ARCHIVES
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 288, No. 1, July, pp. 231-238, 1991
Interaction of Ferredoxin-NADP+ Reductase from Anabaena with Its Substrates Javier
Sancho’
and Carlos
G6mez-Moreno’
Departamento de Bioquimica y Biologia Molecular y Cellular, Facultad de Ciencias. Urhersidad de Zaragoza 50009 Zaragoza, Spain
Received December 7, 1990, and in revised form March 7, 1991
The interaction of ferredoxin-NADP+ reductase from Anabaena variabilis with its subthe cyanobacterium strates, NADP+ and ferredoxin, has been studied by difference absorption spectroscopy. Several structural analogs of NADP+ have been shown to form complexes the stabilities of which are strongly dependent on the ionic strength of the medium. In most cases the binding energy of these complexes and their difference absorption spectra are similar to those reported for the spinach enzyme. However, NADP+ perturbs the absorption spectra of the Anabaena and spinach enzymes in a different way. This difference has been shown to be related to the binding of the nicotinamide ring of NADP+ to the enzymes. These results are interpreted as being due to a different nicotinamide binding site in the two reductases. The enthalpic and entropic components of the Gibbs energy of formation of the NADP+ complex have been estimated. An increase in entropy on NADP+ binding seems to be the main source of stability for the complex. A shift of approximately 40 mV in the redox potential of the couple NADP+/ NADPH has been observed to occur upon binding of NADP’ to the oxidized enzyme. This allows us to calculate the binding energy between the reductase and NADPH. The ability of the reductase, ferredoxin, and NADP+ to form a ternary complex indicates that the protein carrier binds to the reductase through a different site than that of the pyridine nucleotide. o Issl Academic Press,
Inc.
The absorption spectra of flavin coenzymes, FMN and FAD, are perturbed when they become incorporated into proteins. The most prominent change is usually indicated by the appearance of shoulders associated with the main absorption band at approximately 445 nm. These shoul’ Present address: Cambridge Center for Protein Engineering. Medical Res. Center, Hills Rd, Cambridge, CB2 2&H U.K. ’ To whom correspondence should be addressed. FAX: 34-76-567920. 0003.9861/91 $3.00 Copyright 0 1991 by Academic Press, All rights of reproduction in any form
ders have been related to different vibrational levels of the electronic transition (1,2) and have been found in at least 40 flavoenzymes. Owing to the fact that similar changes occur when the flavin coenzyme is dissolved in organic solvents, it has been proposed that these changes are caused by interaction of the organic chromophore with apolar residues in the polypeptide chain (3, 4). It is generally accepted (5,6) that the enhancement of the shoulders in the 445-nm band is due to a decrease in the polarity of the environment of the prosthetic group, though it could simply reflect a more rigid environment of the flavin. Ferredoxin-NADP+ reductase (FNR)” is the flavoenzyme involved in the transfer of electrons from ferredoxin to NADP+, thus providing the reducing factor necessary for most biosynthetic reactions in plants. The enzymes, isolated from either higher plants (7), algae (8), or cyanobacteria (9), all carry a single FAD group that participates in the electron transfer reaction. The absorption spectrum of FNR is similar to that of many other flavoproteins. Early studies on the spinach enzyme had revealed that spectral perturbations appear as a consequence of the formation of a complex with several electron transfer proteins such as ferredoxin, flavodoxin, or rubredoxin (10, 11). Binding of NADP(H) to ferredoxin-NADP+ reductase has also been reported to produce significant changes in its absorption spectrum in the visible region (12). Difference absorption spectroscopy can be used to measure the binding constant for these complexes since they have a well-defined stoichiometry (11-13). Comparison of the interaction between ferredoxin-NADP’ reductase with ferredoxin and with tlavodoxin is of interest since both electron transport proteins are natural substrates for the enzyme in cyanobacteria. Data are presented that show that cyanobacterial FNR forms complexes of similar binding strength with ferredoxin and flavodoxin from dif-
3 Abbreviation
used: FNR, ferredoxin-NADP+
reductase. 231
Inc. reserved.
232
SANCHO
AND
ferent sources. Measurements of the affinity of different NADP+ analogs to ferredoxin-NADP+ reductase from Anabaena provide evidence that different parts of the NADP+ molecule are involved in the binding to the reductase. From the temperature dependence of the dissociation constant of the NADP+-FNR complex, an estimation of the entropic and enthalpic components of the Gibbs energy of binding has been made. MATERIAL
AND
GOMEZ-MORENO
a 0.02
METHODS
Proteins and reagents. Anabaena uariabilis ferredoxin-NADP+ reductase was purified as described earlier (9). Ferredoxin from A. uariubilis and spinach were purified essentially as in (14,15). Azotobacter uinehdii flavodoxin, purified according to (16), was a gift from Dr. D. E. Edmondson (Emory University). Apoflavodoxin was prepared by a modification of the method described in (2). Nucleotides were from Sigma or Boehringer. An extinction coefficient of 15.4 mM-’ cm-’ at 260 nm has been used to quantify 2’-AMP, 2’,5’-ADP, and 2’5.ATP ribose. All other reagents used in this study were of analytical grade. Complex formation. All experiments were performed at 25°C in a thermostated Kontron Uvikon 860 spectrophotometer using two-compartment cuvettes. Samples of 1 ml containing 40 pM Anubaena reductase were placed in the front compartment of the sample and of the reference cuvettes. The same volume of buffer was added to the rear compartments of both cells. A reference spectrum was then run and electronically subtracted from any further spectra. Equal volumes (usually 1 to 5 ~1) of electron-carrier protein, NADP+, or NADP’ analogs were added to the front compartment of the sample cuvette and to the rear compartment of the reference cuvette, using spectrophotometrically calibrated Hamilton microsyringes. An equivalent volume of buffer was added each time to the front compartment of the reference cuvette to compensate for the dilution of the enzyme in the sample cuvette. The difference spectrum was then recorded. The dissociation constants and extinction coefficients of the different complexes were calculated by nonlinear fitting of the data assuming a 1:l stoichiometry. The program described in (17) was used with a modification which takes into account all dilution effects during titration. A 1:2 stoichiometry was not consistent with the experimental data. The total amount of ligand added in each experiment was never higher than 200 ~1. Most of the Kd values here reported were calculated by fitting more than 20 data points to the theoretical equation (18). The buffer used was 50 mM, pH 8.0, Tris/HCl unless otherwise indicated. Redor potential. The redox potential of Anabaena ferredoxinNADP+ reductase was determined by chemical titration with dithionite as described earlier (9) but in the absence of glycerol. The midpoint potential of the reductase in the presence of NADP+ was determined in a mixture of NADP+, glucose 6-phosphate, and glucose-6-phosphate dehydrogenase as an NADPH generating system. The NADP+/NADPH couple was used as the redox indicator. The chemical titration was carried out at 25OC in 50 mM sodium phosphate buffer, pH 7.0, and the titration with NADP+/NADPH in 5 mM buffer at the same temperature. The spectral changes produced during redox titration were determined with a Hewlett-Packard diode array spectrophotometer.
RESULTS
Complex Formation between Ferredoxin-NADP’ Reductase and NADP+ or its Analogs Ferredoxin-NADP+ reductase from Anabaena forms a complex with NADP’ that can be detected and quantified from the spectral changes that are observed in the ultraviolet and visible absorption spectra of the mixture (Fig.
