Site-directed Mutagenesis of Cytochromec 6 from Synechocystis sp. PCC 6803

This paper reports the first site-directed mutagenesis analysis of any cytochrome c 6, a heme protein that performs the same function as the copper-protein plastocyanin in the electron transport chain of photosynthetic organisms. Photosystem I reduction by the mutants of cytochromec 6 from the cyanobacteriumSynechocystis sp. PCC 6803 has been studied by laser flash absorption spectroscopy. Their kinetic efficiency and thermodynamic properties have been compared with those of plastocyanin mutants from the same organism. Such a comparative study reveals that aspartates at positions 70 and 72 in cytochrome c 6 are located in an acidic patch that may be isofunctional with the well known “south-east” patch of plastocyanin. Calculations of surface electrostatic potential distribution in the mutants of cytochromec 6 and plastocyanin indicate that the changes in protein reactivity depend on the surface electrostatic potential pattern rather than on the net charge modification induced by mutagenesis. Phe-64, which is close to the heme group and may be the counterpart of Tyr-83 in plastocyanin, does not appear to be involved in the electron transfer to photosystem I. In contrast, Arg-67, which is at the edge of the cytochrome c 6 acidic area, seems to be crucial for the interaction with the reaction center.

plexes. Although cyanobacteria and eukaryotic green alga can synthesize both Pc and Cyt, Pc seems to have been able to replace the primitive Cyt along evolution of photosynthetic organisms as the copper-protein is the only electron carrier in higher plants. The two metalloproteins are now well characterized, both at the structural and functional levels. Their threedimensional structures have been solved by x-ray crystallography and NMR spectroscopy in several organisms, and their reaction mechanisms have been widely investigated (see Ref. 1 for a recent review).
Alignment of eukaryotic Pc and Cyt molecules according to their dipole moment and surface charge distribution has allowed us the observation of areas in the heme protein similar to those previously reported in Pc: i) a hydrophobic region around the surface-exposed heme edge that could be equivalent to the north patch involving the copper ligand His-87 in Pc, and ii) a negatively charged area in Cyt similar to the east acidic patch around Tyr-83 in Pc (1,2). It should be noted that such an acidic patch is significantly smaller in the metalloproteins isolated from prokaryotes, namely the cyanobacterium Synechocystis sp. PCC 6803 in which the acidic patch is rather southeast-facing, just below Tyr-87.
In eukaryotic systems, site-directed mutagenesis of Pc has supplied relevant information on the role of specific residues in site 1 (or the hydrophobic north pole) and site 2 (or the acidic east face), which both have been proposed to be involved in the interaction with PSI (3)(4)(5)(6). Whereas negative residues in the east face seem to be responsible for electrostatic interactions and complex formation with the positively charged PsaF subunit of PSI, the transfer of electrons from the Pc copper center to the chlorophyll dimer P700 ϩ in PSI could take place via the imidazole ring of His-87 (one of the four ligands to the copper atom) in the north pole.
In Synechocystis, in which wild type (WT) Cyt and Pc interact with PSI according to a simple oriented collisional reaction mechanism (7), we have recently carried out a functional sitedirected mutagenesis analysis of Pc (8). This study showed that most of the mutants react following the same model as the WT protein, with the only exception of the double mutant D44R/ D47R which is able to form a kinetically detectable electrostatic complex with PSI at low ionic strength, as do the eukaryotic copper proteins (9,10). These findings indicate that the reaction mechanism of PSI reduction can be drastically modified by changing specific surface amino acids, mainly the acidic residues in the east face of the protein (8).
This paper describes the kinetic and thermodynamic properties of a set of mutants of Synechocystis Cyt, which has been modified in two specific residues located in a region equivalent to the east face of Pc, as well as in two others close to the heme environment. To the best of our knowledge, this is the first mutagenesis study of any Cyt reported up to now. The thermodynamic parameters of PSI reduction by Cyt mutants are also analyzed in a comparative way with a number of mutants of Pc modified in the east face.

