Ferredoxin:NADP+ oxidoreductase. Equilibria in binary and ternary complexes with NADP+ and ferredoxin.

Ferredoxin:NADP+ oxidoreductase (ferredoxin: NADP+ reductase, EC 1.18.1.2) was shown to form a ternary complex with its substrates ferredoxin (Fd) and NADP(H), but the ternary complex was less stable than the separate binary complexes. Kd for oxidized binary Fd-ferredoxin NADP+ reductase complex was less than 50 nM; Kd(Fd) increased with NADP+ concentration, approaching 0.5-0.6 microM when the flavoprotein was saturated with NADP+ K(NADP+) also increased from about 14 microM to about 310 microM, on addition of excess Fd. The changes in Kd were consistent with negative cooperativity between the associations of Fd and NADP+ and with our unpublished observations which suggest that product dissociation is rate-limiting in the reaction mechanism. Similar interference in binding was observed in more reduced states; NADPH released much ferredoxin:NADP+ reductase from Fd-Sepharose whether the proteins were initially oxidized or reduced. Complexation between Fd and ferredoxin: NADP+ reductase was found to shield each center from paramagnetic probes; charge specificity suggested that the active sites of Fd and ferredoxin:NADP+ reductase were, respectively, negatively and positively charged.

* This work was supported by National Institutes of Health Grant GM21226 and National Science Foundation Grant PCM 79-24877. 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.
2 To whom correspondence should be addressed.
Workers from several laboratories have identified binary Fd.FNR and FNR.NADP(H) complexes (5)(6)(7). Equilibrium studies have shown that complexation alters the oxidationreduction properties of ferredoxin:NADP+ reductase and its substrates; electron transfer from Fd to ferredoxin:NADP+ reductase and from ferredoxin:NADP+ reductase to NADP' is more highly favored in complex than would be predicted from the mid-point potentials of the separate components (8,9,24).
Identification of a ternary complex is less certain. Ricard et al. (10) reported (on the basis of difference mixing spectra) that the associations of Fd and NADP+ with the flavoprotein were independent, and a ternary complex formed. In contrast, we reported preliminary experiments indicating competition between Fd and NADP+ for complex with ferredoxin:NADP+ reductase (11); Davis (12) reported similar results. The latter findings tend to argue against a ternary complex.
We report here a series of experiments designed to determine whether any Fd.FNR. NADP+ complex forms, and if so, whether K d values are different than in the binary Fd. FNR and FNR. NADP+ complexes. The data indicate that ferredoxin:NADP+ reductase will allow simultaneous complexation by Fd and NADP+; however, addition of one substrate decreased the association of the other in a pattern of negative cooperativity.
We will report, separately, rapid kinetic studies which indicate that a ternary complex must participate in NADP+ reduction. The kinetic data suggest that the rate of electron transport may be limited by dissociation of oxidized Fd from ferredoxin:NADP+ reductase; destabilization of Fd. FNR complex by NADP+ may facilitate the overall reaction.
Fd and ferredoxin:NADP+ reductase were purified as reported previously (9,13), with the exception of the last step in ferredoxin:NADP+ reductase purification. After chromatography on DE52 cellulose, we applied the ferredoxin:NADP+ reductase preparation to a 2',5'-ADP-Sepharose column; enzyme and column were in 30 mM Tris, pH 8.0. The column-bound flavoprotein was eluted with the same buffer plus 10 mM NaIP207. Fractions in which A ,~/ A w exceeded 0.12 were pooled, dialyzed uers'sus 50 mM Hepes, pH 8.0, and stored under liquid Nz. Sodium dodecyl sulfate-polyacrylamide gel electrophoresis revealed only one major band in these preparations.

Binary and Ternary Complexes with FNR, Fd, N A D P 8833
If these samples were further purified by affinity chromatography on Fd-Sepharose, minor bands disappeared. Fd-Sepharose was prepared as described by Shin and Oshino (14).