3
345
430
WAVELENGTH
0
50
100
515
c$010
(nm)
150
200
WDP + 1 WW FIG. 1. Anabatma ferredoxin-NADP’ reductase-NADP+ complex. (a) Difference spectra on addition of increasing amounts of NADP’. The experiment was performed at 25’C with the enzyme dissolved in 50 mM Tris/HCl buffer, pH 8.0. Aliquots of a concentrated NADP+ solution (2.4 mM) were added stepwise (1 to 5 ~1 additions) as described under Materials and Methods. (b) Best fit of the change in absorbance (460 nm minus 400 nm) to a 1:l stoichiometry.
la). Plotting the spectral changes vs nucleotide concentration (Fig. lb) yields a saturation curve. The binding constant of the complex has been calculated to be 13.2 PM assuming a 1:l stoichiometry. In order to determine which parts of the nucleotide molecule are essential for binding to the active site in the enzyme several NADP+ analogs have been studied with respect to their ability to bind ferredoxin-NADP+ reductase. The Kd values for these complexes are compared in Table I with dissociation constants measured for the complexes of the same analogs with other NADP(H)-dependent enzymes. NAD’, when added up to a 2 mM con-
233
FERREDOXIN-NADP+ REDUCTASE FROM Anabaena TABLE I Dissociation Constants, Binding Energies, and Extinction Coefficients of the Complexes between NADP+ (or other analogs) and Ferredoxin-NADP+ Reductase or Glutathion Reductase AG; (kcal/mol)
K, (PM)
c (mM-’
cm-‘)
Nucleotide
A. var.
Spinach
GR
A. uar.
Spinach
GR
A. var.
Spinach
GR
NADP’ 2’,5’-ATP ribose
13.25 10.46 15.38 73.49 1500
14 8 2.4 120 1530
70 75 30 500 2900
-6.63 -6.77 -6.54 -5.62 -3.84
-6.60 -6.92 -7.64 -5.33 -3.82
-5.64 -5.60 -6.14 -4.46 -3.45
1.19 1.49 1.14 0.37
0.80 0.71 0.70 0.22
-
2’,5’-ADP 2’-AMP NAD(H)
Note. Data for spinach FNR come from (13) and for glutathion reductase (GR) from (19). The extinction by Anabaena FNR have been calculated from the difference in absorbance at 460 with respect to 400 nm.
centration, was able neither to produce any spectral perturbation nor to displace NADP+ bound to the enzyme (data not shown). This indicates that the 2’-phosphate group-absent in NAD+-is essential for the binding of NADP+ to the FNR from Anabaena. The same requirement has been found for other pyridine nucleotide-dependent enzymes such as glutathione reductase, where the 2’-phosphate group causes a 40-fold increase in the binding constant of the complex (19) with the enzyme. Of all the other nucleotides tested, only 2’-AMP forms a weaker complex than NADP+, which suggests that the negative charge of the NADP+ pyrophosphate bridge, absent in 2’-AMP, stabilizes the complex between the reductase and NADP+. This is confirmed by the fact that 2’,5’-ADP-which has only one of the two phosphate moieties of the bridge-binds the enzyme almost as strongly as NADP+. We have observed that the stability of the complexes between the different nucleotides and the reductase is strongly dependent on the ionic strength of the medium. High ionic strengths, such as 0.5 M NaCl, completely abolish the formation of the complexes, indicating an important contribution of electrostatic interactions to their stability. In a parallel set of experiments, the temperature dependence of the binding constant of the NADP+-FNR complex was studied in order to obtain information on the forces involved in its stabilization. Our data indicate that the binding constant for NADP+ does not change significantly in the range from 7 to 25°C (& from 13.2 to 13.85 PM). This corresponds to a A(& value of approximately -6.6 kcal/mol. Calculation of AH; for the binding of the complex using the Arrhenius plot has proved to be difficult due to the very shallow slope of this plot. From our data we can estimate a m0 value of approximately 0.2 to -1.2 kcal/mol. This low value of the enthalpy of binding implies that the complex is stabilized mainly by entropic forces.
coefficients
of the complexes formed
Complex Formation between Ferredoxin-NADP’ Reductase and Ferredorin or Flavodoxin Ferredoxin-NADP’ reductase from Anabaena forms stable complexes with ferredoxin, from both spinach and Anabaena. The Kd values determined from the spectral changes produced upon mixing both proteins are in the micromolar range, as reported for the spinach reductase (11). It has been proposed that flavodoxin can replace ferredoxin in many of the physiological reactions in which this iron-sulfur protein participates, including those involving ferredoxin-NADP+ reductase (20). Anabaena flavodoxin has been shown to interact with ferredoxinNADP+ reductase from the same organism in what can be considered the preliminary step in the electron transfer reaction between these proteins (21). Since Azotobacter flavodoxin shows a lower activity in the reactions assayed with the reductase, i.e., the cytochrome c reductase and NADP+ photoreductase activities (22), it was of interest to determine whether the two flavoproteins interact to form a complex at all. Data in Table II indicates that the strength of the complex formed by both proteins is, in
TABLE II Dissociation Constants, Binding Energies, and Extinction Coefficients of the Complexes between Different Electron Carrier Proteins and Ferredoxin-NADP+ Reductase from Anabaena Electron
carrier
Anabaena ferredoxin Spinach ferredoxin Azotobacter flavodoxin Apoflavodoxin
Kc, (PM)
AG; (kcal/mol)
3.57 1.25 7.3 73.8
-7.4 -8.0 -7.0 -5.6
6 (mM-’
cm-‘)
1.49 1.57 0.6 0.7
Note. Extinction coefficients have been determined at 460 nm for the ferredoxin complexes and at 400 nm for the complexes with flavodoxin or apoflavodoxin.
234
SANCHO
AND
GGMEZ-MORENO
state exhibits NADP+.
different
binding
sites for ferredoxin
Redox Behavior of Ferredoxin-NADP’ Presence and Absence of NADP’
-‘.O 1 300
400 WAVELENGTH
500
600
(nm)
FIG. 2. Difference spectraof the ternary complex (ferredoxin-FNRNADP’). A 15 WM solution of Anabaena reductase was saturated with Anabaena ferredoxin (53 pM) and a baseline spectrum was recorded. Aliquots of a 2.4 mM NADP+ solution were then added stepwise (5 ~1 each time) to develop the spectral perturbation shown in the figure. Temperature and buffer were as in Fig. 1.
fact, very similar to that of ferredoxin. Furthermore, the ability of flavodoxin from Azotobacter to bind the reductase is not lost when the FMN group is removed from the carrier. The FNR-apo-flavodoxin complex is, however, weaker by a factor of 10 than that in which the holoprotein is involved. None of these reductase-electron-carrier complexes were stable in 0.5 M NaCl, indicating an important contribution of electrostatic interactions to their stability.
and
Reductase in the
In previous work (3) we have reported a -320 mV value for the EL of Anabaena reductase at pH 7.0, as determined by spectrocoulombimetric measurement in the presence of glycerol. We have now obtained the same value in its absence. Batie and Kamin (13) have reported, for the spinach enzyme, that the redox potential of NADP+ became 40 mV more positive when the pyridine nucleotide was bound to ferredoxin-NADP+ reductase. This is an important finding since it allows calculation of the relative affinity of the enzyme for the oxidized and reduced nucleotide substrate. We have performed a similar experiment aimed at determining the stability of both complexes in the cyanobacterial enzyme. To an anaerobic mixture containing 30 PM Anabaena ferredoxin-NADP+ reductase, 370 PM NADP+, and excess glucose 6-phosphate, an anaerobic solution containing catalytic concentrations of glucose-6-phosphate dehydrogenase was added. The spectral changes that occurred upon mixing were followed with a rapid scan spectrophotometer. The NADPH concentration was calculated from the absorbance at 340 nm, which is an isosbestic point for the reduction of the Anabaena reductase. Since the initial concentration of NADP+ was known, the redox potential in the solution could be calculated throughout the reaction. A plot of the absorbances at 459 and 600 nm vs the redox potential in the solution is shown in (Fig. 3). A drop in the absorbance at 459 nm closely parallels an increase in that at 600 nm, which is indicative of the formation of a stable chargetransfer complex between the oxidized reductase and NADPH, as has been described for the spinach enzyme (13). This behavior differs from what was observed when
The Ternary Complex Ferredoxin-FNR-NADP’ Three different molecules participate in the reaction catalyzed by FNR. It is of mechanistic interest to determine whether a ternary complex of the enzyme, the nucleotide, and the electron carrier protein can be formed. Figure 2 shows the differential spectrum obtained by addition of NADP+ to a mixture of the reductase and ferredoxin from Anabaena. Apart from the presence of ferredoxin, the titration was performed under the same conditions as in its absence. The calculated value of the Kd for the Anabaena FNR-NADP+ complex in the presence of a saturating concentration of ferredoxin has been found to be 15.3 FM. This value is very similar to that found in the absence of ferredoxin (see Table I). This indicates that the enzyme from Anabaena in its oxidized
REDOX
POTENTIAL
(mV)
FIG. 3. Redox titration of Anabaena ferredoxin-NADP+ reductase in the presence of NADP+. To an anaerobic solution of Anabaena FNR containing 370 pM NADP+ and excess of glucose 6-phosphate catalytic amounts of glucose-r-phosphate dehydrogenase were added. The absorbance spectrum of the solution was then measured at different times with a rapid scan spectrophotometer. Open symbols: absorbance at 459 nm; closed symbols: 3X absorbance at 600 nm.
