Isolation and Biochemical Characterization of Hypophosphite/ 2-Oxoglutarate Dioxygenase A NOVEL PHOSPHORUS-OXIDIZING ENZYME FROM PSEUDOMONAS STUTZERI WM88*

The htxA gene is required for the oxidation of hypophosphite in Pseudomonas stutzeri WM88 (Metcalf, W. W., and Wolfe, R. S. (1998) J. Bacteriol . 180, 5547– 5558). Amino acid sequence comparisons suggest that hypophosphite:2-oxoglutarate dioxygenase (HtxA) is a novel member of the 2-oxoglutarate-dependent dioxygenase enzyme family. To provide experimental support for this hypothesis, HtxA was overproduced in Escherichia coli and purified to apparent homogeneity. Recombinant HtxA is identical to the native enzyme based on amino terminus sequencing and mass spectral analysis, and it catalyzes the oxidation of hypophosphite to phosphite in a process strictly dependent on 2-oxoglut-arate, ferrous ions, and oxygen. Succinate and phosphite are stoichiometrically produced, indicating a strict coupling of the reaction. Size exclusion analysis suggests that HtxA is active as a homodimer, and maximal activity is observed at pH 7.0 and at 27 °C. The apparent K m values for hypophosphite and 2-oxoglut- arate were 0.58 (cid:1) 0.04 m M and 10.6 (cid:1) 1.4 (cid:2) M , respectively. V max and k cat values were determined to be 10.9 (cid:1) 0.30 (cid:2) mol min (cid:3) superfamily of enzymes and catalyzes the oxidation of hypophosphite to phosphite requiring 2-oxoglutarate and oxygen as cosubstrates in addition to ferrous ions for activity. The reaction proceeds by the incorporation of one of the atoms of dioxygen into hypophosphite to form phosphite. The other atom of dioxygen reacts with 2-oxoglutarate, resulting in oxidative decarboxylation giving rise to succinate and CO 2 . PtxD is a member of the D -isomer-specific 2-hydroxy acid NAD-dependent dehydrogenase protein family (15). This enzyme completes the oxidation of hypophos- phite by oxidizing phosphite to phosphate using NAD (cid:1) as a cofactor. Determine the of Molecular Amino Terminus Sequencing and Mass Spectrometry— Partially pu- rified recombinant and native HtxA were separated from contaminat-ing proteins on a 12% Tris-HCl precast polyacrylamide gel (Bio-Rad) under denaturing conditions. The protein was transferred to a Sequi-Blot polyvinylidene difluoride membrane (Bio-Rad) using a Mini Trans- Blot Electrophoretic Transfer Cell (Bio-Rad) in 10 m M CAPS, 10% methanol buffer, pH 11.0. Protein bands were visualized with Coomas- sie Blue, and the membranes were submitted to the University of Illinois Protein Sciences Facility for amino terminus sequencing by Edman degradation. Matrix-assisted laser desorption ionization mass spectrometry was performed at the University of Illinois Spec- trometry facility using a Voyager-DE STR mass spectrometer (PerSep-tive Biosystems). Enzyme Activity Assays— purification and characterization of HtxA, a continuous coupled spectrophotometric assay using MBP-PtxD as a reporter enzyme used. Activity NADH deter- mined using the PtxD-coupled assay in 20 m M MOPS, pH 7.0, containing 1 m M NAD (cid:1) , 0.25 m M 2-oxoglutarate, 40 (cid:3) M FeCl 2 , 391 (cid:3) g of PtxD, 3.9 (cid:3) g of HtxA, and 0.18–3.0 m M hypophosphite. B , a substrate concentration curve for 2-oxoglutarate was constructed, and the apparent K m for 2-oxoglutarate was determined. The activity assays were done using the PtxD-coupled assay in a 5-cm path length cuvette. The assay contained 20 m M MOPS, pH 7.0, 10 m M hypophosphite, 1 m M NAD (cid:1) , 26 (cid:3) M FeCl 2 , 1.5 units of PtxD, 3.9 (cid:3) g of HtxA, and a 2-oxoglutarate concentration range of 5.26–100 (cid:3) M . All assays were performed in triplicate at room temperature. The kinetic constants were determined with the kinetic analysis program KINSIM (53). by succinate in an end under “Experimental procedures.” The results shown the average of two experiments.

The current view of phosphorus metabolism dictates that, unlike other elements essential for growth, phosphorus does not undergo a biologically catalyzed oxidation-reduction cycle in nature. The biochemistry involving this essential and often limiting nutrient is typically thought to be restricted to the formation and hydrolysis of phosphate esters, in which phosphorus exists in its most oxidized state (P 5ϩ ). However, the number of microorganisms capable of using reduced phosphorus compounds as the sole source of phosphorus or able to synthesize reduced phosphorus compounds clearly demonstrates that this is not the case (1)(2)(3)(4)(5). Although microbial metabolism of reduced phosphorus com-pounds has been documented in the literature for several decades, it has remained unexplored in any detail on either the biochemical or genetic levels until recently. This is especially true with respect to the microbial oxidation of the reduced P i compounds, phosphite (P 3ϩ ) and hypophosphite (P ϩ ).
Microbial growth on hypophosphite or phosphite as the sole source of phosphorus has been reported in such microorganism as Escherichia coli, Bacillus spp., Pseudomonas fluorescens, Klebsiella aerogenes, and Erwinia spp. (2, 4, 6 -9); however, little is known about the biochemistry of hypophosphite or phosphite oxidation in these organisms. In P. fluorescens, activity of partially purified phosphorus-oxidizing enzyme was demonstrated to be NAD ϩ -dependent and specific for phosphite; however, a more detailed analysis was not completed (10). Similarly, hypophosphite oxidation was detected in cell extracts of Bacillus caldolyticus, which was demonstrated to grow on hypophosphite as the sole source of phosphorus (11). This enzyme was also partially purified, but nothing is known about the reaction beyond the requirement for NAD ϩ ; neither the responsible enzyme nor the mechanism of hypophosphite oxidation was determined.
