following from these control measurements, experiments were done on cyt c550 using a 20 lL sample in 10 mM phosphate bu€er with 0.1 M KCl at pH 7.0 with no incubation of the sample. CVs of the bu€er showed the formation of redox signals arising from the electrode surface itself, with potentials around 240 mV. These were attributed to quinonoid species formed on the surface of the PGE. When using a 475 lM cyt c550 sample, in addition to the sharp quinonoid peaks, a broad redox couple was observed around ±90 mV. In order to obtain a better-resolved signal, SWV was performed on the same sample. In this case, the previously seen redox couple was observed, after background and straight-line subtraction, at E1/2=±115 mV with DEP=50 mV (DEP was measured as the di€erence between the reductive and oxidative SWVs, as described by Tong and Feinberg [29] and Bowden and co-workers [28]).

After cleaning the electrode by polishing and sonication, SWV of the bu€er solution showed some remaining signal arising from the protein, indicating that the active species may be adsorbed onto the electrode. This is comparable to the protein ®lm voltammetry used by Armstrong and co-workers [21, 30, 31], in which proteins and enzymes were electrostatically adsorbed onto PGE electrodes. The forward and reverse SWVs also supported this idea. As described by Osteryoung and O'Dea [32], when the reaction of an electroactive species is di€usion limited, the reverse SWV exhibits a substantially smaller re-reduction or re-oxidation current than the oxidation or reduction in the forward SWV. When the reaction is thin layer or surface bound, the forward and reverse SWVs look nearly the same, which is what we observed. Surface adsorption of the protein also explains why the currents measured are small (Fig. 6). Electron transfer proceeds only from those proteins that are bound to the electrode surface. Furthermore, it has been proposed that proteins bound to surface-modi®ed electrodes must adopt a preferred orientation on the surface for electron transfer to occur eciently [33]. Cleaning the electrode again by alumina polishing and brief sonication resulted in the loss of the signals arising from the protein. In order to investigate further the possibility of protein adsorption, the electrode was immersed in the protein solution for 30 min, transferred to the cell, and then 20 lL of phosphate bu€er was added. The experiments showed a redox couple at E1/2=±120 mV with DEP=80 mV. The reductive SWV shifted from ±100 to ±150 mV and increased in area, whereas the oxidative SWV decreased. Cleaning the electrode proved much more dicult in this case than without incubation. The regular combination of alumina polishing and brief sonication, repeated three times, did not remove the protein. However, the protein could be removed by immersing the electrode for 1 h in 2 M CaCl2, similar to the high-salt bu€er used for the initial isolation of the protein. Because it has been observed that phosphate may bind to other c-type cytochromes (presumably via positively charged lysine residues) and may cause an electrostatic change in the stabilization of the heme [34], the same experiments were repeated in MOPS bu€er. In this case, when adding sample to the electrode and immediately conducting SWV (without incubation), the potential was found to be E1/2=±125 mV with DEP=10 mV. The forward and reverse voltammograms indicated a fair amount of di€usion, particularly in the reductive SWVs. However, after a 30 min incubation of the electrode (as described above), the potential was found to be E1/2=±140 mV, with DEP=0 mV. In this case, the forward and reverse SWVs indicated substantial adsorption. As cyt c550 is released from PSII with a salt wash, and the PGE electrode can be cleaned by immersion in 2 M CaCl2, the in¯uence of the electrolyte on the degree of adsorption was also investigated. Results with only

712 Table 1 Comparison of Raman frequencies of cyt c550 with other bis-imidazole hemes

Normal modea

m10 m37 m2 m11 m38 m3 m4 m21 m13

m14 c22 m50 m8 m51 m9 c24

cyt c550 442 nmd

MP8(III)-(His/ImH)b 442 nmd

413 nmd

1637 1601 1584 1566 1554 1503 1372 1314 1229 1216 1191 1147 1129 439 415 404 362 341 318 303 281 238 221 206 177