FERREDOXIN-NADP+
REDUCTASE
ferredoxin-NADP+ reductase was titrated with dithionite in the absence of NADP+ [see (9)]. There, the absorbance at 600 nm rose to a maximum which was only approximately 7% of that at 459 nm and remained at that level until the end of the titration, indicating the formation of the semiquinone free radical that is quite unstable in this enzyme. No charge-transfer complex was formed under those conditions. The midpoint potential for this initial reductive process was calculated by plotting log (ox)/(red) vs the potential assuming that the redox process involves two electrons and also that the millimolar extinction coefficient for the FNR-NADP’ complex is 10.04 mM ’ cm-l and that for the reduced protein is 0.6 rnM-~’ cm-‘. A value of -275 mV is calculated, which is approximately 40-50 mV more positive than the potential described for free NADP+/NADPH in solution at pH 7.0 (-320 mV). It appears that once the charge-transfer complex is formed, and as more negative potentials are generated by the continuous production of NADPH, a second reduction process which corresponds to the main decrease in the absorbance at 459 nm takes place. At the same time only a partial decrease in the absorbance at 600 nm can be observed. This can be interpreted as a consequence of the reduction of FAD in the oxidized reductase-NADPH charge-transfer complex. From the data presented here it is difficult to calculate an accurate value of potential for this process since the reduction does not proceed to completion; however, it appears to approach the previously determined value of -320 mV (9). The residual absorbance observed at 600 nm could be due to the semiquinone radical that remains at this point, since full reduction has not yet occurred.
FROM
235
Anabaena
Anabaena
DISCUSSION The Complex between NADPc and FNR from Anabaena: Comparison with the Complex Formed by FNR from Spinach
The binding of a relatively large molecule, like NADP+, to an enzyme is the result of several simultaneous interactions. In order to assess which groups, within NADP+, favorably contribute to its binding to FNR, the dissociation constants of the complexes formed by the reductase and several NADP+ analogs have been determined from the spectral changes produced during complex formation. Comparison of the binding energies of different pyridine nucleotide analogs to ferredoxin-NADP+ reductase can provide important information on the nature of the forces involved in such binding. Figure 4 shows the energies of interaction in these complexes. The low affinity of the reductase for NADH suggests that the phosphate group in the 2’ position of NADP+ is effectively involved in binding to the enzyme. This has also been described for the spinach enzyme and the NADPH-dependent flavoenzyme glutathione reductase. X-ray data of the later en-
2’5’ADP
FIG. 4. Binding energy diagram of the complexes between NADP+ (and analogs) and several flavo-oxidoreductases. The binding energies of the Anabaena reductase complexes (solid line) have been determined from their dissociation constants (see Materials and Methods). Data for spinach reductase (light line) and for glutathione reductase (high ionic strength; dotted line) come from Refs. (13,351 and (191, respectively.
zyme indicate that the formation of a complex with NADP+ takes place by an induced-fit mechanism (19). In this model the interaction of the 2’-phosphate group of NADP+ with the enzyme contributes to a change in position of several amino acids in the binding site, thus facilitating the right conformation for effective binding. The importance of the 2’-phosphate group is supported by the finding that binding of 2’-AMP to Anabaena ferredoxinNADP+ reductase is much stronger than that of NADH, although 2’-AMP lacks the 5’-phosphate group present in NADH that is known to be important for binding (23). The increase in stability (1 kcal/mol) found for the 2’,5’-ADP complex with respect to 2’-AMP would indicate the existence of a strong interaction between the phosphate group in the 5’ position and a positively charged group in the reductase. We have obtained direct evidence for the presence of an arginine residue, in the reductase from Anabaena (23), which interacts with the pyrophosphate bridge in NADP+. This has also been proposed for the Bumilleriopsis enzyme (12). The similarity of the binding energy values of the NADP’ and 2’,5’-ADP complexes suggests that the nicotinamide nucleotide moiety within NADP+ does not contribute to the stability of the enzyme-NADP+ complex. When the stabilities of the complexes formed by NADP+ and analogs with the reductases from either Anabaena or spinach are compared, similar values are found with only one exception. The 2’,5’-ADP complex with the spinach enzyme is approximately 1 kcal/mol more stable than that with the cyanobacterial enzyme or when compared to the complexes formed with either 2’,5’-ATP ri-
236
SANCHO
AND
bose or NADP+ (See Fig. 4). This indicates that the nicotine nucleotide within NADP+ (absent in 2’,5’-ADP) introduces a destabilizing effect in the spinach enzyme complex that is absent in the complex with the reductase from Anabaena. A similar destabilizing effect has also been observed for glutathion reductase (19). Though this last result should be regarded with caution, since it was obtained at high ionic strength, it appears that in none of these cases the nicotine nucleotide within NADP+ contributes to stabilize the complex between NADP+ and these reductases. Another difference with respect to the interaction of the FNR from several sources with the nicotinamide ring in NADP+ has been found by comparing the difference spectra obtained for the NADP+ and 2’,5’-ATP ribose complexes with the reductases from Anabaena, Bumilleriopsis, and spinach. Figure 5 shows that NADP+ produces a very similar perturbation in the visible spectrum of the enzymes from both Anabaena and Bumilleriopsis. This perturbation is almost identical to that produced by 2’,5’ATP ribose (a molecule similar to NADP+ but lacking the nicotinamide ring) in both Anabaena and spinach FNR. However, in the spinach enzyme, the effect of NADP+ is completely different. This shows that the nicotinamide ring is not able to alter the absorption spectrum of the prosthetic group in the reductases from Anabaena and Bumilleriopsis, suggesting a weak interaction of the nicotinamide with the NADP+ binding site of these enzymes in their oxidized state. On the contrary, the spectrum of spinach FNR is clearly modified by the interaction with the nicotinamide. This points either to a direct interaction between the nicotinamide and the FAD group of the spinach reductase or to a nicotinamide-induced reorganization of the NADP+ binding site that changes the absorption properties of the FAD group. The stability of the FNR-NADP+ complex is rather temperature independent. We estimate that at least 5-6 kcal/mol of the binding energy is due to entropic forces. Studies by Janin and Chothia have suggested an essential role for hydrophobic interactions in the binding of coenzymes to several enzymes (24). They have pointed out that the loss in surface area that often occurs upon binding is enough to account for the stability of these complexes. Nevertheless, electrostatic interactions are also very important in the stability of the NADP+-FNR complex since it is destroyed by high ionic strength media. We therefore propose that the entropic stabilization of the FNR-NADP+ complex is afforded both by the loss of surface area and by electrostatic interactions that release water molecules initially associated with the interacting charges. In spite of the differences observed for the interaction between NADP+ and ferredoxin-NADP’ reductase from both Anabaena and spinach, the shift in the redox midpoint potential of the nucleotide that was reported to occur upon binding to the spinach enzyme also appears to occur in the cyanobacterial enzyme. By comparing the Gibbs
GOMEZ-MORENO 2’5’
NADP+
- ATP ribose
I
WAVELENGTH
(nm)
FIG. 5. Comparison of difference spectra of the complexes between NADP+ or 2’,5’-ATP ribose and ferredoxin-NADP+ reductase from different species. Spectra of the spinach and BumiUeriopsis complexes come from Refs. (13, 35) and (12), respectively.
