Biosynthesis of Pteridines in Escherichia coli

An open reading frame located at 69.0 kilobases on the Escherichia coli chromosome was shown to code for dihydroneopterin aldolase, catalyzing the conversion of 7,8-dihydroneopterin to 6-hydroxymethyl-7,8-dihydropterin in the biosynthetic pathway of tetrahydrofolate. The gene was subsequently designated folB. The FolB protein shows 30% identity to the paralogous dihydroneopterin-triphosphate epimerase, which is specified by the folX gene located at 2427 kilobases on theE. coli chromosome. The folX andfolB gene products were both expressed to high yield in recombinant E. coli strains, and the recombinant proteins were purified to homogeneity. Both enzymes form homo-octamers. Aldolase can use l-threo-dihydroneopterin andd-erythro-dihydroneopterin as substrates for the formation of 6-hydroxymethyldihydropterin, but it can also catalyze the epimerization of carbon 2′ of dihydroneopterin and dihydromonapterin at appreciable velocity. Epimerase catalyzes the epimerization of carbon 2′ in the triphosphates of dihydroneopterin and dihydromonapterin. However, the enzyme can also catalyze the cleavage of the position 6 side chain of several pteridine derivatives at a slow rate. Steady-state kinetic parameters are reported for the various enzyme-catalyzed reactions. We propose that the polarization of the 2′-hydroxy group of the substrate could serve as the initial reaction step for the aldolase as well as for the epimerase activity. A deletion mutant obtained by targeting the folX gene of E. coli has normal growth properties on complete medium as well as on minimal medium. Thus, the physiological role of the E. coli epimerase remains unknown. The open reading frameygiG of Hemophilus influenzae specifies a protein with the catalytic properties of an aldolase. However, the genome of H. influenzae does not specify a dihydroneopterin-triphosphate epimerase.

The pathway of tetrahydrofolate biosynthesis is an important drug target in bacteria and protozoal parasites (1)(2)(3). Thus, the sulfonamides inhibiting dihydropteroate synthase were the first synthetic agents that could be used with high efficiency against a wide variety of microbial and protozoal parasites. Even after more than 60 years, they remain important therapeutic agents (1,4). More recently, trimethoprim was found to act as an inhibitor of bacterial dihydrofolate reductase (5). The combined application of trimethoprim and a sulfonamide is an important protocol for the treatment of bacterial infections (5)(6)(7)(8).
Tetrahydrofolate is biosynthesized by plants and many microorganisms. The first committed step catalyzed by GTP cyclohydrolase I is a mechanistically complex ring expansion reaction conducive to the formation of dihydroneopterin triphosphate (Compound 1) ( Fig. 1) (9). A pyrophosphatase and a phosphatase have been proposed to convert Compound 1 to dihydroneopterin (Compound 3) in two consecutive steps (10), but firm evidence for the involvement of specific enzymes in the conversion of Compound 1 to Compound 3 has not been obtained. Recently, it has been proposed that the release of pyrophosphate from Compound 1 can be catalyzed of two-valent ions such as Mg 2ϩ and Ca 2ϩ (11).
An enzyme catalyzing the epimerization of the folate precursor, dihydroneopterin triphosphate, under formation of 7,8dihydromonapterin (Compound 2) had been observed in Escherichia coli by Heine and Brown (14). Recently, the primary structure of this enzyme has been determined and has been found to show similarity to the dihydroneopterin aldolase domain of a multifunctional folate biosynthetic enzyme from Pneumocystis carinii (15). On the other hand, the epimerase is similar to a reading frame at 69.0 kb 1 of the E. coli chromosome designated ygiG (EMBL accession number L12966) in the close vicinity of the bacitracin resistance gene. This reading frame specifies the dihydroneopterin aldolase of E. coli, which is a potential target for the development of enzyme inhibitors with antibacterial and/or antiprotozoal activity. This paper reports on structural and mechanistic similarities of dihydroneopterin-triphosphate epimerase and dihydroneopterin aldolase. It also shows that the single Hemophilus influenzae homolog of these proteins (specified by the unannotated open reading frame ygiG_haein) is an aldolase-type enzyme.

