Evidence for the hypothesis that 10-formyldihydrofolate is the in vivo substrate for aminoimidazolecarboxamide ribotide transformylase

Abstract

We postulate that 10-formyl-7,8-dihydrofolate (10-HCO-H₂folate), not 10-formyl-5,6,7,8-tetrahydrofolate (10-HCO-H₄folate), is the predominant in vivo substrate for mammalian aminoimidazolecarboxamide ribotide (AICAR) transformylase, an enzyme in purine nucleotide biosynthesis de novo, which introduces carbon 2 (C₂) into the purine ring. 10-HCO-H₂folate exists in vivo as labeled 10-formyl-folic acid (10-HCO-folic acid: an oxidation product of 10-HCO-H₄folate and 10-HCO-H₂folate) and is found after doses of labeled folic acid in humans or laboratory animals. The bioactivity of the unnatural isomer, [6R]-5-formyltetrahydrofolate, in humans is explained by its in vivo conversion to 10-HCO-H₂folate. The structure and active site of AICAR transformylase are not consistent with other enzymes that utilize 10-HCO-H₄folate. Because 10-HCO-H₄folate is rapidly oxidized in vitro to 10-HCO-H₂folate by cytochrome C alone and in mitochondria, it is hypothesized that this process takes place in vivo. In vitro data indicate that 10-HCO-H₂folate is kinetically preferred over 10-HCO-H₄folate by AICAR transformylase and that this enzyme may not have access to sufficient supplies of 10-HCO-H₄folate. Methotrexate blockage of the AICAR transformylase process in patients with rheumatoid arthritis suggests that dihydrofolate (H₂folate) reductase is involved and is consistent with H₂folate and 10-HCO-H₂folate being the product and substrate for AICAR transformylase. The labeling of purine C₂ by an oral dose of [6RS]-5-H[¹³C]O-H₄folate in a human subject is consistent with 10-H[¹³C]O-H₂folate formation from unnatural isomer, [6R]-5-H[¹³C]O-H₄folate, and it being a substrate for AICAR transformylase. In vitro exchange reactions of purine C₂ using H₄folate coenzymes are not duplicated in vivo and is consistent with H₂folate coenzymes being used in vivo by AICAR transformylase.

Keywords

AICAR transformylase 10-formyldihydrofolate folate purine nucleotide biosynthesis

Introduction

The vast majority of the purine ring is biosynthesized de novo rather than coming from the diet.¹ Two folate-dependent transformylases, glycinamide ribotide (GAR) and aminoimidazolecarboxamide ribotide (AICAR) transformylases, are required for purine nucleotide biosynthesis (PNB) de novo in mammals by introducing ring carbons 8 (C₈) and 2 (C₂), respectively.² It is assumed that 10-formyl-5,6,7,8-tetrahydrofolate (10-HCO-H₄folate) is the in vivo substrate for both enzymes, because it has been shown that only 5,6,7,8-H₄folate is capable of enzymatically acquiring a one-carbon unit.^3,4 Therefore, cellular pools of H₄folate are maintained, because 7,8-dihydrofolate (H₂folate) reductase (DHFR) catalyzes reduction of H₂folate to H₄folate.⁵ However, in vitro mammalian AICAR transformylase (not GAR transformylase) is able to utilize both 10-HCO-H₄folate and 10-formyl-7,8-dihydrofolate (10-HCO-H₂folate) with H₄folate and H₂folate as products, respectively.⁶ In this review, we present a hypothesis that 10-HCO-H₂folate (not 10-HCO-H₄folate) is the actual in vivo substrate for mammalian AICAR transformylase. To support this hypothesis, we present in vivo and in vitro findings, including our previously published data and new analyses of those as well as findings by others.

Presence of 10-formyldihydrofolate in vivo

An important prerequisite for 10-HCO-H₂folate being the in vivo substrate for AICAR transformylase in PNB is the proof indicating that this compound exists in vivo. Researchers have identified isotope-labeled or -unlabeled 10-formyl-folic acid (10-HCO-folic acid) in human bile, portal plasma and urine within a few hours after a dose of labeled folic acid or 5-formyl-5,6,7,8-tetrahydrofolate (5-HCO-H₄folate).^7–9 These samples were collected in the presence of ascorbate to prevent the oxidation of reduced folates. In order to acquire a one-carbon unit, folic acid must first be reduced to H₄folate. Therefore, in these studies, H₄folate must have acquired a one-carbon unit to form labeled 10-HCO-H₄folate that must have been oxidized to form 10-HCO-folic acid. The obligate intermediate in this two-step-oxidation process is 10-HCO-H₂folate that was in fact tentatively identified in the human bile.⁷ The first oxidation of 10-HCO-H₄folate to 10-HCO-H₂folate is more rapid than the second oxidation to 10-HCO-folic acid,^6,10 suggesting that 10-HCO-H₂folate was rapidly formed after the dose of folic acid or 5-HCO-H₄folate. Labeled 10-HCO-folic acid was also identified in the rat liver, bile and urine after the dose of [¹⁴C]- and [³H]-labeled folic acid.^11–13 Although labeled 10-HCO-folic acid is detected after a dose of labeled folic acid, it is in relatively small quantities compared with the pool of labeled tetrahydrofolates. Therefore, the in vivo concentration of its precursor, 10-HCO-H₂folate, may also be relatively small.^7,11–13