energies obtained from the redox potential of the couples NADP+/NADPH (EL = -320 mV) and FNR-NADP+/ FNR-NADPH (Eb = -275 mV) with the dissociation constant of the FNR-NADP+ complex (& = 13 PM) we calculate a value of 0.39 /.LM for the dissociation constant of the complex between FNR and NADPH. This indicates that the complex between the enzyme and the oxidized nucleotide is less stable than the complex with the reduced nucleotide. The Complex between FNR from Anabaena and Electron Carrier Proteins All the perturbations induced on the spectrum of ferredoxin-NADP+ reductase by the different electron carrier proteins that we have tested differ from those induced by NADP+ and its analogs, in that they have only positive extinction coefficients in the visible region. The similarity of the difference spectra of the complexes between FNR from Anabaena and ferredoxin from Anabaena and spinach (present work), and those of FNR and ferredoxin from spinach (ll), lettuce (lo), and Bumilleriopsis (12), suggests that the ferredoxin binding site is similar in the reductase from all these origins. We have also determined the stability of the complex between the reductase from Anabaena and Azotobacter flavodoxin and found that it is
FERREDOXIN-NADP’
REDUCTASE
similar to that of the complexes with ferredoxin. Although we cannot conclude that the perturbation observed in the visible region is due only to a perturbation of the FAD chromophore in the reductase, it seems clear that these carriers do perturb the spectrum of the enzyme since apoflavodoxin from Azotobacter, which does not absorb in the visible, is still able to alter the absorption spectrum of Anabaena FNR (not shown). The dissociation constants measured for all the complexes between the Anabaena reductase and ferredoxin or flavodoxin from several origins [present work and (al)] are similar. Cross-linking experiments performed in our laboratory (25) have demonstrated that Anabaena ferredoxin and Azotobacter Aavodoxin do indeed share a common binding site. All these complexes are unstable at high ionic strength, which indicates that they are stabilized by electrostatic interactions. Carboxyl groups in the carriers (26, 27) and lysil (28) and arginine (23) groups in the reductase are likely to be responsible for the electrostatic stabilization of these complexes. The Ternary Complex: Ferredoxin-FNR-NADP+ It has already been suggested (29, 30) from difference spectroscopy studies that the transfer of electrons from reduced ferredoxin to NADP+ takes place with the formation of a ternary complex between both substrates and the enzyme from spinach. The observation we have made, that ferredoxin-NADP+ reductase binds NADP+ even in the presence of ferredoxin, is indicative of the formation of a ternary complex that could be involved in the reaction catalyzed by FNR. This proposal is also supported by other experiments in which it was shown that a covalent cross-linked complex between the reductase and the ferredoxin or the Aavodoxin could still bind to, and also be reduced by, NADPH (25). Measurements of the presteady state kinetics of this reaction in the non-cross-linked complex are currently under way in order to determine if FNR can accept electrons from ferredoxin when bound to NADP+ as a first step in the mechanism of the reaction of this enzyme. The Use of Difference Spectroscopy to Obtain Structural Information Difference absorption spectroscopy has been widely used to detect intermolecular interactions as well as to quantify their strength (31) but it has other potential uses we would like to point out. It can give detailed information concerning the changes that occur in the environment of the chromophoric groups within a protein as a consequence of binding of a ligand. Information of this kind might be attainable if the changes observed in the difference spectra could be related to environmental properties of the chromophore within the protein. In the case of flavoenzymes, circular dichroism studies (2) have shown that the absorption spectrum in the visible region has
FROM
237
Anabaena
350
400
450
WAVELENGTH
500
550
(nm)
FIG. 6. Distribution of peaks from difference spectraalong the absorbance axes. The absorbance axis has been divided into 5-nm windows. The number of peaks that appear in each of the windows in all of the difference spectra reported in this work is represented in this figure.
contributions from different vibronic transitions associated with the two main absorption bands. Indeed, Miiller et al. (1) have correlated temperature-dependent changes in the absorption spectrum of ferredoxin-NADP+ from spinach with five vibronic transitions previously described (3). In trying to relate the peaks that appear in our difference spectra to these transitions we have defined 5-nm windows in the wavelength axis and have counted the number of peaks (both positive and negative) that appeared in each of these windows in all of our difference spectra. Thirty-seven peaks have been considered. Figure 6 shows our results. It emerges that the peaks cluster in five regions with maxima at 397, 432, 457, 480, and 507 nm. These maxima relate well with the wavelengths of the vibronic transitions described previously (375, 435, 455, 489, and 510 nm) (3). Though the nature of these transitions is still not well characterized, progress in our understanding is expected to result from molecular orbital studies (31) and Raman spectroscopy (33,34). Once these transitions are better characterized, difference absorption spectroscopy could give rapid qualitative information on the effect of the binding of ligands and on the environment of the prosthetic group. ACKNOWLEDGMENTS This work was supported in part by Grant BAP 0397.E from the Commission of the European Communities and by Grant BIO 88/0413 from the Comision Interministerial de Ciencia y Tecnologia (Spain). We thank Dr. D. E. Edmondson from Emory University for providing us with A. uinelandii Aavodoxin and for interesting discussions and Drs R. Loewenthal from Cambridge University and H. J. Loewenthal from the Israeli Institute of Technology, for help with the manuscript.
REFERENCES 1. Miiller, F., Mayhew, S. G., and Massey, V. (1973) Biochemistry 12, 4654-4662. 2. Edmondson, D. E., and Tollin, G. (1971) Biochemistry 10, 113124. 3. Gordon-Walker, A., Penzer, G. R., and Radda, G. K. (1970) Eur. J. Biochem. 13, 313-321.
238
SANCHO AND GOMEZ-MORENO
4. Shiga, T., and Shiga, K. (1973) J. Rio&em.
74, 1033109.
5. Macheroux, P., Schmidt, K. U., Steinerstauch, P., Ghisla, S., Colepicolo, R., Buntic, R., and Hasting, J. W. (1987) Biochem. Biophys. Res. Commun. 146, 101-106. 6. Massey, V., and Ganther, H. (1965) Biochemistry
4, 1161-1176.
7. Zanetti, G., Cidaria, D., and Curti, B. (1982) Eur. J. &o&em. 4533458.
126,
8. Bookjans, G., San Pietro, A., and Biiger, P. (1978) Biochem. Btiphys. Res. Commun. 80, 759-765. C., and Edmondson, 9. Sancho, J., Peleato, M. L., Gbmez-Moreno, D. E. (1988) Arch. Biochem. Biophys. 260, 200-207. 10. Nelson, N., and Neumann, J. (1969) J. Biol. Chem. 244,1932-1936. 11. Foust, G. P., Mayhew, S. G., and Massey, V. (1969) J. Biol. Chem. 244,964-970. 12. Bookjans,
G., and Bager, P. (1979) Arch. Biochem. Biophys.
194,
387-393. 13. Batie, C. J., and Kamin, 11,223.
H. (1986) J. Biol. Chem. 261,
14. Susor, W. A., and Krogmann,
11,214-
D. W. (1966) Biochim. Biophys. Acta
120,65-72. 15. Buchanan, B., and Arnon, D. I. (1971) in Methods in Enzymology (San Pietro, A., Ed.), Vol. 23, pp. 413-441, Academic Press, San Diego. 16. Edmondson, D. E., and Tollin, G. (1980) in Methods in Enzymology (San Pietro, A., Ed.), Vol. 69, pp. 392-405, Academic Press, San Diego. 17. Duggleby, R. G. (1981) Anal. Biochem. 110, g-18. 18. Batie, C. J., and Kamin, H. (1984) J. Biol. Chem. 259.8832-8834. 19. Pai, E. F., Karplus, 27.4465-4474.
P. A., and Schulz, G. E. (1988) Biochemistry
20. Mayhew, S. G., and Ludwing, (1975) in The Enzymes (Bayer, P. D., Ed.), 3rd ed., Vol. 12, pp. 57-118, Academic Press, New York. 21. Fillat, M. F., Sandman, G., and Gomez-Moreno, C. (1988) Arch. Microbial. 150, 160-164. 22. vanlin, B., and Bothe, H. (1972) Arch. Microbial. 82, 155-172. C. (1990) Eur. J. 23. Sancho, J., Medina, M., and Gbmez-Moreno, Biochem. 187,39-48. 24. Janin, J., and Chothia, C. (1978) Biochemistry 17,1943-1947. 25. Pueyo, J. J., Sancho, J., Edmondson, D. E., and Gbmez-Moreno, C. (1989) Eur. J. Biochem. 183, 539-544. 26. Vieira, B. J., and Davis, D. J. (1986) Arch. Biochem. Biophys. 247, 140-146. 27. Chan, T. M., Ulrich, P., and Markley, J. L. (1983) Biochemistry 22, 6002-6007. 28. Zanetti, G., Aliverti, A., and Curti, B. (1984) J. Biol. Chem. 259, 6153-6157. 29. Masaki, R., Yosmikawa, S., and Matsubara, H. (1982) Biochim. Biophys. Acta 700, 101-109. 30. Batie, C. J., and Kamin, H. (1984) J. Biol. Chem. 259, 11,97611,985. and 31. Bagshaw, C. R., and Harris, D. A. (1987) Spectrophotometry Spectrofluorimetry (Harris, D. A., and Bashford, C. L., Eds.), pp. 91-113, IRL Press, Oxford. 32. Hall, L. H., Bowers, M. L., and Durfor, C. N. (1987) Biochemistry 26, 7401-7409. 33. Abe, M., and Kyogoku, Y. (1987) Spectrochim. Acta Part A 43, 1027-1037. 34. Bienstock, R. J., Schopfer, L. M., and Morris, M. D. (1987) J. Raman Spectroscopy 18, 241-245. 35. Ricard, J., Nari, J., and Diamantidis, G. (1980) Eur. J. Biochem.
108,55%66.
