Origin of the pH dependence of the midpoint reduction potential in Clostridium pasteurianum ferredoxin:oxidation state-dependent hydrogen ion association.

The origin of the dependence of E1/2 on pH exhibited by Clostridium pasteurianum 2(4Fe-4S) ferredoxin has been investigated. The results show that oxidation state-dependent pK values, which may arise from sites on the iron-sulfur centers, are responsible for the pH effect. Based on a model of two equivalent protonation sites/molecule, values of 7.4 for pKox and 8.9 for pKrd were obtained. The results of experiments which monitor changes in the hydrogen ion concentration with changes in protein oxidation state are reported. The magnitude of the changes in pH on reduction or reoxidation of the protein are in reasonable agreement with the proposed model. The conformation of C. pasteurianum ferredoxin was examined by nmr, epr, and CD spectroscopies to rule out a pH-dependent conformation equilibrium as the origin of the pH effect.

The origin of the dependence of Elfz on pH exhibited by Clostridium pasteurianum 2(4Fe-4S) ferredoxin has been investigated. The results show that oxidation state-dependent pK values, which may arise from sites on the iron-sulfur centers, are responsible for the pH effect. Based on a model of two equivalent protonation sites/molecule, values of 7.4 for p L X and 8.9 for pKrd were obtained. The results of experiments which monitor changes in the hydrogen ion concentration with changes in protein oxidation state are reported. The magnitude of the changes in pH on reduction or reoxidation of the protein are in reasonable agreement with the proposed model. The conformation of C. pasteurianum ferredoxin was examined by nmr, epr, and CD spectroscopies to rule out a pH-dependent conformation equilibrium as the origin of the pH effect.
Many iron-sulfur proteins exhibit pH-dependent midpoint reduction potentials and some examples of these are listed in Table I (1)(2)(3)(4)(5). Although such dependences have been known for years, the question of the origin of the effect has not been addressed in the literature. It appears reasonable that oxidation state-dependent hydrogen ion binding is a feature of the chemistry of Fe-S centers in general, in view of the following: hydrogenase is an iron-sulfur protein; iron-sulfur proteins may be involved in proton translocation in energy-transducing membranes (5, 6); and the observation that many iron-sulfur proteins exhibit pH-dependent midpoint reduction potentials despite varied polypeptide compositions. To test for the existence of oxidation state-dependent hydrogen ion equilibria, the pH dependence of the midpoint reduction potential of Clostridium pasteurianum 2(4Fe-4S) ferredoxin has been examined. This ferredoxin is well characterized, has no amino acid residue with an intrinsic pK between 6 and 9, and exhibits a significant and well defined pH dependence. It is a low molecular weight, clostridial-type ferredoxin with a reduction potential near -400 mV.

MATERIALS AND METHODS
C. pasteurianum was grown and ferredoxin isolated according to the procedure of Rabinowitz (7). Reduction potentials were determined using partially purZ1ed C. pasteurianum hydrogenase, as described by Lode et al. (3). Ferredoxin solutions were prepared from a highly concentrated stock solution by dilution into the buffer of desired pH. All solutions were prepared to be approximately 0.03 mM ferredoxin in 0.1 M Trisacetate-glycine-phosphate buffer containing 0.5 M NaCl. The pH of reoxidized ferredoxin solutions was measured using a Radiometer pH meter, R 26, and this finai pH value was used in the calculations. (Small changes in pH occurred upon reduction of certain samples, although the changes were generally not larger than 0.05 pH unit compared to the starting buffer.) The pD of buffers used for 'H-nmr samples was calculated by adding 0.4 to the pH meter reading (8). Optical measurements were made using a Cary 219 UV-visible spectrophotometer.
Proton nmr spectra were recorded on a Bruker 270 MHz spectrometer at the NIH Regional Facility in New Haven, CT. epr spectra were obtained using a Varian V-4500 X-band spectrometer equipped with a Heli-Tran liquid helium transfer system (Air Products). Circular dichroism spectra were recorded on a Jasco 5-20 spectropolarimeter.
Electrolytically reduced methyl viologen (Princeton Applied Research, Model 380 Constant Potential Coulometry System) was used in experiments which determined the number of micromoles of hydrogen ion bound or released. The pH measurements were made using a Radiometer PHM 26 meter equipped with a combination microelectrode in an anaerobic cell which permitted additions through a serum cap. Since aerobic oxidation of ferredoxin or methyl viologen leads to an increase in pH, care was taken to ensure strict anaerobic conditions. All solutions were prepared in 1.0 mM Tris-C1 buffer containing 0.50 M NaCl. For the reduction experiments, the pH of the methyl viologen solution was adjusted to match that of the protein solution, and for the reoxidation experiments, the pH of the ferredoxin solution was adjusted to match that of the methyl viologen/ferredoxin solution.

