Fur. J. Biochem. 209.735-743 (2992)

0FEBS 1992

Cloning, overexpression and mechanistic studies of carboxyphosphonoenolpyruvate mutase from Streptomyces hygroscopicus Scot1 J. POLLACK

I,

Sally FREEMAN'.', David L. POMPLIANO' and Jeremy R. KNOWLES'

' Departments of Chemistry and Biochemistry, Harvard University, Cambridgc, Massachusetts, USA Pharmaceutical Sciences Institute, Aston University, Birmingham, England (Received May 15, 1992) - EJB 92 0672

The enzyme carboxyphosphonoenolpyruvate mutase catalyses the formation of one of the two C-P bonds in bialaphos, a potent herbicide isolated from Streptomyces hygroscopicus. The gene encoding the enzyme has been cloned from a subgenomic library from S. hygroscopicus by colony hybridisation using an exact nucleotide probe. An open reading frame has been identified that encodes a protein of molecular mass 32700 Da, in good agreement with the subunit molecular mass of the carboxyphosphonoenolpyruvate mutase recently isolated from this source [Hidaka, T., Imai, S., Hara, O., Anzai, H., Murakami, T., Nagaoka, K. & Selo, H. (1990) J . Bacteriol. 172, 3066-330721. 'The gene shares significant sequence similarity with that of phosphoenolpyruvate mutase, an enzyme that catalyses the related interconversion of phosphoenolpyruvate and phosphonopyruvate. When the carboxyphosphonoenolpyruvale-mutase gene was subcloned into the vector pETl1 a, the mutase was expressed as about 20% of the total soluble cellular protein in Escherichia coli. The mutase has been purified to homogeneity in three steps in 40% yield. With malate dehydrogenase/NADH, (hydroxyphosphiny1)pyruvate gives (hydroxyphosphiny1)lactate (k,,, 164 s and K, 680 pM) and this spectrophotometric assay for the product of the mutase rcaction has been employed in the mechanistic studies. The kinetics for the mutase reaction have been evaluated for the substrate, carboxyphosphonoenolpyruvate, and for the putative reaction intermediate carboxyphosphinopyruvate, both of which have been prepared by chemical synthesis. Carboxyphosphonoenolpyruvate is converted to (hydroxyphosphiny1)pyruvate with a k,,, of 0.020 s-' and a K, of 270 pM, and carboxyphosphinopyruvate is converted to (hydroxyphosphiny1)pyruvate with a k,,, of 7.6 x s-' and a K, of 2.2 pM. Although the exogenously added intermediate is not kinetically competent, these results suggest that the mechanism for the mutase reaction involves an initial rearrangement to the intermediate carboxyphosphinopyruvate, followed by decarboxylation to yield the product (hydroxyphosphiny1)pyruvate.

Naturally-occurring phosphonates are an unusual class of metabolites. the biochemical significance of which is only beginning to be understood. Several classes of lower oi-ganisms produce phosphonates, in compounds ranging from the antibiotics produced by Streptomyces to the phosphonolipids that occur in the membranes of several species of ciliated protozoa (Hori et al., 1984; Hilderbrand, 1983). Two recently isolated enzymes reprcsent the only known examples of biological catalysts for the formation of the C-P bonds in these compounds. One of these enzymes, phosphoenolpyruvate mutase (Bowman et al., 1988; Seidel et al., 1988; Hidaka et al., 1989), catalyses the interconversion of phosphoenolpyruvate (PEP, 1 ; Scheme 1) and phosphonopyr-

uvate (2; Scheme 1). This rearrangement is thought to proceed via a covalent phospho-enzyme intermediate, a conclusion that is based primarily on the finding that the overall reaction proceeds with net retention of the configuration at phosphorus (Freeman et al., 1989; Seidel et al., 1990; McQueney et al., 1991). The C-P bond-forming reaction is energetically unfavorable and must be driven forward by a subsequent exergonic reaction (Bowman et al., 1988; Seidel et al., 1988). The second C-P bond-forming enzyme, carboxyphosphonoenolpyruvate mutase (Hidaka et al., 1989; Hidaka et al., 1990), catalyses the formation of the C-P bond of (hydroxyphosphiny1)pyruvate (Scheme 2,4). This reaction involves the rearrangement and decarboxylation of carboxyphosphono-

Correspondence to S. Freeman, Pharmaceutical Sciences Institute. Aston IJniversily, Birmingham, England B4 7ET Ahhreviations. CPEP, carboxyphosphonoerzolpyruvate; PEP, phosphoenolpyruvate. Enzymes. Carboxyphosphonoenolpyruvate mutase (EC 6.4. .-); Phosphoenolpyruvate mutase (EC 6.4.2.9) ; malate dehydrogenase (EC 1.1.1.37). Note. The novel nucleotide sequence data published here have been submitted to the EMBL, GenBank and DDBJ sequence data banks and are availablc under accession number X67953.

Scheme 1. The reaction catalysed by phosphoenolpyruvate mutase.

~

736

-0,

P

1

bialaphos

I;H Ala Ala

Scheme 2. The reaction catalyscd by carboxyphosphonoenolpyruvatemutasc, in the biosynthetic pathway that leads to bialaphos.