OF BIOCHEMISTRY
AND
BIOPHYSICS
Vol. 288, No. 1, July, pp. 231-238, 1991
Interaction of Ferredoxin-NADP+ Reductase from Anabaena with Its Substrates Javier
Sancho’
and Carlos
G6mez-Moreno’
Departamento de Bioquimica y Biologia Molecular y Cellular, Facultad de Ciencias. Urhersidad de Zaragoza 50009 Zaragoza, Spain
Received December 7, 1990, and in revised form March 7, 1991
The interaction of ferredoxin-NADP+ reductase from Anabaena variabilis with its subthe cyanobacterium strates, NADP+ and ferredoxin, has been studied by difference absorption spectroscopy. Several structural analogs of NADP+ have been shown to form complexes the stabilities of which are strongly dependent on the ionic strength of the medium. In most cases the binding energy of these complexes and their difference absorption spectra are similar to those reported for the spinach enzyme. However, NADP+ perturbs the absorption spectra of the Anabaena and spinach enzymes in a different way. This difference has been shown to be related to the binding of the nicotinamide ring of NADP+ to the enzymes. These results are interpreted as being due to a different nicotinamide binding site in the two reductases. The enthalpic and entropic components of the Gibbs energy of formation of the NADP+ complex have been estimated. An increase in entropy on NADP+ binding seems to be the main source of stability for the complex. A shift of approximately 40 mV in the redox potential of the couple NADP+/ NADPH has been observed to occur upon binding of NADP’ to the oxidized enzyme. This allows us to calculate the binding energy between the reductase and NADPH. The ability of the reductase, ferredoxin, and NADP+ to form a ternary complex indicates that the protein carrier binds to the reductase through a different site than that of the pyridine nucleotide. o Issl Academic Press,
Inc.
The absorption spectra of flavin coenzymes, FMN and FAD, are perturbed when they become incorporated into proteins. The most prominent change is usually indicated by the appearance of shoulders associated with the main absorption band at approximately 445 nm. These shoul’ Present address: Cambridge Center for Protein Engineering. Medical Res. Center, Hills Rd, Cambridge, CB2 2&H U.K. ’ To whom correspondence should be addressed. FAX: 34-76-567920. 0003.9861/91 $3.00 Copyright 0 1991 by Academic Press, All rights of reproduction in any form
ders have been related to different vibrational levels of the electronic transition (1,2) and have been found in at least 40 flavoenzymes. Owing to the fact that similar changes occur when the flavin coenzyme is dissolved in organic solvents, it has been proposed that these changes are caused by interaction of the organic chromophore with apolar residues in the polypeptide chain (3, 4). It is generally accepted (5,6) that the enhancement of the shoulders in the 445-nm band is due to a decrease in the polarity of the environment of the prosthetic group, though it could simply reflect a more rigid environment of the flavin. Ferredoxin-NADP+ reductase (FNR)” is the flavoenzyme involved in the transfer of electrons from ferredoxin to NADP+, thus providing the reducing factor necessary for most biosynthetic reactions in plants. The enzymes, isolated from either higher plants (7), algae (8), or cyanobacteria (9), all carry a single FAD group that participates in the electron transfer reaction. The absorption spectrum of FNR is similar to that of many other flavoproteins. Early studies on the spinach enzyme had revealed that spectral perturbations appear as a consequence of the formation of a complex with several electron transfer proteins such as ferredoxin, flavodoxin, or rubredoxin (10, 11). Binding of NADP(H) to ferredoxin-NADP+ reductase has also been reported to produce significant changes in its absorption spectrum in the visible region (12). Difference absorption spectroscopy can be used to measure the binding constant for these complexes since they have a well-defined stoichiometry (11-13). Comparison of the interaction between ferredoxin-NADP’ reductase with ferredoxin and with tlavodoxin is of interest since both electron transport proteins are natural substrates for the enzyme in cyanobacteria. Data are presented that show that cyanobacterial FNR forms complexes of similar binding strength with ferredoxin and flavodoxin from dif-
3 Abbreviation
used: FNR, ferredoxin-NADP+
reductase. 231
Inc. reserved.
232
SANCHO
AND
ferent sources. Measurements of the affinity of different NADP+ analogs to ferredoxin-NADP+ reductase from Anabaena provide evidence that different parts of the NADP+ molecule are involved in the binding to the reductase. From the temperature dependence of the dissociation constant of the NADP+-FNR complex, an estimation of the entropic and enthalpic components of the Gibbs energy of binding has been made. MATERIAL
AND
GOMEZ-MORENO
a 0.02
METHODS
Proteins and reagents. Anabaena uariabilis ferredoxin-NADP+ reductase was purified as described earlier (9). Ferredoxin from A. uariubilis and spinach were purified essentially as in (14,15). Azotobacter uinehdii flavodoxin, purified according to (16), was a gift from Dr. D. E. Edmondson (Emory University). Apoflavodoxin was prepared by a modification of the method described in (2). Nucleotides were from Sigma or Boehringer. An extinction coefficient of 15.4 mM-’ cm-’ at 260 nm has been used to quantify 2’-AMP, 2’,5’-ADP, and 2’5.ATP ribose. All other reagents used in this study were of analytical grade. Complex formation. All experiments were performed at 25°C in a thermostated Kontron Uvikon 860 spectrophotometer using two-compartment cuvettes. Samples of 1 ml containing 40 pM Anubaena reductase were placed in the front compartment of the sample and of the reference cuvettes. The same volume of buffer was added to the rear compartments of both cells. A reference spectrum was then run and electronically subtracted from any further spectra. Equal volumes (usually 1 to 5 ~1) of electron-carrier protein, NADP+, or NADP’ analogs were added to the front compartment of the sample cuvette and to the rear compartment of the reference cuvette, using spectrophotometrically calibrated Hamilton microsyringes. An equivalent volume of buffer was added each time to the front compartment of the reference cuvette to compensate for the dilution of the enzyme in the sample cuvette. The difference spectrum was then recorded. The dissociation constants and extinction coefficients of the different complexes were calculated by nonlinear fitting of the data assuming a 1:l stoichiometry. The program described in (17) was used with a modification which takes into account all dilution effects during titration. A 1:2 stoichiometry was not consistent with the experimental data. The total amount of ligand added in each experiment was never higher than 200 ~1. Most of the Kd values here reported were calculated by fitting more than 20 data points to the theoretical equation (18). The buffer used was 50 mM, pH 8.0, Tris/HCl unless otherwise indicated. Redor potential. The redox potential of Anabaena ferredoxinNADP+ reductase was determined by chemical titration with dithionite as described earlier (9) but in the absence of glycerol. The midpoint potential of the reductase in the presence of NADP+ was determined in a mixture of NADP+, glucose 6-phosphate, and glucose-6-phosphate dehydrogenase as an NADPH generating system. The NADP+/NADPH couple was used as the redox indicator. The chemical titration was carried out at 25OC in 50 mM sodium phosphate buffer, pH 7.0, and the titration with NADP+/NADPH in 5 mM buffer at the same temperature. The spectral changes produced during redox titration were determined with a Hewlett-Packard diode array spectrophotometer.
RESULTS
Complex Formation between Ferredoxin-NADP’ Reductase and NADP+ or its Analogs Ferredoxin-NADP+ reductase from Anabaena forms a complex with NADP’ that can be detected and quantified from the spectral changes that are observed in the ultraviolet and visible absorption spectra of the mixture (Fig.
