Radical S-Adenosylmethionine Enzyme Coproporphyrinogen III Oxidase HemN

The S-adenosylmethionine (AdoMet) radical enzyme oxygen-independent coproporphyrinogen III oxidase HemN catalyzes the oxidative decarboxylation of coproporphyrinogen III to protoporphyrinogen IX during bacterial heme biosynthesis. The recently solved crystal structure of Escherichia coli HemN revealed the presence of an unusually coordinated iron-sulfur cluster and two molecules of AdoMet. EPR spectroscopy of the reduced iron-sulfur center in anaerobically purified HemN in the absence of AdoMet has revealed a [4Fe-4S]1+ cluster in two slightly different conformations. Mössbauer spectroscopy of anaerobically purified HemN has identified a predominantly [4Fe-4S]2+ cluster in which only three iron atoms were coordinated by cysteine residues (isomer shift of δ = 0.43 (1) mm/s). The fourth non-cysteine-ligated iron exhibited a δ = 0.57 (3) mm/s, which shifted to a δ = 0.68 (3) mm/s upon addition of AdoMet. Substrate binding by HemN did not alter AdoMet coordination to the cluster. Multiple rounds of AdoMet cleavage with the formation of the reaction product methionine indicated AdoMet consumption during catalysis and identified AdoMet as a co-substrate for HemN catalysis. AdoMet cleavage was found to be dependent on the presence of the substrate coproporphyrinogen III. Two molecules of AdoMet were cleaved during one catalytic cycle for the formation of one molecule of protoporphyrinogen IX. Finally, the binding site for the unusual second, non iron-sulfur cluster coordinating AdoMet molecule (AdoMet2) was targeted using site-directed mutagenesis. All AdoMet2 binding site mutants still contained an iron-sulfur cluster and most still exhibited AdoMet cleavage, albeit reduced compared with the wild-type enzyme. However, all mutants lost their overall catalytic ability indicating a functional role for AdoMet2 in HemN catalysis. The reported significant correlation of structural and functional biophysical and biochemical data identifies HemN as a useful model system for the elucidation of general AdoMet radical enzyme features.

The biosynthesis of hemes and chlorophylls requires the conversion of coproporphyrinogen III to protoporphyrinogen IX. During this reaction, the propionate side chains on rings A and B of the tetrapyrrole are converted to the corresponding vinyl groups (Scheme 1a) (1)(2)(3). In the absence of molecular oxygen the oxidative decarboxylation of coproporphyrinogen III is catalyzed by the oxygen-independent coproporphyrinogen III oxidase HemN. Escherichia coli HemN has been biochemically characterized (4), and the crystal structure of the HemN protein was recently determined (5).
HemN is a monomeric, iron-sulfur cluster containing protein, which belongs to the AdoMet 1 radical protein family (6). All members of this large enzyme class contain a [4Fe-4S] cluster coordinated by cysteine residues residing in a conserved CXXXCXXC sequence motif. Furthermore, all AdoMet radical proteins require S-adenosylmethionine as a cofactor to initiate radical-based catalysis. The iron-sulfur clusters of several members of the family have been characterized by electron paramagnetic resonance (EPR), resonance Raman, and Mössbauer spectroscopy (7)(8)(9)(10)(11)(12)(13)(14)(15)(16). For pyruvate formate-lyase-activating enzyme (PFL-AE) and biotin synthase (BioB), it was shown using Mössbauer spectroscopy that the [4Fe-4S] cluster contains one iron atom, which is not ligated by a cysteine residue, and that AdoMet binds to this special iron site of the cluster (17,18). Later, AdoMet binding was studied in detail by EN-DOR spectroscopy for PFL-AE and lysine 2,3-aminomutase (LAM) (19 -21). The following initial reaction steps are common for all AdoMet radical enzymes (Scheme 1b) (22)(23)(24). First the iron-sulfur cluster is reduced. For some members of the family, it was shown that this reduction occurs via a reduced flavodoxin in E. coli (25,26). In the next step the reduced iron-sulfur cluster transfers an electron to AdoMet, which subsequently is cleaved to methionine and a reactive 5Ј-deoxyadenosyl radical (27)(28)(29)(30). Then the 5Ј-deoxyadenosyl radical abstracts a hydrogen atom from an appropriately positioned carbon atom creating either a substrate radical (31)(32)(33) or a catalytic glycyl radical on a partner protein (34). In some AdoMet radical reactions, AdoMet is consumed and acts as a co-substrate with release of methionine and 5Ј-deoxyadenosine (34,35). In others, AdoMet is restored and acts as a true cofactor (22,31).
Based on the current knowledge about AdoMet radical enzymes a potential HemN mechanism, shown in Scheme 1c, was proposed (4). The solved HemN crystal structure provided insights into the corresponding enzyme and cofactor architecture. Nevertheless, a crystal structure represents only a single static, sometimes even artificial, view of a protein. Therefore, several functional questions concerning the proposed HemN mechanism remained to be answered.
