Identification, Characterization, and Cloning of a Phosphonate Monoester Hydrolase from Burkholderia caryophilli PG2982*

The glyphosate-degrading bacterium, Burkholderia caryophilli PG2982, was observed to utilize glyceryl glyphosate as a sole phosphorus source. The hydrolysis of glyceryl glyphosate to glyphosate by a phosphonate ester hydrolase (PEH) was identified as the first meta- bolic step in the mineralization pathway. This observation provides the first biological role for a phosphonate ester hydrolase activity. Purified PEH enzyme hydrolyzed several phosphonate esters including p -nitrophen- yl phenylphosphonate, (cid:98) -naphthyl phenylphosphonate, and 5-bromo-4-chloro-3-indolyl phenylphosphonate. The purified PEH also hydrolyzed some phosphodiesters including p -nitrophenyl 5 (cid:42) -thymidine monophosphate and p -nitrophenyl phosphorylcholine. The most catalyti- cally efficient substrate identified was bis-( p -nitrophen-yl) phosphate with a K m of 0.9 m M and a k cat of 6.2 (cid:51) 10 2 min (cid:50) 1 , suggesting that the enzyme may also function

Radioisotope experiments confirmed that glyceryl glyphosate was adsorbed, translocated and relatively stable in plants. 1 Similarly, Escherichia coli cells, which were growthinhibited by 0.5 mM glyphosate, did not display any growth inhibition in the presence of 5 mM glyceryl glyphosate. These preliminary results suggested that glyceryl glyphosate is not a substrate for the previously described, ubiquitous phosphonate monoester hydrolases (1). An enzyme capable of hydrolyzing the phosphonate ester bond of glyceryl glyphosate would likely be unique among currently described enzymes in this class. As described herein, the glyphosate-degrading bacteria Burkholderia caryophilli PG2982 (8) was observed to utilize glyceryl glyphosate as a sole phosphorus source. The hydrolysis of glyceryl glyphosate to glyphosate by a phosphonate ester hydrolase was identified as the first metabolic step in the pathway. The PEH 2 enzyme from PG2982 was more fully characterized, and the gene has been cloned. The purified enzyme exhibited a broad substrate specificity for phosphonate monoesters and phosphodiesters.

MATERIALS AND METHODS
Reagents-All buffer components and 5-bromo-4-chloro-3-indolyl phenylphosphonate were from Research Organics. Enzymes for coupled assays and DNA modifying enzymes were from Boehringer Mannheim. All bacteriological media components were obtained from Difco. Oligonucleotides were obtained by custom synthesis from Midland Scientific. Unless otherwise indicated, all other reagents were the highest quality available from Sigma. The p-nitrophenyl phenylphosphonate was purified before use by extraction of contaminating p-nitrophenol into hexane at pH 5.0 (5) and then a stock solution standardized using an extinction coefficient for p-nitrophenol of 18,320 cm Ϫ1 M Ϫ1 . Unless otherwise specified, all cationic and anionic buffers were prepared as the Cl Ϫ and Na ϩ salts, respectively. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  Synthesis of Glyceryl Glyphosate-Glyceryl glyphosate prepared as described (7) was kindly provided by Dr. Om P. Dingrha of the Monsanto Agricultural Group. The preparation was determined to be Ͼ99% pure by glyphosate analysis and by 31 P NMR. The crystalline compound and 100 mM solutions at neutral pH were stable for greater than 1 year. Radioactive glyceryl glyphosate was obtained from DuPont NEN at a specific activity of 52.3 mCi/mmol. The radioactive compound was unstable over 3-4 months and was routinely purified by anion exchange on a MonoQ HR10/10 (Pharmacia Biotech Inc.) equilibrated in H 2 O and then eluted with a 1600-ml gradient of 0 -500 mM triethylammonium bicarbonate, pH 7.5. Fractions containing glyceryl glyphosate were identified using an analytical HPLC method (described for enzymatic assays below). The pool of glyceryl glyphosate (still contaminated with glyphosate) was fractionated on the same column a second time with an 80-ml gradient of 0 -100 mM triethylamine acetate, pH 5.5. The purified radioactive compound and the stable analog were shown to be identical and Ͼ99% pure by 31 P NMR, by anion exchange chromatography, and by ion-pair chromatography under several conditions. Bacterial Strains and Growth Conditions-Cloning and expression were performed in E. coli MM294 (9) and JM101 (10). DNA template for sequencing and site-directed mutagenesis was prepared from E. coli strain CJ236 (11). Pseudomonas sp. PG2982 has been described (8) and recently renamed as B. caryophilli PG2982. Phenotypic selection with glyceryl glyphosate for the pehA gene in E. coli was performed using minimal medium containing MOPS salts (12) supplemented with 0.05 mM KH 2 PO 4 , 100 g/ml spectinomycin, and 1.5% agarose (Seakem). PG2982 was typically cultured in Dworkin-Foster (DF) minimal salts (13) with the addition of glucose, sodium gluconate, and potassium citrate (each 0.1%) as carbon sources and containing 0.2 mM glyceryl glyphosate or Na 2 HPO 4 as a phosphorus source.
Phosphonate Monoester Hydrolase Assays-A radioactive HPLC assay was routinely used to measure the enzymatic hydrolysis of glyceryl glyphosate. Cells or enzyme were incubated with 100,000 cpm glyceryl [3-14 C]glyphosate at 30°C in a total volume of 100 l. Reactions were quenched by the addition of 100 l of 0.1 M sodium acetate in ethanol, pH 5.5, and precipitates removed by brief centrifugation in a microcentrifuge. Radioactive reaction products were separated by isocratic elution over 15 min at 1 ml/min on a Synchropac AX100 anion exchange HPLC column (Synchrom) equilibrated in 65 mM potassium phosphate buffer, pH 5.5. Peaks were quantitated on-line by a Radiomatic FlowOne radioactive detector (Packard Instruments). Glyceryl glyphosate eluted at 4.9 min and glyphosate at 9.7 min, providing base-line separation.