3
345
430
WAVELENGTH
0
50
100
515
c$010
(nm)
150
200
WDP + 1 WW FIG. 1. Anabatma ferredoxin-NADP’ reductase-NADP+ complex. (a) Difference spectra on addition of increasing amounts of NADP’. The experiment was performed at 25’C with the enzyme dissolved in 50 mM Tris/HCl buffer, pH 8.0. Aliquots of a concentrated NADP+ solution (2.4 mM) were added stepwise (1 to 5 ~1 additions) as described under Materials and Methods. (b) Best fit of the change in absorbance (460 nm minus 400 nm) to a 1:l stoichiometry.
la). Plotting the spectral changes vs nucleotide concentration (Fig. lb) yields a saturation curve. The binding constant of the complex has been calculated to be 13.2 PM assuming a 1:l stoichiometry. In order to determine which parts of the nucleotide molecule are essential for binding to the active site in the enzyme several NADP+ analogs have been studied with respect to their ability to bind ferredoxin-NADP+ reductase. The Kd values for these complexes are compared in Table I with dissociation constants measured for the complexes of the same analogs with other NADP(H)-dependent enzymes. NAD’, when added up to a 2 mM con-
233
FERREDOXIN-NADP+ REDUCTASE FROM Anabaena TABLE I Dissociation Constants, Binding Energies, and Extinction Coefficients of the Complexes between NADP+ (or other analogs) and Ferredoxin-NADP+ Reductase or Glutathion Reductase AG; (kcal/mol)
K, (PM)
c (mM-’
cm-‘)
Nucleotide
A. var.
Spinach
GR
A. uar.
Spinach
GR
A. var.
Spinach
GR
NADP’ 2’,5’-ATP ribose
13.25 10.46 15.38 73.49 1500
14 8 2.4 120 1530
70 75 30 500 2900
-6.63 -6.77 -6.54 -5.62 -3.84
-6.60 -6.92 -7.64 -5.33 -3.82
-5.64 -5.60 -6.14 -4.46 -3.45
1.19 1.49 1.14 0.37
0.80 0.71 0.70 0.22
-
2’,5’-ADP 2’-AMP NAD(H)
Note. Data for spinach FNR come from (13) and for glutathion reductase (GR) from (19). The extinction by Anabaena FNR have been calculated from the difference in absorbance at 460 with respect to 400 nm.
centration, was able neither to produce any spectral perturbation nor to displace NADP+ bound to the enzyme (data not shown). This indicates that the 2’-phosphate group-absent in NAD+-is essential for the binding of NADP+ to the FNR from Anabaena. The same requirement has been found for other pyridine nucleotide-dependent enzymes such as glutathione reductase, where the 2’-phosphate group causes a 40-fold increase in the binding constant of the complex (19) with the enzyme. Of all the other nucleotides tested, only 2’-AMP forms a weaker complex than NADP+, which suggests that the negative charge of the NADP+ pyrophosphate bridge, absent in 2’-AMP, stabilizes the complex between the reductase and NADP+. This is confirmed by the fact that 2’,5’-ADP-which has only one of the two phosphate moieties of the bridge-binds the enzyme almost as strongly as NADP+. We have observed that the stability of the complexes between the different nucleotides and the reductase is strongly dependent on the ionic strength of the medium. High ionic strengths, such as 0.5 M NaCl, completely abolish the formation of the complexes, indicating an important contribution of electrostatic interactions to their stability. In a parallel set of experiments, the temperature dependence of the binding constant of the NADP+-FNR complex was studied in order to obtain information on the forces involved in its stabilization. Our data indicate that the binding constant for NADP+ does not change significantly in the range from 7 to 25°C (& from 13.2 to 13.85 PM). This corresponds to a A(& value of approximately -6.6 kcal/mol. Calculation of AH; for the binding of the complex using the Arrhenius plot has proved to be difficult due to the very shallow slope of this plot. From our data we can estimate a m0 value of approximately 0.2 to -1.2 kcal/mol. This low value of the enthalpy of binding implies that the complex is stabilized mainly by entropic forces.
coefficients
of the complexes formed
Complex Formation between Ferredoxin-NADP’ Reductase and Ferredorin or Flavodoxin Ferredoxin-NADP’ reductase from Anabaena forms stable complexes with ferredoxin, from both spinach and Anabaena. The Kd values determined from the spectral changes produced upon mixing both proteins are in the micromolar range, as reported for the spinach reductase (11). It has been proposed that flavodoxin can replace ferredoxin in many of the physiological reactions in which this iron-sulfur protein participates, including those involving ferredoxin-NADP+ reductase (20). Anabaena flavodoxin has been shown to interact with ferredoxinNADP+ reductase from the same organism in what can be considered the preliminary step in the electron transfer reaction between these proteins (21). Since Azotobacter flavodoxin shows a lower activity in the reactions assayed with the reductase, i.e., the cytochrome c reductase and NADP+ photoreductase activities (22), it was of interest to determine whether the two flavoproteins interact to form a complex at all. Data in Table II indicates that the strength of the complex formed by both proteins is, in
TABLE II Dissociation Constants, Binding Energies, and Extinction Coefficients of the Complexes between Different Electron Carrier Proteins and Ferredoxin-NADP+ Reductase from Anabaena Electron
carrier
Anabaena ferredoxin Spinach ferredoxin Azotobacter flavodoxin Apoflavodoxin
Kc, (PM)
AG; (kcal/mol)
3.57 1.25 7.3 73.8
-7.4 -8.0 -7.0 -5.6
6 (mM-’
cm-‘)
1.49 1.57 0.6 0.7
Note. Extinction coefficients have been determined at 460 nm for the ferredoxin complexes and at 400 nm for the complexes with flavodoxin or apoflavodoxin.
234
SANCHO
AND
GGMEZ-MORENO
state exhibits NADP+.
different
binding
sites for ferredoxin
Redox Behavior of Ferredoxin-NADP’ Presence and Absence of NADP’
-‘.O 1 300
400 WAVELENGTH
500
600
(nm)
FIG. 2. Difference spectraof the ternary complex (ferredoxin-FNRNADP’). A 15 WM solution of Anabaena reductase was saturated with Anabaena ferredoxin (53 pM) and a baseline spectrum was recorded. Aliquots of a 2.4 mM NADP+ solution were then added stepwise (5 ~1 each time) to develop the spectral perturbation shown in the figure. Temperature and buffer were as in Fig. 1.
fact, very similar to that of ferredoxin. Furthermore, the ability of flavodoxin from Azotobacter to bind the reductase is not lost when the FMN group is removed from the carrier. The FNR-apo-flavodoxin complex is, however, weaker by a factor of 10 than that in which the holoprotein is involved. None of these reductase-electron-carrier complexes were stable in 0.5 M NaCl, indicating an important contribution of electrostatic interactions to their stability.
and
Reductase in the
In previous work (3) we have reported a -320 mV value for the EL of Anabaena reductase at pH 7.0, as determined by spectrocoulombimetric measurement in the presence of glycerol. We have now obtained the same value in its absence. Batie and Kamin (13) have reported, for the spinach enzyme, that the redox potential of NADP+ became 40 mV more positive when the pyridine nucleotide was bound to ferredoxin-NADP+ reductase. This is an important finding since it allows calculation of the relative affinity of the enzyme for the oxidized and reduced nucleotide substrate. We have performed a similar experiment aimed at determining the stability of both complexes in the cyanobacterial enzyme. To an anaerobic mixture containing 30 PM Anabaena ferredoxin-NADP+ reductase, 370 PM NADP+, and excess glucose 6-phosphate, an anaerobic solution containing catalytic concentrations of glucose-6-phosphate dehydrogenase was added. The spectral changes that occurred upon mixing were followed with a rapid scan spectrophotometer. The NADPH concentration was calculated from the absorbance at 340 nm, which is an isosbestic point for the reduction of the Anabaena reductase. Since the initial concentration of NADP+ was known, the redox potential in the solution could be calculated throughout the reaction. A plot of the absorbances at 459 and 600 nm vs the redox potential in the solution is shown in (Fig. 3). A drop in the absorbance at 459 nm closely parallels an increase in that at 600 nm, which is indicative of the formation of a stable chargetransfer complex between the oxidized reductase and NADPH, as has been described for the spinach enzyme (13). This behavior differs from what was observed when
The Ternary Complex Ferredoxin-FNR-NADP’ Three different molecules participate in the reaction catalyzed by FNR. It is of mechanistic interest to determine whether a ternary complex of the enzyme, the nucleotide, and the electron carrier protein can be formed. Figure 2 shows the differential spectrum obtained by addition of NADP+ to a mixture of the reductase and ferredoxin from Anabaena. Apart from the presence of ferredoxin, the titration was performed under the same conditions as in its absence. The calculated value of the Kd for the Anabaena FNR-NADP+ complex in the presence of a saturating concentration of ferredoxin has been found to be 15.3 FM. This value is very similar to that found in the absence of ferredoxin (see Table I). This indicates that the enzyme from Anabaena in its oxidized
REDOX
POTENTIAL
(mV)
FIG. 3. Redox titration of Anabaena ferredoxin-NADP+ reductase in the presence of NADP+. To an anaerobic solution of Anabaena FNR containing 370 pM NADP+ and excess of glucose 6-phosphate catalytic amounts of glucose-r-phosphate dehydrogenase were added. The absorbance spectrum of the solution was then measured at different times with a rapid scan spectrophotometer. Open symbols: absorbance at 459 nm; closed symbols: 3X absorbance at 600 nm.