1641 1603 1586 1564 1554 1505 1373 1318 ± ± ± ± ± ± 416

1637 1603 1588 1569 1554 1503 1375 1316 ± ± ± ± ± 445 416 404 362 344 318 ± 271 ± 225 211 ±

362 347 315 298 267 ± ± ± ±

Fe(III)PPIX DME-(MeIm)2c 413 nmd 1639 ± 1582 ± 1554 1506 1374 ± 1235 ± ± ± 1131 ± ± ± ± ± ± ± ± ± ± ± ±

a

Mode numbering according to [24] From [23]. MP8(III)-(His/ImH)=ferric microperoxidase 8 where the iron is coordinated by one histidine side chain and an exogenous imidazole ligand c From [22]. Fe(III)PPIXDME-(MeIm)2=bis(N-methylimidazole) complex of ferric protoporphyrin IX dimethyl ester d Excitation wavelength b

Fig. 5 Low-frequency region of the resonance Raman spectrum of cyt c550 (conditions are the same as in Fig. 4)

Fig. 6 Square wave voltammograms of 475 lM cyt c550 in 10 mM MOPS, pH 7.0, 10 mM KCl, after 30 min incubation at room temperature on a pyrolytic graphite edge electrode. The electrode surface is 2 mm2. A bu€er baseline and a straight line background correction have been subtracted from both the oxidative and reductive SWVs

713

10 mM KCl showed an increased current at E1/2=± 108 mV, with DEP=5 mV (Fig. 6). The forward and reverse SWVs indicated stronger adsorption than in the case of 0.1 M KCl.

Discussion As can be seen in the EPR spectrum of a ferricyanideoxidized sample of His-tagged Synechocystis PCC 6803 PSII, the g tensors of cyt b559 and c550 are very similar and result in signi®cant spectral overlap (Fig. 3a). Spectral deconvolution of these overlapping EPR signals is achieved by exploiting the di€erence in reduction potentials of cyt b559 and c550. Spectrophotometric redox titrations have shown that cyt c550 has a lower reduction potential than cyt b559 [3, 6, 35]. Thus, reduction of the sample with ascorbate results in signals primarily from cyt c550 (Fig. 3b). The assignment of g tensors for cyt b559 (gXX=1.50, gYY=2.22, and gZZ=3.01) and cyt c550 (gXX=1.50, gYY=2.23, and gZZ=2.88) in the PSII complex is con®rmed by the observation of the EPR spectrum of cyt c550 puri®ed from His-tagged Synechocystis PCC 6803 PSII (gXX=1.50, gYY=2.23, and gZZ=2.87) (Fig. 3c). It is notable that the gXX trough is clearly visible in all three EPR spectra. The EPR signals observed for both cyt b559 and cyt c550 are typical of b- and c-type low-spin hemes, respectively. Upon comparison with the crystal-®eld diagram proposed by Peisach and Blumberg [36], both cyt b559 and cyt c550 can be placed in the six-coordinate bis-His domain (category H) as cyt b559 and cyt c550 have rhombicity (V/D) and tetragonal ®eld (D/k) values of 0.73 and 3.685 and 0.772 and 3.098, respectively (calculated by the method of Palmer [37]). It has been shown by studies of the bis(2-methylimidazole) complexes of low-spin ferric porphyrins that the tetragonal and rhombic crystal-®eld splittings and the value of gZZ are sensitive to the relative orientations of the axial ligands [38]. The planes de®ned by the imidazole rings can lie either perpendicular (strained) or parallel (unstrained) with respect to each other. The value of gZZ is larger when the ligands are strained (gZZ3.4±3.8) than for the unstrained geometry (gZZ2.9). The values of gZZ, V/D, and D/k for cyt c550 are consistent with the histidine ligands adopting the unstrained orientation. Similar g tensors of free cyt c550 and cyt c550 bound to His-tagged Synechocystis PSII suggest that the ligation of the heme is equivalent in the free and bound forms. The g tensor for cyt c550 in His-tagged Synechocystis PCC 6803 PSII is also similar to the g tensor of cyt c549 puri®ed from wild-type Synechocystis PCC 6803 PSII [6] and A. nidulans [2]. However, the linewidths of the EPR signals of puri®ed c550 in the present study are much narrower than those in the EPR signals of NaCl-solubilized cyt c549 puri®ed using a DEAE column [2], suggesting that the puri®cation protocol used in the present study does not induce conformational heterogeneity in the sample. It has been seen previously that several