Methods
All experiments, unless otherwise noted, were conducted in 50 mM Hepes buffer, pH 8.0, at room temperature. Fd used had an A120/A276 ratio z 0.46; ferredoxin:NADP+ reductase had an A~M/Azw 2 0.12. Concentrations of Fd, ferredoxin:NADP+ reductase, NADP+, and NADPH were determined using extinction coefficients reported previously (9); we assumed an extinction coefficient of 15,400 cm" M" at 259 nm for 2',5'-ADP (15). Absorbance spectra were recorded using a Cary 219 UV-VIS spectrophotometer linked to a Minc-11 computer. EPR spectra were recorded using a Varian E-9 spectrometer equipped with an Air Products gas-transfer line. Oxygen-free Ar was produced by passing prepurified Ar through a BASF column, then through a bubbler filled with methyl viologen photo-reduced by 5-deazaflavin and EDTA as described by Massey and Hemmerich (35).
NADP(H) was also assayed by stimulation of ferricyanide reduction. The assay contained the same concentrations of ferricyanide, glucose 6-phosphate, and glucose-6-phosphate dehydrogenase as above, but with 1.06 &M ferredoxin:NADP+ reductase and no NADPH. Standard curves were linear from 0.08 to 2 nmol NADP(H).
Spectrophotometric Binding Titrations-Two matched cells were used ferredoxin:NADP+ reductase was in the sample cell, and an equal volume of buffer was in the reference cell. Spectra were recorded of the initial solutions and again after each addition of equal aliquots of ligand to sample and reference cells; mixing was by "plumper" (Calbiochem-Behring). Difference mixing spectra were calculated by subtracting the absorbance at each wavelength of the initial ferredoxin:NADP+ reductase solution from the dilution-corrected absorbance of the mixed ligand/ferredoxin:NADP+ reductase solution.
In some cases, difference mixing spectra were determined by use of tandem cells. Ferredoxin:NADP+ reductase was placed in the front compartment of the cell; an equal volume of ligand was in the rear. The reference cell contained only buffer. Spectra were recorded of the components before and after mixing the contents of the two compartments. Difference spectra were calculated arithmetically.
Calculation of Kd Values-Kd values were calculated either by a computer program adapted from Duggleby (17) or by a program developed in our laboratory for fitting binding curves to data by a nonlinear least squares procedure. The method of Duggleby (17) uses a Gauss-Newton method to approximate Kd and A A , , (absorbance change observed at saturating ligand), and best fit was determined by least squared error; K d values reported with standard errors were determined by this method. The other procedure was to vary M,, manually and calculate a "best" K d for each M,,; in each case, best fit was determined by minimization of the sum of squared error. Both programs used the following theoretical relation: This relation accounts for the fraction of ligand and enzyme in complex and thus was the correct expression for experiments in which a significant fraction of ligand was bound to the enzyme. The two procedures gave the roughly equivalent values for Kd and AA, , .

RESULTS
Demonstration of Ternary Complex Formation-We used difference mixing spectra to investigate the simultaneous association of Fd and NADP+ with ferredoxin:NADP+ reductase. First, Fd was added in a small molar excess to a ferredoxin:NADP+ reductase solution (Fig. 1A); spectrum a is the . NADP+ was then titrated into the Fd. FNR mixture, producing additional perturbations (spectra b-d) in the absorbance spectrum without obvious loss of the Fd. FNR mixing spectrum. Mixing Fd and NADP+ (no ferredoxin:NADP+ reductase) caused no spectral perturbation. Thus, it is likely that the spectral changes in spectra b-d correspond to formation of the ternary complex Fd. FNR. NADP+.
We found that the complex spectra (b-d) could be resolved into perturbations due to Fd. FNR and FNR. NADP+ association; subtraction of the spectrum of the Fd.FNR mixture from succeeding spectra (Fig. 1B) removes the initial Fd. FNR mixing spectrum, giving the perturbations caused by NADP+ in the presence of Fd. The shapes of the difference spectra were essentially the same as those observed in the absence of Fd? The lack of any absorbance changes indicating dissociation of Fd from ferredoxin:NADP+ reductase indicates that NADP+ is able to occupy its site on ferredoxin:NADP+ reductase and produce the FNR.NADP+ difference mixing spectrum without displacing Fd. Hence an Fd. FNR. NADP+ complex must be forming.
The absorbance of the mixture of ferredoxin:NADP+ reductase and ligands minus the sum of the separate absorbances of ferredoxin:NADP+ reductase plus ligand(s).