The release of glycerol from glyceryl glyphosate was measured using a colorimetric assay, which included 0.5 ml of enzyme and 0.5 ml of reaction mix consisting of 1.5 mM ␤-NAD, 0.8 unit of NADH:dye oxireductase, 2 units of glycerol dehydrogenase, 300 mM [NH 4 ] 2 SO 4 , 50 mM Bis-Tris propane, pH 9.0, 0.03% Triton X-100, and 0.08 mM p-iodonitrotetrazolium violet. The reaction proceeded with the formation of a red color, which was measured at 492 nm in a Hewlett-Packard diode array spectrophotometer at fixed time intervals.
The direct hydrolysis of glyceryl glyphosate to glyphosate was further confirmed using 31 P NMR. Purified enzyme was added to 1 ml of 10 mM glyceryl glyphosate in 30% D 2 O, pH 7.0. The phosphonate ester of glyceryl glyphosate was observed to exhibit a major peak at 13.7 ppm and a minor peak at 12.9 ppm, which reflected the equilibrium ratio of 1-to 2-glyceryl ester. A ϩ7.7 ppm signal was observed for an authentic glyphosate standard under the same pH and ionic strength.
Enzyme activity was also followed by monitoring the hydrolysis of p-nitrophenyl phenylphosphonate. Standard assays contained 4 mM p-nitrophenyl phenylphosphonate, 0.5 mM MgCl 2 , and 20 mM diethanolamine, pH 9.0, in a final volume of 1 ml. Enzymatic activity was followed by continuous spectrophotometric detection at 404 nm. Initial rates were determined by linear regression.
Enzyme activity was detected in situ after native polyacrylamide gel electrophoresis (PAGE). Native PAGE gels were prepared as described (14) using the Tris-glycine discontinuous buffer system of Davis (15). The proteins were separated on a 7.5% acrylamide gel at 12 mA overnight, with cooling to 4°C, in a Hoeffer model SE600 vertical slab gel apparatus. Alternatively, proteins were separated on a Phastsystem using 10 -15% gradient native PAGE gels as described (Phastsystem Applications Manual, Pharmacia). After electrophoresis, the gels were stained for activity by pouring a thin layer of 10 mM Bis-Tris propane, pH 9.0, 100 mM KCl, 25 mM ␤-naphthyl phenylphosphonate, and 1% Fast Red.
The in vivo expression of the phosphonate ester hydrolase gene in E. coli was detected using a 5-bromo-4-chloro-3-indolyl phenylphosphonate (XPP) in a histochemical assay. A final concentration of 0.01% XPP from a 10% stock in dimethylformamide was added to the cooled agar medium before pouring the plates. Plates were stored in the dark at 4°C.
Other Enzyme Assays-3Ј,5Ј-Cyclic GMP phosphodiesterase activity was assayed essentially as described (16). DNase activity was determined using the method of Kuntz (17). Assays contained 0.125 mg of DNA, 0.0125 mg of ethidium bromide, and the change in absorbance monitored at 260 nm over 30 min.
Enzyme Purification from PG2982-Five 10-liter fermentations of PG2982 cells were carried out in DF minimal medium to obtain PEH enzyme for purification. Cells were collected by centrifugation and stored at Ϫ20°C until used. All procedures were carried out at 4°C. During the purification, activity assays were based on the release of glycerol from glycerol glyphosate, except during the final purification steps when column fractions were first screened using the pNPP assay. The cells (200 g) were resuspended in 1 liter of 100 mM Tris, pH 8.0, 100 mM KCl, 2 mM dithiothreitol (Buffer I) and lysed by passing through a Manton-Gaulin at 8,000 p.s.i. three times. The cell debris was pelleted by centrifugation at 8,000 ϫ g for 30 min. Protamine sulfate was added to the supernatants to 0.2%, and the precipitated nucleic acids harvested by centrifugation as above. The enzyme activity was found in the pellet and the supernatant. Additional protamine sulfate was added to the supernatant to 0.4% total, and the remaining PEH activity was precipitated. The two protamine sulfate precipitates were combined and resuspended with stirring overnight at 4°C in 10 volumes of 0.5 M KCl in 100 mM Tris, pH 8.0. The suspension was centrifuged at 10,000 ϫ g for 20 min, and the soluble PEH activity recovered in the supernatant. A substantial amount of activity remained in the pellet with the nucleic acids, and therefore the extraction was repeated a second time under identical conditions. The protamine sulfate extract was then brought to 50% [NH 4 ] 2 SO 4 by the addition of a saturated stock solution, and the precipitated protein was collected by centrifugation at 10,000 ϫ g for 20 min. The [NH 4 ] 2 SO 4 pellet was resuspended in 100 ml of Buffer I and dialyzed overnight against two 4-liter changes of 50 mM Tris, pH 8.0, 2 mM dithiothreitol (dialysis buffer). The dialyzed enzyme was loaded on a Q-Sepharose fast flow anion exchange column (4 ϫ 40 cm) equilibrated in dialysis buffer. The PEH protein eluted toward the end of a 2-liter, 0 -500 mM KCl gradient in the same buffer. The fractions containing the highest activity were pooled and PEH precipitated with 50% [NH 4 ] 2 SO 4 as before. The [NH 4 ] 2 SO 4 pellet was then loaded onto a 2.6 ϫ 100-cm Sephacryl S-200 gel filtration column equilibrated in 10 mM Tris⅐Cl, pH 8.0, 50 mM KCl, and 5-ml fractions were collected. The PEH activity eluted just after the void volume, and the fractions containing most of the activity were pooled and loaded onto a FPLC MonoQ anion exchange column. The PEH activity was eluted with a 30-ml 0 -500 mM KCl gradient in 10 mM Tris, pH 8.5. The fractions with the highest activity were pooled and rechromatographed on the MonoQ column using a 0 -500 mM KCl gradient in 60 ml of 10 mM Tris, pH 8.5. Fractions containing the highest activity were again pooled and further purified using native PAGE as described above but with a preparative 4 mm ϫ 8-cm resolving gel. The PEH activity was assayed by overlaying a solution of 4 mM pNPP in 100 mM Tris, pH 8.5, 100 mM KCl. The assay was stopped by rinsing the gel with H 2 O as soon as a yellow band appeared. The band displaying the activity was excised, and the PEH protein was electroeluted using a Bio-Rad Mini Protean II electroeluter in 10 mM Tris, 20 mM glycine, pH 8.5. The PEH protein was then brought to 40% ammonium sulfate using a saturated stock solution and loaded on a FPLC alkyl-Superose HR5/5 column equilibrated in 10 mM TAPS, pH 9.0, 40% [NH 4 ] 2 SO 4 . The PEH activity eluted in the middle of a 25-ml gradient of 40 -0% [NH 4 ] 2 SO 4 .