EXPERIMENTAL PROCEDURES
DNA Techniques-The petJ gene coding for Cyt was cloned in pBluescript II (SK ϩ ) (Stratagene) as described previously (11), thereafter being used as a template for the site-directed mutagenesis protocol. The mutant genes were constructed using the polymerase chain reaction in two steps (12). Cloning and sequencing of the modified petJ genes were carried out as reported previously (8) for Pc mutagenesis. Cloning and sequencing of the modified petE genes encoding Pc have already been published (8). The numbering of residues of Synechocystis Cyt herein used is that corresponding to the heme protein from Monoraphidium braunii (2).
Recombinant Proteins and PSI Particles-The procedures for production and purification of mutant metalloproteins (both Cyt and Pc) were those previously described (8,11) except that the recombinant Cyt was produced in Escherichia coli MC1061 cells (rather than in DH5␣ cells); the final yield of Cyt was 5-10-fold higher in the MC1061 strain. PSI particles were purified by ␤-dodecyl maltoside solubilization as described previously (7).
Redox Titrations-Redox titrations of WT Cyt and its mutants were performed in a dual wavelength spectrophotometer as described previously (8) for Pc, except that differential absorbance changes were monitored at 552Ϫ570 nm. Errors in experimental determinations were less than 5 mV.
Laser Flash Absorption Spectroscopy-Kinetics of flash-induced absorbance changes in PSI were followed at 820 nm as described previously (9). The standard reaction mixture and experimental conditions were as described previously (8), the buffer being 20 mM Tricine/KOH, pH 7.5; for experiments at pH 5.5, the buffer used was 20 mM MES/ NaOH. In all cases, the standard reaction mixture contained 10 mM MgCl 2 , which was omitted in experiments for the analysis of the ionic strength effect. For the thermodynamic analyses, experiments were run at varying temperatures (10). Data collection, as well as kinetic and thermodynamic analyses were as described previously (9,10). Apparent thermodynamic parameters were estimated as in Díaz et al. (13). Electrostatic potential energies were obtained by fitting the calculated apparent activation energies to the Watkins equation (14). The errors in estimated values for the observed pseudo first-order rate constant (k obs ) were less than 10%.
Computer Simulations-The structures of WT Cyt and Pc, as well as that of mutants obtained by specific changes of residues were modeled using the SYBYL program (Tripos Inc.) in a SGI RC10000 workstation. The structure of WT Cyt from Synechocystis was modeled using the three-dimensional crystal structure of Cyt from the green alga M. braunii (2) as a template. The original PDB file was modified to fit the Synechocystis Cyt sequence with the BIOPOLYMER module of SYBYL. The resulting file was first submitted to energy minimization in vacuo up to an RMS energy gradient of 0.41 kJ mol Ϫ1 Å Ϫ1 , using the SANDER module of AMBER 4.1 (15), and then solvated with TIP3T water molecules using the BLOB option of the EDIT module. Solvent was energy minimized and submitted to a 9-ps molecular dynamics (MD) calculation. The whole system was again energy minimized and submitted to a 750-ps MD run at 300 K, from which the last 200 ps of trajectory were extracted and analyzed with CARNAL. The resulting structure (2.4 Å of RMSD from original) was again submitted to energy minimization. The quality of the structure was tested using the PROCHECK program (16). Force field parameters for the heme moiety were those included in the AMBER package (17). Models of the modified proteins were obtained by changing the residues with the BIOPOLYMER module of SYBYL and further energy minimization in vacuo. Surface electrostatic potentials were estimated by using the algorithm of Nicholls and Honig (18), as indicated in the MOLMOL program (19).
The structure of WT Pc from Synechocystis was modeled using the three-dimensional crystal structure of the triple mutant A42D/D47P/ A63L (20). The structure was first submitted to restrained energy minimization, for which the cut-off radius was 6 Å for the full simulation region around the modified residue and 12 Å for the buffer region. Minimizations were performed first by the steepest descent method and second by conjugated gradient minimization until an RMS energy gradient of 0.12 kJ mol Ϫ1 Å Ϫ1 was reached. TRIPOS force field parameters (21) were used. Bond lengths and dihedral angles were restrained for the copper site. A distance-dependent dielectric constant was used to simulate solvation. Surface electrostatic potentials were likewise calculated by using the algorithm of Nicholls and Honig (18), as indicated in the MOLMOL program (19).