di€erent fractions of cyt c549 are obtained in a DEAE ion-exchange column chromatography puri®cation from A. nidulans. The metal-anity puri®cation of cyt c550 from His-tagged Synechocystis PCC 6803 PSII yields a single highly pure fraction of cyt c550. EPR spectra were also recorded of His-tagged Synechocystis PCC 6803 PSII in the presence of cyanide (spectra not shown). The EPR signals in the absence and presence of cyanide were identical, indicating that cyanide does not bind to cyt b559 and cyt c550 [3]. This suggests that the heme in cyt c550 in His-tagged Synechocystis PCC 6803 PSII is six-coordinate with tightly bound axial ligands. The ability to correlate the structure of a heme with its Raman spectrum has been well documented (see [39] and references therein). Several empirical relationships exist which relate the core size of the tetrapyrrole macrocycle as well as the spin and oxidation states of the metal to the Raman spectrum of the heme. The highfrequency portion of the resonance Raman spectrum of cyt c550 is shown in Fig. 4. The mode at 1584 cm±1 is assigned to m2 rather than m19 since totally symmetric vibrations are selectively enhanced by Soret excitation [40]. The mode m4 at 1372 cm±1, predominantly a Ca-N stretch [41], is sensitive to the heme p* orbital occupancy which is in¯uenced by p-backbonding from the iron [39, 42]. It upshifts from 1355 cm±1 in Fe(II) porphyrins to 1375 cm±1 upon oxidation to Fe(III) as electron density is removed from the heme p* orbital. The position of this mode in cyt c550, therefore, serves as the oxidation state marker for the Fe(III). The heme core size is an indicator of the spin state of the iron. In the low-spin form, only the bonding iron dp (t2g) orbitals are occupied, while in the high-spin case, there is occupation of the antibonding (eg) orbitals. Therefore, low-spin iron does not contribute electron density to the heme p* orbitals; the heme is slightly contracted, and so the heme center-to-N distance is short, in the range of 1.98±2.00 AÊ [39, 43]. It has been shown that the energies of high-frequency heme skeletal modes are especially sensitive to the core size, and an empirically determined equation was found which relates the two [44, 45, 46, 47, 48]. Application of this coresize relationship to several of the high-frequency skeletal modes of cyt c550 (m10, m37, m2, m3) results in a core size of 1.99 AÊ, which is characteristic of low-spin ferric hemes [43]. The mode at 1601 cm±1 is assigned to m37 by analogy to the Ni(II)OEP normal coordinate analysis [41], as well as the spectrum of the hemes in cytochrome c554, in which m37 is prominent [49]. The appearance of this mode is indicative of a deviation from D4h symmetry, which could arise from some non-planarity of the heme. Slight ru‚ing or saddling of the heme is supported by the frequency of the m10 mode at 1637 cm±1. It has been shown from Raman experiments and normal mode calculations on planar and non-planar Ni-porphyrins that high-frequency modes are sensitive to the planarity of the porphyrin [50, 51]. Deviations from planarity

