A Novel [2Fe-2S] Ferredoxin from Pseudomonas putida mt2 Promotes the Reductive Reactivation of Catechol 2,3-Dioxygenase*

Catechol 2,3-dioxygenase (XylE) is a component of the TOL plasmid-encoded pathway for the degradation of toluene and xylenes and catalyzes the dioxygenolytic cleavage of the aromatic ring. Purified XylE is oxygen-sensitive and unstable in vitro , particularly in the presence of substituted catechol substrates, but it is stabi-lized in vivo by another protein, XylT, encoded by the xylT gene located just upstream of xylE . In this study, we have purified to homogeneity the XylT product from a recombinant Escherichia coli strain containing a hyper-expressible xylT gene and characterized it as a novel [2Fe-2S] ferredoxin. It is the first example of a soluble ferredoxin with a net positive charge at neutral pH. The EPR signal of the iron sulfur cluster has rhombic symmetry as is the case for plant-type ferredoxins, but the XylT absorbance spectrum resembles more closely that of adrenodoxin

Pseudomonas putida mt2 carries the TOL plasmid pWW0, which encodes a set of enzymes responsible for the transformation of toluene and xylene to central pathway intermediates (1).
The catabolic (xyl) genes are organized in two operons, the so-called upper and meta-operons. The enzymes encoded by the upper operon catalyze the sequential oxidation of toluene to benzoate, whereas the enzymes of the meta-operon convert benzoate to Krebs cycle intermediates (2). In the meta-pathway, the enzyme catechol 2,3-dioxygenase encoded by xylE catalyzes the extradiol cleavage of the aromatic ring. This reaction has been studied in detail, and a general mechanism for the oxidative cleavage of catechol by extradiol dioxygenases has been proposed (3). The two atoms of the oxygen molecule are incorporated in the catechol substrate on two adjacent carbon atoms of the aromatic ring, one of which already carries a hydroxyl substituent of the diol and the other of which is unsubstituted. At the active site, the enzyme possesses a single iron atom that binds the substrate and oxygen and participates in the catalytic cycle.
Structural determination of the related biphenyl 2,3-dioxygenase revealed that the iron atom is bound to the polypeptide through covalent linkages with the side chains of three residues, histidines 146 and 210 and aspartate 260 (4), residues that are conserved among extradiol dioxygenases (5). In the active enzyme, the iron atom is in the ferrous state and is assumed to stay reduced throughout the catalytic cycle. However, in the presence of certain substrates, such as 4-methylcatechol and chlorocatechols, oxidation of the iron atom occurs, accompanied by inactivation of the enzyme (6). Slow but significant inactivation also takes place during enzyme turnover with catechol as substrate (7).
In a recent study, Polissi and Harayama (8) found that a mechanism exists in vivo that hinders irreversible inactivation of catechol dioxygenase. Mutants lacking a functional xylT gene lost the ability to grow on p-xylene and p-toluate as carbon sources. In addition, it was found that 4-methylcatechol, which is an intermediate in the degradation of p-xylene and p-toluate, irreversibly inactivated catechol 2,3-dioxygenase in the xylT mutants, while the enzyme remained active in wild-type P. putida. It was therefore envisioned that the xylT gene product might participate in a mechanism of protection or reactivation of the catechol 2,3-dioxygenase (8). The xylT gene lies immediately upstream of xylE, and its sequence suggests that it may code for a ferredoxin (9).
In this study, we have purified the xylT gene product and tested whether it can reactivate catechol 2,3-dioxygenase in vitro. The xylT gene was hyperexpressed in Escherichia coli yielding a red protein that was purified and characterized as a novel [2Fe-2S] ferredoxin. An experimental system was developed to monitor XylT-dependent reactivation of the catechol 2,3-dioxygenase. It was demonstrated that XylT alone can reactivate catechol 2,3-dioxygenase in a reaction that is rapid and specific and requires reductant. The low amount of XylT de-tected in wild-type P. putida cells suggests that it acts catalytically rather than stoichiometrically.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-E. coli strain DH5␣ (Life Technologies, Inc.) was used as a host for general DNA manipulations, as well as for xylE hypererexpression. Strain BL21(DE3) (Novagen) was employed for hyperexpression of the xylT gene. Cultures were grown on LB medium containing appropriate antibiotics. For XylT hyperproduction, cultures were grown at 28°C in 10-liter fermentors supplemented with 50 M FeSO 4 . For XylE production, E. coli, DH5␣ (pAW31) (10), was grown overnight at 37°C in 2-liter flasks containing 0.8 liter of culture. Bacteria were harvested by centrifugation, washed in 50 mM Tris-HCl, pH 8.0, and then stored at Ϫ20°C. P. putida strains mt2 (1) and KT 2442 (11) were grown on minimal M9 medium supplemented with 5 mM benzoate as sole carbon source. Cultures were incubated at 30°C and 200 rpm in a rotary shaker.
