Spectroscopic Studies on the [4Fe-4S] Cluster in Adenosine 5′-Phosphosulfate Reductase from Mycobacterium tuberculosis

Mycobacterium tuberculosis adenosine 5′-phosphosulfate reductase (MtAPR) is an iron-sulfur protein and a validated target to develop new antitubercular agents, particularly for the treatment of latent infection. The enzyme harbors a [4Fe-4S]2+ cluster that is coordinated by four cysteinyl ligands, two of which are adjacent in the amino acid sequence. The iron-sulfur cluster is essential for catalysis; however, the precise role of the [4Fe-4S] cluster in APR remains unknown. Progress in this area has been hampered by the failure to generate a paramagnetic state of the [4Fe-4S] cluster that can be studied by electron paramagnetic resonance spectroscopy. Herein, we overcome this limitation and report the EPR spectra of MtAPR in the [4Fe-4S]+ state. The EPR signal is rhombic and consists of two overlapping S = ½ species. Substrate binding to MtAPR led to a marked increase in the intensity and resolution of the EPR signal and to minor shifts in principle g values that were not observed among a panel of substrate analogs, including adenosine 5′-diphosphate. Using site-directed mutagenesis, in conjunction with kinetic and EPR studies, we have also identified an essential role for the active site residue Lys-144, whose side chain interacts with both the iron-sulfur cluster and the sulfate group of adenosine 5′-phosphosulfate. The implications of these findings are discussed with respect to the role of the iron-sulfur cluster in the catalytic mechanism of APR.

APS and PAPS are reduced by enzymes in the reductive branch of the sulfate assimilation pathway, producing sulfite and AMP or adenosine 3Ј,5Ј-diphosphate (Scheme 1). These enzymes can be subdivided into two groups according to their substrate preference: the APS reductases (APR) and the PAPS reductases (PAPR) (EC 1. 8.99.4). Functional and structural studies have been used to investigate the chemical reaction mechanism of APR and PAPR enzymes (1,(5)(6)(7)(8). The mechanism involves nucleophilic attack by the active site cysteine on the sulfur atom of APS or PAPS to form an enzyme S-sulfocysteine intermediate, which is cleaved by thiol-disulfide exchange with thioredoxin or glutaredoxin (Fig. 1). The sulfite product is then reduced to sulfide by sulfite reductase (EC 1.8.7.1) and utilized to synthesize cysteine and other essential sulfur-containing biomolecules (9). In the human pathogen Mycobacterium tuberculosis, APR is a validated target against the latent phase of infection (10).
Only a 3Ј-phosphate group distinguishes PAPS from APS. Accordingly, APR and PAPR have nearly identical three-dimensional structures (1.6 Å root mean square deviation of backbone atoms) and share ϳ20% sequence identity, including the active site motif, ECG, and the sulfonucleotide binding pocket (5,11,12). However, a key difference between the two enzymes is that APR contains two conserved cysteine motifs: CXXC and CC. These four additional cysteine residues coordinate a [4Fe-4S] cluster, which is essential for catalytic activity (1,5,13). The only known exception is Physcomitrella patens sulfonucleotide reductase APR, which lacks the cysteine pairs required to bind the cofactor but can utilize both APS and PAPS as substrates (14). However, P. patens APR B has to pay a significant penalty for the absence of the [4Fe-4S] cluster as evidenced by second order rate constants (k cat /K m ) of 3,520 and 37 M Ϫ1 s Ϫ1 with APS and PAPS, respectively. The 2.7 Å crystal structure of Pseudomonas aeruginosa APR (PaAPR) bound to substrate provides valuable insights into the arrangement of active site residues that are conserved among APRs (5). 3 The iron-sulfur cluster is coordinated by Cys-228 and Cys-231, positioned at the tip of a ␤-loop, along with the tandem pair Cys-139 and Cys-140, within an ␣-helix ( Fig. 2A). Coordination by sequential cysteines is highly unusual for [4Fe-4S] clusters and has been characterized in only one other crystal structure, the NuoB subunit of respiratory complex I (15). There are four charged and/or polar NH-S or OH-S hydrogen bonds involving side chains of absolutely conserved residues (Fig. 2, B and C). In particular, the Cys-Cys motif interacts with a pair of basic residues, Arg-143 and Lys-144. Other interactions with the iron-sulfur cluster involve Thr-87 and Trp-246. In the active site, the phosphosulfate group of APS is positioned opposite the [4Fe-4S] cluster, and although no atoms intervene, the sulfate moiety is not in direct contact with the [4Fe-4S] cluster.
Given the unusual Cys-Cys dyad coordination and its requirement for catalytic activity, defining the function and properties of the iron-sulfur cluster in APR has generated considerable interest (1,5,7,16,17). Most proteins containing [4Fe-4S] clusters are redox-active (18 -21); however, the [4Fe-4S] 2ϩ cluster in APR does not undergo redox changes during the catalytic cycle (1). A purely structural role also appears unlikely in light of biophysical data obtained on the apo form of APR (6,13,16) and the fact that APR and PAPR share a common protein fold (5,11,12). Unfortunately, progress in this area has been hampered by the failure to generate a paramagnetic state of the [4Fe-4S] cluster that can be studied by EPR spectroscopy and related methods (16,17,22).