The redox reaction performed by the [4Fe-4S] cluster is the crucial initial step of radical generation. The HemN crystal structure revealed the presence of a [4Fe-4S] cluster ligated by the three cysteine residues of the conserved CXXXCXXC amino acid motif and by carboxyl and amino groups of the methionine part of AdoMet. We have investigated the properties of the iron-sulfur cluster in this protein in solution by EPR and Mössbauer spectroscopy. These are very sensitive techniques for the observation of structurally different forms of the cluster in solution. The characteristics of the cluster in the [4Fe-4S] 2ϩ state were determined using Mössbauer spectroscopy. There, the influence of AdoMet and the substrate on the coordination of the cluster were of interest.
A central step in the proposed mechanism for HemN catalysis is the homolytic cleavage of AdoMet and radical formation, which initiates the decarboxylation of the propionate side chains of the substrate coproporphyrinogen III. In the case of HemN two consecutive decarboxylations are catalyzed to form the reaction product protoporphyrinogen IX. The crystal structure revealed the presence of two bound AdoMet molecules. Therefore, AdoMet cleavage was analyzed for its stoichiometry, turnover number, and possible AdoMet regeneration, and the influence of the substrate on AdoMet cleavage was investigated. Finally, the function of the second AdoMet molecule (AdoMet2), which is unique to the solved HemN crystal structure, was investigated via site-directed mutagenesis of the involved conserved amino acid residues. Mutant proteins were tested for their overall structural integrity and iron-sulfur cluster stability, AdoMet cleavage, and catalytic activity. Obtained results allow for a deeper and exemplary understanding of the structure-function-relationships for this AdoMet radical enzyme.
Preparation of 57 FeCl 3 -22 mg of metallic 57 Fe were dissolved in 377 l of HCl (37%, w/v) at 80°C. The obtained 57 FeCl 3 solution was then added dropwise to 17 ml of 600 mM nitrilotriacetic acid, pH 5.2, under continuous stirring. During this procedure, the pH was held between 3 and 5 by adding 2 M NaOH. The resulting 17.5-ml solution was added to 10 liters of culture medium.
Purification of E. coli HemN-Recombinant E. coli HemN wild-type and mutant proteins were purified anaerobically as previously described (4). Some of the mutant HemN enzymes (Y56A, Y56L) were purified using a modified affinity chromatography protocol. Glycerol in a concentration of 10% (v/v) increased their affinity to the material in a batch purification approach. For Mössbauer spectroscopy 57 Fe was incorporated into HemN by growing E. coli BL21(DE3) carrying pET3ahemN on minimal SMM-medium (37) containing 31 M 57 Fe.
Purification of Thermosynechococcus elongatus Protoporphyrinogen IX Oxidase-Recombinant T. elongatus protoporphyrinogen IX oxidase was produced as a His-Tag fusion protein in E. coli BL21-Codon-Plus(DE3)-RIL cells carrying vector pET-32ahemY. Protein production was induced at A 578 of 0.6 with isopropyl-1-thio-␤-D-galactopyranoside (final concentration 500 M). After induction, cells were grown overnight at 25°C. Purification was performed using the MagneHis TM Protein Purification System (Promega) according to the manufacturer's instructions. After purification, the protein was dialyzed against 20 mM Tris, pH 8.0, 50 mM NaCl, and 0.1% (v/v) Triton X-100 to remove imidazole. SCHEME 1. The oxidative decarboxylation of coproporphyrinogen III to protoporphyrinogen IX (a) and the proposed reaction mechanism of AdoMet radical enzymes (b) and HemN (c). a, HemN catalyzed reaction. b, reaction steps common to all AdoMet radical enzymes including reduction of the iron-sulfur cluster, AdoMet cleavage with formation of a 5Ј-deoxyadenosyl radical and H-atom abstraction from the substrate (R-H). c, in the HemN reaction the H-atom is proposed to be abstracted from the propionate side chain of the substrate. Release of CO 2 and uptake of the remaining electron by an external electron acceptor (still unknown) lead to product formation.
Determination of Protein Concentration-The BCA (bicinchoninic acid) protein assay kit (Sigma), and the Bio-Rad protein assay were used according to the manufacturer's instructions.
Iron Determination-The iron content of recombinant, purified E. coli HemN was determined colorimetrically with o-phenanthroline as previously described (4). UV-visible light absorption spectroscopy for the detection of the [4Fe-4S] cluster was also carried out as described before (4).
Coproporphyrinogen III Oxidase Activity Assay Using Recombinant HemN-The oxygen-independent coproporphyrinogen III oxidase activity assay for wild-type and mutant HemN proteins was performed as described before (4). For the investigation of AdoMet cleavage during HemN catalysis and determination of the methionine/protoporphyrinogen IX ratio, the reaction was performed with slight modifications. [ 14 C] AdoMet (specific activity of 53.6 mCi/1.98 GBq per mmol) was used instead of non-labeled AdoMet. The concentrations in the assay mixtures were 1.5 M HemN, 160 g of E. coli cell-free extract (S250), 20 -30 M [ 14 C]AdoMet, 500 M NADH, and 20 M coproporphyrinogen III in a total volume of 300 l of 50 mM Tris, pH 7.0, 3 mM DTT, 300 mM NaCl, and 0.3% (v/v) Triton X-100. The mixtures were incubated anaerobically at 37°C in the dark. After 15, 30, 60, and 90 min of incubation, 50-l samples were taken from the assay mixture for the determination of protoporphyrinogen IX and methionine formation. Enzymatically formed protoporphyrinogen IX was oxidized by addition of T. elongatus protoporphyrinogen IX oxidase (final concentration 3 M), and the amount of protoporphyrinogen IX was determined by fluorimetric detection of its oxidized form protoporphyrin IX using a PE LS50B luminescence spectrometer (PerkinElmer Life Sciences) as described before (4). For the determination of methionine formation, the proteins in the assay mixture were precipitated by addition of 10% (w/v) PCA and removed by centrifugation. Methionine formation was determined by HPLC analysis of the supernatant as described below.
