Site-specific Mutations in Human Ferredoxin That Affect Binding to Ferredoxin Reductase and Cytochrome P45OsCc*

Ferredoxins found in animal mitochondria function in electron transfer from NADPH-dependent ferre- doxin reductase (Fd-reductase) to cytochrome P450 enzymes. To identify residues involved in binding of human ferredoxin to its electron transfer partners, neutral amino acids were introduced in a highly conserved acidic region (positions 68-86) by site-directed mutagenesis of the cDNA. Mutant ferredoxins were produced in Escherichia coli, and separate assays were used to determine the effect of substitutions on the capacity of each mutant to bind to Fd-reductase and cytochrome P450.,, and to participate in the cholesterol side chain cleavage reaction. Replacements at several positions (mutants D68A, E74Q, and D86A) did not significantly affect activity, suggesting that acidic residues at these positions are not required for binding or electron transfer interactions. In contrast, substitutions at positions 76 and 79 (D76N and D79A) caused dramatic decreases in activity and in the affinity of ferredoxin for both Fd-reductase and P450.,,; this suggests that the binding sites on ferredoxin for its redox partners overlap. Other substitutions (mu- tants D72A, D72N, E73A, E73Q, and D79N), however, caused differential effects on binding to Fd-reductase and P450.,,, suggesting that the interaction sites are not identical. We propose a model in which Fd-reductase activity was determined assuming 20 (mM.cm)-'. Under these Fd-reductase is lim- iting; rates observed for wild type and mutant ferredoxins were independent of the amount of cytochrome c present when tested at cytochrome c ranging from separate experiments.


Site-specific Mutations in Human Ferredoxin That Affect Binding to
Ferredoxin Reductase and Cytochrome P45OsCc* (Received for publication, April 19, 1991) Vincent M. Coghlan  Ferredoxins found in animal mitochondria function in electron transfer from NADPH-dependent ferredoxin reductase (Fd-reductase) to cytochrome P450 enzymes. To identify residues involved in binding of human ferredoxin to its electron transfer partners, neutral amino acids were introduced in a highly conserved acidic region (positions 68-86) by site-directed mutagenesis of the cDNA. Mutant ferredoxins were produced in Escherichia coli, and separate assays were used to determine the effect of substitutions on the capacity of each mutant to bind to Fd-reductase and cytochrome P450.,, and to participate in the cholesterol side chain cleavage reaction. Replacements at several positions (mutants D68A, E74Q, and D86A) did not significantly affect activity, suggesting that acidic residues at these positions are not required for binding or electron transfer interactions. In contrast, substitutions at positions 76 and 79 (D76N and D79A) caused dramatic decreases in activity and in the affinity of ferredoxin for both Fd-reductase and P450.,,; this suggests that the binding sites on ferredoxin for its redox partners overlap. Other substitutions (mutants D72A, D72N, E73A, E73Q, and D79N), however, caused differential effects on binding to Fd-reductase and P450.,,, suggesting that the interaction sites are not identical.
We propose a model in which Fd-reductase and P450.,, share a requirement for ferredoxin residues Asp-76 and Asp-79 but have other determinants that differ and play an important role in binding. This model is consistent with the hypothesis that ferredoxin functions as a mobile shuttle in steroidogenic electron transfer, and it is considered unlikely that a functional ternary complex is formed.
The animal mitochondrial ferredoxins are small (-14 kDa), acidic proteins that contain a single [2Fe-2S] cluster. They function as central components in electron transfer from NADPH-dependent ferredoxin oxidoreductase (Fd-reductase)' to cytochrome P450 enzymes. These mitochondrial cytochrome P450 enzymes are involved at key steps in a variety of processes, including the first step in steroid hormone biosynthesis, the conversion of cholesterol to pregnen-* This work was supported by National Institutes of Health Research Grant GM43548. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom correspondence should be addressed Dept. of Physiology and Biophysics, University of California, Irvine, CA 92717.