Herein, we report the EPR spectra of MtAPR in the [4Fe-4S] ϩ state. The EPR spectrum of MtAPR displays a rhombic signal but is complex and consists of at least two overlapping S ϭ 1 ⁄ 2 species. Mössbauer studies of the native and reduced forms confirm the presence of a [4Fe-4S] 2ϩ/1ϩ cluster. APS binding to MtAPR is accompanied by marked sharpening of the EPR signal and an increase in intensity, which is not observed among a panel of substrate analogs, including ADP. In addition, kinetic and EPR investigation of the K144A variant of MtAPR demonstrates a key function for this residue in catalysis and as a link between APS and the iron-sulfur cluster. These data, together with known functional and structural information, directly implicate the iron-sulfur cluster in the catalytic mechanism of APS.

EXPERIMENTAL PROCEDURES
Materials-APS (Ն95%) was obtained from Biolog Life Sciences Institute (Bremen, Germany). ADP and AMP were purchased from Sigma-Aldrich (St. Louis, MO). ADP␤F was synthesized as described previously (23). The structure and purity (Ն98%) was confirmed by 1 H and 31 P NMR (data not SCHEME 1. Reaction catalyzed by sulfonucleotide reductases.  shown). 10-Methyl-3-sulfopropyl-5-deazaisoalloxazine potassium salt (deazaflavin) was a generous gift from Prof. David Ballou (University of Michigan). Titanium(III) citrate was prepared anaerobically from a 15% titanium(III) chloride solution in 1 M HCl with an equimolar amount of citrate (trisodium salt) and neutralized to pH 7.0 with saturated sodium bicarbonate.
Mutagenesis and Protein Expression-The construction of the expression vector encoding wild-type and C256S APR from M. tuberculosis cloned into the vector pET24b has been described previously (24). The K144A variant was generated from the wild-type MtAPR template using a QuikChange sitedirected mutagenesis kit (Stratagene, La Jolla, CA) and the following primer sequence: 5Ј-GCTGCCGGTTGCGCAAG-GTCGTTCCCCTGGG-3Ј. Plasmids encoding wild-type, C256S, or K144A MtAPR pET24 and pACYC (containing genes encoding the isc operon of six accessory proteins required for iron-sulfur cluster biosynthesis in Azotobacter vinelandii under the control of an arabinose-inducible promoter) (25) were cotransformed into Escherichia coli BL21(DE3) (Novagen, San Diego, CA) and plated on L-agar 50 g/ml kanamycin and 100 g/ml carbenicillin. A single colony was picked and added to 5 ml of L-broth plus antibiotics and grown overnight with shaking at 37°C. This culture was used as a 0.5% 1 liter of L-broth plus antibiotics and grown with shaking at 37°C until absorbance at 600 nm reached ϳ0.6. Arabinose and iron citrate were added to final concentrations of 20 and 0.8 mM, respectively, and the culture was grown as above for 1 h. At this point, the flasks were removed from the incubator. Isopropyl ␤-D-thiogalactoside was added to a final concentration of 0.3 mM, and the flasks were returned to the incubator and grown overnight at 18°C with shaking at 200 rpm. The cultures were harvested by centrifugation (4°C, 4,300 ϫ g). After removal of the supernatant, the pellets were stored at Ϫ80°C until required.
All of the purification steps were carried out at 4°C. Cell pellets were resuspended in 30 ml of Buffer A (20 mM sodium phosphate, 0.5 M NaCl, 10 mM imidazole, pH 7.4) supplemented with 0.1 mM PMSF, 10 g/ml DNase, 5 g/ml lysozyme and lysed by sonication. The lysates were centrifuged (20,000 ϫ g, 15 min) and loaded onto a 5-ml HiTrap chelating column (GE Healthcare, Amersham, Piscataway, NJ) equilibrated in the same buffer. Unbound material was washed off with 50 ml of Buffer A, and bound proteins were then eluted with Buffer B (20 mM phosphate, 0.5 M NaCl, 250 mM imidazole, pH 7.4). Fractions containing wild type or K144A were pooled, concentrated by centrifugation (Amicon 10-kDa cutoff; Millipore, Billerica, MA), and loaded onto a 16/60 Superdex 200 size exclusion column previously equilibrated in Buffer C (50 mM Tris-HCl, 150 mM NaCl, 5 mM DTT, 10% glycerol, pH 8.5, at 4°C). Fractions containing wild-type or K144A MtAPR were pooled, snap-frozen in liquid nitrogen, and stored at Ϫ80°C. Protein concentrations were determined using the extinction coefficient, ⑀ 280 ϭ 36,815 M Ϫ1 cm Ϫ1 , obtained from quantitative amino acid analysis (1).
Preparation of MtAPR for EPR and Mössbauer Spectroscopy-Samples of wild-type, C256S, or K144A MtAPR suitable for Mössbauer or EPR spectroscopy were prepared inside of an anaerobic chamber with O 2 levels Յ1 ppm. Purified MtAPR was exchanged into anaerobic buffer containing 50 mM Tris-HCl, 150 mM NaCl (pH 8.5 at 4°C), and 10% glycerol. To reduce the cluster in MtAPR, the reactions contained 250 M enzyme, 25 mM sodium oxalate, 250 M deazaflavin in a total volume of 250 l. When included, substrate or other analogs were added to a final concentration of 1 mM and incubated with MtAPR for 10 min at 25°C prior to photoreduction. The reaction mixtures were transferred to EPR tubes, chilled in an ice salt bath (Ϫ6°C), and irradiated with light from a 100-W quartz halogen lamp (Applied Photophysics, Surrey, UK) for 30 min. After illumination, the samples were immediately frozen in liquid nitrogen and analyzed by low temperature EPR. Mössbauer spectra were recorded on proteins that contained 57 Fe in place of natural abundance iron. 57 Fe was incorporated into MtAPR via supplementation of E. coli growth medium, and samples contained 1 mM protein and 2 mM APS when appropriate. After 10 min of incubation with substrate, the samples were transferred to Mössbauer cups and frozen in liquid nitrogen.
