EPR and Mossbauer Studies of Protocatechuate 4,5=Dioxygenase CHARACTERIZATION OF A NEW Fez+ ENVIRONMENT*

Protocatechuate 4,5-dioxygenase from Pseudomonas testosteroni has been purified to homogeneity and crystallized. The iron containing, extradiol dioxygenase is shown to be composed of two subunit types (a, M, = 17,700 and B, M, = 33,800) in a 1:l ratio; such a composition has not been observed for other extradiol dioxygenases. The 4.2 K Mossbauer spectrum of native protocatechuate 4,5-dioxygenase prepared from cells grown in “Fe-enriched media consists of a doublet with quadrupole splitting, AEQ = 2.22 mm/s, and isomer shift bFe = 1.28 mmfs, demonstrating a high spin Fez‘ site. These parameters, and the temperature dependence of AEQ, are unique among enzymes but are strikingly similar to those reported for the reaction center of the photosynthetic bacterium Rhodopseu- domonas sphaeroides R-26, suggesting very similar ligand environments. The Fez+ of protocatechuate 4,5- dioxygenase can be oxidized, for instance by HzOZ, to yield high spin Fe3+ with EPR g values around g = 6 (and g = 4.3). In the oxidized state, protocatechuate 4,ti-dioxygenase is inactive; the iron, however, can be rereduced by ascorbate to yield active enzyme.


Protocatechuate 4,5-dioxygenase from
Pseudomonas testosteroni has been purified to homogeneity and crystallized. The iron containing, extradiol dioxygenase is shown to be composed of two subunit types such a composition has not been observed for other extradiol dioxygenases. The 4.2 K Mossbauer spectrum of native protocatechuate 4,5-dioxygenase prepared from cells grown in "Fe-enriched media consists of a doublet with quadrupole splitting, AEQ = 2.22 mm/s, and isomer shift bFe = 1.28 mmfs, demonstrating a high spin Fez' site. These parameters, and the temperature dependence of AEQ, are unique among enzymes but are strikingly similar to those reported for the reaction center of the photosynthetic bacterium Rhodopseudomonas sphaeroides R-26, suggesting very similar ligand environments. The Fez+ of protocatechuate 4,5dioxygenase can be oxidized, for instance by HzOZ, to yield high spin Fe3+ with EPR g values around g = 6 (and g = 4.3). In the oxidized state, protocatechuate 4,ti-dioxygenase is inactive; the iron, however, can be rereduced by ascorbate to yield active enzyme. Our data suggest that protocatechuate binds to Fe2+; the spectra indicate that the ligand binding is heterogeneous. The Mossbauer spectra observed here are fundamentally different from those reported earlier (Zabinski, R., Miinck, E., Champion, p., and Wood, J. M. (1972) Biochemistry 11,[3212][3213][3214][3215][3216][3217][3218][3219]. The spectra of the earlier (reconstituted) preparations, which had substantially lower specific activities, probably reflect adventitiously bound Fe3+. We discuss here how adventitiously bound iron can be identified and removed.
The Fe2+ which is present in native protocatechuate 4,5-dioxygenase and its complexes with substrates and inhibitors reacts quantitatively with nitric oxide to produce a species with electronic spin S = 312. The EPR and Mossbauer spectra of these complexes compare favorably with EDTA*Fe(II)*NO. We have studied the latter complex extensively and have analyzed the Mossbauer spectra with an S = 3/2 spin Hamiltonian. EPR spectra show that protocatechuate 4,5-diox-* 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.  ygenase-NO complexes with substrates or inhibitors are heterogeneous and consist of several well defined subspecies. The data show that NO, and presumably also 02, has access to the active site Fe2+ in the enzymesubstrate complex. The use of EPR-detectable NO complexes as a rapid and sensitive tool for the study of the EPR silent active site iron of extradiol dioxygenases is discussed.
Protocatechuate 4,5-dioxygenase (EC 1.13.11.8) catalyzes the incorporation of molecular oxygen into PCA' to produce a-hydroxy-y-carboxy-muconic semialdehyde (1). This "extradiol" cleavage is the essential ring opening step in the biodegradation of many aromatic compounds by nonfluorescent pseudomonads. The enzyme has been purified from Pseudomonas testosteroni (1,2) as well as Pseudomonas species (3). An analogous step in aromatic metabolism by fluorescent pseudomonads is catalyzed by the intradiol cleaving enzyme protocatechuate 3,4-dioxygenase. The latter enzyme has recently been shown to possess a new class of iron center and a unique mechanism for substrate and oxygen activation by a variety of experimental approaches (4,5). Similar details of the iron site structure and the mechanism of protocatechuate 4,5-dioxygenase are largely unexplored.