Recently we isolated Pseudomonas stutzeri WM88, an organism with the ability to oxidize hypophosphite and phosphite (12). Genetic analysis of hypophosphite and phosphite oxidation in P. stutzeri WM88 led to the identification of two regions of the chromosome involved in utilization of these compounds as the sole phosphorus sources, one region for the oxidation of hypophosphite and the other for phosphite oxidation. Furthermore, these studies showed that the genes involved in phosphite oxidation were also required for growth on hypophosphite, suggesting that hypophosphite oxidation to phosphate proceeds via a phosphite intermediate. Sequence analysis of the chromosomal region for hypophosphite oxidation revealed five open reading frames, htxABCDE, which appear to form a transcriptional unit. Analysis of the predicted amino acid sequence of the htxA gene product indicates that it is most similar to members of the 2-oxoglutarate-dependent dioxygenase family, having 26% identity to proline 4-hydroxylase from Dactylosporangium (13). The putative htxBCDE products likely comprise a binding protein-dependent hypophosphite transporter. Of the five open reading frames identified, only the htxA gene was required for hypophosphite oxidation in P. stutzeri WM88. This, in addition to the amino acid sequence similarities of hypophosphite:2-oxoglutarate dioxygenase (HtxA) 1 to members of the 2-oxoglutarate-dependent dioxygenase family of enzymes, strongly suggests that htxA encodes a novel enzyme responsible for hypophosphite oxidation, HtxA. Sequence analysis of the phosphite oxidation region suggested the presence a binding protein-dependent phosphite transporter, encoded by ptxABC, and an NAD:phosphite oxidoreductase (PtxD), encoded by ptxD (12).
Based on these data, a biochemical pathway for the oxidation of hypophosphite to phosphate was proposed ( Fig. 1) (12). In this pathway hypophosphite is first oxidized to phosphite by the HtxA protein. The phosphite produced in this reaction is subsequently oxidized to phosphate by the PtxD protein.
Strong biochemical evidence has been provided for the second step in this putative pathway (14,15). The responsible enzyme, PtxD, has been characterized in pure form. This enzyme is highly specific for its substrates and stoichiometrically produces phosphate and NADH from phosphite and NAD ϩ . However, biochemical support for the initial step, oxidation of hypophosphite to phosphite, has yet to be provided. In this paper we address this deficiency by reporting the purification and initial biochemical characterization of HtxA, thus completing the first biochemical characterization of a pathway for the oxidation of the reduced inorganic phosphorus compounds.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Culture Conditions-E. coli DH5␣ (16) was used as a host for cloning experiments and as a host for vector pMAL-c2X (New England Biolabs Inc., Beverly, MA) and its derivatives. E. coli BL21(DE3) (17) was used as a host for expression vector pETlla (Novagen, Madison, WI) and its derivatives. E. coli strains were grown in Luria-Burtani medium with either 100 g/ml ampicillin or 50 g/ml carbenicillin when appropriate. P. stutzeri strain WM536 is a spontaneous smooth colony mutant of the original phosphite and hypophosphite-oxidizing isolate, WM88, and P. stutzeri WM567 is a spontaneous streptomycin-resistant derivative of WM536 (12). P. stutzeri strains were grown in either Luria-Bertani medium or MOPS minimal medium with 0.4% glucose and the appropriate phosphorus source (18). Phosphorus solutions were made fresh and sterilized by filtration immediately before use. For large scale expression of native or recombinant proteins, cultures were grown in a 30-liter stainless steel bioreactor (model P30A, B. Bruan Biotech, Allentown, PA) at 30°C. To remove residual phosphate remaining after the cleaning processes, all glassware for growth and media preparation was soaked in ultrapure deionized water with several changes. The bioreactor vessel was washed additionally with 0.1 M nitric acid and rinsed with ultrapure deionized water. For the preparation of phosphate-free solid glucose MOPS minimal medium, the agar was rinsed before use with several changes of ultrapure deionized water.
Construction of HtxA Expression Vector pAG4 -Standard methods for DNA manipulation and cloning were used throughout (19). Plasmid pAG4, which carries HtxA under control of the T7 promoter, was con-structed by PCR amplification of the htxA gene from P. stutzeri WM88 chromosomal DNA using Taq polymerase (Invitrogen) and the following primers: 5Ј-GCGCGCGCCATATGTTTGCAGAGCAGCAACGC-3Ј, which introduces an NdeI site at the translational start codon, and 5Ј-GGATCCCCTCAGTGAGTTAAAGAC-3Ј, which introduces a BamHI site just past the translational stop codon of htxA (restriction sites are underlined). The resulting PCR product was digested with NdeI and BamHI and ligated into the same sites of vector pET11a (Novagen). The sequence of the cloned htxA gene was determined using standard T7 promoter and terminator primers and was identical to the sequence determined previously.
Construction of Maltose-binding Protein (MBP)-PtxD Fusion Expression Vector-The ptxD gene was amplified by PCR using Pfu Turbo polymerase (Statagene Cloning Systems, La Jolla, CA) and pWM237 (12) as the template, with the following primers: 5Ј-ATGCTGC-CGAAACTCGTTATAACTCACC-3Ј and 5Ј-GGATCCAAGCTTTCAA-CATGCGGCAGGCTCGG-3Ј. The reverse primer introduces a HindIII restriction site (underlined) immediately after the ptxD translational stop codon. The PCR fragment was digested with HindIII and cloned into the XmnI-HindIII site of the MBP fusion vector pMAL-c2X (New England Biolabs), creating pAW32. Creation of the correct fusion was verified by sequencing using the malE and M13/pUC sequencing primers (New England Biolabs).