B = BamHl Bg = sgll E = EcoRl

RV = E&V S = SsH X =Xhol

3 kb

f BamHl

I

5b

Fig. 1. The 20-kb fragment comprising the bialaphos biosynthetic genes and the 3-kb BarnHI restriction fragment carrying the CPEP mutase gene.

cnolpyruvate (CPEP; 3, Scheme 2), giving (hydroxyphosphiny1)pyruvate. The decarboxylation drives the overall equilibrium toward C-P bond formation. This transformation is a key step in the biosynt.hesis of the herbicide bialaphos by Streptonzyces hygroscopicus. A third enzyme catalyses the methylation that results in the formation or the second C-P bond of bialaphos. To gain a better understanding of the mechanism of this class of enzyme-catalysed phospho group transfer reactions, we set out Lo clone and overexpress the genes that encode the enzymes responsible for C-P bond formation. The cloning of the PEP mutase gene from Tetrahyrnena pyrijorrnis and its ovcrexprcssion in Escherichiu coli has recently been described (Seidel et al., 1992). We report here the cloning of the gene for carboxyphosphonoenolpyruvate (CPEP) rnutase from S. hygroscopirus, overexpression of the gene product in E . coli, and mechanistic studies on the recombinant enzyme. To facilitate these mechanistic studics, we have prepared product 4 (Scheme 3) and characterised its interconversion with (hydroxyphosphiny1)lactate with malate dehydrogenase,"ADH. Substrate 3 (Scheme 3 ) was also prepared by chemical synthesis; only analytical quantities of substrate 3 had been isolated previously (Hidaka et al., 1990). We have prepared and determined the reaction kinetics of carboxyphosphinopyruvate (Scheme 3, 5), the putative reaction intermediate with

CPEP mutase. Two key mechanistic questions are whether the decarboxylation event is spontaneous or enzyme catalysed and whether it precedes, accompanies or follows the rearrangement. Carboxyphosphinopyruvate (Scheme 2, 5 ) is the putative intermediate in the pathway if the initial rearrangement is followed and driven forward by the decarboxylation (Scheme 2 ) .

EXPERIMENTAL PROCEDURES

Materials Genomic DNA was obtained from the bialaphos-producing strain S. hygroscopicus, ATCC 21705. E. coli strain AG1 (Stratagene) was used as host for construction of the library and for the large-scale preparation of plasmid DNA for sequencing and subcloning. E. coli strain BL21(DE3) (Novagen) was used for protein expression. For routine transformations, cells were madc competent by the CaClz method (Sambrook et al., 1989). Luria-Bertani broth (Gibco, BRL) was used for all cultures of E. coli.