A number of other extradiol dioxygenases which catalyze the cleavage of catechol and its derivatives have been purified. Those isolated from pseudomonads reportedly contain ferrous ion (6,7) as an essential metal cofactor, although this has only been directly demonstrated for catechol 2,3-dioxygenase (8). In contrast, the intradiol cleaving enzymes, protocatechuate 3,4-dioxygenase (4) and catechol 1,2-dioxygenase (9), contain only Fe3+. This difference in the redox state of the iron is frequently cited as a fundamental distinguishing feature between the intra-and extradiol cleaving enzymes (10). An earlier Mossbauer investigation of 57Fe reconstituted P. testosteroni protocatechuate 4,5-dioxygenase (11) which showed that the enzyme did not fall into this general pattern was thus unexpected and is reinvestigated in this report.
Extradiol dioxygenases tend to present many more experimental obstacles than the intradiol enzymes. For example, they are quite labile and often undergo loss of activity during turnover (1,12). They also lack the visible spectrum and other convenient spectroscopic features of intradiol dioxygenases. As a result, progress in the investigation of these enzymes in general, and of protocatechuate 4,S-dioxygenase in particular, has been slow. We have recently (13) found conditions which allow preparation of P. testosteroni protocatechuate 4,5-diox-The abbreviations used are: PCA, protocatechuic acid SDS, sodium dodecyl sulfate. ygenase with much higher specific activity and stability than previously reported. The enzyme was also isolated from bacteria which had been cultured on media enriched in ,-"Fe so that Mossbauer spectroscopy could be performed without danger of creating artifacts, a possibility inherent in a reconstitution process. Preliminary Mossbauer and EPR results obtained with this enzyme (13, 14) were markedly different from t,hose previously reported (11). The enzyme contained approximately equal amounts of ferric and ferrous ion, and the iron species which had been observed in reconstit.uted preparations of Zabinski et 01. (1 1) was absent,.
In the present study we report the purification to homogeneity, crystallization, and partial characterization of P. testnsteroni protocatechuate 4,5-dioxygenase. Techniques are described which result in the specific removal of the majority of the ferric ion without activity loss. Mossbauer spectra of the resulting enzyme show that the remaining ferrous ion is high spin and resides in an environment which thus far has only been observed for photosynthetic reaction centers.

MATERIALS A N D METHODS ANI) RESULTS?
Purification of Protocatechuote, 4..5-Divxygrnase-The purification of protocatechuate 4,5-dioxygenase as described under "Materials and Methods" is summarized in Table I. T h e entire procedure is carried out in buffer containing 10% glycerol, 2 mM cysteine, and 100 VM ferrous ion. Although the purification procedure can be carried out in the absence of iron and cysteine in t.he steps preceding DE52 chromotography, these stabilizers are required to maintain activity in the later steps. The affinity chromatography step proved to be an effective method to remove contaminating proteins which were not readily separated by the other procedures. Vanillate was used rather than PCA as an affinity ligand in preparing this column because vanillate is less readily oxidized.
No metabolism of PCA linked to Sepharose has been observed.
T h e specific activity of this protocatechuate 4,5-dioxygenase preparation (specific activity = 212) is approximately 15 times that previously reported (2, ll), and comparable to that reported for purified P. species protocatechuate 4,S-dioxygenase (specific activity = 168) (3). T h e stabi1it.y of the purified enzyme is discussed in the "Miniprint."  Purified protocatechuate 4,5-dioxygenase failed to crystallize in aerobic solution but crystallized readily under strictly anaerobic conditions as described under "Materials and Methods." Fig. 1 shows the crystals which resulted from incubation in 40% methylpentanediol solution at pH 8. Similar crystals were obtained when 30% ethanol was used in place of the methylpentanediol. The specific activity of the enzyme was not increased by the crystallization procedure.
Molecular Weight and Subunit Structure-The molecular weight of protocatechuate 4,5-dioxygenase was estimated to be approximately 142,000 by elution from an agarose A-1.5m column (2.5 X 90 cm) equilibrated in standard purification buffer and calibrated with standard proteins as shown in Fig.  2. Polyacrylamide gel electrophoresis in the presence of SDS showed at least two subunits with relative M, = 17,700 ( a ) and 33,800 ( 8 ) (Fig. 3). No other stained protein bands were visible on these gels showing that the enzyme is homogeneous by this criteria. The subunits were resolved by chromatography on agarose A-5m equilibrated in 6 M guanidine HCI as shown in Fig. 3. Amino acid analysis of the enzyme and the resolved subunits (Table 11) shows that the amino acid content of the enzyme is accounted for by a 1:l ratio of subunits. Moreover, the integrated absorbance of the elution profile shown in Fig. 3 is consistent with a 1:l ratio of subunits with the extinction coefficients listed in Table 11. The integrated intensities of the Coomassie blue stained bands on the SDS gels shown in Fig.  3 are also consistent with a 1:l ratio of subunits.