RESULTS AND DISCUSSION
A site-directed mutagenesis study of Synechocystis Cyt has been performed here to investigate its structural and functional analogies with Pc. First, the three-dimesional structure of Cyt was modeled so that the mutants could properly be designed. Modeling makes evident that Cyt does possess an acidic cluster at the south-east face formed by Asp-70, Asp-72, Glu-74, Asp-75, and Asp-2. On the other hand, Tyr-83 has widely been reported to be a highly conserved residue in the east face of Pc (in the Introduction). Interestingly, Ullmann et al. (22) have recently proposed not only that an aromatic residue at position 64 in eukaryotic Cyt may be the counterpart of Pc Tyr-83, but also that the cation-system between the aromatic ring of Phe-64 and the guanidinium group of Arg-67 could play a special role in electron transfer. On the basis of these data, we have mutated four specific residues in Cyt, two of them in the acidic area and the two others just outside or at the border: Asp-70 and Asp-72 were replaced by arginines; Phe-64, which is close to the heme group, was substituted by alanine; and Arg-67, which is at the edge of the east face, was changed to aspartate (Fig. 1). Table I shows that the redox potential value of the heme group is not significantly affected by replacement of any one of the two aspartates by arginines, but it decreases in ϳ30 mV when Phe-64 or Arg-67 are mutated. In all cases, however, the comparative analysis of Cyt mutants by UV/visible spectroscopy reveals no changes in the heme environment (data not shown).
The capability of Cyt mutants to reduce the photooxidized chlorophyll molecule P700 in PSI particles was determined by laser flash-induced absorption spectroscopy. With the four mutants, the PSI reduction kinetic trace is monoexponential, as is it with WT Cyt, but the rate constant is drastically changed depending on the residue modified. As can be inferred from the kinetic profiles in Fig. 2 (upper), R67D is impaired for the reduction of PSI, but the D70R mutant is even more efficient than WT Cyt. Fig. 2 (lower) shows that the observed rate constant (k obs ) for PSI reduction with all mutants increases linearly with increasing donor protein concentration up to 200 M. As is the case not only with WT Cyt but also with WT Pc (7), such a linear dependence is interpreted by assuming that there is no formation of any kinetically detectable (or long-lived) intermediate complex between the metalloprotein and PSI. The reaction mechanism should thus involve a simple oriented collision of Cyt molecules with the membrane-anchored PSI complex. The experiments in Fig. 2 were run at pH 7.5, but similar data were obtained at pH 5.5 (not shown).
The bimolecular rate constant (k bim ) for the overall reaction can be obtained from the linear plots in Fig. 2. The resulting values are presented in Table I, both at pH 5.5 and 7.5. As can be seen, the mutant F64A behaves like the WT molecule, but the modification of charges in other mutants alters their rate constants: D70R and D72R yield values for k bim that are 2-or 3-fold higher than those with WT Cyt, whereas the k bim values with R67D are 5-8-fold lower than those with the WT protein.
These findings indicate that electrostatic repulsions between residues 70 and 72 and their counterparts in PSI are hindering the interaction of the WT Cyt and that Arg-67 plays a critical role in the interaction with PSI (see below).
Taking into account the electrostatic nature of the reaction between Cyt and PSI, the effect of ionic strength on the secondorder rate constant was analyzed. As can be seen in Fig. 3, k bim with WT Cyt increases with increasing NaCl concentration up to reach a maximum value at high ionic strength, indicative of repulsive interactions between the heme protein and PSI (9). The F64A mutant shows an ionic strength dependence similar to WT Cyt, as would be expected for a mutation that does not involve any charged residue. The mutant R67D, in its turn, is much less efficient in transferring electrons to PSI. Replacement of aspartate by arginine at positions 70 and 72 renders the mutant proteins slightly ionic strength-dependent, but such a substitution in position 70 makes the Cyt molecule exhibit an ionic strength dependence that is opposite that of WT Cyt.