737 Chemical syntheses

column (200 ml) of Dowex 50 X4 (H' form). The column was eluted with water (800 ml) and the pH of the eluate was Trisodium carboxyphosphoenolpyruvate { trisodium salt adjusted to 6.5 with 5 M NaOH. The eluate was diluted to 4 1 of'[ ( 1-carboxyetheyl) oxy]hydroxy phosphinecarboxylic acid and loaded onto a column (100ml) of AG1-8X (HCOOoxide] form) equilibrated with 10 mM triethylammonium bicarbonThe key stcp in the synthesis of compound 3 (Scheme 3) ate, pH 7.0 (buffer A). The column was eluted with a linear was similar to that described for the reaction of bis(trimethy1- gradient of this buffer (10-500 mM, 1 1 1 1). Fractions sily1)methyl phosphite with ethyl bromopyruvate (Sekine et (20 ml) were collected and the A z S s was measured to detect al., 1982). To a solution of ethyl bromopyruvate (6.77g, the presence of the pyruvoyl group (Anderson et al., 1984). 35 mmol) in dry benzene (50 ml) under argon at 0'C was Fractions containing the desired compound were pooled and added bis(trimethylsily1)methoxycarbonylphosphite (Issleib et concentrated. Isopropanol (50 ml) was added and removed al., 1985; 10.3 g, 38 mmol, 1.1 equivalents) dropwise over by rotary evaporation three times, to drive off excess buffer 5 min. The mixture was stirred at 0°C for 30 min, allowed to A. The p H of the solution was adjusted to 7 with NaOH at reach room temperature, then stirred for an additional frequent intervals. The residue was dissolved in water (100 ml) 100 min. [(1-Ethoxycarbonylethenyl)oxy]trimcthylsilyloxy- and applied to a cation exchange column (150 ml) of Dowex phosphine carboxylic acid methyl ester was isolated by frac- 50 X4 (Nat form). The column was then eluted with water tional distillation [(6.32 g, 20 mmol, 59%; boiling point 134(500 ml). Thc eluate, containing disodium (hydroxyphosphi140°C at 99.99 Pa); hP (101.3 MHz, CDC13), 18.4 ppm (s, 'H ny1)pyruvate (Scheme 3, compound 4), was concentrated and decoupled); hH (250.1 MHz, CDCI3) 5.94 ppm ( l H , dd, JPH dissolved in water (50 ml) for assay. Using the semicarbazide ~ 4 : ~ , . , , ~ 2Hz, . 0 vinyl), 5.65 ppm (IH, dd, J P H ~ J g r m ~Hz, 2 . 4 assay developed by Anderson et al. (1984) for phosphonovinyl), 4.23 ppm (2H. 4. JHH7.1 Hz, -OCH2CH3),3.83 ppm pyruvate, the yield of (hydroxyphosphiny1)pyruvate was (3H, d. J p H 1.1 Hz, PC(O)OCH3), 1.28 ppm (3H, t , J H H 8.8 mniol, 30.5% (assuming an absorption coefficient at 7.1 Hz, -OCH2CI13) and 0.35 ppm (9H, s, OSi(CH3),); hC 253 nm of 10000 M-lcm-', as for phosphonopyruvate). (62.9 MHz, CDC13) 143.35 ppm (d, JpC8.7 Hz, C=CH2), Using the malate dehydrogenase assay, the yield was 8.6 mmol 111.7 ppm (d, Jpc 5.0 Hz, C=CH2), 62.0 ppm (s, (30%); hp (121.5 MHz, H 2 0 with DzO inner lock) 17.7 ppm -OCH,CH,), 52.5 ppm (d, Jpc 5.8 Hz, OCH,), 14.0 ppm (s, (s, 'H decoupled; dt, JpH550.3, Jplr18.9 Hz, 'H coupled); bH (500 MHz, H 2 0 with d4-methanol inner lock, referenced to -OCH2CH,), 0.65 ppm (s,OSi(CH,),J. A solution of NaOH (1.5 g, 38 mmol, 3 equivalents) in HzO at 4.85 ppm) 7.33 ppm (IH, dt, J p H 550.3, J H H 1.7 Hz), 1.7 Hz); 6, (100.6 MHz, water (20 ml) was added rapidly to a stirred sample of [(l- 3.37 ppm (2H, dd, JPH18.9 Hz, Jrllr cthoxycarbonylethenyl) oxy] trimethylsilyloxyphosphine - car - H 2 0 with d,-methanol inner lock, referenced to C D 3 0 D at boxylic acid methyl ester (3.9 g, 13 mmol) at 0°C. The mixture 49.0 ppm) 199.9 ppm (br s, CH,-CO), 168.9 ppm (s, -COO-), was stirred at 0°C for 1 h, and was then left overnight at room 46.7 ppm (d, Jpc 71.7 Hz, P-C'Hz); mjz (negative-ion temperature. Methanol (40 ml) was added and the mixture FAB) 173 Da (1000/, M-Na+). The observed exact mass was filtercd to remove a small amount of insoluble material. 172.9623 Da; C3H3Na05Prequires 172.9616 Da. Another addition of methanol (40 ml) precipitated trisodium carboxyphosphoenolpyruvate (Scheme 3, compound 3), which was dried under vacuum over P 2 0 5 to give a white solid (tIydroxyphosphiny1)lactate [free acid o j (3-carhoxy-2(1.74 g, 6.6 mmol, 53%; melting point 275-280°C with dehydroxjfethyl)hydroxyphosphine oxide] composition; (C 1 8 . 0 0 %H ~ ~3.01%, P 11.48%. C4Hz07PNa3 Malate dehydrogenase (0.4 ml, 1000 U) was added to a requires C 18.34%, H 0.77%, P 11.82%); hP (101.3 MHz, D20). - 1.77 ppm (s, 'H decoupled); ak1(250.1 MHz, D 2 0 , stirred solution of (hydroxyphosphiny1)pyruvate (1.OX mmol referenced to HOD at 4.85 ppm) 5.54 ppm (lH, dd, JPH in 6.2 ml water), NADH (0.877 g, 1.I9 mmol) and 30 ml 0.1 Mes, pH 6.8. After 72 h at room temperature, the reaction zJgen,%1.7 Hz) and 5.20 ppm (lH, dd, J P H ~ J g e m z Hz); l.8 6,(62.9 MHz, D 2 0 ) 179.6 ppm (d, Jpc235.4 Hz, PCOO-), mixture was diluted to 1250 ml and loaded onto an AG1XX anion-exchange column (30 ml), equilibrated with 10 mM 173.4ppm(d,Jpc:6.1Hz,CCOO~),151.8ppm(d,JpC9.2H~, C=CH2) and 106.0 ppm (d, JpC 4.8 I-Iz, C=CHz); u(KBr), buffer A. The column was eluted with a linear gradient of 1603 cm-' and 1411 c m - '; m/z (positive ion FAB) peaks 10-500 mM buffer A (0.5 1 + 0.5 1). 31P-NMRspectroscopy consistent with molecular ion not observed, 197 Da (100%; showed that the fractions eluting at approximately 200 mM buffer A contained (hydroxyphosphiny1)lactate. These were product of decarboxylation). concentrated, then isopropanol (3 x 5 ml) was added and removed to eliminate excess buffer A. The residue was dissolved (Hydroxyphosphinyl)p-vruvate[disodium phosphinopyruvate; in water (50 ml) and applied to a Dowex-50 (50 ml, Na' form) disodiuni salt oj (3-carhoxy-2-0x0-ethyl) hydroxyphosphine cation-exchange column. The column was eluted with water OXillC] and the first 100 ml was concentrated. The sample contained Using thc transamination conditions described for the a small amount of NADH or NAD' , which was removed by preparation of phosphonopyruvate (Sparkes et al., 1990), a diluting in water (20 ml) and passing down a Dowex-50 (50 ml, solution of 2-amino-2-carboxyethylphosphonousacid (Ding- H ' form) column. The column was eluted with water and 20wall et al., 1989; 4.65 g, 30 mmol, 1.04 equivalents), glyoxylic ml fractions were collected. By 'P-NMR spectroscopy, it was acid monohydrate (2.66 g, 29 mmol) and Cu(02CCH3), determined that (hydroxyphosphiny1)lactic acid was present H 2 0 (5.99 g, 30 mmol, 1.04 equivalents) in 60 ml 1 M pyri- in fractions 2 and 3, which were concentrated to give a yellow dine and 1 M acetic acid was stirred at room temperature for gum (0.133 g, 0.869 mmol, 73%); hP (101.3 MHz, D 2 0 ) 11 h. The mixture was then evaporated to dryness and water 30.83 ppm (s, 'H decoupled) ( d x 4, JPH564 Hz, JPII14.9 Hz) (20 inl) was added and removed by rotary evaporation twice, which is slowly converted, by deuterium exchange, into to drive off residual pyridine and acetic acid. The residue was 30.46 ppm [ t (lines of equal height), JPD 86.5 Hz, 'H dissolved in water (100 ml) and loaded onto a cation-exchange decoupled] for the P-D compound; 6,, 7.19 ppm (lH, ddd, J,,