In contrast to the chromatographic and electrophoretic indications of protein purity detailed above, ultracentrifugation experiments showed that the preparation contained species of different molecular weights. For example. plots of r'l ~'cr.su.s In (protein concentration) generated from sedimentation equilibrium data, were slightly curved. A similar conclusion could be drawn from gel electrophoresis experiments conducted under nondenaturing conditions which exhibited a broad, diffuse protein staining band. The specific stain for enzymatic activity described under "Materials and Methods" showed, however, that protocatechuate 4,5-dioxygenase was present throughout this band. Furthermore, when the gel containing the broad protein band was placed at the top of a slab gel containing SDS and electrophoresis conducted in a second dimension, only the protein staining bands associated with the subunits were observed. Together these data suggest that the molecular weight heterogeneity is observed because protocatechuate 4,5-dioxygenase can exist in several different forms in solution, all of which are composed of either the same subunits or suhunits of similar molecular weights. Since we have no indication of families of (Y and fi subunits differing in amino acid compositions, the different forms of the enzyme probably result from a dynamic equilibrium of quaternary structures or from a differential interaction with the stabilizing agents (see below). Iron analysis of the enzyme treated to remove excess ferric ion (see below) showed 0.5 f 0.05 Fe/n@ structure ( M , = 51,500) or approximately one Fe/(tufi). ( M , = 103,000). Thus, (a@). Fe may be the dominant structural unit of t,he enzyme, although the higher observed molecular weight and the various indications of molecular weight heterogeneity cited above suggest that there is a significant contribution from higher polymers. We also cannot discount t.he presence of apoenzyme which would decrease the apparent iron to protein ratio, but this possibility is made less likely by the consistent, half-integer ratio observed, and the high specific activity of the preparation.
Protocatechuate 4,5-dioxygenase prepared as described here contains approximately equal amounts of Fe" and Fe"'. The latter is largely EPR silent. In the "Miniprint" under "Specific Removal of Ferric Iron" we discuss in detail how the iron content of the enzyme was monitored by EPR and Miissbauer spectroscopies during the development of the purification procedure. Simultaneous evaluation of these spectra led to the conclusion that the majority of the Fe:" was bound adventitiously, probably in the form of EPR silent aggregates.
A method was developed to specifically remove this  listed the parameters obtained at various temperatures. The fits to the whole series of spectra revealed that the low energy absorption line was more intense by about 5-10% than the high energy line. This suggests that a minor high spin ferric impurity contributes to the absorption of the low energy line.
The Mossbauer parameters obtained here are quite different from those observed for Fez+ sites in enzymes. We noted, however, a striking similarity of our parameters with those reported for the reaction center of the photosynthetic bacterium R. sphaeroides R-26 (25,35). The parameters for AEQ are compared in Table 111. The isomer shift of the reaction center iron, = 1.17 mm/s (see Table I  In the absence of an applied magnetic field, no magnetic hyperfine interactions are observed for high spin Fez+ compounds. A strong applied field, however, can polarize the electronic system and sizable magnetic hyperfine interactions may be observed. For reduced Pseudomonas aeruginosa protocatechuate 3,4-dioxygenase such studies (26) have revealed the presence of internal fields up to 25 Tesla. We have studied the protocatechuate 4,5-dioxygenase sample in an external field of 6.0 Tesla. The spectrum was poorly resolved; it re-vealed, however, that the internal fields, at 4.2 K are not larger than about 1 Tesla. Moreover, the 6.0-Tesla spectrum of protocatechuate 4,5-dioxygenase was essentially the same as that observed for the R. sphueroides reaction center.
EPR studies of protocatechuate 4,5-dioxygenase frequently reveal a signal around g = 6 which disappears after addition of ascorbate (see Fig. 5 below). We have added ascorbate to a preparation of protocatechuate 4,5-dioxygenase which exhibited a g = 6 resonance. The Mossbauer spectra revealed that the decrease of the g = 6 signal intensity was accompanied by an increase of the intensity of the ferrous doublet discussed here suggesting that the g = 6 signal species results from oxidation of the AEQ = 2.23 mm/s component.    a Total iron estimated by colorimetric assay (-t5%). *Estimated by quantitation of EPR spectra (&lo%). E Calculated by subtracting EPR-quantitated Fe from total iron and normalizing to the maximum observed (the dEstimated from quantitation of the S = Yz signal in the EPR spectrum of the NO-treated sample (+IO%).

ascorbate-treated sample).
