Biosynthesis of Tetrahydrofolate

7,8-Dihydroneopterin aldolase catalyzes the formation of the tetrahydrofolate precursor, 6-hydroxymethyl-7,8-dihydropterin, and is a potential target for antimicrobial and anti-parasite chemotherapy. The last step of the enzyme-catalyzed reaction is believed to involve the protonation of an enol type intermediate. In order to study the stereochemical course of that reaction step, [1′,2′,3′,6,7-13C5]dihydroneopterin was treated with aldolase in deuterated buffer. The resulting, partially deuterated [6α,6,7-13C3]6-hydroxymethyl-7,8-dihydropterin was converted to partially deuterated 6-(R)-[6,7,9,11-13C4]5,10-methylenetetrahydropteroate by a sequence of three enzyme-catalyzed reactions followed by treatment with [13C]formaldehyde. The product was analyzed by multinuclear NMR spectroscopy. The data show that the carbinol group of enzymatically formed 6-hydroxymethyl-dihydropterin contained2H predominantly in the pro-Sposition.

Tetrahydrofolate and its derivatives are essential cofactors of one-carbon metabolism. Although plants and many microorganisms obtain folate coenzymes by de novo synthesis, vertebrates are absolutely dependent on nutritional sources (1). Insufficient supply of the vitamin is conducive to anemia in adults and to neural tube malformation in human embryos (2).
The folate biosynthetic pathway is a well established drug target for antimicrobial as well as antiparasite therapy. Sulfonamides, the first synthetic antimicrobial and antiparasitic drugs with a broad action spectrum, act via inhibition of dihydropteroate synthase, the penultimate enzyme of the dihydrofolate biosynthetic pathway (Fig. 1, step E) (5)(6)(7), and trimethoprim acts against a variety of bacterial pathogens via inhibition of dihydrofolate reductase (Fig. 1, step G) (8 -10).
The rapid development of microbial resistance against all antibiotics in current use has generated an urgent requirement for novel anti-infective agents. Because the folate pathway is already a well established drug target, it appears worthwhile to explore other folate biosynthetic enzymes besides dihydropteroate synthase and dihydrofolate reductase. This paper describes studies on the mechanism of dihydroneopterin aldolase. [1Ј,2Ј,3Ј,4Ј,5Ј-13 C 5 ]GTP was prepared as described (11). All other reagents used were of the highest purity available. A Nucleosil C18 HPLC 1 column (4 ϫ 250 mm) was from Schambeck, Bad Honnef, Germany. Superdex 75, Superdex 200, Q-Sepharose Fast Flow, and DEAE-Sepharose Fast Flow were purchased from Amersham Biosciences. Nanosep 30K Omega Ultrafilter and Omega membrane 10K were purchased from Pall-Gelman (Dreieich, Germany).
Enzymes-Alkaline phosphatase was from Roche Molecular Biochemicals. Recombinant GTP cyclohydrolase I and dihydroneopterin aldolase of Escherichia coli were prepared as described (11,12).
Construction of Expression Plasmids-The folK gene and folP genes of Haemophilus influenzae were amplified by PCR using the plasmids GHIHC06 and GHIDF84 (Table I) as templates. The oligonucleotides HKsense and HKanti (Table II) served as primers for amplification of folK, and the oligonucleotides HPsense and HPanti served as primers for amplification of folP. The resulting DNA fragments served again as templates for PCR with the primer kEcoRI for both reactions and the oligonucleotides HKanti and HPanti for amplification of folK and folP, respectively. After restriction with EcoRI and BamHI, the amplificates were ligated into the vector pNCO113 affording the plasmids pHPPK and pDHPS comprising folK and folP gene, respectively. The expression plasmids were transformed into E. coli M15[pRep4]. Kanamycin (20 mg/liter) and ampicillin (50 mg/liter) were added for the maintenance of the plasmids in the host strain.
The folA gene of E. coli was amplified by PCR using chromosomal DNA as template and the oligonucleotides EC-FolA-NdeI-sense and EC-FolA-BamHI-anti (Table II) as primers. The amplificate was cleaved with NdeI and BamHI and was then ligated into the expression vector pET-5a that had been digested with the same enzymes. The resulting expression plasmid designated pDHFR was transformed into E. coli BL21(DE3) [pLysS]. Chloramphenicol (20 mg/liter) and ampicillin (50 mg/liter) were added for the maintenance of the plasmids in the host strains. * This work was supported by the Deutsche Forschungsgemeinschaft, the Fonds der Chemischen Industrie, the Hans-Fischer-Gesellschaft, and by EC Grant FMRX-CT980204. 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.
