Mossbauer Studies of Escherichia coli Sulfite Reductase Complexes with Carbon Monoxide and Cyanide EXCHANGE COUPLING AND INTRINSIC PROPERTIES OF THE [4Fe-4S] CLUSTER*

Mossbauer studies of the hemoprotein subunit (SiR) of E. coli sulfite reductase have shown that the siro- heme and the f4Fe-4SI cluster are exchange-coupled. Here we report Mossbauer studies of SiR complexed with either CO or CN- and of SiR in the presence of the chaotropic agent dimethyl sulfoxide (MezSO). The spectra of one-electron-reduced SiR. CN show that all five iron atoms reside in a diamagnetic environment; the ferroheme.CN complex is low spin and the [4Fe-451 cluster is in the 2+ oxidation state. Titration with ferricyanide affords a CN- complex of oxidized SiR in which the siroheme iron is low spin ferric, with the cluster remaining in the 2+ state. At low temperatures, paramagnetic hyperfine interactions are observed for the iron sites of the cluster, suggesting that it is ex-change-coupled to the heme iron. Reduction of one- electron-reduced SiR. CN and SiR. CO yields complexes with “g = 1.94”-type EPR signals showing that the second electron is accommodated by the iron-sulfur cluster. The fully reduced complexes yield well re- solved Mossbauer spectra which were analyzed in the spin Hamiltonian formalism. The analysis shows that the cluster subsites are equivalent in pairs,

Mossbauer studies of the hemoprotein subunit (SiR) of E. coli sulfite reductase have shown that the siroheme and the f4Fe-4SI cluster are exchange-coupled.
Here we report Mossbauer studies of SiR complexed with either CO or CN-and of SiR in the presence of the chaotropic agent dimethyl sulfoxide (MezSO). The spectra of one-electron-reduced SiR. CN show that all five iron atoms reside in a diamagnetic environment; the ferroheme.CN complex is low spin and the [4Fe-451 cluster is in the 2+ oxidation state. Titration with ferricyanide affords a CN-complex of oxidized SiR in which the siroheme iron is low spin ferric, with the cluster remaining in the 2+ state. At low temperatures, paramagnetic hyperfine interactions are observed for the iron sites of the cluster, suggesting that it is exchange-coupled to the heme iron. Reduction of oneelectron-reduced SiR. CN and SiR. CO yields complexes with "g = 1.94"-type EPR signals showing that the second electron is accommodated by the iron-sulfur cluster. The fully reduced complexes yield well resolved Mossbauer spectra which were analyzed in the spin Hamiltonian formalism. The analysis shows that the cluster subsites are equivalent in pairs, one pair having properties reminiscent of ferric sites whereas the other pair has features more typical of ferrous sites. The Mossbauer spectra of oxidized SiR kept in 60% (v/v) Me,SO are virtually identical with those observed for SiR in standard buffer, implying that the coupling is maintained in the presence of the chaotrope. Fully reduced SiR displays an EPR signal with g values of g = 2.53, 2.29, and 2.07. In 60% MezSO, this signal vanishes and a g = 1.94 signal develops; this transition is accompanied by a change in the spin state of the heme iron from S = 1 (or 2) to S = 0.
The NADPH-sulfite reductase of Escherichia coli is a large (Mr = 685,000) oligomeric protein (a8&) which catalyzes the 6-electron reductions of sulfite to sulfide and of nitrite to ammonia. The substrate-binding site resides on the @-subunit (termed SiR) which contains one siroheme, an isobacterio-* 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.
Fellow of the Medical Scientist Training Program at Duke University.
Recipient of National Science Foundation Grant PCM 7924877 and Veterans Administration Project Grant 7875-01.
Recipient of National Science Foundation Grant PCM 8005610.
chlorin, and one [4Fe-4S] cluster per mol. SiR can be isolated as a monomer by treatment of the holoenzyme with 4 M urea followed by DEAE chromatography. When provided with a suitable artificial electron donor such as reduced methylviologen, SiR is catalytically competent.