EPR Spectroscopy-X-band EPR spectra of photoreduced samples were recorded on a Bruker EMX spectrometer (Billerica, MA) equipped with an Oxford Instruments ITC4 temperature controller, a Hewlett-Packard model 5340 automatic frequency counter, and a Bruker gaussmeter. The figure legends contain relevant instrumental parameters. The sample buffer was used to record base lines under conditions identical to those in which the sample spectra were obtained. These base lines were subtracted from the MtAPR spectra shown in the figures. Spin concentrations in MtAPR samples were determined by double integration of the EPR signal over a range of 2 kgauss and comparison with double integrals of 1 mM Cu(ClO 4 ) 2 in sample buffer. EPR spectra of cryoreduced samples were recorded on a Bruker ER-200DSRC spectrometer equipped with an Oxford Instruments ESR 910 continuous flow cryostat. Simulations of EPR spectra were performed using Spin Count (version 2.6.7) created by Professor M. P. Hendrich at Carnegie Mellon University. Spin Count is available online.
Mössbauer Spectroscopy-Mössbauer spectra were recorded on a spectrometer from WEB Research (Edina, MN) operating in the constant acceleration mode in transmission geometry. The spectra were recorded with the temperature of the sample maintained at 4.2 K in an externally applied magnetic field of 53 mT oriented parallel to the ␥-beam. The quoted isomer shifts were relative to the centroid of the spectrum of a foil of ␣-Fe metal at room temperature. The data analysis was performed using the program WMOSS from WEB Research.
Cryoreduction of MtAPR by Low Temperature ␥-Radiolysis-Samples containing 250 M MtAPR were loaded into EPR tubes or Mössbauer cups and flash frozen in liquid nitrogen inside the glove box. When appropriate, APS was added to a final concentration of 1 mM and incubated with protein for 10 min at 25°C prior to freezing. The samples were ␥-irradiated ( 60 Co; total dose of 4 Mrad) at the ␥-irradiation facility of the Breazeale nuclear reactor at the Pennsylvania State Univer-sity. During irradiation, the samples were maintained at 77 K by immersion in liquid N 2 .
MtAPR Activity Assay-Reactions were carried out at 30°C. Unless otherwise indicated, the buffer consisted of 100 mM bis-tris propane (pH 7.5) and 100 mM NaCl supplemented with 5 mM DTT and 10 M E. coli thioredoxin. Production of 35 SO 3 2Ϫ from [ 35 S]APS was monitored using charcoal-based separation and scintillation counting as reported previously (26). The substrate was incubated with excess enzyme to ensure single-turnover conditions (Ͼ2.5-fold molar excess of enzyme). The reaction progress curve was plotted as the fraction of product versus time and was fit by a single exponential where F is the fraction product, A is the fraction of substrate converted to product at completion, k obs is the observed rate constant, and t is time. The reactions were followed for Ն5 half-lives except for very slow reactions. Under single-turnover conditions, it is expected that the concentration dependence of the enzyme will be hyperbolic (Equation 1).
To determine k cat /K m , a concentration of enzyme was chosen that was at least 5-fold below the K m value. Although we refer to the K1 ⁄ 2 for maximal activity as K m , we note that the K1 ⁄ 2 for single turnover is not necessarily the same as the K m for the multiple turnover reaction because the latter can be affected by the rate of product release. For conditions in which [E] Ͻ Ͻ K m , the second order rate constant, k cat /K m ϭ k obs /[E]. We note that the reported values of k cat /K m are for single-turnover conditions, but the measurement is equivalent to steady state k cat /K m . At a saturating concentration of enzyme, the observed single-turnover rate constant reaches a maximum, k max . To determine k max , the concentration of enzyme was varied by at least 3-fold to establish that the observed rate was independent of the concentration of enzyme, indicating that the enzyme was in excess and at a saturating concentration (i.e. k obs ϭ k max ). The reported values of k cat /K m and k max are the averages of at least three independent determinations. Unless otherwise indicated, the standard deviation was Յ15% of the value of the mean.
