Site-directed Mutagenesis of Cytochrome c6 from Anabaena Species PCC 7119 IDENTIFICATION OF SURFACE RESIDUES OF THE HEMEPROTEIN INVOLVED IN PHOTOSYSTEM I REDUCTION*

A number of surface residues of cytochrome c6 from the cyanobacterium Anabaena sp. PCC 7119 have been modified by site-directed mutagenesis. Changes were made in six amino acids, two near the heme group (Val-25 and Lys-29) and four in the positively charged patch (Lys-62, Arg-64, Lys-66, and Asp-72). The reactivity of mutants toward the membrane-anchored complex photosystem I was analyzed by laser flash absorption spectroscopy. The experimental results indicate that cytochrome c6 possesses two areas involved in the redox interaction with photosystem I: 1) a positively charged patch that may drive its electrostatic attractive movement toward photosystem I to form a transient complex and 2) a hydrophobic region at the edge of the heme pocket that may provide the contact surface for the transfer of electrons to P700. The isofunctionality of these two areas with those found in plastocyanin (which acts as an alternative electron carrier playing the same role as cytochrome c6) are evident.

A number of surface residues of cytochrome c 6 from the cyanobacterium Anabaena sp. PCC 7119 have been modified by site-directed mutagenesis. Changes were made in six amino acids, two near the heme group (Val-25 and Lys-29) and four in the positively charged patch (Lys-62, Arg-64, Lys-66, and Asp-72). The reactivity of mutants toward the membrane-anchored complex photosystem I was analyzed by laser flash absorption spectroscopy. The experimental results indicate that cytochrome c 6 possesses two areas involved in the redox interaction with photosystem I: 1) a positively charged patch that may drive its electrostatic attractive movement toward photosystem I to form a transient complex and 2) a hydrophobic region at the edge of the heme pocket that may provide the contact surface for the transfer of electrons to P 700 . The isofunctionality of these two areas with those found in plastocyanin (which acts as an alternative electron carrier playing the same role as cytochrome c 6 ) are evident.
In cyanobacteria and green algae, cytochrome (Cyt) 1 c 6 acts as a soluble redox carrier that can replace plastocyanin in the transport of electrons between the two membrane-embedded complexes Cyt b 6 f and photosystem I (PSI) (see Refs. 1 and 2 for reviews). In higher plants, however, the copper protein is the only electron carrier. The cyanobacterium Anabaena, like some other organisms, is able to synthesize either Cyt c 6 or plastocyanin (which both exhibit a basic isoelectric point of ϳ9) as a function of copper concentration in the growing medium (3).
The structures and functions of these two metalloproteins have been extensively studied in a wide range of organisms (4 -11). According to our laser flash-induced kinetic studies, the reaction mechanism of PSI reduction follows three different models: type I, which involves an oriented collision between the two redox partners; type II, which proceeds through the formation of a transient complex prior to electron transfer; and type III, which requires an additional rearrangement step so as to make the redox centers orientate properly within the complex (9). During the evolution of photosynthetic organisms, interaction between a positively charged Cyt c 6 and PSI was first optimized (as is the case with Anabaena Cyt c 6 , which follows the type III mechanism) and only later in evolution was a more complex kinetic mechanism developed with plastocyanin (2).
The three-dimensional structures of Cyt c 6 from the two eukaryotic green algae Monoraphidium (12) and Chlamydomonas (13) and from the cyanobacterium Synechococcus (14) have been solved. The analysis of the Cyt c 6 molecule compared with the plastocyanin structure allowed us to identify a hydrophobic region around the solvent-exposed heme propionates that resembles the north pole of plastocyanin as well as a negatively charged patch in eukaryotic Cyt c 6 similar to the east face of eukaryotic plastocyanin (12). In Anabaena, neither Cyt c 6 nor plastocyanin exhibits the acidic patches at their east face, which is rather positively charged (2).
