Magnetic Susceptibility Studies of Native and Thionine-oxidized Molybdenum-Iron Protein from Azotobacter uinelandii Nitrogenase*

The difference between the magnetic susceptibilities of native and thionine-oxidized molybdenmn-iron protein from Azotobacter vinelandii nitrogenase was meas- ured by the nuclear magnetic resonance method. Reversible oxidation of the MoFe protein by 4 to 8 electron eq of thionine/mol made the protein more paramagnetic than it was in the native state. The NMR suscep- tibility results were analyzed in terms of a model for the spin states of the iron centers in the MoFe protein based on low temperature electron paramagnetic resonance and Mossbauer spectral studies. The model pro-poses that the native protein contains 2 “M” centers (S = ”/2) and 4 rrP” centers (S = O)/mol and that the oxidized protein has diamagnetic M centers and paramagnetic P centers with S z 3/2. Assuming that this model holds at 280 K, the NMR susceptibility results show that the effective magnetic moment of the oxidized P centers is larger than that of the native M centers. Based on an analysis in terms of spin only magnetic moments, the susceptibility results suggest that the P centers in the oxidized protein are S = 6/2 systems. comparison of their low temperature EPR spectra assay of the acetylene reduction activity in the presence of excess Fe protein The specific activities of the MoFe protein preparations used in this work ranged from 1,540 to 2,000 nmol of ethylene produced/min/mg of MoFe protein. Protein concentrations were determined by the biuret reaction using bovine serum albumin as a standard (15). Analysis of a typical MoFe protein preparation for molybdenum (16) and iron (17) showed the presence of 1.6 mol of Mo and 23 mol of Fe/mol of MoFe protein of 243,000-dalton molecular mass. Preparation of Samples for Susceptibility Studies-For susceptibility the inner and outer chambers of a double septum-sealed NMR cell were and fiied with oxygen-free at least six times. Reagent solutions were prepared in double septum-sealed tonometers. As an added precaution against the influx of oxygen, a 0.1 M solution of sodium dithionite with methyl viologen as an indicator of reduction potential placed in the outer chambers of the double septum-sealed containers. A long-needled gas-tight syringe used to add protein solutions and other reagents to the anaerobic NMR cells. a typical an anaerobic NMR cell be

7 Present address, Department of Chemistry, Massachusetts Institute of Technology, Cambridge, MA 02139. molecular weight fragment that can be isolated (7) from the MoFe protein.
The Mossbauer spectrum of the native protein (1) shows that 16 of the remaining iron atoms reside in diamagnetic environments. Since these iron atoms are associated with the bulk of the protein molecule, rather than the MoFe cofactor, their environments are referred to as P centers (3). Mossbauer spectra of the MoFe protein reveal the presence of a third iron environment, referred to as the S centers (3), which contain 2 iron atoms/mol.
The MoFe protein can be reversibly oxidized by the dye thionine (3). The oxidation proceeds in two distinct phases. The protein can be oxidized by 4 electron eq without any loss of the EPR signal from the M centers. After oxidation by 4 electron eq, the iron atoms in the P centers give rise to a Mossbauer spectrum characteristic of S 2 % spin systems.
After oxidation of the protein by 6 electron eq, the EPR of the M centers disappears and the Mossbauer spectrum shows that the M center iron atoms are now in a diamagnetic environment. On the basis of these results it has been proposed that there are 2 M centers/molecule, with each center containing 6 iron atoms, and 4 P centers/molecule, each with 4 iron atoms. Chemical studies (8,9) have confinned the presence of four Fe4Ss centers in the noncofactor part of the protein. The environment of the S center iron atoms, as characterized by the Mossbauer spectra, undergoes no apparent change during thionine oxidation of the protein.
Measurement of the magnetic susceptibility can provide insight into the spin state of the iron-sulfur centers in proteins. In this paper, we report on our efforts to understand the changes due to thionine oxidation in the spin state of the P centers through a measurement of the change in the magnetic susceptibility of the MoFe protein caused by thionine oxidation. A preliminary report of our results has been published elsewhere (10).