The resolution does not allow us to quantitate the two species to better than 5 5 % ; the fit displayed in Fig. 4 8 suggests that 60% of the total ferrous ion belongs to doublet I. The parameters obtained for site I1 suggest that substrate is bound to the Fez+ site. The parameters of site I are rather close to those observed for the native enzyme suggesting uncomplexed enzyme. Two observations, however, argue against this interpretation. First, a 3-fold further increase in the substrate concentration did not result in any spectral changes. Secondly, reaction of the enzyme substrate complex with NO yields two major S = 3/2 species, both of which are different from the species observed when native enzyme is reacted with NO (see Fig. 6B below). We have also studied the sample at 150 K; at this temperature, the quadrupole splittings are AE& = 2.10 mm/s and AEQ(II) = 2.65 mm/s. EPR Spectra of Protocatechuate 4,5-Dioxygenase-The relative concentration of ferric ion, which remains in preparations after Tiron treatment, is variable, but usually very low; the sample used for Mossbauer spectroscopy ( Fig. 4; EPR spectrum shown in Fig. 5A) contained only 0.06 Fe3+/(~/3)2 structural unit. All of this Fe3+ appears to be EPR active.
Since we have observed an approximate inverse correlation between the concentration of residual Fe3+ and specific activity, it seems very unlikely that ferric ion has a catalytic role. Nevertheless, the residual Fe"+ is tightly bound, and its reduction by mild reductants leads to activation of the enzyme, suggesting that at least a portion of the Fe3+ is the oxidized form of the active site ferrous ion. This interpretation would imply that the residual ferric ion concentration could be accounted for in terms of spontaneous oxidation of the Fez+ occurring, for example, during handling procedures, purification, or freeze-thaw cycles. EPR spectra of a preparation of protocatechuate 4,5-dioxygenase which had retained a relatively large amount of Fe3+ after treatment to remove adventitious iron are shown in Fig. 5 , B-G, and rough quantitations of the EPR active species of this sample are summarized in Table IV. Two types of iron are present (Fig. 5B): near axial type with g values near g = 6, and a rhombic type resonating near g = 4.3. Both sets of resonances are rather broad suggesting that the species are heterogeneous. Reduction of the sample by ascorbate (Fig. 5C) resulted in elimination of the EPR signal from the axial species as well as part of the signal from the rhombic species. Concurrently, the specific activity of the sample increased by 33%. Quantitation of the spectra suggested that the axial species accounted for approximately 16% of the total iron in the sample while the rhombic species accounted for approximately 25%, of which one third was reducible by ascorbate. In other samples, the concentration of the axial type ferric species ranged from 0 to about 16% while the rhombic species ranged from about 3 to 25% of the total. Ascorbate treatment failed to increase the specific activity of the samples with no g = 6 type ferric species, despite the fact that part of the rhombic species was always reduced. Anaerobic addition of PCA or any other substrate as well as most nonmetabolized substrate analogs caused the axial signal to disappear from the EPR spectrum and the rhombic signal to increase (Fig. 50). A shoulder is apparent on the low field side of the g = 4.3 resonance shown in Fig. 5 0 , suggesting that the increase in signal is due to a new, slightly less rhombic, species. Since the substrate complex of the ascorbate-reduced enzyme failed to show a similar increase in the g = 4.3 signal (Fig. 5E), it is likely that the intensity increase around g = 4.3 originated from the ascorbate-reducible iron with g values near g = 6 in Fig. 5B. This implies some sort of interaction of the axial ferric ion with substrate. The observation that the rhombic ferric species is only partially reduced by ascorbate suggests that it is composed of two or more subspecies. We cannot eliminate the possibility that oxidation of some of the ferrous ion also contributes to one of these specie^.^ The properties of the iron components observed in protocatechuate 4,5-dioxygenase are summarized in Table V. Enzyme-Nitric Oxide Complexes-During the past few years it has been shown that nitric oxide can complex with ferrous ion of several non-heme iron proteins (14,(29)(30)(31) to produce species with electronic spin S = 3/2. The identification of this spin state is firmly established by the observation of EPR resonances around g = 4 and 2. A similar EPR signal, with g values at g = 4.1, 3.9, and 2.0, has also been reported for an EDTA.Fe(I1) .NO complex (30). The observed EPR signals of the S = 3/2 complexes can be described with the spin Hamiltonian: H e = D[SZ -5/4 + E/D(SZ + S:)] + go@g.g (1) where go = 2 and D and E are parameters describing the zero field splitting. The NO complexes studied thus far have D > 3Broad EPR signals near g = 4.3 are commonly observed in biological preparations. Signals of this type are observed for iron in solutions containing buffer ions, iron nonspecifically associated with proteins, and iron chelated by catechols as well as from active site iron in many proteins. Since the g = 6 type iron is, by contrast, rare in non-heme proteins and seldom, if ever, observed for nonactive site iron, we believe that the g = 6 type iron is the principal form of the oxidized active site iron of our enzyme. Nevertheless, the g = 4.3 signal increases in intensity when the enzyme is oxidized and decreases when the enzyme is rereduced; thus, it is possible that some of the oxidized active site iron resonates near g = 4.3. Since full activity is not recovered after rereduction, however, this signal may originate from enzyme molecules damaged in some way by the oxidation process. Reacts with NO to give S = Yz species.