Determination of Substrate Affinity-The apparent dissociation constant (K d ) for [ 35 S]APS from K144A MtAPR-ligand complexes was measured using an ultrafiltration binding assay reported by Hernick and Fierke (27). Because the chemical step (i.e. S-sulfocysteine formation) is rate-limiting (26), the K1 ⁄ 2 is equal to K d of APS for K144A MtAPR. In brief, the concentration of substrate was kept low (i.e. below the K d ) and constant, and the concentration of the enzyme was varied (0 -80 M). K144A MtAPR was incubated in 100 mM bis-tris propane, pH 7.5, at 30°C for 15 min prior to the assay to allow for ligand equilibration. Assay mixtures were then transferred into ultrafiltration devices (Microcon 30-kDa cutoff; Millipore), and the free and bound ligand were separated by centrifuging the samples at 3,000 rpm for 2.5 min. Equal volumes of the filtrate and retentate were removed and quantified using scintillation counting. The ratio of EL/L total was determined as a function of [E] total , and the K d value was obtained by fitting Equation 2 to these data. 2ϩ Cluster in MtAPR-We have previously reported conditions for the purification of MtAPR (1). However, yields from these preparations were low because of the large quantity of MtAPR present in the insoluble protein fraction. To improve the yield and stability of purified MtAPR, we coexpressed the MtAPR gene with the gene products of the isc operon required for iron-sulfur cluster biosynthesis in A. vinelandii (25). Under these conditions the yield of MtAPR was typically 7 mg/liter of culture, which represents an improvement over the ϳ1 mg/liter obtained when MtAPR is overexpressed in the absence of the isc proteins. The specific activity of the purified enzyme was 5 M min Ϫ1 mg of protein Ϫ1 with thioredoxin and DTT as reductants. The UV-visible absorbance spectrum of MtAPR showed a maximum in the visible range at 410 nm that is consistent with the presence of bound iron (supplemental Fig. S2). Analysis of iron content by inductively coupled plasma resonance spectrometry for MtAPR indicated that each mole of protein contained 3.5 Ϯ 0.4 mol of iron, which is indicative of four iron atoms in the cluster. The intensity of the 410-nm peak was unaffected by the addition of sodium dithionite (data not shown) but increased slightly upon the addition of APS (supplemental Fig. S2). These results are analogous to those found for PaAPR (6). The minor increase in absorption at 410 nm could reflect substrate-dependent conformational changes within the C-terminal region and a concomitant alteration in cluster environment.
Photoreduction of the [4Fe-4S] 2ϩ cluster in MtAPR-Isolated APR from higher plants and P. aeruginosa exhibits weak isotropic signals at g ϭ 2.01 attributed to a small proportion of [3Fe-4S] ϩ cluster and at g ϭ 4.3 from high spin Fe(III) (16,17). When MtAPR was prepared according to the previous method (13), similar resonances were observed (data not shown). However, such EPR signals were not present in samples of MtAPR produced via the improved coexpression system. Earlier attempts to generate new EPR signals in assimilatory APR from higher plants and bacteria by titrating the enzyme with dithionite, Ti(III)citrate or photochemical reduction with the deazaflavin/oxalate system have proven unsuccessful (16,17,22). Similarly, our earlier studies of MtAPR found no evidence for the presence of [4Fe-4S] ϩ after treatment with dithionite (13).
In the present study, we first explored reduction of the cluster in MtAPR using Ti(III)citrate. New EPR signals were observed; however, interpretation of the spectra was confounded by nonspecifically bound Ti(III)citrate that gave rise to an isotropic signal at g ϭ 1.94 (data not shown). Next, we tested photoreduction of the [4Fe-4S] 2ϩ center in MtAPR in the presence of deazaflavin/oxalate. The resulting EPR spectrum is broad but shows rhombic symmetry with apparent g values of 2.04, 1.94, and 1.75 (Fig. 4A). The EPR signal also gives evidence for a second component with apparent g values at 2.13 and 1.85; however, the broad resonances precluded an accurate simulation of the two paramagnetic species. Spin quantitation of the EPR signals from g ϭ 2.5 to 1.3 indicate 0.3 eq of spins/mole of enzyme. On the basis of the observed g values, the resonances can be attributed to either an S ϭ 1 ⁄ 2 [2Fe-2Fe] ϩ or [4Fe-4S] ϩ cluster. The temperature dependence of the EPR signal (data not shown) indicated that it was maximal between 8 and 10 K and was no longer visible at temperatures above 12 K, using a microwave power of 10 milliwatts. This behavior suggests that the paramagnetic signal arises from a [4Fe-4S] ϩ cluster. By contrast, [2Fe-2S] ϩ clusters are slow relaxing and observable by EPR at temperatures above 70 K (28).
Interaction of the Photoreduced [4Fe-4S] ϩ Cluster with Substrate and Analogs-The [4Fe-4S] cluster at the active site of APR is required for catalytic activity (1); however, the mechanistic details are unknown and remain a central question for this family of enzymes. Although the iron-sulfur cluster in APR does not undergo redox activity during the catalytic cycle (1), the 1ϩ state of APR can serve as a useful tool for mechanistic studies analogous to other enzymes that harbor redox-inactive iron-sulfur clusters such as aconitase (29). Therefore, to gain insight into the functional role of the clus-ter, we next investigated whether substrate binding would perturb the EPR spectrum of reduced MtAPR. Two types of protein-ligand complexes were prepared: (i) wild-type MtAPR treated with APS to afford the S-sulfocysteine intermediate form of the enzyme bound to AMP and (ii) the catalytically inactive variant, C256S MtAPR bound to APS. Established procedures for complex formation were performed (1,13), and the resultant samples were then subjected to photoreduction.
Compared with wild-type MtAPR (Fig. 4A), the EPR spectrum of C256S MtAPR bound to APS is markedly sharper in appearance with apparent g values of 2.02, 1.91, and 1.76 (Fig.  4B). The EPR signal of the enzyme S-sulfocysteine intermediate bound to AMP also exhibits increased resolution with apparent g values of 2.03, 1.91, and 1.75 (Fig. 4C). Additionally, both spectra (Fig. 4, B and C) indicate the presence of a second paramagnetic species. Spin quantitation of EPR signals from g ϭ 2.7 to 1.6 indicate a respective 0.45 and 0.4 eq of spin/mol for the C256S and wild-type MtAPR complexes. The heterogeneity of these samples was essentially unaffected by the addition of 2.5 M urea, changes in sample pH between the ranges of 7.5 and 9.5 and variation in temperature or microwave power levels (data not shown).