' The abbreviations used are: Fd-reductase, NADPH-dependent ferredoxin oxidoreductase; P45OS,,, cytochrome P45OS,, (P450XJA1). olone catalyzed by the cholesterol side chain cleavage cytochrome P450 (P450,,,). Two different mechanisms have been proposed for ferredoxin-mediated electron transfer from Fdreductase to P450,,,. In one model, ferredoxin acts as an electron shuttle, initially forming a complex with Fd-reductase, dissociating after accepting an electron, and then associating with and transferring the reducing equivalent to P450,,,. This role for ferredoxin as a mobile electron carrier in the cholesterol side chain cleavage reaction is supported by kinetic evidence that maximal activity can be achieved at significantly less than 1 reductase/P450,,, when ferredoxin is saturating (I), by potentiometric data that indicate that dissociation of ferredoxin from Fd-reductase is reduction-induced (2), and by spectroscopic results that suggest sequential formation of 1:l complexes between ferredoxin and Fd-reductase and between ferredoxin and P450,,, (1,3). In an alternative model, the three proteins function as a ternary complex. Support for this model comes from reports that a complex with 1:l:l molar ratio of the three redox partner proteins was isolated (4) and that chemically cross-linked binary complexes between Fd-reductase and ferredoxin can transfer electrons to P450,,, (5,6). One distinction between the two models is that ternary complex formation requires distinct binding sites on ferredoxin for Fd-reductase and P450,,,.
The associations between Fd-reductase, ferredoxin, and P450,,, are strongly dependent on ionic strength (2), indicating an important role for electrostatic interactions. The amino acid sequences of mitochondrial ferredoxins from human (7), cow (8), pig (9), and chick (10) indicate that the protein is highly acidic (PI 4.0-4.5) and contains two conserved regions of 6 or more acidic residues that are uninterrupted by basic amino acids (positions 27-47 and 68-86, Fig. 1). Chemical modification studies using bovine ferredoxin have indicated that acidic residues within the second region (positions 68-86) are important for binding to Fd-reductase (11). However, due to lack of specificity in chemical labeling, it was not possible to unequivocally identify the specific residues involved or to distinguish which residues were important for association with Fd-reductase and which contribute to P450,,, binding.
In an effort to define the Fd-reductase and P450,,,-binding domains of human ferredoxin, we have utilized site-directed mutagenesis to introduce mutations into the cloned cDNA. These mutations resulted in ferredoxins with neutral amino acid substitutions at acidic positions in the region containing residues 68 through 86. Mutant proteins were prepared using an expression system developed for producing human ferredoxin in Escherichia coli (12) and were analyzed in vitro for their ability to support P450-catalyzed cholesterol side chain cleavage, to associate with Fd-reductase, and to bind P45OS,,. Mutagenesis-Synthetic 20-mer oligonucleotides containing single or double point mutations were obtained from either Operon Technologies (Alameda, CA) or the UCR Biotechnology Instrumentation Facility (University of California, Riverside, CA). Synthetic oligonucleotides were purified on 14% denaturing polyacrylamide gels (13) followed by electroelution (14). Oligonucleotide-directed mutagenesis was performed on ferredoxin DNA vector pHFdxl (12) using either a gapped-duplex method (Boehringer Mannheim mutagenesis kit) or a polymerase chain reaction method developed by Nelson and Long (15). In both cases, mutated DNA was treated with BamHI and Hind111 and subcloned into the expression vector pMb3 (16). The resulting plasmids were sequenced using a double-stranded dideoxy method (13) to confirm that mutagenesis was limited to the predicted sites.