It has been reported that an oral dose of the unnatural isomers, [6R]-5-HCO-H₄folate and [6S]-5,10-methenyltetrahydrofolate (5,10-CH = H₄folate), were 25% and 50% as bioactive compared with [6S]-5-HCO-H₄folate, respectively, in humans.¹⁴ As described, both [6R]-5-HCO-H₄folate and [6S]-5,10-CH = H₄folate are converted chemically in the gastrointestinal tract to [6S]-10-HCO-H₄folate.¹⁴ Oxidation of the pteridine ring of [6S]-10-HCO-H₄folate (unnatural isomer) to the 7,8-dihydro-oxidation state (therefore, destroying the chiral center) converts this unnatural isomer to 10-HCO-H₂folate. This folate is now bioactive, because it is a substrate for AICAR transformylase producing H₂folate that can be enzymatically reduced to H₄folate by DHFR.⁶ The fact that [6S]-5,10-CH = H₄folate was more bioactive than [6R]-5-HCO-H₄folate is consistent with the fact that the former is an intermediate in the conversion of [6R]-5-HCO-H₄folate to 10-HCO-H₂folate.¹⁴ Chemical formation of [6S]-10-HCO-H₄folate and its oxidation to 10-HCO-H₂folate must have been rapid because the experiments lasted only 4 h. Baggott and Tamura¹⁴ were not the first to present data suggesting that [6R]-5-HCO-H₄folate is bioactive. Although Baker et al.¹⁵ did not report this, re-evaluation of their data using an 8-h follow-up after oral doses of [6RS]-5-HCO-H₄folate (believed to be only 50% bioactive) and folic acid indicated that the unnatural isomer, [6R]-5-HCO-H₄folate, was ∼80% as bioactive compared with folic acid in humans.^16,17 Devito et al.¹⁸ also reported that an intravenous dose of [6RS]-5-HCO-H₄folate produced 19% more bioactive folates than one-half that of the [6S]-5-HCO-H₄folate dose in healthy subjects. Therefore, formation of [6S]-10-HCO-H₄folate from [6R]-5-HCO-H₄folate and its oxidation to 10-HCO-H₂folate may take place in sites other than the gastrointestinal tract. In addition, using high-performance liquid chromatography with a microbiological assay, Baggott and Tamura¹⁴ identified 10-HCO-H₂folate and 10-HCO-folic acid in 2-h urine samples after doses of either [6R]- or [6S]-5-HCO-H₄folate,¹⁴ indicating that metabolites of these natural and unnatural isomers must undergo in vivo oxidation to 10-HCO-H₂folate. Based on the above data, Baggott and Tamura¹⁴ conclude that 10-HCO-H₂folate exists in vivo.

In vitro data supporting our hypothesis

Unique properties of AICAR transformylase

Enzymes that utilize H₂folates include thymidylate synthase, DHFR and AICAR transformylase. Thymidylate synthase catalyzes the production of H₂folate from an H₄folate, and DHFR reduces H₂folate to H₄folate. On the other hand, AICAR transformylase is unique because both substrate and product can be H₂folates. Further, it is important to note that AICAR transformylase has neither structural similarity to GAR transformylase¹⁹ nor amino acid sequence for the 10-HCO-H₄folate binding site that is present in GAR transformylase, 10-HCO-H₄folate synthetase and 10-HCO-H₄folate dehydrogenase of a variety of species such as Escherichia coli and humans.^20,21 Site-directed mutagenesis of a suspected 10-HCO-H₄folate binding site with a minor homology to the site in GAR transformylase has no effect on AICAR transformylase activity.²¹ Another aspect of the uniqueness in AICAR transformylase is that it is the only folate-dependent enzyme with a homodimer quaternary structure and the active site at the dimmer interface; thus, the monomer is inactive.¹⁹ The pterine ring of folate analogs binds differently to AICAR and GAR transformylases. The ring is more exposed to the bulk water (possibly more exposed to oxidation) in AICAR transformylase.¹⁹ Considering the above, it would be reasonable that folate coenzyme preference by AICAR transformylase evolved with a preference for 10-HCO-H₂folate in a way not similar to other 10-HCO-H₄folate-utilizing enzymes.

In vitro lability of 10-HCO-H₄folate

10-HCO-H₄folate is extremely labile to oxidation in vitro. It has been reported that a trace amount of iron in high-grade reagents catalyzes the oxidation of 10-HCO-H₄folate to 10-HCO-H₂folate.²² Therefore, 10-HCO-H₂folate cannot be prepared in vitro without being contaminated with oxidation products.²³

Oxidized cytochrome C reacts with 10-HCO-H₄folate with a second-order rate constant of 1.3 × 10⁴/mol/L × s.¹⁰ As cytochrome C concentration is 100–200 µmol/L in the intermembrane space, and mitochondrial folate concentration is 1–5 µmol/L, the pseudo-first-order oxidation rate suggests that 10-HCO-H₄folate would have a half-life of 1 s or less in the intermembrane space.¹⁰ As the protein porin forms channels in the outer mitochondrial membrane that allow free diffusion of molecules of 10,000 Da or less,²⁴ 10-HCO H₄folate and its polyglutamates found in the cytoplasm could come in contact with cytochrome C in the intermembrane space and diffuse out to the cytoplasm as 10-HCO-H₂folate to be utilized by AICAR transformylase. The above process would explain how isolated rat liver mitochondria rapidly form 10-HCO-H₂folate from 10-HCO-H₄folate in the solution outside the organelle.²⁵ Therefore, 10-HCO-H₄folate must be short-lived if it is formed in the intermembrane space. However, it is possible that 10-HCO-H₄folate may still be able to play a role as a one-carbon donor to AICAR transformylase before it is oxidized. In conclusion, the oxidation of 10-HCO-H₄folate to 10-HCO-H₂folate occurs so readily suggesting the in vivo existence of the latter.