In Fig. lB, the NADP+ binding spectra have an upward deflection from 550 to 350 nm as compared to the spectra observed in the absence of Fd (6); this deflection was not observed if mixing spectra were produced by use of tandem cells. Such a deflection might result from small errors in corrections made for dilution.
We also observed apparent formation of ternary complex when Fd was titrated into FNR. NADP' complex (0.6 p~ ferredoxin:NADP+ reductase, 2 mM NADP+). As Fd was added, we observed absorbance changes corresponding to the appearance of the Fd. FNR complex, but could not detect significant disappearance of FNR. NADP+ complex; thus, order of addition of substrate was not important in our equilibrium experiments.
In view of data (presented below) suggesting interference between Fd and NADP+ in complex formation with the NADP+ reductase, we used ultrafiltration to determine whether Fd remained associated with ferredoxin:NADP+ reductase under the conditions of Fig. 1. Amicon PM-30 membranes retained ferredoxin:NADP+ reductase entirely, but Fd only slightly (<20% by absorption spectra of the filtrate). After addition of equimolar ferredoxin:NADP+ reductase to a 25 p~ Fd solution, essentially no Fd was found in the filtrate. Addition of NADP+ (up to 1.6 mM) did not cause any measurable appearance of Fd in the filtrate; however, after the solution was brought to 1 M NaCl by addition of solid NaCl (Fd'FNR complex is salt-dissociable (7, 14)), Fd was again found in the filtrate. Thus, a t concentrations of NADP+ which result in formation of NADP+ binding spectra, most Fd remained bound to ferredoxin:NADP+ reductase, confirming the conclusions drawn on the basis of difference mixing spectra.
Evidence for Interaction-Although Fig. 1 demonstrates ternary complex formation, we found that the associations of Fd and NADP+ with ferredoxin:NADP' reductase were not independent. Higher NADP+ concentrations were required to form the FNR. NADP' difference mixing spectrum in the presence of Fd than in its absence (Fig. 2). The K d of the complex between ferredoxin:NADP+ reductase and NADP+ was only about 14 p~ in the absence of Fd (in agreement with Dykes and Davis (18), who presented evidence suggesting that association of a second NADP+ at a low affinity site (& 2 1 mM) caused overestimation of K d in previous studies (5, 7)). Use of changes in A510-&50 as an indication of FNR. NADP' complex obviates the need to consider contributions from the low affinity site (18). When Fd was present, K d for FNR. NADP' complex was much greater than when Fd was absent ( Fig. 2B and Table I); no absorbance changes were observed which would indicate association of a second NADP' (data not shown).
Kd values in Table I were (Fig. 3). Fd.FNR complex was determined by A,,,-A,* (A47o is at or near an isosbestic point in the FNR. NADP+ difference mixing spectrum); thus, changes in the association of ferredoxin:NADP+ reductase and NADP+ should not interfere. With no NADP+ present, Fd-FNR complex was essentially fully formed a t 1:l FdFNR; We did not distinguish between binary Fd.FNR and ternary Fd. FNR. NADP+ complexes; we assumed that both complexes give identical absorbance changes. Hence Kd(Fd) was ([FNR-NADP+] +

Of K d ( F d t e r ) when [NADP+] >> 320 pM ( K d ( N A D P + t e r ) ) .
[ Kd cannot be accurately determined from such a titration (best fit was 20 nM), but K d must be less than 50 nM to give a sharp break at 1:l FdFNR.
As NADP' was increased, &Fd) increased, becoming measurable. The variation of K d versus [NADP+] showed saturation above 1 mM NADP+, as predicted by Equation 3. The curve drawn through the data is that predicted in Kd(Fdbj) = &(Fdter) = 0.55 pM. The data are in fair agreement with the theoretical line, but the wide data scatter permits only rough quantitation. The noise in these determinations results from use of low ferredoxin:NADP+ reductase concentrations which yield small absorbance changes upon complex formation. Certainly, increased with NADP' as we would expect; in addition, &Fd) seemed to become fairly constant at very high NADP+ concentrations.