Preparation of Mutant Ferredoxins-Expression in E. coli (12) and purification of ferredoxin protein (17) was as previously described, except that transformed MZ-1 cells were induced at an As",, of 0.35, and specific cleavage of fusion proteins was achieved using Factor Xa protease from InFerGene (Benicia, CA) a t a ratio of 1:lOOO (w/w) for 2 h a t 25 "C. In the case of folding mutants (E74A and E74Q, see below), apoprotein was partially purified from the insoluble fraction of lysed cells and resuspended in 100 ml of 6 M urea, 1 mM dithiothreitol, and 100 mM Tris-HC1, pH 7.4. Reconstitution of [2Fe-2S] clusters was achieved by gradual addition of FeCl,, and Na2S to final concentrations of 1 mM each. The mixture was diluted %fold with 50 mM Tris-HC1 (pH 7.4), and the reconstituted mutant ferredoxins were partially purified by ion exchange chromatography followed by gel filtration. Instability precluded purification of these refolded mutants to homogeneity; ratios of A4L4/A276 were less than 0.4. The ratio A414/A27R for all other ferredoxin preparations was greater than 0.7.
Visible absorbance spectra were recorded at room temperature using a Cary 17D spectrophotometer interfaced to a Zenith 2-100 computer (On-Line Instrument Systems, Inc., Jefferson, GA). Ultraviolet and visible CD spectra were recorded a t room temperature in 0.05-and 1-cm cuvettes, respectively, using a Jasco 5720 spectropolarimeter. Two to four scans were routinely averaged data were collected at 0.5-nm intervals and were not smoothed. Samples for CD contained 20 WM ferredoxin in 33 mM potassium phosphate, pH 7.2.
Assays-Cholesterol side chain cleavage assays were performed as previously described (18,22). The experiments were initially carried out using a ferredoxin concentration of 0.25 p M (approximately K,/ 2 for wild type ferredoxin); additional experiments were performed using ferredoxin mutants D76N and D79N a t a concentration of 2 &I . Reaction mixtures contained 25 nM P45OS,,, 30 nM Fd-reductase, and a NADPH-regenerating system. Incubations were for 5 min a t 37 "C in 33 mM potassium phosphate, p H 7.2, 0.1% Tween 20, and 70 FM cholesterol; pregnenolone produced was measured by radioimmunoassay (18,23). We have previously shown that both naturally occurring (17) and recombinant (12) human ferredoxins are functionally equivalent to bovine adrenal ferredoxin (adrenodoxin) when reconstituted in vitro with bovine adrenal Fd-reductase (adrenodoxin reductase) and bovine adrenal P450,,,.
Cytochrome c reduction was assayed in 33 mM potassium phosphate, p H 7.2 a t room temperature. Reaction mixtures contained 10 nM Fd-reductase, 20 p~ horse heart cytochrome c (Sigma type VI) and a NADPH-regenerating system. Reduction was monitored spectrophotometrically a t 550 nm, and activity was determined assuming Ae550 = 20 (mM.cm)-'. Under these conditions, Fd-reductase is limiting; rates observed for wild type and mutant ferredoxins were independent of the amount of cytochrome c present when tested at cytochrome c concentrations ranging from 10 to 40 FM.
Spectral titrations were performed as described (24) using 0.6 FM P450,,., in 33 mM potassium phosphate, p H 7.2, 0.1% Tween 20, and 70 PM cholesterol a t room temperature. Specified amounts of ferredoxin were added from concentrated stock solutions such that final dilutions of reaction mixtures were less than 4%, and dilutions were corrected for in calculating difference spectra. The concentrations of free ferredoxin were calculated using the following equation Values of kinetic parameters were determined by least squares linear regression analysis of the data from two separate experiments.