In vitro enzyme kinetics of AICAR transformylase

There is in vitro evidence indicating that 10-HCO-H₂folate is the preferred substrate for AICAR transformylase. 10-HCO-H₂folate has a ∼5-fold kinetic advantage (V _m/K _m) over 10-HCO-H₄folate for AICAR transformylases in human leukemia cell and rat bone marrow,⁶ and in human recombinant AICAR transformylase.²⁶ This kinetic advantage is largely due to a lower K _m (tighter binding) for 10-HCO-H₂folate, and this is consistent with 10-HCO-H₂folate, rather than 10-HCO-H₄folate being the preferred in vivo substrate. However, it is not known whether polyglutamates of 10-HCO-H₂folate would have the same kinetic advantage over polyglutamates of 10-HCO-H₄folate.

It is likely that 10-HCO-H₄folate, which is utilized by GAR transformylase, is supplied in vivo by an enzyme complex, originally described in vitro by Benkovic and colleagues.^27,28 This complex contains the trifunctional-folate-metabolizing protein (TFM), serine hydroxymethyltransferase (SHMT) and GAR transformylase. The biological purpose of the complex would be to furnish or channel the labile 10-HCO-H₄folate to GAR transformylase immediately after 10-HCO-H₄folate is produced.

Consistent with this channeling hypothesis, Benkovic and colleagues²⁹ recently reported in cultured human cancer cells that the protein with GAR transformylase activity forms clusters in the cytoplasm. In contrast, AICAR transformylase does not form such clusters, and TFM is evenly distributed in the cytoplasm. Therefore, it is likely that any 10-HCO-H₄folate produced by TFM that is in close proximity to a GAR transformylase cluster must be preferentially utilized by this enzyme and stands little chance of ‘finding’ AICAR transformylase. This suggests that 10-HCO-H₄folate produced by TFM must diffuse a greater distance with an increased risk of oxidation before its utilization by AICAR transformylase. In addition, Baggott et al.⁶ found in human leukemia cells that 10-HCO-H₄folate K _m's were 4.9 and 420 µmol/L for GAR and AICAR transformylases, respectively. Thus, GAR transformylase has a higher affinity for 10-HCO-H₄folate than AICAR transformylase.

Taken together, the above data suggest that 10-HCO-H₄folate is channeled to GAR transformylase by forming clusters with a binding affinity advantage over AICAR transformylase; therefore, less 10-HCO-H₄folate is available to AICAR transformylase. It is likely that folate coenzymes are effectively cycled and channeled to each enzyme in the complex (TFM, SHMT and GAR transformylase) at the H₄folate-oxidation state. This series of reactions are shown in Reaction Sequence 1 (Table 1).

Table 1

Reaction Sequence 1. Utilization of the carbon 3 of serine by GAR transformylase

Enzyme	Reaction
SHMT	Serine + H₄folate ⇄ glycine + 5,10-CH₂-H₄folate
TFM	5,10-CH₂-H₄folate + NADP ⇄ 5,10-CH = H₄folate + NADPH
TFM	5,10-CH = H₄folate + H₂O ⇄ 10-CHO-H₄folate
GAR transformylase	10-CHO-H₄folate + GAR→ formyl-GAR + H₄folate
Net	Serine + NADP + H₂O + GAR → glycine + NADPH + formyl-GAR

GAR, glycinamide ribotide; NADP, nicotinamide adenine dinucleotide phosphate; NADPH, reduced form of nicotinamide adenine dinucleotide phosphate; 5,10-CH = H₄folate, 5,10-methenyltetrahydrofolate; TFM, trifunctional-folate-metabolizing protein; SHMT, serine hydroxymethyltransferase

No enzyme complex that generates 10-HCO-H₄folate and immediately channels or furnishes it to AICAR transformylase is needed because the first oxidation product of 10-HCO-H₄folate is 10-HCO-H₂folate that is utilized by this transformylase. The oxidation of 10-HCO-H₄folate could occur in a reaction with oxidized cytochrome C and is probably not simply an in vitro phenomenon, because the oxidation is fast enough to support respiration in isolated rat liver mitochondria.²⁵ The second-order rate constant for the oxidation of 10-HCO-H₂folate to 10-HCO-folic acid by oxidized cytochrome C is slow.¹⁰ Therefore, the relatively stable 10-HCO-H₂folate requires less protection from oxidation than 10-HCO-H₄folate and can be utilized by AICAR transformylase as shown in Reaction Sequence 2 (Table 2).