Iron Environment of Protocatechuate 4,5-Dioxygenase
>90% of the iron in fully active samples.
Fe", g = 6.4 and 5.5 Oxidized form of active site iron. and for a discussion of the Mossbauer spectra resulting from such a system, see Ref. 45. Fig. 5F shows the EPR spectrum which results when an anaerobic solution of protocatechuate 4,5-dioxygenase is exposed to NO gas. The observed g values at g = 4.1, 3.9, and 1.99 result from a species in an almost axial environment, E/ D = 0.016. Quantitation of the spectrum shows that the signal accounts for approximately 70% of the entire iron content of the sample. The EPR signals of the Fe3+ species are essentially unchanged by the presence of NO. Thus the S = 3/2 signal appears to be uniquely associated with the active site ferrous ion of the enzyme. This is supported by the observation that reduction of the sample with ascorbate either before or after introduction of NO causes an increase in the S = 3/2 signal (Fig. 5G). This increase is approximately proportional to the amount of ascorbate reducible Fe3+ in the original sample and proportional to the fractional increase in activity (Table IV).
Anaerobic addition of substrate either before or after exposure of the enzyme to NO results in formation of at least two S = 3/2 species (E/D = 0.031 and 0.065) as shown in Fig.  6B.4 Approximate quantitations of the species were obtained by estimation of the area of the low field resonances as described under "Materials and Methods." The spin quantitation accounts for approximately 50 and 85% of the total iron before and after ascorbate reduction, respectively; this correlates well with the Fez+ content of the sample (Table   IV) .
Other substrates and inhibitors of protocatechuate 4,5dioxygenase also elicit complex S = 3/2 EPR spectra (Fig. 6,   C-E). Each spectrum appears to consist of two or more components suggesting that all of the substrate and substrate analog complexes are comprised of multiple species. The E/ A minor species with E/D = 0.015 is usually also observed in this spectrum. This signal could represent the NO complex of native enzyme which had failed to bind substrate. However, attempts to eliminate this spectrum by subtracting that of the native NO complex were unsuccessful suggesting that there are a t least three distinct enzyme-substrate complexes with NO. D values for the complexes formed with metabolizable substrate analogs are nearly identical with those observed for the PCA complex (Fig. 6, C and D), but the distribution of the species varies significantly. In the case of other analogs, such as 3-amino-4-hydroxybenzoic acid (Fig. 6E), the E/D values of the spectra observed are all different from those observed in the PCA complex. In each case, however, the total S = 3/ 2 concentration is approximately equal to the total ferrous ion concentration of the sample.
Nitric oxide can be displaced from protocatechuate 4,5dioxygenase by removing excess NO under vacuum and then exposing the buffered sample to air (pumping alone will not remove the bound NO). The S = 3/2 EPR spectrum was eliminated by this procedure and the full catalytic activity was observed. Fig. 7 A shows a Mossbauer spectrum5 of protocatechuate 4,5-dioxygenase complexed with substrate and NO. In order to assess the similarities between the S = 3/2 species observed for the proteins and simple inorganic complexes, we have also studied the spectra of EDTA. Fe(I1). NO. A comparison of the spectra shown in Fig. 7, A and B, shows striking similarities between the two comdexes. This is the sample of Fig. 4 to which NO was added anaerobically. In the region around g = 4, the EPR spectrum was the same as that shown in Fig. 6B, i.e. a t least two S = 3/2 species are present. Since all of the species have quite similar electronic systems we do not expect to resolve these differences in the Mossbauer spectra.  The theoretical curves shown in Fig. 7 were generated by computer simulations from the Hamiltonian H = He + H , using the parameters listed in the legend to Fig. 7. The parameters quoted were determined in the same way as described elsewhere (46). The values for E / D and go are known from EPR. A. is determined from the overall magnetic splitting. The sign of A. and the magnitude of D are obtained from the 5.0-Tesla spectrum of Fig. 7C. V,, and V , and therefore AEQ and q are determined from the line positions of the spectra recorded in parallel (Fig. 7) and transverse (spectrum not shown) magnetic fields. A more detailed analysis revealed that the z axis of the electric field gradient tensor is tilted by about 15" relative to the z axis which defines the zero field splitting frame. Within the uncertainties the parameter set for the EDTA. Fe(I1). NO complex fits the spectrum of Fig. 7A also. (Since we have not studied the enzyme complex in a strong applied magnetic field, the Mossbauer data give no information on the magnitude of D. We know however, from EPR that D is comparable to that of the EDTA. Fe ( 11). NO complex.) Inactivation of Protocatechuate 4,5-Dioxygenase-Inactivation of protocatechuate 4,5-dioxygenase by treatment with H2Oz (Fig. 8A), FeCNZ-, or extended exposure to air results in EPR active species with major resonances near g = 6 and g = 4.3, similar to the minority sDecies observed in native freezing; B, sample from A was made anaerobic and exposed to NO gas; C, sample from A was treated with 10 mM ascorbate for 5 min; D, sample from C was made anaerobic and exposed to NO gas. The protein and measurement conditions were as described in the legend to Fig. 5. Scale factors are referenced to the spectra shown in Fig. 5. enzyme (Fig. 5B). As shown in Table IV, the total Fe3+ present after Hz02 treatment accounts for approximately 96% of the total iron in the sample. Under these conditions, the enzymatic activity and the EPR signal intensity of the S = 3/2 NO complex are correspondingly reduced. These results show that loss of activity occurs upon oxidation of the active site iron. Treatment of the inactivated enzyme with ascorbate restores approximately 50% of both the original activity and the S = 3/2 EPR signal from the NO complex (Fig. 80).