+

738 564.1 Hz, J H H 1.9, Jll,, 2.5, P-H), 4.59 ppm (lH, ddd, J p H 14.0, J H H 8.7, JHH5.2), 2.40-2.04 ppm (2H. m, P-CH2); 6c 179.4ppin(d,JpC12.2,COOH),68.5 ppm(d,JpC3.6,CHOH), -11.2" (c 4.45, 0.1 M 37.2 ppm (d, Jpc 92.2, P-CH2); HC1); mjz (positive-ion FAB) 155 Da (100Oi0, MfH'). The observed exact mass was 155.0110 Da; C3H805Prequires 155.0109 Da. Trisodium carbox~phosphinopyrut.ate[trisodium salt of (3carbo,~y-2-oxo-eth)l)hydroxyphusphinecarbox~dicacid oxideJ A solution of disodium (hydroxyphosphiny1)pyruvate (Scheme 3, compound 4) (1.8 mmol) in water (10 ml) was passed down a column (15 mlj of Dowex-50 (Hf form). The column was then eluted with water (100 ml) and the solution was concentrated. The 31P-('H-coupled)-NMR spectrum in H 2 0 / D 2 0gave two peaks at 18.8 ppm (approximately 60%, dt, JPH 565.8, 18.7 Hz) and 26.5 ppm (approximately 40%, dt, JPH 561.9, 36.1 Hz) for the free acid of phosphinopyruvate and its lactone. [On addition of aqueous NaOH to pH 8.5, the 31P-('H-coupled)-NMR spectrum gave a peak at 17.6 ppm (dt, JPH550.0, 18.8 Hz)]. The residue was dried by the addition and removal of dry dioxane (2 x 5 mlj under reduced pressure. Dichloromethane (10 ml) and bis(trimethylsily1)acetamide (1.07 g, 5.3 mmol, 3 equivalents) were added and the solution was stirred under argon. The 31P-(1Hdecoup1ed)-NMR spectrum gave a major peak at 143.1 ppm for bis(trimethylsilyloxy)-[2-(trimethylsilyloxy)-2-(trimethylsilyloxycarbonylj-ethenyllphosphine.After 2 h, methyl chloroformate (0.33 g, 3.5 mmol, 2 equivalents) was added, and the solution was stirred for an additional 1.5 h. The 31P('H-coupled)-NMR spectrum gave 3.77 ppm (approximately 90%, d, Jprr11.7 Hz) for bis(trimethylsilyloxy)-[2-(trimethylsilyloxy)-[2-(trimethylsilyloxy) - 2 - (trimethylsilyloxycarbony1)ethenyllphosphine-carboxylic acid methyl ester. The volatile components were removed under high vacuum. A solution of NaOH (3 equivalents) in water (10 ml) was added to the residue with stirring. The solution was washed with dichloromethane (2 x 5 ml). The aqueous layer was diluted to 1 1 and loaded onto a column (20 mlj of AGl-X8 (HCOOform). The column was eluted with a linear gradient of aqueous NaCl (0-300 mM, 500 ml 500 ml) and fractions (15 ml) were collected. The fractions were assayed at 255 nm to detect the presence of the pyruvoyl group (Anderson et al., 1984) and selected fractions were examined by 31PNMR spectroscopy. Trisodium carboxyphosphinopyruvate (Scheme 3, compound 5) eluted at approximately 250 rnM NaCl and the fractions were concentrated to 5 ml under reduccd pressure. The assay with semicarbazide (Anderson et al., 1984) showed the yield of compound 5 (Scheme 3), to be 0.7 mmol (40%); fiP (121.5 MHz, H 2 0 j 17.9 ppm (s, 'H decoupled; t, JPH16.25 Hz, 'H coupled); aH (400 MHz, H 2 0 with d,-methanol inner lock, referenced to HzO at 4.85 ppm), 3.41 ppm (2H, d, JPH16.7 Hz); dC(100.6 MHz, H 2 0 with d4methanol inner lock, referenced to d4-methanol at 49.0 ppmj, 200.0 ppm [d, Jpc 4.2 Hz, CH,-C(O)], 179.7 ppm (d, Jpc 179.9 Hz, PC02-), 169.5ppm [s, -C(0)C02-], 42.5 ppm (d, Jpc 70.5 Hz, P-CH,); miz (negative ion FAB) 239 Da (20%, M-Na'). The observed exact mass was 238.9339Da; C4H207PNa2requires 238.9333 Da.

+

Cloning and overcxpression of carhoxyphosphonoenolpyruvatemutase Syntlzetic oligonucleotides Oligonucleotides were synthesised on a Milligen/Biosearch 7500 DNA synthesizer (Millipore) and purified using

oligonucleotide purification cartridges (Applied Biosystems). The oligonucleotides synthesized were (5' to 3'): I, GCGCCGGCTCCGACACAGCACACCGAGGAGAGAACAG ; 2, TGACCAAGGCACGTACGTT: 3, GAGGACTTCACGA-TCATC; 4, CATGGTCGAGGGAGGCAAGAC; 5, GAAGAGTTCGGCGAAGGACATC; 6, CATCGATCTCGTCGCGTAC; 7 , TTGGCGCTCAGAGCGTCGTA; 8, AACAGCTATGACCATG; 9, GTAAAACGACGGCCAGT; 10, CGCGTGGATCCCATATGGCCGTGACCAAGGCACGTACGTTC; 11, CTTGGCGTTGGTGGCCTGCTCCGAGAC . DNA manipulations

Standard DNA manipulations were performed according to methods described in Sambrook et al. (1989). Cloning and subcloning were performed in the plasmid pBS( +) (Stratagene). The plasmid p E T l l a (Novagenj was used for expression. Strains carrying these plasmids were maintained on plates or in liquid culture in the presence of ampicillin (100 pg/ml). Restriction endonucleases, T4 DNA ligase and calf intestinal alkaline phosphatase were obtained from New England Biolabs, Gibco BRL, Stratagene or Boehringcr Mannheim. Restriction-endonuelease digestions were performed in the buffers recommended by the manufacturers. Preparation of a size-selected subgenomic library Genomic DNA was isolated from S . hygroscopicus ATCC 21705 by the method of Hopwood et al. (1985). Genomic DNA (16 pl, 3.8 pg) was treated with the restriction endonuclease Burnt11 (20 U j at 37'C for 12 h. The mixture was separated on a Seaplaque (American Bioanalytical) lowmelt agarose gel ( l % j . A gel fragment of 1.0-3.5 kb was excised and the DNA extracted. DNA concentrations were determined by comparisons with standard samples using ethidium bromide visualization. BamHI-digested and dephosphorylated pBS( +) (38 ng) and genomic DNA inserts (75 ng) were ligated together and epicurean E. coli strain AGl was transformed directly with the ligation mixture to yield a library of approximately 10000 transformants. Library screening by colony hyhridisntion Transformation mixtures were plated onto 150-mm agar plates at a colony density of 2500/plate. Nitrocellulose HA filters (Millipore) were prepared and probed as described by Woods (1 984). Oligonucleotide 1 (256 ng, end-labeled with 5000 Ci/mmol [32P]ATP, purchased from Amersham) was used as a probe. Hybridisation temperatures of 42 "C and 44 "C and filter-washing temperatures of 50 - 60°C were employed. D N A sequencing