Assay for AdoMet Cleavage-HemN-mediated AdoMet cleavage was studied in the presence and absence of the substrate coproporphyrinogen III. For this purpose the assay was performed as described above in the presence and absence of the substrate. Alternatively, the assay was performed without E. coli cell-free extract. In this case sodium dithionite served as the reductant for the iron-sulfur cluster of HemN, and the concentrations were 40 M HemN, 2 mM sodium dithionite, 80 M [ 14 C]AdoMet, and 140 M coproporphyrinogen III in 50 mM Tris, pH 7.0, 3 mM DTT, 300 mM NaCl, and 0.3% (v/v) Triton X-100. The mixtures were incubated at 18°C for 3 h in the dark. For the investigation of AdoMet cleavage by the mutant proteins, the assay was carried out using [ 14 C]AdoMet with the following assay composition: 10 M HemN (wild-type or mutant HemN), 80 M [ 14 C]AdoMet, 2 mM dithionite, 100 M coproporphyrinogen III in a total volume of 25 l of 50 mM Tris, pH 7.0, 3 mM DTT, 300 mM NaCl, and 0.3% (v/v) Triton X-100. All mixtures for testing mutant proteins were incubated anaerobically at 37°C in the dark for 90 min. After incubation, the reaction was stopped by addition of 10% (w/v) PCA, and precipitated proteins were removed by centrifugation. Formation of [ 14 C]methionine was determined by HPLC analysis of the supernatant.
HPLC Analysis-The samples for HPLC analysis were prepared as described above. After removal of the precipitated proteins the solution was filtered through a cellulose acetate membrane syringe filter (Nalge Nunc International, Rochester, NY). 20 l of the sample were loaded onto a 4.6 ϫ 250-mm ODS Hypersil-C 18 reversed phase column (Techlab GmbH, Erkerode, Germany) with a pore width of 120 Å. Separation was performed at 38°C at a flow rate of 500 l/min using 50 mM NH 4 H 2 PO 4 (pH 2.5) as mobile phase. [ 14 C]AdoMet and [ 14 C]methionine were detected by measurement of radioactivity using a flowthrough scintillation counter (Raytest Isotopenmessgerä te GmbH, Straubenhardt, Germany) and in parallel by photometric diode array analysis at 200 -650 nm. For the quantification of the enzymatically formed methionine, 250-l fractions were collected, and amounts of [ 14 C]AdoMet and [ 14 C]methionine were determined using a Liquid Scintillation Analyzer Tri-Carb 2900 TR (PerkinElmer Life Sciences). The scintillation mixture employed was OptiPhase HiSafe 2 (PerkinElmer Life Sciences). The HPLC system employed was Jasco 1500 (Jasco, Gross-Umstadt, Germany). [ 14 C]AdoMet and L-methionine (Sigma) were used as standards.
Preparation of EPR Samples-Samples for EPR were prepared under strict anaerobic conditions and contained 160 M HemN, 10 mM sodium oxalate, and 50 M 5-deazaflavin in a total volume of 250 l of 50 mM Tris, pH 8.0, 3 mM DTT, and 300 mM NaCl. For reduction of the iron-sulfur cluster the samples were illuminated for 1 h using a commercial slide projector. The samples were frozen in liquid nitrogen directly after illumination.
EPR Spectroscopy-EPR spectra were obtained at X-band using a Bruker ELEXSYS E500 spectrometer, equipped with an Oxford Instruments ESR900 liquid helium cryostat. Acquisition conditions are given in the figure legend. Spin intensities were quantitated by double integration and calibrated against a sample of known concentration of copper (2ϩ) sulfate/EDTA.

Preparation of Samples for Mössbauer Spectroscopy-For
Mössbauer spectroscopy the final HemN concentration was 760 M. Sample preparation was performed under strict anaerobic conditions. For the sample containing HemN plus AdoMet a 10-fold molar excess of AdoMet (7.5 mM) over HemN (760 M) was added, and the mixture was incubated for 2 h at 18°C. In the sample of HemN plus AdoMet and substrate, the concentrations were 570 M HemN, 5 mM AdoMet, and 500 M coproporphyrinogen III. This sample was incubated for 2 h at 18°C. The solutions were transferred to 400-l Mössbauer cups and frozen in liquid nitrogen.
Mössbauer Spectroscopy-Mössbauer spectra were recorded using a spectrometer in the constant acceleration mode. Isomer shifts are given relative to ␣-Fe at room temperature. The spectra obtained at 20 mT were measured in a He-bath cryostat (Oxford MD 306) equipped with a pair of permanent magnets. These spectra were analyzed by leastsquare fits using Lorentzian line shape. For the high-field spectra (4 T) a cryostat equipped with a superconducting magnet was used (Oxford Instruments, Spectromag 4000). These spectra were simulated by the nuclear Hamiltonian in Equation 1.