Table 2

Reaction Sequence 2. Formate and 10-HCO-H₂folate utilized by AICAR transformylase

Enzyme	Reaction
TFM	H₄folate + ATP + formate ⇄ 10-HCO-H₄folate + ADP + PO₄
Oxidation*	Substrate_OX + 10-HCO-H₄folate → 10-HCO-H₂folate + substrate_RED
AICAR transformylase	10-HCO-H₂folate + AICAR⇄ formyl-AICAR + H₂folate
AICAR transformylase^†	Formyl-AICAR ⇄ IMP + H₂O
DHFR	H₂folate + NADPH → H₄folate + NADP
Net	Substrate_OX + ATP + formate + AICAR + NADPH → Substrate_RED + ADP + PO₄ + IMP + H₂O + NADP

10-HCO-H₂folate, 10-formyl-7,8-dihydrofolate; AICAR, aminoimidazolecarboxamide ribotide; TFM, trifunctional-folate-metabolizing protein; IMP, inosine 5′-monophosphate; DHFR, 7,8-dihydrofolate (H₂folate) reductase; NADP, nicotinamide adenine dinucleotide phosphate; NADPH, reduced form of nicotinamide adenine dinucleotide phosphate; OX, oxidized; RED, reduced

*This could be a reaction with cytochrome C or other compounds

^†It is assumed that AICAR transformylase also possesses the IMP cyclohydrolase activity

In conclusion, the data in this section suggest that 10-HCO-H₂folate, an oxidation product of 10-HCO-H₄folate, is the preferred substrate for AICAR transformylase and 10-HCO-H₄folate is the preferred substrate for GAR transformylase. This conclusion appears to be reasonable because it would be biologically wasteful for 10-HCO-H₂folate, an intact folate molecule, to be a metabolically inert dead-end catabolite.

In vivo data in humans and animals supporting our hypothesis

Methotrexate therapy of patients with rheumatoid arthritis

Smoleńska et al.³⁰ measured the change in whole-blood hypoxanthine concentrations in adult patients with rheumatoid arthritis (RA) shortly before (baseline) and 2 h after an oral dose of methotrexate (MTX) (7.5 mg; patient's first MTX dose). There was a significant 36% decrease in mean hypoxanthine from baseline. The mean blood uric acid concentration was also decreased to a significant 32% from baseline, suggesting an overall decrease in PNB.

The total absorption of oral MTX is ∼80%, of which ∼70% would be absorbed in 2 h.³¹ Thus, the above patients would have absorbed 4.2 mg of the 7.5-mg dose (7.5 mg × 0.8 × 0.7). Excluding urinary MTX excretion, in vivo retention would be ∼9 × 10⁻⁶ mol (4.2 mg) or in a 70-kg patient ∼1.3 × 10⁻⁷ mol/kg in 2 h. Morgan et al.³² showed that low-dose MTX therapy significantly increases urinary aminoimidazolecarboxamide (AICA) excretion in 24 h, indicating an interference with AICAR transformylase by the drug. Therefore, it is likely that this interference is the target resulting in a decreased inosine 5′-monophosphate (IMP) biosynthesis, thus decreasing blood hypoxanthine. The question remains as to whether the interference with AICAR transformylase is directly caused by MTX. The competitive inhibition constant (K _i) for MTX of AICAR transformylase is 4.0 × 10⁻⁵ mol/L,³³ and it is ∼300 times higher than the estimated in vivo MTX retention of ∼1.3 × 10⁻⁷ mol/kg, which is negligible compared with this K _i. It is concluded that MTX cannot directly inhibit AICAR transformylase.

Inhibition of an enzyme other than AICAR transformylase is likely to be responsible for decreased blood hypoxanthine³⁰ and increased urinary AICA excreted in these patients.³² One logical choice of an enzyme affected by a low MTX concentration is DHFR. The K _i for MTX for human DHFR is 6.1 × 10⁻¹² mol/L,³⁴ and ∼1.3 × 10⁻⁷ mol/kg is ∼20,000 times higher than the K _i. Even such a low MTX concentration would produce potent DHFR inhibition, which could simply reduce the H₄folate-pool leading to an indirect interference with AICAR transformylase. However, this reduction in the H₄folate-pool may not be the actual mechanism, because it would have to occur within 2 h. In fact, Priest and co-workers³⁵ found that low MTX concentrations do not reduce the H₄folate-pool but increase 10-HCO-H₄folate in cultured cells. Pools of other H₄folate compounds were basically unchanged; therefore, there is no evidence that the intracellular H₄folate-pool was depleted by MTX even though cell growth was reduced. Dervieux et al.³⁶ reported that erythrocyte H₄folate-pools in patients with RA were decreased only 13% from baseline after ∼18 weekly doses of MTX (7.5–15 mg/week). Therefore, it is likely that the first MTX dose in patients in the investigation by Smoleńska et al.³⁰ resulted in a trivial reduction in the H₄folate-pool in erythrocytes.

Therefore, an alternative mechanism of MTX interference with AICAR transformylase should be investigated. This may involve the utilization of 10-HCO-H₂folate by the enzyme as shown in Reaction Sequence 2. This sequence is driven towards the biosynthetic direction by the DHFR-catalyzed reduction of H₂folate to H₄folate. The MTX inhibition of DHFR would interfere with AICAR transformylase, as the equilibrium of this step actually lies in the direction of AICAR formation.²⁶ Reaction Sequence 2 suggests how DHFR, 10-HCO-H₂folate and H₂folate are involved in the overall net reaction. The production of H₂folate by AICAR transformylase would explain how low MTX concentrations interfere with the net reaction. In conclusion, 10-HCO-H₂folate as the in vivo substrate for AICAR transformylase explains how low-dose MTX therapy in patients with RA affects this enzyme.

