Synthesis of (6R)- and (6S)-5,10-dideazatetrahydrofolate oligo-γ-glutamates: Kinetics of multiple glutamate ligations catalyzed by folylpoly-γ-glutamate synthetase

Abstract

Folylpoly-γ-glutamate synthetase (FPGS, EC 6.3.2.17) catalyzes the ATP-dependent ligation of glutamic acid to reduced folates including (6S)-5,6,7,8-tetrahydrofolate (H₄PteGlu), as well as to anticancer drugs such as 5,10-dideaza-5,6,7,8-tetrahydrofolate ((6R)-DDAH₄PteGlu₁, (6R)-DDATHF, Lometrexol ™). Synthesis of unlabeled mono- and polyglutamates, DDAH₄PteGlu_n (6R, n = 1–6; 6S, n = 1–2), as well as (6R)-DDAH₄Pte[¹⁴C]Glu₁, was effected from (6R)- or (6S)-5,10-dideazatetrahydropteroyl azide and glutamic acid, H-Glu-γ-Glu_n-γ-Glu-OH (n = 0–4), or [¹⁴C]glutamic acid, respectively. These compounds were evaluated as FPGS substrates to determine steady-state kinetic constants. Michaelis–Menten kinetics were observed for (6R)-DDAH₄PteGlu₁, the isomer corresponding to H₄PteGlu, whereas marked substrate inhibition was observed for (6S)-DDAH₄PteGlu_n (n = 1–2) and (6R)-DDAH₄PteGlu_n (n = 2–5), but not (6R)-DDAH₄PteGlu₆. Multiple ligation of glutamate renders a quantitative analysis of these data difficult. However, approximate values of K_M = 0.65–1.6 µM and K_I = 144 –417 µM for DDAH₄PteGlu_n were obtained using a simple kinetic model.

Introduction

Folylpoly-γ-glutamate synthetase (FPGS, EC 6.3.2.17) is the enzyme responsible for the intracellular poly-γ-glutamylation of natural folates as well as many antifolates (Scheme 1). This poly-γ-glutamylation leads to increased intracellular retention and (often) affinity for the target enzymes.^1–3 The antifolate (6R)-5,10-dideaza-5,6,7,8-tetrahydrofolate ((6R)-DDAH₄PteGlu₁, (6R)-DDATHF, LY264618, Lometrexol™) has been evaluated in clinical trials as a treatment for breast and non-small cell lung cancers as well as melanoma, sarcoma, and others.^4–7 DDAH₄PteGlu₁ is an inhibitor of de novo purine biosynthesis that specifically inhibits the reaction catalyzed by glycinamide ribonucleotide formyltransferase (GARFT, EC 2.1.2.1).⁸ The (6R)- and (6S)-diastereomers of DDAH₄PteGlu₁ are equally effective at inhibiting the growth of cancer cell lines in vitro and tumors in vivo.^9,10 The finding that the antiproliferative effect of DDAH₄PteGlu₁ is due to inhibition of GARFT sparked much interest in this compound as a drug candidate.^4,8 In contrast, two well known antifolates, aminopterin and methotrexate, both act by inhibition of dihydrofolate reductase (DHFR, EC 1.5.1.3).

Scheme 1.

The reaction of FPGS with (6R)-DDAH₄PteGlu₁ (1). The structure of (6S)-tetrahydrofolic acid is shown for comparison.

Early work with DDAH₄PteGlu₁ indicated that cells treated with (6R)-DDAH₄PteGlu₁ rapidly convert the drug to its pentaglutamate and hexaglutamatemetabolites.¹¹ Rapid conversion to polyglutamate forms has also been observed with purified hFPGS¹² and in the mouse.¹³ It is an excellent substrate for purified mammalian FPGS with kinetic constants similar to the best natural substrates, 10-formyltetrahydrofolate (10-formyl-H₄PteGlu) and tetrahydrofolate (H₄PteGlu).^8,10,14,15 The (6R)- and (6S)-diastereomers of DDAH₄PteGlu₁ have K_M values with human FPGS of 1.7 and 1.0 µM respectively. Substrate inhibition was observed, but only by the (6S)-diastereomer.¹⁰ It was also found that the poly-γ-glutamate forms are much more potent inhibitors of GARFT, with K_I values for DDAH₄PteGlu₁ and DDAH₄PteGlu₅ being 39 nM and 0.39 nM, respectively.¹⁶ This increase in potency for the target enzyme upon polyglutamylation has been observed for many anti-folate compounds.¹⁷ Resistance of cancer cells to DDAH₄PteGlu₁ has been observed to arise from a decrease in polyglutamylation.^11,18,19 These data strongly suggest that the polyglutamylated form of DDAH₄PteGlu₁ is responsible for its cytotoxic activity.

DDAH₄PteGlu₁ was first synthesized as a mixture of diastereomers about the C-6 position as reported by Taylor et al.¹⁴ Since that time a number of other racemic syntheses have been reported.^20,21 Access to the individual diastereomers of DDAH₄PteGlu₁ was originally obtained by fractional crystallization or chiral HPLC of DDAH₄PteGlu₁ diethyl ester, d-10-camphorsulfonic acid salt, followed by careful saponification.^10,22 An asymmetric synthesis of (6R)-DDAH₄PteGlu₁, the diastereomer selected for clinical trials, has been reported.²³

In this paper, we report the syntheses of (6R)-DDAH₄PteGlu_1–6, (6S)-DDAH₄PteGlu_1–2, and (6R)-DDAH₄Pte[¹⁴C]Glu₁ via a 5,10-dideaza-5,6,7,8-tetrahydropteroyl azide (DDAH₄Pte–N₃) intermediate that allows for the incorporation of completely unprotected l-glutamate, poly-γ-glutamate peptides, or l-[U–¹⁴C]glutamate, respectively, as the final step in the synthesis.^24,25 The (6R)- and (6S)-isomers of DDAH₄Pte–N₃ were derived from the individual enantiomers of methyl 2-pivaloyl-5,10-dideaza-5,6,7,8-tetrahydropteroate, which, in turn, were purified from the racemic mixture²⁶ by chiral HPLC. The only previously reported synthesis of radiolabeled (6R)-DDAH₄PteGlu₁ involved a multi-step procedure in which the unstable isotope (³H) was incorporated to form a racemic precursor midway through the synthesis.²⁷ Resolution of racemic [³H]DDAH₄PteGlu₁ was required to obtain the desired single diastereomer for use in metabolic studies.¹¹ In contrast, in the synthesis described herein, the ¹⁴C label is introduced into a single isomer, via (6R)-DDAH₄Pte–N₃, in the final step.

The full range of poly-γ-glutamate forms of individual folates and antifolates have been synthesized previously by both solid-phase^28–30 and solution-phase^29,31–35 chemistries. In this research, the poly-γ-glutamate peptides were synthesized by solid-p ase peptide synthesis utilizing l-glutamate protected as the N-Fmoc and α- or γ-tert-butyl ester. The procedure included a capping step that prevented further elongation of any unreacted N-terminal amines. The biochemical evaluation of these compounds as substrates for hFPGS was then pursued. We report steady-state kinetics data (K_M, V_max, V/K) for all newly synthesized polyglutamates, and also document marked substrate inhibition by all compounds with the exception of (6R)-DDAH₄PteGlu₁ and (6R)-DDAH₄PteGlu₆.

Results and discussion

Chemistry

Synthesis of (6RS)-5,10-dideaza-5,6,7,8-tetrahydrofolic acid (6RS-1) was achieved via the route illustrated in Scheme 2. This route is based on the work of Taylor et al.,²⁶ but is modified to allow the incorporation of unprotected glutamate, radiolabeled glutamate, or poly-γ-glutamate peptides as the final synthetic step in the synthesis. This was achieved through the use of an acyl azide, 9, derived from (6RS)-methyl 2-pivaloyl-5,10-dideaza-5,6,7,8-tetrahydropteroate (6RS-7). The racemic ester, 6RS-7, was separated into its individual enantiomers by chiral HPLC, enabling ready access to either the (6R)- or (6S)-isomer of DDAH₄PteGlu₁ and DDAH₄Pte polyglutamates.

Scheme 2.

Synthesis of DDAH₄PteGlu₁ (1) from protected DDAH₄Pte–OH (7).

Deprotection of 7 was achieved by basic hydrolysis in 1 M NaOH_aq. Isolation of 5,10-dideaza-5,6,7,8-tetrahydropteroic acid (8) involved acidification of the basic product solution with HCl to precipitate the insoluble salt, which was then collected by centrifugation. 5,10-Dideaza-5,6,7,8-tetrahydropteroyl azide (9) was formed by the treatment of 8 with ethyl chloroformate in the presence of triethylamine in DMSO to produce the mixed anhydride, followed by displacement with azide anion. The product was then recovered by precipitation with aqueous HCl followed by centrifugation. HPLC analysis of the reaction mixture prior towork-up consistently indicated 95–100% conversion to product. However, during the acidic aqueous workup, a significant portion of the product (5–10%) was hydrolyzed back to 8. Since other work-up options provided no improvement in product purity and 8 would not interfere in the following steps, 9 contaminated with a small amount of 8 was carried through the subsequent transformations.