By applying the formalism developed by Watkins (14), the bimolecular rate constant extrapolated to infinite ionic strength, that is the diffusion-limited rate constant (k ϱ ), can be calculated, a parameter that provides information on the intrinsic reactivity of redox partners in the absence of electrostatic interactions. The k ϱ values in Table I indicate that F64A is the only mutant that approaches WT Cyt, thereby suggesting that the other mutations involve changes that are not only electrostatic in nature. The small salt dependence of k bim for D70R impeded an accurate estimation of k ϱ , and the slow   kinetics of the R67D mutant did not allow us to get a precise value for k ϱ .
A thermodynamic analysis was also performed to get further insights into the reaction mechanism of PSI reduction by WT and mutant Cyt. In all cases, the temperature dependence of the observed rate constant yields linear Eyring plots with no breakpoints, from which the values for the apparent activation enthalpy (⌬H ), entropy (⌬S ), and free energy (⌬G ) for the overall reaction can be calculated. As can be seen in Table II, the most drastic difference in the activation parameters is observed with R67D: its free energy value at pH 7.5 is ϳ5 kJ mol Ϫ1 higher than that of WT Cyt, as expected from its rather inefficient interaction with PSI. Such an increase in free energy is mainly because of the change in enthalpy (but not in entropy) of the system, thereby indicating that the mutation is altering just the electrostatic interactions between reaction partners. Noteworthy also is the increase in the entropic term (ϳ14 J mol Ϫ1 K Ϫ1 ) with the D70R mutant.
As reported previously (23), the bulk electrostatic effect of substitutions of charged residues can be roughly approximated by assuming that the change in the activation free energy of the overall reaction with mutants as compared with that with the WT protein (⌬⌬G ) is directly proportional to the net charge difference between WT and mutant metalloproteins (⌬q). Thus, the real effect of the mutations can be estimated by using the Watkins equation (14), where V ii is the net electrostatic potential for the interaction, which depends on the electrostatic charges of the reaction partners; X k is the Debye-Hü ckel term that accounts for the effect of charge screening as a function of ionic strength; and k ϱ is the diffusion-limited kinetic rate constant. We can thus write that, where ⌬V el (⌬V el ϭ ⌬V ii ⅐ RT) stands for the difference in electrostatic energy between the WT and mutant proteins. From Equation 2, the difference between the apparent activation free energies for WT and mutant proteins must be proportional to X k . As expected, the experimental data in Fig. 4 (left) indicate that ⌬⌬G changes with ionic strength according to Equation 2, but such an energetic term does not approach 0 at high ionic strength with some Cyt mutants. This finding suggests that k ϱ is being affected not just by purely electrostatic changes, in good agreement with the entropy-dependent change in ⌬G observed with D70R (see above). To account for these discrepancies, a new term should be added to Equation 2, in which the relative value of k ϱ with mutants as compared with that with the WT protein is considered. The experimental data in Fig. 4 were thus fitted to Equation 3, from which the values for ⌬V el with the mutants D70R and D72R were estimated (Table II). F64A exhibits a 0 value for ⌬V el , as expected from its kinetic behavior close to the WT Cyt; with D67R, in turn, it was not possible to determine a value for ⌬V el as ⌬⌬G depends linearly on ionic strength (see Fig. 4). A comparative thermodynamic study was carried out with a set of mutants of Synechocystis Pc, which was modified by replacing specific residues in the east patch. We have previously reported that the values of k obs for PSI reduction, at the ionic strengths used throughout this work, depend linearly on Pc concentration with all mutants of Synechocystis Pc-even with the double mutant D44R/D47R, which reaches a saturation plateau at lower ionic strength that suggests the formation of an electrostatic transient complex with PSI (8).