¹³C enrichment at C₂ and C₈ of purine after a [6RS]-5-H[¹³C]O-H₄folate dose

In this section, we attempt to explain how 10-HCO-H₂folate is the in vivo substrate for AICAR transformylase by re-analyzing our published data of ¹³C enrichment at C₂ and C₈ of purine after a [6RS]-5-H[¹³C]O-H₄folate dose. The ¹³C enrichment at C₂ and C₈ of urinary uric acid after a 25-mg oral dose of [6RS]-5-H[¹³C]O-H₄folate was measured in an adult.^37,38 Figure 1 shows the relationship between the log of the smoothed C₂/C₈ enrichment ratios and the void number. This logarithmic transformation of the data was used to readily identify ratios <1.0 (negative numbers). Although considerable variations in the C₂/C₈ ratios existed, there are basically two phases after the dose. The first phase consisted of a period of enrichment ratios <1, and the second phase was a large increase of the enrichment ratio with a peak value of ∼10.

Figure 1

C₂/C₈ ¹³C enrichment ratios of urinary uric acid after an oral dose of 25 mg [6RS]-5-H[¹³C]O-H₄folate in a human. The smoothed 1:2:3:2:1 weighted running average of the % ¹³C enrichments of urinary uric acid at the C₂ and C₈ positions were calculated after an oral dose of 25 mg [6RS]-5-H[¹³C]O-H₄folate.³⁸ Closed circles represent the log of the ¹³C enrichment ratio (C₂/C₈) that was plotted against the void number. The data from the first two and last two voids were lost in this process. Open circles represent data that were negative numbers, thus presumed to be low ratios (i.e. negative log C₂/C8 enrichment). The line is a best-fit spine curve. C₂, carbon 2; C₈, carbon 8

The previously held notion that an equal utilization by both GAR (C₈) and AICAR (C₂) transformylases, of [6R]-10-H[¹³C]O-H₄folate (formed by rapid enzymatic reactions from [6S]-5-H[¹³C]O-H₄folate), is not consistent with the data shown in Figure 1. If an equal utilization were true, the C₂/C₈ ratios should have been 1.0 (log = 0) for all voids. To explain these data, we postulate the following hypotheses. The first phase is the result of the rapid enzymatic formation of [6R]-10-H[¹³C]O-H₄folate that is utilized primarily by GAR transformylase. This folate is channeled to and preferentially utilized by GAR transformylase (see the section ‘In vitro lability of 10-HCO-H₄folate’); hence, C₂/C₈ ratios are <1.0. In the second phase, a slow non-enzymatic formation of 10-H[¹³C]O-H₂folate from the unnatural isomer [6R]-5-H[¹³C]O-H₄folate predominates,¹⁴ resulting in high C₂/C₈ ratios, since 10-H[¹³C]O-H₂folate can only be utilized by AICAR transformylase.^6,14 These data resemble what one would predict, if [6R]-10-H[¹³C]O-H₄folate is rapidly formed, channeled to and utilized only by GAR transformylase, and 10-H[¹³C]O-H₂folate is slowly formed and utilized only by AICAR transformylase. Human data reported by Baggott and Tamura¹⁴ are also consistent with 10-HCO-H₂folate being the in vivo substrate for AICAR transformylase.

Carbon exchange at the second position of the purine ring

A C₂ to C₈ exchange in the purine ring was reported by Warren et al.³⁹ When an avian liver preparation containing GAR and AICAR transformylases was incubated with IMP, GAR and H₄folate, both AICAR and formyl-GAR were readily formed (Table 3, Reaction Sequence 3). The C₂ from IMP (in equilibrium with formyl-AICAR) was transferred to H₄folate, and the resulting 10-HCO-H₄folate was utilized to form formyl-GAR. Reaction Sequence 3 shows these reactions for which the apparent equilibrium lies in favor of formyl-GAR and AICAR formation because the GAR transformylase reaction is irreversible. When this formyl-GAR is metabolized to IMP, the former carbon at the C₂ position now appears in the C₈ position. Because GAR transformylase cannot utilize 10-HCO-H₂folate, this exchange requires that H₄folate and 10-HCO-H₄folate are the coenzymes for the AICAR transformylase reaction.⁶ In theory, this exchange shown in Reaction Sequence 3 can occur in vivo.

Table 3

Reaction Sequence 3. Exchange of carbon 2 of IMP to carbon 8 of IMP

Enzyme	Reaction
AICAR transformylase	IMP + H₂O⇄ formyl-AICAR
AICAR transformylase	Formyl-AICAR + H₄folate ⇄ 10-HCO-H₄folate + AICAR
GAR transformylase	10-HCO-H₄folate + GAR → formyl-GAR + H₄folate
Net	IMP + H₂O + GAR → AICAR + formyl-GAR

IMP, inosine 5′-monophosphate; AICAR, aminoimidazolecarboxamide ribotide; 10-HCO-H₂folate, 10-formyl-7,8-dihydrofolate; GAR, glycinamide ribotide

Using an avian liver preparation, Flaks et al.⁴⁰ reported an exchange reaction of C₂ of IMP with the carbon 3 of serine (Reaction Sequence 4, Table 4). This exchange involves TFM, SHMT and H₄folate coenzymes (not H₂folate coenzymes). In theory, this exchange can occur in vivo, and the serine formed by this reaction would be readily oxidized to CO₂.