Tryptic Mapping-Tryptic maps were obtained for both the 66-and 59-kDa polypeptides to compare their similarity and to obtain tryptic fragments for protein sequence analysis. 600 g of purified PEH protein was subjected to full reduction and alkylation with iodoacetic acid (18). The 66-and 59-kDa polypeptides were then separated from each other by electrophoresis on a 3-17% acrylamide gradient SDS-PAGE minigel (Jule, Inc.) run at 30 mA. The two polypeptides were visualized by a brief staining (15 min) in 0.3% Coomassie Blue R-250 in H 2 O, excised, and electroeluted into 25 mM Tris, 192 mM glycine, and 0.1% SDS (Bio-Rad MiniPROTEAN II electroelution chamber). The eluted polypeptides were dialyzed against H 2 O for 4 h, precipitated with five volumes of ice cold acetone, resuspended in 0.1 M NaOH at 40°C, and then desalted into 0.1 M ammonium bicarbonate, pH 8.1 using a Sephadex G-25 column. The polypeptides were digested with trypsin (1:25 wt/wt) overnight at 37°C, and the tryptic peptides were separated by RP-HPLC, using a Brownlee RP-300 Aquapore C8 column developed with a 0 -70% acetonitrile gradient in 0.1% trifluoroacetic acid over 60 min.
Amino Acid Sequencing-Automated Edman degradation chemistry was used for amino acid sequence analysis. A model 470A gas phase sequencer (Applied Biosystems, Inc.) was employed for the degradations using the standard sequencer cycle, 03RPTH (19). The respective PTH-derivatives were identified by RP-HPLC analysis in an on-line fashion employing a model 120A PTH analyzer (Applied Biosystems, Inc.) fitted with a 2.1-mm (inner diameter). PTH-C18 column (Brownlee).
Enzyme Purification from E. coli-The plasmid pMON9428 was transformed into E. coli W3110 and grown in a 10-liter fermenter in LB broth at 30°C. The culture was induced for 2 h with 50 g/ml nalidixic acid and then harvested by centrifugation in a Sharples solid bowl centrifuge. The cell pellet was stored at Ϫ20°C until use. One kilogram of paste was thawed and resuspended in 3 liters of 100 mM Tris, pH 8.5, and passed twice through a Manton-Gaulin press at 8,000 p.s.i. with cooling to 15°C. The crude extract was obtained after centrifugation at 8,000 ϫ g for 25 min. Protamine sulfate (pH 8.0) was added dropwise to 0.2% with stirring and nucleic acids removed by centrifugation as above. The PEH protein was precipitated by addition of an equal volume of a saturated solution of ammonium sulfate, harvested by centrifugation at 10,000 ϫ g for 20 min, resuspended, and dialyzed into anion exchange buffer (AE buffer) including 10 mM Bis-Tris propane, pH 9.0. A portion of the recovered protein (34 g; the remaining 40 g was frozen at Ϫ70°C) was overloaded onto a 5 ϫ 30-cm Q-Sepharose FF (Pharmacia) column equilibrated in AE buffer. The column was washed with AE buffer plus 200 mM KCl and the PEH eluted with a 200 -750 mM KCl gradient (4 liters). As a final purification, 5.4 g of protein was brought to 5% ammonium sulfate and loaded on a 5 ϫ 20-cm phenyl-Sepharose column equilibrated in 5% ammonium sulfate, 20 mM Bis-Tris propane, pH 9.0. A gradient of 5-0% ammonium sulfate was formed over 2 liters and the PEH activity eluted with an additional 1 liter of 0% ammonium sulfate buffer. The active fractions were pooled and dialyzed extensively against 20 mM Bis-Tris propane, pH 7.0, and 0.5 mM EDTA.
Enzyme Activity pH Optimum-The optimum pH for activity was determined using p-nitrophenyl phenylphosphonate as a substrate. A three-buffer system of constant ionic strength (20) was used, which consisted of 0.052 M MES, 0.052 M HEPES, and 0.1 M diethanolamine. The actual pH values after dilution of the buffer with substrate were recorded. Enzyme was preincubated at each pH for 2 min, and then assays were run for 5 min. Assays were stopped by the addition of diethanolamine base to 0.1 M and the absorbance at 404 nm recorded. A second set of 5-min assays was performed after first incubating the enzyme at various pH values for 12 min at 30°C.
Metal Ion Analysis-The trace metal analysis was performed by inductively coupled plasma-atomic emission spectroscopy. The samples were prepared into acidic aqueous solutions using a low temperature ashing method. About 1 g of sample was weighed and transferred into a quartz boat. The boat was loaded in a vacuum chamber and ashed by oxygen plasma overnight. The ash residues were digested with 1 ml of concentrated nitric acid on a hot plate at about 150°C. The solution was dried slowly under nitrogen purge to reduce contamination problems. The dry residue was dissolved in 1 ml of 5% nitric acid solution. The solution then was measured by a Jarrell-Ash inductively coupled plasma-atomic emission spectroscopy system. Molecular Techniques-Restriction maps, cloning, Southern blots, and other DNA manipulations were performed using standard techniques (21) except when referenced otherwise. Genomic DNA from PG2982 was prepared as described previously (22) and used as a PCR template, for Southern analysis, and for making a genomic library.