Extensive mutagenesis studies of plastocyanin have supplied relevant information on the role of specific residues located both in its north and east faces (5,8,(15)(16)(17). Recently, we performed a site-directed mutagenesis analysis of Synechocystis Cyt c 6 (18). Aspartates 70 and 72 appear to be located in a negatively charged region of Cyt c 6 that may be isofunctional with the well known "south-east" acidic patch of plastocyanin. In addition, Phe-64 (which is close to the heme group and could be the counterpart of Tyr-83 in plastocyanin (19)) does not appear to be involved in the electron transfer to PSI. In contrast, Arg-67, which is located at the edge of the Cyt c 6 acidic area, seems to be crucial. This paper reports the kinetic and thermodynamic characterization of PSI reduction by a set of site-directed mutants of Anabaena Cyt c 6 . The results demonstrate that a single mutation of specific residues in the hydrophobic or positively charged area of Cyt c 6 can promote drastic changes in the reaction mechanism. The two functional areas of Cyt c 6 involved in PSI photoreduction have been identified.

EXPERIMENTAL PROCEDURES
Purification of Native Cytochrome c 6 -The hemeprotein from Anabaena sp. PCC 7119 was purified as described previously (11), but with slight modifications. Cyt c 6 samples were applied to a CM-cellulose column after oxidation with potassium ferricyanide. Elution of the adsorbed proteins was performed with a 50 M potassium ferricyanide solution in a linear gradient of 2-30 mM potassium phosphate buffer, pH 7.0. The fractions containing pure Cyt c 6 were pooled, concentrated on an Amicon pressure-dialysis cell fitted with a YM-10 membrane, and stored at Ϫ80°C. Cyt c 6 concentration was determined spectrophotometrically using an absorption coefficient of 26.2 mM Ϫ1 cm Ϫ1 at 553 nm for the reduced protein (20).
Construction of Mutants-The mutant petJ genes were constructed by polymerase chain reaction with the QuickChange kit (Stratagene) using oligonucleotides of 32 base pairs, 15 ng of DNA templates, and 12 min of extension time. The previously described expression construction for the petJ gene from Anabaena (21) was used as a template. The DNA sequencing service MediGenomix carried out the nucleotide sequence analysis. Other molecular biology protocols were standard (22).

Production of Recombinant Proteins and Purification
Procedures-Escherichia coli MC1061-transformed cells were grown in M9 medium (22) supplemented with 1 g/liter Tryptone, 6 mg/liter Fe(III) ammonium citrate, and 100 g/ml ampicillin. Cells from 10-liter microaerobic cultures were collected, and the periplasmic fraction was extracted according to the method of Hoshino and Kageyama (23) as modified by Eftekhar and Schiller (24). The resulting suspension was extensively dialyzed against 2 mM potassium phosphate, pH 7.0. From this point, the purification procedure was that for native Cyt c 6 (see above), with the exception of mutants K66E and D72K, which were eluted with buffer gradients ranging from 1 to 10 mM and from 2 to 120 mM, respectively. In all cases, 50 M potassium ferricyanide was added to the gradient solutions to keep Cyt c 6 oxidized. Protein concentration was determined as described previously (21).
Redox Titrations-The redox potential value for the heme group in each Cyt c 6 mutant was determined as reported previously (21,25), for which the differential absorbance changes at 553 minus 570 nm were followed. Errors in the experimental determinations were less than Ϯ5 mV.
Preparation of PSI Particles-PSI particles were isolated from Anabaena cells by ␤-dodecyl maltoside solubilization (26,27). The chlorophyll/P 700 ratio of the resulting PSI preparations was 140:1. The P 700 content in PSI samples was calculated from the photoinduced absorbance changes at 820 nm using the absorption coefficient of 6.5 mM Ϫ1 cm Ϫ1 determined by Mathis and Sétif (28). Chlorophyll concentration was determined according to Arnon (29).
Laser Flash Absorption Spectroscopy-The kinetics of flash-induced absorbance changes in PSI were followed at 820 nm as described previously (9,18). Experimental conditions and the standard reaction mixture were also as reported previously (17); the buffer used throughout this work was 20 mM Tricine/KOH, pH 7.5. Unless otherwise indicated, low and high ionic strengths refer to the absence and addition of 10 mM MgCl 2 , respectively. Data collection and kinetic and thermodynamic analyses were carried out as reported by Hervás et al. (9,10). Apparent thermodynamic parameters were estimated as described Díaz et al. (6) by fitting the experimental data to the Watkins equation (30). The values for the rate constant for electron transfer and K A (see below) were determined according to the formalism by Meyer et al. (31).