Construction of htx Promoter-lacZ Fusion Strain, WM2940 -The 1.0-kbp region of DNA directly upstream from the htxA ribosomal binding site was amplified by PCR using Taq polymerase (Invitrogen) and the following primers: 5Ј-GGCGGCACTAGTGGATCCCCGATTCG-TACCGGGTGGC-3Ј, which introduces an SpeI site, and 5Ј-GGATC-CGCGGCCGCAAGGTCTTCCAACGAATAATC-3Ј, which introduces an NotI restriction site for facilitation of cloning (restriction sites underlined). The resulting PCR product was digested with the appropriate enzymes and ligated into the same sites of the broad host range cloning vector pWM263 (12) to create pAW35. The lacZ gene was amplified from E. coli genomic DNA using Taq polymerase and the following primers: 5Ј-GGCGGCGCGGCCGCAGGAAACAGCTATGACCATG-3Ј and 5Ј-GGCGGCGCGGCCGCTTATTTTTGACACCAGACCA-3Ј, which insert NotI sites at the 5Ј-and 3Ј-ends of the PCR product (sites underlined). The resulting product was digested with NotI and inserted into the same site of pAW35 to create pAW36. Construction of the correct transcriptional htx promoter-lacZ fusion was verified by DNA sequencing. Plasmid pAW36 was transformed into the transfer-competent E. coli strain BW20767 (20) with selection for ampicillin resistance, followed by mating of the transformants with P. stutzeri strain WM567 with selection on glucose MOPS minimal medium containing ampicillin as described previously (21). For expression analysis of the htx promoter-lacZ fusion, a P. stutzeri exconjugate (WM2940) harboring pAW36 was grown on glucose MOPS minimal medium containing carbenicillin and either 2 mM P i and 0.15% glucose (excess phosphorus) or 0.1 mM phosphite, phosphate, or hypophosphite and 1% glucose (limiting phosphorus). Cells were harvested at stationary phase (A 600 about 1.0) by centrifugation, and extracts were made as described above, using ␤-galactosidase buffer (50 mM Tris-Cl, pH 8.0, 10 mM KCl, 1 mM MgSO 4 , 50 mM ␤-mercaptoethanol) to resuspend the cells. Continuous ␤-galactosidase assays were performed in a 1-ml volume of ␤-galactosidase buffer with the addition of 2.7 mM -nitrophenyl-␤-D-galactoside and 0.05 ml of extract containing about 0.1 mg of protein. Activity was monitored as an increase in absorption at 420 nm.
Expression and Purification of MBP-PtxD-WM2021 (E. coli DH5␣ transformant harboring pAW32) was grown in 30 liters of Luria-Burtani medium, 0.2% glucose medium with carbenicillin in a 30-liter stainless steel bioreactor at 25°C. The culture was induced for expression at an A 600 of about 0.4 by the addition of isopropyl-1-thio-␤-Dgalactopyranoside (IPTG) to a final concentration of 0.3 mM and was incubated additionally at 25°C for 12 h, then harvested by centrifugation. Preparation of cell extracts and all purification steps were performed at 4°C. Approximately 26 g (wet weight) of cell paste was resuspended in 40 ml of column buffer (20 mM MOPS, 10% glycerol, 100 mM NaCl, 1 mM dithiothreitol, pH 7.4) with the addition of about 15 mg of DNase I, and the cells were broken by passage through a French pressure cell twice at 12,000 p.s.i. The lysate was centrifuged at 20,000 ϫ g for 30 min, and the supernatant was centrifuged further at 270,000 ϫ g for 45 min to separate the soluble portion of the extract from the membrane components. The supernatant was recovered to obtain the high speed extract that was used in subsequent purification steps.
High speed extract containing about 300 mg of protein at 3 mg/ml was loaded at 0.75 ml/min onto a 2.5-cm (inner diameter) ϫ 20-cm amylose/agarose column (New England Biolabs) preequilibrated in col-FIG. 1. Proposed biochemical pathway for the oxidation of hypophosphite to phosphate in P. stutzeri WM88. HtxA is a novel member of the 2-oxoglutarate-dependent dioxygenase superfamily of enzymes and catalyzes the oxidation of hypophosphite to phosphite requiring 2-oxoglutarate and oxygen as cosubstrates in addition to ferrous ions for activity. The reaction proceeds by the incorporation of one of the atoms of dioxygen into hypophosphite to form phosphite. The other atom of dioxygen reacts with 2-oxoglutarate, resulting in oxidative decarboxylation giving rise to succinate and CO 2 . PtxD is a member of the D-isomer-specific 2-hydroxy acid NAD-dependent dehydrogenase protein family (15). This enzyme completes the oxidation of hypophosphite by oxidizing phosphite to phosphate using NAD ϩ as a cofactor. umn buffer. Unbound protein was removed with about 10 column volumes of column buffer at 1.5 ml/min until protein was no longer detected in the eluent by UV absorption at 280 nm. MBP-PtxD was eluted with column buffer containing 20 mM maltose at a flow rate of 1.5 ml/min. The purest fractions determined by SDS-PAGE analysis and specific activity were pooled, desalted, and concentrated in an ultrafiltration cell (Amicon Inc., Beverly, MA) equipped with an Amicon Dia-Flow membrane with molecular exclusion size of 10,000 Da. Purified MBP-PtxD was stored at Ϫ70°C in 20 mM MOPS, 10% glycerol, 1 mM dithiothreitol, pH 7.25.
Expression and Purification of Recombinant HtxA-WM804 (E. coli BL21(DE3) harboring pAG4) was grown at 37°C in a 30-liter stainless steel bioreactor in Luria-Burtani medium containing carbenicillin. Protein expression was induced when the culture reached mid-log phase (A 600 about 0.6) by the addition of IPTG to a final concentration of 1 mM. The cells were incubated for an additional 1.5 h and harvested by centrifugation.
Crude extracts were prepared by resuspending about 8 g of cells (wet weight) in 20 ml of buffer A (20 mM MOPS, 10% glycerol, 75 mM NaCl, pH 8.0). Approximately 10 mg of DNase I was added, and the cells were broken by passage through a French pressure cell twice at 12,000 p.s.i. High speed extracts were obtained as described above.
Chromatography for the purification of both recombinant and native HtxA was performed using AKTA fast protein liquid chromatography (Amersham Biosciences) at 4°C. Recombinant HtxA was purified on a 10-mm (inner diameter) ϫ 100-mm POROS HQ anion exchange column (PerSeptive Biosystems, Inc. Framingham, MA), preequilibrated in buffer A, by applying high speed extract containing about 200 mg of protein at 24 mg/ml at 0.5 ml/min. Unbound protein was removed with 40 column volumes of buffer A at 3.0 ml/min. Recombinant HtxA was eluted with a 30-column volume linear gradient of 75-300 mM NaCl in 20 mM MOPS, 10% glycerol, pH 8.0, collecting 2.0-ml fractions. The purest fractions determined by HtxA specific activity and SDS-PAGE analysis were pooled, desalted, and concentrated using the ultrafiltration cell and membrane described above. The purest pools from five such purifications were combined and stored at Ϫ70°C in 20 mM MOPS, 15% glycerol, pH 7.25, for use in the studies described below.