Using the deazaflavin-EDTA photoreduction system originally described by Massey and Hemmerich (l), Janick and Siegel (2) have shown that SiR is readily reduced by up to 2 electrons. The first added electron is accommodated by the siroheme and results in the loss of the S = 5/2 EPR signal characteristic of oxidized SIR. Addition of a second electron yields an EPR-active species with g values at g = 2.53, 2.29, and 2.07 (-0.65 spin/SiR in phosphate buffer) and two minority S = 3/ 2 species (together -0.15 spin/SiR). The addition of 60% (v/v) Me2SO' to fully reduced SiR produces drastic changes in the optical spectrum with a concomitant appearance of a strong "g = 1.94" EPR signal.
We have demonstrated (3,4) that the siroheme and the iron-sulfur cluster are exchange-coupled in the three oxidation states of SiR and in a turnover complex of SiR with nitrite (a ferroheme. NO intermediate). These studies have shown that Mossbauer spectroscopy is uniquely suited to study this novel system. Here we have extended those studies to complexes of SiR with carbon monoxide and cyanide. Optical and EPR data (5,6) have shown that these ligands bind stoichiometrically and that they coordinate to the heme prosthetic group. Complexes with CN-can be prepared in states containing either a ferri-or a ferroheme, whereas CO binds only in those states where the heme iron is ferrous.
Studies of the reduced inhibitor complexes are of particular interest for the following reason. In a low spin ferrous state, the siroheme iron is not suitable for the development of interatomic exchange. Therefore, the spectroscopic features of the two prosthetic groups are affected only in minor ways by the coupling between them. Thus, these states provide us with an opportunity to study the intrinsic spectroscopic properties of the [4Fe-4S] cluster. Such information is indispensable for an understanding of the exchange-coupling phenomena in SiR.

Sulfite Reductase
Complexes with Carbon Monoxide and C.yanide Mosshauer samples of SIR ligated to CO or CN-were prepared by partially photoreducing anaerobic sclutions of 250-300 p~ SiR, 50 p~ deazaflavin, 10 mM EDTA, and 1 mM CN-or 0.6 mM CO in EPR tubes to allow rapid hinding of the ligand to the siroheme. The partially reduced SiR'CN samples (at least 1-electron reduced) were either directly transferred to a Mossbauer cup in a Thunherg tube under Ar or fully oxidized by addition of ferricyanide (added in small aliquots while the optical spectrum was monitored) before transfer. The samples were frozen in liquid nitrogen. The oxidation state of the complex was determined optically by comparison with spectra obtained during a ferricyanide titration of fully reduced SIR. CN (see Fig. 2 of Ref. 5). Since photoreduction of ,!3iR.CN past the 1-electron state was very slow, the fully reduced SiR. CN sample was prepared by addition of dithionite (2 mM final concentration) to the EPR tube, followed by anaerobic transfer to a single Mosshauer cup in a Thunherg tube under Ar and then frozen in liquid nitrogen. For SiR.CO, 150 pl of the partially reduced SiR. CO was transferred to each of two Mossbauer cups in a Thunherg tube under Ar and an anaerobic EPR tube. The samples were further illuminated until complete reduction was achieved as determined using the optical spectrum of the sample in the EPR tube. The samples were then frozen in liquid nitrogen for EPR and Mosshauer analysis. The two Mosshauer cups were placed hack to hack in the spectrometer for analysis. (A single 300-pl sample would have been optically too dense to permit complete photoreduction.) Optical spectroscopy was performed using an Aminco DW-2 dual beam spectrophotometer. EPR spectra were recorded as described previously (2, 5, 10) using a Varian E9 spectrometer equipped with an Air Products He gas transfer cryostat. Spin concentrations were determined by quantitation with a cupric EDTA standard. The Mosshauer spectrometers used were of the constant acceleration type, as descrihed in the accompanying paper (4). The isomer shifts, 6, are quoted relative to the centroid of the spectrum of iron metal recorded a t room temperature.