Structure Simulation-The structures of WT and mutant Cyt c 6 were modeled using the SYBYL program (Tripos Associates) in an SGI RC10000 workstation. The three-dimensional crystal structure of Cyt c 6 from the green alga Monoraphidium braunii (12) was used as a template. Sequence alignment and subsequent amino acid substitution were performed with the BIOPOLYMER module of SYBYL Version 6.4. Force field parameters for the heme moiety were those in the AMBER package (32). The resulting file was first submitted to energy minimization in vacuo up to a root mean square energy gradient of 0.41 kJ mol Ϫ1 Å Ϫ1 using the SANDER module of AMBER Version 4.1 (33). During these calculations, the backbone heavy atoms of ␣-helix regions were restrained at their position by a harmonic force of 62.7 kJ mol Ϫ1 Å Ϫ1 . Then, the whole system was solvated with three-point water molecules using the BLOB option of the EDIT module. Solvent was energyminimized and submitted to a 9-ps molecular dynamics calculation. The whole system was again energy-minimized and submitted to a 1250-ps molecular dynamics run at 300 K. A total of 10 samples from the last 400 ps of trajectory were quenched by freezing the system in six steps of 1.5 ps. The qualities of the resulting structures were tested using the PROCHECK program (34). Surface electrostatic potentials were estimated using the algorithm of Nicholls and Honig (35), as indicated in the MOLMOL program (36).

RESULTS
To analyze the role of specific residues of Anabaena Cyt c 6 in the reaction mechanism of PSI reduction, six amino acids (two near the heme group and four in the positively charged east face) were chosen for mutations ( Fig. 1). At the edge of the heme crevice, modifications were made at residues 25 and 29: Val-25, which is located near the heme ␤-meso-position, was substituted by alanine and glutamate; and Lys-29, which exhibits its side chain lying close to propionate 7, was replaced with histidine. Val-25 is located in the middle of the hydrophobic patch, but Lys-29 is not. However, mutation of the latter to histidine was considered to be interesting because the residue at position 29 is lysine in all Cyts c 6 with the exception of that from Monoraphidium, in which it is histidine (actually, the EPR spectra of Monoraphidium Cyt c 6 suggested an unusual histidine-histidine axial coordination for the heme iron, a ligand system that is not possible in the rest of Cyts c 6 with just one histidine residue (37)). In addition, basic residues at positions 28 -30 in type I cytochromes have been proposed to con-  6 showing the residues modified by mutagenesis. The molecule is oriented to show the six residues mutated; the "east" positively charged patch (equivalent to the acid face of eukaryotic cytochrome c 6 and plastocyanin) is in front, whereas the "north" hydrophobic patch is on the left. The heme group is depicted in green.
trol the redox potential of the heme group through stabilization of propionate 7 (38). On the east face, modifications included the replacement of Lys-62 and Lys-66 by glutamates; Asp-72, which is also located in the middle of the east patch, was changed to lysine. Finally, Arg-64, which is at the edge of the east patch and has been shown to be crucial in Synechocystis Cyt c 6 (18), was replaced with glutamate.