Cloning and Hyperexpression of the xylT Gene in E. coli-The xylT gene was amplified by polymerase chain reaction using Taq polymerase (Promega), plasmid pAW31 (10), as a template and the following primers: AJ021, 5Ј-cgggcatATGAACAGTGCCGGCTACG-3Ј; AJ022, 5Ј-gcg-cggatcCTCATGACGTCACCTCTTC-3Ј. NdeI and BamHI sites were included in these primers to facilitate subsequent cloning into the expression plasmid. A fragment of the expected size was purified and cloned into pGEM-T (Promega) to give plasmid pAJ90. The xylT insert was checked by DNA sequencing (12) and then subcloned into pET9a (Novagen). The resulting plasmid called pAJ91 was introduced into the E. coli strain BL21(DE3) (Novagen). Induction of xylT expression was initiated with 1 mM isopropyl ␤-D-thiogalactoside when the bacterial density reached an A 600 of 1.8. Bacteria were harvested 6 h later.
Purification of XylT-The purification procedure was performed at 4°C under argon using Tris-HCl buffer, pH 9.0, containing 2 mM dithionite. Bacterial cells from 20 liters of culture were resuspended in 120 ml of 50 mM Tris-HCl, pH 9.0, containing 5 mM EDTA and subjected to lyzozyme treatment (0.2 mg/ml) for 15 min. The bacterial extract was sonicated three times for 30 s and then diluted to 190 ml with buffer containing 5 mM MgCl 2 and 5 units/ml benzonase (Merck). Ammonium sulfate was then added to 1 M final concentration, and the protein extract was centrifuged at 20,000 ϫ g for 25 min. The supernatant fraction was applied to a 30-ml phenyl-Sepharose column (Amersham Pharmacia Biotech) preequilibrated with 1 M ammonium sulfate in Tris buffer. A red-brown protein remained bound on top of the column. After extensive washing with buffered 1 M ammonium sulfate, the red protein was eluted with Tris-HCl, pH 9.0, and then diluted 10-fold in this buffer and passed through a 40-ml column of DEAE-cellulose (DE52, Whatman). The unbound protein fraction containing XylT was adjusted to 1 M ammonium sulfate and then applied to a 15-ml column of phenyl-Sepharose. The XylT fraction was eluted in a small volume of Tris buffer and then loaded on a gel filtration column (AcA 54, Biosepra), Ø ϭ 2.5 ϫ 110 cm, equilibrated with Tris buffer containing 0.2 M NaCl. The column was developed at a flow rate of 20 ml/h. The XylT fraction was concentrated to approximately 8 ml by ultrafiltration through an 8-kDa Omega membrane (Pall Filtron). XylT was stored as pellets in liquid nitrogen.
Purification of XylE-Bacterial cells were rinsed with 25 mM Tris-HCl, pH 7.5, containing 10% isopropyl alcohol (v/v) and 1 mM EDTA (hereafter referred to as buffer A). Cells were broken by sonication as described for XylT purification, and the resulting preparation was centrifuged for 20 min at 20,000 ϫ g. The supernatant fraction was loaded onto a 40-ml DEAE-cellulose column (DE52), which was sequentially developed by solutions of buffer A containing NaCl at increasing concentrations from 0.1 to 0.4 M (0.1 M increments). Dioxygenase activity was recovered in fractions eluting between 0.3 and 0.4 M NaCl. These fractions were pooled and adjusted to 1 M ammonium sulfate and then applied to a 40-ml phenyl-Sepharose column (Amersham Pharmacia Biotech). The column was developed with a linear gradient from 1 to 0 M ammonium sulfate in buffer A. Active fractions were loaded on a 25-ml column of Q-hyper D (Biosepra), which was developed with a linear gradient from 0 to 1 M NaCl. XylE eluted at approximately 0.6 M NaCl. It was concentrated by ultrafiltration through a 30-kDa Omega membrane and stored in liquid nitrogen. The purified preparations thus obtained exhibited specific activities of between 125 and 175 units/mg.
Purification and Immobilization of Rabbit XylT Antibodies-An antiserum was raised in rabbits against purified recombinant XylT. XylTspecific IgG antibodies were purified from the antiserum by immunoaffinity according to a previously published procedure (13). Briefly, XylT (0.3 mg in SDS mixture) was adsorbed onto pieces of nitrocellulose membrane (7 cm 2 ). After washing, the XylT-coated membrane was incubated with 0.5 ml of antiserum for 2 h at room temperature. After extensive washing, antibodies were eluted from the membrane by a solution of 5 mM glycine-HCl, 0.5 M NaCl, 0.5% Triton X-100, pH 2.3, and the eluate was immediately neutralized. The antibody preparation was used in Western blot analysis at a 1:250 dilution. Anti-XylT IgG was also immobilized on protein A-Sepharose (Sigma). A sample equivalent to 0.15 ml of packed resin was incubated with 0.4 ml of antiserum in two Spin-X centrifuge filter units (Costar) for 1 h at room temperature under constant agitation. The resin was washed three times with 0.5 ml of 0.2 M sodium borate, pH 9. Bound IgG antibodies were cross-linked to protein A by incubation for 30 min with 20 mM dimethylpimelimidate (Sigma) in 0.5 ml of borate buffer. The resin was washed and incubated for 2 h in 0.2 M ethanolamine, pH 8.0, and finally stored in 10 mM phosphate, 140 mM NaCl, 2.7 mM KCl, 0.5% Tween 20, pH 7.4.