Aerobic incubation of protocatechuate 4,5-dioxygenase with 4-methylcatechol or one of several other nonmetabolized substrate analogs yields an inactive enzyme exhibiting an EPR spectrum with a large signal near g = 4.3, this signal accounts for approximately 90% of the iron in the sample. The EPR spectra taken after the addition of NO to this sample exhibited an S = 3/2 signal with only 10% of the intensity that is observed when active enzyme is reacted with NO. Furthermore, the activity of the inactivated sample was not restored by treatment with ascorbate or Fe2+. Reductants with a lower redox potential than ascorbate were also ineffective.

Fe2+
Active Site-In this study we have demonstrated that the active site iron of native protocatechuate 4,5-dioxygenase is high spin ferrous. The observed Mossbauer parameters are unique among enzymes containing Fez+. The Mossbauer spectra are, however, strikingly similar to those reported for the iron of the reaction center from R. sphaeroides R-26. The similarity between the two sites as expressed by the AEQ values (Table 111) is also apparent when the spectra obtained " in 6-Tesla applied fields are compared. The iron sites of both Iron Environment of Protocatechuute 4,5-Dioxygenase proteins exhibit unusually small internal magnetic fields at 4.2 K. The near identity of the Mossbauer spectra implies close structural similarities of the iron sites. Two recent extended x-ray fine structure studies (33,34) of the R. sphueroides reaction center suggest an octahedral iron environment consisting mainly of nitrogen and oxygen ligands. Bunker et al. (34) suggest that the data are best fitted by assuming that four histidines and two oxygenous ligands are coordinated to the iron. Such an environment is in accord with the isomer shifts observed here.
Since we have only recently arrived at metal pure preparations, we have not yet performed systematic studies of the redox properties of the protocatechuate 4,5-dioxygenase iron. Our preliminary data suggest that oxidation of the iron yields a high spin ferric species with EPR resonances at g = 6, a signal which is rarely observed (4) for non-heme iron. The latter species is rereduced readily by ascorbate suggesting a site with a high redox potential, a feature which protocatechuate 4,5-dioxygenase shares with the reaction center (33). It is likely that valuable clues about the reaction center iron can be obtained from further studies of protocatechuate 4,5dioxygenase. In particular, it would be interesting to learn whether the Fez+ of the reaction center reacts with NO.
The Mossbauer spectra obtained here are fundamentally different from those observed by Zabinski et al. (11). Their spectra did not reveal any trace of high spin ferrous material. We have elaborated in the "Miniprint" on the instability of the protein and on its affinity for binding iron in multiple sites. We have described in detail how we improved the preparations by closely monitoring all iron environments with both Mossbauer and EPR spectroscopy. Our present preparations have specific activities higher by at least a factor of 15 than those of Zabinski et al. (11); moreover, their reported reconstitution procedures do not work with our preparation. We have applied our preparative procedure to isolate protocatechuate 4,5-dioxygenase from the same strain of P. testosteroni used in the earlier study; the resulting enzyme displays identical EPR and Mossbauer spectra to those reported here. It appears to us, then, that the spectra of Zabinski et al. (11) reflect adventitiously bound iron, possibly similar to that designated above as "fast relaxing Fe3+." We must point out that although the iron observed in the earlier study (11) appears to be high spin ferric in character, it is entirely different from the g = 6 species which we have implicated here to be the oxidized form of the ferrous active site iron.