Initially, the synthesis of 9 was attempted by treating 8 with diphenylphosphoryl azide, as has been previously reported for similar antifolates.²⁵ Unfortunately, this reaction proceeded very slowly, with less than 50% conversion after prolonged reaction (9 days) based on HPLC analysis, in contrast to 85% conversion of 4-deoxy-4-amino-10-methylpteroic acid to the corresponding azide over a much shorter reaction time (2 days). This marked difference in reaction rate is presumably due to decreased nucleophilicity of the conjugate base of the carboxylic acid 8 (pK_a = ca. 4.3) vs. hat of the corresponding pteroic acid analogue (pK_a = ca. 5.0).³⁶ Therefore, formation of a mixed anhydride followed by reaction with azide ion³⁷ was investigated and proved to be effective in converting 8 to 9 in nearly quantitative yield. The coupling reactions between 9 and l-glutamate or poly-γ-glutamates to produce 5,10-dideaza-5,6,7,8-tetrahydrofolic acid (1) or its polyglutamates (2 through 6) were performed in DMSO in the presence of Et₃N (vide infra). The optimal reaction conditions consisted of two equivalents of 9 per equivalent of amino acid or peptide with a reaction time of two days at room temperature. Any of the parent acid, 8, formed from excess 9 due to the presence of adventitious water in the reaction mixture, or during workup, could be recovered.

The poly-γ-glutamates required for the preparation of DDAH₄PteGlu_n were synthesized by standard solid-phase Fmoc-peptide synthesis procedures.³⁸ Two differently protected Fmoc-l-glutamic acid building blocks were utilized in the solid-phase synthesis. N-α-Fmoc-l-glutamic acid α-tert-butyl ester (Fmoc-Glu-(OH)Ot-Bu) was required to obtain the desired γ-linkages, while N-α-Fmoc-l-glutamic acid γ-tert-butyl ester (Fmoc-Glu(Ot-Bu)–OH) was used to load the p-benzyloxybenzyl alcohol (Wang) resin (Scheme 3). Acetic anhydride and dimethylaminopyridine (DMAP) in DMF were used in a capping step to acylate any free N-termini remaining after each peptide coupling step, thereby preventing further elongation of any unreacted peptide. Fmoc-Glu(Ot-Bu)–OH was chosen for resin loading because it has been reported that the γ-linkage to the resin is more labile than the standard α-linkage.²⁸ Also, as this is the least efficient step (40–50% yield), the fact that the Fmoc-Glu(Ot-Bu)–OH is five times less costly than Fmoc-Glu-Ot-Bu makes its use in this reaction more attractive. This method allows for the facile synthesis of any γ-glutamyl peptide on a 100 mg scale in a few days using manual methods. Purification is achieved by reversed-phase, semi-preparative HPLC followed by lyophilization to yield the trifluoroacetate salt of the peptide. Purified γ-glutamyl peptides were coupled to 5,10-dideaza-5,6,7,8-tetrahydropteroyl azide (9) under the optimized conditions described above (Scheme 4).

Scheme 3.

Solid-phase synthesis of poly-γ-glutamate peptides.

Scheme 4.

Synthesis of DDAH₄Pte polyglutamates (2–6) from DDAH₄Pte–N₃ (9). Synthesis of 6R-2 and 6S-2 utilized crude H-Glu-γ-Glu-OH peptide in the coupling reaction. All product yields are determined by UV/vis spectroscopy.

Biochemical studies

The biochemical studies utilized an end-point assay that quantitated the total amount of L-[³H]glutamic acid ligated to the folate substrate by FPGS. Assays were performed at fixed, saturating concentrations of ATP (5 mM, ca. 100 × K_M) and l-glutamic acid (2 mM, ca. 10 × K_M) with varying concentrations of the DDAH₄PteGlu_n being characterized. Both diastereomers of DDAH₄PteGlu₁ exhibited Michaelis–Menten kinetics at folate concentrations up to at least ∼15 × K_M (Fig. 1A). The kinetic constants that were obtained (Table 1) agree with the reported values of 1.7 and 1.0 µM for the (6R)- and (6S)-diastereomers, respectively.¹⁰

Fig. 1.

Steady-state kinetics: hFPGS-catalyzed ligation of l-glutamate to each of the C-6 stereoisomers of DDAH₄PteGlu₁, (6R)- (●), and (6S)- (○). hFPGS was expressed in E. coli and partially purified. Each assay was done in duplicate and the data points shown are the average values ± range. A. Curves are the best fit of the data in the non-inhibitory concentration range to the Michaelis–Menten equation. B. Curves are the best fit of all data to the Michaelis–Menten equation (6R), or eqn (2) (6S).

Table 1.

Kinetic constants of the C-6 stereoisomers of DDAH₄PteGlu₁

	Michaelis–Menten fita		Substrate inhibition fitb
	Vmax/µM h−1	KM/µM	Vmax/µM h−1	KM/µM	KI/µM
6R	3.15 ± 0.11	1.93 ± 0.23	NAc	NAc	NA
6S	3.31 ± 0.15	1.06 ± 0.20	4.01 ± 0.20	1.56 ± 0.23	76 ± 10

^aValues obtained using the Michaelis–Menten equation to fit the data in Fig. 1A.

^bValues obtained using eqn (2) to fit the (6S)-isomer data in Fig. 1B.

^cK_M = 3.67 ± 0.55 µM and V_max = 3.94 ± 0.10 µM h⁻¹ were obtained when the Michaelis–Menten equation was used to fit the data for the (6R)-isomers in Fig. 1B.

However, as is apparent from Fig. 1B, the data for the (6R)-diastereomer at concentrations >30 µM deviate significantly from the best fit to the Michaelis–Menten equation for a single E·S complex. A model which invokes the formation of two different E·S complexes (eqn (1)) yielded a better fit to the entire range of data for the (6R)-diastereomer (Fig. 1B) and resulted in values of K_M1 and K_M2 that differed by two orders of magnitude (K_M1 ≪ K_M2, ca. 1 and 100 µM respectively). However, the value

$$\upsilon =\frac{V_{\text{max}}\left[S\right]}{K_{M1}+\left[S\right]}+\frac{V_{\text{max}}\left[S\right]}{K_{M2}+\left[S\right]}$$ (1)

of K_M2 differs markedly from the steady state K_M value determined for the (6R)-DDAH₄Pte–Glu₂ species (vide infra), the presumed substrate involved in the second E·S complex. This model does not appear to be adequate to explain the deviation from Michaelis–Menten kinetics and was not pursued further.

At higher concentrations of (6S)-DDAH₄PteGlu₁ (30–300 µM), substrate inhibition was noted (Fig. 1B). This is consistent with what is known in the literature.^8,10 In order to obtain K_M, K_I, and V_max values from these data, a substrate inhibition equation for a simple system, eqn (2), was used.³⁹

$$\upsilon =\frac{V_{\text{max}}}{1+\frac{K_M}{\left[S\right]}+\frac{\left[S\right]}{K_1}}$$ (2)

It was expected that the (6R)- and (6S)-diastereomers of the poly-γ-glutamate conjugates would exhibit similar properties as was observed for the respective monoglutamate diastereomers in terms of substrate inhibition. However, as can be seen in Fig. 2, both the (6R)- and (6S)-isomers of the diglutamate, DDAH₄PteGlu₂, exhibit striking substrate inhibition that occurred at much lower relative concentrations (∼5 × K_M) than that observed for (6S)-DDAH₄PteGlu₁. Due to the low concentration at which the substrate inhibition became significant, the Michaelis–Menten equation was used for fitting only the data in the non-inhibitory region (≤5 µM). The kinetic constants for the diastereomers of DDAH₄PteGlu₂ are summarized in Table 2. The observed substrate inhibition could be associated with either non-sequential kinetics and/or partial sites reactivity in an oligomeric protein. However, there is no supporting evidence for either of these scenarios for FPGS.

Fig. 2.

Steady-state kinetics: hFPGS-catalyzed ligation of l-glutamate to each of the C-6 stereoisomers of DDAH₄PteGlu₂, (6R)- (●), and (6S)- (○). hFPGS was expressed in E. coli and partially purified. Each assay was done in duplicate and the data points shown are the average values ± range. Curves shown are the best fit of all data to eqn (2). Data in the non-inhibitory range were fit to the Michaelis–Menten equation.

Table 2.