Immunochemical Analysis of XylT in P. putida Extracts-P. putida cultures (200 ml) in midlog phase (A 600 ranging between 1.0 and 1.5) were harvested by centrifugation for 15 min at 15,000 ϫ g. Bacterial pellets were washed and resuspended in 0.5 ml of 10 mM potassium phosphate containing 10 mM EDTA, pH 7.5. Bacteria were disrupted by ultrasonication for 2 min on ice. The crude extracts were then centrifuged for 30 min at 220,000 ϫ g in a Beckman TL-100 ultracentrifuge. Samples of the high speed supernatant fractions, equivalent to 3.4 mg of protein, were incubated with immobilized anti-XylT IgG (20 or 40 l) in a total volume of 0.3 ml. Incubation was carried out for 1 h at 4°C in Spin-X centrifuge filter units (Costar) maintained under constant agitation (Eppendorf thermomixer). The beads of immobilized IgG-Sepharose were washed three times with 0.5 ml of 10 mM potassium phosphate, 0.15 M NaCl, 0.1% Triton X-100, pH 7.5. Bound proteins were then eluted by treatment for 5 min at 95°C in 25 l of SDS mixture. Samples of the eluted proteins were separated by electrophoresis on a 16% polyacrylamide gel containing 13% glycerol in a Tris-Tricine 1 buffer system (14). Proteins were electroblotted on a nitrocellulose membrane as described previously (15), which was then developed with purified anti-XylT IgG (1:250 dilution) and a goat anti-rabbit peroxidase conjugate (1:10,000 dilution; Sigma) as primary and secondary antibodies, respectively. The blot was revealed with an enhanced chemiluminescence kit (Amersham Pharmacia Biotech).
Analytical Methods-UV-visible absorption spectrophotometry, EPR spectroscopy, and estimation of protein-bound iron were performed as described previously (16). For preparation of the nitrosyl complex of the dioxygenase, enzyme samples were made anaerobic under an argon phase, transferred into EPR tubes, and slowly bubbled with NO gas for 3 min.
Proteins were analyzed by slab gel electrophoresis under denaturing conditions (SDS-PAGE) according to Jouanneau et al. (13). Protein concentrations were determined using the BCA assay (Pierce).
XylE activity was assayed as described by Nozaki et al. (14). Briefly, 10 -50-l samples were mixed in a cuvette containing 3 ml of 0.330 mM catechol in 50 mM potassium phosphate, pH 7.5. The increase in absorption at 375 nm was recorded over 30 s using an HP8452 spectrophotometer.
Measurement of Redox Potential-Redox titration of XyT was performed spectrophotometrically using 5-deazaflavin as a photoreductant and benzyl-viologen as a dye indicator (EЈ 0 ϭ Ϫ320 mV versus normal hydrogen electrode). Spectrophotometric measurements were made at 25°C inside a glove box (Jacomex) containing argon purified to less than 1 ppm O 2 (15). A 1-cm light path cuvette contained, in a 820-l volume, 5 M 5-deazaflavin, 15 mM sodium oxalate, 50 M benzylviologen, and 27 M XylT in 50 mM Tris-HCl, pH 8.5. The mixture was exposed to the light of a slide projector for short intervals of time (5-60 s), and absorbance changes were monitored after 30-s equilibration in darkness. The percentage of XylT reduction was determined from absorbance readings at 440 nm (an isosbestic point between the oxidized and reduced forms of benzyl-viologen), using ⑀ 440 (ox) ϭ 8.60 mM Ϫ1 ⅐cm Ϫ1 and ⑀ 440 (red)ϭ 4.885 mM Ϫ1 ⅐cm Ϫ1 for the oxidized and reduced forms of XylT, respectively. The extent of benzyl-viologen reduction was calculated from the absorption at 600 nm after correction for the absorption of ferredoxin at this wavelength: ⑀ 600 (ox) ϭ 3.51 mM Ϫ1 ⅐cm Ϫ1 , and ⑀ 600 (red) ϭ 2.69 mM Ϫ1 ⅐cm Ϫ1 for the oxidized and reduced forms of XylT, respectively.
Inactivation and Reactivation of XylE-The conditions of inactivation were adapted from Bartels et al. (6) and Wasserfallen (10). Catechol 2,3-dioxygenase was diluted to 50 nM in a total volume of 500 l of 0.1 M potassium phosphate, pH 7.5, containing 1% isopropyl alcohol and incubated with 400 M 4-methylcatechol for 30 min at room temperature. In control experiments, an equivalent amount of enzyme was treated similarly, except that 4-methylcatechol was omitted. After extensive dialysis twice against 1000 volumes of 0.1 M potassium phosphate, pH 7.5, containing 10% isopropyl alcohol, the enzyme samples were made anaerobic in 4-ml glass vials and kept under argon during reactivation.
Chemical reactivation of XylE was carried out in the presence of 1 mM ferrous iron and 1 mM cysteine (17). XylT-dependent reactivation was performed under argon in 4-ml vials containing 50 nM inactivated XylE, 2 M 5-deazaflavin, 1 mM glycine, 0.1 M potassium phosphate, pH 7.5, and a variable concentration of XylT in a final volume of 200 l. Reaction vials were kept at 20°C in a thermostatted water bath and were illuminated by the light of a slide projector for defined time periods. After light exposure, the samples were allowed to equilibrate for 30 s in the dark before measuring XylE activity. Assays were done in triplicate on 10 l of reaction mixture. Ferredoxins other than XylT were tested under the same conditions.