The activation parameters of the electron transfer reaction between modified Pc and PSI have thus been analyzed here in a similar way than those with Cyt and its mutants. Fig. 4 (right) shows the effect of ionic strength on ⌬⌬G for the mutants of Synechocystis Pc. The experimental data indicates that ⌬⌬G changes with ionic strength according to Equation 2, thereby indicating that the changes in this area alter the rate constants as a consequence of the changes in the bulk electrostatics of the interaction. In principle, the relative effect of amino acid substitutions can directly be related to the net change in charge ⌬q. Actually, the values for ⌬⌬G at low ionic strength with D44R/D47R (⌬q ϭ 4) are approximately 2-fold higher than those with D44K and D44A/D47A (⌬q ϭ 2). Table II shows the apparent activation parameters, as calculated from the Eyring plots under standard conditions, for the reduction of PSI by WT and mutant Pc. It also shows the values for ⌬V el obtained by fitting the experimental data in Fig.  4 to Equation 2. As expected, the differences in ⌬G between WT and any mutant Pc are proportional to ⌬V el , the D44R/ The ⌬V el values change from one to another Pc mutant, even when there is no difference in ⌬q. The mutations are thus affecting in a different manner the electrostatic interaction energy, with the relative effect of mutation on ⌬V el depending not only on ⌬q but also on the specific location of the modified residue in the protein surface. Actually, when mutations involving the same net charge difference are performed in residues Asp-44 and/or Asp-47, the ⌬V el value with mutants at position 44 is always larger than that with mutants at position 47 (see Table II). Such a finding indicates that the changes in reactivity mainly depend on the net charge difference ⌬q, but it is specifically the local negative potential in the proximity of residue 44 that determines the repulsion between reaction partners.
A computer simulation of the distribution of surface electrostatic potential in the mutants of Cyt was compared with that in the mutants of Pc (Fig. 5). Asp-70 in Cyt and Asp-44 in Pc are located near the interphase between the acidic patch and a positively charged region, whereas Asp-72 in Cyt and Asp-47 in Pc are just located in the middle of the acidic patch (see also Fig. 1). Hence, the positive charge introduced upon mutation of any one of the latter residues to arginine is screened by the surrounding negative ones, whereas mutations at the interphase between the two regions result in a rather large modification of the electrostatic potential. In Cyt, the replacement of Asp-70 by arginine induces a spread of the positively charged region, whereas the mutation D72R has a reduced effect on the electrostatic potential. In Pc, the single mutation in D44K as well as the double replacement in D44A/D47A induce a significant increase in the positively charged area closer to the hydrophobic north pole. In addition, the double substitution in D44R/D47R induces an inversion of the electrostatic potential at the eastern patch of Pc, so enabling a large, long range electrostatic attraction toward the reaction center.
In the MD model of Synechocystis Cyt, Phe-64 points toward inside the protein, with the aromatic ring dwelling close to the heme group in such a way that a small reorganization of the Arg-67 side chain could bring the guanidinium group toward the aromatic ring of Phe-64. However, the much lower reactivity of the Cyt R67D mutant cannot be explained on Ullmann's hypothesis (see above). In fact, the replacement of Phe-64 by alanine yields a mutant with the same kinetic efficiency as the WT molecule, thus indicating not only that the aromatic residue is not crucial for the interaction of Cyt with PSI but also that the cation-interaction is not necessary for electron transfer to PSI. We should however bear in mind that Cyt has to interact also with the cytochrome b 6 f complex, for which the aromatic residue would be required (22,24). In a similar way, Tyr-83 in Pc has been reported to be involved in the interaction with cytochrome f but not with PSI (4,25,26). In this context, Sebban-Kreuzer et al. (27) have shown that a tyrosine residue at position 64 is essential for electron transfer between Desulfovibrio vulgaris cytochrome c 553 and its formate dehydrogenase. Whether or not Phe-64 plays any role in the redox interaction with cytochrome f, its vicinity both to the heme group and to Arg-67 would explain why the mutation in Phe-64 leads to changes in the redox potential value (see Table I).
We conclude that Cyt possesses a negatively charged cluster that is isofunctional with the "south-east" patch of Pc. Despite the large positively charged region on the north-east face of WT Cyt, such an acidic patch should be responsible for the repulsive interactions with PSI at low ionic strength.
FIG. 5. Electrostatic potential distribution on the surface of wild-type (WT) and mutants of cytochrome c 6 (upper) and plastocyanin (lower). Surface electrostatic potential was calculated as indicated under "Experimental Procedures." Simulations were performed assuming an ionic strength of 40 mM at pH 7. Negative and positive potential regions are depicted in red and blue, respectively. The relative orientation of the heme and copper proteins is as described in the legend to Fig. 1.