RESULTS
Production of Mutants-Two types of amino acid substitutions were made for acidic residues between positions 68 and 86 as shown below. 70 75 80

90 -L -D -A -I -T -D -E -E -N -D -M -L -D -L -D -L -T -D -R -S -R -L -
In most cases, aspartate or glutamate was replaced with the corresponding amide to maintain polarity at that position and to attempt to minimize structural perturbations. Alanine, although smaller and nonpolar, was used to replace acidic amino acids at some positions as an alternative neutral residue. With the exception of substitutions at Glu-74, each of the expressed mutants yielded soluble ferredoxin that incorporated an iron-sulfur cluster in uiuo. Replacement of glutamate at position 74 with either alanine (E74A) or glutamine (E74Q), however, resulted in formation of insoluble apoprotein in E. coli. By reconstitution in vitro with iron and sulfur (see "Experimental Procedures"), we were able to partially purify a limited amount of mutant holoprotein E74Q, but the preparation was unstable and exhibited UV absorbance indicating the presence of impurities and apoprotein. All other mutant proteins exhibited UV and visible absorption spectra indistinguishable from that obtained for wild type recombinant human ferredoxin (Ref. 12; data not shown). These results indicate that mutations introduced at positions 68, 72, 73, 76, 79, and 86 did not disrupt incorporation of the [2Fe-2S] cluster.
To determine whether changes in structure had occurred as a result of amino acid substitutions, CD spectra were recorded for mutant and wild type ferredoxins. Far UV and visible CD spectra for wild type human ferredoxin and for two mutants in which binding to Fd-reductase and P450,,, were affected (see below) are presented in Fig. 2.
In the far UV region, no major differences in CD were observed between wild type human ferredoxin and mutants D68A, D72A, D72N, E73A, E73Q, D76N, D79A, D79N, and D86A. These results suggest that amino acid substitutions at these positions did not introduce any large changes in polypeptide backbone structure. Instability and the presence of contaminants precluded reliable CD measurements in the far UV for preparations of E74Q.
In the visible range, mutants D68A, D72A, D72N, E73A, E73Q, D76N, D79A, D79N, and D86A each exhibited CD spectra that were indistinguishable from wild type ferredoxin. Samples contained 20 PM ferredoxin in 33 mM potassium phosphate, These results suggest that substitutions at these positions did not alter the environment in the immediate vicinity of the [2Fe-2S] cluster. Mutant E74Q exhibited visible CD spectra that were qualitatively similar to wild type ferredoxin in terms of band shape and position of peaks, but it was not possible to accurately determine A6 values due to instability of the sample.
Reconstitution Assay-Each mutant was initially tested for the capacity to participate in the cholesterol side chain cleavage reaction by reconstitution of mutant ferredoxins with bovine Fd-reductase and bovine P450,,,. Activity in this assay requires binding of ferredoxin to Fd-reductase and to P450,, and productive electron transfer. These experiments were initially carried out using a limiting concentration of ferredoxin (wild type Km/2), in order to ensure that differences in the affinity of mutant ferredoxins for the redox partners would be apparent. The results are summarized in Table I.
The mutants could be classified into three groups based on their specific activity. Mutants D68A and D86A exhibited activity that was similar to that of wild type ferredoxin (>2 ng of pregnenolone/min) indicating that binding and electron transfer were not significantly affected by these substitutions. Mutants D72A, D72N, E73A, and E73Q showed activity that was reduced to slightly less than half that of wild type (1.1-1.4 ng of pregnenolone/min) suggesting that acidic residues at these positions may be important for association of ferredoxin with one or both redox partner proteins. For mutants D76N, D79A, and D79N, activity was too low (cO.1 ng of pregnenolone/min) to be accurately measured at the ferredoxin concentration used in this experiment. In an attempt to measure the specific activity of mutants D76N, D79A, and D79N, an 8-fold higher concentration (2 PM ferredoxin) was tested in a separate experiment; under these conditions, all three mutants exhibited a capacity to support cholesterol side chain cleavage, but with low specific activity (0.42, 0.60, 0.67 ng of pregnenolone/min, respectively). The finding of activity only at elevated ferredoxin concentrations suggests that the reduced activity observed in mutants D76N, D79A, and D79N most likely arises from decreased binding to Fd-reductase and/or P450,,, and not from a failure to transfer electrons.