Simulation of the signal from wild-type MtAPR treated with APS suggests that it is the sum of at least two S ϭ 1 ⁄ 2 components (Fig. 5). One species has narrow lines and g values at 2.04, 1.94, and 1.76 (component A), whereas the other has broad lines and exhibits principle g values at 2.10 and 1.88 (component B) with an intensity ratio of 1:1.2. Although simulated and experimental spectra are in overall agreement, some discrepancies remain, particularly with respect to signal amplitudes. The addition of a third spectral component did not improve the fit quality. However, residual differences between the simulated and experimental spectra may result from weak signal intensity, minor fluctuations in temperature ϳ10 K, or the large number of variables required to simulate a [4Fe-4S] ϩ cluster.
Next, we tested whether the observed APS-dependent changes in the EPR spectra of reduced MtAPR were specific to the substrate. To investigate this possibility, we analyzed the spectra of photoreduced MtAPR bound to AMP, ADP, and ADP␤F. The EPR signal of MtAPR bound to AMP (Fig.  4D) was similar in shape and intensity to enzyme alone (Fig. 4A). By contrast, the spectrum in the presence of ADP (Fig. 4E) showed a significant reduction in signal intensity. To determine whether the observed changes in the EPR spectrum were due to the additional negative charge at a ␤-phosphate group (i.e. ADP) relative to a ␤-sulfate group (i.e. APS), we examined fluorine substitution of a ␤-nonbridging oxygen atom. Interestingly, however, the EPR spectrum of MtAPR bound to ADP␤F (Fig. 4F) was essentially the same as ADP (Fig. 4E). Spin quantitation of MtAPR bound to ADP or ADP␤F indicate less than 0.1 eq of spins per mole of enzyme in each sample. The marked increase in signal resolution and intensity in the presence of APS, but not upon the addition of substrate analogs or product, reflects a unique state of the active site and cluster environment, which might be related to catalytic activity.
Characterization of K144A MtAPR-The preceding EPR data support the existence of mid-range electrostatic interactions between the cluster and APS. Given that the side chain of Lys-144 is positioned between coordinating Cys-140 and the ␤-sulfate group of APS (Fig. 2C), we reasoned that this residue might help mediate this interaction. To investigate this possibility, we generated the K144A MtAPR variant and characterized this protein through kinetic and EPR studies. An effect of 63,000-fold on k cat /K m is observed upon mutation of Lys-144 to Ala (Table 1; see also supplemental Fig. S4). Because the chemical step (i.e. S-sulfocysteine formation) is ratelimiting for the reaction of APS and MtAPR (26), this value represents the effect of removal of the Lys residue on the overall binding and chemical transformation. The mutation decreases the value of k max by 270-fold, suggesting that the Lys side chain stabilizes the transition state relative to the ground state complex. To further explore the molecular recognition of APS, we measured the K d value of the substrate for K144A MtAPR. Relative to wild-type MtAPR, the affinity of APS for K144A is decreased by 400-fold.
The UV-visible absorbance spectrum of K144A MtAPR showed a maximum in the visible range at 410 nm that is consistent with the presence of bound iron (data not shown). Analysis of iron content by inductively coupled plasma resonance spectrometry for K144A MtAPR indicated that each mole of protein contained 3.3 Ϯ 0.4 mol of iron, which is consistent with four iron atoms in the cluster. Photoreduction of K144A MtAPR gave rise to a rhombic signal with resonances at g ϭ 2.12, 1.99, and 1.82 (Fig. 6, A and C). However, the signal intensity was decreased relative to wild-type MtAPR, and spin quantitation accounted for less than 0.05 eq of spins/ mole of enzyme. In the presence of saturating APS, the feature at g ϭ 1.75 disappears, and the remaining signal exhibits g values at 2.12 and 1.99 (Fig. 6, B and D). A low field isotropic Fe 3ϩ signal accounting for less than 0.01 eq of spin per mole of enzyme was also observed in the presence of APS, consistent with a small degree of cluster degradation (data not shown). Although the EPR spectra for both samples are relatively broad and weak, a modest reduction in the magnetic heterogeneity of MtAPR can be observed. Together, the kinetic and EPR data indicate a key role for Lys-144 in chemistry and substrate binding and show that this residue helps modulate APS-dependent changes in the iron-sulfur cluster environment.  (30,31). To this end, the 2ϩ state of MtAPR was incubated in the presence or absence of APS, frozen in liquid nitrogen, and then exposed to high energy ␥-irradiation at 77 K to produce the [4Fe-4S] ϩ cluster.