‡ directing the expression of the folK gene of H. influenzae was grown aerobically at 37°C in LB medium containing kanamycin (20 mg/liter) and ampicillin (50 mg/liter) to an optical density of 0.7 (600 nm). IPTG was added to a concentration of 1 mM, and incubation was continued for a period of 4 h. Cells were harvested by centrifugation, washed with 0.9% sodium chloride, and stored at Ϫ20°C.
Frozen cell mass (5 g) was suspended in 35 ml of 50 mM Tris hydrochloride, pH 8.0. The suspension was subjected to ultrasonic treatment and centrifuged. The supernatant was applied to a Q-Sepharose Fast Flow column (2 ϫ 15 cm) that had been equilibrated with 20 mM Tris hydrochloride, pH 8.0. The column was washed with 150 ml of 20 mM Tris hydrochloride, pH 8.0, and developed with a linear gradient of 0 -1 M sodium chloride containing 20 mM Tris hydrochloride, pH 8.0 (total volume, 300 ml; flow rate, 1 ml min Ϫ1 ). 6-Hydroxymethyl-7,8-dihydropterin pyrophosphokinase was eluted from 100 to 130 ml. Fractions were combined and concentrated by ultrafiltration (Omega membrane 10K). The solution was placed on a column of Superdex 75 (2.6 ϫ 60 cm) which was then developed with 20 mM Tris hydrochloride, pH 8.0, containing 100 mM sodium chloride at a flow rate of 3 ml min Ϫ1 . Fractions were combined and concentrated by ultrafiltration. The concentration of 6-hydroxymethyl-7,8-dihydropterin pyrophosphokinase was monitored photometrically using an absorbance coefficient of 12,090 M Ϫ1 cm Ϫ1 (280 nm). The specific activity was 395 mol mg Ϫ1 h Ϫ1 . The enzyme was stored at Ϫ70°C in 60 mM potassium phosphate, pH 7.6.
Purification of Recombinant Dihydropteroate Synthase-E. coli strain M15[pRep4,pDHPS] directing the expression of the folP gene of H. influenzae was grown in LB medium containing 20 mg of kanamycin and 50 mg of ampicillin per liter. The cultures were incubated with shaking at 37°C to an optical density of 0.7 (600 nm). IPTG was added to a concentration of 2 mM, and incubation was continued for a period of 3 h. Cells were harvested by centrifugation, washed with 0.9% sodium chloride, and stored at Ϫ20°C.
Frozen cell mass (4 g) was thawed in 30 ml of 50 mM Tris hydrochloride, pH 8.0. The suspension was subjected to ultrasonic treatment and centrifuged. The supernatant was loaded on a Q-Sepharose Fast Flow column (2 ϫ 15 cm) that had been equilibrated with 20 mM Tris hydrochloride, pH 8.0, containing 1 mM EDTA. The column was washed with 200 ml of 20 mM Tris hydrochloride, pH 8.0, containing 1 mM EDTA, and developed with a linear gradient of 0 -1 M sodium chloride in 20 mM Tris hydrochloride, pH 8.0, containing 1 mM EDTA (flow rate, 1 ml min Ϫ1 ; total volume, 400 ml). Dihydropteroate synthase was eluted from 120 to 140 ml. Fractions were combined and concentrated by ultrafiltration (Omega membrane 10K). The solution was placed on top of a Superdex 200 column (2.6 ϫ 60 cm) that was developed with 60 mM potassium phosphate, pH 7.8, at a flow rate of 3 ml min Ϫ1 . The concentration of dihydropteroate synthase was monitored photometrically using an absorbance coefficient of 24,070 M Ϫ1 cm Ϫ1 (280 nm). The specific activity was 4.1 mmol mg Ϫ1 h Ϫ1 . The enzyme was stored at Ϫ70°C in 60 mM potassium phosphate, pH 7.6.
Purification of Recombinant Dihydrofolate Reductase-The recombinant E. coli strain BL21(DE3)[pLysS,pDHFR] directing the expression of the folA gene of E. coli was grown in LB medium containing chloramphenicol (20 mg/liter) and ampicillin (50 mg/liter) at 37°C to an optical density of 0.8 (600 nm). IPTG was added to a concentration of 1 mM, and incubation was continued for a period of 4 h. Cells were harvested by centrifugation, washed with 0.9% sodium chloride, and stored at Ϫ20°C.