Sulfite Reductase Complexed with Carbon
Monoxide-Murphy et al. (11) have shown that CO binds tightly to the siroheme iron of reduced SIR. Since CO is a strong field ligand, the siroheme iron of SiR"-CO is expected to be low spin ferrous (S, = 0). In the fully reduced state, SiR2-. CO,' the complex exhibits an EPR signal of the g = 1.94 type with g values at 2.029, 1.925, and 1.910 (5). This signal, which is characteristic of the 1+ oxidation state of the [4Fe-4S] cluster, accounts for 1 spin/siroheme. Fig. 1 shows a Mossbauer spectrum of SiR". CO recorded at 4.2 K in zero field. The spectrum exhibits a sharp quadrupole doublet superimposed on a broad component. The latter belongs to the paramagnetic [4Fe-4SI1+ cluster. In general, one records spectra of paramagnetic species in applied magnetic fields H > 10 milliTeslas. In zero field, such spectra tend to be broader and more poorly resolved. We have used this to our advantage; because the [4Fe-4SI1+ cluster spectrum is broad, the spectrum of the siroheme appears as a conspicuous quadrupole doublet in Fig. 1. By matching two Lorentzian lines to the doublet, we found that it accounts for -20% of the total absorption, i.e. the doublet has the proper intensity to be assigned to the siroheme'C0 complex. The observed isomer shift 6 = 0.23 k 0.02 mm/s is typical of ferrous heme. CO complexes. The value for the quadrupole splitting, AE, = 0.80 t 0.03 mm/s, is somewhat larger than those observed for other heme.CO complexes (see Table 9 of Ref. 12,and Ref. 13). The parameters obtained here strongly implicate a low spin ferrous siroheme.
Since low spin ferrous hemes have no thermally accessible excited electronic states, their spectra have temperature-in-We will identify the oxidation states of SIR and of its complexes by the superscripts 0, 1-, and 2-, which refer to the oxidized, 1electron-and 2-electron-reduced states, respectively.   The spectra were prepared from the raw data by removing the 20% Contribution of the siroheme. CO complex. The 6-Tesla spectrum was recorded in a larger velocity range and scaled into the range of the spectra of A and B; the scaling procedure smooths the counting statistics between adjacent velocity channels. The solid lines drawn through the data are theoretical curves generated from Equation 1 as described in the text. In B and C, a decomposition into site I and site I1 (dashed) is indicated. dependent quadrupole splittings. Within the resolution of our data, the siroheme.CO complex fits this pattern. Since the spectrum of the siroheme. CO is well defined, its contribution can be reliably subtracted from the raw data to yield the spectra of the [4Fe-4SI1+ cluster. Fig. 2 shows spectra of the [4Fe-4SI1+ cluster recorded at 4.2 K in external magnetic fields of 60 milliTeslas (A and R ) and 6.0 Teslas (C). The spectra (discussed below) exhibit magnetic hyperfine interactions, as expected for an EPRactive species with slow electronic spin relaxation. At higher temperatures, above 50 K, the relaxation rate becomes fast compared with the nuclear precession frequencies. Consequently, the magnetic hyperfine interactions are averaged out and only quadrupole doublets are observed. A representative spectrum, taken at 110 K, is shown in Fig. 3A. The solid line indicates the 4.2 K spectrum of the siroheme. CO (determined from Fig. l), shifted by -0.03 mm/s to account for the second order Doppler shift and scaled to represent 20% of the total Fe. The spectrum in Fig. 3 Table I. A detailed analysis of the 4.2 K spectra displayed in Fig. 2 shows that the four sites of the iron-sulfur cluster are still equivalent in pairs even when the magnetic hyperfine inter-  12 In Equation 1, S = I/z is the cluster spin and g the electronic g tensor, known from EPR. A is the magnetic hyperfine tensor and V,,, V,,, and V,, are the principal axis components of the electric field gradient tensor; q = ( V, -V,,)/V, is the asymmetry parameter; AEQ = (eQV,,/2)( 1 + $/3)"' is the quadrupole splitting. The parameters obtained from a series of computer simulations are listed in Table 11.