714

cause a downshift in the m10 mode. This was demonstrated in the MP8(III) model system, in which the addition of imidazole to form the six-coordinate (His/ ImH) and (ImH)2 complexes caused m10 to upshift from 1637 to 1641 cm±1 as the saddling of the heme was relaxed [23]. The appearance of m10 at 1637 cm±1 in cyt c550 suggests that the heme is ru‚ed or saddled, as is common for many c-type cytochromes. Also of interest is the ligand-sensitive mode m11 at 1566 cm±1. This mode, primarily a Cb-Cb stretch, is modulated according to the r-donor properties of the axial ligands [52]. For example, in ferric c-type hemes, m11 appears at 1566 cm±1 when the axial ligand is an Ncontaining base but at 1562 cm±1 with S-containing bases [52, 53]. The appearance of m11 at 1566 cm±1 is a clear indicator of bis-His axial ligation in cyt c550. This is in agreement with the EPR results discussed above and with results obtained by NMR spectroscopy [54]. Recent studies on model porphyrins show that deviations of the macrocycle from planarity can decrease their reduction potentials substantially [55, 56]. The Raman spectrum of cyt c550 indicates that, while the heme shares the conservative ru‚ing of other c-type hemes [57], it is not so distorted as to be the origin of the extremely low reduction potential. In order to measure the redox properties of cyt c550 in the absence of redox mediators, we used direct electrochemistry with a PGE electrode. SWV was used because CV showed only a very weak signal. Although SWV has become a very powerful and widely applicable pulsed voltammetry technique, there are still relatively few instances where it has been used rather than CV [28, 32, 58]. As described in the Results, there is good evidence that the protein binds to the PGE surface and the E1/2 measured is for the bound protein. For example, the thin-layer behavior [32] and lack of a peak separation in MOPS bu€er after incubation of the sample [59] indicate adsorption. Also, the ratio between the currents of the forward and reverse voltammograms for the sample incubated in MOPS bu€er is close to 1. The most obvious non-ideality in the SWV in Fig. 6 is that the full-width at half-maximum is more than the theoretical value of 90 mV [28]. The non-ideality of the SWV response is partly due to the heterogeneous structure of the electrode surface. This causes adsorption heterogeneity and dispersion of both the reduction potential and the electron-transfer rate constants due to variations in the electron-transfer rate, which result from protein units which are bound in slightly di€ering orientations on the PGE [60, 61]. Although phosphate promotes adsorption, it can also result in non-ideal electron transfer and a deteriorated signal during incubation. The use of MOPS bu€er with a lower electrolyte concentration was found to improve electrostatic adsorption but without the drawbacks of phosphate. Three major factors contribute to the reduction potential of a heme group: (1) the nature of the axial ligands, (2) the solvent exposure of the heme, and (3) the

protein environment around the heme-binding site. Of these, the ®rst can change the reduction potential by 150±200 mV. For example, a change in the eclipsed or staggered orientation of the His plane versus the heme pyrrole nitrogens can cause variations in the potential of up to 200 mV [62]. The second contributing factor, solvent exposure, can have an even larger in¯uence. For example, changing the exposure of the heme to water from 21% in Rhodospirillum rubrum ferricytochrome c2 to 36% in Desulforomonas acetoxidans ferricytochrome c7 changes the reduction potential from 323 mV to ±102 mV, respectively. This is a net decrease of 425 mV brought about by only a 15% point change in the solvent exposure of the heme [63]. Conversely, decreasing the solvent dielectric by going from water to a nonaqueous solvent can increase the potential by as much as 250 mV. The third factor depends on the presence of charged amino acids near the heme and the way in which the heme macrocycle is distorted by steric crowding in the binding pocket. The results obtained by EPR and Raman spectroscopy show that the structure of the heme in cyt c550 and its axial ligands are very much like many other c-type cytochromes. Therefore, it does not appear that an unusual heme structure is the source of the low reduction potential of cyt c550. Rather, the degree of solvent exposure of the heme appears to be the major factor that determines the unusually low reduction potential. This is supported by a comparison of the reduction potentials determined for cyt c550 by redox titrations in solution (EM=±260 mV [3, 6]) to that observed by SWV for the protein adsorbed onto a PGE surface (E1/2=±108 mV, Fig. 6). This di€erence may re¯ect a change in the reduction potential when cyt c550 is bound to a surface. Adsorption onto the PGE surface may exclude water from the region of the heme-binding pocket, which would raise the reduction potential. It appears that cyt c550 from Synechocystis PCC 6803 has a more solventexposed heme when in solution, causing a lower reduction potential when measured by redox titration. Because spectroscopic measurements of proteins (particularly c-type cytochromes) adsorbed onto PGE surfaces indicate that they retain their native structures [31], it is unlikely that the potential measured by SWV is a€ected by a perturbation of the structure of cyt c550 when it is bound to the electrode surface. Changes in the reduction potential due to solvent exposure have been observed electrochemically for several cytochromes [64]. For example, the reduction potential of cyt b5 measured electrochemically on surface-modi®ed electrodes is 40±100 mV higher than when measured by other potentiometric methods, depending on the type of surface-modi®ed electrode used [65]. The authors of this study report that the degree of solvent exposure of the heme edge modulates the reduction potential, and that complexation of the protein onto the electrode surface excludes water from the region of the exposed heme surface. The reduction potential increases with decreasing heme surface