MATERIALS AND METHODS
Susceptibility Measurements-Magnetic susceptibilities were measured by the NMR method (11). NMR spectra were obtained using a Bruker WH-270 spectrometer using 5-mm coaxial NMR tubes (Wilmad Glass Co., Buena, NJ, Model 517). For anaerobic work, the inner tubes of the coaxial cells were fitted to 715 standard taper outer joints. This allowed the NMR tubes to be attached to the double septum seal devices that have been described previously for use in low temperature EPR studies (12). After the addition of MoFe protein and the other reagents to the inner tubes (see below) the tubes were permanently sealed with a gas-oxygen torch flame.
Enzyme Samples-The MoFe protein from Azotobacter vinelandii was prepared by the method of Shah and Brill (13). For Fourier transform NMR experiments, the solvent proton content of the samples was reduced by the following modification of the protein crystallization procedure. After the initial centrifugation of the crystals of the MoFe protein, the pellet was resuspended in 0.04 M NaC1, 0.025 M potassium phosphate, DzO (99.7 atom '% D) buffer with pD = 7.35. After incubation at 37 "C for 1  All solutions used in the preparation of the enzyme samples contained 1 mM sodium dithionite. The integrity of the MoFe protein preparations used in this work was verified by comparison of their low temperature EPR spectra with previously published results (6) and assay of the acetylene reduction activity in the presence of excess Fe protein (14). The specific activities of the MoFe protein preparations used in this work ranged from 1,540 to 2,000 nmol of ethylene produced/min/mg of MoFe protein. Protein concentrations were determined by the biuret reaction using bovine serum albumin as a standard (15). Analysis of a typical MoFe protein preparation for molybdenum (16) and iron (17) showed the presence of 1.6 mol of Mo and 23 mol of Fe/mol of MoFe protein of 243,000-dalton molecular mass.
Preparation of Samples for Susceptibility Studies-For susceptibility measurements, the inner and outer chambers of a double septum-sealed NMR cell were evacuated and fiied with oxygen-free argon at least six times. Reagent solutions were prepared in double septum-sealed tonometers. As an added precaution against the influx of oxygen, a 0.1 M solution of sodium dithionite with methyl viologen as an indicator of reduction potential was placed in the outer chambers of the double septum-sealed containers. A long-needled gas-tight syringe was used to add protein solutions and other reagents to the anaerobic NMR cells.
For a typical measurement, an anaerobic NMR cell would be prepared containing 0.2 ml of a 64 mg/ml solution of MoFe protein.
It has been found (3) that the presence of a trace of methyl viologen is necessary to ensure the complete reduction of the P centers even in the presence of an excess of sodium dithionite. Consequently, methyl viologen and additional sodium dithionite were added to final concentrations of 10 p~ and 50 p~, respectively. Samples were then allowed to incubate for 30 min at 26-28 "C. Excess sodium dithionite was consumed by the addition of clostridial hydrogenase and ferredoxin to concentrations of 0.8 and 3 p~, respectively, followed by incubation for 30 min at 26-28 "C. Sodium 2.2-dimethyl-2-silapentane sulfonate was added as a chemical shift reference at a final concentration of 2 mM. The protein sample was then in what we shall refer to as the native state, with both the M and P centers reduced.
Samples of native MoFe protein were oxidized by the addition of thionine as a 3.6 mM solution. Solutions of thionine were prepared in neat D,O since the dye is only sparingly soluble in 0.25 M KC1. Thionine solutions were standardized spectrophotometrically using an extinction coefficient of 6,000 cm" M" at 602 nm. Addition of thionine to the sample of MoFe protein caused some local precipitation. The precipitate redissolved in 5-10 min leaving the sample free of turbidity provided that strictly anaerobic conditions had been maintained. To check the reversibility of the oxidation, some samples were oxidized by 6 or 7 eq of thionine/mol of protein and then reduced with a 4-fold excess of sodium dithionite.
After making all additions and flame sealing the inner tubes, a 1.6 mM solution of DSS' was placed in the outer annulus of the coaxial NMR cell. The NMR spectrum of the sample was then measured and the resonance frequencies of the DSS methyl protons in both the inner tube and outer annulus solutions were recorded. The sealed samples could be saved for future study by rapidly freezing them in a stirred isopentane bath maintained at -140 "C followed by storage in liquid nitrogen.