In the following, we discuss briefly the salient features of the spectra; for this discussion it is assumed that-A and g are isotropic. In the presence of an applied field H , each "7Fe nucleus senses-an effective field Neff which is+the su-m of an internal field Hint and the applied field; He,, =  A comparison of the spectra of Fig. 2, B and C, shows that they contain a component whose magnetic splitting increases with increasing applied field, i.e. Heft > Hi,, and therefore A > 0. Virtually ail absorption a t Doppler velocities around -2 and "3 mm/s in Fig. 2C  The well defined outer features of the 6.0-Tesla spectrum of c o~p o n e n t 11 determine its parameters to a large extent. Detailed computer simulations revealed that the magnetic hyperfine tensor must be anisotropic, and that AEQ > 0. After a reasonable set of parameters was found, the theoretical  field, i.e. A < 0. Moreover, the analysis showed that the A tensor of component I is quite isotropic and that its components are larger in magnitude than those of spectral component 11. We will argue below that the irons of site I are somewhat ferric in character whereas those of site I1 have features of Fe(I1).
For the data analysis, we have assumed that all tensors in Equation 1 have the same principal axis system. Since the g values are virtually isot_ropic, it is not possible to correlate the components of g and A. Furthermore, since the A tensor of site I is isotropic, its components cannot be correlated with the electric field gradient tensor. For site 11, however, the components of the field gradient tensor can be correlated with those of the anisotropic A tensor; the field gradient components are adjustable by the value for ?. The spectra displayed in Fig. 2 show that the overall agreement between experiment and theory is quite good; thus, we felt no need to rotate the A tensor of site I1 relative to the field gradient tensor.
Once reasonable values for the parameters were obtained, we performed an extensive series of simulations aimed at optimizing the parameter set with respect to the whole set of experimental data. These calculations also yielded reasonable estimates for the uncertainties; the estimates are based on visual inspections and are quoted in Table 11.
Sulfite Reductase Complexed with Cyanide-Using optical and EPR spectroscopy, Janick and Siege1 (5)   essentially the same values for AEQ and 6. The observed isomer shift, 6 = 0.45 f 0.02 mm/s, unambiguously shows the cluster to be in the 2+ oxidation state. Furthermore, the values of 6 and AEQ = 1.00 f 0.03 mm/s are the same as those observed for the cluster of oxidized and 1-electron-reduced SiR (3, 41, of SiR".NO (4) and SiR'.CN (see below); in all these states, the cluster is in the 2+ state. The quadrupole splitting observed here is independent of temperature for 1.5 K 5 T 5 200 K. The spectrum shown in Fig. 4B was taken at 4.2 K in a field of 6.0 Teslas. The solid line is a theoretical spectrum computed with the assumption that all five iron sites are diamagnetic. The triplet structure of the 6.0-Tesla spectrum results from a powder average of a diamagnetic species with AEQ = 1.0 mm/s whose nuclei experience an effective field of 6.0 Teslas, i.e. Hint = 0. The positions of the three major bands are practically fixed by the magnitude of AEQ and the value of Heff. The excellent agreement between the theoretical curve and the data shows that SiR" . CN is a diamagnetic compound. In a minor way, the spectrum depends on the signs of AEQ and the values of the asymmetry parameters 9. Values for 7 < 0.7 produce a shoulder on the low energy side of the central band if AEQ > 0 whereas the shoulder is observed on the high energy side for AEQ < 0.
Clearly, the majority of the iron sites of SiR1-.CN must have AEQ > 0. It is reasonable to assume that these sites belong to the [4Fe-4SIZ+ cluster, since positive AEQ values are also suggested from the analyses of the SiRo (3) and SIR". NO (4) spectra. For the fit displayed in Fig. 4B, we used AEQ > 0 for all sites of the iron-sulfur cluster, together with 9 = 0 for two cluster subsites and 9 = 0.5 for the other two sites. For the siroheme iron, we used AEQ < 0 and 9 = 0.5. Since we are dealing with a multiparameter problem, the 7 values are not uniquely determined.