DNA sequencing was performed using the dideoxy chain termination method (Sanger el al., 1977) using plasmid DNA prepared by the boiling method (Del Sal et al., 1989). Sequencing reactions were carried out following the Seyuenase 2 protocol using modified T7 DNA polymerase (US Biochemical). The protocol was modified by carrying out the extension step at 37°C (instead of 25 "C) and by adding singlestranded binding protein (1.0 p1 of a 0.5 mg/ml solution, US Biochemical) to the reaction mixture during the extension

739 step. After addition of the 'Stop' solution, proteinase K (0.1 pg, US Biochemical)was added and the reaction mixtures were heated at 65°C for 20 min. The samples were heated at 80' C for 2 min immediately prior to loading the gel. Subcloning the CPEP mutase gene into the expression vector pETlla Polymerase chain reaction

To a 0.6-ml microcentrifuge tube was added water (73 pl), 10 x Taq polymerase buffer (10 pl, Promega, adjusted to 25 mM MgCl,), 2 mM dNTP mixture (10 pl), 50 pM primer 10 (1 pl), 50 pM primer 11 (1 pl) and supercoiled pBS-BAM3 template DNA (1 pl, 50 ng). After mixing, the sample was overlaid with light mineral oil (50 pl). Automated reactions wcre carried out in a programmable thermal cycler (PTC-100, MJ Research). 'The mixture was heated at 95°C for 5 min and cooled to 55°C. Tuy polymerase (2.5 U, Promega) was added. After 2 min at 55 "C and 2 inin at 72 "C, 25 cycles were carried out as follows: 94'C for 1 min, 55°C for 1 min, 72°C for 2 min. After the last cycle, the sample was heated at 72'C for 10 min. Assembling the CPEP mutuse gene

The polymerase-chain-reaction-amplified fragments were washed with water in a microconcentrator (Millipore) and digested with N d d and Sphl restriction endonucleases. The reaction products were fractionated on a low-melt agarose gel consisting of a mixture of 3% Nusieve and 1% Seaplaque (American Bioanalyticdl). The appropriate band (150 bp) was excised and the DNA isolated. The pBS plasmid containing the 3.0-kb BamH1 insert (pBS-RAM3) was digestcd with Splzl and BumHl and the appropriate fragment (0.88 kb) was isolated from a Seaplaque gel (l0/o). Likewise, pBS(+) was digcsted with NdeI and RamHI and the 2.4-kb fragment was isolated. The A7&l- SphI fragment (50 ng), the SphI - BamHl fragment (200 ng) and the NdeI - BamHI fragment (120 ng) wcre ligated together in a single reaction. Double-stranded plasmid DNA (pBS-CPMI) was isolated from B. coli transformants and sequenced. Subcloning into p E T l l a

The plasmids pBS-CPMI and PET1 l a were each digested with NdeI and Barn111 restriction endonucleases and the appropriate fragments were excised and extracted from a lowmelt gel (1%). The PET1 l a fragment was dephosphorylated and a ligation mixture of the two fragments was used to transform E. coli AG1. Growth qf cell cultures.for protein expression

Competent E. coli BL21(DE3) cells were transformed with pETl la-CPM1 and plated on plates containing ampicillin (100 pg/ml). Single bacterial colonies were used to inoculate starter cultures (10 ml) and grown at 37'C for 6 h. Four 10ml cultures wcre used to inoculate larger cultures (4 x 1 1, in 4 1 baffled flasks). Cells were grown at 37°C with shaking (275 rpm) to an Asoo of 1.O. Isopropylthio-8-D-galactoside (Gibco BRL) was added to a final conccntration of 0.4 mM and the cells wcre grown for an additional 2.5 h. Thc cclls were centrifuged at 5000 g for 10 min and the supernatant decanted.

Protein determination

Protein concentrations were determined by the method of Bradford (2976) using bovine serum albumin as a calibration standard. Protein samples were analysed by SDS/PAGE according to the method of Laemmli (1970). Enzyme assay with malate dehydrogenase

(Hydroxyphosphiny1)pyruvate was shown to be a substrate for malate dehydrogenase, with a K, of 0.68 mM and a k,,, of 164 s-l (Freeman et al., 1992). The product (hydroxyphosphiny1)lactate had [ x ] ' ~of~ - 11.2' (c 4.45, 0.1 M HCl), the sign of which is the same as the compound isolated from S. hygroscopicus {[xIzoD -6.5" (c l , O . l M HCI)} (Seto ct al., 1983). CPEP mutase activity was determined using a coupled assay with malate dehydrogenase. CPEP mutase was added to a solution containing 50 mM Mes, pH 6.5 (0.48 ml), 0.1 M MnC12 (10 pl), NADH (5 pl of a 10 mg/ml solution), malate dehydrogenase (bovine heart, Sigma, 5 p1 containing 50 U, 1 U being the conversion of 1 pmol oxaloacetate and NADH to malate and NAD/min at pH 7.5 and 25 "C, 14 pg protein) and CPEP (50 pl, 3.8 mM final concentration) at 30°C. The consumption of NADH was monitored at 340 nm. Kinetic parameters were determined by the method of initial rates. Solutions containing (hydroxyphosphiny1)pyruvate were assayed as above (in the absence of CPEP, CPEP mutase, and MnCI2), initiating the reaction with the addition of malate dehydrogenase. Spectrophotometric measurcments were made on a Uvikon 860 (Kontron Instrumcnts), a Hewlett Packard model 4582A or a Perkin Elmer 554 UViVis spectrophotometer. (Hydroxyphosphiny1)pyruvate formation durmg the CPEP mutase reaction could also be monitored directly by 31P-NMR. MgC12 (2 mM) rather than MnCl, was present in the mixture and malate dehydrogenase and NADH were omitted. NMR experiments were carried out on a Bruker WM300 spectroincter. Purification of CPEP mutase