EXPERIMENTAL PROCEDURES
Materials-6-Hydroxymethylpterin, D-dihydroneopterin, L-dihydromonapterin, D-dihydromonapterin, L-neopterin, L-monapterin, D-dihydroneopterin monophosphate, D-neopterin, and dihydrobiopterin * This work was supported in part by grants from the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie and by Grant ERBCHRXCT930243 from the European Community. 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  . Goldstar DNA polymerase was obtained from Eurogentec (Seraing, Belgium). GTP cyclohydrolase I was purified from a recombinant E. coli strain as described earlier (9).
Microorganisms and Plasmids-The bacterial strains and plasmids used in this study are summarized in Table I.
An aliquot of this reaction mixture (20 l) was added to 80 l of a solution containing 80 mM PIPES, pH 6.2, and 8 mM MgCl 2 . The epimerase reaction was started by the addition of protein solution. The reaction mixture was incubated at 55°C for 10 min. The reaction was terminated by the addition of a solution (30 l) containing 1% I 2 and 2% (w/v) KI in 1 M HCl. The samples were incubated at room temperature for 5 min. Excess iodine was reduced by the addition of 2% ascorbic acid (10 l, w/v). A solution (210 l) containing 1.5 M Tris-HCl, pH 8.5, 4.8 mM ZnCl 2 , 4.5 mM EDTA, and 0.5 units of alkaline phosphatase (Boehringer Mannheim, grade II) was added. The mixture was incubated at 37°C for 90 min. Trichloroacetic acid (50 l, 40%, w/v) was added. Aliquots of 10 l were analyzed by HPLC using a reversed-phase column (4 ϫ 250 mm) of Nucleosil RP18 and an eluent containing 0.5% (w/v) H 3 BO 3 , pH 4.7. The effluent was monitored fluorometrically (excitation, 365 nm; emission, 446 nm). The flow rate was 2 ml/min. The retention volume was 7 ml for D-erythro-neopterin and 9 ml for L-threoneopterin (monapterin).
Assay of Dihydroneopterin Aldolase-The assay is based on a method described earlier (12). Assay mixtures contained 100 mM Tris-HCl, pH 8.0, 5 mM dithiothreitol, 70 M dihydroneopterin, and protein. The mixture was incubated at 70°C for 10 min. The reaction was terminated by the addition of 30 l of 1 M HCl containing 1% I 2 and 2% (w/v) KI. The samples were incubated at room temperature for 5 min. Excess iodine was reduced by the addition of 2% ascorbic acid (10 l, w/v). The samples were analyzed by reversed-phase HPLC using a column (4 ϫ 250 mm) of Nucleosil RP18 and an eluent containing 30 mM HCOOH and 7% MeOH. 6-Hydroxymethylpterin was detected fluorometrically (excitation, 365 nm; emission, 446 nm). The retention volume was 6.5 ml.
Construction of Expression Plasmids for the folX and folB Genes-The expression plasmids for the folX and folB genes were constructed as described by Richter et al. (16). The folX and folB genes of E. coli were amplified by PCR with chromosomal DNA from E. coli XL1-Blue as template. The folB gene of H. influenzae was amplified by PCR using the plasmid GHIFL0I as template ( Table I). The folX gene was amplified with the oligonucleotides EpiExsense as forward primer and EpiExanti as reverse primer (Table II). The folB gene of E. coli was amplified with ECAL1 as forward primer and ECAL2 as reverse primer (Table II), and the folB gene of H. influenzae was amplified with HAEALD1 as forward primer and HAEALD2 as reverse primer (Table  II). Each amplification product served as template for a second PCR with the universal forward primer kEcoRI (Tables II and Table III). The reverse primer EpiExanti was used for the folX gene. The reverse primer ECAL2 was used for the folB gene of E. coli, and the reverse primer HAEALD2 for the folB gene of H. influenzae. The PCR products (folB of H. influenzae and folX of E. coli) were cleaved with EcoRI and BamHI and ligated in the expression vector pNCO113, which had been digested with EcoRI and BamHI, whereas the folB gene of E. coli was cleaved with PstI and EcoRI and then ligated into the expression vector pNCO113, which had been digested with PstI and EcoRI. The expression plasmids designated pEPI (containing the folX gene of E. coli), pEAL (containing the folB gene of E. coli), and pHIAL containing the folB gene of H. influenzae) ( Table I) were transformed into E. coli M15(pREP4). Kanamycin (20 g/ml) and ampicillin (180 g/ml) were added for the maintenance of the plasmids in the host strain.