Expression and Purification of Native HtxA-For the expression of HtxA in its native host, 30 liters of P. stutzeri WM567 was grown in a 30-liter bioreactor at 30°C in MOPS minimal medium containing 0.4% glucose, 2 mM hypophosphite, and Antifoam 289 (Sigma). Cells were harvested at stationary phase (A 600 about 1.4) by centrifugation. Approximately 25 g of cells was resuspended in 60 ml of buffer B (20 mM MOPS, 10% glycerol, pH 8.0). Crude and high speed extract were prepared as described above.
Purification of native HtxA from P. stutzeri WM567 required a threestep purification procedure. High speed extract containing about 300 mg of protein was loaded at a flow rate of 1.5 ml/min onto a Hi Prep 16/10 DEAE-Sepharose column (Amersham Biosciences) preequilibrated in buffer B. Unbound sample was removed with 5 column volumes of buffer B. HtxA was eluted from the column with a linear 0 -0.4 M NaCl gradient in buffer B over 40 column volumes. The purest fractions based on SDS-PAGE analysis and specific activity of HtxA were pooled, desalted into buffer B, and concentrated using a Centriprep 30 concentrator (Amicon). The concentrated pools of three such purifications were then loaded onto a 4.6-mm (inner diameter) ϫ 250-mm POROS HQ anion exchange column at a flow rate of 2.0 ml/min. Unbound protein was removed with 5 column volumes buffer B, and HtxA was eluted with a linear 0 -0.4 M NaCl gradient in buffer B, over 30 column volumes. The purest fractions from two such purifications were pooled, concentrated, and desalted into buffer B and loaded onto a 4.6-mm (inner diameter) ϫ 250-mm POROS NH-2-oxoglutarate affinity column (see below) at a flow rate of 2.0 ml/min. Unbound protein was removed with 5 column volumes buffer B, and HtxA was eluted with a linear 0 -100 mM 2-oxoglutarate gradient over 40 column volumes. The purest fractions were pooled, concentrated, and desalted into 20 mM MOPS, 10% glycerol, pH 7.2, using a Centriprep 30 concentrator, and stored at Ϫ70°C for future use.
Coupling 2-Oxoglutarate to POROS NH Affinity Resin-To make a 2-oxoglutarate cosubstrate affinity column for the purification of native HtxA, 2-oxoglutarate was randomly coupled to POROS NH activated affinity resin as follows. POROS NH powder (2.7 g) was resuspended in 13.5 ml of 0.1 M 2-oxoglutarate in deionized water at pH 5.0. With continuous stirring, 1.5 ml of 1.0 M EDAC was slowly added to the slurry to a final concentration of 0.1 M. The pH was periodically adjusted back to 5.0 with 2.0 M NaOH during the 1st h of the reaction until stabilized. The mixture was gently mixed for 24 h at room temperature. After the 24-h period, the coupled resin was washed with 750 ml of 0.1 M Tris, pH 8.3, containing 0.5 M NaCl to remove unreacted substrates. Tris was replaced with deionized water, and the coupled resin was packed into the column.
Amino Terminus Sequencing and Mass Spectrometry-Partially purified recombinant and native HtxA were separated from contaminating proteins on a 12% Tris-HCl precast polyacrylamide gel (Bio-Rad) under denaturing conditions. The protein was transferred to a Sequi-Blot polyvinylidene difluoride membrane (Bio-Rad) using a Mini Trans-Blot Electrophoretic Transfer Cell (Bio-Rad) in 10 mM CAPS, 10% methanol buffer, pH 11.0. Protein bands were visualized with Coomassie Blue, and the membranes were submitted to the University of Illinois Protein Sciences Facility for amino terminus sequencing by Edman degradation. Matrix-assisted laser desorption ionization mass spectrometry was performed at the University of Illinois Mass Spectrometry facility using a Voyager-DE STR mass spectrometer (PerSeptive Biosystems).
Enzyme Activity Assays-For purification and characterization of HtxA, a continuous coupled spectrophotometric assay using MBP-PtxD as a reporter enzyme was used. Activity was measured by NADH production monitored as an increase in absorbance at 340 nm. The extinction coefficient of 6,220 M Ϫ1 cm Ϫ1 was used to calculate NADH production, and enzyme activities are given in standard enzyme units (mol of NADH produced/min). The standard assay mixture contained 20 mM MOPS, pH 7.0, 1 mM 2-oxoglutarate, 2 mM hypophosphite, 1 mM NAD ϩ , 20 M FeCl 2 , 0.2-1 mg of purified PtxD, and 5-10 g of purified HtxA in a 1.0-ml volume. All activity assays were done at room temperature unless otherwise stated. Because HtxA activity rapidly diminishes in assay buffer (significant loss is observed within first 2 min at low protein concentrations) spectrophotometric data were collected once/s for 1-2 min. Only the linear portion of the curve, which was typically about 20 s, was used in activity calculations. Anaerobic activity assays were done using substrate and enzyme solutions, which were made anaerobic with multiple cycles of alternating vacuum and nitrogen gas exchange; the assays were carried out in gas-tight cuvettes. For the determination of pH optimum, 100 mM Tris, 50 mM glacial acetic acid, 50 mM MES buffer was used, and the pH was adjusted appropriately with HCl or NaOH. The ionic strength of this buffer remained constant over the pH range tested (22). Stoichiometry studies and analysis of the use of alternative substrates were done by measuring succinate production in an end point assay using a commercially available kit (Roche Molecular Biochemicals). 31 P NMR Assays-31 P NMR spectra were acquired with a Varian Unity 500 equipped with a 5 mm Nalorac QUAD probe at the Varian Oxford Instruments Center for Excellence in NMR Laboratory at the University of Illinois. The HtxA assay solution contained 20 mM MOPS, pH 7.0, 5 mM hypophosphite, 3 mM 2-oxoglutarate, 100 M FeCl 2 , and 0.5 mg/ml HtxA in a 3-ml volume. A similar assay without HtxA was used as the enzyme-free control. Both mixtures were incubated with constant mixing at room temperature for about 2 h, after which samples were removed, D 2 O was added to 10% final concentration, and the samples were analyzed with 31 P NMR. The phosphite only standard contained 5 mM phosphite in 20 mM MOPS buffer. Proton-coupled 31 P spectra were acquired with an acquisition time of 0.655 s with 2,000 transients and using a pulse width of 5.2 s. An external reference of 85% phosphoric acid set at 0 ppm was used.