Kinetic constants of the C-6 stereoisomers of DDAH₄PteGlu₂

	Michaelis–Menten fita		Substrate inhibition fitb
	Vmax/µM h−1	KM/µM	Vmax/µM h−1	KM/µM	KI/µM
6R	1.84 ± 0.24	0.87 ± 0.33	1.64 ± 0.11	0.65 ± 0.17	157 ± 25
6S	2.83 ± 0.29	1.16 ± 0.35	2.65 ± 0.15	0.95 ± 0.20	92 ± 20

^aValues obtained using the Michaelis–Menten equation to analyze the data in the non-inhibitory region (≤5µM). Similar values were obtained when the data were analyzed utilizing the Hanes linear transform.⁴⁰

^bValues obtained using eqn (2).

This substrate inhibition may be explained by two molecules of substrate binding to a single hFPGS simultaneously. The first molecule of substrate would bind in the active site correctly oriented for turnover, while the second molecule would bind to another site that inhibits catalysis. This second site could either be an allosteric binding site involved in feedback inhibition or an incorrect binding of the second folate substrate molecule in the active site. Based on the crystal structure of a binary complex (L. casei FPGS·ATP), the C-terminal domain of FPGS has a very similar secondary structure to that of DHFR from several species,⁴¹ indicating a possible role in folate binding. Preliminary docking studies performed based on the crystal structure of a ternary complex (L. casei FPGS·AMPPCPP·5,10-CH₂-H₄PteGlu₁)⁴² indicates the possible presence of a second folate substrate binding site on the N-terminal domain of the protein (Dr Heather Carlson, University of Michigan, personal communication). Recent work by Mathieu et al. supports this finding. A second folate substrate binding site with dihydropteroate bound was observed when a structure of the bifunctional E. coli FolC enzyme, harboring dihydrofolate synthetase and folylpolyglutamate synthetase activities, was determined.⁴³

The trend of substrate inhibition carried through the entire series of (6R)-DDAH₄PteGlu_n (n = 3–5, Fig. 3). Of the polyglutamate compounds, only the hexaglutamate exhibited Michaelis–Menten behaviour. Table 3 lists the kinetic constants obtained from the (6R)-DDAH₄PteGlu_n compounds.⁴⁴ As glutamate chain-length increased (n = 1–5), the V/K values first increased approximately 2.5-fold when going from the mono- to diglutamate species. This difference arises from a much lower K_M value for the latter. The tri-, tetra-, and pentaglutamates exhibited much lower, but similar, V/K values than those observed for the shorter chain length substrates due to much reduced V_max values. These data follow the same general pattern reported by Chen et al. for pteroyl and (6S)-tetrahydropteroyl poly-γ-glutamate_n (n = 1–5).⁴⁵ A similar trend has been reported for FPGS from L. casei, an enzyme for which a crystal structure is available.⁴² Analysis of this structure indicates that the γ-carboxyl group of the monoglutamate substrate does not reach the ATP and is therefore incapable of forming the γ-glutamyl phosphate⁴⁶ without significant, and energetically costly, conformational changes. However, the diglutamate is perfectly accommodated in the active site.⁴⁷ There is no indication of how longer chain length products could be accommodated within the active site. In the current work, a very low value of V/K for (6R)-DDAH₄PteGlu₆ was determined, due primarily to a significantly increased value of K_M.

Fig. 3.

Steady-state kinetics: hFPGS-catalyzed ligation of l-glutamate to (6R)-DDAH₄PteGlu_n (n = 3–6). hFPGS was expressed in bacculovirus-infected SF9 insect cells and partially purified. Each assay was done in duplicate and the data points shown are the average values ± range. n = 3 (■), 4 (▲), 5 (□), and 6 (△). Curves are the best fit of all data to the Michaelis–Menten equation (n = 6) or eqn (2) (n = 3–5).

Table 3.

Kinetic constants of (6R)-DDAH₄PteGlu_n^a

n	Vmax/µM h−1	KM/µM	KI/µM	(V/K)/h−1
1	3.94 ± 0.10	3.67 ± 0.55	NA	1.07 ± 0.16
2	1.64 ± 0.08	0.65 ± 0.12	157 ± 25	2.52 ± 0.48
3	0.30 ± 0.03	0.96 ± 0.43	179 ± 91	0.31 ± 0.14
4	0.71 ± 0.07	1.60 ± 0.55	144 ± 56	0.44 ± 0.16
5	0.22 ± 0.01	1.05 ± 0.24	417 ± 152	0.21 ± 0.05
6	0.59 ± 0.02	14.5 ± 1.9	NA	0.04 ± 0.01

^aValues obtained using the Michaelis–Menten equation (n = 1 and 6) or eqn (2) (n = 2–5).

One caveat in these analyses is that multiple glutamate ligations are occurring during the reactions, i.e., the di-, tri-, tetra-, etc. glutamate products are formed since each product is also a substrate for further ligation. This complexity makes the following assumptions in the standard Michaelis–Menten analysis invalid: (1) only a single substrate and single enzyme–substrate complex are involved, (2) the enzyme–substrate complex breaks down directly to form free enzyme and product, and (3) enzyme, substrate, and enzyme–substrate complex are at equilibrium, i.e., the dissociation of ES to E + S is rapid in comparison to the formation of E + P.⁴⁸

To establish that multiple glutamate ligations were occurring to a single folate substrate, product distribution analysis was done at varied (6R)-DDAH₄Pte[¹⁴C]Glu₁ concentrations. When these reactions were analyzed, it was determined that multiple products were produced and that the product distribution varied with changing concentrations of (6R)-DDAH₄PteGlu₁ (Fig. 4). The observed trend is that progressively shorter chain length products are produced with increasing concentrations of (6R)-DDAH₄PteGlu₁. The results of this experiment are very similar to those observed for rat liver FPGS with (6RS)-H₄PteGlu as the substrate.⁴⁹ In that case, multiple glutamate ligations to (6RS)-H₄PteGlu at 5 µM (∼K_M) yielded H₄PteGlu_4–5 as the predominant products, following initial formation of H₄PteGlu₂ and H₄PteGlu₃ that built up to a low level and subsequently decreased. In contrast, (6RS)-H₄PteGlu at 35 µM (∼10 × K_M) yielded predominately H₄PteGlu₂ and H₄PteGlu₃; H₄PteGlu_4–5 products were only observed at longer reaction times.

Fig. 4.

Product distribution of reactions of (6R)-DDAH₄Pte[¹⁴C]Glu₁ with purified hFPGS expressed in bacculovirus-infected SF9 insect cells. The concentrations of the glutamate and ATP substrates were 5mM and 10 mM respectively. Reactions contained the following concentrations of (6R)-DDAH₄Pte[¹⁴C]Glu₁: A. 2.5 µM. B. 25 µM. C. 250 µM.

The product distribution results at a low concentration, 2.5 µM, agree well with what has been observed in whole CCRF-CEM cells¹¹ and in the livers of mice¹³ that were treated with (6R)-DDAH₄PteGlu₁. In cells treated with 10 µM (6R)-DDAH₄PteGlu₁, polyglutamate products (n = 2–6) were observed even at the earliest time point, 4 h, with the pentaglutamate product predominating at longer incubation times. Analysis of the livers of mice dosed with 0.72 mg kg⁻¹ of (6R)-DDAH₄PteGlu₁ indicated that the pentaglutamate product was formed relatively quickly with no detectable diglutamate product and smaller amounts of tri-, tetra- and hexaglutamate products.

In an attempt to explain these product distributions, a kinetic modeling study was undertaken. This study utilized the K_M and V_max values listed in Table 3 and a minimum mechanism based on sequential elongation of the substrate. Product release between each catalytic step was assumed, i.e., a distributive mechanism, so that the Michaelis–Menten kinetic equation could be utilized to analyze these data. Substrate inhibition by the polyglutamates was ignored because of the low concentrations of products formed in the reactions. Modeling of this distributive mechanism, while able to produce different product distributions with differing initial monoglutamate substrate concentrations, was unable to mimic the product distributions actually observed. A model in which product release was not mandated between each catalytic step, i.e. a processive mechanism, could not be readily developed. However, a processive mechanism was assessed directly by subsequent substrate-trapping and pulse-chase experiments, the results of which will be published separately.

Conclusions

In this paper, we report the synthesis of the antifolate drug Lometrexol™ ((6R)-DDAH₄PteGlu₁), its poly-γ-glutamates ((6R)-DDAH₄PteGlu_2–6), and a radiolabeled form ((6R)-DDAH₄Pte[¹⁴C]Glu₁). In addition, the (6S)-isomers, (6S)-DDAH₄PteGlu₁ and (6S)-DDAH₄PteGlu₂, were synthesized. To obtain the single isomer materials, a chiral HPLC separation of (6RS)-methyl 2-pivaloyl-5,10-dideazatetrahydropteroate (6RS-7) was developed. This separation was achieved on a preparative scale (∼5 g) and resulted in material with an isomeric purity of >98%. The isomerically pure materials were subsequently used in the preparation of all DDAH₄PteGlu_n compounds as well as (6R)-DDAH₄Pte[¹⁴C]Glu₁. These compounds were then evaluated as substrates for recombinant human cytosolic FPGS, and the kinetic data augmented by reaction product analysis.