Construction of the Vector Used for the Deletion of the folX Gene-The folX gene and the adjacent regions were amplified by PCR with EpiKpn as forward primer and EpiPst as reverse primer (Table II) using chromosomal E. coli XL1 DNA as template. The amplification product was cleaved with KpnI and PstI and then ligated into the vector pMAK705, which had previously been digested with KpnI and PstI. The vector contains the temperature-sensitive pSC101 origin of replication and a chloramphenicol resistance gene. The resulting plasmid, pMAK705E, was digested with EcoRV. The folX gene contains two internal EcoRV sites, and digestion results in the excision of a DNA fragment of 105 bp. A kanamycin resistance cassette was amplified by PCR using the forward primer kan-Rbs-1 and the reverse primer Kana4 (Table II) with the plasmid pREP4 (Table I) as template. The EcoRV-digested plasmid pMAK705E was T-tailed with Taq polymerase, and the product was ligated with the kanamycin resistance cassette, yielding the plasmid pMAK705EK (Table I and Fig. 2).
Replacement of the Chromosomal folX Gene with Its Inactivated Variant-The replacement of the chromosomal folX gene is a modification of the methods described by Hamilton et al. (17), Russel and Model (18), and Fermér and Swedberg (19). The plasmid pMAK705EK was transformed into the recA ϩ E. coli strain SK6600. The plasmid replicates at 30°C, but not at 44°C. Growth at 44°C on plates that contain chloramphenicol selects clones with the plasmid integrated into the chromosome by homologous recombination. This results in a tandem arrangement of the intact folX gene adjacent to the inactivated folX  (Table II), complementary to the E. coli chromosome in the direct vicinity of the 3Ј-end of the folX insert, should result in the amplification of a 1.4-kb fragment.
One of 12 clones screened with this method showed the correct pattern of amplificates. The resolution of this cointegrate was achieved by overnight incubation of the bacterial strain in LB liquid medium containing 20 mg/liter chloramphenicol at 30°C. Cells were then diluted 10 5 -fold and grown on plates containing chloramphenicol at 30°C. Plasmid isolation of the resulting colonies showed that ϳ50% of the colonies now contained an intact folX gene on the plasmid and therefore the defective folX gene on the chromosome. Curing of the clones was achieved by growing them at 44°C on LB plates. The resulting clones were sensitive to chloramphenicol. Epimerase activity could not be   detected in cell extracts of the mutants. PCR amplification using the forward primer Episense and the reverse primer Epianti (Table II) resulted in amplificates of ϳ1400 bp, whereas the intact folX gene yields amplificates of ϳ300 bp. For the comparison of growth rates, SK6600 cells with intact or defective folX genes were grown on LB medium as well as on M9 minimal medium supplemented with vitaminfree casamino acids (Gibco BRL).
Purification of Recombinant Dihydroneopterin Aldolase-E. coli strain M15(pREP4) carrying the expression plasmid pEAL was grown to a density of 0.6 A 600 nm units in shaking flasks with LB medium containing 20 g/ml kanamycin and 180 g/ml ampicillin at 37°C. Isopropyl-␤-D-thiogalactopyranoside was added to a concentration of 2.0 mM, and incubation was continued for a period of 6 h. Cells were harvested by centrifugation; washed with 50 mM Tris-HCl, pH 8.0; and stored at Ϫ20°C.