Due to its instability at 37 "C, it was not possible to characterize mutant E74Q in this assay. Binding to Fd-reductase-To determine the extent to which mutations affect binding of ferredoxin to Fd-reductase, we utilized an assay involving cytochrome c as an electron acceptorlindicator. Although this reaction does not occur physiologically, it has been widely used to analyze the kinetics of association of ferredoxin with Fd-reductase (2,11,17,25). Under the assay conditions employed, the experimentally derived K, values are equivalent to dissociation constants ( K d ) for the Fd-reductase:ferredoxin complex (2). Kinetic results obtained in this assay are plotted in Fig. 3 and are summarized in Table  I. At saturating concentrations, all ferredoxins were active, but mutants could be classified into two groups (low and high affinity), based on their apparent K, for Fd-reductase. Low affinities were observed for mutants D76N, D79A, and D79N; apparent K , values obtained for these mutants (-1.9-2.4 PM) were more than 100-fold greater than wild type. In contrast, K , values obtained for mutants D68A, E73A, E73Q, and D86A (17-19 nM) were very close to that determined for wild type ferredoxin (17 nM). Small but reproducible increases in K , values were observed for mutants D72A and D72N (25-28 nM), but these differences (-1.4-1.6fold increase in K,) are much less than the >100-fold increases in K , values observed for mutants D76N, D79A, and D79N. Only minor differences in V,,, values were observed between mutant ferredoxins in the low affinity group (D76N, D79A, and D79N; range 0.60-0.74 nmol of cytochrome c reduced/min) and those in the high affinity group (D68A, D72A, D72N, E73A, E73Q, and D86A; 0.64-0.79 nmol of cytochrome c reduced/min); this suggests that the reduced activities of mutants D76N, D79A, and D79N are due to decreased binding affinity and not to defects in electron transfer. Mutant E74Q exhibited activity in this assay, but because a limited amount of this unstable mutant was available, only two concentrations were examined. The estimated As a further test of electron transfer function, additional experiments were performed in which we added approximately 100-fold molar excess of mutants D76N, D79A, or D79N (1 PM) to reactions containing a limiting amount of wild type ferredoxin (10 nM). The measured activities were additive (data not shown); wild type activity was not inhibited in the presence of any of these low affinity mutants. This result indicates that when D76N, D79A, or D79N binds to Fdreductase, electron transfer proceeds at a rate similar to that achieved with wild type ferredoxin.
Taken together, these results indicate that Asp-76 and Asp-79 are critical for high affinity binding of ferredoxin to Fdreductase, whereas Asp-68, Asp-72, Glu-73, Glu-74, and Asp-86 are not essential. None of these acidic residues appear to be specifically required for electron transfer once the complex is formed.
Binding to P45OS,,"To test whether amino acid substitutions affect binding of ferredoxin to P450,,, we utilized a spectral binding assay. In the presence of detergent and limiting cholesterol, P450,,, exists predominantly as the low spin, substrate-free form with a Soret maximum near 417 nm (22). Binding of ferredoxin to P450,,, induces cholesterol binding and results in conversion of the cytochrome to a high spin form having a Soret peak near 392 nm (26). Thus, binding of ferredoxin to P450,,, can be followed by monitoring difference spectra resulting from the shift in the Soret spectrum. Difference spectra recorded during titration of P45OS,, with either wild type ferredoxin or mutant D76N are shown in Fig.  4. In both cases, spectral changes corresponding to a shift in P450,,, from low to high spin were associated with ferredoxin binding, but much larger changes were induced by wild type ferredoxin than by mutant D76N. The data from these titrations and from similar experiments with other mutants are plotted in Fig. 5; the calculated spectral dissociation constants ( K J and AA,,, values are summarized in Table I respectively). These results are similar to the results obtained in the Fd-reductase assay in which no major changes in affinity were observed (see above) and suggest that Asp-68 and Asp-86 are not essential for binding of ferredoxin to either redox partner. In contrast, dramatic changes in apparent affinity for P450,,, were exhibited by; mutants D76N and D79A; the K, values for these mutants (5.0 and 6.6 pM, respectively) were increased 6-8-fold over wild type ferredoxin. The behavior of mutants D76N and D79A is similar to that observed in the Fd-reductase cytochrome c reduction assay (see above) in which significantly reduced affinities for Fd-reductase were apparent. These results suggest that Asp-76 and Asp-79 are important in the binding of ferredoxin to both of its redox partner proteins. Overlap must therefore exist between the binding sites on ferredoxin for Fd-reductase and P450,,,. Interestingly, replacement of Asp-79 with asparagine (mutant D79N) had only a minor influence on the ability of ferredoxin to bind P450,, (Ks = 1.1 p~) .
