Concentration-Dependent Processivity of Multiple Glutamate Ligations Catalyzed by Folylpoly-γ-glutamate Synthetase

Abstract

Folylpoly-γ-glutamate synthetase (FPGS, EC 6.3.2.17) is an ATP-dependent ligase that catalyzes formation of poly-γ-glutamate derivatives of reduced folates and anti-folates such as methotrexate and 5,10-dideaza-5,6,7,8-tetrahydrofolate (DDAH₄PteGlu₁). While the chemical mechanism of the reaction catalyzed by FPGS is known, it is unknown whether single or multiple glutamate residues are added following each folate binding event. A very sensitive high performance liquid chromatography method has been used to analyze the multiple ligation reactions onto radiolabeled DDAH₄PteGlu₁ catalyzed by FPGS in order to distinguish between distributive or processive mechanisms of catalysis. Reaction time courses, substrate trapping, and pulse-chase experiments were used to measure folate release during multiple glutamate additions. Together, the results of these experiments indicate that hFPGS can catalyze the processive addition of approximately four glutamate residues onto DDAH₄PteGlu₁. The degree of processivity was determined to be dependent on the concentration of the folate substrate, thus suggesting a mechanism for the regulation of folate polyglutamate synthesis in cells.

Folylpoly-γ-glutamate synthetase (FPGS¹, EC 6.3.2.17) catalyzes the intracellular poly-γ-glutamylation of natural folates, such as (6S)-5,6,7,8-tetrahydrofolate ((6S)-H₄PteGlu₁), and antifolates, such as (6R)-5,10-dideaza-5,6,7,8-tetrahydrofolate (DDAH₄PteGlu₁, DDATHF) (Figure 1). These polyglutamylated compounds exhibit increased affinity for the target enzymes and enhanced intracellular retention (1-4). DDAH₄PteGlu₁ has been evaluated in clinical trials under the trade name Lometrexol as a cancer chemotherapeutic agent (5-8) while a closely related compound, pemetrexed (Alimta), is now approved for the treatment of several cancers (9). The mechanism of action of DDAH₄PteGlu₁ as an antitumor agent involves inhibition of glycinamide ribonucleotide formyltransferase (GARFT, EC 2.1.2.1), an essential enzyme in the de novo purine biosynthetic pathway (10).

Figure 1.

The reaction catalyzed by FPGS. Note that the total number of glutamate residues added (m) is one less than the total number of glutamates in the product, H₄PteGlu_n; i.e., n = m +1.

FPGS has been found to convert DDAH₄PteGlu₁ rapidly to polyglutamate forms in cell-free reactions (11), in cells (12), and in the mouse (13). Thus, the polyglutamate forms of DDAH₄PteGlu₁ predominate within the cell and account for the cytotoxic action of the drug. Indeed, the hexaglutamate (DDAH₄PteGlu₆) metabolite has 10-fold lower K_D for GARFT than does the parent compound (14). Also, a decrease in polyglutamylation of DDAH₄PteGlu₁ has led to resistance due to decreased cytotoxic effects in cancer cells (12, 15-17). Thus, the mechanism of FPGS-catalyzed ligation of multiple moles of glutamate to DDAH₄PteGlu₁ is central to our understanding of the action of DDAH₄PteGlu₁ and related drugs in vivo.

The chemical reaction catalyzed by FPGS involves simple amide bond formation via a γ-glutamyl phosphate intermediate (18). However, the reaction requires three substrates, a folate or antifolate, ATP, and glutamate, and produces three products, a γ-glutamyl peptide metabolite of the folate or antifolate, ADP, and inorganic phosphate (P_i). Investigation of this reaction is further complicated by the fact that the folate products of this reaction can generally act again as FPGS substrates. This has been shown to be the case as FPGS alone is capable of catalyzing formation of the entire range of folylpoly-γ-glutamates found in cells (11-13, 19, 20). Since FPGS is able to further utilize the folate products of its reaction as substrates, it may be considered a polymerase which repeatedly adds a glutamate monomer onto the growing folylpoly-γ-glutamate oligomer. An important kinetic property of polymerases is the degree of processivity of the reaction (21, 22) as defined by the number of monomer additions, or turnovers, catalyzed per folate substrate binding event. Most relevant to the current study are non-template polymerization reactions such as those involved in the biosynthesis of oligosaccharides. For example, a peptidoglycan glycosyltransferase has been found to synthesize a 40-mer peptidoglycan polymer without the aid of an existing template via a processive mechanism (23, 24).

Previous studies of the FPGS-catalyzed reaction involved determination of the steady-state kinetic mechanism. An Ordered Ter Ter mechanism was determined for the reaction catalyzed by FPGS isolated from three different sources utilizing different folate or antifolate substrates to which only a single glutamate was added (25-27). In all cases, the substrates bind in the order of ATP, folate/antifolate, and glutamate and the products are released in the order of ADP, folate/antifolate-Glu, and P_i (Figure 2A). If this kinetic mechanism were to hold for substrates that are converted to polyglutamates, a processive reaction, in which the ligation product remains bound while both ADP and P_i are released, would be precluded. The ligation product would also need to remain bound during the binding of another molecule each of ATP and glutamate in preparation for the next turnover (e.g., Figure 2B). DDAH₄PteGlu₁, like the naturally occurring tetrahydrofolates, is rapidly converted to DDAH₄PteGlu_5-6 in human cancer cells (12), murine liver (13), and with purified recombinant human cytosolic FPGS (11) with little accumulation of short chain polyglutamates. This evidence is suggestive of a processive mechanism for human FPGS, which is investigated in the current studies.