Other Methods-Protein concentration was determined using Coomassie Plus reagent (Pierce Chemical Co.) with bovine serum albumin as the standard and following manufacturer's instructions. SDS-PAGE was done in a Mini-PROTEAN II Cell (Bio-Rad) using 12% polyacrylamide slab gels and the Laemmli buffer system (23). DNA sequencing was performed at the W. M. Keck Center for Comparative and Functional Genomics, University of Illinois.

Hypophosphite Oxidation Is Catalyzed by HtxA Protein-To
demonstrate that htxA encodes a protein that catalyzes the oxidation of hypophosphite to phosphite in a 2-oxoglutaratedependent manner, a coupled continuous HtxA activity assay Hypophosphite/2-Oxoglutarate Dioxygenase was developed using MBP-PtxD as a reporter enzyme. Accordingly, enzymatic oxidation of hypophosphite to phosphite is coupled to phosphite-dependent NAD ϩ reduction, which is monitored at 340 nm (see Fig. 1). However, using this assay, hypophosphite oxidation could not consistently be detected in crude extract of P. stutzeri WM88 grown on hypophosphite as the sole source of phosphorus; therefore, HtxA was overproduced in E. coli. Cell extract from E. coli WM804, in which HtxA is overexpressed, catalyzes hypophosphite oxidation with a specific activity of 2.40 units/mg. This activity is strictly dependent on the addition of 2-oxoglutarate and ferrous ions and was not detected in extract from E. coli BL21(DE3) harboring the pET11a expression vector only. These data strongly support the hypothesis that HtxA is a 2-oxoglutarate-dependent hypophosphite dioxygenase.
Purification of Recombinant and Native HtxA-Overexpression in E. coli BL21(DE3) produces very high levels of soluble HtxA in the IPTG-induced extracts allowing about 95% homogeneous protein to be obtained after a single anion exchange chromatography step (Table I and Fig. 2). Further purification using 2-oxoglutarate cosubstrate affinity chromatography was not helpful and resulted in significant loss in activity accompanied by an insignificant increase in purity. Additional purification attempts, including anaerobic purification, also resulted in a significant and irreversible loss of activity.
The nucleotide sequence of htxA revealed the presence of two putative translation start sites for HtxA which would change the size of the protein by 14 amino acids. To ensure that the recombinant form of HtxA was expressed from the correct translational start site and that it is identical to the native enzyme isolated from P. stutzeri WM536, attempts to purify native HtxA were made. Extracts of P. stutzeri WM536 grown with 2 mM hypophosphite as the sole source of phosphorus were fractionated with tandem anion exchange chromatography followed by 2-oxoglutarate affinity chromatography. Partially purified HtxA was obtained (about 30% homogeneity) and subjected to amino terminus sequencing and mass spectral analysis. The amino-terminal sequence of both the native and recombinant forms of the enzyme was determined to be MFAEQQREYLDKGYT, which is in complete agreement with the predicted amino acid sequence of htxA (as annotated in GenBank, accession no. AFO61267). Mass spectral analysis yielded peaks at 32,485 Ϯ 40 and 32,475 Ϯ 40 Da for the native and recombinant forms of HtxA, respectively, which are consistent with the predicted molecular mass of the monomer of 32,503 Da. Based on these analyses, the native and recombinant forms of the enzyme are indistinguishable, and recombinant HtxA was used in all further studies.
HtxA Is a 2-Oxoglutarate-dependent Hypophosphite Dioxygenase-Purified HtxA demonstrates a strict requirement for ferrous ions, oxygen, 2-oxoglutarate, and hypophosphite for activity (Fig. 3). Only in the presence of all of the components required for the continuous coupled assay was an increase in absorbance at 340 nm observed. The end products succinate and phosphite were produced in equimolar quantities, demonstrating a 1:1 stoichiometry for the reaction, with 0.524 Ϯ 0.032 mol of succinate/0.510 Ϯ 0.010 mol of phosphite (as measured by PtxD-catalyzed NADH production) produced after a 20-min incubation time. Given the strict substrate specificity of PtxD determined in a previous study (15), it seems evident that phosphite is the phosphorus product of the HtxA reaction. However, to identify the product of the HtxA reaction unequivocally, 31 P NMR was used to analyze the reaction products (Fig.  4). With the addition of 3 mM 2-oxoglutarate in the assay, 2.85 mM phosphite was produced in the complete assay mixture (Fig. 4C). These data show that phosphite is the only phosphorus product made upon incubation of hypophosphite with HtxA and that phosphite production is stoichiometric with 2-oxoglutarate consumption.
Biochemical Characterization of Purified HtxA-HtxA was rapidly inactivated in assay buffer (see below); however, purified HtxA was stable upon storage in 20 mM MOPS, 15% glycerol, pH 7.25 at Ϫ70°C indefinitely and at 0°C for several days. Size exclusion analysis of the purified enzyme suggests a native molecular mass of 68,794 Da, consistent with HtxA being active as a homodimer (the amino acid sequence predicts a dimer molecular mass of 65,006 Da).
The optimal pH for HtxA was established using both a Universal buffer system (100 mM Tris-Cl, 50 mM acetic acid, 50 mM MES), in which the ionic strength remains constant over the  pH range tested, and in 20 mM MOPS buffer. In both buffers, the pH optimum was found to be 7.0; however, activity in MOPS buffer was slightly higher and was sustained over a broader range of pH, from 6.25 to 7.5 (Fig. 5A). Activity in BisTris and phosphate buffers was much lower than what was observed in MOPS buffer (data not shown), therefore, MOPS buffer was used in all characterization studies. The temperature optimum was determined to be between 25 and 30°C with a broad range of activity from 15 to 35°C (Fig. 5B). HtxA showed a marked decrease in activity upon addition of NaCl at concentrations ranging from 25 to 300 mM (Fig. 5C).