This result is different from that obtained in the Fd-reductase assay in which D79N behaved like D79A and exhibited dramatically decreased affinity (>lOO-fold, see above) for Fd-reductase. This finding indicates that a polar residue is sufficient to fulfill P450,, binding requirements at position 79, whereas an acidic group at this position is critical for Fd-reductase binding.
Significant differences in P450,,, binding were also observed for mutants D72A, D72N, E73A, and E73Q; K, values for these mutants (range 2.5-3.1 p~) were 3-4-fold higher than wild type. These results differ from those of Fd-reductase binding experiments, in which D72A, D72N, E73A, and E73Q each exhibited apparent affinity for Fd-reductase that was similar (E73A and E73Q) or only slightly reduced (D72A and D72N) when compared to the affinity of wild type ferredoxin for Fd-reductase. These findings suggest that, whereas some acidic residues (Asp-76 and Asp-79) make an important contribution to the binding of both P450,,, and Fd-reductase, other positions (Asp-72 and Glu-73) appear to be relatively more important for interaction with P450.,, than for association with Fd-reductase.
Only limited measurements were possible for E74Q due to instability of the protein, but the data obtained at two concentrations indicate that this mutant exhibits an apparent binding affinity for P450,, similar to that observed for wild type ferredoxin (Ks -0.6 p~) .
Because only minor differences in binding affinity were also observed for this mutant in the Fd-reductase assay, Glu-74 (like Asp-68 and Asp-86) does not appear to be required for interaction with either Fd-reductase or P450,,,.

DISCUSSION
Previous efforts to identify specific groups that are involved in electrostatic interactions between ferredoxin and its redox partners have utilized chemical modification. In ferredoxin, acidic residues appear to be most important; labeling of lysines in bovine adrenal ferredoxin had no effect on binding (27), but modification at acidic residues inhibited binding to Fdreductase (11). In these experiments, modifications appeared to occur predominantly at ferredoxin positions 74, 79, and 86, but multiple and nonstoichiometric labeling made it impossible to determine the extent to which individual amino acid residues contribute to Fd-reductase binding. Moreover, it is possible that the bulky labeling reagent may have also prevented interactions at neighboring sites (e.g. position 76). These problems are circumvented in the present study by the use of site-directed mutagenesis. As with chemical labeling, however, it is possible that modifications can alter native protein structure. This limitation is illustrated by mutations at ferredoxin position 74 where substitutions of alanine or glutamine for glutamate resulted in proteins with markedly decreased stability. Although mutant E74Q was unstable, it exhibited activity in the Fd-reductase cytochrome c reduction assay and induced a spectral shift in P450,,, with binding affinity approximating that observed for wild type ferredoxin. These results suggest that although Glu-74 may play some role in maintaining ferredoxin structure, it is not essential for binding to either Fd-reductase or P450.,,.