Cloning the pehA Gene-PCR was used to amplify gene segments between the sequences encoding the N terminus and the T37 tryptic fragment and the T20 and T37 tryptic fragments. The PCR primers were designed using a codon preference table developed for PG2982 genes. 3 The PCR primer, ATC GTA/G GAT CAG TGC CGC GCA GAT TTC ATC CCG CAT CTA ATG, was made from the codons predicted from the N-terminal amino acid sequence. A slash indicates an equimolar mixture of two nucleotides at that position. Two PCR primers, GAA/G GAC/T ATC/T TGG CTN CC and GAA/G GAC/T ATC/T TGG TTA/G CC, were made from the T20 tryptic peptide sequence and mixed 2:1, respectively. A single primer, TGG/ACCG/C GTC/T TCA/G TC, was made from the predicted anti-codons for the T37 tryptic fragment sequence. The PCR reactions were performed with a 40°C annealing step using the Taq polymerase under standard conditions (Perkin Elmer).
The 450-bp T20-T37 PCR product was used to screen a PG2982 genomic cosmid library essentially as described (21). The PG2982 genomic library was prepared from a partial HindIII digest ligated into the HindIII site of cosmid pHC79 (23) as described previously (22). 4 Colonies that hybridized to the probe on each of the duplicate filters were further screened (confirmed) by PCR using the T20 and T37 primers. Three cosmid clones were selected and cosmid DNA prepared using a rapid alkaline lysis method (25). The cosmid DNA was then digested with BglII, BamHI, ClaI, NcoI, HindIII, and EcoRI, and the pehA gene was mapped by Southern analysis (24) using the 32 P-labeled N-terminal-T37 PCR product as a probe. The results were used to partially construct the restriction map in Fig. 3. The 3-kb NcoI and 2.2-kb HindIII fragments containing the pehA gene were ligated into a pUC118-derived vector (modified to contain a NcoI site in the polylinker) and Bluescript pSK (Clontech), respectively. The NcoI and HindIII fragments were mapped using common restriction enzymes and the data combined to produce the restriction map in Fig. 3.
Nucleotide Sequencing-Single-stranded DNA template was used to sequence both strands of the entire pehA gene. Single-strand DNA was produced using the M13 helper phage M13KO7 (Bio-Rad) and purified using a standard protocol (21). DNA sequencing utilized the Sequenase 2.0 kit (U. S. Biochemical Corp.). Initially, the degenerate PCR primers designed from tryptic fragment sequences were used as sequencing primers. As sequence data became available, new primers were synthesized until a complete set of primers were available every 250 bp for both strands, allowing the gene to be completely sequenced on both strands.
Site-directed Mutagenesis-The primer, GTAAGCCTCGGAAATA-AAGATCTCACCATGGCCAGAAAAAATGTCCTG, was used to insert BglII and NcoI recognition sites at the starting methionine of the pehA gene using Kunkel mutagenesis (11). A 5Ј NcoI-BamHI fragment of the pehA gene was ligated with a BamHI to SalI 3Ј fragment into an E. coli expression vector containing the recA promoter and T7 phage gene10 leader (26) and the resulting plasmid designated pMON9428.
The primer TTGCTCCTGAGCTCAATGGTTGC was used to insert a SacI site just 3Ј to the predicted stop codon. To facilitate further cloning, the primers GAAACGCGGATCTCTTGCAGAGGT, ATACGGAAGCT-CTCGGCATTGTA, GAGCCTTCCGCCCATGAAAGAACGAGCC, and CAGATTGCTGAACTCATGCGGGTC were used to remove internal BamHI, HindIII, NcoI, and EcoRI sites, respectively. After mutagenesis, the pehA gene resided on a NcoI-SacI fragment cloned into pBluescript SK(ϩ) and was designated pMON9432. The pehA gene was then resequenced with the deoxyinosine reagent set (Sequenase 2.0 kit) to confirm correct mutageneses and activity was verified in E. coli. The 3Ј-noncoding sequence between the newly introduced SacI site and a vector SacI site was removed from pMON9432 yielding pMON9434, and pehA activity was again confirmed in E. coli by the in situ hydrolysis of XPP.
Miscellaneous Procedures-All nucleic acid sequences were analyzed on a VAX using the GCG sequence analysis programs (27). Data base searches were performed using the BLAST algorithm against the nonredundant sequence data bases at the National Center for Biotechnology Information (28). Enzyme kinetic data were analyzed using the ENZFITTER program (29).

Identification of Glyceryl Glyphosate Phosphonate Ester
Hydrolase Activity-B. caryophilli PG2982 has been previously characterized for its ability to utilize glyphosate as a sole phosphorus source (8). In this study, PG2982 was observed to utilize glyceryl glyphosate as a sole phosphorus source, suggesting the presence of a phosphonate monoester hydrolase activity. The glyceryl glyphosate phosphonate esterase activity was confirmed by directly demonstrating the conversion of glyceryl glyphosate to glyphosate using in vivo and in vitro radioactive assays. Intact PG2982 cells and crude extracts were incubated with glyceryl [3-14 C]glyphosate and the products identified by HPLC anion exchange chromatography. The only radioactively labeled specie formed from glyceryl [3-14 C]glyphosate was glyphosate, thereby confirming that hydrolysis of the ester was the first step in the mineralization of glyceryl glyphosate. The PG2982 PEH appeared to be expressed constituitively and was unaffected by growth in DF medium with 0.2 mM phosphate, in L-broth, or in M9 medium with 100 mM phosphate (data not shown).