The fully reduced cyanide complex, SiR2-. CN, exhibits a g = 1.94-type signal with g values at 2.03, 1.93, and 1.91, values almost identical with those observed for the corresponding CO complex. We have prepared a sample by adding dithionite, as described above. Examination of the Mossbauer spectra revealed that the sample contained about 15% of SiR". CN. After removing this contribution from the raw data, we obtained spectra very similar to those recorded for SiR*-. CO. A spectrum taken under the same conditions as that shown in Fig. 1 revealed that the siroheme.CN moiety contributes a quadrupole doublet at 4.2 K with AEQ = 0.72 k 0.05 mm/s and 6 = 0.39 f 0.03 mm/s, confirming our expectations that the heme. CN complex is low spin ferrous. (Surprisingly, there have been no reports of Mossbauer spectra for any ferrous heme. CN complexes.) After subtracting the siroheme. CN doublet from the data, we obtained spectra for the [4Fe-4SI1+ cluster which were almost identical with those shown in Figs. 2 and 3B. AS an example, 4.2 K spectra of the cluster as observed in SiR2-. CN (hashmarks) and SiR2-. CO (full circles) are shown in Fig.  5A.
The fully oxidized cyanide complex, SiRO. CN, exhibits an EPR spectrum with g values at g = 2.39, 2.33, and 1.67 (see Ref. 6 and Footnote 3 of Ref. 5 ) . These g values are characteristic of low spin ferric isobacteriochlorin complexes (16). An EPR spectrum taken on the sample studied here revealed that more than 95% of the detectable EPR signals were attributable to SiR'. CN.
We will discuss the high temperature Mossbauer data first. At temperatures above 70 K, the electronic spin relaxation  The solid line in R is the result of a computer simulation using Equation 1 for the spectrum of the (4Fe-4SJ2+ cluster; the spectrum has been scaled to 80% of the total absorption and i t s base-line has been displaced downward to facilitate comparison with the data. The dashed curue is a spectral simulat,ion for the low spin ferric. heme. CN complex in the framework of the ligand field model described in the text. Fig. 6A. It Fig. 6A), we obtained AEQ = 1.35 t 0.08 mm/s and 6 = 0.19 k 0.05 mm/s at 195 K. At 90 K, the quadrupole splitting is slightly larger, AEQ = 1.44 k 0.08 mm/s. The values for AEq and 6 agree very well with those typically observed for protoporphyrin-cyanide complexes (see Table 4 of Ref. 17) and those reported for the cytochrome a3.CN complexes of the Thermus thermophilus (18) and the beef heart cytochrome oxidases (see Footnote 7 in Ref. 18).

CN is shown in
A low temperature spectrum of SiR('.CN at 4.2 K in a 60-milliTesla parallel field is shown in Fig. 6B. We will focus first on the siroheme spectrum. A comparison of the spectra in Fig. 6, A and B, shows that the quadrupole doublet of the siroheme has vanished and that a broad magnetic component has appeared. The lack of resolution precludes a quantitative evaluation of this component. We can, however, compare the shape and the overall splitting of the spectrum with the features predictable from the EPR data and with information reported for other heme. CN complexes. We have therefore calculated theoretical spectra in the framework of the ligand field model proposed by Griffith (19) and further developed by Blumberg and Peisach (20) and Oosterhuis and Lang (21). In this model, the g values determine the ratios A/X and V/X where A and V are tetragonal and rhombic distortion parameters, respectively, and where X is the spin-orbit coupling constant. The observed g values yield A/X = 1.79, V/X = 2.95, and h = 0.71 for the orbital reduction factor. From these parameters, and using P = -(4.2 mm/s)N2 = -1.8 mm/s to scale the A tensor (for definitions, see Refs. 17 and 2l), we have computed the t.heoretica1 spectrum (dashed curue) shown in Fig. 6B. Clearly, the experimentally observed spectrum for the siroheme iron is not in disagreement with the theory.