715

exposure due to the exclusion of water [64]. In addition, the reversible reduction of cyt c at a gold electrode showed a decrease in the reduction potential from 286 mV to ±200 mV as the concentration of guanidine was increased owing to the presence of unfolded cyt c [66]. It has been found that the reduction potential of unfolded cyt c is about ±200 mV, as compared to 260 mV for the folded protein [62]. Refolding of the protein involves burial of the heme. Our observation that the reduction potential of cyt c550 is higher when measured by SWV relative to that measured in solution is further evidence that cyt c550 does not unfold when it binds to the PGE surface. The recent crystal structure of PSII from Synechococcus elongatus at 3.8 AÊ resolution contains cyt c550 bound to the protein complex [15]. It can be seen that the heme edge is located very near the surface of the protein, and would probably be signi®cantly solvent exposed when cyt c550 is free in solution. However, when bound to PSII, the heme edge is oriented toward the membrane, facing the manganese-stabilizing protein and the periplasmic surface of PSII. Based on our work and the observations that solvent exclusion raises the reduction potential of other cytochromes [62, 64, 65], we propose that the reduction potential of cyt c550 when bound to PSII is signi®cantly more positive than measured by redox titrations. Using PGE, the protein is electrostatically bound to the electrode surface and the heme-binding site will be less exposed to water, resulting in a higher reduction potential. If a similar e€ect occurs upon the binding of cyt c550 to the PSII complex, then this value of the reduction potential may more closely re¯ect that of the protein in its physiological state. However, the value when cyt c550 is bound to the PGE is still well below the minimum of around 0 mV found for other c-type cytochromes. This may indicate that, even using direct electrochemistry and PGE electrodes, the heme could still be more solvent exposed than in vivo, and its physiological reduction potential may be even higher than that measured. It would certainly be easier to conceive of a biological function for cyt c550 if its reduction potential while bound to PSII was greater than ±108 mV. With an E1/2 below ±100 mV, there are no likely electron donors to cyt c550 on the periplasmic side of PSII; ferredoxin, a possible reductant, is found on the cytoplasmic side of the thylakoid membrane. Measurements of the reduction potential of cyt c550 bound to PSII would help to understand its function. However, such measurements are problematic because, under the highly reducing conditions required for such a measurement, the tetranuclear manganese cluster of PSII disassembles, which alters the binding anities of the extrinsic polypeptides to PSII [67]. Therefore, measurement of the reduction potential of cyt c550 in a PSII preparation by using a redox titration is likely to yield a value for the dissociated protein. However, it has been found that use of ascorbate as a reductant does not disrupt the manganese cluster [67]. Therefore, the data

shown in Fig. 3, which show that ascorbate does not reduce cyt c550 bound to PSII, demonstrate that the potential of cyt c550 bound to PSII is lower than approximately 0 mV. In closing, we note that the electrochemical measurement of the reduction potential of cyt c550 may underestimate the physiological value if cyt c550 is less solvent exposed in vivo than on the electrode surface. Thus, we conclude that the physiological potential of cyt c550 bound to PSII is likely to be between the upper limit of about 0 mV, de®ned by the lack of reduction by ascorbate in an intact PSI complex, and the value measured electrochemically of ±108 mV. Acknowledgements This work was supported by grant GM32715 from the National Institutes of Health (G.W.B.) and by grants from the National Science Foundation and the Camille and Henry Dreyfus Foundation (J.C.dP.). J.S.V. and M.J.R. contributed equally to this work.