The PEH reaction products were further authenticated in crude extracts of PG2982 incubated with 10 mM cold glyceryl glyphosate for 4 h at 30°C. The enzyme-dependent formation of glycerol was verified by coupling the reaction to glycerol dehydrogenase, and the appearance of glyphosate was verified using a HPLC assay. The purified PG2982 PEH (see below) was also incubated with 10 mM glyceryl glyphosate in 30% D 2 O, and the time-dependent formation of glyphosate was confirmed with 31 P NMR.
An E. coli phosphonate ester hydrolase activity has been described previously (1) and was evaluated for activity against glyceryl glyphosate, in order to investigate whether the glyceryl glyphosate phosphonate ester hydrolase activity was unique to PG2982. The growth of E. coli was expected to be similarly inhibited by glyphosate and glyceryl glyphosate, if the E. coli phosphonate ester hydrolase activity was able to hydrolyze the phosphonate monoester bond of glyceryl glyphosate. However, glyceryl glyphosate was observed to be at least 50fold less inhibitory than glyphosate to E. coli JM101 when plated on minimal medium (data not shown). Consistent with these results, no hydrolysis of glyceryl [3-14 C]glyphosate was observed when E. coli strains JM101 and MM294 were grown in the presence of the radioactive compound for 48 h in DF or MOPS minimal medium with either 0.2 mM or 0.01 mM (limiting) phosphate or in LB medium. The washed E. coli cell pellets contained a substantial amount of intact radioactive substrate, indicating that the lack of hydrolysis was not due to poor uptake (data not shown). The recovery of intact glyceryl glyphosate in these experiments implied that the putative E. coli phosphonate esterases were unable to hydrolyze the phosphonate ester bond. Several commercial enzyme preparations capable of hydrolyzing p-nitrophenyl phenylphosphonate were tested for their ability to hydrolyze glyceryl glyphosate. Phosphodiesterase I (Sigma P6903; 0.14 mol/min) and 5Ј nucleotidase (Sigma N-4005 from Crotalus adamanteus venom; 2.5 mol/min) were incubated with glyceryl [3-14 C]glyphosate (8 mM) in 100 mM diethanolamine, pH 9.0, for 60 min at 30°C. No hydrolysis of glyceryl glyphosate was observed (Ͻ0.1 nmol/ min). In summary, the PG2982 PEH activity appeared to be novel because other known phosphonate ester hydrolases were unable to hydrolyze glyceryl glyphosate.
Characterization and Partial Sequence of the PG2982 Phosphonate Monoester Hydrolase-To aid in further characterization of the PEH enzyme, the corresponding gene was cloned. The cloning strategy began with the purification of the enzyme from PG2982 in order to obtain amino acid sequence information. Purification was assisted by the development of a qualitative colorimetric assay, which measured the release of glycerol from glyceryl glyphosate. At the end of the purification, the PG2982 PEH activity appeared homogeneous, as evidenced by a single silver-stained band after native PAGE. A band of phosphonate ester hydrolase activity, which corresponded to the single silver-stained protein, was evident when the gel was incubated with ␤-naphthyl phenylphosphonate or when gel slices were incubated with glyceryl [3-14 C]glyphosate. These results demonstrated that the silver-stained protein was the PEH enzyme and, importantly, that the enzyme possessed a broad substrate specificity beyond glyceryl glyphosate. Separation on a isoelectric focusing gel (range of pH 4 -6) revealed a single band with a pI of 4.2 that stained in situ with ␤-naphthyl phenylphosphonate. During purification, the enzyme eluted as a single peak on Sephacryl S200 with an apparent native molecular mass of 240 kDa.
The purified PEH enzyme, which migrated as a single band by native PAGE, was resolved by SDS-PAGE revealing 66-and 59-kDa polypeptides (Fig. 1). Protein sequence and tryptic map analyses were employed to decide if the presence of the two polypeptides resulted from partial proteolysis or if they represented heteromeric subunits of the phosphonate monoester hydrolase. Initially, a single N-terminal sequence was obtained from the mixture of both polypeptides (Table I). Following purification of each polypeptide using preparative SDS-PAGE (Fig. 2), the individual N-terminal sequences were found to be identical and confirmed the sequence obtained from the mixture (Table I). Furthermore, tryptic profiles of the two polypeptides appeared nearly identical (Fig. 2). The similar tryptic profiles and identical N-terminal sequences suggested that the two polypeptides were encoded by the same gene and probably resulted from either post-translational modification or limited proteolysis during enzyme isolation. The possibility of alternate translation start sites was ruled out since the two polypeptides had identical N-terminal sequences. In addition to the N terminus, sequences were obtained for two tryptic fragments of the 66-kDa polypeptide, T20 and T37, which were subsequently used to clone the pehA gene (Table I).
Cloning the Phosphonate Monoester Hydrolase pehA Gene-Probes for the PG2982 pehA gene were obtained by PCR using degenerate primers designed from the tryptic fragment sequences described above. A 450-bp product was amplified using primers designed from the T20 and T37 tryptic peptide sequences, and a 880-bp fragment was amplified using primers designed from the N terminus and T37 tryptic peptide sequences. The T20-T37 450-bp PCR product was used as a probe to obtain a full-length pehA gene from a PG2982 cosmid library. Three cosmid clones were identified from screening 1800 colonies and the pehA gene mapped to a 3.2-kb SalI fragment (Fig.  3). Starting with degenerate sequencing primers designed from the N-terminal and tryptic fragment sequences, the entire gene was sequenced on both strands (Fig. 4). The starting methionine was identified as the only in-frame methionine between the encoded N-terminal amino acid sequence and an upstream in-frame stop codon. The N-terminal and tryptic fragment sequences (T20, T32, and T37) obtained from the purified PEH protein were identified in the deduced amino acid sequence, confirming the intended gene had been cloned (Fig. 4). The predicted size of the protein encoded by the pehA open reading frame was 58.2 kDa, which was 12% smaller than the 66-kDa polypeptide previously observed by SDS-PAGE for the purified PG2982 protein (Fig. 1), indicating the protein migrated somewhat anomalously on SDS-PAGE. The predicted pI was 5.8 and varied significantly from the observed pI of 4.2, likely reflecting intramolecular interactions of ionizable groups within the PEH polypeptide.