Comparison of the spectra in Fig. 6, A Fig. 6B is caused by the presence of magnetic hyperfine interactions. Since the broadening persists in zero field, the cluster belongs to a system of half-integer spin. For This description yields a reasonable representation of the broadening. We have also studied the sample at 4.2 K in a 6.0-Tesla field. The [4Fe-4S] cluster spectrum was well described by two components with effective magnetic fields of 5.2 and 6.9 Teslas, corresponding to internal fields in agreement with the above quoted A values. We will argue below that the magnetic hyperfine broadening of the cluster absorption lines reflects exchange interactions between the siroheme and the iron-sulfur cluster. The data presented here show that the siroheme is low spin ferrous in both SIR'-.CN and SiR2-.CN. Interestingly, the Mossbauer parameters of the siroheme iron are different for these states. For SiR2-.CN, we obtained AEQ = 0.72 f 0.05 mm/s and 6 = 0.39 f 0.03 mm/s at 4.2 K; these parameters could be extracted quite easily from the zero field spectrum. Precise parameters of the siroheme doublet of SiR". CN are more difficult to determine since the doublet is completely masked by the contribution of the [4Fe-4S] cluster. Since SiRo. CN and SIR". CN both have the iron-sulfur cluster in the 2+ state, we have produced a difference spectrum by subtracting the 140 K spectra of the two samples. This procedure should cancel the contribution of the iron-sulfur cluster. From an analysis of the resultant difference spectrum (22), we obtained for the siroheme.CN moiety of SiR". CN the values AEQ = 0.9 f 0.1 mm/s and 6 = 0.42 -+ 0.05 mm/s. Taking the second order Doppler shift into account, this isomer shift was extrapolated to 6 = 0.45 mm/s at 4.2 K. Since the spectrum of SIR". CN exhibits a temperature-independent quadrupole splitting, we can assume that the siroheme. CN moiety has AEQ = 0.9 mm/s at 4.2 K as well. Thus, the siroheme . CN complex has different values for AEQ in SiR2-.
CN and SiR".CN; the isomer shifts, however, are the same within the uncertainties.
Oxidized and Fully Reduced SiR in 60% Dimethyl Sulfoxide-Janick and Siege1 (5) have reported EPR and optical studies of SIR in the presence of MezSO in various concentrations. In the presence of 60% Me2S0, the optical spectrum of SiRo is t,he same as that observed in standard buffer; the EPR spectra show a slight shift in g values from 2 = 6.63, 5.24, and 1.98 to g = 6.54, 5.35, and 1.99, i.e. the rhombicity parameter of the zero-field splitting term shifts from E / D = 0.029 to 0.025. The Mossbauer spectra (22) of the 60% MezSO sample are virtually identical with those reported (3) for the native enzyme. The small differences observed can be attributed entirely to the differences in E / D . Since the two chromophores are exchange-coupled in SiRO, the similarity of the spectra observed in standard buffer and in 60% MezSO solution implies that the siroheme and the iron-sulfur cluster remain exchange-coupled in the presence of 60% Me2S0.
The addition of 60% Me2S0 to SiR2-changes the EPR and the optical spectra dramatically (5). The g = 2.29 and the S = 3/2 species observed in SiR2-(2) disappear and the only observable EPR-active species exhibits a g = 1.94-type signal with g = 2.033, 1.928, and 1.905 (see Table I  K in a 60-milliTesla field is shown in Fig. 5B (hashmarks).
The spectrum is a superposition of a magnetically split component which is identical with the spectrum (full circles) of the [4Fe-4SI1+ cluster of SiR2-.C0, and of a quadrupole doublet (bracket) with Mossbauer parameters AEQ = 1.45 f 0.03 mm/s and 6 = 0.44 k 0.02 mm/s, with AEQ independent of temperature for T < 200 K. These parameters suggest strongly that the siroheme iron is low spin ferrous, a conclusion which agrees with the interpretation of the optical spectra (5). In the absence of Mossbauer data of suitable siroheme model complexes, an interpretation of the values obtained for AEQ and 6 in terms of ligand structure is not yet possible. While it is conceivable that MezSO occupies the sixth coor-dination position of the heme iron, the strength of this ligand is probably insufficient to affect a transition to a low spin configuration (23,24). More likely, the chaotropic Me2S0 has perturbed the protein conformation sufficiently to allow an endogenous ligand to gain access to the heme iron.