For EPR measurements, XylE was inactivated with 100 mM or 200 mM 4-methylcatechol, dialyzed extensively, and then concentrated to a final concentration ranging from 105 to 126 M. Part of the preparation was subjected to reactivation for 5-10 min in the light in the presence of 10 mM glycine, 2.5 M 5-deazaflavin, and XylT, as indicated. Another part, used as a control, was incubated under identical conditions except that XylT was omitted.

RESULTS
To facilitate isolation of the xylT gene product, its coding sequence was cloned in the expression vector pET9a under the transcriptional control of the strong T7 promoter. While verifying the DNA sequence of xylT, we found the first two bases of codon 76 to be inverted compared with the previously reported sequence (GC instead of CG) (9). This inversion would result in an alanine-to-glycine change in the deduced polypeptide sequence. We observed the same difference upon sequencing the xylT region of plasmid pAW31. The new xylT sequence resembles more closely those of four homologous genes currently in the EMBL/DDBJ/GenBank TM data base.
Expression of xylT in the E. coli strain BL21(DE3) gave rise to the production of a polypeptide of the expected size (approximately 15 kDa) that was visualized as a prominent band by SDS-PAGE analysis of whole cell lysates. Since the xylT gene product contains a potential binding site for a [2Fe-2S] cluster (9), we expected the recombinant protein to be red-brown like ferredoxins containing this type of cluster. However, the protein extract prepared from E. coli hyperexpressing xylT at 37°C was hardly colored. This suggested that the recombinant protein accumulated essentially in the apoform in E. coli. Induction of xylT expression at 28°C results in the formation of a greater amount of the recombinant protein as the holoform. This was confirmed by EPR measurements of the signal produced by the Fe-S center of recombinant XylT (see below) made directly on dithionite-reduced crude extracts of E. coli. The signal intensity was approximately 4-fold greater with extracts from cells grown at 28°C than with cells grown at 37°C, whereas a control extract from uninduced E. coli gave a negligible signal. Hyperexpression of xylT was therefore carried out at 28°C. Due to the instability of the recombinant protein in air (see below), purification was performed under strict anaerobiosis. The xylT gene product could be purified to near homogeneity by the three-step procedure illustrated in Fig. 1. An average of 1.0 mg of purified XylT was recovered per liter of culture with a yield close to 10%.
Molecular Properties of XylT-The purified XylT protein migrated as a single band on SDS-PAGE with an apparent M r of 15,000 (Fig. 1). N-terminal sequence analysis clearly identified this protein band as the xylT gene product and showed that the initial methionine was not removed in E. coli. The expected molecular mass of the XylT polypeptide as deduced from its polynucleotide sequence is 12,034 Da. The molecular mass of the native XylT protein estimated by gel filtration was about 12 kDa, indicating that it is monomeric. The XylT polypeptide is characterized by a predominance of basic over acidic residues (9) and has a theoretical isoelectric point of 8.27 (Table I).
Although this value was not confirmed experimentally, the basic character of XylT was consistent with its lack of binding to DEAE-cellulose in the 7-9 pH range and its retention on a strong anion exchange column only at pH 9.0 (Fig. 1, lane 2). The XylT protein is relatively unstable due to its sensitivity to oxygen and temperature. The protein denatures when exposed to air, resulting in both a loss of its absorbance in the visible region and its precipitation. From the decrease in its absorption at 416 nm, which followed first order kinetics, the half-life of XylT in air was estimated to be 69 min at ambient temperature. Based on the same criterion, XylT appeared to be unstable at 37°C (Table I), consistent with the finding that it was mainly present in apoform in E. coli grown at this temperature.
Spectroscopic Properties-The UV-visible absorption spectra of the oxidized and reduced forms of XylT are presented in Fig.  2. Protein purified under reducing conditions spontaneously oxidized upon removal of dithionite by anaerobic gel filtration, probably because of residual oxygen present in the argon (approximately 200 ppm). The spectrum of oxidized XylT exhibited three maxima due to the chromophore at 336, 416, and 456 nm, as well as a shoulder near 540 nm. The absorption of the polypeptide at 278 nm is not very pronounced, consistent with the absence of tryptophan and the low content of aromatic residues in XylT sequence. Upon reduction, a general decrease of the absorption in the visible region occurred, and a peak appeared at 540 nm (Fig. 2). Based on protein determination using the BCA assay, and given the molecular mass of XylT (12 kDa), the absorption coefficient of the oxidized form was calculated to be 9.52 mM Ϫ1 ⅐cm Ϫ1 at 416 nm. All of these properties are similar to those of other [2Fe-2S] ferredoxins (18). Determination of the iron content of XylT indicates the presence of one [2Fe-2S] cluster per polypeptide chain (Table I).
EPR analysis of purified XylT in the reduced form gave a signal centered at g ϭ 1.94 with general rhombic symmetry that was detectable at temperatures at least as high as 40 K. This signal may be attributed to a [2Fe-2S] cluster with an S ϭ 1 ⁄2 spin state (Fig. 3). Double integration of the EPR signal of spectrum II in Fig. 3 and comparison with a standard, i.e. the [2Fe-2S] ferredoxin from Rhodobacter capsulatus (19), gave a value of about 0.9 spin/mol of protein, thus providing further evidence that XylT contains one [2Fe-2S] cluster.