We have recently started a Mossbauer study of protocatechuate 2,3-dioxygenase isolated from Bacillus macerans (36). The material is not yet metal pure. However, approximately 40% of the total Fe in the sample contributes to a doublet with Mossbauer parameters similar to those obtained here for the ferrous site. Tatsuno et al. (8) have reported Mossbauer spectra of 57Fereconstituted catechol 2,3-dioxygenase from Pseudomonas arvilla. The Mossbauer spectra show a ferrous site with AEQ = 3.28 mm/s, i.e. an iron environment quite different from that observed here. Curiously, addition of a 10-fold excess of substrate did not affect the spectra. Since these authors have studied reconstituted material and since the values of AEQ and are close to those frequently observed for adventitiously bound iron, we grew P. arvilla on an 57Fe-enriched medium. Our values for AE, and 6Fe for the purified catechol 2,3-dioxygenase agree with those reported by Tatsuno et al. (8). The comparison of the Mossbauer parameters show that the iron environments of the P. testosteroni and P. arvilla enzymes differ significantly in spite of the similar catalyzed reactions and the mutual requirement for reduced iron.
Nitric Oxide Complexes-We have shown above that NO reacts with the ferrous iron of protocatechuate 4,5-dioxygenase and that the S = 3/2 EPR signals account quantitatively for this iron. Furthermore, the Mossbauer spectrum of Fig.  7 A reveals that all iron of the enzyme-substrate complex has reacted with NO. Addition of nitric oxide to Fez+ containing buffer did not elicit an S = 3/2 species. Thus, the nitric oxide reaction can be used for a quantitative assessment of the amount of ferrous ion in the sample, a welcome spectroscopic tool in the absence of a visible absorption. The Mossbauer spectra shown in Fig. 7 show that the NO complexes of Fe(I1) .EDTA and protocatechuate 4,5-dioxygenase-substrate are characterized by almost identical sets of fine structure and hyperfine structure parameters. This would suggest that both complexes are structurally quite similar. From this observation one may be inclined to deduce that the electronic structure of native protocatechuate 4,5-dioxygenase is also very similar to that of Fe(I1). EDTA. This conclusion, however, is not supported by the data. The 4.2 K Mossbauer spectra of Fe(II).EDTA, with AEQ = 3.43 mm/s and 6pe = 1.31 mm/s are very different from those observed for protocatechuate 4,5-dioxygenase; in particular, the 6.0-Tesla spectra of the two compounds (data not shown) differ substantially. Furthermore, the enzyme-substrate spectra are also quite unlike those of Fe(I1) .EDTA.
So far a proper description of the electronic structure of the S = 3/2 complexes has not emerged in the literature. The S = 3/2 state could reflect either a high spin 3d7 configuration, or an intermediate spin complex of ferric iron (with NO electron withdrawing), or a ferrous complex with NO with three unpaired electrons in appropriate molecular orbitals. It is noteworthy that the electronic Zeeman term of EDTA. Fe(I1) .NO is isotropic and that the magnetic hyperfine interaction seems to be isotropic as well (the Mossbauer data show that A, = A, to within a few per cent: A, cannot be determined to better than k 20%). The isomer shift, = 0.66 mm/s, argues against a ferric configuration. This conclusion is suggested by comparison of our data with those reported by Structure-We have found that P. testosteroni protocatechuate 4,5-dioxygenase has a more complex subunit structure, nominally (a,@,)Fe, than other extradiol dioxygenases. Both P. arvilla catechol 2,3-dioxygenase (23,39) and Pseudomonas ovalis 3,4-dihydroxyphenylacetate 2,3-dioxygenase (38) reportedly have a4 structures with 1-3 and 4-5 irons, respectively, depending on growth and purification conditions. The recently reported manganese containing 3,4-dihydroxyphenylacetate 2,3-dioxygenase from Bacillus brevis (51) also has an cy4 subunit structure. In contrast, the subunit structure of all well characterized catecholic intradiol dioxygenases is (aPFe),, or (a&Fe),, where n values of 1, 4, 5, 8, and 10 have been reported (40,42,52). Among the protocatechuate 3 4dioxygenases the 01 subunits are remarkably similar in terms of their physical properties and preliminary evidence suggests that the substrate-binding site is located on this subunit (5, 32). The CY subunit of protocatechuate 4,5-dioxygenase has approximately the same size as these subunits; however, more physical and structural studies need to be done before a meaningful comparison can be made. Within the experimental uncertainties the @ subunit is the same size as the single subunit of known extradiol dioxygenases. Since the amino acid sequences of catechol 2,3-dioxygenase (50) and protocatechuate 3,4-dioxygenase (47,48) have recently been reported, we will be able to search for regions of homology in each of Iron Environment of Protocatechuate 4,5-Dioxygenase the protocatechuate 4,5-dioxygenase subunits. The apparent differences in the iron sites of catechol 2,3-dioxygenase and our enzyme, together with the fact that the gene of the former enzyme is plasmid borne may indicate, however, that the enzymes have evolved separately for some time.