The complete deduced amino acid sequence of the pehA gene was used to search for similarities with other known proteins using the BLAST searching algorithm (27). Data base hits were individually evaluated in an attempt to identify evolutionarily distant similarities. In particular, no significant homologies to known phosphodiesterases were observed. Some similarity to a conserved family of sulfatases was noted; however, further analysis revealed that the pehA gene failed to share many of the highly conserved sulfatase consensus sequences. As an additional significance test of the sulfatase homology, the aryl sulfatases from Abalone entrails (Sigma catalog no. S9629) and from Patella vulgata (Sigma catalog no. S8504) (0.14 mol/min) were incubated with glyceryl [3-14 C]glyphosate, and the reaction products were analyzed by anion exchange HPLC. Neither enzyme hydrolyzed glyceryl glyphosate at the minimal detectable rate of Ϸ0.14 nmol/min/mg. Furthermore, purified PEH (see below) did not hydrolyze p-nitrophenyl sulfate, a substrate common to many aryl sulfatases. Therefore, the similarity between the PG2982 pehA gene and the aryl sulfatase family may have little significance. No other similarities were observed that would provide clues to the evolutionary origin of the PEH protein.
Expression of the Phosphonate Monoester Hydrolase in E. coli-The PG2982 phosphonate monoester hydrolase gene was engineered for overexpression in E. coli. First, a NcoI site was inserted at the starting methionine codon by site-directed mutagenesis (also resulting in the insertion of an alanine at position two). A 2.2-kb NcoI to SacI restriction fragment containing the pehA gene and some 3Ј-untranslated region was assembled behind a RecA promoter with the T7 phage gene10 leader resulting in plasmid pMON9428 (Fig. 5). The expression of the PG2982 phosphonate ester hydrolase in E. coli was confirmed using in vivo and in vitro radioactive glyceryl glyphosate HPLC assays and in situ XPP staining. Significant hydrolysis of glyceryl [3-14 C]glyphosate to glyphosate was observed when cells of E. coli transformed with pMON9428 were incubated with the radioactive compound for one hour. Furthermore, crude extracts of E. coli transformed with pMON9428 efficiently hydrolyzed glyceryl glyphosate. Finally, E. coli transformed with pMON9428 turned blue when plated on medium containing XPP while wild-type E. coli colonies were white. These results provided conclusive evidence that the 2.2-kb fragment did encode the PG2982 phosphonate ester hydrolase. Kunkel mutagenesis was used to introduce a SacI site in pMON9428 2 nucleotides downstream of the predicted stop codon and the 3Ј-noncoding region was removed as a SacI fragment. PEH activity was still observed in E. coli using in situ XPP staining, which confirmed the deduced 3Ј-untranslated region was not essential for enzyme activity.
Characterization of the Phosphonate Monoester Hydrolase-The PG2982 phosphonate monoester hydrolase was overex-TABLE I Amino acid sequencing of the 66-and 59-kDa PEH polypeptides A single N-terminal sequence was determined from a purified PEH preparation, which contained both 66-and 59-kDa polypeptides. N-terminal sequences were also obtained for the individual 66-and 59-kDa PEH polypeptides after they were purified from each other using preparative SDS-PAGE. Tryptic peptides were isolated by RP-HPLC after trypsin digestion of the 66-kDa polypeptide. Unequivocal sequence was obtained for three tryptic peptides designated T20, T32, and T37.  2. Tryptic maps reveal the 66-kDa and 59-kDa polypeptides are highly related. The 66-kDa and 59-kDa polypeptides, which correlated with phosphonate monoester hydrolase activity (Fig. 1), were each purified by preparative SDS-PAGE. A portion of each preparation was analyzed by SDS-PAGE using a 10% Phastgel and silver stain (right). Each preparation appeared to be composed substantially of a single polypeptide. Each polypeptide was then reductively alkylated and subjected to trypsin digestion. Tryptic maps were obtained by separating tryptic peptides on a Brownlee C8 reverse phase column using an acetonitrile gradient in 0.1% trifluoroacetic acid and UV detection at 214 nm. The tryptic map for the 66-kDa polypeptide (top left) and the 59-kDa polypeptide (bottom left) appeared nearly identical, suggesting the two polypeptides originated from the same gene. pressed in E. coli transformed with pMON9428 and then purified 3.5-fold to homogeneity (Table II). The heterologously expressed enzyme had an apparent native molecular mass of 240 kDa, determined by chromatography on a Sephacryl S400 column. A single polypeptide of 66 kDa was observed on SDS-PAGE, confirming the enzyme was a homotetramer. The pI of the heterologously expressed protein was 4.2. Therefore, the enzyme expressed in E. coli appeared identical to the native enzyme found in PG2982 using these extrinsic criteria. Amino acid composition of the purified protein also was consistent with the deduced amino acid sequence of the pehA gene (data not shown). The variation of PEH activity with pH was determined using a three-buffer system at constant ionic strength (Fig. 6). The pH was varied between 5 and 10, and PEH reactions were run for 5 min at 30°C after 2-min and 12-min preincubations. PEH activity was stable over the course of the experiment since similar rates were observed after either a 2or a 12-min preincubation at the various pH conditions. A bell-shaped curve was observed for PEH activity with a maximum activity at pH 9.0.