The observed biochemical properties of the (6R)- and (6S)-diastereomers of DDAH₄PteGlu₁, including the segregation of substrate inhibition to the (6S)-diastereomer, are comparable to that previously reported in the literature. The properties of the single diastereomers of DDAH₄PteGlu₂ are reported for the first time and show no such separation of the substrate inhibition. Indeed, all of the (6R)-DDAH₄PteGlu_2–5 compounds show substrate inhibition. (6R)-DDAH₄PteGlu₆ is the only polyglutamate substrate not to exhibit substrate inhibition.

When the kinetic constants for the (6R)-DDAH₄PteGlu_n compounds are compared (Table 3), it is seen that the diglutamate has the highest V/K value. The value for the diglutamate is over two-fold greater than that obtained for monoglutamate indicating that the diglutamate is the preferred substrate for hFPGS. This is primarily due to the much lower K_M for the diglutamate vs. the monoglutamate which overcomes the lower V_max of the former. The rest of the poly-γ-glutamates, tri- through hexaglutamate, exhibit V/K values of less than half that of the monoglutamate, indicating that they are much poorer substrates than the mono- and diglutamate compounds.

It is also apparent that the product distributions obtained utilizing (6R)-DDAH₄Pte[¹⁴C]Glu₁ exhibit a significant dependence on substrate concentration (Fig. 4). The experimentally observed product distributions are similar to those observed in cells and whole animals. This similarity of product distributions (in vitro vs. in vivo) provides strong support for the validity of applying conclusions obtained from this work to what may be occurring in vivo.

Experimental

Materials

Common solvents and reagents were obtained from commercial sources and were of the highest purity available. l-[2,3-³H]-Glutamic acid (NET-395) and l-[U-¹⁴C]glutamic acid (NEC-290E) were obtained from Perkin–Elmer Life Sciences. p-Benzyloxybenzyl alcohol (Wang) resin, N-α-Fmoc-l-glutamic acid α-tert-butyl ester (Fmoc-Glu(OH)–Ot-Bu), N-α-Fmoc-l-glutamic acid γ-tert-butyl ester (Fmoc-Glu(Ot-Bu)–OH), N-hydroxybenzotriazole–H₂O (HOBt), benzotriazole-l-yloxytris-(pyrrolidino)phosphonium hexafluorophosphate (PyBOP), and 2-(1H-benzotriazole-l-yl)-1,1,3,3-tetramethyluronium hexafluorophosphate (HBTU) utilized for solid phase peptide synthesis were obtained from NovaBiochem. CDCl₃, D₂O, d-TFA, and NaOD were from Isotec and d₆-DMSO was from Cambridge. (6R)-5,10-Dideaza-5,6,7,8-tetrahydrofolic acid sodium ((6R)-DDAH₄PteGlu₁–Na, Lometrexol™ sodium, Lot 235MH8) and (6RS)-5,10-dideaza-5,6,7,8-tetrahydropteroyl polyγ-glutamates (DDAH₄PteGlu₂, Lot 266251 and DDAH₄PteGlu₅, Lot 266578) were kind gifts of Dr Chuan Shih, Eli Lilly & Co, Indianapolis, IN. These compounds were used only in initial studies on assay development and for comparison of physical properties with newly synthesized compounds. (6RS)-Methyl 2-pivaloyl-5,10-dideazatetrahydropteroate (6RS-7) was synthesized as described by Taylor²⁶ with significant procedural modifications. Details may be found in the Supplementary Information. Spin-X HPLC 0.2 µm nylon microcentrifuge filters (spin filters) were obtained from Costar. Human folylpolyglutamate synthetase was expressed in E. coli or bacculovirus-infected SF9 insect cells^12,45 and partially purified as described previously.⁵⁰ Homogenous hFPGS, expressed in bacculovirus-infected SF9 insect cells and purified as described previously,¹² was a generous gift of Dr Richard Moran. hFPGS concentration was determined by the method of Bradford⁵¹ corrected for protein purity as determined by quantitative PAGE.

General procedures

All synthetic reactions were carried out in oven-dried glassware. Moisture-sensitive reactions were carried out under a dry atmosphere of nitrogen or argon. Solvents and liquid reagents were measured and dispensed in oven-dried glass syringes or dry nitrogen-flushed disposable syringes with oven-dried needles. Et₃N, pyridine, MeOH, and CH₃CN were distilled from CaH₂. THF was distilled from benzophenone and sodium. ¹H and ¹³C NMR spectra were recorded on a Bruker 300 or 500 MHz instrument using either TMS or solvent peaks as an internal reference. When an internal reference was not available (e.g. a ¹³C NMR spectrum in D₂O) the frequency of the solvent deuterium was utilized. ¹H NMR assignments of diastereotopic protons are noted as, e.g. “1.86 (1 H, m, C(8)H₂), 2.07 (1 H, m, C(8)H₂)” and signify that each of the two methylene protons attached to carbon 8 have unique chemical shifts of 1.86 and 2.07. All chemical shifts (δ) are reported in ppm. Assignments for nuclei in the γ-glutamyl peptide portion use the prime notation (e.g., C(9′)) to distinguish them from nuclei in the didezatetrahydropteroyl portion (Supplementary Information, Schemes S2 and S3). Assignment of peaks as indicated was aided and confirmed by the acquisition of various 2D spectra (COSY, HETCOR, COLOC). Nominal and high resolution mass spectra utilizing electrospray ionization with positive detection utilized Na⁺ and/or H⁺ matrices. Nominal mass spectral data are reported only for those fragments with a relative intensity of ≥20% of the base peak. UV/visible spectra were recorded on a Beckman Model 640B spectrophotometer. Melting points were determined with a MEL-TEMP apparatus and are reported uncorrected. Scintillation counting was carried out with a Packard 1600 scintillation counter using Bio-Safe II scintillation cocktail from Research Products Inc. and 20 mL glass scintillation vials with foil-lined lids or 5.5 mL polypropylene multi vials from Life Science Products. The analytical HPLC system used was manufactured by Varian and consisted of a Model 410 auto-sampler, two Model 210 Prostar pumps, a Model 330 photo diode array detector, and a Model 701 fraction collector. The semi-preparative HPLC system was manufactured by Rainin and consisted of two Rabbit Model HPX pumps fitted with 50 mL min⁻¹ pump heads, a dynamic mixer, and a single wavelength UV detector. Varian Star v6.3 software controlled both systems. All HPLC was run at ambient temperature. Optical rotation was determined with a Perkin–Elmer Model 241 polarimeter using the sodium D line source (589 nm) with an integration time of 5 s, at ambient temperature, a cell length of 10.002 cm, and with a solution concentration of 10 mg mL⁻¹ unless otherwise noted. [α]_d values are given in 10⁻¹ deg cm² g⁻¹ and are uncorrected. All kinetics data were fitted using KaleidaGraph v3.5, Synergy Software (Reading, PA).

Chromatography methods

Size-exclusion chromatography

A 1.6 × 80 cm column was packed with Bio-Gel P-2 (Bio-Rad, fine mesh) that had been previously swelled and equilibrated with aqueous 50 mM ammonium bicarbonate, pH 8.1. Flow through the column was gravity controlled with a 10 cm column head whose level was maintained with a siphoning system. Fractions (150-drop, ∼3.4 mL) of column eluant were collected beginning with column loading. Elution of products was monitored by UV/visible absorbance, scintillation counting, and/or ninhydrin staining.

Semi-preparative reversed-phase (RP)–HPLC

All separations utilized a Varian Dynamax 21.4 × 250 mm, Microsorb 60–8, C18 column with a flow rate of 21 mL min⁻¹. Methods 1–4 utilized Eluant A consisting of 0.1% w/v TFA in ddH₂O, pH 1.8. Eluant B consisting of 0.1% w/v TFA in CH₃CN, and detection at 214 nm with gradient profiles as follows:

• Method 1: 0 min, 1% B; 2 min, 1% B; 15 min, 10% B.

• Method 2: 0 min, 1% B; 5 min, 1% B; 15 min, 10% B; 20 min, 100% B.

• Method 3: 0 min, 5% B; 5 min, 5% B; 20 min, 15% B; 25 min, 50% B.

• Method 4: 0 min, 2% B; 2 min, 2% B; 20 min, 15% B; 25 min, 50% B.

  Methods 5–8 utilized an Eluant A consisting of 20 mM ammonium acetate, pH 6.5, in ddH₂O, an Eluant B consisting of MeOH, and detection at 280 nm with gradient profiles as follows:

• Method 5: 0 min, 5% B; 25 min, 20% B; 30 min, 50% B; 35 min, 50% B.

• Method 6: 0 min, 2% B; 25 min, 20% B; 30 min, 50% B; 35 min, 50% B.

• Method 7: 0 min, 2% B; 2 min, 2% B; 25 min, 15% B; 30 min, 50% B; 35 min, 50% B.