The EPR and electronic absorption spectra of Anabaena Cyt c 6 were not changed by the mutations (data not shown), but its midpoint redox potential (E m ) could be significantly affected ( Table I). The E m value was unchanged when the residue mutated was at the east face, including Arg-64, but it was ϳ50 mV lower when the mutations were located near the heme group. Only minor differences were observed in the 1 H NMR spectra and nuclear Overhauser effect intensities of heme resonances with protons from other residues. 2 The kinetics of PSI reduction by WT Cyt c 6 are biphasic, but those with the mutants (with the exception of D72K) are monoexponential, lacking the fast phase typically observed with the WT species. As shown in Fig. 2, the kinetics corresponding to D72K and WT Cyt c 6 exhibited a sharp initial fast phase (with a rate constant that was independent of donor protein concentration) followed by a slower decay. In contrast, the oscilloscope traces with mutants V25A and V25E fit to single exponential curves. Fig. 3 shows that the observed pseudo first-order rate constant (k obs ) of PSI reduction by any mutant (with the exception of D72K) varied linearly with Cyt c 6 concentration. This can be interpreted by assuming that there is no formation of any stable complex between PSI and Cyt c 6 , and the reaction thus follows a collisional kinetic mechanism (type I). With mutant D72K, however, the protein concentration dependence of k obs exhibited a saturation profile, which indicates the formation of a bimolecular Cyt c 6 ⅐PSI complex prior to electron transfer. The extrapolated rate constant at infinite D72K concentration is similar to its electron transfer rate constant, which is obtained directly from the fast kinetic phase, thus suggesting a type II mechanism.
Such a saturation profile with D72K was even more evident at lower ionic strengths, which increase attractive electrostatic interactions. As shown in Fig. 4, the D72K mutant showed efficient complex formation, with an equilibrium constant (K A ) of 1.06 ϫ 10 5 M Ϫ1 at low ionic strength. Extrapolation of the observed rate constant to infinite D72K concentration yielded a value that approached one-half the experimental electron transfer rate constant, which indicates that this mutant may follow a type III mechanism at low ionic strength. This finding also suggests that the rearrangement step (see above) is not limiting at high ionic strength. Fig. 4 also shows that the V25A mutant was likewise able to form a transient complex with PSI at low ionic strength, even though its kinetic profiles of PSI reduction were monophasic at any ionic strength. The K A value for complex formation between V25A and PSI is 1.07 ϫ 10 4 M Ϫ1 , which is 10 times lower than that with D72K. These data indicate that V25A follows a type II mechanism in the absence of MgCl 2 and a type I mechanism when 10 mM MgCl 2 is added.
The bimolecular rate constant for the overall reaction (k bim ) of PSI reduction, which can be calculated from the linear plots in Fig. 3, is smaller with any mutant than with WT Cyt c 6 ( Table I). In the hydrophobic patch, replacement of Val-25 with alanine or glutamate promoted a decrease in k bim of ϳ2 and 60  times, respectively, whereas mutation of Lys-29 to histidine involved a parallel decrease of ϳ50% as compared with WT Cyt c 6 . Substitution of Arg-64, in its turn, by glutamate induced the rate constant to decrease by a factor of 5. In the east face, replacement of Lys-62 and Lys-66 with glutamate made the k bim value decrease to half that of WT Cyt c 6 . In the case of D72K, it should be noted that the bimolecular rate constant in Table I refers to its association constant, which is kinetically different from the other k bim values in the table. Its electron transfer rate constant, calculated directly from the fast phase, compares well with that of WT Cyt c 6 , as does the maximum percentage of fast phase (Table I).
Taking into account the electrostatic nature of the interaction of Cyt c 6 with PSI, a detailed analysis of the effect of ionic strength on k bim was performed. Fig. 5 shows that the k bim values with WT Cyt c 6 monotonically diminished with increasing NaCl concentration, thereby indicating the existence of attractive electrostatic interactions between the reaction partners as described previously (10,11). A similar effect of ionic strength on k bim was observed with all mutants, but some differences could be found among them. Actually, the k bim values with mutants V25A and D72K are significantly higher than that with WT Cyt c 6 at low ionic strength, but they decreased drastically upon small additions of NaCl. Mutant K29H showed an ionic strength dependence similar to that of WT Cyt c 6 , although its electron transfer efficiency was lower in the whole range analyzed. The k bim values with all the other mutants are lower than that with WT Cyt c 6 , indicating that the changes in net electrostatic charge alter the attractive interactions between PSI and mutant proteins.