In an earlier report, we concluded that ferredoxin:NADP+ reductase formed only binary complexes with its substrates (11). This conclusion was based on preliminary difference mixing spectra and upon elution of ferredoxin:NADP+ reductase from affinity columns (see below).
We found that titration of Fd into 100 pM NADP' + 32 pM ferredoxin:NADP+ reductase produced perturbations in Ab10 relative to changes in A456, suggesting a t least partial loss of FNR. NADP+ complex. We would now explain these observations as the result of the increased K d for ternary complex; spectrum. Apparent Kd,Fd) has been found to increase with NADP+ at this ionic strength (data not shown); this probably explains at least part of the dissociation.
Other Measures of Binding-We used several other measures of binding to assure that the observed interaction was not an artifact of difference mixing spectra. We titrated NADP+ into 53 p~ ferredoxin:NADP+ reductase f 75 p M Fd; free unbound NADP+ was determined as the concentration of NADP+ passing through an Amicon CF-25 m e m b~a n e .~ Bound NADP+ was determined as the difference between total NADP+ added and free NADP'. K d values were found to be 7 p~ (-Fd) and 156 p~ (+Fd); these values are in reasonable agreement (given the greater uncertainty of the filtration experiment) with those determined using difference mixing spectra (see Table I). At the higher NADP+ concentrations, bound NADP+ was evaluated as the difference between two larger numbers. Ferredoxin:NADP+ reductase binds to 2',5'-ADP-Sepharose (11, 19) and Fd-Sepharose (14); we have previously reported using these immobilized ligands to investigate interactions between Fd and NADP+ (or NADP+ analogues) in their associations with ferredoxin:NADP+ reductase.
Ferredoxin:NADP+ reductase bound to 2',5'-ADP-Sepharose, but was eluted by 5 p~ Fd in a sharp band the first fractions contained almost 1:l Fd:FNR (11). Analogously, 3 mM NADP+, but not NAD, was found to elute ferredoxin:NADP+ reductase from Fd-Sepharose (11). In more recent work, we have found (data not shown) that 2',5'-ADP would elute most Fd bound to FNR-Sepharose (ferredoxin:NADP+ reductase was covalently linked to cyanogen bromide-activated Sepharose). These observations, while qualitative, corroborate our conclusion that occupation of the NADP+-binding site decreases the association of ferredoxin:NADP+ reductase and Fd.
The Association of Fd, Ferredoxin:NADP+ Reductase, and N A D P i n Reduced States: EPR Studies-The preceding experiments all examine the interactions of Fd, ferredoxin:NADP+ reductase, and NADP+ in the fully oxidized state. We extended our studies to more reduced (and thus possibly more catalytically relevant) states. Paramagnetic probes were used to establish the existence of reduced Fd. FNR complexes and conversely, to investigate the effect of complexation on the proximity of the redox centers of ferredoxin:NADP+ reductase (FAD neutral semiquinone) and Fd (Fe2S,) to solvent.
Rapidly relaxing paramagnetic species (such as Dy3+, a rare earth) will strongly affect neighboring paramagnetic species through dipolar interactions (20)(21)(22). These interactions can cause the EPR signals of the affected centers to broaden and also to relax more rapidly (thus relieving power saturation); the closer the probe to the redox center, the stronger the interaction (20, 21). Thus, interaction is strongest when the redox center is exposed to solvent or when the probe binds to the protein. In this study, we used two probes: Dy3+ chelated by EDTA, a negatively charged probe, and Dy3+ chelated by o-phenanthroline, a positively charged probe. Excess dithionite was present in all samples; thus, ferredoxin:NADP+ reductase was also reduced. Dy-o- The CF-25-Centriflo membranes were found to retain ferredoxin:NADP+ reductase and Fd completely; no retention of NADP+ was observed in the absence of ferredoxin:NADP+ reductase. The neutral ferredoxin:NADP' reductase semiquinone, produced by aerobic addition of NADPH (23), was also accessible to Dy (Fig. 4B). Ferredoxin FNR,,,i,,,,,,). We could not determine whether or not NADP(H) was also bound to ferredoxin:NADP+ reductase.