Here I denotes the nuclear spin quantum number, Q the nuclear quadrupole moment of the excited nuclear state, V zz the z-component of the electric-field gradient (efg) tensor and ϭ (V xx Ϫ V yy )/V zz the asymmetry parameter of the efg, and g N the nuclear g-factor. Circular Dichroism (CD) Analysis of Wild-type and Mutant HemN-For CD measurements, protein solutions were dialyzed against 50 mM sodium phosphate buffer, pH 7.6, and 10 mM NaCl. CD spectra of protein samples (0.125 mg/ml) in quartz cuvettes of 1-mm path length were recorded as an average of ten scans with a Jasco J810 spectropolarimeter over a range of 200 -250 nm on a millidegree ellipticity scale. Dialysis and measurements were carried out under anaerobic conditions.

RESULTS AND DISCUSSION
The reduction and oxidation of the iron-sulfur cluster is the initial critical step of HemN catalysis. Previously, the cluster of recombinant, anaerobically purified E. coli HemN was spectroscopically identified using UV-visible light absorption spectroscopy (4). In the solved HemN crystal structure one view of the coordination chemistry of the cluster was obtained (5). However, questions regarding the properties of the reduced cluster in the [4Fe-4S] 1ϩ state remained. Moreover, the functional change in the coordination sphere of the cluster in response to the binding of AdoMet and the substrate were of interest.
To answer these questions and further characterize the detailed functional nature of the iron-sulfur cluster of E. coli HemN in solution, we performed EPR and Mössbauer spectroscopic investigations. Such spectroscopic investigations in combination with the data obtained from the x-ray crystallographic study (5) allow a very precise description of the electronic and structural properties of the iron-sulfur cluster of HemN.
EPR Spectrum of the HemN Iron-Sulfur Cluster Reduced in the Absence of S-Adenosylmethionine-HemN (2 mol of iron/ mol HemN) at a concentration of 160 M was reduced with oxalate/deazaflavin and light as described in detail under "Experimental Procedures" and frozen in liquid nitrogen. The EPR spectrum of the reduced (oxalate/deazaflavin) enzyme in the absence of S-adenosylmethionine produces an EPR spectrum (Fig. 1, trace a) at 15 K typical of a [4Fe-4S] center with g values g 1 ϭ 2.06, g 2 ϭ 1.94, g 3 ϭ 1.89. Quantitation of the number of spins (see "Experimental Procedures") indicates that they are not stoichiometric given the concentration of the enzyme, with ϳ20% of the iron-sulfur centers in the enzyme contributing to the signal observed. This results from a combi-nation of incomplete reduction of the iron-sulfur center, and incomplete incorporation of the cluster into all HemN molecules during overexpression of the protein (4). As reported previously (4), attempts to reconstitute the cluster in vitro did not yield a higher iron content but resulted in precipitation of black iron sulfide. Therefore, HemN was used as isolated containing 2 mol of iron/mol HemN for all experiments, assuming, after the results of EPR and Mössbauer studies, that half of the protein population was in the apo form and the other half in the holo form carrying a [4Fe-4S] cluster.
The EPR spectrum of the iron-sulfur center at 15 K also appears heterogeneous, in that the linewidth is quite large and the lineshape is non-gaussian (i.e. the g ϭ 1.89 region where it appears quite broad and appears to contain two overlapping components). Measuring the spectrum at 6.8 K (Fig. 1, trace b) resolves two components in the spectrum. This indicates that the iron-sulfur center exists in two different forms with slightly different g-values. This shift in g-value of the iron-sulfur center must arise from a change in the structure of the cluster rather than a change in the binding pocket, although it could be caused by a change in bond length of as little as 0.1 Å. Because the iron-sulfur center spectrum is obtained in the absence of AdoMet, it is possible that the two different structures result from two different ligands binding to the uncoordinated iron, i.e. water in one case and an amino acid residue in the other.
Although EPR has thus proved to be a sensitive probe of the iron-sulfur center, its environment and ligands, unfortunately attempts to obtain EPR spectra of HemN with bound AdoMet failed. We were unable to reduce the cluster in the presence of the cofactor. Moreover, we did not detect an EPR signal when AdoMet was added to the reduced cluster. Therefore, to study the influence of AdoMet binding to the cluster on its electronic properties under functional conditions, i.e. in solution, we performed Mössbauer spectroscopy.
Mössbauer Spectra of HemN-For Mössbauer measurements 57 Fe was incorporated into the protein during cell growth and protein production. The iron content of the purified protein was 2 mol of iron/mol of HemN (see above). Mössbauer spectra were recorded for samples containing HemN, HemN plus AdoMet, and HemN plus AdoMet and coproporphyrinogen III. The Mössbauer spectra of solely HemN without further additions are shown in Fig. 2A (2) mm/s). The ligand of the fourth iron site is unknown in the absence of AdoMet. Possible candidates could be either an amino acid residue of the polypeptide other than cysteine or alternatively any molecule contained within the buffer solution (see EPR spectra above). The results obtained from the HemN spectra are consistent with the previous finding that only the three cysteine residues Cys 62 , Cys 66 , Cys 69 within the CXXXCXXC motif of E. coli HemN are involved in iron-sulfur cluster coordination (4). The results are also in agreement with the results of biochemical and biophysical investigations of other AdoMet radical enzymes (13,15) and with the recently obtained x-ray crystallographic data for HemN (5), biotin synthase (39), and MoaA (40).