Frozen cell mass (8.3 g) was suspended in 50 ml of 50 mM Tris-HCl, pH 8.0, containing 0.2 mg/ml DNase and 0.02 mg/ml RNase. Bacteria were lysed by ultrasonic treatment, and cell debris was removed by centrifugation. The supernatant was heated to 95°C in a water bath for 3.5 min and was then cooled in ice water. The precipitate was removed by centrifugation. The supernatant was loaded on a Sepharose Q column (2 ϫ 11 cm; Pharmacia) that had been equilibrated with 20 mM Tris-HCl, pH 8.0. The column was developed with a linear gradient of 0 -1 M NaCl. Fractions were combined and concentrated by ultrafiltration (YM-30 membrane). Concentrated fractions (8 ml) were further purified by gel filtration on a Superdex 200 column (2.6 ϫ 60 cm; Pharmacia). The column was developed with 20 mM Tris-HCl, pH 8.0, containing 100 mM NaCl at a flow rate of 3 ml/min. Dihydroneopterin aldolase from H. influenzae could be purified using the same procedure.
Purification of Recombinant Epimerase-E. coli strain M15(pREP4) harboring the expression plasmid pEpi was grown in LB medium containing 150 mg/liter ampicillin and 20 mg/liter kanamycin. The culture was incubated at 37°C with shaking. At a cell density of 0.6 A 600 nm units, isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 2 mM, and incubation was continued overnight. The cells were harvested by centrifugation and stored at Ϫ20°C.
Frozen cell mass (8.5 g) was thawed and resuspended in 80 ml of 50 mM Tris-HCl, pH 8.0, containing 0.2 mg/ml DNase and 0.02 mg/ml RNase. Cells were disrupted by ultrasonic treatment. The suspension was centrifuged, and the supernatant was loaded on a Sepharose Q column (2.5 ϫ 12 cm). The column was developed with a linear gradient of 0 -0.4 M NaCl containing 20 mM Tris-HCl, pH 8.0. Fractions were combined, concentrated by ultrafiltration (YM-30 membrane), and subjected to heat treatment at 80°C for 4 min. The precipitate was removed by centrifugation. The enzyme was further purified by gel filtration on a Superdex 200 column (2.6 ϫ 60 cm). The column was developed with 20 mM Tris-HCl, pH 8.0, containing 100 mM NaCl. The flow rate was 3 ml/min.
Determination of Kinetic Parameters-Steady-state kinetic experiments were performed at 37°C. K m and V max were determined by least-square fit according to the Michaelis-Menten equation using the programs Origin (Microcal) and Excel (Microsoft). The equilibrium constant for epimerization of dihydroneopterin triphosphate and dihydromonapterin triphosphate catalyzed by dihydroneopterin-triphosphate epimerase was estimated from the composition of reaction mixtures that had reached a state of apparent equilibrium.
Analytical Ultracentrifugation-Experiments were performed with an analytical ultracentrifuge (Optima XL-I, Beckman Instruments) equipped with absorbance and interference optics. Experiments were performed in double sector cells with aluminum centerpieces and sapphire windows. Fluorochemical FC43 (10 l; Beckman Instruments) was added to the sample compartment for sedimentation equilibrium experiments. The partial specific volume was estimated on the basis of amino acid compositions (20). Protein (ϳ2 mg/ml) dissolved in 100 mM sodium phosphate, pH 7.0 ( ϭ 1.0228 g cm Ϫ3 ), was centrifuged at 4°C and 7000 rpm for 72 h. Protein concentration was monitored by laser interferometry.

Recombinant Expression of Dihydroneopterin Aldolases-We
have recently reported the sequence of dihydroneopterintriphosphate epimerase from E. coli (15). A data base search indicated that an open reading frame in the neighborhood of the bacA gene located at 69.0 kb on the E. coli chromosome specifies a similar protein (30% identical amino acid residues). This open reading frame, which was subsequently designated folB, also shows similarity to putative dihydroneopterin aldolases from a variety of microorganisms (Fig. 3) (2,(21)(22)(23)(24)(25)(26)28). 2 The folB gene of E. coli was amplified by PCR with chromosomal DNA and placed under the control of the lac operator and a T5 promoter in the plasmid pNCO113. A recombinant E. coli strain carrying this plasmid expressed a 12.5-kDa peptide as shown by SDS-polyacrylamide gel electrophoresis. Cell extracts of the recombinant strain had a dihydroneopterin aldolase activity of 180 mol/min/mg as compared with a value of 41 nmol/min/mg in the E. coli wild-type strain DSM613. This finding provided strong support for the hypothesis that the folB gene codes for dihydroneopterin aldolase.