Frozen cell mass (5.5 g) was suspended in 40 ml of 20 mM potassium phosphate, pH 7.0, containing 100 mM potassium chloride and 1 mM EDTA. The suspension was subjected to ultrasonic treatment and centrifuged. The supernatant was applied to a DEAE-Sepharose FF column (2 ϫ 15 cm) that had been equilibrated with 20 mM potassium phosphate, pH 7.0, containing 100 mM potassium chloride and 1 mM EDTA. The column was developed with a linear gradient of 0 -0.5 M potassium chloride (flow rate, 1 ml min Ϫ1 ; total volume, 400 ml). Dihydrofolate reductase was eluted from 240 to 290 ml. The concentration of dihydrofolate reductase was monitored photometrically using an absorbance coefficient of 31,100 M Ϫ1 cm Ϫ1 (280 nm) (17). The specific activity was 17 mol mg Ϫ1 h Ϫ1 . The enzyme was stored at Ϫ70°C in 60 mM potassium phosphate, pH 7.6.
Synthesis of [11-13 C 1 ]6-(RS)-5,10-Methylenetetrahydropteroate-To a reaction mixture containing 60 mM potassium phosphate, pH 7.6, and 1 mM dihydropteroic acid in a total volume of 1 ml, sodium borohydride was added to a final concentration of 10 mM under a stream of nitrogen (22). After 20 min, the excess of sodium borohydride was destroyed by the addition of 20 l of 17 M acetic acid, and the solution was adjusted to pH 7.1 with 10 N NaOH. Mercaptoethanol was added to a final concentration of 2 mM, and [ 13 C]formaldehyde was added to a concentration of 3 mM. The formation of 6-(R,S)-[11-13 C 1 ]5,10-methylenetetrahydropteroate was monitored by HPLC and by NMR spectrometry.  1. Biosynthesis of tetrahydrofolate (compound 7). The reaction catalyzed by dihydroneopterin aldolase (FolB, reaction C) is shown in a box.
A mixture containing 60 mM potassium phosphate, pH 7.6, 10 mM [1Ј,2Ј,3Ј,4Ј,5Ј-13 C 5 ]GTP, and 20 mg of GTP cyclohydrolase I of E. coli in a total volume of 1 ml was incubated at 37°C for 5 h. A solution (161 l) made up from 100 l of 60 mM potassium phosphate, pH 7.6, 10 l of 100 mM magnesium chloride, 1 l of 100 mM zinc chloride, and 50 l of alkaline phosphatase solution (50 units) was added, and incubation at 37°C was continued for 2 h. The solution was passed through an ultrafilter (Nanosep 30K Omega) by centrifugation at 20,000 ϫ g. The ultrafiltrate was collected under a stream of nitrogen.
NMR Spectroscopy-NMR measurements were performed at 27°C using a Bruker DRX 500 spectrometer operating at 500.13 and 125.7 MHz for 1 H and 13 C NMR experiments, respectively. The spectrometer was equipped with a lock-switch unit for 2 H-decoupling experiments using the lock channel.
One-dimensional 1 H and 13 C NMR experiments and two-dimensional COSY, NOESY, ROESY, HMQC, and HMQC-TOCSY experiments were performed with standard Bruker software (XWIN-NMR 3.0). 13 C spectra were obtained using a dual 13 C/ 1 H probe head, and proton detection experiments were performed using a 1 H/ 13 C/ 15 N inverse triple resonance probe head. The mixing time was 500 ms for NOESY, 300 ms for ROESY, and 60 ms for TOCSY transfer.

RESULTS
Dihydroneopterine aldolase had been proposed to cleave 7,8dihydro-D-neopterin (2, Fig. 2) by a retroaldol-type mechanism, although no experimental evidence has been reported (24). The hypothetical reaction sequence requires the protonation of carbon 6␣ of the postulated intermediate (8). The proton for that reaction step could be supplied by an acidic group of the protein, by the solvent, or by internal return of a proton abstracted from the substrate in an earlier reaction step. In order to determine the stereochemical features of the protonation, we decided to study the incorporation of deuterium from solvent D 2 O into the heterocyclic reaction product by multinuclear NMR spectroscopy.