The line shape of the EPR signal was, however, unusual for this type of cluster in that the resonance lines were relatively broad, and, as a consequence, the g x component of the tensor was not well resolved (Fig. 3, spectrum II). The broadening might reflect some microheterogeneity of the protein preparation in the vicinity of the Fe-S cluster. Considering the relative instability of XylT, the question arose whether the anomalies detected by EPR spectroscopy resulted from conformational changes in the protein during the purification. This hypothesis was assessed by recording EPR spectra in the E. coli soluble extract (Fig. 3, spectrum I) as well as at the subsequent steps of the purification (data not shown). The signal observed did show small differences compared with that of purified XylT (Fig. 3, spectrum II), indicating that XylT may have undergone minor changes during isolation. The EPR spectrum of the protein in the cell extract (Fig. 3, spectrum I) exhibited a greater apparent complexity, which was primarily due to the contribution of E. coli proteins. Interestingly, the addition of glycerol to purified XylT protein before freezing the EPR tube caused a significant change in the high field region of the spectrum (Fig. 3, spectrum  III), reflecting a more pronounced rhombic symmetry with g x,y,z ϭ 1.88, 1.94, and 2.04. Hence, the EPR data give clear evidence that XylT can adopt at least two slightly different conformations at low temperature. It is unknown whether the heterogeneity detected in the frozen XylT sample reflects that of a sample at ambient temperature. The addition of glycerol to XylT had no detectable effect on the absorbance spectrum of the oxidized form. It is nevertheless plausible that the two signals detected by EPR with purified XylT (Fig. 3, spectra II and III) correspond to two conformations that could have been trapped upon freezing. These properties seem to denote an unusual flexibility of the protein around the cluster binding site.
To further characterize the [2Fe-2S] cluster of XylT, its redox potential was determined by spectrophotometric titration using benzyl-viologen as potential indicator. The calculated midpoint redox potential was Ϫ373 Ϯ 6 mV at pH 8.5 and 25°C (Fig. 4).
XylT-dependent Reactivation of Catechol 2,3-Dioxygenase-To examine reactivation of XylE by XylT, XylE was purified and subjected to controlled inactivation. Catechol 2,3-dioxygenase is inactivated by exposure to air or during the catalytic cycle, especially when substituted catechols such as 3and 4-methylcatechol are substrates (20,21). Inactivation is thought to be caused primarily by the oxidation of the ferrous iron present at the active site of the enzyme, but ultimately the oxidized iron atom may be released from the enzyme (10). The preparation of inactive enzyme used in the experiments described below was obtained by incubating purified XylE with 4-methylcatechol in buffer containing 1% isopropyl alcohol. Such a treatment resulted in 90 -95% inactivation in 30 min. The dioxygenase preparation thus obtained could be fully reactivated by subsequent incubation with ferrous iron and cysteine, conditions that are known to promote enzyme recovery in vitro (7,17). The rate of chemical reactivation is rather slow and takes about 1 h for completion. A slow reactivation was also obtained by incubation of the inactive dioxygenase with 5-deazaflavin, which generates strong reductants when exposed to light. In contrast, incubation of the dioxygenase with both 5-deazaflavin and stoichiometric amounts of XylT promoted rapid reactivation of the enzyme (Fig. 5). The rate as well as the extent of the reactivation was dependent on the XylT concentration. In control experiments, it was found that the dioxygenase remained inactive when incubated with XylT in the dark or with heat-denatured XylT (Table II). These results indicate that the XylT-dependent reactivation did not merely result from substitution of the ferric iron in the active site of the enzyme by ferrous iron originating from the Fe-S cluster of XylT. Rather, the requirement of a source of reductant (the photoactivated deazaflavin) suggested that reactivation of the dioxygenase involved an electron transfer from XylT  to the active site of the enzyme. When XylT was replaced by either spinach ferredoxin or a R. capsulatus ferredoxin (FdVI) at concentrations 10-fold higher than that of the dioxygenase, no significant reactivation was detected (data not shown). Spinach ferredoxin and R. capsulatus FdVI are representative members of two major subgroups of [2Fe-2S] ferredoxins, those found in plants, on the one hand (22), and those similar to adrenodoxin, on the other (23,24). Since neither of the two ferredoxins was competent in reactivating XylE in vitro, it may be concluded that XylT catalyzes a specific reactivation.