FERREDOXIN-NADP+
REDUCTASE
ferredoxin-NADP+ reductase was titrated with dithionite in the absence of NADP+ [see (9)]. There, the absorbance at 600 nm rose to a maximum which was only approximately 7% of that at 459 nm and remained at that level until the end of the titration, indicating the formation of the semiquinone free radical that is quite unstable in this enzyme. No charge-transfer complex was formed under those conditions. The midpoint potential for this initial reductive process was calculated by plotting log (ox)/(red) vs the potential assuming that the redox process involves two electrons and also that the millimolar extinction coefficient for the FNR-NADP’ complex is 10.04 mM ’ cm-l and that for the reduced protein is 0.6 rnM-~’ cm-‘. A value of -275 mV is calculated, which is approximately 40-50 mV more positive than the potential described for free NADP+/NADPH in solution at pH 7.0 (-320 mV). It appears that once the charge-transfer complex is formed, and as more negative potentials are generated by the continuous production of NADPH, a second reduction process which corresponds to the main decrease in the absorbance at 459 nm takes place. At the same time only a partial decrease in the absorbance at 600 nm can be observed. This can be interpreted as a consequence of the reduction of FAD in the oxidized reductase-NADPH charge-transfer complex. From the data presented here it is difficult to calculate an accurate value of potential for this process since the reduction does not proceed to completion; however, it appears to approach the previously determined value of -320 mV (9). The residual absorbance observed at 600 nm could be due to the semiquinone radical that remains at this point, since full reduction has not yet occurred.
FROM
235
Anabaena
Anabaena
DISCUSSION The Complex between NADPc and FNR from Anabaena: Comparison with the Complex Formed by FNR from Spinach
The binding of a relatively large molecule, like NADP+, to an enzyme is the result of several simultaneous interactions. In order to assess which groups, within NADP+, favorably contribute to its binding to FNR, the dissociation constants of the complexes formed by the reductase and several NADP+ analogs have been determined from the spectral changes produced during complex formation. Comparison of the binding energies of different pyridine nucleotide analogs to ferredoxin-NADP+ reductase can provide important information on the nature of the forces involved in such binding. Figure 4 shows the energies of interaction in these complexes. The low affinity of the reductase for NADH suggests that the phosphate group in the 2’ position of NADP+ is effectively involved in binding to the enzyme. This has also been described for the spinach enzyme and the NADPH-dependent flavoenzyme glutathione reductase. X-ray data of the later en-
2’5’ADP
FIG. 4. Binding energy diagram of the complexes between NADP+ (and analogs) and several flavo-oxidoreductases. The binding energies of the Anabaena reductase complexes (solid line) have been determined from their dissociation constants (see Materials and Methods). Data for spinach reductase (light line) and for glutathione reductase (high ionic strength; dotted line) come from Refs. (13,351 and (191, respectively.
zyme indicate that the formation of a complex with NADP+ takes place by an induced-fit mechanism (19). In this model the interaction of the 2’-phosphate group of NADP+ with the enzyme contributes to a change in position of several amino acids in the binding site, thus facilitating the right conformation for effective binding. The importance of the 2’-phosphate group is supported by the finding that binding of 2’-AMP to Anabaena ferredoxinNADP+ reductase is much stronger than that of NADH, although 2’-AMP lacks the 5’-phosphate group present in NADH that is known to be important for binding (23). The increase in stability (1 kcal/mol) found for the 2’,5’-ADP complex with respect to 2’-AMP would indicate the existence of a strong interaction between the phosphate group in the 5’ position and a positively charged group in the reductase. We have obtained direct evidence for the presence of an arginine residue, in the reductase from Anabaena (23), which interacts with the pyrophosphate bridge in NADP+. This has also been proposed for the Bumilleriopsis enzyme (12). The similarity of the binding energy values of the NADP’ and 2’,5’-ADP complexes suggests that the nicotinamide nucleotide moiety within NADP+ does not contribute to the stability of the enzyme-NADP+ complex. When the stabilities of the complexes formed by NADP+ and analogs with the reductases from either Anabaena or spinach are compared, similar values are found with only one exception. The 2’,5’-ADP complex with the spinach enzyme is approximately 1 kcal/mol more stable than that with the cyanobacterial enzyme or when compared to the complexes formed with either 2’,5’-ATP ri-
236
SANCHO
AND
bose or NADP+ (See Fig. 4). This indicates that the nicotine nucleotide within NADP+ (absent in 2’,5’-ADP) introduces a destabilizing effect in the spinach enzyme complex that is absent in the complex with the reductase from Anabaena. A similar destabilizing effect has also been observed for glutathion reductase (19). Though this last result should be regarded with caution, since it was obtained at high ionic strength, it appears that in none of these cases the nicotine nucleotide within NADP+ contributes to stabilize the complex between NADP+ and these reductases. Another difference with respect to the interaction of the FNR from several sources with the nicotinamide ring in NADP+ has been found by comparing the difference spectra obtained for the NADP+ and 2’,5’-ATP ribose complexes with the reductases from Anabaena, Bumilleriopsis, and spinach. Figure 5 shows that NADP+ produces a very similar perturbation in the visible spectrum of the enzymes from both Anabaena and Bumilleriopsis. This perturbation is almost identical to that produced by 2’,5’ATP ribose (a molecule similar to NADP+ but lacking the nicotinamide ring) in both Anabaena and spinach FNR. However, in the spinach enzyme, the effect of NADP+ is completely different. This shows that the nicotinamide ring is not able to alter the absorption spectrum of the prosthetic group in the reductases from Anabaena and Bumilleriopsis, suggesting a weak interaction of the nicotinamide with the NADP+ binding site of these enzymes in their oxidized state. On the contrary, the spectrum of spinach FNR is clearly modified by the interaction with the nicotinamide. This points either to a direct interaction between the nicotinamide and the FAD group of the spinach reductase or to a nicotinamide-induced reorganization of the NADP+ binding site that changes the absorption properties of the FAD group. The stability of the FNR-NADP+ complex is rather temperature independent. We estimate that at least 5-6 kcal/mol of the binding energy is due to entropic forces. Studies by Janin and Chothia have suggested an essential role for hydrophobic interactions in the binding of coenzymes to several enzymes (24). They have pointed out that the loss in surface area that often occurs upon binding is enough to account for the stability of these complexes. Nevertheless, electrostatic interactions are also very important in the stability of the NADP+-FNR complex since it is destroyed by high ionic strength media. We therefore propose that the entropic stabilization of the FNR-NADP+ complex is afforded both by the loss of surface area and by electrostatic interactions that release water molecules initially associated with the interacting charges. In spite of the differences observed for the interaction between NADP+ and ferredoxin-NADP’ reductase from both Anabaena and spinach, the shift in the redox midpoint potential of the nucleotide that was reported to occur upon binding to the spinach enzyme also appears to occur in the cyanobacterial enzyme. By comparing the Gibbs
GOMEZ-MORENO 2’5’
NADP+
- ATP ribose
I
WAVELENGTH
(nm)
FIG. 5. Comparison of difference spectra of the complexes between NADP+ or 2’,5’-ATP ribose and ferredoxin-NADP+ reductase from different species. Spectra of the spinach and BumiUeriopsis complexes come from Refs. (13, 35) and (12), respectively.