Using the Watkins equation (30), the bimolecular rate constant extrapolated to infinite ionic strength (k inf ) (which facilitates the analysis of the intrinsic reactivity of redox partners in the absence of electrostatic interactions) can be calculated from the experimental data. As shown in Table I, the k inf values with Cyt c 6 mutated at the positively charged patch are very similar to that with WT Cyt c 6 , a fact that can be explained by assuming that the changes in reactivity induced by these mutations are mainly due to electrostatic, and not structural, effects. The only exception is R64E, for which the k inf value is 4 -5 times lower than that with WT Cyt c 6 , a finding suggesting that the change in electrostatic charge is not the only factor affecting its reactivity toward PSI, as reported previously for Synechocystis Cyt c 6 (18). The effect of the K29H mutation seems to be merely electrostatic in nature, as k inf is similar to that with WT Cyt c 6 . The two mutations at position 25 involve modifications that make k inf lower (the V25E mutant, in particular, exhibits a k inf value that is 70 times lower than that with WT Cyt c 6 ).
To gain further insights into the nature of the interactions between PSI and Cyt c 6 , a thermodynamic analysis of PSI reduction by the Cyt c 6 mutants was performed. In all cases, the temperature dependence of the observed rate constant (k obs ) yielded linear Eyring plots with no breakpoints, from which the values for the apparent activation enthalpy (⌬H ‡ ), entropy (⌬S ‡ ) and free energy (⌬G ‡ ) of the overall reaction could be calculated. For comparative purposes, Table II shows the differences in such activation parameters between WT Cyt c 6 and every mutant. The greatest difference was observed with V25E, whose free energy change is 9.41 kJ mol Ϫ1 higher than that of WT Cyt c 6 , as expected from its inefficient interaction with PSI. This difference is due mainly to a decrease in the entropic term by ϳ28 J mol Ϫ1 K Ϫ1 . Also interesting is mutant R64E, whose free energy change is 4.15 kJ mol Ϫ1 higher than that of WT Cyt c 6 , a fact that is due to changes in both the enthalpic and entropic terms. Mutant V25A behaved differently than any other mutant; in fact, the free energy term of its reaction is similar to that of WT Cyt c 6 despite the fact that both entropy and enthalpy show dramatic changes as compared with the thermodynamic parameters of WT Cyt c 6 .
To check whether the differences in ⌬G ‡ between WT and mutant Cyt c 6 (⌬⌬G ‡ ) were due to electrostatic interactions, the experimental data were fitted to the Watkins equation (30). As shown in Fig. 6, the ⌬⌬G ‡ values with D72K and V25A perfectly fit the Watkins equation, and those of K29H, K62E, and K66E roughly fit it. Mutants V25E and R64E showed a linear NaCl dependence of ⌬⌬G ‡ at high ionic strength that caused the data to deviate from the Watkins equation. DISCUSSION Up to now, the only site-directed mutational analysis of any Cyt c 6 was recently reported by De la Cerda et al. (18) in the cyanobacterium Synechocystis. This hemeprotein, which is almost neutral, reacts with PSI according to a simple collisional model (type I). On the contrary, Anabaena Cyt c 6 is a positively charged protein that exhibits strong electrostatic attractions toward PSI (9,11) and that reacts with the photosystem following a more complex three-step mechanism (type III). The FIG. 4. Dependence upon hemeprotein concentration of k obs for PSI reduction by WT cytochrome c 6 and its mutants V25A and D72K at low ionic strength. The solid lines for the mutants correspond to theoretical fittings according to Meyer et al. (31). The experimental conditions were as described in the legend to Fig. 2, except that magnesium chloride was omitted from the reaction mixture.
FIG. 5. Effect of ionic strength on k bim for PSI reduction by WT cytochrome c 6 and its mutants. Experimental conditions were as described in the legend to Fig. 4, but the ionic strength was adjusted to the desired value by adding small amounts of a concentrated NaCl solution.
goal of this study was to elucidate the role played by some specific amino acids of Anabaena Cyt c 6 in such an attractive interaction with PSI and to investigate the possible involvement of its hydrophobic north area in the type III reaction mechanism.