As described above, maximal HtxA activity was shown to be dependent on exogenous ferrous ions, although about 2% residual activity was observed in the absence of exogenous Fe(II) ( Table II). This is likely caused by small amounts of remaining enzyme bound ferrous ions that were retained during enzyme purification. FeCl 2 concentrations ranging from 20 to 120 M supported full HtxA activity; however, addition of EDTA to the reaction mixture completely abolished the activity of HtxA. A variety of ferrous salts supported equal levels of HtxA activity when provided at 100 M (Table II). The alternative divalent cations Ni(II), Co(II), Mn(II), Ca(II), and Mg(II) could not substitute for ferrous ions when provided at 100 M as chloride salts, and each inhibited HtxA activity by about 10 -30% when provided in addition to 100 M FeCl 2 (Table II).
HtxA activity diminishes rapidly in assay buffer. Inactivation of other 2-oxoglutarate-dependent dioxygenases has been prevented by the addition of ascorbate to the assay mixture, which is thought to prevent oxidation of enzyme-bound ferrous ions, resulting in inactivation of the enzyme (24,25). However, addition of ascorbate at 100 M to the HtxA assay results in a 20% decrease in specific activity. Furthermore, preincubation of HtxA in the reaction mixture in the presence and absence of ferrous ions before assaying enzyme activity resulted in an equally rapid decrease in HtxA activity. Thus, HtxA inactivation in the reaction mixture is independent of the presence of ferrous ions.
Kinetic Analysis of Hypophosphite Oxidation Catalyzed by HtxA-The oxidation of hypophosphite to phosphite by HtxA follows Henri-Michaelis-Menten kinetics. The kinetic constants were determined using the continuous coupled assay with MBP-PtxD as the reporter enzyme. V max and k cat were determined to be 10.9 Ϯ 0.30 mol min Ϫ1 mg Ϫ1 and 355 min Ϫ1 , respectively (Fig. 6). The apparent K m values were determined to be 0.58 Ϯ 0.04 mM for hypophosphite (Fig. 6A) and 10.6 Ϯ 1.4 M for 2-oxoglutarate (Fig. 6B). To determine the very low K m of 2-oxoglutarate, a 5-cm path length cuvette was used to increase the sensitivity of the assay.
Substrate and Cosubstrate Specificity of HtxA-The substrate specificity of HtxA was examined using an end point assay in which accumulation of succinate was measured using a highly sensitive enzyme assay for the detection of succinate (Table III). Both inorganic and organic compounds that are structurally analogous or that have similar oxidation states to hypophosphite were examined. Of the alternative substrates tested, sulfite, nitrite, methylphosphonate, methylphosphinate, dimethylphosphinate, taurine, and formaldehyde were not oxidized by HtxA. Formate and arsenite, and to a lesser extent, phosphite, each yielded succinate after a prolonged incubation with 66 g/ml HtxA and 4 mM substrate. Surprisingly, the amount of succinate produced with formate and arsenite as substrates exceeded that produced with hypophosphite as substrate. To address this further, the specific activity of HtxA with each substrate supporting activity was determined under conditions identical to those above with the exception of the enzyme concentration, which was reduced to 19.3 g/ml to slow the reaction enough to make the intermediate measurements possible (Table III). With hypophosphite as the substrate, the specific activity of HtxA was 11.0 units mg Ϫ1 . The specific activities acquired with formate and arsenite were 10.0 and 5.3 units mg -1 , respectively. Activity was not detected with phosphite as the substrate using this assay. Thus, HtxA demonstrates relatively relaxed substrate specificity, being able to oxidize several substrate analogs, a common characteristic among the members of the 2-oxoglutarate-dependent dioxygenase enzyme family (26 -28).
The cosubstrate range of HtxA was also examined (Table IV). The continuous coupled assay was used to determine whether other 2-oxoacids could support the oxidation of hypophosphite. Among the alternative cosubstrates tested, oxaloacetate, 2-oxovalerate, 2-oxocaproate, and 2-oxobutyrate could not substitute for 2-oxoglutarate as cosubstrate, even when provided at 5 mM. At this concentration, only 2-oxoadipate, and to a much lesser extent, pyruvate, resulted in the oxidation of hypophosphite. When provided at 0.5 mM, only 2-oxoadipate supported HtxA activity, resulting in about 37% of the activity observed with 2-oxoglutarate.
Effects of Substrate Analogs and Reaction Products on the Activity of HtxA-Inhibition by substrate analogs was examined using the continuous coupled HtxA activity assay ( Table  V). The resulting specific activities were compared with the activity observed with no inhibitor present. Of the substrate analogs tested, nitrate and formate severely inhibited HtxA activity, resulting in only 28.5 and 9.7% activity, respectively. HtxA activity was only mildly inhibited by the presence of nitrite, arsenite, sulfate, methylphosphonate, and aminoethylphosphonate. Product inhibition was examined by adding succinate or phosphate to the assay. The presence of phosphate slightly enhanced activity, whereas succinate resulted in about 50% inhibition of HtxA activity.
The Expression of HtxA Is Phosphate Starvation-inducible-Because of the relaxed substrate specificity of HtxA shown above, a genetic approach was taken to elucidate further the in vivo role of this enzyme with regard to its substrate. If the in vivo substrate of HtxA is an alternative phosphorus source such as hypophosphite or phosphite, then expression of HtxA might be induced under conditions of phosphate starvation. To examine the regulation of expression of the htx locus with Hypophosphite/2-Oxoglutarate Dioxygenase growth on various phosphorus sources, expression analysis from the htx promoter was performed in P. stutzeri WM2940, which carries a plasmid-borne htx promoter-lacZ transcriptional fusion. With growth on limiting phosphate, or on hypophosphite or phosphite as the sole source of phosphorus with excess carbon, the expression of ␤-galactosidase from this fusion is induced 10 -15-fold, relative to expression with growth on excess phosphate. These data strongly suggest a role for HtxA in acquiring an alternative source of phosphorus from hypophosphite and phosphite and further support that hypophosphite is the in vivo substrate of this enzyme. DISCUSSION Although there have been numerous accounts of microorganisms that can grow on hypophosphite as the sole source of phosphorus, the biochemical process by which this occurs had not been examined in detail, leaving a novel and significant area of phosphorus metabolism largely unexplored. The isolation and characterization of HtxA presented in this paper represent the first detailed biochemical analysis of an enzyme in pure form devoted to the oxidation of the reduced inorganic phosphorus compound, hypophosphite.