• Method 8: 0 min, 2% B; 2 min, 2% B; 25 min, 10% B; 30 min, 50% B; 35 min, 50% B.

Analytical RP-HPLC

• Method 1:²⁴ Eluant A: 50 mM sodium phosphate buffer, pH 7.0. Eluant B: CH₃CN. Column: Chrompack Kromasil 100 Å, C18, 4.6 × 250 mm. Flow rate: 0.7 mL min⁻¹. Gradient: 0 min, 2% B; 25 min, 50% B; 35 min, 50%B. Monitored at 280 nm or scintillation counting of 30 s fractions.

• Method 2: Eluant A: 0.1% w/v TFA in ddH₂O, pH 1.8. Eluant B: 0.1% w/v TFA in CH₃CN. Column: Vydac 218TP54, 5 µm, C18 protein and peptide, 4.6 × 250 mm. Flow rate: 1.0 mL min⁻¹. Isocratic at 13% B. Monitored at 280 nm.

• Method 3: Eluant A: ddH₂O. Eluant B: CH₃CN. Column: Vydac 218TP54, 5 µm, C18 protein and peptide, 4.6 × 250 mm. Flow rate: 1.0 mL min⁻¹. Gradient: 0 min, 30% B; 1 min, 30% B; 20 min, 60% B; 25 min, 100% B; 40 min, 100% B. Monitored at 280 nm.

Ion-pair HPLC

Eluant A: 100% ddH₂O. Eluant B: 65% CH₃CN, 35% ddH₂O. Both eluants A and B contain 1 mM KH₂PO₄, 5 mM tetrabutylammonium phosphate (TBAP), 7 mM NaCl, and 3 mM NaN₃. Column: Vydac 218TP54, 5 µm, C18 protein and peptide, 4.6 × 250 mm. Flow rate: 1.0 mL min⁻¹. Gradient: 0 min, 10% B; 5 min, 10% B; 10 min, 36% B; 20 min, 40% B; 50 min, 55% B; 53 min, 100% B, 58 min, 100% B. Monitored at 280 nm or by the scintillation counting of fractions. This method was developed from a previously published procedure.⁵²

Ion-pairing buffers for HPLC were prepared as follows. 1 L of a 10 × stock solution was prepared and consisted of 10 mM KH₂PO₄, 50 mM TBAP, 70 mM NaCl, and 30 mM NaN₃ in ddH₂O. For both eluants A and B, 100 mL of the 10 × stock solution was measured using a graduated cylinder and added to a 1 L volumetric flask. The graduated cylinder was rinsed two times with 100 mL ddH₂O and once with 50 mL ddH₂O. The eluants A and B were brought up to 1 L with ddH₂O and CH₃CN respectively and filtered through a 0.2 µm nylon filter.

Equilibration of the HPLC column was achieved in two steps. The first step was preparation of the column for the ion-pairing buffers and consists of washing the column at 1 mL min⁻¹ with ddH₂O (A) and CH₃CN (B). Gradient: 0 min, 100% B; 5 min, 100% B; 8 min, 50% B; 10 min, 50% B. The flow was stopped and the pumps were switched to ion-pair containing eluants A and B. The flow was resumed at 1 mL min⁻¹ and 100% B until the operating pressure had been reached. Gradient: 0 min, 100% B; 10 min, 100% B; 15 min, 10% B; 30 min, 10% B.

Chiral-HPLC (analytical)

Eluant A : ddH₂O. Eluant B: CH₃CN. Column: Chiral Technologies (DAICEL Chemical Industries, LTD) Chiralcel OJ–R, 4.6 × 150 mm. Isocratic at 40% B. Flow rate: 0.5 mL min⁻¹. Monitored at 280 nm.

Chiral-HPLC (preparative)

Eluant A: hexane. Eluant B: EtOH. Column: Chiralcel OJ, 2 cm × 25 cm. Isocratic at 70% B. Flow rate: 30 mL min⁻¹. Run time: 22 min. Sample concentration: 1.25 mg mL⁻¹ in 25% A: 75% B. Monitored at 240 nm.

(6R)- and (6S)-Methyl 2-pivaloyl-5,10-dideaza-5,6,7,8-tetrahydropteroate (6R-7 and 6S-7)

(6RS)-Methyl 2-pivaloyl-5,10-dideaza-5,6,7,8-tetrahydropteroate (6RS-7, 4.75 g) was resolved into its individual isomers by preparative chiral-HPLC as described above (see Supplementary Information, Figure S1). This separation was done at the Analytical Development Department, Pfizer Global Research and Development, Ann Arbor, MI. The (6R)-isomer (6R-7) (1.6 g) was obtained in 98.2% chiral purity (t_r = 15.2 min) as determined by analytical chiral HPLC (see Supplementary Information, Figure S2). [α]_d = −44.5 (MeOH–CHCl₃ 1.5 : 98.5); m/z (ESI⁺) 435.2020 (M + Na⁺. C₂₂H₂₈N₄NaO₄ requires 435.2008) 436 (25%), and 435 (100). The (6S)-isomer (6S-7) (1.8 g) was obtained in 98.2% chiral purity (t_r = 10.6 min) as determined by analytical chiral HPLC (see Supplementary Information, Figure S2). [α]_d = +39.7 (MeOH–CHCl₃ 1.5 : 98.5); m/z (ESI⁺) 435.2015 (M + Na⁺. C₂₂H₂₈N₄NaO₄ requires 435.2008) 436(25%), and 435(100). It is noted that the optical rotations of the isomers are not equal and opposite but chiral HPLC analysis provides strong evidence of extremely high stereochemical purity of both diastereomers.

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroic acid (6R-8)

(6R)-2-Pivaloyl-5,10-dideaza-5,6,7,8-tetrahydropteroic acid methyl ester (6R-7) (250 mg, 0.242 mmol) was suspended in 6.1 mL 1 M NaOH. A reflux condenser was fitted to the flask and the reaction was heated at reflux temperature for 30 min after the starting material had completely dissolved. The flask was then cooled to ambient temperature, and the solution was filtered through a paper filter, which was then washed with water. The pH of the filtrate was adjusted to ∼ 1 with 6.7 mL of 1 M HCl. The resulting suspension was lyophilized to a small volume and resuspended in water for centrifugation. The precipitate was collected by centrifugation (15 min, 4 °C, 12000g) to obtain the desired product as a pellet. The pellet was then washed with ddH₂O (2 × 10 mL), CH₃CN (2 × 10 mL), and Et₂O (1 × 10 mL). The product was dried under high vacuum in the presence of P₂O₅. Ion-pair HPLC indicated >99% product purity. (196 mg, 92%); [α]_d = −49.5 (0.1 M NaOH) (lit.,²³ [α]_d = −48.3 (1 N NaOH, from the acid hydrolysis of (6R)-DDAH₄PteGlu₁) and [α]_d = −52.6 (1 N NaOH)); δ_H (300 MHz; 0.1 M NaOD in D₂O) 1.60 (2 H, m, C(10)H₂), 1.73 (1 H, m, C(6)H), 1.97 (1 H, dd, J = 8.8, 15.5 Hz, C(5)H₂), 2.53 (1 H, m, C(9)H₂), 2.72 (2 H, m, C(5)H₂ and C(9)H₂), 2.85 (1 H, m, C(7)H₂), 3.23 (1 H, m, C(7)H₂), 7.30 (2 H, d, J = 8.1 Hz, C(12)H and C(16)H), 7.74 (2 H, d, J = 8.1 Hz, C(13)H and C(15)H); m/z (ESI⁺) 315.1448 (M + H⁺. C₁₆H₁₉N₄O₃ requires 315.1457) 316(23%), 315(100), and 274(28); Ion-pair HPLC, t_r = 15.8 min. Spectral data are identical to those reported for this compound in the literature.²³

(6S)-5,10-Dideaza-5,6,7,8-tetrahydropteroic acid (6S-8)

(6S)-2-Pivaloyl-5,10-dideaza-5,6,7,8-tetrahydropteroic acid methyl ester (6S-7) was used in a procedure identical to that described for the synthesis of (6R)-5,10-dideaza-5,6,7,8-tetrahydropteroic acid (6R-8). Ion-pair HPLC indicated >99% product purity. (197 mg, 93%); [α]_d = +44.5 (0.1 M NaOH) (lit.,^53,54 [α]_d = +40.7 (1 N NaOH) and [α]_d = +48.3 (1 N NaOH, from the acid hydrolysis of (6S)-DDAH₄PteGlu₁)); δ_H (300 MHz; 0.1 M NaOD in D₂O) 1.59 (2 H, m, C(10)H₂), 1.73 (1 H, m, C(6)H), 1.96 (1 H, dd, J = 8.8, 15.5 Hz, C(5)H₂), 2.52 (1 H, m, C(9)H₂), 2.71 (2 H, m, C(5)H₂ and C(9)H₂), 2.85 (1 H, m, C(7)H₂), 3.23 (1 H, m, C(7)H₂), 7.29 (2 H, d, J = 8.1 Hz, C(12)H and C(16)H), 7.73 (2 H, d, J = 8.1 Hz, C(13)H and C(15)H); m/z (ESI⁺) 315.1456 (M + H⁺. C₁₆H₁₉N₄O₃ requires 315.1457) 316(24%), 315(94), 305(48), 297(21), 275(24), 274(100), 268(50), 233(31), and 79(40); Ion-pair HPLC, t_r = 15.9 min.