References 1. Kerfeld CA, Krogmann DW (1998) Annu Rev Plant Physiol Plant Mol Biol 49:397±425 2. Hoganson CW, Lagenfelt G, AndreÂasson L-E (1990) Biochim Biophys Acta 1016:203±206 3. Holton RW, Myers J (1967) Biochim Biophys Acta 131:375± 384 4. Boussac A, Sugiura M, Inoue Y, Rutherford AW (2000) Biochemistry 39:13788±13799 5. Kuhl H, RoÈgner M, van Breeman JFL, Boekema EJ (1999) Eur J Biochem 266:453±459 6. Navarro JA, HervaÂs M, De la Cerda B, De la Rosa MA (1995) Arch Biochem Biophys 318:46±52 7. Alam J, Sprinkle J, Hermodson MA, Krogmann DW (1984) Biochim Biophys Acta 766:317±321 8. Pettigrew GW, Moore GR (1987) In: Cytochromes c: biological aspects. Springer, Berlin Heidelberg New York, pp 113±229 9. Krogmann DW, Smith S (1990) Curr Res Photosynth II:687± 690 10. Krogmann DW (1991) Biochim Biophys Acta 1058:35±37 11. Shen JR, Inoue Y (1993) Biochemistry 32:1825±1832 12. Shen JR, Vermaas W, Inoue Y (1995) J Biol Chem 270:6901± 6907 13. Shen JR, Ikeuchi M, Inoue Y (1997) J Biol Chem 272:17821± 17826 14. Shen JR, Qian M, Inoue Y, Burnap RL (1998) Biochemistry 37:1551±1558 15. Zouni A, Witt H-T, Kern J, Fromme P, Krauss N, Saenger W, Orth P (2001) Nature 409:739±743 16. Rei¯er MJ, Chisholm DA, Wang J, Diner BA, Brudvig GW (1998) In: Garab G (ed) Photosynthesis: mechanisms and effects. Kluwer, Dordrecht, pp 1189±1192 17. Tang X-S, Diner BA (1994) Biochemistry 33:4594±4603 18. Chua N-H (1980) Methods Enzymol 69:434±446 19. Kievit O, Brudvig GW (2001) J Electroanal Chem 497:139±149 20. Bard AJ, Faulkner LR (1980) Electrochemical methods. Wiley, New York 21. Armstrong FA (1997) In: Lenaz G, Milazzo G (eds) Bioelectrochemistry of biomacromolecules. BirkhaÈuser, Basel, pp 205± 255 22. Callahan PM, Babcock GT (1981) Biochemistry 20:952±958 23. Othman S, Le Lirzin A, Desbois A (1994) Biochemistry 33:15437±15448 24. Hu S, Morris IK, Singh JP, Smith KM, Spiro TG (1993) J Am Chem Soc 115:12446±12458