For the purpose of NMR analysis, we designed a reaction sequence in order to convert the enzyme product, 6-hydroxymethyl-7,8-dihydropterin (3), into 5,10-methylenetetrahydropteroate (11, Fig. 3). In that tricyclic compound, carbon atom 9 which is equivalent to the 6␣ carbinol group of 6-hydroxymethyl-7,8-dihydropterin (3) is embedded into a five-membered, relatively rigid ring system, which was expected to enable the stereospecific assignment of the diastereotopic protons of the position 7 and 9 methylene groups of 11 by NMR spectroscopy using carbon 6 as a chiral reference center.
The reaction steps designated A, C, D, and E in Fig. 1 are catalyzed by enzymes of the tetrahydrofolate pathway and all were documented in the literature (4,6,(25)(26)(27)(28)(29)(30). The details of removal of the triphosphate motif under physiological conditions in the biosynthetic pathway (Fig. 1, reaction B) are still not known with certainty; however, alkaline phosphatase could be used for the dephosphorylation of 1.
The enzymatic reduction of dihydropteroate (5) affording 6-(S)-tetra-hydropteroate (6-S-10) (Fig. 3) has not been described to the best of our knowledge, but we could show that the non-physiological reaction can be catalyzed at an appreciable rate by dihydrofolate reductase of E. coli. The conversion of 6-(S)-tetrahydropteroate into 6-(R)-5,10-methylenetetrahydropteroate (6-R-11) could be performed by treatment with formaldehyde following a suggestion of R. Matthews. 2 It should be noted that the experimental strategy required the recombinant expression of the folK, folP, and folA genes in order to prepare 6-hydroxymethyldihydropterin pyrophosphokinase, dihydropteroate synthase, and dihydrofolate reductase as described under "Experimental Procedures." In order to check the feasibility of the planned NMR analysis, we prepared racemic 6-(R,S)-5,10-methylenetetrahydropteroate (11) by sodium borohydride reduction of a commercial sample of 7,8-dihydropteroate (5) followed by condensation with formaldehyde (Fig. 3). In order to improve the sensitivity and selectivity of product analysis, we used 13 C-labeled formaldehyde in the condensation reaction. 1 H and 13 C NMR sig-2 R. Matthews, personal communication. nals of the resulting 6-(R,S)-[11-13 C 1 ]5,10-methylenetetrahydropteroate were assigned by two-dimensional COSY, NOESY, ROESY, and HMQC spectroscopy (Tables III and IV).
The 1 H NMR signals for the diastereotopic protons at C-11, C-9, and C-7 of 6-(R,S)-[11-13 C]5,10-methylenetetrahydropteroate were found to be well separated from each other (Table  III and Fig. 4). More specifically, the resonances for H-11 protons and the C-11 carbon atom in 6-(R,S)-[11-13 C]5,10-methylenetetrahydropteroate were assigned by their enhanced intensities in two-dimensional 1 H 13 C HMQC experiments. The expected 1 H 13 C couplings of the H-11 protons with the attached 13 C-11 were also observed in the one-dimensional 1 H NMR spectra of 6-(R,S)-[11-13 C]5,10-methylenetetrahydropteroate (cf. Fig. 5A). The doublet of triplets centered at 3.20 ppm showed coupling to the NH-8 proton in the two-dimensional COSY experiment. Moreover, the coupling signature of the well resolved signal in the one-dimensional 1 H NMR experiment (Table III and Fig. 4) indicated coupling to NH-8 (coupling constant, 4 Hz). Consequently, the signal at 3.20 ppm was assigned as one of the H-7 protons (H-7e). The signal for the second H-atom (H-7d) attached to C-7 was assigned by HMQC spectroscopy where the signals for H-7e and H-7d were both correlated to the same 13 C NMR signal at 38.9 ppm. The signals for H-9 methylene protons could then be assigned via the 1 H 1 H correlation pattern gleaned from the COSY experiment as well as from 1 H 13 C correlations in the HMQC experiment (Table III). Stereo-specific assignments were obtained from ROESY and NOESY experiments (Fig. 5 and Table IV). Thus, negative cross-peaks between resonances for H-9b, H-6, and H-7e (Fig. 5C), respectively H-9a and H-7d (Fig. 5B) indicated syn orientation of H-6, H-9b, and H-7e.