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl azide (6R-9)

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroic acid (6R-8) (47 mg, 0.13 mmol) was added under Ar to a dry flask containing a stir bar. The flask was sealed with a septum under an atmosphere of Ar (balloon) before the addition of 1 mL anhydrous DMSO. The solution was cooled in a 20° C water bath before the addition of Et₃N (107 µL, 0.77 mmol). After 15 min, cold ethyl chloroformate (14 µL, 0.15 mmol) was added. After 45 min, NaN₃ (10.5 mg, 0.16 mmol) was then added. The reaction was allowed to stir for 1 h at ambient temperature. Reaction completion was verified by RP-HPLC, Method 1. The reaction solution was diluted with 6 mL of 0.1 M HCl for precipitation. The precipitate was collected by centrifugation (15 min, 4° C, 12000g). The supernatant was removed and the pellet was washed with: 1 × 10 mL ddH₂O; 1 × 10 mL CH₃CN; and 1 × 5mL Et₂O. The pellet was allowed to air-dry for 15 min before drying under high vacuum in the presence of P₂O₅ for several hours. Total mass recovery, 49 mg, a mixture of 6R-9 (90.5%) and 6R-8 based on RP-HPLC (Method 1); Ion-pair HPLC, t_r = 30.6 min. This material was used in subsequent coupling reactions without further purification. Because of concerns about stability, ¹H NMR spectra were not obtained. These data for racemic 9 are reported in the Supplementary Information.

(6S)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl azide (6S -9)

(6S)-5,10-Dideaza-5,6,7,8-tetrahydropteroic acid (6S-8) was used in a procedure identical to that described for the synthesis of (6R)-5,10-dideaza-5,6,7,8-tetrahydropteroyl azide (6R-9). Total mass recovery, 51 mg, a mixture of 6S-9 (89.6%) and 6S-8 based on RP-HPLC (Method 1); Ion-pair HPLC, t_r = 30.4 min. This material was used in subsequent coupling reactions without further purification. Because of concerns about stability, ¹H NMR spectra were not obtained. These data for racemic 9 are reported in the Supplementary Information.

Solid phase synthesis of glutamyl-γ-glutamate peptides (H-Glu-γ-Glu_n-γ-Glu-OH, n = 0–4). Typical procedure.³⁸ l-Glutamyl-γ- l-glutamate (γ-Glu₂)

Resin loading

.714 g (4.03 mmol) Fmoc-Glu(Ot-Bu)–OH, 0.545 g (4.03 mmol) HOBt, and 2.06 g (3.95 mmol) PyBOP were dissolved in 10 mL of anhydrous DMF in a dry test tube with gentle mixing. After 15 min, 1.5 mL (8.64 mmol) of DIPEA was added to the test tube with gentle mixing. After 5 min, this solution was added to a reaction tube containing 0.905 g (1.08 mmol) of swelled Wang resin. After 90 min of agitation by inversion at ambient temperature, the reaction was stopped by filtering the resin and washing with 3 × 10 mL DMF and 2 × 10 mL CH₂Cl₂. Loading yield was determined by Fmoc quantification as follows. Two samples of loaded resin (approximately 1 mg each) were accurately weighed out and placed in two test tubes. A third tube lacking resin was used as the blank. 3 mL of 20% piperidine in DMF was added to each tube with agitation for 3 min. The solutions were then placed in quartz cuvettes and their absorbance at 290 nm measured. After subtracting the blank absorbance, eqn (3)³⁸ was used to determine that a loading yield of 0.53 mmol g⁻¹ (44%) was achieved.

$$\text{mmol}{g}^{-1}=\frac{O{D}_{290}}{\text{mg resin}\times 1.65}$$ (3)

Capping

A solution of 0.20 mL (2.16 mmol) acetic anhydride and 0.132 g (1.08 mmol) DMAP in 10 mL of anhydrous DMF was added to the resin. The mixture was agitated by inversion for 60 min at ambient temperature. The reaction was stopped by filtering the resin and washing with 3 × 10 mL DMF and 2 × 10 mL CH₂Cl₂.

Deprotection

50 mL of a 20% (v/v) solution of piperidine in anhydrous DMF was prepared. 10 mL of this solution was added to the resin with agitation. After 3 min, the resin was filtered and the above deprotection steps were repeated four times. After the final filtration, the resin was washed with 3 × 10 mL DMF and 2 × 10 mL CH₂Cl₂.

Coupling

0.808 g (1.90 mmol) Fmoc-Glu-Ot-Bu, 0.257 g (1.90 mmol) HOBt, and 0.702 g (1.85 mmol) HBTU were dissolved in 10 mL of anhydrous DMF in a dry test tube with gentle mixing. After 15 min, 0.66 mL (3.80 mmol) of DIPEA was added to the test tube with gentle mixing. After 5 min, this solution was added to a reaction tube containing the resin to which the C-terminal Glu_n was attached. After 90 min of agitation by inversion at ambient temperature, the reaction was stopped by filtering the resin and washing with 3 × 10 mL DMF and 2 × 10 mL CH₂Cl₂.

Capping and deprotection

As above.

Cleavage/deprotection

The resin was transferred to a 25 mL round-bottomed flask and mixed with 9.5 mL TFA, 2.5 mL ddH₂O, and 2.5 mL TIS. After 90 min at ambient temperature with occasional swirling, the solution was filtered and the resin was washed with TFA. The combined filtrates were concentrated in vacuo. The residue was then diluted with ddH₂O and lyophilized to remove trace TFA. The residue was dissolved in ddH₂O with the resulting solution being filtered with a 0.2 µm PTFE syringe filter prior to lyophilization. Mass spectral analysis indicated the presence of a significant amount of the mono-tert-butyl ester so the deprotection step was repeated as above except utilizing 9.5 mL TFA, 0.25 mL ddH₂O, and 0.25 mL TIS. Lyophilization yielded the crude dipeptide (172 mg, 93%). A portion of this material, 154 mg, was then purified by semi-preparative RP-HPLC (Method 1), t_r = 6.4 min, to obtain the TFA salt of the dipeptide after lyophilization. Total yield 75 mg (40% based on total resin loading, 45% based on fraction of crude material purified). δ_H (500 MHz; D₂O; numbering system for Glu_2–6, see Supplementary Information Scheme S2) 1.96 (1 H, m, C(8′)H₂), 2.14 (3 H, m, C(3′)H₂ and C(8′)H₂), 2.46 (4 H, m, C(4′)H₂ and C(9′)H₂), 3.87 (1 H, dd, J = 4.0, 6.1 Hz, C(2′)H), 4.35 (1 H, t, J = 4.4 Hz, C(7′)H); δ_C (500 MHz; D₂O) 26.1 (C8′), 26.2 (C3′), 30.5 (C9′), 31.5 (C4′), 52.7 (C7′), 53.7 (C2′), 173.3 (C=O), 174.9 (C=O), 175.7 (C=O), 177.5 (C=O), m/z (ESI⁺) 277.1. (C₁₀H₁₇N₂O₇ requires 277.1). This material was identical to a sample of l-Glu-γ-l-Glu prepared by solution methods (see Supplementary Information).

An identical procedure varying only in the number of cycles of coupling, capping, and deprotection was used to synthesize all other peptides of the series H-Glu-γ-Glu_n-γ-Glu-OH, n = 0–4.