energies obtained from the redox potential of the couples NADP+/NADPH (EL = -320 mV) and FNR-NADP+/ FNR-NADPH (Eb = -275 mV) with the dissociation constant of the FNR-NADP+ complex (& = 13 PM) we calculate a value of 0.39 /.LM for the dissociation constant of the complex between FNR and NADPH. This indicates that the complex between the enzyme and the oxidized nucleotide is less stable than the complex with the reduced nucleotide. The Complex between FNR from Anabaena and Electron Carrier Proteins All the perturbations induced on the spectrum of ferredoxin-NADP+ reductase by the different electron carrier proteins that we have tested differ from those induced by NADP+ and its analogs, in that they have only positive extinction coefficients in the visible region. The similarity of the difference spectra of the complexes between FNR from Anabaena and ferredoxin from Anabaena and spinach (present work), and those of FNR and ferredoxin from spinach (ll), lettuce (lo), and Bumilleriopsis (12), suggests that the ferredoxin binding site is similar in the reductase from all these origins. We have also determined the stability of the complex between the reductase from Anabaena and Azotobacter flavodoxin and found that it is
FERREDOXIN-NADP’
REDUCTASE
similar to that of the complexes with ferredoxin. Although we cannot conclude that the perturbation observed in the visible region is due only to a perturbation of the FAD chromophore in the reductase, it seems clear that these carriers do perturb the spectrum of the enzyme since apoflavodoxin from Azotobacter, which does not absorb in the visible, is still able to alter the absorption spectrum of Anabaena FNR (not shown). The dissociation constants measured for all the complexes between the Anabaena reductase and ferredoxin or flavodoxin from several origins [present work and (al)] are similar. Cross-linking experiments performed in our laboratory (25) have demonstrated that Anabaena ferredoxin and Azotobacter Aavodoxin do indeed share a common binding site. All these complexes are unstable at high ionic strength, which indicates that they are stabilized by electrostatic interactions. Carboxyl groups in the carriers (26, 27) and lysil (28) and arginine (23) groups in the reductase are likely to be responsible for the electrostatic stabilization of these complexes. The Ternary Complex: Ferredoxin-FNR-NADP+ It has already been suggested (29, 30) from difference spectroscopy studies that the transfer of electrons from reduced ferredoxin to NADP+ takes place with the formation of a ternary complex between both substrates and the enzyme from spinach. The observation we have made, that ferredoxin-NADP+ reductase binds NADP+ even in the presence of ferredoxin, is indicative of the formation of a ternary complex that could be involved in the reaction catalyzed by FNR. This proposal is also supported by other experiments in which it was shown that a covalent cross-linked complex between the reductase and the ferredoxin or the Aavodoxin could still bind to, and also be reduced by, NADPH (25). Measurements of the presteady state kinetics of this reaction in the non-cross-linked complex are currently under way in order to determine if FNR can accept electrons from ferredoxin when bound to NADP+ as a first step in the mechanism of the reaction of this enzyme. The Use of Difference Spectroscopy to Obtain Structural Information Difference absorption spectroscopy has been widely used to detect intermolecular interactions as well as to quantify their strength (31) but it has other potential uses we would like to point out. It can give detailed information concerning the changes that occur in the environment of the chromophoric groups within a protein as a consequence of binding of a ligand. Information of this kind might be attainable if the changes observed in the difference spectra could be related to environmental properties of the chromophore within the protein. In the case of flavoenzymes, circular dichroism studies (2) have shown that the absorption spectrum in the visible region has
FROM
237
Anabaena
350
400
450
WAVELENGTH
500
550
(nm)
FIG. 6. Distribution of peaks from difference spectraalong the absorbance axes. The absorbance axis has been divided into 5-nm windows. The number of peaks that appear in each of the windows in all of the difference spectra reported in this work is represented in this figure.
contributions from different vibronic transitions associated with the two main absorption bands. Indeed, Miiller et al. (1) have correlated temperature-dependent changes in the absorption spectrum of ferredoxin-NADP+ from spinach with five vibronic transitions previously described (3). In trying to relate the peaks that appear in our difference spectra to these transitions we have defined 5-nm windows in the wavelength axis and have counted the number of peaks (both positive and negative) that appeared in each of these windows in all of our difference spectra. Thirty-seven peaks have been considered. Figure 6 shows our results. It emerges that the peaks cluster in five regions with maxima at 397, 432, 457, 480, and 507 nm. These maxima relate well with the wavelengths of the vibronic transitions described previously (375, 435, 455, 489, and 510 nm) (3). Though the nature of these transitions is still not well characterized, progress in our understanding is expected to result from molecular orbital studies (31) and Raman spectroscopy (33,34). Once these transitions are better characterized, difference absorption spectroscopy could give rapid qualitative information on the effect of the binding of ligands and on the environment of the prosthetic group. ACKNOWLEDGMENTS This work was supported in part by Grant BAP 0397.E from the Commission of the European Communities and by Grant BIO 88/0413 from the Comision Interministerial de Ciencia y Tecnologia (Spain). We thank Dr. D. E. Edmondson from Emory University for providing us with A. uinelandii Aavodoxin and for interesting discussions and Drs R. Loewenthal from Cambridge University and H. J. Loewenthal from the Israeli Institute of Technology, for help with the manuscript.
REFERENCES 1. Miiller, F., Mayhew, S. G., and Massey, V. (1973) Biochemistry 12, 4654-4662. 2. Edmondson, D. E., and Tollin, G. (1971) Biochemistry 10, 113124. 3. Gordon-Walker, A., Penzer, G. R., and Radda, G. K. (1970) Eur. J. Biochem. 13, 313-321.
238
SANCHO AND GOMEZ-MORENO
4. Shiga, T., and Shiga, K. (1973) J. Rio&em.
74, 1033109.
5. Macheroux, P., Schmidt, K. U., Steinerstauch, P., Ghisla, S., Colepicolo, R., Buntic, R., and Hasting, J. W. (1987) Biochem. Biophys. Res. Commun. 146, 101-106. 6. Massey, V., and Ganther, H. (1965) Biochemistry
4, 1161-1176.
7. Zanetti, G., Cidaria, D., and Curti, B. (1982) Eur. J. &o&em. 4533458.
126,
8. Bookjans, G., San Pietro, A., and Biiger, P. (1978) Biochem. Btiphys. Res. Commun. 80, 759-765. C., and Edmondson, 9. Sancho, J., Peleato, M. L., Gbmez-Moreno, D. E. (1988) Arch. Biochem. Biophys. 260, 200-207. 10. Nelson, N., and Neumann, J. (1969) J. Biol. Chem. 244,1932-1936. 11. Foust, G. P., Mayhew, S. G., and Massey, V. (1969) J. Biol. Chem. 244,964-970. 12. Bookjans,
G., and Bager, P. (1979) Arch. Biochem. Biophys.
194,
387-393. 13. Batie, C. J., and Kamin, 11,223.
H. (1986) J. Biol. Chem. 261,
14. Susor, W. A., and Krogmann,
11,214-
D. W. (1966) Biochim. Biophys. Acta
120,65-72. 15. Buchanan, B., and Arnon, D. I. (1971) in Methods in Enzymology (San Pietro, A., Ed.), Vol. 23, pp. 413-441, Academic Press, San Diego. 16. Edmondson, D. E., and Tollin, G. (1980) in Methods in Enzymology (San Pietro, A., Ed.), Vol. 69, pp. 392-405, Academic Press, San Diego. 17. Duggleby, R. G. (1981) Anal. Biochem. 110, g-18. 18. Batie, C. J., and Kamin, H. (1984) J. Biol. Chem. 259.8832-8834. 19. Pai, E. F., Karplus, 27.4465-4474.
P. A., and Schulz, G. E. (1988) Biochemistry
20. Mayhew, S. G., and Ludwing, (1975) in The Enzymes (Bayer, P. D., Ed.), 3rd ed., Vol. 12, pp. 57-118, Academic Press, New York. 21. Fillat, M. F., Sandman, G., and Gomez-Moreno, C. (1988) Arch. Microbial. 150, 160-164. 22. vanlin, B., and Bothe, H. (1972) Arch. Microbial. 82, 155-172. C. (1990) Eur. J. 23. Sancho, J., Medina, M., and Gbmez-Moreno, Biochem. 187,39-48. 24. Janin, J., and Chothia, C. (1978) Biochemistry 17,1943-1947. 25. Pueyo, J. J., Sancho, J., Edmondson, D. E., and Gbmez-Moreno, C. (1989) Eur. J. Biochem. 183, 539-544. 26. Vieira, B. J., and Davis, D. J. (1986) Arch. Biochem. Biophys. 247, 140-146. 27. Chan, T. M., Ulrich, P., and Markley, J. L. (1983) Biochemistry 22, 6002-6007. 28. Zanetti, G., Aliverti, A., and Curti, B. (1984) J. Biol. Chem. 259, 6153-6157. 29. Masaki, R., Yosmikawa, S., and Matsubara, H. (1982) Biochim. Biophys. Acta 700, 101-109. 30. Batie, C. J., and Kamin, H. (1984) J. Biol. Chem. 259, 11,97611,985. and 31. Bagshaw, C. R., and Harris, D. A. (1987) Spectrophotometry Spectrofluorimetry (Harris, D. A., and Bashford, C. L., Eds.), pp. 91-113, IRL Press, Oxford. 32. Hall, L. H., Bowers, M. L., and Durfor, C. N. (1987) Biochemistry 26, 7401-7409. 33. Abe, M., and Kyogoku, Y. (1987) Spectrochim. Acta Part A 43, 1027-1037. 34. Bienstock, R. J., Schopfer, L. M., and Morris, M. D. (1987) J. Raman Spectroscopy 18, 241-245. 35. Ricard, J., Nari, J., and Diamantidis, G. (1980) Eur. J. Biochem.
108,55%66.