CPEP mutase was purified by a modification of the procedure of Hidaka et al. (1990). All purification steps were carried out at 4°C. Cell paste (13 g from a 4-1 culture) was suspended in 50mM Mes, pH6.5, (30ml) containing dithiothrcitol (1 mM) and pheiiylmethylsulphonyl fluoride (1 mM). Cells were passed twice through an Aminco French press at medium pressure (1300 psi). The cell debris was removed by centrifugation at 200000 g for 90 min. The supernatant was stirred, whilc solid ammonium sulphatc was addcd over 30 niin to 50% saturation. The mixture was then centrifuged at 16000 g for 20 min. To the supernatant was added ammonium sulphate to 80% saturation, and the mixture was then centrifuged as before. The pellet was dissolved in 10 mM potassium phosphate, pH 6.5 (15 ml), and loaded onto a column (100 ml) of hydroxyapatite (Bio Gel HTP, Bio-Rad Laboratories), equilibrated with the same buffer. The column was washcd with the 10 mM loading buffer (500 ml), and thcn eluted with a linear gradient of 10 - 200 mM potassium phosphate, pH 6.5 (1 1 + 11). Fractions containing CPEP mutase activity were pooled and concentrated by ultrafiltration with Diaflo PMlO membranes (Amicon Division, W. R. Grace). The buffer was changed to 50 mM Tris/HCl, pH 7.5, and the sample was loaded onto a Mono Q HR lOjl0 FPLC column equilibrated with the same buffer. The protein was

740 eluted with a linear gradient (30 min, 2 ml/min) of 0 - 0.8 M NaCl in 50 mM Tris/HCl, pH 7.5. Fractions containing CPEP mutase activity were pooled and concentrated by ultrafiltration. The buffer was changed to 50 mM Mes, pH 6.5, for storagc and assay.

RESULTS AND DISCUSSION Cloning and sequencing of the carboxyphosphonoenoZpyruvate mutase gene from Streptomyces hygroscopicus

T7 promoter. The vector is used in conjunction with E. coli strain BL21(DE3), which carries a chromosomal copy of the gene of T7-RNA polymerase. The T7 polymerase gene and the CPEP mutase gene are both under lac control, suppressing basal expression until induction, after which the CPEP mutase gene is expressed to very high levels. In order to subclone the gene into the pETIla vector, polymerase-chain-reaction amplification was used to introduce a NdeI site at the 5' end of the gene. Restriction digestion of the amplified fragment afforded an 150-bp iVde1- SphI fragment. The remainder of the gene, a Sph - BamHI fragment (0.88 kb), was derived directly from pBS-BAM3 by restriction digestion. The 3' BamHI site, located after the TAG stop codon in the CPEP mutase open reading frame (Fig. 2), provides the 3' restriction site for subcloning into the expression vector. The two gene fragments (150 bp and 880 bp) were then ligated together with an appropriately digested pBS fragment in a single reaction. The gene was then subcloned into the expression vector pETlla. The resulting plasmid, pETl1a-CPEPM, is stable in E. coli strains AGZ and BL21(l)E3). BL21(DE3) cells, transformed with the plasmid PET1 laCPEPM, were induced with isopropylthio-P-D-galactosidc and the level of protein expression was monitored by SDS/ PAGE. CPEP mutase was expressed at about 20% of the total soluble cell protein. To facilitate the purification of the enzyme, we have devised a continuous assay in which (hydroxyphosphinyl)pyruvate, the product of the mutase reaction, acts as a substrate for malate dehydrogenase [with (hydroxyphosphiny1)pyruvate as substrate, K, = 680 pM and k,,, = 164 s-'I. This NADH-dependent reaction is readily monitored spectrophotometrically (Freeman et al., 1992). With the availability of synthetic CPEP, this assay was used to locate and measure enzymatic activity during the purification. The enzyme was purified to homogeneity in a sequence of three steps (Table 1) with an overall yield of 40% from the crude cell extract.