The compounds 10 and 11 are oxygen sensitive; partial degradation of 5,10-methylenetetrahydropteroate (11) was observed after a few hours even at 5°C. We therefore decided to perform the reaction steps in Fig. 3 as a one-pot reaction with direct NMR analysis. As a substitute for product purification, we made use of 13 C labeling (Fig. 6) in order to enhance the sensitivity and selectivity of the NMR analysis. As described below, this approach enabled the acquisition of 13 C-filtered 1 H NMR data, thus virtually eliminating the 1 H signals of unlabeled reagents.
Two-dimensional HMQC spectra of 6-(R)- [6,7,9,11-13 C 4 ]5,10methylenetetrahydropteroate obtained in water resp. in D 2 O are shown in Fig. 7. In the experiment with deuterium-enriched buffer, the intensities of the signals assigned as H-9 pro-S and H-9 pro-R are reduced to relative values of 14 and 34% as compared with product formed in experiments without deuterium enrichment (Table V). Thus, H-9 pro-R contained 66% deuterium, and H-9 pro-S contained 86% deuterium. These data are discussed in more detail below. In line with earlier work (11), 77% deuterium introduced by the action of GTP cyclohydrolase I was found at H-7 pro-R ; no significant amount of deuterium was detected in the H-7 pro-S position. DISCUSSION In an earlier study (11), we showed that the action of GTP cyclohydrolase in D 2 O afforded [7-2 H 1 ]dihydroneopterin triphosphate with 2 H located at the 7-pro-R position (Fig. 8). In that study, the stereochemistry of deuterium incorporation was determined by transformation of the product into tetrahydrobiopterin by the catalytic action of 6-pyruvoyl-tetrahydropterin synthase and sepiapterin reductase. The stereochemical assignments of the protons at the position 7 methylene group could then be determined using C-6 of tetrahydrobiopterin as a chiral reference.
In the present study, the deuterium found at C-7 of 6-(R)- [6,7,9,11-13 C 4 ]5,10-methylenetetrahydropteroate (11) has been incorporated from solvent by GTP cyclohydrolase I. Hence, the deuterium at C-7 of compound 11 must be located in the pro-R position (i.e. H-7d in Fig. 3). Because we could show that H-7d and H-6 in 11 have anti orientation (Table III), the findings confirm that dihydrofolate reductase converts dihydropteroate into 6-(S)-tetrahydropteroate. Thus, our working hypothesis that the reduction of dihydropteroate and dihydrofolate by dihydrofolate reductase affords products with the same stereochemical features (see above) is now confirmed by direct evidence.
The residual 1 H in the H-7 pro-R position (23%) of 11 indicates that the solvent had not been completely deuterated. This finding was not unexpected because H 2 O/D 2 O exchange had been performed by lyophilization of the multienzyme mixture. The impact of incomplete deuteration of the solvent may have been enhanced by a deuterium isotope effect because release of a proton from bulk water or from an acidic group in exchange with the bulk solvent would occur preferentially by comparison with the release of a deuteron.
Whereas the hypothetical reaction mechanism in Fig. 2 suggests the incorporation of precisely one deuterium equivalent at C-9 of 11 by protonation of the hypothetical intermediate 8 (Fig. 2), the data in Table V show a considerable excess of 2 H. Specifically, 86% deuterium was found in the pro-S position and 66% in the pro-R position at C-9 of 11. This could be due to stereospecific protonation of the enol 8 in the pro-S position in conjunction with wash-in of additional deuterium by spontaneous or enzyme-mediated exchange processes.
The 6␣ hydrogen atoms of compound 3 have increased CH acidity due to activation by the aminal motif in the dihydropyrazine ring. This activation should also extend to the adjacent ring atom, C-7, but the acidity of the C-6␣ and C-7 protons may well differ quantitatively, and this could explain the absence of excess deuterium in the position 7 methylene group of compound 11.
The data leave no doubt that the protonation of the intermediate 8 by dihydroneopterin aldolase occurs preferentially in the pro-S position (Fig. 8). However, due to the excess deuterium at the position 6␣ carbon of compound 3, the degree of stereospecificity of the protonation of the intermediate 8 cannot be determined unequivocally.
The present data do not answer the question whether the donor for proton transfer to the 6␣ carbon of the enol 8 is an acidic group of the protein or a solvent water. However, it is clear that the protonation of the enol does not occur by return of the proton that has been abstracted from the substrate. Thus, if the protonation of 8 occurs by an amino acid residue involved in the formation of an anion from the substrate, 2, that residue would have to be water-accessible, and the abstracted proton would have to be exchanged with bulk solvent.  [6,7,9,11-13 C 4