Mutations of Val-25 indicate that this residue may contribute to the specific topology of the north hydrophobic area of Cyt c 6 in its interaction with PSI. Similar conclusions were inferred from mutants at the north pole of eukaryotic plastocyanin (5,8), in which the fast phase of electron transfer to PSI cannot be detected. This suggests that the mutant proteins (both Cyt c 6 and plastocyanin) are unable to reach the optimal orientation required for their redox centers to transfer electrons to PSI. It is interesting to compare the drastic effect induced by mutation of Val-25 to glutamate with mutation to alanine; the severe kinetic phenotype of V25E can be attributed to the presence of a negative charge in the middle of a normally hydrophobic region.
The K29H mutant possesses a redox potential value that is 65 mV more negative than that of WT Cyt c 6 . This is probably due to the proximity of Lys-29 to heme propionate 7, in agreement with previous reports (38). Even though the driving force of electron transfer from K29H to PSI is higher, the kinetic profile of PSI reduction by the mutant loses the first fast phase, and its k bim value is about half that of WT Cyt c 6 . This can be ascribed in part to electrostatic effects, as the dependence of ⌬⌬G ‡ on ionic strength fits the Watkins equation (30), and the value for k inf compares well with that of WT Cyt c 6 . The change in redox potential induced by the mutation clearly does not affect reactivity with PSI, suggesting that complex formation is the rate-limiting step of the overall reaction.
Mutations of Lys-62 and Lys-66 at the positively charged patch of Cyt c 6 demonstrate that this region is responsible for the attractive electrostatic interactions with PSI. In Anabaena, the PsaF subunit does not seem to be directly involved in the interaction of PSI with its donor proteins, as is the case in Synechocystis (16,39). Hence, the positive charges of Anabaena Cyt c 6 may interact with certain negatively charged areas in the PsaA/PsaB heterodimer of PSI to form a transient electrostatic complex. The surface electrostatic potential of WT and mutant Cyt c 6 was then calculated. Fig. 7 shows that the WT molecule has an extensive area of positive potential at its west face, i.e. close to the hydrophobic patch at the edge of the heme  cleft. Mutants K66E, K62E, and R64E present a reduction in charge of the positive patch, although the change in the orientation of the dipole moment is Ͻ10°. This is consistent with the kinetic behavior exhibited by these three mutants, in which the ionic strength dependence of the interaction with PSI is not so much evident. It is noteworthy that the effect of such mutations on the reaction rates can be closely correlated to changes in the positive electrostatic potential at the edge of the hydrophobic area. In fact, replacement of Arg-64 (which is located in the basic patch, adjacent to the hydrophobic region) by an acidic residue promotes a drastic reduction in size of the positive patch. Mutation of Lys-66 to glutamate likewise induces a diminution of the basic patch, but it does not alter the electrostatic surface potential at the edge of the hydrophobic area.
Mutant D72K is of special relevance as it clearly supports our proposal that the positively charged patch of Cyt c 6 is responsible for the interaction with PSI. Replacement of Asp-72 by a positive residue like lysine makes the positive potential region spread out (Fig. 7), thus favoring the attractive interactions with any negatively charged area of PSI. Like WT Cyt c 6 , D72K is the only mutant that exhibits the fast kinetic phase of PSI reduction.
The thermodynamic analysis of PSI reduction by the mutants of Cyt c 6 revealed that all mutations induce changes in ⌬H ‡ and ⌬S ‡ , but only some of them (V25A, V25E, and R64E) are significant. The large entropic changes of such mutants suggest that the solvent molecules are tuning the reactivity of the donor proteins toward PSI. The kinetic behavior of V25A indicates that the electrostatic interactions are predominant over hydrophobic forces.
To conclude, the experimental data reported here reveal that in Anabaena Cyt c 6 , there are at least two interaction sites (which would be isofunctional with two similar areas in plastocyanin) for electron transfer to PSI: (i) a positively charged area, responsible for the electrostatic interactions forming the transient complex with the photosystem; and (ii) a hydrophobic region surrounding the heme pocket, responsible for the formation of the contact interphase with PSI allowing electrons to go from the heme iron to P 700 ϩ . The main difference with respect to Synechocystis Cyt c 6 (18) is found at the electrostatically charged patch, which explains the differences in their respective reaction mechanisms of PSI photoreduction (10).