Cryoreduction of the [4Fe-4S] 2ϩ Cluster in MtAPR-
The resulting spectrum in the absence of substrate (supplemental Fig. S5) is ill defined and relatively broad. However, the addition of substrate APS was accompanied by a marked increase in signal intensity and resolution with apparent g values at 2.08, 2.04, and 2.02 (supplemental Fig. S5). The observation of substrate-dependent changes in the EPR spectrum is in qualitative agreement with our experimental results from photoreduction. However, we note that the EPR spectra of cryoreduced MtAPR are distinct from those of the photochemically reduced enzyme. These findings may indicate that the substrate interacts differently with the cluster in the ϩ2 and ϩ1 states. An alternative possibility is that the observed EPR signals correspond to a thiyl radical formed by homolytic scission of the S-sulfocysteine bond. In support of this pro-

Effect of K144A mutation on APR-catalyzed reduction and binding of APS
Rate constants for single-turnover reactions were determined at 30°C in 100 mM bis-tris propane buffer, 5 mM DTT, 10 M thioredoxin as described under "Experimental Procedures." In all cases, the protein was in excess over substrate, with at least 2.5-fold more protein than substrate.  Anaerobic 250 M K144A MtAPR alone (A and C) or incubated with 1 mM APS (B and D) for 10 min at 25°C was then photoreduced as described under "Experimental Procedures." In A and B, EPR signal intensities have been scaled to match those in Fig. 4. In C and D, the intensity of the EPR signals has been scaled 7.5-fold. The EPR spectra were recorded at 10 K, and the instrument parameters were: microwave power, 10 milliwatt; receiver gain, 2 ϫ 10 4 ; modulation amplitude, 10 G; and microwave frequency, 9.43 GHz.
posal, the relatively high g values are consistent with literature data for a sulfur-centered species (32,33). In addition to the aforementioned g values, the cryoreduced samples contained an extremely intense signal isotropic signal at g ϭ 2, which could arise from other nonspecifically reduced organic radicals present in the sample (supplemental Fig. S5). 4.2-K/53-mT Mössbauer spectra of samples recorded before and after cryoreduction (supplemental Fig. S6) indicate that ϳ50% of the [4Fe-4S] 2ϩ clusters were reduced to the ϩ1 state under these conditions. Thus, Mössbauer analysis confirms the presence of the [4Fe-4S] ϩ cluster in cryoreduced samples; however, the ␥-irradiation also appears to have generated extremely intense, new paramagnetic species that are distinct from a simple one-electron reduction of the ϩ2 cluster. Lastly, no significant changes in the Mössbauer spectra from substrate-bound and substrate-free cryoreduced MtAPR were observed (supplemental Fig. S6).
Ferricyanide Oxidation of the [4Fe-4S] 2ϩ Cluster in MtAPR-The [4Fe-4S] 2ϩ cluster in PaAPR is partially converted to the [3Fe-4S] ϩ form by treatment with the oxidant potassium ferricyanide (16). To assess this possibility for MtAPR, a stoichiometric amount of potassium ferricyanide was added to the enzyme. The resulting EPR signal shows a well defined EPR resonance at g ϭ 2.03 and a poorly resolved high field component at g ϭ 1.99 (supplemental Fig. S7). This pattern of EPR signals is very similar to those of S ϭ 1 ⁄ 2 [3Fe-4S] ϩ clusters in aconitase (29) and endonuclease III (34). Spin quantitation of the EPR signals from g ϭ 2.1 to 1.8 indicates less than 0.1 eq of spins/mole of enzyme. The EPR spectrum of oxidized MtAPR also exhibited a signal accounting for less than 0.01 eq of spins/mol of enzyme at g ϭ 4.3 that is characteristic of high spin Fe(III). When ferricyanide was added to MtAPR that had been pretreated with APS, the intensity of the signal at g ϭ 4.3 was increased by 3-fold (data not shown). Attempts to purify the [3Fe-4S] ϩ cluster form of MtAPR were unsuccessful because the cluster rapidly decomposed upon oxidant removal. Nonetheless, these data are consistent with the existence of a labile iron site within the [4Fe-4S] 2ϩ cluster of MtAPR. Given the constraints of tandem Cys coordination (5) and the proximity of Lys-144, the iron coordinated to Cys-140 may correspond to the displaced atom.

DISCUSSION
Iron-sulfur clusters are amazingly versatile cofactors with functions in electron transfer, Lewis acid-assisted enzyme catalysis, radical generation, oxidation of a wide variety of substrates under anaerobic conditions, and protein structure (35)(36)(37)(38)(39)(40). Although numerous studies indicate that the ironsulfur cluster is essential for APR activity (1,5,7,13), the specific role of the iron-sulfur cluster has been elusive. Progress on this front has been limited, in part, by the inability to generate a paramagnetic state of the [4Fe-4S] cluster that can be studied by EPR spectroscopy. The present study is the first reported case in which the [4Fe-4S] 2ϩ cluster of APR has been reduced to the S ϭ 1 ⁄ 2 [4Fe-4S] ϩ form.
Photoreduction in the presence of deazaflavin and oxalate turned out to be the most effective method to generate the paramagnetic 1ϩ state of MtAPR. By contrast, attempts to photochemically reduce APR from plants and bacteria such as P. aeruginosa and Bacillus subtilis have not been successful (13,22). Active site residues that interact with the iron-sulfur cluster and/or the substrate (e.g. Thr-87, Arg-143, Lys-144, Arg-242, Arg-245, and Trp-246) are highly conserved among APRs and therefore are unlikely to account for EPR behavioral differences. On the other hand MtAPR is a monomer in solution, whereas APR from other species exists as a homodimer or homotetramer (1,17). Along these lines, the structure of PaAPR shows that the face of the iron-sulfur cluster opposite the active site is buried at the interface between two monomeric subunits (5). Studies on the effect of solvent on redox potentials of model clusters indicate that water raises the reduction potential of the [4Fe-4S] 2ϩ/ϩ couple (41,42). Hence, increased solvent accessibility to the cluster of monomeric MtAPR may account for the ability to generate the 1ϩ state.