Mössbauer Spectra of HemN Plus AdoMet- Fig. 2B shows the Mössbauer spectrum of HemN plus AdoMet (parameters in Table I; see also Supplementary Materials). As in the case of the HemN spectrum without additions two different iron sites of a [4Fe-4S] 2ϩ cluster were identified for the HemN spectrum in the presence of AdoMet. However, for the latter spectrum, a change in the isomer shift of the non-cysteine-ligated iron atom (from ␦ 2 ϭ 0.57 (3) mm/s to ␦ 2 ϭ 0.68 (3) mm/s) was observed, which indicates that AdoMet only binds at one iron of the cluster, namely this unique, non-cysteine-ligated iron atom. AdoMet binding to this unique iron site of the [4Fe-4S] cluster was previously observed by Mössbauer studies of PFL-AE (17) and biotin synthase (18). Furthermore, our results from the Mössbauer spectrum of HemN in the presence of AdoMet are consistent with the observed coordination of AdoMet to the iron-sulfur cluster in the crystal structure of E. coli HemN (5) in which it was observed that AdoMet binds at the iron by means of the carboxyl and the amino group of the methionine part of AdoMet. Thus AdoMet coordination to the fourth iron site of the cluster is analogous in solution and in the crystalline state of HemN.
Mössbauer Spectra of HemN Plus AdoMet and Coproporphyrinogen III-The spectrum of HemN in the presence of AdoMet and substrate basically resembles the pattern of the spectrum of HemN plus AdoMet (see Supplementary Materials). The fit parameters for these spectra are, within error margins, identical with those obtained for HemN plus AdoMet. These Mössbauer results have thus shown for the first time that the substrate does not interact directly with the iron-sulfur cluster, and that no change in the coordination of AdoMet to the cluster occurs upon the presence of the substrate, at least as long as the cluster is in the [4Fe-4S] 2ϩ state. Matters may change upon reduction of the cluster into its active [4Fe-4S] 1ϩ state. One might argue that there is no direct evidence that the substrate is bound by the enzyme at all. However, the fact that under almost identical assay conditions a positive influence of the substrate coproporphyrinogen III on the cleavage of AdoMet was demonstrated (see below) argues for the observed results.
Conclusions from Mössbauer Spectra and Crystallographic Studies of HemN Plus AdoMet-The data obtained from the Mössbauer and the x-ray crystallographic studies of the HemN iron-sulfur cluster in the presence of AdoMet were compared. The isomer shift value for the AdoMet-ligated iron site can be directly correlated with specified distances between the iron and the ligand atoms in the case of HemN. Thus, the observed isomer shift of ␦ 2 ϭ 0.68 (3) mm/s is associated with the iron site, which is ligated by the O-(2.3 Å) and the N-ligand (2.6 Å) of the methionine part of AdoMet as observed in the crystal structure of HemN (Fig. 4A). Therefore, although the observa-tion that AdoMet binds to the iron-sulfur cluster is not new, our Mössbauer results provide essential information in correlation with the structural data, namely the possibility to combine the structural features (distances between the atoms) with the electronic properties (Mössbauer parameters) of the [4Fe-4S] cluster of HemN.
AdoMet Is Cleaved During the Oxygen-independent Coproporphyrinogen III Oxidase Reaction-The outlined Mössbauer studies in combination with the solved crystal structure of HemN revealed the coordination of AdoMet to the iron-sulfur cluster. The proposed enzyme mechanism for HemN catalysis suggests an electron transfer from the cluster to AdoMet. This results in the homolytic cleavage of AdoMet with the formation of methionine and a putative 5Ј-deoxyadenosyl radical (Scheme 1b) (4). For other members of the AdoMet radical family, AdoMet cleavage was shown before (29,34,(41)(42)(43)(44). For HemN we previously only demonstrated the absolute requirement of AdoMet for catalysis (4). To unambiguously demonstrate that AdoMet is enzymatically cleaved by HemN, we analyzed the reaction products of a HemN assay in the presence of S-adeno-  Table I. (Fig. 2) ͓4Fe-4s͔ 2ϩ Cys ligand (solid line 1) syl-L-[methyl-14 C]methionine via HPLC. A control reaction mixture containing all reagents except for HemN and the E. coli cell-free extract exhibited only one radioactive peak characteristic of the chromatographic behavior of AdoMet with a retention time of about 7 min (Fig. 3A, top). The HPLC analysis of an activity assay mixture containing HemN and E. coli cell-free extract demonstrates the almost complete conversion of [ 14 C]AdoMet into [ 14 C]methionine with a retention time of 11 min (Fig. 3A, bottom). Another control reaction was performed in which all reaction components were present except for recombinant, purified HemN. HPLC analysis of this mixture showed that a small amount of AdoMet is cleaved with minor methionine formation catalyzed by HemN and other enzymes within the E. coli cell-free extract. For this mixture we also observed a second unidentified reaction product (retention time, 18 min), which is also caused by the action of enzymes of the cell-free extract (Fig. 3A, middle). Clearly, only in the presence of purified HemN was AdoMet completely cleaved and a single peak at a retention time of about 11 min representing methionine was then detected (Fig. 3A, bottom). These results clearly show that AdoMet is cleaved during HemN catalysis and further support the predicted reaction mechanism. AdoMet Cleavage Is Dependent on the Presence of the Substrate Coproporphyrinogen III-For some of the AdoMet radical enzymes it was observed that AdoMet cleavage occurs only in the presence of their respective substrates (30,33,34,41,45). To test HemN for this potential behavior we performed assays for AdoMet cleavage using different reduction systems for the iron-sulfur cluster of HemN in the presence and absence of the substrate coproporphyrinogen III. The results of these experiments are summarized in Table II. For the reaction mixtures in which the physiological electron donor system (components of the E. coli cell-free extract) was employed, AdoMet cleavage was observed only in the presence of the substrate. Mixtures containing all components including HemN in which the substrate coproporphyrinogen III was absent showed no AdoMet cleavage.