The finding that each of the stable mutant proteins exhibit visible absorption and CD spectra indistinguishable from wild type ferredoxin indicates that the mutations did not significantly alter the environment around the [2Fe-2S] center. CD spectra recorded in the far UV indicate that gross changes in polypeptide backbone structure were also absent in these mutants. The atomic structure of an animal ferredoxin has not been solved, and without crystallographic data on wild type and mutant proteins, we cannot rule out small conformational changes; however, the spectroscopic data suggest that large deviations from wild type activity observed among mutants most likely result from neutralization of acidic residues and not changes in protein structure. Preliminary 'H NMR data support this conclusion: no significant differences in well defined aromatic and hyperfine shifted regions were apparent in spectra recorded at 600 MHz for wild type recombinant human ferredoxin and mutants D76N and D79A.' Models for Fd-reductase and P450,, binding that are consistent with the biochemical properties of these mutant ferredoxins are shown in Fig. 6. In these models, Asp-76 is critical for binding both Fd-reductase and P450,,,. A polar residue at position 79 is also important for binding of both proteins, but only Fd-reductase association requires a negative charge at this position. Asp-72 and Glu-73 appear to play only a minor role in Fd-reductase association, but contribute significantly to P450,,, binding. Asp-68, Glu-74, and Asp-86 do  not appear to be directly involved in binding to either Fdreductase or P450,,,.
Amino acid replacement at positions 76 (D76N) and 79 (D79A) had the largest effects on ferredoxin activity, producing dramatic decreases in affinity for both Fd-reductase and P450,,. It is possible that these substitutions each cause conformational changes that affect distinct binding domains (one for Fd-reductase and another for P450,,,), but the lack of any detectable changes in the visible or far UV CD argue against this. A more plausible explanation is that residues from Fd-reductase and P450,,, interact with Asp-76 and Asp-79 when each forms a complex with ferredoxin. Such overlap in binding sites would seem likely to preclude simultaneous binding of Fd-reductase and P450,, and suggests that ferredoxin binds the two proteins individually. Thus, our model is consistent with the proposal of Lambeth et al. (25) in which ferredoxin acts as a mobile electron shuttle, sequentially binding to Fd-reductase and then to P450,,,. This interpretation is inconsistent with the proposal that ferredoxin functions in a 1:l:l ternary complex with Fd-reductase and P450,,, (28-30). The proposals for a ternary complex are largely based on results from chemical cross-linking studies, and it is quite possible that such modified complexes have no physiological relevance. In fact, functional activity of a cross-linked ternary complex has not been demonstrated.
The large changes in affinity observed for mutant ferredoxins containing a substitution at position 76 or 79 indicate that acidic residues play an important role in electron transfer partner recognition and complex formation. The decreased binding affinities observed for ferredoxin mutants D76N, D79A, and D79N in the Fd-reductase interaction correspond to increases in free energy for association of approximately 2.8-3.0 kcal/mol each.3 Similarly, increases in K, values for binding P450,,, observed with mutants D76N and D79A correspond to free energy changes of approximately 1.1 and 1.3 kcal/mol. These increases in free energy are consistent with loss of stabilizing ion pairs (31, 32), but some of the charged residues could also be involved in forming hydrogen bonds (33). The involvement of an acidic residue in hydrogen bonding may account for the differences in P450,,, binding observed for mutants at position 79; replacement of Asp-79 with a nonpolar amino acid (alanine) inhibited P450,,, binding, whereas substitution with the corresponding amide produced only a small increase in K,.
Although the results demonstrate that Asp-76 and Asp-79 are critical for binding of ferredoxin to both Fd-reductase and P450,,,, it is likely that additional residues play an important role in these associations. The methods employed in this study may be useful for further mapping of Fd-reductase and P450,,,-binding domains and serve to refine our preliminary models. The specific residues in Fd-reductase and P450,,, that interact with ferredoxin have not been unequivocally identified, but chemical modification of lysine amino groups in both proteins has been shown to inhibit ferredoxin binding (34-37). Site specific mutagenesis may also be useful to help identify the complementary charges in Fd-reductase and P450.,, that interact with the acidic ferredoxin residues identified in this study.