RESULTS
The values of the frequency shift between the solution containing the MoFe protein and the reference solution were measured for 12 separate samples of MoFe protein at 280 K (Table I). These data were taken from samples prepared and NMR measurements made on three separate occasions. We report the mean value of A f and the standard deviation of the mean value determined from four to six measurements of A f on the same sample. The actual protein concentrations of the samples vaned due to the addition of different volumes of reagents. The concentrations reported in Table I were   " The numbers in parentheses are the standard deviations of the oxidation can be largely reversed by reduction with sodium dithionite. Thus, changes in the susceptibility can be interpreted as arising from a reversible oxidation rather than from an irreversible destruction of the protein.
A change in the chemical shift of the probe species due to a specific interaction with the solute is a potential source of error in susceptibility measurements by the NMR technique. The chemical shifts of DSS and acetate ion were measured as a function of concentration in the presence of 0.1 m~ MoFe protein and an internal chemical shift reference of 7 mM dioxane. The ratio of concentrations of DSS or acetate to the concentration of the MoFe protein was varied from 0.4:l to 12.0:l. Over this range, the change in the chemical shifts of acetate and DSS relative to that of dioxane was less than 0.4 Hz. This suggests that at least for the native protein there is no specific interaction between DSS or acetate and the protein that leads to a change in the resonance frequencies of DSS or acetate.
There is a significant amount of scatter in the A f values for oxidized MoFe protein samples with nominally equal degrees of oxidation. The scatter might be due to differences in the amount of sodium dithionite remaining after incubation with hydrogenase. Preparations of the MoFe protein contain a 5fold molar excess of sodium dithionite over protein. A 10% variation in the concentration of dithionite remaining after treatment with hydrogenase could easily change the effective degree of oxidation by as much as 1 electron eq/mol of protein.
There is considerably less scatter in the values of Af for the native protein samples.

DISCUSSION
The magnetic field experienced by a nuclear spin in a solution sample depends on the bulk magnetic susceptibility of the solution. Consequently, the NMR frequency of a probe nucleus in a sample solution will shift by A f (in Hz) upon the addition of mg/ml of a solute with mass susceptibility x. The susceptibility of the solute may be determined from A f using the relation In Equation 1, fis the resonance frequency in Hertz, x. and do are, respectively, the mass susceptibility and density of the reference solution, and d, is the density of the sample solution containing the solute of interest. Equation 1 is written for the case where the applied magnetic field is along the cylindrical axis of the sample tube (11).
The values of XO, do, and d, must be known in order to

Magnetic Susceptibility
of Nitrogenase obtain a value of x from Af. These quantities can be calculated from published susceptibility, density, and partial specific volume data (18-21). Although the limited precision of the available density data permit only a rough estimate of d, - do, a simple calculation shows that for the samples used in this work, the changes observed in the fiist term of Equation 1 are much larger than the changes in the second term. For the samples used in this work, the second term in Equation 1 has an approximately constant value of -52.6 X IO-* c.g.s., whereas the value of the fvst term changes from 7.63 X lo-* c.g.s. for a typical sample of native MoFe protein to an average value of 25.6 x lo-' c.g.s. for the thionine-oxidized protein.
Consequently, the uncertainty in d8 -do will not have a major impact on the accuracy of a measurement of the change in solute susceptibility due to thionine oxidation. Magnetic susceptibility measurements by the NMR method provide a value for the sum of the paramagnetic (xp) and the diamagnetic ( x d ) susceptibility of the solute. We wish to measure the change in XI, caused by thionine oxidation in order to study changes in the spin states of the iron-sulfur centers of the MoFe protein. The change in xp may be obtained directly from the measured change in x provided that thionine oxidation does not change x d . This appears to be a reasonable assumption to make in light of the finding (22) that the diamagnetic susceptibility of carbon monoxyhemoglobin is equal to the weighted sum of the diamagnetic susceptibilities of its constituent amino acids and porphyrin prosthetic groups. This shows that even gross changes in protein conformation have no major effect on protein diamagnetism. Consequently, we will assume that thionine oxidation causes no significant changes in Xd and that changes in the measured value of x directly reflect changes in xp.
The molar magnetic susceptibility xm is equal to the mass susceptibility times the molecular weight. The molar para- where g is the electronic g factor, provided that there are no significant contributions to the magnetic moment from incompletely quenched orbital angular momentum, or from excited electronic states (23).
According to the previously proposed model ( = (4pp2 -2p~')P'N/3kT (4) From the data in Table I, the average molar susceptibility of the native protein was found to be -0.110 +. 0.002. The average molar susceptibilities of samples oxidized (nominally) by 5, 6, 7, and 8 electron eq of thionine/mol of protein was -0.065 -C 0.006. We found, using Equation 4, that the difference in effective magnetic moments (4pp2 -2@M2) was 101 +.