XylT-mediated Reduction of the Active Site of the Dioxygenase-The experiments described above demonstrated that XylE reactivation requires a source of reductant and XylT as mediator, suggesting that enzyme recovery occurs through reduction of the iron atom in the catalytic site. To demonstrate that such a reduction takes place when the dioxygenase is reactivated by XylT, we have monitored the redox state of the mononuclear iron center of XylE by EPR spectroscopy. The active site of the native enzyme contains a high spin Fe(II), which is EPR-silent (3). In contrast, the 4-methylcatechol-inactivated enzyme preparation gave an EPR spectrum with dominant features at g ϭ 7.54, 4.4, and 1.91 as well as a smaller signal at g ϭ 5.82, indicative of the presence of a ferric iron with an S ϭ 5 ⁄2 ground state (Fig. 6a). In control experiments, it was shown that the signal observed actually came from the enzyme-bound iron and not from a chelate of free iron with 4-methylcatechol or its product. The EPR signal of the 4-methylcatechol-inactivated dioxygenase was clearly distinct from the signal arising from a nonspecific oxidation of the active site iron, provoked, for example, by H 2 O 2 treatment (H 2 O 2 -inactivated enzyme gave an EPR signal centered near g ϭ 4.3; data not shown). Thus, the signal shown in Fig. 6a might denote the presence of a ligand molecule resulting from the oxidation of 4-methylcatechol remaining bound to the enzyme catalytic site, despite the extensive dialysis step that preceded the EPR analysis. A preparation of inactivated dioxygenase, having about 7% residual activity, was subjected to reactivation by XylT at a XylT/XylE molar ratio of 1:10. The specific activity of the enzyme increased up to 50% of the initial activity within 5 min. EPR analysis of the reactivated sample (Fig. 6b) revealed a complete disappearance of the enzyme-bound Fe(III) signal. This result suggests that XylT reactivates XylE through reduction of Fe(III) to Fe(II) in the enzyme active site. To assess this as- FIG. 4. Redox titration of XylT. Reductive titration of XylT was carried out by stepwise photoreduction using 5-deazaflavin as a catalyst (see "Experimental Procedures"). Intermediate redox potentials (Eh) were determined from the extent of benzyl-viologen reduction. The ratio of the oxidized (XylTox) and reduced (XylTred) forms of XylT was calculated from the absorption at 440 nm. Data points could be correlated through a log curve fit with a regression coefficient R of 0.95. The deduced EЈ 0 of XylT is Ϫ373 Ϯ 6 mV at pH 8.5.

FIG. 5. Kinetics of XylT-dependent XylE reactivation.
Samples of inactivated XylE were incubated in anaerobic vials in a total of 200 l of reactivation mixture (see "Experimental Procedures"). Vials contained 50 nM XylE and the following concentration of XylT: 5 nM (diamonds), 50 nM (squares), or 250 nM (circles). A control vial contained no XyT (triangles). At time 0, vials were exposed to light. XylE activity was assayed on 5-50-l samples anaerobically withdrawn from the vials at the times indicated.  6. EPR spectra of XylE inactivated by 4-methylcatechol and reactivated by XylT. A sample of XylE (126 M) was inactivated by 100 mM 4-methylcatechol then dialyzed and concentrated as described in detail under "Experimental Procedures." After recording the EPR spectrum (spectrum a), the same sample was subjected to reactivation by XylT (11 M) for 5 min and again analyzed by EPR (spectrum b). Note that XylE was diluted to 110 M in the reactivation mixture. Spectra were recorded at 4 K under the conditions given in Fig. 3 except that microwave power was 1 milliwatt and microwave frequency was 9.652 GHz. A narrow signal at g ϭ 4.3, best observed in spectrum b, is due to free Fe(III). The feature around g ϭ 2 noted by an asterisk is certainly due to contaminating Cu 2ϩ . sumption, the following series of experiments was carried out.
First, we checked whether the extent of Fe(III) reduction as estimated by EPR analysis correlated with the extent of enzyme activity recovery. A sample of inactivated dioxygenase showing 14.3% residual activity (25 units/mg protein) was subjected to EPR analysis and spin quantitation. For this purpose, the low field component of the signal at g ϭ 7.54 was integrated and compared with the corresponding signal of the beef liver catalase used as a reference (25). This estimation gave a minimal value of 0.38 Fe(III)/tetramer. By chemical assay, the enzyme sample was found to contain 0.75 Fe atom/mol, suggesting that part of the iron was in the reduced Fe(II) form. After reactivation, the enzyme activity reached 85 units/mg protein, which corresponds to reactivating about one-third of the inactivated enzyme molecules. Thus, there is a good correlation between the amount of Fe(III)-containing dioxygenase and the extent of XylT-mediated reactivation achieved. We noticed that XylE inactivation was always accompanied by a variable but significant loss of iron. For example, in the experiment described above, iron loss accounted for 56% of the initial iron content of the enzyme. Since the apoenzyme was not reactivated in the XylT-dependent reaction described in this study, it follows that only a fraction of the inactive dioxygenase can be reactivated in the in vitro system employed here.
Second, experiments were performed to monitor the reduction status of the iron atom at the dioxygenase active site upon XylT-dependent reactivation. Nitric oxide binds the Fe(II) atom of dioxygenase to form a nitrosyl complex with an electronic spin of S ϭ 3 ⁄2 (26), a property that was exploited to monitor the changes in the nitrosyl-Fe(II) complex signal intensity upon reactivation of XylE. The same samples that gave the spectra shown in Fig. 6 as well as the uninactivated enzyme were equilibrated with NO gas and analyzed by EPR (Fig. 7). The uninactivated enzyme sample exhibited a two-component signal with g x ϭ 4.16 and g y ϭ 3.91, thereafter referred as signal A (Fig. 7d). The enzyme inactivated with 4-methylcatechol gave a signal (Fig. 7a) composed of a signal similar to signal A and a minor signal with g x ϭ 4.1 and g y ϭ 3.99 (signal B). Several S ϭ 3 ⁄2 species had already been observed for the NO complex of catechol 2,3-dioxygenase (26). Analysis of another inactivated enzyme sample, which had completely lost dioxy-genase activity (residual activity Ͻ1%), yielded no detectable Fe(II)-NO signal, suggesting that the signal shown in Fig. 7a arose from the fraction of enzyme that was still active. After XylT-dependent reactivation, the signal of the nitrosyl complex globally increased in intensity (Fig. 7b) compared with that of the inactive sample, indicating that XylT mediated reduction of Fe(III) to Fe(II) at the enzyme active site. This result extends our observation that the Fe(III) signal of the inactivated enzyme disappeared upon reactivation and provides further evidence that enzyme reactivation occurred through reduction of the iron atom at the active site. The calculated difference between spectrum a and spectrum b, given as trace c in Fig. 7, may be seen as the superposition of three signals: a major signal B, a minor signal A, and a new signal (signal C) with g x ϭ 4.17 and g y ϭ 3.88. Some of the signals we observed might result from the binding at the XylE active site of a ligand resulting from oxidation of 4-methylcatechol. Existence of such a ligand was already suggested by the analysis of the Fe(III) signal of the inactive enzyme (Fig. 6a).