716 25. Albery WJ, Eddowes MJ, Hill HAO, Hillman AR (1981) J Am Chem Soc 103:3904±3910 26. Bond AM (1994) Inorg Chim Acta 226:293±340 27. Taniguchi I, Toyosawa K, Yamaguchi H, Yasukuochi K (1982) J Chem Soc Chem Commun 1032±1033 28. Reeves JH, Song S, Bowden EF (1993) Anal Chem 65:683±688 29. Tong J, Feinberg BA (1994) J Biol Chem 269:24920±24927 30. Armstrong FA (1990) Struct Bonding 72:137±219 31. Armstrong FA, Heering HA, Hirst J (1997) Chem Soc Rev 26:169±179 32. Osteryoung J, O'Dea JJ (1986) In: Bard AJ (ed) Electroanalytical chemistry. Dekker, New York, pp 209±308 33. Kasmi AE, Wallace JM, Bowden EF, Binet SM, Linderman RJ (1998) J Am Chem Soc 120:225±226 34. Koller KB, Hawkridge RM (1985) J Am Chem Soc 107:7412± 7417 35. Cramer WA, Whitmarsh J (1977) Annu Rev Plant Physiol 28:133±172 36. Peisach J, Blumberg WE (1971) In: Gould RF (ed) Bioinorganic chemistry. American Chemical Society, Washington, pp 271±291 37. Palmer G (1985) Biochem Soc Trans 13:548±560 38. Walker FA, Huynh BH, Scheidt WR, Osvath SR (1986) J Am Chem Soc 108:5288±5297 39. Spiro TG (1988) Resonance Raman spectra of heme and metalloproteins. Biological applications of Raman spectroscopy, vol 3. Wiley, New York 40. Rakshit G, Spiro TG (1974) Biochemistry 13:5317±5322 41. Abe M, Kitagawa T, Kyogoku Y (1978) J Chem Phys 69:4526± 4534 42. Spiro TG (1983) In: Lever ABP, Gray HB (eds) Iron porphyrins, part II. Addison-Wesley, Reading, Mass., pp 91±159 43. Collins DM, Countryman R, Hoard JL (1972) J Am Chem Soc 94:2066±2072 44. Ozaki Y, Iriyama K, Ogoshi H, Ochiai T, Kitagawa T (1986) J Phys Chem 90:6105±6112 45. Parthasarathi N, Hanson C, Yamaguchi S, Spiro TG (1987) J Am Chem Soc 109:3865±3871 46. Scholler DM, Ho€man BM (1979) J Am Chem Soc 101:1655± 1662

47. Spaulding LD, Chang CC, Yu N-T, Felton RH (1975) J Am Chem Soc 97:2517±2525 48. Spiro TG, Stong JD, Stein P (1979) J Am Chem Soc 101:2648± 2655 49. Heibel G, Griebenow K, Hildebrandt P (1991) Biochim Biophys Acta 1060:196±202 50. Desbois A (1994) Biochimie 76:693±707 51. Pi€at C, Melamed D, Spiro TG (1993) J Phys Chem 97:7441± 7450 52. Kitagawa T, Ozaki Y, Teraoka J, Kyogoku Y, Yamanaka Y (1977) Biochim Biophys Acta 494:100±114 53. Larsen RW, Chavez MD, Nunez DJ, Davidson MW, Kna€ DB, Krulwich TA, Ondrias MR (1990) Arch Biochem Biophys 283:266±270 54. Aguiar AP, Costa HS, Louro RO, Xavier AV, Turner DL (1998) Inorg Chim Acta 273:196±200 55. Barkigia KM, Chantranupong L, Smith KM, Fajer J (1988) J Am Chem Soc 110:7566±7567 56. Ravikanth M, Chandrashekar TK (1995) Struct Bonding 82:105±188 57. Hobbs JD, Shelnutt JA (1995) J Protein Chem 14:19±25 58. Willit JL, Bowden EF (1990) J Phys Chem 94:8241±8246 59. Smith ET, Feinberg BA (1990) J Biol Chem 265:14371±14376 60. Nahir TM, Bowden EF (1996) J Electroanal Chem 410:9±13 61. Bowden EF, Clark RA, Willit JL, Song S (1993) In: Schultz FA, Taniguchi I (eds) Redox mechanisms and interfacial properties of molecules of biological importance. The Electrochemical Society, Pennington, NJ, pp 34±45 62. Cusanovich MA, Meyer TE, Tollin G (1988) In: Eichhorn GL, Marzilli LG (eds) Heme proteins. Elsevier, New York, pp 37±91 63. Dolla A, Blanchard L, Guerlesquin F, Bruschi M (1994) Biochimie 76:471±479 64. Tezcan FA, Winkler JR, Gray HB (1998) J Am Chem Soc 120:13383±13388 65. Rivera M, Seetharaman R, Girdhar D, Wirtz M, Zhang X, Wang X, White S (1998) Biochemistry 37:1485±1494 66. Bixler J, Bakker G, McLendon G (1992) J Am Chem Soc 114:6938±6939 67. Stewart DH (1998) PhD Thesis, Yale University