RESULTS AND DISCUSSION
The pH dependence of the midpoint reduction potential, El/*, for C. pasteurianum ferredoxin was determined from pH 6.2 to pH 8.9. At pH values lower than 7.4 exhibits a dependence of -16 mV/pH unit and at pH values greater than 7.4, a dependence of -30 mV/pH unit is observed. These results are consistent with a number of previously reported results' (I, 3, 9, 10, 11).
Two mechanisms by which the observed midpoint reduction potential can exhibit a pH dependence are: oxidation-reduction equilibria involving hydrogen ion binding, and pH-dependent protein conformation changes.
To test the possibility that oxidation state-dependent hydrogen ion binding occurs, the experimental data were compared to a calculated pH dependence curve. A pH dependence can be incorporated into the Nernst equation using the model

Reduction-linked Hydrogen Ion Binding in a Ferredoxin
of two equivalent sites/protein molecule: The diagram above is only schematic. It is not required that the site of hydrogen ion binding be the 4Fe-4S center, but only that the equilibrium for binding be oxidation state-dependent. E& and E& are the midpoint reduction potentials of the unprotonated and protonated forms, respectively. KO, and Krd are the equilibria for proton association to the oxidized and reduced forms of the protein and any electrostatic interaction with a titrable site will induce an oxidation statedependent pK. The change between the intrinsic midpoint potentials of the unprotonated and protonated forms and E&) induced by the electrostatic interactions is equal to 59(pKd -pId,) mV (25 "C). In other words, an electrostatic interaction will always yield an oxidation state-dependent ion association. For a one-electron reduction at 25 "C, the apparent midpoint reduction potential in millivolts is given by Equation 1 (12): (1) Pseudomonas putida -30 7.5-?
A reproducible difference was observed between the solution of oxidized and reduced ferredoxin (Table 11) indicating the presence of sites of oxidation state-dependent proton binding. Using a titration curve determined for an identical sample of protein, the hydrogen ion binding to the protein on reduction was quantitated from the observed pH difference (Table 11). These experimentally obtained results were compared to a calculated difference in bound hydrogen ion based on the model of two equivalent sites with pK,, = 7.4, pKrd = pK,d, El/z = -371 mV.

Change in hydrogen ion concentration on reduction of C. pasteurianum ferredoxin
An 800-pl sample of 0.53 mM C. pasteurianum ferredoxin in 1 m and 3) and the buffered protein solution (Experiments 4, 5, and 6).
Tris-HC1 buffer/500 mM NaCl, at the indicated pH value, was made Because of the relatively large effect of addition of methyl viologen to anaerobic by continuous flushing of a stirred sample with oxygen-free a sample of buffer containing no protein, a correction was made to nitrogen in a cell which also contained a combination pH electrode. include this factor in a calculation of the total number of micromoles The sample was reduced by addition of 100 pl of 50 mM reduced of H+ bound on reduction, as indicated in the table. A calculation of methyl viologen. The number of micromoles of hydrogen bound was an expected number of micromoles of hydrogen ion bound on reducdetermined from the observed pH values by comparison with titration tion was made based on a model of two equivalent sites of hydrogen curves obtained for solutions of the buffer alone (Experiments 1, 2, ion binding/ferredoxin molecule, with pK,, = 7.4 and pKrd = 8.9.