The recombinant protein was purified by chromatography as described under "Experimental Procedures." A typical experiment is summarized in Table IV. The purified protein migrated as a single band on SDS-polyacrylamide gels. The purification factor of 2.9 indicates that the protein had been expressed to a level of ϳ35% of cellular protein in the recombinant E. coli strain.
The H. influenzae genome contains only one open reading frame (ygiG), located at 298.5 kb on the chromosome, that is similar to the folB gene of E. coli (56% identical amino acid residues). The H. influenzae gene was expressed in a recombinant E. coli strain and purified in the same way as the E. coli aldolase. As expected on the basis of the sequence similarity, the recombinant protein catalyzed the retroaldol cleavage of dihydroneopterin. The kinetic parameters were similar to those of the E. coli aldolase.
Recombinant Expression of Epimerase-The folX gene of E. coli specifying dihydroneopterin-triphosphate epimerase was expressed under the control of the lac promoter and T5 operator. The recombinant protein was purified by a sequence of ion-exchange chromatography, heat treatment, and gel permeation chromatography ( Table V). The purified protein migrated as a single band of 14 kDa on SDS-polyacrylamide gels.
Quaternary Structure-The molecular masses of the proteins under study were determined by sedimentation equilibrium analysis. The molecular masses are summarized in Table VI and suggest that the three proteins under study form homo-octamers.
Kinetic Properties of Aldolases-In light of the sequence similarity between epimerase and aldolase, it was in order to perform a detailed comparison of their catalytic and kinetic properties. Initial enzyme experiments were performed at the temperature optimum of the enzymes studied (70°C for dihydroneopterin aldolase and 55°C for dihydroneopterin-triphosphate epimerase) (Tables IV and V). The more detailed kinetic analyses reported in Table VII were performed at 37°C.
Products were determined by fluorescence-monitored HPLC after dehydrogenation and dephosphorylation as required. The results shown in Table VII and Fig. 4 indicate that the aldolases specified by the folB genes of E. coli and H. influenzae catalyze the cleavage of dihydroneopterin. The velocity was 127 mol/mg/h in comparison with 234 nmol/h as determined by Mathis and Brown (12). The remarkable difference can be explained by the pH at which Mathis and Brown determined their kinetic properties. Their activities were measured at pH 9.6, but we have found a dramatic decrease in activity at pH Ͼ8.5.
However, the enzyme could also catalyze the epimerization of position 3Ј of dihydroneopterin, yielding dihydromonapterin. The rates of the forward and reverse reactions were similar. Moreover, we found that the epimer dihydromonapterin could also serve as a substrate for aldol cleavage of the side chain, yielding hydroxymethyldihydropterin. All cleavage and epimerization reactions catalyzed by the enzyme have similar velocities. The kinetic properties of the aldolase from H. influenzae are shown in parentheses in Fig. 4.
The naturally occurring substrate of the aldolases in the biosynthetic pathway of tetrahydrofolate is dihydroneopterin. However, the epimer dihydromonapterin can be converted to 6-hydroxymethyldihydropterin without prior epimerization to dihydroneopterin as the velocity of the cleavage reaction for both epimers is at least six times higher than the epimerization reaction of both substrates (Table VII). It should be noted that the folB gene of E. coli catalyzes the epimerization reaction more efficiently then the H. influenzae protein. Table VII indicates that in H. influenzae, the velocity of epimerization, as well as the velocity of the formation of hydroxymethylpterin from dihydromonapterin, is considerably lower as compared with the E. coli enzyme.
The monophosphate of dihydroneopterin can be cleaved by the aldolases of E. coli and H. influenzae, but the velocity is Ͻ1% as compared with the unphosphorylated substrates. The triphosphates of dihydroneopterin and dihydromonapterin were neither cleaved nor isomerized by the aldolases under study.