Purification of poly(γ-l-glutamate) peptides

The crude TFA salts of the peptides were dissolved in ddH₂O to yield solutions with a concentration of ∼12 mg per 50 µL. 50 µL of these solutions were then individually used for purification via semi-preparative RP-HPLC as follows:

l-Glutamyl-(γ-l-glutamate)₂ (γ-Glu₃)

Semi-preparative RP-HPLC (Method 2) yielded 10.9 mg, 21.0 µmol, of the TFA salt of the tripeptide after lyophilization. δ_H (300 MHz; D₂O) 1.87 (2 H, m, C(8′)H₂ and C(13′)H₂), 2.07 (4 H, m, C(3′)H₂, C(8′)H₂, and C(13′)H₂), 2.33 (6 H, m, C(4′)H₂, C(9′)H₂, and C(14′)H₂), 3.93 (1 H, t, J = 6.4 Hz, C(2′)H), 4.23 (2 H, m, C(7′)H and C(12′)H); Semi-preparative RP-HPLC (Method 2), t_r = 16.0 min; m/z (ESI⁺) 406.2 (C₁₅H₂₄N₃O₁₀ requires 406.2) 406 (23%), 304 (100), and 282 (54).

l-Glutamyl-(γ-l-glutamate)₃ (γ-Glu₄)

Semi-preparative RP-HPLC (Method 3) yielded 8.4 mg, 13.0 µmol, of the TFA salt of the tetrapeptide after lyophilization. δ_H (300 MHz; D₂O) 1.96 (3 H, m, C(8′)H₂, C(13′)H₂, and C(18′)H₂), 2.17 (5 H, m, C(3′)H₂, C(8′)H₂, C(13′)H₂, and C(18′)H₂), 2.45 (8 H, m, C(4′)H₂, C(9′)H₂, C(14′)H₂, and C(19′)H₂), 3.97 (1 H, t, J = 6.4 Hz, C(2′)H), 4.32 (2 H, m, C(7′)H and C(12′)H), 4.38 (1 H, m, C(17′)H); Semi-preparative RP-HPLC (Method 3), t_r = 5.7 min; m/z (ESI⁺) 535.3 (C₂₀H₃₁N₄O₁₃ requires 535.2) 557 (20%), and 535 (100).

l-Glutamyl-(γ-l-glutamate)₄ (γ-Glu₅)

Semi-preparative RP-HPLC (Method 4) yielded 8.9 mg, 11.4 µmol, of the TFA salt of the pentapeptide after lyophilization. δ_H (300 MHz; D₂O) 1.96 (4 H, m, C(8′)H₂, C(13′)H₂, C(18′)H₂, and C(23′)H₂), 2.17 (6 H, m, C(3′)H₂, C(8′)H₂, C(13′)H₂, C(18′)H₂, and C(23′)H₂), 2.45 (10 H, m, C(4′)H₂, C(9′)H₂, C(14′)H₂, C(19′)H₂, and C(24′)H₂), 3.97 (1 H, t, J = 6.4 Hz, C(2′)H), 4.32 (3 H, m, C(7′)H, C(12′)H, and C(17′)H), 4.39 (1 H, m, C(22′)H); Semi-preparative RP-HPLC (Method 4), t_r = 10.7 min; m/z (ESI⁺) 664.4 (C₂₅H₃₈N₅O₁₆ requires 664.2) 664(100%).

l-Glutamyl-(γ-l-glutamate)₅ (γ-Glu₆)

Semi-preparative RP-HPLC (Method 4) yielded 7.7 mg, 8.5 µmol, of the TFA salt of the hexapeptide after lyophilization. δ_H (300 MHz; D₂O) 1.96 (5 H, m, C(8′)H₂, C(13′)H₂, C(18′)H₂, C(23′)H₂, and C(28′)H₂), 2.17 (7 H, m, C(3′)H₂, C(8′)H₂, C(13′)H₂, C(18)H₂, C(23′)H₂, and C(28′)H₂), 2.46 (12 H, m, C(4′)H₂, C(9′)H₂, C(14′)H₂, C(19′)H₂, C(24′)H₂, and C(29′)H₂), 3.96 (1 H, t, J = 6.4 Hz, C(2′)H), 4.33 (4 H, m, C(7′)H, C(12′)H, C(17′)H, and C(22′)H), 4.39 (1 H, m, C(27′)H); Semi-preparative RP-HPLC (Method 4), t_r = 11.4 min; m/z (ESI⁺) 793.5 (C₃₀H₄₅N₆O₁₉ requires 793.3) 794 (22%), and 793 (100).

5,10-Dideaza-5,6,7,8-tetrahydropteroylglutamyl-γ-glutamates (DDAH₄PteGlu_n, n = 1–6) (1–6)

An aqueous solution of the l-glutamate or poly-γ-glutamate (γ-Glu_2–6) was placed into a 10 × 75 mm test tube and concentrated to dryness on a vacuum centrifuge overnight at ambient temperature. When the tube was removed from the vacuum, a small, dry stir bar was placed in the tube and it was immediately sealed with a rubber septum. The tube was then gently flushed for 15 min with a stream of dry Ar introduced with a needle through the septum. The vent was then removed and the Ar line was replaced with an Ar-filled balloon. A solution consisting of (6R)-, or (6S)-5,10-dideazatetrahydropteroyl azide · HCl (9) (2 equiv.) in anhydrous DMSO was added to the tube with stirring followed by Et₃N (1 equiv. for each COOH of the amino acid or peptide, 1 equiv. for the amine salt (if applicable) of the amino acid or peptide, and 4 equiv. for 9). The reaction was monitored by HPLC and stopped after 2 days by the addition of aqueous NH₄HCO₃ buffer (pH 8.1). The reaction solution was then concentrated on a vacuum centrifuge overnight with heating (35–40 °C) for product analysis (HPLC) followed by purification by size-exclusion chromatography. All DDAH₄PteGlu_2–6 compounds were further purified by semi-preparative RP-HPLC, as detailed below, unless otherwise noted. Product yields were determined by UV absorption (272 nm, 0.1 M NaOH, ε = 11 700 M⁻¹ cm⁻¹).¹⁰

(6R)-5,10-Dideaza-5,6,7,8-tetrahydrofolic acid, (6R)-DDAH₄-PteGlu₁ (6R-1)

Yield, 87%; λ_max (0.1 M NaOH)/nm 242 and 272; [α]_d = −29.3 (0.1 M NaOD in D₂O) ([α]_d = −27.2 (0.1 M NaOD in D₂O, from an authentic sample of (6R)-DDAH₄PteGlu₁–Na); lit.,¹⁰ [α]_d = −21.1 (1 N NaOH in methanol, from the in situ saponification of the d-10-(+)-camphorsulfonic acid salt of (6R)-DDAH₄PteGlu₁ diethyl ester)); δ_H (500 MHz; 0.1 M NaOD/D₂O) 1.65 (2H, dd, J = 7.6, 15.0 Hz, C(10)H₂), 1.79 (1 H, m, C(6)H), 2.01 (2 H, dd, J = 8.9, 15.4 Hz, C(5)H₂, C(3′)H₂), 2.15 (1 H, m, C(3′)H₂), 2.30 (2 H, dd, J = 7.6, 7.9 Hz, C(4′)H₂), 2.58 (1 H, dd, J = 4.9, 15.6 Hz, C(9)H₂), 2.79 (2 H, m, C(5)H₂ and C(9)H₂), 2.89 (1 H, m, C(7)H₂), 3.28 (1 H, br d, J = 10.2 Hz, C(7)H₂), 4.31 (1 H, m, C(2′)H), 7.41 (2 H, d, J = 7.7 Hz, C(12)H and C(16)H), 7.74 (2 H, d, J = 7.7 Hz, C(13)H and C(15)H); δ_C (500 MHz; 0.1 M NaOD/D₂O) 26.3, 28.9, 31.5, 33.0, 34.7, 45.7, 56.4, 87.5, 127.8, 129.2, 131.4, 148.1, 161.0, 161.6, 170.6, 174.1, 179.4, 182.7; m/z (ESI+) 444.1873 (M + H⁺. C₂₁H₂₆N₅O₆ requires 444.1883) 445 (25%), and 444 (100); Ion-pair HPLC, t_r = 18.4 min. Spectral data (UV/vis, ¹H NMR, and ¹³C NMR) are identical to those obtained with an authentic sample of (6R)-DDAH₄PteGlu₁.

(6S)-5,10-Dideaza-5,6,7,8-tetrahydrofolic acid, (6S)-DDAH₄-PteGlu₁ (6S-1)

Yield, 92%; λ_max (0.1 M NaOH)/nm 242 and 272; [α]_d = +29.2 (0.1 M NaOD in D₂O) (lit.,¹⁰ [α]_d = +31.1 (1 N NaOH in methanol, from the in situ saponification of the d-10-(+)-camphorsulfonic acid salt of (6S)-DDAH₄PteGlu₁ diethyl ester); δ_H (500 MHz; 0.1 M NaOD/D₂O) 1.65 (2H, dd, J = 6.7, 14.3 Hz, C(10)H₂), 1.79 (1 H, m, C(6)H), 2.02 (2 H, dd, J = 7.4, 15.4 Hz, C(5)H₂, C(3′)H₂), 2.14 (1 H, m, C(3′)H₂), 2.30 (2 H, dd, J = 6.7, 7.9 Hz, C(4′)H₂), 2.58 (1 H, m, C(9)H₂), 2.79 (2 H, m, C(5)H₂ and C(9)H₂), 2.89 (1 H, dd, C(7)H₂), 3.28 (1 H, br d, J = 11.7 Hz, C(7)H₂), 4.31 (1 H, m, C(2′)H), 7.42 (2 H, d, J = 8.2 Hz, C(12)H and C(16)H), 7.74 (2 H, d, J = 8.2 Hz, C(13)H and C(15)H); δ_C (500 MHz; 0.1 M NaOD/D₂O) 26.3, 28.8, 31.4, 32.9, 34.7, 45.7, 56.4, 87.4, 127.8, 129.1, 131.4, 148.0, 161.0, 161.6, 170.6, 174.1, 179.4, 182.7; m/z (ESI⁺) 444.1879 (M + H⁺. C₂₁H₂₆N₅O₆ requires 444.1883) 445 (23%), and 444 (100); Ion-pair HPLC, t_r = 17.7 min.