The overall physical size and specific activity of our enzyme are very similar to those reported for protocatechuate 4,5dioxygenase isolated from P. species (3). The principal distinguishing characteristic is the requirement for ethanol as a stabilizing agent for the latter enzyme. We do not find ethanol to be an effective stabilizer for the P. testosteroni enzyme.
The two enzymes are inactivated by aerobic storage, H202, metals such as Cu, and by substrate turnover as are most other known iron containing extradiol dioxygenases. The subunit structure of P. species protocatechuate 4,5-dioxygenase has not been reported.
Our work was facilitated greatly by the recognition that the enzyme is much more stable if prepared from bacteria which were well out of log phase growth at the time of harvest. This suggests that some permanent change in the structure of the enzyme occurs during the post log phase period. It is possible that a post-translational modification of some sort takes place, but our efforts to identify a modified amino acid have thus far been unsuccessful.
The iron site of the native enzyme is apparently quite homogeneous. In contrast, the anaerobically formed substrate complex is markedly heterogeneous. Similar heterogeneities have also been observed for the anaerobic PCA complex with protocatechuate 3,4-dioxygenase (4,41). Recent studies of the latter enzyme in the presence of transition state analogs suggest that only the initial, relatively weak complex is heterogeneous (41). A high affinity, homogeneous complex appears to form coincident with or after Oz binding. A similar mechanism would also be consistent with the properties of the extradiol protocatechuate 4,5-dioxygenase observed here.
Inactivation by Oxidation-The assignment of Fe2+ as the active site iron is supported by the observation that two fundamentally different inactivation processes lead ultimately to oxidation of the iron. EPR spectra of protocatechuate 4,5dioxygenase recorded after treatment of the enzyme with HZOz or Fe(CN)i-clearly show that the Fez+ has become oxidized as suggested by numerous less direct techniques for other extradiol dioxygenases similarly inactivated. The oxidized iron gives broad EPR features in spectral regions characteristic of axial and rhombic environments showing that numerous species have been generated. Nevertheless, the iron appears to remain bound to the enzyme since we have not observed g = 6 type signals in Fe/substrate/buffer solutions in the absence of enzyme. This type of inactivation, whether induced by oxidants or occurring spontaneously during purification, is reversible a t least in part by the addition of mild reductants such as ascorbate. The inability of ascorbate to effect complete reactivation may indicate that the inactivation is progressive in nature and is only initiated by the oxidation. The second type of inactivation results when specific substrate analogs, such as 4-methylcatechol, interact with the enzyme under aerobic conditions. The g = 4.3 type EPR spectrum of this form shows no evidence for proteinbound iron and is identical in shape to the spectrum of free Fe-catechol complex. Since 57Fez+ will exchange under anaerobic conditions with the active site iron, the iron dissociation constant must be relatively high compared to those of hemes and other iron containing proteins. Thus, it is possible that the inactivating inhibitors, which are good iron chelators, shift the equilibrium significantly in favor of iron release. We have observed that the free Fez+-inhibitor complexes oxidize rapidly under aerobic conditions. Since the resulting com-plexes have very low redox potentials and low dissociation constants, the inactivaticjn process would be difficult to reverse. However, we have also observed that the inactivation proceeds much more rapidly with 4-methylcatechol than with other analogs which are equally effective as chelators. This suggests that the destabilization is initiated by binding of the inhibitor rather than dissociation of the iron.
Mechanistic Implications-Past studies have provided evidence that the presence of Fez+ is essential for maintaining or restoring the activity of most extradiol catecholic dioxygenases. The present study shows that active protocatechuate 4,5-dioxygenase contains an Fe2+ site and that this site has a unique environment, distinct from that of P. arvilla catechol 2,3-dioxygenase and other enzymes containing ferrous ion. The presence of Fe2+ in the extradiol cleaving enzymes provides a clear contrast to intradiol dioxygenases which perform an analogous reaction on the same substrate, but utilize exclusively Fe3+ throughout the catalytic cycle (4, 5). Our studies have shown that the iron of protocatechuate 4,5dioxygenase remains ferrous in the substrate complex prepared anaerobically. Moreover, our data show that small molecules such as NO have access to the ferrous ion even when substrate is bound. We thus suspect that the metal ion is accessible to O2 as well, and a reaction mechanism in which Oz is activated by binding directly to the Fez+ should be considered. On the other hand, we have observed no evidence for such an oxy complex. Moreover, the characteristic reactions and components of other systems which activate oxygen by direct interaction with the metal, such as rapid autoxidation and the participation of low redox potential centers (49), appear to be absent.