Table 4

Reaction Sequence 4. Exchange of carbon 2 of IMP to carbon 3 of serine

Enzyme	Reaction
AICAR transformylase	IMP + H₂O ⇄ formyl-AICAR
AICAR transformylase	Formyl-AICAR + H₄folate ⇄ 10-HCO-H₄folate + AICAR
TFM	10-HCO-H₄folate + H₂O ⇄ 5,10-CH = H₄folate
TFM	5,10-CH = H₄folate + NADPH ⇄ 5,10-CH₂-H₄folate + NADP
SHMT	Glycine + 5,10-CH₂-H₄folate ⇄ serine + H₄folate
Net	IMP + H₂O + NADPH + glycine ⇄ AICAR + NADPH + serine

IMP, inosine 5′-monophosphate; AICAR, aminoimidazolecarboxamide ribotide; 10-HCO-H₂folate, 10-formyl-7,8-dihydrofolate; TFM, trifunctional-folate-metabolizing protein; 5,10-CH = H₄folate, 5,10-methenyltetrahydrofolate; NADP, nicotinamide adenine dinucleotide phosphate; NADPH, reduced form of nicotinamide adenine dinucleotide phosphate; SHMT, serine hydroxymethyltransferase

It should be emphasized that exchange reactions (Sequences 3 and 4) require that H₄folate coenzymes function in the AICAR transformylase process. However, it remained to be seen whether these exchange reactions actually occur in vivo. Using adenine labeled with ¹³C at the C₂ position and ¹⁴C at the C₈ position, Abrams⁴¹ evaluated the possibility of these exchanges in rats. Adenine is readily metabolized to IMP.² The C₂/C₈ labeling ratio was not changed compared with the starting material in liver RNA adenine and guanine days after the dose of the double-labeled adenine. Abrams⁴¹ concluded that C₂ exchanges described above do not occur in vivo. However, the fact that no exchange reactions are detected indicates that H₂folate coenzymes are utilized by AICAR transformylase.

Bennett and Karlsson⁴² studied the exchange of 2-[¹⁴C]adenine and 8-[¹⁴C]adenine in mice. Measuring hepatic RNA adenine- and guanine-specific activities, they found no difference comparing C₂ to C₈ in experiments that lasted for 2 weeks, and concluded that C₂ was not involved in exchange reactions described above. Of particular interest is the measurement of expired ¹⁴CO₂ from the C₂- and C₈-labeled compounds in mice. After 20 h, ¹⁴CO₂ production from both compounds was small, and there was no marked difference in ¹⁴CO₂ production from 2-[¹⁴C]adenine compared with 8-[¹⁴C]adenine. These findings are also evidence indicating that the exchange reactions discussed above do not occur in vivo in mice.

In humans, however, data are limited. Baggott et al.⁴³ conducted a human study to measure incorporation of [¹³C]formate into C₂ and C₈ of urinary uric acid. In three-day experiments, two subjects had little incorporation into C₈, but C₂ incorporation was significant. Therefore, the C₂ to C₈ exchange (Reaction Sequence 3) was small. However, a subject incorporated significant amounts into the C₈ position as well as into C₂. Therefore, the data in this subject could be used to test the possible C₂ to C₈ exchange. If C₈ incorporation (all or part) was the result of the exchange from C₂, more enrichment at C₈ should occur in a time-dependent manner as C₂ becomes more enriched. Thus, both C₈ enrichment and the C₈/C₂ enrichment ratio should have increased with time. Figure 2 shows both of these values plotted over a three-day period after the [¹³C]formate dose, and linear regression of the data in both plots indicates negative slopes. Thus, there was no evidence of C₂ to C₈ exchange in this subject. The findings in humans by Baggott et al.⁴³ are, therefore, in agreement with the animal studies.^41,42 It is possible that in vitro exchange reactions are not operable in vivo, because enzymes are not in close proximity and do not exchange folate coenzymes or other factors. However, if H₂folate coenzymes are the in vivo substrate for AICAR transformylase, these exchanges should not occur in vivo.

Figure 2

C₈/C₂ ¹³C enrichment ratios and % ¹³C enrichment at the C₈ position of urinary uric acid after an oral dose of 1.0 g of sodium [¹³C]formate in a human. The smoothed 1:2:1 weighted running averages of % enrichment at the C₂ and C₈ positions (%C₈) and the enrichment ratio (C₈/C₂) of urinary uric acid after an oral dose of 1.0 g of sodium [¹³C]formate were plotted against the void number for subject C.³⁸ The data from the first and last void were lost in this process. Linear regression lines are negative, but are not significant (P > 0.05)

Summary

We conclude that 10-HCO-H₂folate is the in vivo predominant substrate for mammalian AICAR transformylase, and summarize as follows. The in vivo existence of 10-HCO-H₂folate is likely due to the lability to oxidation of 10-HCO-H₄folate, suggesting that 10-HCO-H₂folate must have a metabolic role rather than being a dead-end catabolite. AICAR transformylase does not have a 10-HCO-H₄folate binding site that is present in other 10-HCO-H₄folate-utilizing enzymes and has a unique active site formed by a homodimer. Enzyme-kinetic data suggest the utilization of 10-HCO-H₂folate by AICAR transformylase and utilization of only 10-HCO-H₄folate by GAR transformylase. The labile 10-HCO-H₄folate is likely channeled to GAR transformylase in vivo and the relatively stable 10-HCO-H₂folate is utilized by AICAR transformylase. These findings are consistent with labeling patterns in the purine ring after a dose of [6RS]-5-H[¹³C]O-H₄folate in a human. The unique sensitivity of AICAR transformylase to the low-dose MTX therapy in patients suggests that formation of H₂folate from 10-HCO-H₂folate and inhibition of DHFR are involved in the transformylation process. The facile in vitro exchanges of the C₂ position of purine in the presence of H₄folates and enzymes are not duplicated in vivo, suggesting that H₂folates (which cannot produce facile in vitro exchanges), not H₄folates, are involved in the in vivo AICAR transformylase (C₂) process.