(6R)-[¹⁴C]5,10-Dideaza-5,6,7,8-tetrahydrofolic acid, (6R)-DDAH₄Pte[¹⁴C]Glu₁

A 10 mL solution (ethanol–H₂O, 2 : 98) containing 1.0 mCi of l-[U-¹⁴C]glutamic acid was evaporated to dryness on a vacuum centrifuge. This residue was used in a reaction with 6R-9 generally as described above but with special precautions in place to prevent the inadvertent release of radioactive material. Two purifications by size-exclusion chromatography were required to obtain product of the desired radiochemical purity. Yield, 62%; Ion-pair HPLC, t_r = 17.5–18.5 min.; Purity, 96%; Specific activity = 249 Ci mol⁻¹.

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl-l-glutamate-γ-l-glutamate (6R-2)

Crude γ-Glu₂ was used for this reaction. Purified using semi-preparative RP-HPLC (Method 5), t_r = 20.0 min. Yield, 39%; λ_max (0.1 M NaOH)/nm 242 and 272; m/z (ESI⁺) 573.2 (M + H⁺. C₂₆H₃₃N₆O₉ requires 573.2); m/z (ESI⁺) 573.2305 (M + H⁺. C₂₆H₃₃N₆O₉ requires 573.2309) 574 (32%), 573 (100), 349 (21), and 305 (48); Ion-pair HPLC, t_r = 22.2 min.

(6S)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl-l-glutamate-γ-l-glutamate (6S-2)

Crude γ-Glu₂ was used for this reaction. Purification was by size-exclusion chromatography only. Yield, 31%; λ_max (0.1 M NaOH)/nm 242 and 272; m/z (ESI⁺) 573.2324 (M + H⁺. C₂₆H₃₃N₆O₉ requires 573.2309) 574 (29%), 573 (100), 305 (50), 274 (22), and 261 (34); Ion-pair HPLC, t_r = 20.8 min.

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl-l-glutamate-(γ-l-glutamate)₂ (6R-3)

Purified using semi-preparative RP-HPLC (Method 5), t_r = 15.0 min. Yield, 78%; λ_max (0.1 M NaOH)/nm 242 and 272; δ_H (500 MHz; 0.1 M NaOD/D₂O; numbering system, see Supplementary Information, Scheme S3) 1.72 (2 H, m, C(10)H₂), 1.89 (3 H, m, C(6)H, C(8′)H₂, and C(13′)H₂), 2.07 (5 H, m, C(5)H₂, C(3′)H₂, C(8′)H₂, and C(13′)H₂), 2.23 (2 H, dd, J = 7.8, 8.3 Hz, C(9′)H₂), 2.31 (2 H, dd, J = 7.7, 8.2 Hz, C(4′)H₂), 2.48 (2 H, dd, J = 6.4, 7.8 Hz, C(14′)H₂), 2.63 (1 H, m, C(9)H₂), 2.84 (2 H, m, C(5)H₂ and C(9)H₂), 2.95 (1 H, m, C(7)H₂), 3.34 (1 H, br d, J = 11.5 Hz, C(7)H₂), 4.12 (2 H, m, C(7′)H and C(12′)H), 4.40 (1 H, dd, J = 4.5, 8.9 Hz, C(2′)H), 7.46 (2 H, d, J = 7.9 Hz, C(12)H and C(16)H), 7.76 (2 H, d, J = 7.9 Hz, C(13)H an C(15)H); m/z (ESI⁺) 724.2527 (M + Na⁺. C₃₁H₃₉N₇NaO₁₂ requires 724.2554) 702 (100%); Ion-pair HPLC, t_r = 29.3 min.

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl-l-glutamate-(γ-l-glutamate)₃ (6R-4)

Purified using semi-preparative RP-HPLC (Method 6), t_r = 12.0 min. Yield, 59%; λ_max (0.1 M NaOH)/nm 242 and 272; m/z (ESI⁺) 853.2989 (M + Na⁺. C₃₆H₄₆N₈NaO₁₅ requires 853.2980) 832 (20%), 831 (100), and 427 (50); Ion-pair HPLC, t_r = 37.0 min.

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl-l-glutamate-(γ-L-glutamate)₄ (6R-5)

Purified using semi-preparative RP-HPLC (Method 7), t_r = 13.5 min. Yield, 53%; λ_max (0.1 M NaOH)/nm 242 and 272; m/z (ESI⁺) 982.3409 (M + Na⁺. C₄₁H₅₃N₉NaO₁₈ requires 982.3406) 960 (20%), 492 (30), 491 (63), and 381 (100); Ion-pair HPLC, t_r = 43.4 min.

(6R)-5,10-Dideaza-5,6,7,8-tetrahydropteroyl-l-glutamate-(γ-l-glutamate)₅ (6R-6)

Purified using semi-preparative RP-HPLC (Method 8), t_r = 15.5 min. Yield, 55%; λ_max (0.1 M NaOH)/nm 242 and 272; m/z (ESI⁺) 1111.3868 (M + Na⁺. C₄₆H₆₀N₁₀NaO₂₁ requires 1111.3832) 1089 (20), 556 (100), 545 (30), and 381 (41); Ion-pair HPLC, t_r = 48.4 min.

Preparation of DDAH₄PteGlu_n solutions

Each DDAH₄PteGlu_n was dissolved in 1–2 mL of 20–100 mM NaOH. The resulting solutions were filtered through a spin-filter. All solutions were then stored at −20 °C. Concentrations of the solutions were determined based on absorbance at 272 nm (0.1 M NaOH, ε = 11,700 M⁻¹ cm⁻¹).¹⁰ All solutions of DDAH₄PteGlu_n were freshly prepared and purity confirmed by HPLC. Extended storage (>2 months, −20 °C) of solutions with [NaOH] > 20 mM led to partial hydrolysis of the polyglutamate chain yielding a mixture of shorter chain compounds.

hFPGS Assays

Total glutamate incorporation

Determination of steady-state kinetic constants^49,55. General conditions

Final concentrations of 100 mM Tris (pH 8.85), 10 mM MgCl₂, 20 mM KCl, 10 mM NaHCO₃, 0.5 mg mL⁻¹ bovine serum albumin, 5 mM ATP, 2 mM l-[³H]glutamate (speci.c activity ∼2.6 cpm pmol⁻¹), 0–300 µM DDAH₄PteGlu_n, 14 nM hFPGS, 250 µL reaction volume, 60 min reaction time, and 37 °C. The solution was incubated at 37 °C for 5 min prior to addition of partially purified hFPGS expressed in either E. coli (n = 1–2) or bacculovirus-infected SF9 insect cells (n = 3–6) to initiate the reaction. Reactions were carried out in 1.5 mL, silanized Eppendorf tubes and were stopped by the addition of 1 mL of ice-cold buffer consisting of 5 mM glutamate, pH 7.5.

Assay products were purified with DEAE-cellulose (DE-52) ion-exchange chromatography using a buffer consisting of 10 mM Tris, and 110 mM NaCl, conductivity ∼11.5 mS, at pH 7.5 to remove unincorporated [³H]glutamate. Folate products were eluted with 0.1 M HCl. The amount of tritium incorporation into the folate products was determined by liquid scintillation counting. All reactions were carried out in duplicate and the blank reactions lacked the folate substrate.

Analysis of product formation

General conditions

As above, except for final concentrations of 20 mM MgCl₂, 5 mM dithiothreitol, 10 mM ATP, and 5 mM glutamate. The concentrations of (6R)-DDAH₄Pte[¹⁴C]Glu₁ utilized were 2.5, 25, or 250 µM with specific activities of 249, 24.9, and 2.49 Ci mol⁻¹ respectively. A 5 min reaction time was utilized with a final [hFPGS] = 1 µM and volume of 80 µL.

The solution was incubated at 37 °C for 5 min prior to reaction initiation by addition of purified hFPGS. The reactions were stopped by addition of 50% w/v trichloroacetic acid, with mixing, to a final volume of 5% w/v. For the control reaction, trichloroacetic acid was added immediately prior to the addition of hFPGS and analysis showed no product formation. The tubes were stored at −80 °C. The samples were clarified prior to HPLC analysis by centrifugation (16 000g, 10 min).

Ion-pair HPLC method

A portion (75 µL) of each sample was analyzed using the ion-pair HPLC method described above except fractions (30 s) were collected over the range t_r = 10–60 min.

Liquid scintillation counting method

The 30 s fractions were collected directly into 5.5 mL plastic scintillation vials. Bio-Safe II scintillation cocktail (4.0 mL) was added to each vial before capping and mixing by vortexing. Radioactivity (DPM) of each fraction was determined by liquid scintillation counting.