TABLE I Mössbauer parameters obtained from the fit and simulation of HemN spectra in the absence or presence of AdoMet
Alternatively, we used sodium dithionite as a reductant for the iron-sulfur cluster of HemN. Using this artificial electron donor system we observed that AdoMet was cleaved to a low extent even in the absence of the substrate. This low AdoMet cleavage activity of HemN in the absence of the substrate is probably caused by the strong reducing power of dithionite and does not represent the physiological situation. Moreover, AdoMet cleavage was significantly stimulated for this artificial system by the presence of coproporphyrinogen III. A comparable behavior was observed for the AdoMet radical enzymes anaerobic ribonucleotide reductase-activating enzyme (ARNR-AE) and BioB (26,29,41,42). These enzymes also cleaved AdoMet in the absence of their substrates only if strong artificial electron donor systems like deazaflavin/light or dithionite were employed for the reduction of the corresponding ironsulfur cluster. However, when the physiological electron donor systems (flavodoxin-reductase/flavodoxin) were used, AdoMet cleavage only occurred in the presence of the corresponding substrates. For PFL-AE and LAM it was observed that AdoMet cleavage is dependent on the presence of their substrates using the artificial and the physiological electron donor systems (33,  b For the amounts of employed coproporphyrinogen III see "Experimental Procedures." c The AdoMet cleavage activity of HemN in the presence of coproporphyrinogen III was set to 100%, and the activities in the absence of the substrate were related to that. This was done independently for the different assay systems (cell free extract/sodium dithionite).
d The turnover was estimated assuming that 50% of the employed HemN was in its active form (see text). The amount of formed protoporphyrin IX was related to the amount of employed HemN in the active form.
e Not detectable. f The assay mixture with dithionite as a reductant for the iron-sulfur cluster of HemN contained no final electron acceptor. Therefore, there was no protoporphyrinogen IX formation. 34). It seems to be a common feature of proteins within the AdoMet radical protein family that AdoMet cleavage is triggered by the presence of their respective substrates. An explanation for this behavior is provided by the assumption that the thermodynamically unfavorable AdoMet cleavage has to be coupled to the more favorable reactions of H-atom abstraction and the resulting radical formation (26).
AdoMet Is Consumed During HemN Catalysis with Cleavage of Two AdoMet Molecules During Formation of One Protoporphyrinogen IX-The role of AdoMet in the different catalyses performed by the various AdoMet radical enzymes is either that of a true cofactor or that of a co-substrate. Our proposed mechanism for the HemN reaction implies that AdoMet is consumed during catalysis and therefore acts as a co-substrate. In order to test this hypothesis we performed enzyme assays with the subsequent quantitation of the formed protoporphyrinogen IX and methionine. We observed that HemN incubated with E. coli cell-free extract and an excess of non-labeled AdoMet performed several catalytic turnovers (Table II). Similar results were obtained with [ 14 C]AdoMet. The turnover was estimated under the assumption that only 50% of the HemN molecules contained an intact [4Fe-4S] cluster, which was deduced from the Mössbauer data (89% [4Fe-4S] 2ϩ ) and iron analysis (2 mol iron/mol HemN). Protoporphyrinogen IX formation was followed fluorimetrically and methionine formation was followed by radiochemical analysis after HPLC separation of [ 14 C]AdoMet and [ 14 C]methionine as described under "Experimental Procedures." We observed that protoporphyrinogen IX formation and methionine formation both increased in parallel over several catalytic turnovers. Moreover, as shown in Fig. 3B, quantitation of the amounts of formed methionine and protoporphyrinogen IX after 15, 30, 60, and 90 min of incubation revealed an average methionine/protoporphyrinogen IX ratio of 1.9:1 for each time point. The values are averages of three independent measurements with individual ratios ranging between 1.7 and 2.3. These results clearly establish that AdoMet is consumed during HemN catalysis and that two AdoMet molecules are cleaved during the formation of one molecule protoporphyrinogen IX. This stoichiometry makes perfect sense with respect to the requirement for the decarboxylation of two propionate side chains of the substrate coproporphyrinogen III. The need for two AdoMet molecules to be cleaved is also in good agreement with the observation of two AdoMet molecules bound in the crystal structure of HemN (5). One of these AdoMet molecules (AdoMet1) functions as the ligand to the fourth iron atom of the [4Fe-4S] cluster (Ref. 5 and Fig.4) as found for all AdoMet radical enzymes studied so far. The second AdoMet molecule (AdoMet2) observed in the HemN structure is not involved in iron-sulfur cluster coordination. It was proposed that it was functional because of its almost perfect coordination provided by the surrounding conserved amino acid residues (Ref. 5 and Fig. 4B).