In addition, we have considered the possibility that XylT might reactivate XylE by transferring Fe(II) atoms, instead of electrons, to the inactive enzyme. We have measured the EPR signal intensity of the dithionite-reduced XylT cluster before and after XylE reactivation and found it to be unaltered by the XylE reactivation procedure. This rules out the possibility that the ferredoxin functions by transferring iron atoms to the enzyme active site.
Estimation of XylT Content of P. putida mt2-Preliminary attemps to directly detect XylT in crude extracts of P. putida mt2 by Western blot analysis were unsuccessful, because the sensitivity of the method (about 5 ng/lane on a polyacrylamide slab gel) was not adequate to visualize the very low level of XylT in the extract. It was therefore necessary to specifically concentrate XylT by immunoaffinity capture from the soluble protein fraction on immobilized anti-XylT IgG (see "Experimental Procedures"), followed by Western blot analysis. Material recovered in this way from 3.4 mg of soluble protein from P. putida mt2 gave a detectable XylT-specific signal (Fig. 8, lanes  4 and 5). No signal was observed in the case of the control strain KT 2442, which lacks the pWWO plasmid. The level of the XylT protein in P. putida mt2 was tentatively estimated from the signal intensity observed on the Western blot. Given that about 15 ng of XylT was recovered from a sample containing 0.7 mg of soluble protein (Fig. 7, lane 5), XylT would account for 0.0021% of the soluble protein. This was also of interest to determine the XylT/XylE molar ratio in P. putida protein extracts. The amount of XylE was estimated indirectly from activity assays, based on the assumption of an average specific activity of 300 units/mg for the purified enzyme (26). The XylE activity in the P. putida soluble extract was 0.85 Ϯ 0.09 units/mg (n ϭ 3), from which we calculated a XylT/XylE FIG. 7. EPR analysis of the nitrosyl complexes of inactivated and reactivated forms of XylE. The two XylE samples (inactivated and reactivated) already analyzed as indicated in Fig. 6, were equilibrated with NO gas. Spectrum a, nitrosyl complex of inactivated XylE; spectrum b, nitrosyl complex of XylT-reactivated XylE; spectrum d, nitrosyl derivative of untreated XylE. Trace c was obtained by subtracting spectrum a from spectrum b after the two spectra were normalized to the same XylE concentration. Recording conditions were as in Fig. 6. Gain was 5-fold greater in a and b than in d. ratio of 0.091. This figure should be considered only approximate, since both values are minimum estimates; on the one hand, the yield of XylT capture by immunoaffinity may be less than 100%, on the other hand, not all of the XylE molecules in the P. putida extract may be fully active. DISCUSSION Hyperexpression of the xylT gene of the TOL plasmid pWWO in the heterologous E. coli host allowed us to purify and characterize the XylT protein as a novel [2Fe-2S] ferredoxin. Based on UV-visible absorbance and EPR spectroscopy, XylT exhibited properties grossly similar to plant-type ferredoxins (27). These similarities reflect the common nature of their prosthetic group, a [2Fe-2S] cluster with rhombic EPR signal. As revealed by resolution of the crystal structure of model plant-type ferredoxins, the cluster is bound to the polypeptide chain through four cysteinyl ligands, which have a defined and conserved spacing in the sequence (28,29). The XylT sequence contains six cysteines, four of which, located at positions 41, 46, 49, and 81, conform to the conserved spacing and would therefore serve as potential ligands of the cluster (9). However, the possibility that cysteine 36 serves as an alternative ligand cannot a priori be ruled out.
On the other hand, XylT differs from other [2Fe-2S] ferredoxins in many respects. Its overall amino acid sequence shows little similarity with known ferredoxins. A unique feature of the XylT polypeptide is the predominance of basic over acidic residues, which confers on the protein a net positive charge at neutral pH. This is the first known example of a ferredoxin with a basic character, and this property might be related to its specific function. We have shown that representative members of the two major classes of [2Fe-2S] ferredoxins failed to substitute for XylT in XylE reactivation. In the case of R. capsulatus FdVI, an analogue of the Pseudomonas putidaredoxin (24), the high value of its midpoint redox potential (EЈ 0 ϭ Ϫ255 mV) 2 compared with that of XylT could explain its lack of activity. On the other hand, the EЈ 0 value of the spinach ferredoxin (Ϫ420 mV; Ref. 30) is close to that of XylT and thus cannot account for its absence of reactivity toward XylE. This plant ferredoxin is, however, known for its very acidic isoelectric point (pI ϭ 5.4) and a highly negative net charge, which might hinder productive interaction with XylE. It may thus be inferred that the positive net charge of the XylT protein is a prerequisite for its molecular interaction with XylE. This assumption is currently being evaluated through cross-linking experiments between XylT and XylE. The unusual basic character of XylT, and its specificity with respect to XylE reactivation, could also explain why in P. putida, which most likely contains other ferredoxins, XylT is so essential for maintenance of XylE activity and hence bacterial growth on para-substituted derivatives of toluene (8).