Sepharose Binding Studies-We next investigated the effect of NADPH on stability of the Fd. FNR complex. We used the association of ferredoxin:NADP+ reductase with Fd-Sepharose as a convenient method of determining Fd . FNR association anaerobically. Ferredoxin:NADP+ reductase bound to Fd-Sepharose precipitated with the agarose beads; free reductase and FNR. NADP(H) complex remained in solution. Fig.  5 shows the data from two titrations. First, NADPH was titrated into an anaerobic mixture of buffer and ferredoxin:NADP' reductase bound to Fd-Sepharose (0). Little ferredoxin:NADP+ reductase was free of the Fd-Sepharose a t the beginning of the titration. 500 PM NADPH released much of the ferredoxin:NADP' reductase; however, addition of solid NaCl (bringing the solution to 0.5 M NaCI) released additional ferredoxin:NADP+ reductase from the Fd-Sepharose. Thus, NADPH decreases the association of initially oxidized Fd and ferredoxin:NADP+ reductase. We next titrated NADPH into dithionite-reduced Fd.FNR-Sepharose (Fig. 5, 0). As expected, reduction of the proteins decreased Fd. FNR complexation (9); more ferredoxin: NADP+ reductase was free in solution initially. Less NADPH was required to release ferredoxin:NADP+ reductase than in the previous titration; as before, some ferredoxin:NADP+ reductase remained bound to Fd-Sepharose even at very high NADPH.
These data can be simulated by assuming partial competition between NADPH and F d however, we do not present theoretical lines for negative cooperativity because too many variables are unknown to allow much confidence in such a calculation. We cannot be sure that all Fd molecules were bound to the Sepharose in the same manner; thus, ferredoxin:NADP+ reductase-binding sites might be heterogeneous. We could also not determine the extent of reduction of Fd and ferredoxin:NADP+ reductase by NADPH in the first titration. Changes in redox state have significant effects on Kd values of both Fd. FNR and FNR. NADP(H) complexes (9, 24).
Nonetheless, we can conclude that association of NADPH with ferredoxin:NADP+ reductase reduces the affinity of the flavoprotein for Fd-Sepharose. Some ferredoxin:NADP+ reductase remained associated with Fd-Sepharose at very high NADPH concentrations; this is compatible with formation of a ternary Fd. FNR. NADPH-Sepharose. It seems likely that in the reduced states, NADP(H) and Fd interact in the same fashion as in the oxidized state.
In a complementary experiment, we mixed Fd with a mixture of ferredoxin:NADP+ reductase and NADPH. Ferredoxin:NADP+ reductase and NADPH form a stable twoelectron reduced charge-transfer specie$ addition of oxidized anaerobic Fd from a side arm of the anaerobic cuvette caused loss of the long wavelength band diagnostic of the chargetransfer complex. In addition, absorbance changes at 456 nm suggested some electron transfer from NADPH to ferredoxin:NADP+ reductase. Thus, Fd altered the relationship between the flavin and pyridine nucleotide centers of ferredoxin:NADP+ reductase and NADP(H); we could not determine whether the NADP(H) dissociated from the flavoprotein or whether Fd merely abolished the charge-transfer interactions.
Stopped flow studies were conducted to investigate the rates at which Fd and NADP+ associate or dissociate with ferredoxin:NADP+ reductase. We sought to determine whether the changes in Kd resulted from increases in the rate of dissociation (kOrr) or from decreases in the rate of association (kJ ( K d = kofr/kon). The association of NADP+ with ferredoxin:NADP+ reductase was too fast to be observed (complete within 3 ms). But, we did observe a very fast increase in AbI0 ( k > 1000 s-') when we mixed NADP+ with preformed Fd. FNR complex (data not shown). This suggests that k,, decreases when Fd is bound to ferredoxin:NADP+ reductase, consistent with the increase in Kd; as the observed reaction was at the limit of observation, this conclusion is very tentative.
Association of Fd with ferredoxin:NADP+ reductase or with FNR.NADP+ was too fast to be observed. Similarly, the saltinduced dissociation of Fd.FNR complex was also complete within 3 ms. Thus, our data do not permit us to explain the changes in K d in terms of changes in association and dissocia-~-C. J. Batie and H. Kamin, unpublished observations. tion rates; we can be sure, however, that the tight association of Fd and ferredoxin:NADP+ reductase is not the result of a very slow dissociation of Fd.