Metal Ion Activation of the Phosphonate Monoester Hydrolase-The effect of inorganic ions on PEH activity was evaluated using the purified enzyme. In general, anions such as Cl Ϫ , CH 3 COO Ϫ , and SO 4 2Ϫ salts stimulated activity 20 -40% at 100 mM. Heavy metal ions such as Cu 2ϩ and Zn 2ϩ were potent inhibitors. Co 2ϩ appeared to stimulate activity; however, subsequent analysis indicated that the Co 2ϩ per se chemically catalyzed the hydrolysis of glyceryl glyphosate at pH 9.0. Manganese ions stimulated the purified enzyme 2.5-fold, while Mg 2ϩ , Ca 2ϩ , Fe 2ϩ , and Fe 3ϩ had little or no effect.
The stimulatory effect of Mn 2ϩ ions on PEH activity was further explored. PEH activity was fully stimulated by Mn 2ϩ concentrations as low as 0.1 M. The low concentration of Mn 2ϩ required for full activation confirmed that the PEH enzyme had high affinity for Mn 2ϩ and suggested the possibility that Mn 2ϩ might be required for enzyme catalysis. To test this hypothesis, a PEH preparation, with a specific activity of 6.9 mol/min/mg at saturating pNPP in the presence of saturating Mn 2ϩ , was dialyzed against EDTA and subjected to metal ion analysis. The total remaining manganese was determined to be 1.58 M, while the protein content, determined by amino acid analysis, was 464 M. In this dialyzed preparation, only 1.58 M enzyme or 0.34% of the total enzyme would be catalytically active with a predicted specific activity of 0.02 mol/min/mg, assuming that Mn 2ϩ was essential for activity. However, the V max of the dialyzed "metal"-free preparation was 0.6 mol/min/mg at saturating pNPP, which suggested that Mn 2ϩ was not essential for catalysis. Upon addition of excess MnCl 2 to the dialyzed apoenzyme preparation, the PEH V max was 8.7 mol/min/mg at saturating pNPP and was comparable to the original activity of 6.9 mol/min/mg, indicating that the apoenzyme was not irreversibly inactivated during the experiment. Overall, a 14-fold activation of the apoenzyme by Mn 2ϩ ions was observed. The difference from the 2.5-fold activation observed earlier indicated the Mn 2ϩ binding sites were partially saturated on the native protein. Interestingly, the activation resulted from an increased rate of catalysis rather than an increase in substrate binding affinity because the observed K m value (1.5 mM) for pNPP in the absence of MnCl 2 was slightly higher in the presence of MnCl 2 (3.3 mM). One possible role of the Mn 2ϩ cofactor may be to stabilize negative charges on a transition state intermediate.
Substrate Specificity of the Phosphonate Monoester Hydrolase-The PEH enzyme has been shown to hydrolyze both glyceryl glyphosate and ␤-naphthyl phenylphosphonate. The Overlapping sequence data was obtained for both strands of the coding region. The deduced amino acid sequences are underlined that correspond to the N terminus and tryptic fragment amino acid sequences obtained for the purified PEH protein (Table I).

FIG. 5. Plasmid map of pMON9428 E. coli expression vector. An
NcoI site was introduced at the starting methionine of the pehA gene, which concomitantly introduced an alanine residue at position 2. The pehA gene was subcloned into an E. coli expression vector containing the recA promoter with a T7 gene10 translational enhancer (24). The vector was maintained in E. coli with 100 g/ml spectinomycin and the promoter induced with 50 g/ml nalidixic acid. kinetics of PEH for glyceryl glyphosate were difficult to determine due to an unusually high K m estimated to be 49 Ϯ 13 mM (Fig. 7). The relatively high K m was not an artifact of expressing the PG2982 gene in E. coli, since similar kinetics were observed in crude extracts of PG2982. Furthermore, the solubility of glyceryl glyphosate at pH 9 was Ͼ1.0 M and did not affect the experiment. The k cat for hydrolysis of glyceryl glyphosate was estimated at 2.8 ϫ 10 1 min Ϫ1 , resulting in a specificity constant of 5.7 ϫ 10 2 min Ϫ1 M Ϫ1 (Table III). The poor catalytic efficiency for the xenobiotic, glyceryl glyphosate, was consistent with the idea that this enzyme was evolutionarily optimized for an, as of yet, unidentified intracellular reaction. A survey of commercially available colorimetric substrates was undertaken to provide clues to the identity of the intracellular PEH reaction. The enzyme hydrolyzed phosphonate monoesters including p-nitrophenyl phenylphosphonate (Table  III), 5-bromo-4-chloro-3-indolyl phenylphosphonate, and ␤-naphthyl phenylphosphonate. The latter compounds were visually identified as substrates by the rapid, enzyme-dependent appearance of an insoluble pigment. Phosphate diesters were considered to be more physiologically relevant and structurally related to phosphonate monoesters. The purified PEH catalyzed the hydrolysis of bis-(p-nitrophenyl) phosphate and p-nitrophenyl thymidine-5Ј-phosphate (Table III); however, no significant rate was observed for tris-(p-nitrophenyl) phosphate. Based on the observed rate for thymidine-5Ј-phosphate, DNA and cGMP were tested as substrates but were not hydrolyzed. Additionally, no activity was observed with either pnitrophenyl phosphate or p-nitrophenyl acetate.
The most catalytically efficient substrate was bis-(p-nitrophenyl) phosphate with a K m of 0.9 mM and a k cat of 6.2 ϫ 10 2 min Ϫ1 and a catalytic efficiency of 7.2 ϫ 10 5 min Ϫ1 M Ϫ1 (Table  III). Interestingly, the least catalytically efficient substrate was glyceryl glyphosate despite the fact that the enzyme was first identified as a glyceryl glyphosate hydrolase. Relative to many other enzymes, all of the substrates identified for PEH are only poorly hydrolyzed, with k cat /K m values that are significantly lower than diffusion controlled association rates. DISCUSSION B. caryophilli PG2982 is a well known bacterium able to mineralize glyphosate (8) and was tested for the ability to hydrolyze glyceryl glyphosate. As it turned out, PG2982 was able to utilize glyceryl glyphosate as a sole phosphorus source due to the activity of a phosphonate monoester hydrolase with broad substrate specificity. This observation provides the first, albeit tenuous, biological role for a phosphonate ester hydrolase. The PG2982 hydrolase was constituitively expressed when cells were grown in DF medium with 0.2 mM phosphate, L-broth, or M9 medium with 100 mM phosphate, demonstrating that expression was not controlled by the phosphate operon. Therefore, the glyceryl glyphosate phosphonate monoester hydrolase did not appear related to the glyphosate degradation pathway described by Moore et al. (8) since glyphosate degradation enzymes were reported to be linked to the phosphate starvation operon (28).