13.
The metal contents of our preparations were somewhat -2 p~' for several half-integer spin quantum numbers for the oxidized P centers ( Table 11). There is good agreement between the experimentally determined value and the calculated value for S = 94 P centers. Furthermore, the changes in the calculated value of 4pp2 -2pM2 due to integer changes in P center spin quantum number are much larger than the uncertainty in the experimentally determined value. This demonstrates that the NMR method is sufficiently sensitive to detect integer spin state changes in the P centers.
We now summarize explicitly the assumptions made in our analysis of the susceptibility data. We know that in the native protein at low temperature the M centers are paramagnetic and the P centers are diamagnetic. The reverse situation holds for the fully oxidized protein. Furthermore, the M centers in the native protein behave as S = % spin systems. We assumed that raising the temperature from 4 to 280 K does not alter this situation. We assumed that the effective magnetic moment of the M centers may be calculated from the spin quantum number using Equation there are no contributions from unquenched orbital angular momentum to the magnetic moment, so that a meaningful "spin only" magnetic moment may be calculated. We have also assumed that thionine oxidation does not alter the contribution of the S iron atoms to the susceptibility and that thionine oxidation does not alter the diamagnetic susceptibility of the MoFe protein.
We have presented an elementary analysis of the susceptibility of the MoFe protein in an effort to understand the spin state changes which accompany thionine oxidation. In view of the complex magnetic behavior of the metal centers in ironsulfur proteins, and the limited data available, a more sophisticated treatment of our results is unwarranted. Susceptibility data on both the native and oxidized proteins over a broad temperature range are clearly needed if the electronic structures of the metal centers in the MoFe protein are to be understood. The NMR technique is limited to temperatures between 270 K (the approximate freezing point of the samples) and 305 K (the temperature at which the MoFe protein comes We assume that the M centers in the native protein are S = 3/2 systems and that the effective magnetic moments and spin quantum numbers are related by Equation 3 in the text.

Spin quantum number for
Oxidized P centers . This is too narrow a range over which to establish a meaningful temperature dependence for the susceptibility. Susceptibility studies on only the native protein from 4 to 280 K would add materially to the utility of the NMR data by providing a known value for the effective magnetic moment of the M centers. This value would provide a benchmark for the calculation of effective magnetic moments of the P centers in oxidized proteins from NMR data.
In conclusion, we have shown that at 280 K, the thionineoxidized MoFe protein is more paramagnetic than the native protein.
The results showed that the effective magnetic moment of the oxidized P center is significantly larger than that of the native M center. Assuming that at 280 K the native M centers have effective magnetic moments corresponding to spin only S = Yi systems, we conclude that the P centers in the fully oxidized protein behave as S = 5/2 systems. This result is consistent with the findings of recent EPR studies on partidy oxidized MoFe protein that show spectra with electronic g values consistent with S = 5/2 spin systems? Finally, our results suggest that the NMR susceptibility technique may be a useful adjunct to EPR and Mossbauer spectroscopy for the study of the metal centers in nitrogenase.