DISCUSSION
Ferredoxin:NADP+ reductase has long been known to have binding sites for Fd and NADP+ (5-7); Ricard et al. (10) have also reported formation of ternary complex. The present data support the conclusion that both substrates can associate simultaneously with ferredoxin:NADP+ reductase; there are at least two substrate-binding sites. This conclusion is supported by the simultaneous appearance of Fd. FNR and FNR. NADP+ mixing spectra when both ligands are at high concen- The changes in apparent Kd(Fd) probably do not result from two classes of Fd-binding sites. No spectral perturbations were observed, indicating an FdFNR stoichiometry greater than 1:1. In addition, the spectral changes associated with complex formation were the same whether or not NADP+ was present; thus, Fd was probably occupying the same site. Similar arguments make it unlikely that changes in apparent &(NADP+) result from multiple binding sites. We could explain the data fairly well by a model of cooperative interactions (Scheme I); thus, although ternary complex (Fd.FNR. NADP( H)) can form, enzyme-substrate interactions are weaker than in binary enzyme-substrate complex.
Titration of NADPH into ferredoxin:NADP+ reductase bound to Fd-Sepharose was found to release most, but not all, of the ferredoxin:NADP+ reductase from the Fd-Sepharose. Ferredoxin:NADP+ reductase was released whether the proteins were initially oxidized or dithionite-reduced. Hence, we conclude that the partial competition observed in the oxidized states probably applies as well to reduced (and presumably more catalytically relevant) states. The dithionite-reduced Fd. FNR complex was disrupted a t lower NADPH concentrations than the initially oxidized Fd-FNR complex; this is probably a reflection of increase in Kd for Fd.FNR complex upon reduction of Fd (9).
Our conclusions are different from those of Ricard et al.
(lo), who reported that the Kd values for Fd . FNR or FNR, NADP+ complex (determined by difference mixing spectra) did not change if the other substrate was present. In light of their report, we also used binding assays in other than mixing spectra. K d values for FNR. NADP+ complex were similar whether binding was assayed by difference mixing spectra or ultrafiltration. In addition, qualitative studies using immobilized ligands confirmed that NADP+ (or its analogue, 2',5'-ADP) caused partial dissociation of Fd. FNR complex. We cannot explain the differences between the studies; in our hands, mixing experiments designed to meet the conditions of the studies of Ricard et al. (10) indicated that Fd interfered with FNR.NADP+ association. The two studies also differ with regard to the effect of Fd on the absorbance changes associated with FNR. NADP+ binding. Ricard et at. (10) found that Fd caused a large decrease in maximal absorbance changes, but with no change in shape of the FNR-NADP+ difference mixing spectrum. We, however, found that at saturating NADP', the absorbance perturbations were as large (possibly even somewhat larger) when Fd was present than when Fd was absent. Two paramagnetic probes differing in charge (positively charged Dy-o-phenanthroline and negatively charged Dy-EDTA) were used to perturb the EPR spectra of ferredoxin:NADP' reductase and Fd (semiquinone and reduced, respectively). The positively charged probe broadened the signal of the acidic (negatively charged) ferredoxin; Dy-EDTA did so also, but to a lesser extent. On the other hand, Dy-EDTA was much more effective in perturbing the radical signal of ferredoxin:NADP+ reductase than was Dy-o-phenanthroline. Mixing Fd and ferredoxin:NADP+ reductase substantially decreased the line broadening by either agent; thus, one-electron reduced (Fdoridized. FNR8emiq,,in,,ne) and three-electron reduced (Fdreducd. FNRredufed) complexes form.
The x-ray crystal structure of algal Fd shows the iron-suafur center to be quite exposed (both Fe molecules within 5 A of solvent); this may explain the accessibility of Fd to both probes. Superimposing the sequence of spinach Fd (25) on the structure of the algal Fd (26, 27) reveals that the portion of the molecule closest to the iron-sulfur center has a net negative charge; this could explain the greater effectiveness of Dyo-phenanthroline.