HtxA catalyzes the first reaction in a pathway required for the acquisition of phosphate from the inorganic reduced phosphorus compound, hypophosphite. The biochemical characterization of HtxA presented here classifies this enzyme as a novel 2-oxoglutarate-dependent dioxygenase. This family of enzymes is remarkably diverse in the reactions its members catalyze, which include amino acid hydroxylations, secondary metabolite biosynthesis, and degradation of alternative carbon and sulfur sources (for review, see Refs. 29 and 30). The commonality among all of the members of this family is that they all require activation of a molecule of dioxygen by enzyme-bound ferrous ions to generate a highly reactive ferryl oxidant. The formation of the ferryl species is linked to the oxidative decarboxylation of 2-oxoglutarate, giving rise to succinate and CO 2 , and mediates hydroxylation of the substrate (31,32). Thus, all members of this family are dependent on ferrous ions, oxygen, and 2-oxoglutarate (or a similar 2-oxoacid) for activity. Although the substrates acted upon by the members of this family are diverse, all that have been characterized, to our knowledge, act on organic substrates (29,30). In contrast, of the substrates tested in this study, HtxA showed the highest activity with the inorganic substrate hypophosphite, making it the first enzyme in this family to have an inorganic substrate. Several organic phosphorus compounds thought to be possible alternative substrates for HtxA (methylphosphinate, dimethylphosphinate, and methylphosphonate) were not oxidized by this enzyme.
The low degree of amino acid sequence identity among 2-oxoglutarate-dependent dioxygenases mirrors their catalytic diversity. Among the few highly conserved residues found in many members of this family are those that have been identified to be involved in binding of the ferrous ion. These residues form a conserved motif, designated the 2-His-1-carboxylate facial triad, typical of non-heme Fe(II) enzymes (33). Advanced  spectroscopic techniques (34), site-directed mutagenesis studies (35)(36)(37)(38)(39)(40), and examination of the crystal structures of cephalosporin synthase and the mechanistically related, isopenicillin N synthase (41,42), have further elucidated the role of these residues in binding Fe(II), oxygen, and 2-oxoglutarate. The 2-His-1-carboxylate motif is also present in HtxA and its closest sequence homologs (Fig. 7). Prolyl 4-hydroxylase shows the highest degree of sequence similarity to HtxA, sharing 26% amino acid sequence identity. Phytanoyl-CoA hydroxylase and an enzyme involved in mitomycin C biosynthesis, MmcH, share weaker homology to HtxA, with 23 and 22% identity, respectively. However, all share a variation to the common His-X-Asp-X 53-57 -His motif that is characteristic to members of the 2-oxoglutarate-dependent dioxygenases superfamily and is thought to form the 2-His-1-carboxylate facial triad ( Fig. 7 and Ref. 38). Within the group of HtxA homologs, the histidine residues believed to be involved in Fe(II) binding are spaced further apart in the amino acid sequence, giving a His-X-Asp-X 72-101 -His-X 10 (Arg/Lys)X-Ser motif that was found to be a distinct characteristic of this subgroup of enzymes, which do not share significant homology to other members of the 2-oxoglutarate-dependent dioxygenases family in the data base (40). Other conserved residues are also evident within the sequence of these proteins, but a function for these has yet to be ascribed.

Hypophosphite/2-Oxoglutarate Dioxygenase
HtxA catalyzes a strictly coupled oxidation of hypophosphite, producing equimolar amounts of succinate and phosphite, demonstrating the expected stoichiometry of the 2-oxoglutaratedependent dioxygenase reaction. Interestingly, an uncoupled reaction of proline 4-hydroxylase has been well characterized, in which 2-oxoglutarate is oxidatively decarboxylated to succinate and CO 2 , without the concomitant hydroxylation of the substrate (24,25,43). However, HtxA showed no such uncoupled reaction because succinate was not produced upon incubation of the enzyme with 2-oxoglutarate in the absence of hypophosphite.
The effect of ascorbate on HtxA activity was explored because it has been reported to stabilize the activity of numerous members of this family of enzymes during the reaction. This effect is attributed to the role of ascorbate as a reducing agent that has been shown to counteract Fe(II) oxidation to inactive Fe(III), believed to occur as a side reaction during the hydroxylation of the substrate (24,25,44). Stabilization of activity in the presence of ascorbate has been observed in avian proline 4-hydroxylase (24,25,44) as well as in other 2-oxoglutaratedependent dioxygenases such as desacetoxyvindoline 4-hydroxylase, 2,4-dichlorophenoxyacetate/2-oxoglutarate dioxygenase, oxygenative alkylsulfatase, and 2-oxoglutarate-dependent tau-rine dioxygenase (27,30,45,46). In contrast, HtxA showed a marked decrease in activity in the presence of ascorbate, further demonstrating that the instability of the enzyme during the reaction is independent of ferrous ion oxidation. The nature of the instability appears to be simply the result of dilution of the enzyme, as has also been observed in 2-oxoglutarte-dependent taurine dioxygenase (27) and has not been explored further.