DISCUSSION
We have demonstrated previously (3) and in the accompanying paper (4) that the siroheme and the iron-sulfur cluster are exchange-coupled in SiRo (in standard buffer and in 60% Me2SO), in SiR", in SiR2-, and in the turnover complex, SiR".NO (in standard buffer as well as in 2 M urea). The data obtained here suggest that the coupling is also maintained in SiRO. CN. The arguments supporting this claim are as follows. The presence of paramagnetic hyperfine interactions in the 4.2 K Mossbauer spectra of the siroheme and the iron-sulfur cluster demonstrates that each prosthetic group belongs to a system of half-integer electronic spin. This spin system must be common to both groups, because if both groups belonged to separate spin systems (one yielding the observed EPR signal, the other EPR silent), then an even number of electrons would be required to produce the diamagnetic state of SiR". CN, in contrast with the experiments (5) which show that SiR". CN is produced from SiRO. CN by a 1-electron reduction.
In SiRO, SiR", SiR2-, and SiR".NO the heme iron has unpaired d electrons and, thus, these states are suitable for the development of interatomic exchange interactions. In contrast, the complexes of SiR2-with CO and CNas well as SiR2-in 60% MezSO have a low spin ferrous ( S , = 0) heme iron. In this state, the heme iron has a closed shell configuration which is unfavorable for the development of exchange interactions. Therefore, these three complexes do not convey information, through the observation of magnetic effects, about the coupling of the chromophores.
Although we cannot prove that the two chromophores are directly linked by a bridging ligand in SiR2-.C0 and SiR2-. CN, there is clear evidence that the two chromophores are interacting. First, the optical spectra of the heme show distinct changes when the iron-sulfur cluster is reduced from the 2+ to the 1+ state (5). Second, the redox potential of this step increases by 70 mV when CN-is replaced by CO (6). Furthermore, the quadrupole splitting of the low spin ferrous herne.CN complex changes when the iron-sulfur cluster is reduced. Finally, the magnetic hyperfine broadening observed for the [4Fe-4SI2+ cluster of SiRo.CN suggests exchangecoupling of the prosthetic groups. The observations listed here suggest the maintenance of coupling but they do not prove it, because even if the chemical bond between the heme and the cluster were broken in the reduced CO and CN complexes, the chromophores would be expected to remain close enough for some interactive contact.
The [4Fe-4S] cluster has been observed here in two oxidation states. In Sil3O.CN and SiR".CN, it is in the 2+ state, whereas the 1+ state is observed for the 2-electron-reduced states of the SiR complexes. Our data show that the four iron sites of the [4Fe-4SI2+ cluster have, even within our good resolution, identical values for AEQ and 6. This is true in all states for which the SiR cluster is observed in the 2+ state. Furthermore, AEQ is independent of temperature for 1.5 K 5 T 5 200 K. This temperature independence implies that there are no thermally accessible excited states with orbital character different from that of the ground state. In this respect, the SiR cluster is similar to the cluster of phosphoribosylpyrophosphate amidotransferase from Bacillus subtilis (25); all other clusters studied thus far exhibit temperaturedependent AEQ values. Our high field studies of SiR'-.CN prove that the [4Fe-4S] cluster of SiR has a diamagnetic ground state in the 2+ state, in agreement with information obtained for [4Fe-4SI2+ clusters of other proteins and model complexes. This result demonstrates that the iron-sulfur cluster of SiR is intrinsically the same as the familiar clusters of other systems. This is significant in view of our observations that the electronic ground state of the cluster has acquired, through exchange-coupling to the siroheme, paramagnetism in some states of SiR, including SiR'.CN.