The EPR signal of reduced MtAPR arises from a mixture of [4Fe-4S] ϩ clusters with S ϭ 1 ⁄ 2, possibly reflecting the existence of distinct conformational states. Multiple S ϭ 1 ⁄ 2 ground states have been observed for other [4Fe-4S] ironsulfur enzymes such as the corrinoid protein from Clostridium thermoaceticum (43), human DNA primase (44), and the ribosomal RNA methyltransferase, RumA (45). The apparent lack of changes in the EPR spectra between pH 6.5 and 9.5 suggests that the complexity does not result from differences in protonation state of residues near the iron-sulfur cluster. Freezing of samples after photoreduction can also led to multiple signals from the same paramagnetic center: one from aggregated protein molecules and the other from dispersed molecules (46). To investigate this possibility, we recorded the EPR spectrum for cryoreduced MtAPR. In these experiments, the EPR signal that arises from the [4Fe-4S] ϩ form of MtAPR is weak relative to other radical species generated during the ␥-irradition process. Nonetheless, magnetic heterogeneity is still apparent in the spectrum (supplemental Fig. S5). Treatment with 1-2.5 M urea has also been reported to convert multiple isomeric states of an [4Fe-4S] ϩ cluster into a single one (43), but this was not the case with MtAPR. This observation may indicate the existence of more subtle differences between cluster forms in MtAPR such as changes in the orientation of a Cys-S␥-Fe bond or in hydrogen bonding to a sulfur atom.
Regarding possible conformational changes, PaAPR has been fortuitously crystallized with APS bound in two of the four monomeric subunits (5). A comparison of bound and unbound states reveals minor structural changes in residues adjacent to the iron-sulfur cluster (Fig. 2, B and C). In the absence of substrate, Lys-144 is present in an extended conformation such that the distance between the side chain and the S␥ atom of Cys-140 is ϳ5 Å. By contrast, the subunit with APS bound shows that the side chain of Lys-144 adopts a bent rotamer conformation, which brings this residue within interaction range (i.e. 3.5 Å) of Cys-140 S␥. In turn, the Cys-140 side chain moves slightly upward (i.e. 5°rotation at the ␤-carbon). It is possible then, that the heterogeneity observed in the EPR spectra may be related to conformational dynamics of Lys-144, Cys-140, and/or Cys-256 within the C terminus.
In support of this hypothesis, a decrease in EPR signal complexity was observed with K144A MtAPR (Fig. 6). Structural changes at these key residues may also account for the observation that the iron-sulfur cluster of APR is protected from oxidation upon addition of APS (13) and that the resonance Raman spectra exhibit changes in iron-S␥ stretching modes when substrate is bound (6).
Previous models for the role of the iron-sulfur cluster in APR catalysis have postulated that one or two nonbridging sulfate oxygen(s) establish a direct interaction with an iron atom (i.e. monodentate or bidentate coordination of the substrate) (47). The interacting iron atom could facilitate nucleophilic attack at the sulfate sulfur by acting as a Lewis acid, analogous to aconitase (29). Because the cluster in APR is ligated to the protein via four cysteine residues, the coordination number of the interacting iron site would increase to 5 or 6. Based on studies of direct interaction between substrate or cofactor and [4Fe-4S] 2ϩ clusters in aconitase (29), pyruvate formate-lyase activase (48), and biotin synthase (49), significant perturbation of the Mössbauer parameters for an interacting iron site in MtAPR would be expected (48,50,51). The present work, however, shows no discernable change in the Mössbauer spectrum upon the addition of APS to MtAPR (supplemental Fig. S3), suggesting that the cluster and the substrate do not establish a direct interaction. This proposal is fully consistent with the crystal structure of PaAPR bound to APS, which shows that the sulfate oxygens are 7 Å from the closest iron atom and 6 Å from the S␥ atom of Cys-140 ( Fig.  2B) (5).
On the other hand, substrate binding to MtAPR led to a marked increase in the intensity and resolution of the EPR signal and to minor shifts in principle g values that were not observed with closely related substrate analogs, such as ADP and ADP␤F (Fig. 4). These findings correlate with observed differences in the dissociation constants (K d values) reported for these ligands (26) and can be rationalized on the basis of differences in anionic charges: ␤-sulfate has a net negative charge of one distributed over three oxygen atoms (i.e. Ϫ1/3 charge each; K d ϭ 0.25 M), whereas ␤-phosphate has a net negative charge of two distributed over three oxygen atoms (i.e. Ϫ2/3 charge each; K d ϭ 5 M), and ␤-fluorophosphate has a net negative charge of one distributed over two oxygen atoms (i.e. Ϫ1/2 charge each; K d ϭ 2.5 M). By contrast, the similarity of EPR spectra obtained in the presence or absence of AMP (Fig. 4, A and D) suggests that the ␣-phosphate group is too distant from the cluster to exert any significant effect. Taken together, these data are indicative of mid-range electrostatic interactions between the iron-sulfur cluster and the ␤-functional group of the ligand, which are finely attuned to the electrostatic properties of the sulfate in APS.