HtxA was able to catalyze the oxidation of both arsenite and formate, although when comparing specific activities, hypophosphite was a slightly better substrate. Arsenite and formate were among those chosen as alternative substrates to examine based on their chemical and/or similarity to hypophosphite. Given these similarities and the relaxed substrate specificity observed in many of the members of this enzyme family (27,45,46), it is not surprising that these compounds could be oxidized. This is especially true in light of the highly reactive dioxygen-derived ferryl species generated upon binding of 2-oxoglutarate and oxygen at the ferrous ion binding site thought to be involved in the hydroxylation of the substrate (for review, see Refs. 29 and 47). To be oxidized via this reaction, the only requirement for the compound would be that it fit into the active site in juxtaposition with this highly reactive ferryl species. Given the similarities of arsenite and formate to hy-FIG. 7. Amino acid sequence alignment of HtxA with members of the 2-oxoglutarate-dependent dioxygenase superfamily. FASTA3 (54) searches using HtxA identified several proteins sharing a significant degree of sequence identity from 22 to 26%. All are characterized or putative members of the 2-oxoglutarate-dependent dioxygenase superfamily. The top three proteins were aligned with HtxA using ClustalW (55) and are as follows: HtxA, 2-oxoglutarate-dependent hypophosphite dioxygenase from P. stutzeri WM88; P4H, L-proline 4-hydroxylase from Dactylosporangium spp.; PAHX, phytanoyl-CoA dioxygenase from Homo sapiens; MmcH, enzyme involved in mitomycin C biosynthesis from Streptomyces lavendulae. Swiss Protein accession numbers for the sequences used are O69060, O06499, O14832, and Q9X556, respectively. Of the four enzymes shown, L-proline 4-hydroxylase has been studied the most extensively with regard to residues involved in forming the ferrous ion binding site (35). The residues comprise the 2-His-1-carboxylate facial triad (33) and are conserved among these four proteins. They are indicated above the corresponding residues in the alignment. The binding motif shared by this subfamily of 2-oxoglutarate-dependent dioxygenases is written above the conserved sequences (40). Strictly conserved residues among the four proteins are shaded in dark gray. Similar residues are light gray.
pophosphite, this seems the most likely explanation for the relaxed substrate specificity of HtxA. Because formate can act as a substrate, it is also not surprising that formate significantly inhibits HtxA activity, probably because of competitive binding with hypophosphite at the active site. This further supports the conclusion that formate acts simply as a structural analog to hypophosphite. The structural similarity between these two compounds has been demonstrated previously by the ability of hypophosphite to act as a substrate analog to both pyruvate formate lyase (48) and formate hydrogenlyase and regulatory proteins involved in their expression (49). Although the specific activity of HtxA with arsenite as a substrate is significant, arsenite is a poor inhibitor of HtxA activity, suggesting weak binding of arsenite in the active site (relative to hypophosphite). Thus, arsenite is also probably not the true substrate of HtxA. Considering both the biochemical evidence presented here and the regulation of the htx and ptx loci by phosphate starvation (this report and Ref. 15), hypophosphite is almost certainly the in vivo substrate for HtxA, with the role of providing P. stutzeri WM88 with an alternate phosphorus source via oxidation of inorganic reduced phosphorus compounds.
Compared with other members of the 2-oxoglutaratedependent dioxygenase enzyme family, HtxA shows somewhat strict cosubstrate specificity. Similar to taurine dioxygenase, HtxA was able to use only 2-oxoadipate to a significant degree in the place of 2-oxoglutarate (27). This is in contrast to most members of this family, which are able to use a broad range of 2-oxoacids, with or without a second carboxyl group (45,46). Both the cosubstrate specificity and the apparent K m for 2-oxoglutarate fall within the ranges observed for other members of this family.
Because of the dearth of knowledge regarding biological oxidation of reduced phosphorus compounds and of the presence of these compounds in the soil, examining the substrate specificity of HtxA was essential to understanding the physiological role of this enzyme in phosphorus metabolism. It is clear from the genetic analysis that the htxA gene allows P. stutzeri WM88 to grow on hypophosphite as a sole phosphorus source and at a rate similar to growth on phosphate (12). In addition, examination of the regulation of expression of the ptx and htx operons indicates that both are highly expressed under conditions of phosphate starvation, including growth on both limiting phosphate and on hypophosphite or phosphite as the sole phosphorus sources. Sequence analysis of the region upstream from the htxA translation start site revealed the presence of a Pho box (data not shown), a conserved binding sequence for transcriptional activation via PhoB (50). In numerous microorganisms, PhoB is the response regulator responsible for activating transcription of the Pho regulon, which is comprised of numerous phosphate starvation-inducible loci, all involved in the assimilation of phosphorus in response to phosphate starvation. These genetic data strongly suggest that the in vivo role of these genes is to allow use of an alternative phosphorus source, such as phosphite and hypophosphite.
What is less clear is the extent to which hypophosphite is present in nature. Although it is widely used industrially as a reducing agent, and microbe-mediated reduction of phosphate to phosphite, hypophosphite, and phosphine has been reported (3,51), no direct measurements of inorganic reduced phosphorus compounds in the soil have been documented. The apparent K m of HtxA for hypophosphite of 0.58 mM is quite likely to be sufficiently low to support growth of this organism on even the very low concentrations of hypophosphite one might expect to find in the environment. This is particularly true given that htxA appears to be cotranscribed with a binding protein-dependent transporter that is required, in addition to htxA, for growth on hypophosphite in the heterologous hosts, E. coli and Pseudomonas aeruginosa (12). Such transporters are known to be able to accumulate very high intracellular levels of their substrate, with concentration gradients of up to 100,000-fold (52), which would put intracellular hypophosphite concentrations well within the range allowing HtxA to catalyze in vivo hypophosphite oxidation.
Finally, hypophosphite oxidation has been studied previously in B. caldolyticus, and activity was found to be dependent on NAD ϩ (11), and anaerobic oxidation of hypophosphite has been documented in uncharacterized Bacillus isolate (8). In contrast to these findings, HtxA does not require NAD ϩ for oxidation of hypophosphite, and oxygen is absolutely required for the HtxA reaction. This indicates that HtxA is a very different enzyme from those studied in two Bacillus species, supporting the possible existence of multiple pathways for the oxidation of hypophosphite in diverse microorganisms and the idea that hypophosphite oxidation is an important activity for survival in the environment. Given the large number and diversity of microorganisms reported to oxidize hypophosphite or phosphite, it seems clear that oxidation of reduced phosphorus compounds is not an uncommon activity and should be considered a significant aspect of phosphorus biochemistry. A more profound understanding of the environmental significance of reduced phosphorus biochemistry and of the reaction mechanisms involved awaits additional detailed analyses of the enzymes catalyzing these reactions and of the genes encoding them.