The present data indicate that the flavin of ferredoxin:NADP+ reductase is exposed to solvent at a positively charged site. We are thus in agreement with Zanetti et al. (28) who concluded that the FAD within ferredoxin:NADP+ reductase is exposed at the 8-position but not the pyrimidine portion of FAD. Zanetti (29) has also reported a lysine residue essential to NADP+ binding; this might contribute one positive charge. It has also been suggested that about 5 tyrosyl residues are buried on formation of the Fd.FNR complex (30).
The charge specificity of line broadening is consistentwith description of Fd. FNR complex as a salt between negatively charged Fd and positively charged ferredoxin:NADP+ reductase. The sensitivity of complex to increasing ionic strength is consistent with predominantly electrostatic interactions (7). Complex formation between Fd and ferredoxin:NADP+ reductase may bring the centers close together to facilitate electron transfer, and in the process bury them within a protein-protein complex, shielding them from the probes. Without further data, we cannot tell whether or not the extent of exposure of either FAD or iron-sulfur center to solvent differs in Fd. FNR and Fd. FNR. NADP(H) complexes.
The ternary Fd. FNR . NADP+ complex identified here may participate in catalysis. Masaki et al. (14) have investigated the steady state kinetics of electron transfer from Fd to NADP+ catalyzed by the ferredoxin:NADP+ reductase from Spirulina, a blue-green alga; the kinetic pattern was consistent with a requirement for a ternary (Fd.FNR.NADP+) complex in the catalytic cycle. In addition, they found that electron transfer was inhibited by excess reduced Fd when NADP+ was near its K,. We have identified a ternary Fd.FNR. NADP' complex and found that Fd will inhibit the association of NADP+ with ferredoxin:NADP' reductase.
Rapid kinetic studies of electron transfer among Fd, ferredoxin:NADP+ reductase, and NADP+ (these data will be reported in another paper in this series) indicate that ternary Fd. FNR . NADP+ complex must participate in catalysis. Our studies indicate that the rate of electron transfer from Fd to NADP+ (uia ferredoxin:NADP+ reductase) may be limited by dissociation of oxidized Fd. Other workers (36) have reported that electron transfer from Fd to ferredoxin:NADP+ reductase, in the absence of oxidized Fd, is very fast. Masaki et al. (4) found that oxidized Fd was a potent competitive inhibitor with respect to reduced Fd. Thus, destabilization of the Fd-FNR complex may facilitate photosynthetic NADP+ reduction.
Chemically analogous electron transfer systems are found in adrenal mitochondria and in Pseudomonas bacteria; they catalyze electron transfer from NAD(P)H to cytochromes P-450 uia an FAD-containing protein and an Fe2Sz* protein (22, 31). Ternary NADP(H)-adrenodoxin reductase-adrenodoxin complex has been identified; it has been shown that electron transfer from NADPH to adrenodoxin proceeds uia the ternary complex (22, 32). Lambeth et al. (22, 32) did not test for interactions between NADP(H) and adrenodoxin in formation of ternary complex with the reductase. The mechanism of the putidaredoxin/putidaredoxin reductase system (the bacterial iron-sulfur protein and flavoprotein (31)) has not yet been described, nor has ternary complex been reported. Thus, it is not clear whether a destabilized ternary complex facilitates other electron transfers among pyridine nucleotides, flavoproteins, and iron-sulfur proteins.
The decrease in NADP(H) binding caused by Fd may explain earlier observations that Fd inhibits the diaphorase and transhydrogenase activities of ferredoxin:NADP+ reductase (33, 34). Nakamura and Kimura (33) described the inhibition of diaphorase activity (electron transfer from NADPH to dichlorophenol-indophenol) as being partially competitive with respect to NADPH. We have also observed inhibition of ferricyanide reductase activity by Fd, competitively with respect to NADPH (data not shown). Nelson and Neumann (34) reported that Fd inhibited electron transfer from NADPH to NAD. Decreased NADPH binding caused by association of Fd with ferredoxin:NADP' reductase could explain both results.
These studies to not address the question of the mechanism of interaction between Fd and NADP+. Steric hinderence between Fd and NADP+ could provide an explanation, or, alternately, occupation of one site could cause changes in ferredoxin:NADP+ reductase conformation which decrease the association of the other substrate with ferredoxin:NADP+ reductase. Resolution of these possibilities may have to wait upon further clarification of the structure of ferredoxin: NADP+ reductase.