In addition to differences in formal anionic charge, the sulfur atom of sulfate is larger and more electronegative and forms shorter bonds with oxygen relative to a phosphorous atom. As a consequence, the oxygen atoms associated with a ␤-sulfate group are associated with less negative charge density relative to oxygen atoms attached to phosphate (52). As compared with APS, the increase in negative charge density associated with ␤-nonbridging oxygen atoms of ADP and ADP␤F might strengthen their interaction with the positively charged Lys-144 side chain such that this residue moves away from the S␥ atom of coordinating Cys-140. Alternatively, or in addition, considering that the [4Fe-4S(Cys-S␥) 4 ] cluster has a net charge of Ϫ2 (53), repulsive electrostatic interactions could arise when ADP or ADP␤F are bound in the active site. Either scenario could account for the observed decrease in analog affinity and hamper the ability of the iron-sulfur cluster in MtAPR to accept an additional electron (i.e. adopt the reduced 1ϩ state).
In general, spin quantitation of the EPR signals revealed less than 0.5 spin/mol MtAPR. These less than unitary values for the spin integration of the reduced spectra can be attributed to incomplete reduction of the [4Fe-4S] 2ϩ cluster that may result from insufficient light intensity or illumination time. However, changes in these parameters were accompanied by noticeable cluster degradation and were therefore not pursued further. Despite the modest efficiency of reduction, variations in spin quantitation between substrate-bound (or substrate analog) and substrate-free states of MtAPR were observed. These differences are likely due to ligand-dependent structural and electrostatic changes in cluster environment, as discussed above.
The positively charged side chain of Lys-144 interacts with both APS and coordinating Cys-140, forming a link between the substrate and the iron-sulfur cluster. Although strict conservation of Lys-144 implies an important role, this study is the first to probe the precise function of this residue. Using site-directed mutagenesis, we demonstrate that Lys-144 is necessary for both substrate binding and transition state stabilization (Table 1). Moreover, the significant decrease in EPR signal intensity for K144A MtAPR (Fig. 6) suggests that the proximity of the positively charged side chain of this residue may be an important factor in stabilizing the reduced state of the cluster. Future measurements of midpoint reduction potentials could provide additional support for this proposal.
Comparison of the EPR spectra between wild-type, C256S (Fig. 4), and K144A MtAPR (Fig. 6) shows that the Lys residue plays an important role in modulating substrate-dependent changes in signal resolution and intensity. However, we note that the EPR spectrum of K144A still exhibits subtle differences in the presence or absence of APS. In particular, the APS-bound state shows small changes in signal intensity at g ϭ 2.12 and 1.99 and a decrease in the spectral feature at g ϭ 1.82. These differences in EPR spectra may be attributable to the intrinsic electrostatic interaction between the iron-sulfur cluster and the substrate.
On the basis of the results reported herein, we propose a catalytic role for the iron-sulfur cluster in the mechanism of APS reduction, specifically in the sequence of events leading up to and through sulfuryl group transfer (Fig. 7). Prior to APS binding, the [4Fe-4S(Cys-S␥) 4 ] 2Ϫ cluster may serve to preorganize the positively charged side chain of Lys-144 and possibly Arg-245 (Fig. 2, B and C) within the active site, so that the substrate can establish interactions with these residues. Based on the Mössbauer data and structural studies, APS binds to the [4Fe-4S] 2ϩ state of APR but does not appear to come into direct contact with the cluster. Rather, EPR in-vestigation of the [4Fe-4S] ϩ and [3Fe-4S] ϩ states indicate that mid-range electrostatic interactions arise between APS and the iron-sulfur cluster. The charge from and polarization within the [4Fe-4S(Cys-S␥) 4 ] 2Ϫ cluster could serve to activate the sulfate group of APS, thereby facilitating S-OP cleavage and S-S bond formation in the reaction. In the absence of an iron-sulfur cluster, PAPR may achieve something similar via repulsion between the extra 3Ј-phosphate group of PAPS and the sulfate end of the 5Ј-phosphosulfate. The sulfate group would then be primed for nucleophilic attack by Cys-256, leading to formation of the S-sulfocysteine enzyme intermediate. In support of this proposal, experimental and theoretical studies of sulfuryl group transfer in sulfate esters (52,54) indicate the need for a strong polarizing agent. Regarding Lys-144, our analysis demonstrates an essential role for this residue in enzyme catalysis. Although only speculative at present, it is plausible that the active site Lys-144 cation may orient the incoming Cys nucleophile with respect to the sulfate moiety and/or act as a "molecular guide wire" during sulfuryl transfer; either possibility could result in transition state stabilization. Analogous functions for Lys and Arg residues in the active site of other enzymes that catalyze sulfuryl transfer have been described (55,56).
In APR from B. subtilis, on the basis of its short half-life (t1 ⁄ 2 , ϳ12 min), an oxygen-sensing function for has been proposed for the iron-sulfur cluster (22). By comparison, the iron-sulfur in MtAPR is significantly more stable with a half-life of ϳ6 h (13). Nevertheless, it is important to note that the model we have proposed above does not preclude the possibility that the cluster plays a regulatory role. Rather, the iron-sulfur cluster of APR may function both in a catalytic capacity and, in some organisms, as an oxygen-sensitive switch.
In summary, the data presented in this work are consistent with a catalytic function for the iron-sulfur cluster in APR. On the basis of these data, we have proposed that the cofactor plays a role in preorganizing active site residues and in substrate activation. Additional experimental and theoretical tests of these proposals are currently underway and will be reported in due course. Moreover, the finding that MtAPR can be reduced to the [4Fe-4S] ϩ state opens the door for other forms of spectroscopy such as electron nuclear double resonance and electron spin echo envelope modulation that can provide further insight into the relationship between the substrate and iron-sulfur cluster.