Reduction-linked Hydrogen Ion
Binding in a Ferredoxin 8.9. As shown in Table 11, the agreement is very good between the experimental and predicted decrease in hydrogen ion concentration on reduction of ferredoxin. A second type of experiment determined the amount of hydrogen ion released upon oxidation of reduced ferredoxin. Ferredoxin was anaerobically reduced with methyl viologen, the pH was determined, and the sample was reoxidized using a minimum amount of potassium ferrocyanide. Once again, a difference in pH between reduced and oxidized ferredoxin was observed, and this difference was used to determine the amount of hydrogen released by comparison to a titration curve. As can be seen in Table 111, the agreement between experiment and theory is excellent. Control experiments indicate that the known destructive side reactions of ferricyanide with ferredoxin do not alter the pH in the period of time required for the reoxidation experiment (2-5 min). The average quantitative result for the reduction and oxidation experiments is 95% of the predicted result.
The quality of the fit of the experimental data to the calculated curve in Fig. 1, the experiniental demonstration of a change in hydrogen ion concentration upon change of ferredoxin oxidation state, and the quantitative agreement between this change and that theoretically predicted, all strongly indicate that the origin of the pH dependence of the midpoint reduction potential is oxidation state-dependent hydrogen ion binding.
To c o n f i this hypothesis the alternate possible origin of a pH-dependent ElIz, a pH-dependent protein conformation, was examined. In order for a pH-dependent change in conformation to affect the reduction potential without oxidation state-dependent proton association, it must be coupled to some alteration of the 4Fe-4S clusters. In order to detect the presence of a change in protein conformation with pH, nmr, epr, and CD spectra were recorded as a function of pH.
The 'H-nmr spectra of oxidized C. pasteurianum ferredoxin at pD 6.8 and pD 8.9 appear in Fig. 2 protons (13,14), which appear as single proton resonances shifted downfield due to the paramagnetism of oxidized 4Fe-4S centers at room temperature (13). The position of these resonances depends on contact shift interactions and are thus sensitive to the orientation of the P-carbon to sulfur bonds (15). The resonance positions exhibited by the P-cysteinyl protons, therefore, provide information concerning the geometry of Fe-S centers and their neighboring atoms. Thus, the 'H-nmr spectra are useful in identifying conformation changes at the cluster. As can be seen from Fig. 2, the resonances in the 10-20 ppm region occur at identical field strengths at pD 6.8 and pD 8.9. It therefore can be concluded that no detectable conformation differences exist at the immediate environment of the clusters in oxidized C. pasteurianum ferredoxin between these two pD values.
To further examine conformation as a function of pH, and to include the reduced state of ferredoxin, epr spectra were compared at the pH values shown in Fig. 3. Two sets of spectra appear; the first pair (Fig. 3A) are the spectra of partially reduced ferredoxin samples (-15% reduction); the second pair (Fig. 3B) are the spectra of more fully reduced ferredoxin samples (75-90% reduced). Slightly reduced samples of ferredoxin exhibit spectra which arise predominantly from reduced 4Fe-4S centers on one-electron reduced molecules. Fully reduced samples exhibit a more complex spectrum which arises from interaction between two paramagnetic reduced 4Fe-4S centers (16). In either case, the linewidths and g values of the signals are sensitive to the bonding geometry of the 4Fe-4S centers. The spectrum of a slightly reduced sample exhibits a small component of the two-electron reduced spectrum, visible as small peaks centered around 3.3 kG. These features correspond to the two major peaks of the fully reduced ferredoxin spectrum.
The observation that the spectra in Fig. 3 are essentially superimposable at the pH values given, indicates that not only is the geometry at a single cluster conserved (partially reduced samples) but also that the conformations at pH 7 and pH 9 are similar enough to conserve the spin interaction between paramagnetic centers in the same molecule. (The small differences that are visible between the spectra of the two near fully reduced samples may be assigned to slightly different degrees of reduction.) The epr data presented corroborate the 'H-nmr findings, indicating that the conformation of C. pasteurianum ferredoxin is pH independent. CD spectra of oxidized ferredoxin were recorded from 300 to 800 nm at pH 6.5 and pH 8.5 and were found to be identical. The CD spectrum of reduced ferredoxin from 300 to 800 nm was also pH-independent from pH 6.5 to 8.3. The results presented in this paper demonstrate that the pH dependence of the midpoint reduction potential arises from oxidation state-dependent hydrogen ion equilibria. It is not possible to assign the sites of these equilibria to specific atoms of the protein based on the available data. In addition to sulfur atoms in the Fe4S4SsCw clusters, the titrable sites include Asp (positions 6, 27, 33, 35, 39), Glu positions 17 and 55, Lys (position 3), Tyr (position 2), and the carboxyl and amine terminals. The fitted pk values of 7.4 and 8.9 suggest that the sulfur atoms or the NH2-terminal may be the site of protonation. The characteristic pK values of glutamic and aspartic acid residues and the COOH-terminal are too low. The pK values of tyrosine and lysine residues are high (usually near lo), and thus also are not likely to be the important sites of protonation.

Reduction-linked Hydrogen Ion Binding in a Ferredoxin
Several arguments favor the s u l h atoms as the probable et al. (17), that the 4Fe-4S center in this protein is solvent inaccessible at pH 7, and therefore could not have sulfur atoms as sites of oxidation state-dependent protonation. Also Job and Bruice (18) have reported a pK of 7.4 for the ironsulfur core of a water-soluble synthetic analogue of 4Fe-4S protein clusters, with the formula Fe4S4(SCH2CH&00)46-. Also, one would expect a hydrogen ion equilibrium at sulfur sites to be intrinsically oxidation state-dependent because of increased charge on reduction of the iron-sulfur center. Evidence of cluster protonation is not seen in the nmr or epr spectra at low pH (<7), but consideration of the nature of these spectra indicates that the effect of the protonation may not be observed. For example, if the site of protonation is a cysteinyl sulfur atom, the nmr resonance positions of only half the P-cysteinyl proton resonances are resolved, and therefore the protonation of the cysteinyl-sulfur atom would not necessarily be detected. In the epr spectrum, hyperfine interaction between a proton bound to the cysteinyl-sulfur and the unpaired electron spin would be present, but may be too small to be visible in a frozen solution spectrum with significant g value anisotropy.
The evidence presented in this paper, in conjunction with the generally observed negative dependence of EI/Z on pH, indicate that oxidation state-dependent hydrogen ion binding may be a general feature of iron-sulfur protein chemistry. Indeed, it is possible that the enzymatic activity exhibited by hydrogenase and the involvement of iron-sulfur centers in energy conservation at Site I in mitochondria are specializations of such a general feature.