In the 2-electron-reduced states of SIR, the [4Fe-4S] cluster is observed in the 1+ state. Under conditions where the magnetic hyperfine interactions are averaged out, i.e. at higher temperatures in the limit of fast electronic spin relaxation, the 1+ states yield spectra consisting of two well resolved quadrupole doublets, each doublet representing two iron sites. The two types of sites exhibit quite different magnetic hyperfine interactions at 4.2 K. Site I has an isotropic A tensor with negative components whereas site I1 is characterized by an A tensor with substantial anisotropies and positive components. Since the iron sites of [4Fe-4S] clusters have a tetrahedral ligand coordination, they have high spin electronic configurations. A comparison of the parameters shows that site I has features reminiscent of ferric ions whereas site I1 has some ferrous character. This is borne out by the observation that site I1 exhibits the larger values for AEQ and 6, although the differences between the two sites are considerably less pronounced than the differences observed for the suitable model, ferric and ferrous rubredoxin (26,27). The magnetic hyperfine interactions of high spin ferric ions derive from the isotropic Fermi contact term which is negative. In contrast, the A tensors of high spin ferrous sites are considerably anisotropic by virtue of large spin-dipolar and orbital contributions. The A values of site I and I1 reflect this pattern. It is noteworthy that the "ferrous" site has positive A values; A tensors with positive components result when the local spin is oriented antiparallel to the system spin in a spin-coupled cluster (28,29). In emphasizing that sites I and I1 are ferric and ferrous in character, respectively, we do not mean to imply that we are dealing here with trapped valences as observed for [2Fe-2S] ferredoxins. However, the degree of electron delocalization in reduced [4Fe-4S] clusters is not as pronounced as is often assumed in the literature.
We do not know whether the differences between sites I and I1 are primarily imposed by the protein environments or whether they reflect an intrinsic electronic property of ironsulfur clusters. Recent studies of a centrosymmetric [2Fe-2S] model complex (30,31) suggest that site inequivalencies can be an intrinsic electronic property of iron-sulfur clusters. Alternatively, one might attribute the differences observed here to the fact that the cluster is linked to the siroheme. This explanation is implausible in view of the fact that very similar Mossbauer parameters have been obtained for the ferredoxin from Bacillus stearothermophilus (14) and for a reconstituted [4Fe-4S] cluster in a ferredoxin from D. gigas (15). The capacity of [4Fe-4S] clusters to develop pronounced localized valences has recently been demonstrated for aconitase; upon binding of citrate, one site of the [4Fe-4S] cluster of that enzyme acquires a distinctly high spin ferrous character (32).
This paper together with our other studies of SiR (3, 4) provide the first Mossbauer results for Fe-isobacteriochlorins. Since the optical and EPR properties of siroheme (2,5) and Fe-isobacteriochlorin model compounds (16) are quite different from those of porphyrins, it was of interest to see whether the Mossbauer spectra displayed unusual features attributable to the doubly reduced macrocycle. Regarding the isomer shifts, we have observed no differences between the values for similar siroheme and porphyrin complexes. Values determined for the quadrupole splittings are also comparable, except for that of the ferrous siroheme.CO complex. The value of AEQ for the latter is distinctly larger than those reported for porphyrin. CO complexes. However, until the ligand trans to the coordinated CO has been identified, it is not possible to decide whether this is attributable to the bacteriochlorin macrocycle. In the ferric states, we have observed magnetic hyperfine interactions in addition to electric hyperfine interactions. An analysis (3) of the spectra of the high spin ferric siroheme yielded parameters which are quite typical for porphyrins. In fact, the values found for the zero-field splitting and the magnetic hyperfine coupling constant of the siroheme of SiR' match those reported for metmyoglobin (33, 34). The EPR results for the low spin Fe(II1). siroheme-CN complex indicate ligand field parameters distinct from those of porphyrin. CN complexes. These differences seem to be reflected in the magnetic features of the Mossbauer spectra as well. (However, since the siroheme spectrum is partially masked by the contribution of the [4Fe-4S] cluster, we are prevented from making a thorough analysis of the siroheme spectrum.) Mossbauer studies of appropriate low spin ferric and low spin ferrous bacteriochlorin model complexes should be illuminating. However, one should keep in mind that the ligand trans to the CO and the CN-ligand possibly provides the link to the iron-sulfur cluster, therefore complicating interpretation of such model complex data.