Mutagenesis of the AdoMet2 Binding Site-Among the AdoMet radical proteins a hitherto unique feature of HemN is the presence of two molecules of AdoMet within the active site of the enzyme. To investigate the functional significance of the second AdoMet molecule (AdoMet2), which is not involved in iron-sulfur cluster coordination, site-directed mutagenesis of relevant amino acid residues was performed. According to the crystal structure of E. coli HemN (5), amino acid residues Tyr 56 , Glu 145 , Phe 310 , Gln 311 , and Ile 329 , which are all conserved among HemN enzymes albeit to a varying degree, are adequately positioned to coordinate AdoMet2 (Fig. 4B). Single amino acid exchanges were conducted, replacing Tyr 56 and Phe 310 by alanine and leucine, Glu 145 by alanine and isoleucine, and Gln 311 and Ile 329 again by alanine. The mutated enzymes were produced in E. coli BL21(DE3) and purified to apparent homogeneity under conditions identical to the wildtype enzyme. All enzyme variants were subjected to CD spectroscopy and light scattering experiments. Obtained protein parameters were identical to those measured for the wild-type enzyme (see Fig. 2 of Supplementary Materials). These observations indicated overall structural integrity of the HemN mutant enzymes. All mutant proteins were then analyzed for their iron-sulfur cluster content, their ability to cleave AdoMet and their overall coproporphyrinogen III oxidase activity.
HemN Mutants Still Carried Iron-Sulfur Clusters-For all proteins UV-visible light absorption spectra were recorded under strictly anaerobic conditions and, in addition, the iron content was analyzed colorimetrically. Both methods revealed that proteins Q311A and I329A contained the same amount of iron-sulfur cluster as the wild-type HemN. In all other mutant proteins the [4Fe-4S] cluster content was slightly reduced (Table III). However, the [4Fe-4S] cluster was never entirely absent (minimum 20% of wild type). The iron content was also reflected by the color of the protein solutions: mutant proteins Y56A, Y56L, E145A, and E145I appeared colorless, F310A and F310L were slightly yellow, and Q311A and I329A exhibited the same yellow-brown color as wild-type HemN.
AdoMet using dithionite as a reductant was determined (Table III). Under these conditions wild-type HemN cleaved two AdoMet molecules per molecule protein, corresponding to the two AdoMet bound to the enzyme. It should be noted that under these assay conditions there is no overall catalytic CPO activity because the final electron acceptor for the reaction is missing (Scheme 1c). AdoMet cleavage was also determined using E. coli cell-free extract. In the presence of the electron acceptor (provided in the E. coli cell-free extract) multiple rounds of AdoMet cleavage were observed for the wild type (Table III). The mutant proteins with regard to their AdoMet cleavage function can be divided into two groups.
1) For HemN Y56A, Y56L, E145A, and E145I absolutely no AdoMet cleavage was detected. It can be argued that in these protein variants the loss of AdoMet2 binding was not the only effect of the introduced mutation. According to the HemN crystal structure Glu 145 interacts with the ribose of AdoMet2, but also forms a salt bridge to the sulfonium sulfur of AdoMet1 in its R-configuration (Fig. 4B). Consequently, binding of both AdoMet molecules might be affected in mutant proteins E145A and E145I. Tyr 56 is not located in close vicinity to AdoMet1 and an explanation for the observed total failure to cleave AdoMet is not obvious. As mentioned above, both HemN Y56A and Y56L revealed a relatively low iron content. Possibly, the active site architecture in these mutants is more drastically disturbed than in others, although their overall structural integrity was shown by CD spectroscopy. Because of their complete lack of activity, the mutant proteins of this first group are only of limited use for the understanding of AdoMet2 function. Nevertheless, their behavior demonstrates the importance of amino acid residues within the AdoMet2 binding site.