XylT exhibits a number of other distinctive properties. Although its cluster resembles that of plant-type ferredoxins, in that it gives a rhombic EPR signal, XylT absorbance in the visible region of the spectrum is closer to that of adrenodoxin (31) and putidaredoxin (32), which instead yield a nearly axial EPR signal (33). The EPR analysis revealed that XylT has intermediate properties, which might reflect greater flexibility of the protein around the cluster. In addition, the very strong effect of glycerol on the EPR signal suggests that the [2Fe-2S] cluster is exposed to the solvent. In accord with this assumption, and in contrast to most [2Fe-2S] ferredoxins, XylT is a relatively unstable protein exhibiting a considerable propensity to lose its cluster, especially when exposed to air. This property might appear somewhat paradoxical for a protein involved in the repair of an enzyme that has sustained oxidative damage. However, the intracellular microenvironment of aerobic microorganisms is characterized by a fairly low oxygen tension due to the membrane-borne respiratory chain, and bacterial cells might thus afford conditions conducive to XylT stability. In this respect, aerobically grown E. coli bacteria turned out to be a convenient system for the biosynthesis of XylT and its purification. The protein yield, about 1 mg/liter of culture, was, however, relatively low compared with the heterologous expression of more stable [2Fe-2S] ferredoxins like human ferredoxin (34), Anabaena ferredoxin (35), R. capsulatus FdIV and FdV (19,36), or Fdx1 from Sphingomonas sp. RW1 (37).
XylT is also novel in its function. We have demonstrated that XylT specifically mediates the reduction of the active site iron in inactive catechol 2,3-dioxygenase, which results in the reactivation of the enzyme. This result confirms and extends the prediction made by Polissi and Harayama (8) on XylT function, based on the characterization of xylT deletion mutants.
In the majority of cases, ferredoxins function as electron carriers in association with oxidoreductases and participate in this way in the oxidation or reduction of various substrates or metabolites (27). XylT therefore represents a rare example of a ferredoxin that does not provide electrons for the reduction of a substrate but rather for the reactivation of an enzyme. An analogous case of ferredoxin-mediated enzyme protection has been described for Azotobacter vinelandii (38,39). A [2Fe-2S] ferredoxin in this bacterium was found to confer protection against oxygen-mediated inactivation upon nitrogenase. Although the mechanism of this protection is not known in detail, it was shown that A. vinelandii [2Fe-2S] ferredoxin forms a tight complex with nitrogenase that is more oxygen-stable than the free enzyme (39). Hence, in contrast to XylT, the A. vinelandii ferredoxin does not reactivate an enzyme (nitrogenase) but rather prevents its inactivation. XylT therefore appears to be the first ferredoxin reported to catalyze the reactivation of an enzyme in a reaction that shows a relatively narrow specificity. It may thus be considered to be an activase.
The extent of XylT-dependent reactivation of various preparations of XylE was somewhat variable. This is assumed to reflect the variable degree of irreversible inactivation of different enzyme preparations through loss of iron. We indeed showed that the relatively harsh conditions employed to obtain inactivation in vitro caused release of iron from a substantial proportion of enzyme molecules. We also showed that XylT can only reactivate the fraction of inactive enzyme that retains iron. Another plausible explanation of variable dioxygenase inactivation would be a tight binding of the product or an intermediate of 4-methylcatechol oxidation at the enzyme active site, which might hinder reactivation by XylT.
Evidence that XylT acts as the XylE-specific activating protein in the natural host is provided by immunochemical demonstration of the XylT product in P. putida extracts. The level detected was extremely low. However, given that catalytic amounts of XylT were sufficient to reactivate XylE in vitro, it seems reasonable to predict that XylT should efficiently reactivate XylE in vivo. It remains to be seen whether increasing the in vivo level of XylT, by hyperexpression of its structural gene, might improve the biodegradation of toluene or analogous molecules by P. putida.
Since the discovery of the xylT gene (9) 113 residues) and share a high degree of similarity with XylT. In particular, the XylT analogues are predicted to contain the four conserved cysteines that are potential ligands of the [2Fe-2S] cluster, as well as two additional conserved cysteines at positions equivalent to Cys 19 and Cys 36 in XylT. This suggests that the latter two cysteines might form a disulfide bond. In addition, their sequences indicate that the XylT analogues have basic isoelectric points, located between 7.5 and 8.4. This information, combined with the fact that the xylT gene homologues are located upstream of extradiol dioxygenase-encoding genes, indicates that these analogues have a similar function. The XylT-like proteins appear as a new family of ferredoxins that might have practical relevance if they can be used to engineer microorganisms with improved capabilities in bioremediation.