Finally, it may be difficult, if not impossible, to prove that 10-HCO-H₂folate is the in vivo substrate for mammalian AICAR transformylase; however, we can turn the table on this argument. Simply stated, at present, we believe that it is difficult (if not impossible) to design experiments proving that 10-HCO-H₄folate is the one and only substrate for AICAR transformylase in vivo and we are not aware of any experiments that attempted to prove this. It would be biologically wasteful for 10-HCO-H₂folate to be a metabolically inert dead-end catabolite. It is possible that mammalian AICAR transformylase evolved with metabolic flexibility to utilize both 10-HCO-H₂folate and 10-HCO-H₄folate in order to metabolically ‘salvage’ former utilizing the PNB pathway. Changing the relatively in vivo amounts of 10-HCO-H₄folate and 10-HCO-H₂folate could be a regulatory mechanism in the PNB pathway.

References

Berthold

, Crain

, Gouni

, Reeds

, Klein

. Evidence for incorporation of intact dietary pyrimidine (but not purine) nucleotides into hepatic RNA. Proc Natl Acad Sci USA 1995;92:10123–7

Garrett

, Grisham

. The synthesis and degradation of nucleotides. In: Biochemistry. 3rd edn. Belmont, CA: Thomson Brooks/Cole, 2005:853–78

Jaenicke

, Brode

. Untersuchungen über Einkohlenstoffkörper. I. Die Tetrahydrofolatformylase aus Taubenleber Reinigung und Mechanismus. Biochem Zeitsch 1961;334:108–32

Blakley

. The reaction of tetrahydropteroylglutamic acid and related hydropteridines with formaldehyde. Biochem J 1959;72:707–15

Blakley

. Dihydrofolate reductase. In: Blakley

, Benkovic

eds. Folates and Pterins. Vol. 1. New York: John Wiley, 1984:191–253

Baggott

, Johanning

, Branham

, Prince

, Morgan

, Eto

, Vaughn

. Cofactor role for 10-formyldihydrofolic acid. Biochem J 1995;308:1031–6

Pratt

, Cooper

. Folates in plasma and bile of man after feeding folic acid-³H and 5-formyltetrahydrofolate (folinic acid). J Clin Invest 1971;50:455–62

Whitehead

, Pratt

, Viallet

, Cooper

. Intestinal conversion of folinic acid to 5-methyltetrahydrofolate in man. Br J Haematol 1972;22:63–72

McLean

, Chanarin

. Urinary excretion of 5-methyl-tetrahydrofolate in man. Blood 1966;27:386–8

10.

Baggott

, Robinson

, Johnston

. Bioactivity of [6R]-5-formyltetrahydrofolate, an unnatural isomer, in humans and Enterococcus hirae, and cytochrome c oxidation of 10-formyltetrahydrofolate to 10-formyldihydrofolate. Biochem J 2001;354:115–22

11.

Murphy

, Keating

, Boyle

, Weir

, Scott

. The elucidation of the mechanism of folate catabolism in the rat. Biochem Biophys Res Commun 1976;71:1017–24

12.

Barford

, Blair

. Effect of an implanted Walker tumour on metabolism of folic acid in the rat. Br J Cancer 1978;38:122–9

13.

Pheasant

, Connor

, Blair

. The metabolism and physiological disposition of radioactively labelled folate derivatives in the rat. Biochem Med 1981;26:435–50

14.

Baggott

, Tamura

. Bioactivity of oral doses of unnatural isomers, [6R]-5-formyltetrahydrofolate and [6S]-5,10-methenyltetrahydrofolate, in humans. Biochim Biophys Acta 1999;1472:323–32

15.

Baker

, Frank

, Feingold

, Ziffer

, Gellene

, Leevy

, Sobotka

. The fate of orally and parenterally administered folates. Am J Clin Nutr 1965;17:88–95

16.

Baggott

, Tamura

, Baker

. Reevaluation of the metabolism of oral doses of racemic carbon-6 isomers of formyltetrahydrofolate in humans. Br J Nutr 2001;85:653–7

17.

Baggott

, Tamura

. Metabolism of 10-formyldihydrofolate in humans. Biomed Pharmacother 2001;55:454–7

18.

DeVito

, Kozloski

, Tonelli

, Johnson

. Bioequivalence of oral and injectable levoleucovorin and leucovorin. Clin Pharmacol 1993;12:293–9

19.

Wolan

, Greasley

, Wall

, Benkovic

, Wilson

. Structure of avian AICAR transformylase with a multisubstrate adduct inhibitor β-DADF identifies the folate binding site. Biochemistry 2003;42:10904–14

20.

Cook

, Lloyd

, Wagner

. Isolation and characterization of cDNA clones for rat liver 10-formyltetrahydrofolate dehydrogenase. J Biol Chem 1991;266:4965–73

21.

Rayl

, Moroson

, Beadsley

. The human purH gene product, 5-aminoimidazole-4-carboxamide ribonucleotide formyltransferase/IMP cyclohydrolase. Cloning, sequencing, expression, purification, kinetic analysis, and domain mapping. J Biol Chem 1996;271:2225–33

22.