Figure 2.

A. The kinetic mechanism of FPGS as determined previously under non-processive ligation conditions. B. A possible kinetic mechanism that allows for processive glutamate ligations onto a single folate substrate. Any scheme in which the folate is the first substrate to bind and the folate-Glu product is the last to be released is compatible with a processive mechanism. A rearrangement step that places folate-Glu in the scheme in place of folate would be required for a processive mechanism. “Folate” is used as a generic term for the folate or antifolate substrate. “Folate-Glu” is also a generic term to denote the ligation product containing a single additional glutamate.

The possibility of processivity in the reaction catalyzed by hFPGS was studied directly. DDAH₄PteGlu₁ was chosen as the folate acceptor substrate for several reasons. Isotopically labeled DDAH₄PteGlu₁ and the unlabeled polyglutamates are available via a convenient synthetic route (28), these compounds display excellent chemical stability (e.g., redox and light), and DDAH₄PteGlu₁ is efficiently converted to longer chain polyglutamates. In addition, determination of the steady-state kinetics of hFPGS-catalyzed glutamate ligation for each of several DDAH₄PteGlu poly-γ-glutamates with hFPGS has recently been achieved (28) providing the necessary foundation for investigating the two mechanistic possibilities. In the previous work, modeling studies that assumed a strictly distributive mechanism were unable to adequately reproduce the observed pattern of substrate consumption and product distribution when [DDAH₄PteGlu₁] ≤ 10 K_M (28). However, the work also indicated that at very high DDAH₄PteGlu₁ concentrations, 250 μM (~100 K_M), a processive mechanism was not operative as only the DDAH₄PteGlu₂ product was observed (28). The choice of DDAH₄PteGlu₁ and hFPGS provides data on an important enzyme mechanism question in the context of a clinically important enzyme-substrate pair. In this paper, we report the results of a detailed kinetics investigation of the reaction catalyzed by hFPGS including 1) a time course study monitoring the formation of all polyglutamate products and subsequent kinetic analysis, 2) substrate trapping experiments to address the validity of the reported kinetic mechanism, and 3) pulse-chase experiments (29, 30) to probe directly a possible processive mechanism of FPGS-catalyzed multiple ligations.

EXPERIMENTAL PROCEDURES

Materials

Common solvents and reagents were obtained from commercial sources and were of the highest purity available. DDAH₄PteGlu₁, DDAH₄Pte[¹⁴C]Glu₁, and DDAH₄PteGlu_n, (n = 2-6) were prepared as described previously (28). Each DDAH₄PteGlu_n (n = 1-6) was dissolved in 1-2 mL of 20-100 mM NaOH.² The resulting solutions were clarified by filtration with a spin-filter. All solutions were then stored at −20 °C. Solution concentrations were determined based on absorbance at 272 nm (0.1 M NaOH, ε = 11,700 M⁻¹ cm⁻¹) (31). Spin-X HPLC 0.2 μm nylon microcentrifuge filters (spin filters) were obtained from Costar. Partially purified hFPGS, expressed in E. coli or baculovirus-infected SF9 insect cells (11, 32) and partially purified as described previously (33), was a generous gift of Dr. John J. McGuire. Homogenous hFPGS was expressed in baculovirus-infected SF9 insect cells and purified as described previously (11). hFPGS concentration was determined by the method of Bradford (34) and corrected for protein purity, if necessary, based on quantitative PAGE.

General Procedures

Scintillation counting used either Bio-Safe II (Research Products Inc.) or Ultima Gold (Perkin Elmer Life Sciences) scintillation cocktail. The HPLC system consisted of an auto-sampler, two pumps, a photo diode array detector, a fraction collector controlled by Varian Star v6.3 software, and was run at ambient temperature. All kinetics data were fitted using KaleidaGraph v3.5, Synergy Software (Reading, PA). Dynafit v3.28.046, BioKin Ltd. (Pullman, WA) (35) was used for kinetics modeling.