The PG2982 enzyme was designated a phosphonate monoester hydrolase since the enzyme was purified to homogeneity based on the hydrolysis of glyceryl glyphosate. However, glyceryl glyphosate is a xenobiotic, and the enzyme likely evolved from a gene encoding a protein with activity against a naturally occurring substrate. The original biological role of the PEH is still unclear, since the purified enzyme exhibited a  6. The pH optimum for PEH activity. A three-buffer system with constant ionic strength of 0.1 M between pH 5 and 10 was employed to measure PEH activity with 4 mM p-nitrophenyl phenylphosphonate as a substrate. The actual pH values after dilution of the buffer with substrate were recorded. Enzyme was preincubated at each pH for 2 min, and then assays were run for 5 min. Assays were stopped by the addition of diethanolamine base to 0.1 M and the absorbance at 404 nm recorded. A second set assays were performed after first incubating the enzyme at the various pH values for 12 min at 30°C.
FIG. 7. The kinetics of glyceryl glyphosate hydrolysis by purified PEH. PEH activity was measured at increasing concentrations of glyceryl glyphosate. Reactions were performed in 20 mM diethanolamine, pH 9.0, 500 M MnCl 2 at 30°C. The appearance of glyphosate was assayed at fixed time points using a radioactive HPLC analysis. The appearance of glyphosate was previously observed to be linear with time and with enzyme concentration. The K m and k cat were determined by fitting the data to the Michaelis-Menten equation using nonlinear regression analysis (29). broad substrate specificity. Hydrolysis of bis-(p-nitrophenyl) phosphate suggested the enzyme was a general phosphodiesterase, while p-nitrophenyl thymidine-5Ј-phosphate hydrolysis indicated the enzyme may be specifically a 5Ј-nucleotidase. However, no activity against DNA or cyclic GMP was observed. Clearly, the enzyme was not a general phosphatase or esterase, since little activity was observed for p-nitrophenyl phosphate or p-nitrophenyl acetate. Comparison of the amino acid sequence to protein sequence data bases revealed a evolutionarily distant similarity to the family of aryl sulfatases, yet the purified enzyme was unable to hydrolyze p-nitrophenyl sulfate. Likewise, two commercially available aryl sulfatases were unable to hydrolyze glyceryl glyphosate. Further biochemical analysis, particularly testing 5Ј nucleotides as substrates, and genetic disruption of the pehA gene may reveal the role of this enzyme in PG2982 metabolism.
As a family of compounds, alkyl and alkoxy esters, with 3 or more carbons, of glyphosate were observed to exhibit at least 10-fold less vegetative phytotoxicity than glyphosate in herbicide field trials at Monsanto (7). The differences in phytotoxicity were unlikely due to transport differences, since these compounds were similar with respect to these properties. Phosphonate esterases with broad substrate affinities are widespread in nature, including plants (1), and were expected to hydrolyze the glyphosate esters resulting in apparent phytotoxicities similar to glyphosate, given phosphonate monoester hydrolysis is amenable to enzyme catalysis. However, the common plant phosphonate esterases appeared to have little activity on glyphosate esters. Likewise, E. coli enzymes were not observed to hydrolyze these glyphosate esters.
The occurrence of a unique enzyme that will hydrolyze a phosphonate ester of glyphosate may find interesting uses in genetics as a conditionally lethal gene. The hydrolysis of glyceryl glyphosate is not growth-inhibitory to PG2982, since it can efficiently metabolize glyphosate (8). However, in plants, fungi, and E. coli, where glyphosate is a potent inhibitor of aromatic amino acid biosynthesis, the release of glyphosate from a nontoxic phosphonate monoester would conceivably result in cell death. The pehA gene seems particularly well suited for heterologous expression and use as a conditionally lethal gene in plants. The PEH activity is encoded by a single polypeptide and does not require an unusual cofactor for activity, although the enzyme is stimulated by Mn 2ϩ ions, which are found in plant cells. The broad pH optimum for PEH activity makes the enzyme suitable for plastid or cytosolic expression. Certainly, the unknown intracellular function of the pehA gene makes it difficult to predict what effects might be observed on plant metabolism; however, there were no discernible effects when the active protein was overexpressed within E. coli cells. Current research is exploring the potential of the pehA gene as a conditional lethal gene in plant genetics (30). The kinetic constants of PEH substrates were determined. Reactions were performed in 20 mM diethanolamine, pH 9.0, 500 M MnCl 2 at 30°C. For p-nitrophenyl substrates, the appearance of p-nitrophenol was measured continuously by spectrophotometric detection at 405 nm. Reaction rates were calculated by least squares analysis of the change in absorbance over time. The hydrolysis of glyceryl glyphosate was assayed at fixed time points using a radioactive HPLC analysis. Glyceryl glyphosate 49 Ϯ 13 2.8 ϫ 10 1 Ϯ 5 5.7 ϫ 10 2 p-Nitrophenyl phenylphosphonate 2.3 Ϯ 0.08 4.0 ϫ 10 2 Ϯ 5 1.8 ϫ 10 5 Bis-p-nitrophenyl phosphate 0.9 Ϯ 0.03 6.2 ϫ 10 2 Ϯ 7 7.2 ϫ 10 5 p-Nitrophenyl thymidine-5Ј-phosphate 4.7 Ϯ 0.8 8.9 ϫ 10 1 Ϯ 9 1.9 ϫ 10 4