Baggott

, Robinson

, Eto

, Johanning

, Cornwell

. Iron compounds catalyze the oxidation of 10-HCO-5,6,7,8 tetrahydrofolic acid to 10-HCO-dihydrofolic acid. J Inorg Biochem 1998;71:181–7

23.

Poe

, Benkovic

. 5-Formyl- and 10-formyl-5,6,7,8-tetrahydrofolate. Conformation of the tetrahydropyrazine ring and formyl group in solution. Biochemistry 1980;19:4576–82

24.

Garret

, Grisham

. Electron transport and oxidative phosphorylation. In: Biochemistry. 3rd edn. Belmont, CA: Thomson Brooks/Cole, 2005:640–73

25.

Brooks

, Baggott

. Oxidation of 10-formyltetrahydrofolate to formyldihydrofolate by complex IV of rat mitochondria. Biochemistry 2002;41:5633–6

26.

Wall

, Shim

, Benkovic

. Human AICAR transformylase: role of the 4-carboxamide of AICAR in binding and catalysis. Biochemistry 2000;39:11303–11

27.

Caperelli

, Benkovic

, Chettur

, Benkovic

. Purification of a complex catalyzing folate cofactor synthesis and transformylation in de novo purine biosynthesis. J Biol Chem 1980;255:1885–90

28.

Smith

, Mueller

, Wasserman

, Taylor

, Benkovic

. Characterization of the enzyme complex involving the folate-requiring enzymes of de novo purine biosynthesis. Biochemistry 1980;19:4313–21

29.

, Kumar

, Sheets

, Benkovic

. Reversible compartmentalization of de novo purine biosynthetic complexes in living cells. Science 2008;320:103–6

30.

Smoleńska

, Karnowska

, Zarówny

, Simmonds

, Smoleński

. Effect of methotrexate on blood purine and pyrimidine levels in patients with rheumatoid arthritis. Rheumatology 1999;38:997–1002

31.

Fox

, Morgan

, Smith

, Robbins

, Choc

, Baggott

. Combined oral cyclosporin and methotrexate therapy in patients with rheumatoid arthritis elevates methotrexate levels and reduce 7-hydroxymethotrexate levels when compared with methotrexate alone. Rheumatology 2003;42:989–94

32.

Morgan

, Oster

, Lee

, Alarcón

, Baggott

. The effect of folic acid and folinic acid supplements on purine metabolism in methotrexate-treated rheumatoid arthritis. Arthr Rheum 2004;50:3104–11

33.

Allegra

, Drake

, Jolivet

, Chabner

. Inhibition of phosphribosylaminoimidazolecarboxamide transformylase by methotrexate and dihydrofolic acid polyglutamates. Proc Natl Acad Sci USA 1985;82:4881–5

34.

Jarabak

, Bachur

. A soluble dihydrofolate reductase from human placenta: purification and properties. Arch Biochem Biophys 1971;142:417–25

35.

Bunni

, Doig

, Donato

, Kesavan

, Priest

. Role of methylenetetrahydrofolate depletion in methotrexate-mediated intracellular thymidylate synthesis inhibition cultured L1210 cells. Cancer Res 1988;48:3398–404

36.

Dervieux

, Greenstein

, Kremer

. Pharmacogenomic and metabolic biomarkers in the folate pathway and their association with methotrexate effects during dosage escalation in rheumatoid arthritis. Arthr Rheum 2006;54:3095–103

37.

Gorman

, Tamura

, Baggott

. Mass spectrometric method for detecting carbon 13 enrichment introduced by folate coenzymes in uric acid. Anal Biochem 2003;321:188–91

38.

Baggott

, Gorman

, Morgan

, Tamura

. ¹³C-enrichment at carbons 8 and 2 or uric acid after ¹³C-labeled folate dose in man. Biochem Biophys Res Commun 2007;361:307–10

39.

Warren

, Flaks

, Buchanan

. Integration of enzymatic transformylation reaction. J Biol Chem 1957;229:627–40

40.

Flaks

, Warren

, Buchanan

. Further studies of the inosinic acid transformylase system. J Biol Chem 1957;228:215–29

41.

Abrams

. Stability of the adenine ring structure in the rat. Biochim Biophys Acta 1956;21:439–40

42.

Bennett

, Karlsson

. Metabolism of adenine in mice. J Biol Chem 1957;229:39–50

43.

Baggott

, Gorman

, Tamura

. ¹³C Enrichment of carbons 2 and 8 of purine by folate-dependent reactions after [¹³C]formate and [2-¹³C]glycine dosing in adult humans. Metabolism 2007;56:708–15

Evidence for the hypothesis that 10-formyldihydrofolate is the in vivo substrate for aminoimidazolecarboxamide ribotide transformylase

Abstract

Keywords

Introduction

Presence of 10-formyldihydrofolate in vivo

In vitro data supporting our hypothesis

Unique properties of AICAR transformylase

In vitro lability of 10-HCO-H4folate

In vitro enzyme kinetics of AICAR transformylase

In vivo data in humans and animals supporting our hypothesis

Methotrexate therapy of patients with rheumatoid arthritis

13 C enrichment at C2 and C8 of purine after a [6RS]-5-H[13C]O-H4folate dose

Carbon exchange at the second position of the purine ring

Summary

References

In vitro lability of 10-HCO-H₄folate

¹³C enrichment at C₂ and C₈ of purine after a [6RS]-5-H[¹³C]O-H₄folate dose