Kinetic Properties of Epimerase-As shown in Table VII and Fig. 4, dihydroneopterin-triphosphate epimerase uses the triphosphates of dihydroneopterin and dihydromonapterin efficiently as substrates for epimerization. A value of 0.77 was estimated for the equilibrium constant on the basis of the steady-state kinetic parameters. Dihydroneopterin and dihydromonapterin can also serve as substrates for epimerization, but the reaction velocities of the unphosphorylated substrates are slower by more than a factor 700. However, the unphosphorylated substrates are also slowly converted to hydroxymethyldihydropterin. The velocity of the aldolase-type reaction catalyzed by epimerase is Ͻ1% as compared with the cleavage of dihydroneopterin by the FolB protein. The cleavage reactions cannot be attributed to a contamination of recombinant dihydroneopterin aldolase with dihydroneopterin-triphosphate epimerase specified by the chromosomal folX gene since aldolase and epimerase are efficiently separated by the anion-exchange   chromatography step (epimerase elutes at a sodium chloride concentration of ϳ150 mM, whereas aldolase elutes at ϳ600 mM NaCl). The physiological role of the FolB protein in the pathway of tetrahydrofolate biosynthesis is well understood. However, the role of the epimerase occurring in E. coli, but not in H. influenzae, is not clear. We therefore constructed a folX deletion mutant by gene targeting in E. coli using the method described by Kushner and co-workers (17). The construction of the required plasmids and the gene deletion are described under "Experimental Procedures." PCR analysis of chromosomal DNA confirmed that the chromosomal gene of the mutant carries an insertion of 807 bp after bp 149 of the folX gene. Cell extracts of the mutant are devoid of epimerase activity. Since the enzyme assay used is very sensitive, any residual activity below the detection limit would have been Ͻ0.5% as compared with the E. coli wild-type strain. The folX mutant showed normal growth properties on minimal medium as well as on complete medium. Thus, the biological role of epimerase remains unknown. DISCUSSION The enzymes specified by the folX and folB genes of E. coli are similar in terms of their primary structure and quaternary structure. The sequence similarity extends to dihydroneopterin aldolases of several other microorganisms (Fig. 3). Some of these proteins are large multidomain proteins catalyzing several steps in the biosynthesis of tetrahydrofolate.
It appears safe to assume that the folB gene is involved in the pathway of tetrahydrofolate biosynthesis, where it is responsible for the shortening of the position 6 side chain of the pteridine precursor. On the other hand, the role of the dihydroneopterin-triphosphate epimerase is still unknown. Growth tests of E. coli with a defective folX gene showed no apparent phenotype.
Pteridines with the threo configuration of the trihydroxypropyl side chain that is generated by the epimerization reaction have been found in a variety of organisms (29,30), but their metabolic role is unknown. Even if these stereoisomers have a specific metabolic function, a specific epimerase may not be required for their formation since, at least in E. coli and H. influenzae, they can be formed by the aldolase specified by the folB gene.
H. influenzae has a folB gene (i.e. the unannotated reading frame ygiG_haein) specifying a protein whose catalytic properties are very similar to those of the FolB protein of E. coli. However, the genomes of H. influenzae and Bacillus subtilis appear to be devoid of folX genes specifying a pteridine epimerase, in line with the hypothesis that the epimerase is not an essential protein in bacteria.
A hypothetical reaction mechanism is shown in Fig. 5. We propose tentatively that both reaction types can be initiated by protonation of N-5 followed by deprotonation at the acidic C-1Ј of dihydroneopterin-or dihydromonapterin-type substrates. Epimerase-as well as aldolase-type reactions can be catalyzed by both the FolB and FolX proteins. This raises the question of a common transition state for both types of reaction. A retroaldol cleavage of the C-C bond between C-1Ј and C-2Ј is proposed to be the crucial reaction step for both enzyme reactions. Epimerization at C-2 would result from reversal of the cleavage reaction without stereochemical control. The different product ratios of FolB-and FolX-type proteins could result from differences in compartmentalization between the different exit pathways originating from the common transition state.
Epimerization could occur if the reaction products resulting from cleavage of the ␣,␤-carbon bond, instead of being released from the enzyme, react under regeneration of a pteridine, albeit without stereo control. Partitioning between the reaction pathways for cleavage and isomerization could then depend on the compartmentalization of the cleavage fragments at the active site. This would imply that in the case of epimerase, the release of the cleavage products from the enzyme could be hindered by the protein conformation, and release would occur only after religation of the cleavage fragments.