Hidaka and coworkcrs havc identified the gene for CPEP mutase from S. hygroscopicus by examining mutant strains blocked at various steps in the biosynthesis of the herbicide bialaphos (Anzai et al., 1987; Hidaka et al., 1989, 1990). This work resulted in the isolation of CPEP mutase and allowed the correlation of its N-terminal sequence with the sequence of a gene that restored bialaphos production in a mutant blocked at this step. The mutase gene was found on a 3-kb BumHl restriction fragment (Fig. 1). We have used this information to clone the gene from a size-selected sub- genomic library of S. hygroscopicus DNA by colony hybridisation. Seven hybridisation-positive clones were obtained, one of which gave a restriction digest pattern in accord with Anzai et al. (1987). This clone (pBS-BAM3) was used in subsequent manipulations. DNA sequencing was carried out using a modified Sanger dideoxynucleotide chain termination protocol. Oligonucleotide 2, whose sequence was based on the reported DNA sequence of Anzai et al. (1987), was used as a primer in the sequencing of double-stranded pBS-BAM3. The resulting sequencing information was used sequentially to define primers 3 - 7. In addition, subcloncs of the 3-kb BarnHI fragmcnt were constructed (BamHI - EcoRI fragments of 2.2 kb and 0.8 kb; Pig. 1) and sequenced using pBS-specific primers 8 and 9. A partial sequence of the 3-kb BarnHI fragment is shown in Fig. 2. Using the start site that agrees with the N-terminal sequence of the purified protein (Anzai et al., 1987), an open reading frame encoding a protein of 295 amino acids was identificd. The calculated molecular mass for the protein (32700 Da) is in agreement with that found for the purified Synthesis of substrate 3, product 4 and intermediate 5 enzyme (32000 f TOO0 Da; Hidaka et al., 1990). Hidaka and To characterize CPEP mutase, we required a sufficient coworkers have recently reported the sequence of the CPEP mutase gene froln s. ~ygroscapicus strain ~ ~ 1 2 in 9 3the supplp of substrate 3 (Scheme 2) for assays during purification, NBRF Protein Database, Their sequence resembles the se- and for kinetic and mechanistic analysis. Hidaka et al. (1990) had isolated O d y analytical quantities of substrate 3 (Scheme quence reported here with the following exceptions: Thr, ~1~ and Ala residues at positions 68, 70 and 76 of this sequence 2) from fermentation broths, and reported that it was an are substituted by Ile, Leu and Thr respectively, in the Hidaka unstable compound. w e have chemically synthesized substrate sequence. These differences probably result from the fact that 3 (Scheme 2), however, and found it to be stable in aqueous over several days at room temperature. The key step a mutagenized high production strain 01S. ~ ~ y g r ~ s c ~was p ~ c u solution s in our synthesis of substrate 3 (Scheme 3) is a Perkov reaction used in the work of Hidaka et al. (1990). Interestingly, the predicted amino acid sequence of CPEP mutase is similar to (Sekine et al., 1982) between his(trimethylsily1) methoxycarthat of PEP mutase from T . pyr$ormis. The two sequences bonylphosphitc and ethyl bromopyruvate. The rcsultillg are 30.3% identical Over a 180-amino-acjd stretch (Seide] et triester was deprotected under mild alkaline conditions to al., 1992). PEP mutase catalyses a closely related rearrange- afford substrate 3 (Scheme 3). (Hydroxyphosphiny1)pyruvate (product 4 Scheme 2 ) was ment (Scheme I), and is also present in S. hygroscopicus (Hidaka et al., 1989): catalysing an earlier step in bialaphos prepared by the transamination of (hydroxyphosphiny1)alanine with glyoxylic acid-Cu(OAc)2 using the conditions debiosynthesis. scribed by Sparkes et al. (1984) (Scheme 3). The putative intermediate ( 5 , Scheme 2) was prepared Carboxyphosphonoenolpyruvatemutase overproduction from (hydroxyphosphinyljpyruvateby a sequence of silylation and purification with bis(trimethylsilyl)acetamide, acylation with methyl We chose to overexpress the CPEP mutase gene in E. cofi chloroformate, and alkaline hydrolysis (Scheme 3). Carboxyusing the vector pETl1 a. This vcctor allows the regulated phosphinopyruvate is stable for at least two weeks under the expression of the target gene under control of the powcrful assay conditions used to monitor the mutase reaction.

741 SStI gy~gctcatcaaygtqggcgytccgcgccggggcgaccgcctg~ccaaytacaaccaqcttctccgqctcg acgagtccgtggcgg~~cccgt~a~ccyccccty~cttccctgaaccgctggtgcygccccycyccggctcc gacacagcacaccgaqgagagaacag ATG GCC GTG ACC AAG GCA CGT ACG T T C CGC GAG 33 met ala val thr lys ala arg thr phe arg g i u

ccc GAG ATC c ' r c GTC C ~ . K ccc AGT GCC TAC GAC GCT CTG AGC l e u met. a s r i ala pro glu ile leu val val p ~ SCL c ald t y r asp ala leu ser

CTG ATG AAC GCC

87

GCC . A G GTC ATC CAG CAG GCC GGC T T C CCC GCG GTG CAC ATG ACC GGC TCG GGC 1 4 1

a l a lys val ile gln yln

dld

yly phe p r o ala v a l hi:; met thr- gly ser gly

SphI ACG TCA GCC AGC ATG C T C GGC C T C CCC GAC CTG GGC T

ttx ser ala ser met l e g~ l y leu pro asp lec gly p CAG GCC P.CC

ACC AGT GTC 'TCG GAG 1 9 5 thr ser va1 ser glu

A 4 C GCC .SAG APC ATC GTG CTC GCC GTG GAC G T A CCG GTG ATC ATG 2 4 9

gln ala thr asn ala lys asn ile v a l leu ala v a l asp vai pro val i l e met EcoRI G3.C GCC GAC GCC GGT TAC GGC AAC GCC ATG TCG GTG TGG CGG GCC ACC AGG GA?. 3 0 3

asp ala asp ala gly tyr gly asn ala met ser val trp arg ala thr arg glu TTC GAG CGG GTC GGC ATC GTC GGC TAC CAC CTG GAG GAC CAG GTG AAC CCC AAG 3 5 7 gly ile val gly tyr his l e u g l u dsp gln vd1 d:in p r o lys T J a l

phe glu arg

C T GAA GGC AAG CGG CTG ATC TCG ACC GAG GAG A'TG ACC GGC 4 1 1 c y s gly his l e u glu g l y l y s arg leu ile s e r thr glu glu met thr- gly

TGT CGC CAC

ATC GAG G C G

A X

xc

GTC GAG GC:

lys ile y - u dla ala val g:,i

r;.v

CGC ZAG

,=la arg glu 2 . s ~

GX:

TTC' ACG ATC AT? GCG 4 6 5

asp phe t t r i;e

ile ala

ACC GAT GCC CGC GAG TCG T T C GGC CTG GAC GAG GCC ATC CGC CGC TCC CGC 514

ary t h r a s p a l a arg g l u ser phe g l y l e u asp qlu ala ile arg arg ser arg GAG TAC GTG GCG GCC

GCC GAC TGC ATC T T C CTG GAG GCC ATG C T C GAC GTC 5 7 3

glu tyr val ala ala

ala asp cys i i e phe leu g l u ala met lcu asp val

CGG GTP. CGC G K GAG ATC: GA'P GCC CCC CTG C T C GCX AAC ATG 6 2 7 y l i i glu met l y s ary v a l arg dsp glu ile asp ala pro leu leu ala asn met