2) HemN F310A, F310L, Q311A, and I329A represent the more interesting group of mutant proteins because they retained their ability to cleave AdoMet to a certain extent (Table  III). As observed for wild-type HemN AdoMet cleavage was strongly increased in the presence of the substrate compared with assays without substrate (data not shown), indicating that substrate binding was not affected by the introduced mutations. The values given in Table III were measured in the presence of the substrate. As mentioned above, under these assay conditions (in the absence of the final electron acceptor) wild-type HemN cleaved two AdoMet molecules per molecule enzyme. The ratio of AdoMet cleavage with respect to the amount of enzyme was also determined for the mutant proteins. Between 0.20 and 0.95 molecules of AdoMet per molecule HemN were measured (Table III). Clearly, for the mutant proteins AdoMet cleavage never exceeded one molecule of AdoMet per molecule of enzyme. Most interestingly, mutant proteins Q311A and I329A, which contained the full amount of iron-sulfur cluster also cleaved only one AdoMet molecule per molecule protein. In contrast to the wild type, the presence of substrate and electron acceptor did not significantly change AdoMet cleavage of the mutant enzymes (Table III). Under these conditions, wild-type HemN performs multiple rounds of catalysis including AdoMet cleavage. Our interpretation of these results is that for the mutant proteins within this second group the AdoMet1 molecule is still bound and thus can be cleaved upon reduction of the iron-sulfur cluster. In contrast, AdoMet2 binding is abolished because of the introduced mutations, and this explains the failure of the mutant proteins to cleave a second AdoMet as is observed for wild-type HemN.
All HemN Mutants in the AdoMet2 Binding Site Are Catalytically Inactive-Finally, the overall CPO activity of all mutant proteins was determined in the assay using E. coli cell-free extract. Mutant enzymes Q311A and I329A were still able to bind and cleave AdoMet1 and their iron-sulfur cluster content was identical to wild-type HemN. We expected either detectable CPO activity if AdoMet2 binding is not of functional importance or complete loss of CPO activity if AdoMet2 is of physiological significance.
With the exception of F310A, which carried traces of residual CPO activity, all generated mutant proteins failed to reveal detectable protoporphyrinogen IX formation (Table III). For the first group of mutant proteins this result is not surprising, because these enzymes also failed to cleave AdoMet, which is a prerequisite of HemN catalysis. More interestingly, the mutant proteins of the second group also did not show any detectable CPO activity although they were obviously still able to bind and cleave AdoMet1. Therefore, we are able to conclude that the loss of enzyme activity is caused by the failure of these enzyme variants to bind AdoMet2.
Conclusions from the HemN Mutagenesis Studies-The obtained results strongly indicate that the second AdoMet molecule bound in the crystal structure of HemN (AdoMet2) is of functional importance. Of course, the possibility that both AdoMet molecules cleaved during one catalytic cycle consecutively bind to the AdoMet1 binding site in wild-type HemN cannot be entirely ruled out at this stage. On the other hand, if this was the case, at least mutant proteins Q311A and I329A should exhibit a wild-type behavior because their iron-sulfur cluster content and AdoMet1 binding were not affected. Therefore, the observed combination of strongly reduced AdoMet cleavage and complete lack of protoporphyrinogen IX formation The iron content (mol iron/mol protein) was calculated after determination of protein and iron concentration of each sample. Protein concentrations were measured using the Bio-Rad protein assay, iron concentrations were measured colorimetrically with o-phenanthroline (see "Experimental Procedures"). For wild-type HemN a maximal iron content of 2 mol iron/mol protein was obtained as outlined in the text. This iron content was set to 100%, and the iron contents of the mutant proteins were related to that. Standard deviation did not exceed 15% of the given values.
b UV-visible light absorption spectra were recorded under anaerobic conditions for a protein concentration of 4 mg/ml. The presence of an iron-sulfur cluster was indicated by an absorption maximum at 410 nm. The intensity of this absorption maximum of wild-type HemN was set to 100%, and all other intensities were related to that.
c AdoMet cleavage without the terminal electron acceptor was determined using ͓ 14 C͔AdoMet and dithionite as a reductant for the ironsulfur cluster in the presence of substrate as described under "Experimental Procedures." The amount of determined methionine was related to the amount of enzyme employed. In the absence of substrate AdoMet cleavage was significantly reduced. d AdoMet cleavage with the terminal electron acceptor was determined using ͓ 14 C͔AdoMet and E. coli cell-free extract in the presence of substrate as described under "Experimental Procedures." As in the previous column, the amount of determined methionine was related to the amount of enzyme employed. e CPO activity was measured using non-labeled AdoMet and E. coli cell-free extract as described under "Experimental Procedures." Wildtype CPO activity was set to 100%, and the activities of the mutant proteins were related to that.
f Not detectable.
in these mutants of the AdoMet2 binding site is a convincing argument for an essential catalytic role for AdoMet2. If AdoMet cleavage was reduced for a different reason (irrespective of the presence or absence of AdoMet2), one would definitely expect to observe some residual protoporphyrinogen IX formation. The substrate molecule coproporphyrinogen III is an asymmetric molecule with regard to its substituents at the tetrapyrrolic ring system. Therefore, simple rotation at the active site is impossible. Obviously, two active site locations for the two consecutive decarboxylation steps are required. Other ringmodifying enzymes, like uroporphyrinogen III decarboxylase, solve this problem by using a dimeric protein with two active sites for one substrate molecule localized at the dimer-dimer interface (46,47). HemN, however, is a monomer. These structural prerequisites make the existence of two functional AdoMet molecules for two independent decarboxylation reactions very likely. The two co-crystallized AdoMet molecules are in adequate distance to allow these two decarboxylations at the coproporphyrinogen III macrocycle (5). Clearly, the final proof for AdoMet2 being physiologically relevant would be provided by successful crystallization and structure determination of wild-type HemN with bound substrate. This is currently being addressed in our laboratory.