GAG GAG ATG ?AG

GTC GAG GGA GGC A?lG ACG CCC TGG CTG ACC ACC AAG GAG CTG GAG TCG ATC GGC 681 val glu g l y gly lys thr pro t i p leu i.111 I.hr ly:; g l u l e u ylu s e r i l e yly TAC AAC CTG iCG ATC TAT CCC; r'TG T C C GGC T C G ATC GCC GCC t y r asn 1eG ala ile tyr jro 1 ~ s~e 1 r g l y trp met a;a ala ,CGC M G CTG T T C ACC GAG CTG AGG GAG GCC GGC 4 C C ACC CAG AAG T T C TGG GAC 789

arcj lys leu phe thr qlu leu arq qlu ala g l y thr thr qln l y s phe trp a s p AAG ATG TCC w c rxtr GAA C ' K wc GAG GTC TTC GAG TAC TCC 8 4 3 asp met g l y leu lys met ser phi. dla g l u leu phe g l u v d l phe g l u tyr s e r

GAC ATC G C ~ ' TTG

~ A GATC T C C GAG C T T GAG GCC CGC TTC GTC CGC GAC CAG GAC TGA ccccgggtct ;ys ile ser qlu leu g l u ala a r g phe val srg asp gln asp stop gctcydaacctccccggaaqtccccagaaattccttcgatcccacgcggcacgggagacaccatqttc~tc gacggagcactgcagccgqcccqccaccacatcactqtqtacgacqctqgacgggcgagtccatcg~atcc BamHI

Fig. 2. Genumic sequence of the CPEP mutase gene and the upstream (5') and downstream (3') regions. Coding sequence is in upper-case letters; non-coding sequence is in lowcr-case letters. Table 1. Purification of recombinant CPEP mutase.

Purification stcp

Crude extract Ammonium sulfate Hydroxyapalile FPLC (Mono Q)

Purification parameters Total protein

Total units

Unit yield

Specific activity

Purification factor

mg

U 3.55 2.34 1.98 1.43

%

U/mg x lo3

zoo

1.5 15.0 33.0 31.6

-fold I .o

413 156 60 38

66 56 40

Kinetic and mechanistic studies 'The enzyme catalyzes the conversion of CPEP to (hydroxyphosphiny1)pyruvate with a k,,, of-0.020 s and a K,, of 270 pM (pH 6.5). The reaction shows a metal dependence: the enzyme is about tcn limes more active in the presence of Mn(l1) compared with the activity in the presence of Mg(I1). ~

2.0 4.4 5.0

The chemical stability of compound 5 (Scheme 2) rules out the possibility that the decarboxylation after the presumed enzyme-catalyzed carboxyphospho-group transfer is a spontaneous, non-enzyme-catalysed event. Indeed, we find that CPEP mutase accelerates the decarboxylation of carboxyphosphinopyruvate, as determined both by 31P-NMR and by

742

5

Scheme 3. The syntheses of carboxyphosphonocnoZpyruvate (3), (hydroxyphosphiny1)pyruvatc (4) and carboxypbosphinopyruvate(5). (a) NaOH ( 3 equivalents); (h) glyoxylic acid, C U ( O Z C C H ~. )HzO; ~ (c) Dowex-H+; (d) Dowex-Na' ; (c) bis(trimcthylsily1)acetamide; (4 methyl chloroformate.

the coupled-enzyme assay for (hydroxyphosphiny1)pyruvate. The decarboxylation reaction proceeds with a k,,, of 7.6 x s-' and a K, of 2.2 pM. The fact that CPEP mutase catalyzes the decarboxylation of compound 5 (Scheme 2) supports the notion that the enzyme mechanism involves the rearrangement of CPEP (3, Scheme 2) to 5 (Scheme 2), and that this step is followed by decarboxylation. This final exergonic step shifts the overall equilibrium toward C-P bond formation. Although the apparent k,,, of CPEP mutase with the notional intermediate 5 (Scheme 2) is much lower than the k,,, for the overall conversion of CPEP to (hydroxyphosphinyl)pyruvate, the decarboxylation of 5 (Scheme 2) is enzyme catalyzed, and it has been pointed out that the rate of binding of an exogenously added intermediate to an enzyme can limit the overall rate of the reaction (Cleland, 1990). That is, even if the intermediate is on the normal enzymatic reaction pathway, the species may appear not to be kinetically competent. Based on steady-state thermodynamic relations between free and enzyme-bound substrate and intermediate species, the association constants for the intermediate and substrate are related to the equilibrium constants for the conversion of substrate to intermcdiatc on the enzyme (&) and in solution (Keq)as follows:

Although the values of Keg and Keq for CPEP mutase remain to be determined directly, these relationships are consistent with the low apparent k,,, of the enzyme with 5 (Scheme 2). Combining Eqns (1) and (2) gives

(3) kl = k - 1KbqKa(substrate)lKesK,(intermediate) . Assuming that the observed turnover number of 7.6 x 10P4s-l reflects thc value of k l for the rate-limiting formation of a productive enzyme-intermediate complex, the estimated K& and Keg values above are valid and Ka(intermcdiatej z Ka(substralc), we have k- I x 6 x l o p 6 s- '. In other words, the intermediate would dissociate from the enzyme approximately once every 3000 turnovers (bascd on a value fork,,, of 0.020 s P for compound 3). The enzyme may bind the intermediate not only with a low off-rate to minimize dissociative loss, but also with a low on-rate to avoid a tightly bound species that would sit in a thermodynamic well (Cleland, 1990). The direct determination of Kb, and Keqwill be needed to resolve the question of the kinetic plausibility of carboxyphosphinopyruvate as an intermediate, yet the clear catalysis of its decarboxylation by the enzyme persuades us of the likelihood that the chemically logical route, of rearrangement followed by decarboxylation, is in fact followed by the mutase. We have cloned and overexpressed, in E. coli, the gene for Ka(intermediatc)/Ka(sub6trate) = Kkq/Keq . (1) carboxyphosphonoenolpyruvate mutase from S. hygroscoIn this equation, the term Ka(inlermediate) includes the associ- picus and have chemically synthesized its substrate its, product ation of the intermediate with the enzyme [Xalintzrmediate)] and and a putative reaction intermediate. lnitial mechanistic studits conversion from a collisional complex to its productive ies suggest that the CPEP mutase reaction proceeds via a enzymic form as follows: rearrangement of the carboxyphospho group, followed by decarboxylation. The availability of large quantities of the Kn