High Performance Liquid Chromatography (HPLC)

The conditions for ion-paired HPLC were as follows: 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, 4.6 × 250 mm. Flow rate - 1.0 mL/min. 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. The effluent was monitored at 280 nm or by the determination of radioactivity in each fraction by liquid scintillation counting. Fractions (30 sec, 0.5 mL) for radiation chromatograms were collected directly in 5.5 mL polypropylene multi-vials for t_r = 10-60 min. Representative radiation chromatograms are shown in Figure 3 with panel A consisting solely of DDAH₄Pte[¹⁴C]Glu₁ and panel B of a mixture of biosynthetic DDAH₄Pte[¹⁴C]Glu_n. For analysis of reaction products, eight fractions (3-4 min, 3-4 mL) were collected corresponding to DDAH₄PteGlu_n (n = 1-6) and a background (BGD). Fractions for n = 1 (two fractions, 14-18 min and 18-21 min), n = 2 (21-25 min), n = 3 (28-32 min), n = 4 (35-39 min), n = 5 (41.5-45.5 min), n = 6 (46.5-49.5 min), and a background (55-59 min) were collected directly into 20 mL glass scintillation vials. The times used to delimit the fractions were determined immediately prior to each set of HPLC analyses by injection of a mixture of biosynthetic DDAH₄Pte[¹⁴C]Glu_n.

Figure 3.

Representative radiation chromatograms of the ion-pair HPLC method. A. Chromatogram of DDAH₄Pte[¹⁴C]Glu₁ isolated from a FPGS-catalyzed reaction at t = 0 min. B. Chromatogram of biosynthetic DDAH₄Pte[¹⁴C]Glu_n isolated from a FPGS-catalyzed reaction of DDAH₄Pte[¹⁴C]Glu₁ and L-glutamic acid to illustrate the separation of the various chain length polyglutamate products. BGD = background.

Liquid Scintillation Counting

Bio-Safe II scintillation cocktail (4.0 mL) was added to each fraction collected for the radiation chromatogram. Each vial was analyzed for two min to determine the DPM values. Ultima Gold scintillation cocktail (10.0 mL or 13.3 mL of cocktail for 3 mL or 4 mL of sample, respectively) was added to the fractions collected for reaction product analysis. Each vial was analyzed for two cycles of five min each to determine the DPM values.

Quantitative Analysis of Product Formation

The number of moles of the substrate and each of the polyglutamate products was determined by fractionating the eluant from a HPLC column corresponding to the position of authentic standards. The radioactivity measurements were corrected for background and the total radioactivity corrected for recovery from the column. The corrected DPM values were then converted to moles using the known specific activity of the DDAH₄Pte[¹⁴C]Glu₁, yielding the moles of each species of DDAH₄PteGlu_1-6 present.

Determination of reaction time course

For these experiments, the following conditions were employed: 100 mM Tris (pH 8.85), 20 mM MgCl₂, 20 mM KCl, 10 mM NaHCO₃, 100 mM 2-mercaptoethanol, 0.5 mg/mL bovine serum albumin (BSA), 10 mM ATP, 5 mM glutamate, 14 nM partially purified hFPGS, and 2.5 or 25 μM DDAH₄Pte[¹⁴C]Glu₁ (specific activity of 120.1 or 24.0 Ci/mole respectively) in a total volume of 880 μL. All reactions were assayed at 37 °C. The assay solution, lacking hFPGS, was preincubated at 37 °C for 5 min. The reactions were then initiated by the addition of ice-cold hFPGS with gentle mixing. After the desired reaction time had elapsed, an 80 μL aliquot was immediately transferred to a microcentrifuge tube and the reaction was stopped by immersion in a boiling water bath for four min. For t = 0, the aliquot was heated in a boiling water bath for two min prior to the addition of an appropriate amount of the enzyme solution after which it was immediately returned to the boiling water bath for an additional two min. The tubes were then chilled on ice. The solutions were clarified by centrifugation (8,000g, 10 min). The supernatant was transferred to a spin filter and filtered by centrifugation (2,000g, 2 min). The tube was rinsed three times with 250 μL of ddH₂O, and the rinses were then filtered as above. The filter membranes were found to retain only very low levels of radioactivity. The filtered solutions were evaporated at ambient temperature overnight on a vacuum centrifuge, the resulting residues were reconstituted with 80 μL ddH₂O, and stored at −80 °C pending analysis by ion-pair HPLC.

Calculation of the concentration of hFPGS·DDAH₄PteGlu₁ (E·S) complex

[ES] was determined based on the total hFPGS concentration, [E]_t, DDAH₄PteGlu₁ concentration, [S], and the K_M value for DDAH₄PteGlu₁(Eq. 1) (36).

$$\left[\mathrm{ES}\right]=\frac{{\left[\mathrm{E}\right]}_{\mathrm{t}}\ast \left[\mathrm{S}\right]}{{\mathrm{K}}_{\mathrm{M}}+\left[\mathrm{S}\right]}$$ (1)

Substrate trapping experiments (Supporting Information, Figure S2)

For these experiments, the following conditions were employed: 100 mM Tris (pH 8.80), 10 mM MgCl₂, 20 mM KCl, 10 mM NaHCO₃, 5 mM DTT, 0.5 mg/mL BSA, 10 mM ATP, 5 mM glutamate, 25 μM DDAH₄PteGlu₁ (preincubation and trapping solutions), and 105 nM homogeneous hFPGS in a final volume of 80 μL. All reactions were assayed in triplicate at 37 °C and stopped by the addition of TCA (50% w/v to a final concentration of 5% w/v). A preincubation solution was prepared that lacked ATP, glutamate, and hFPGS, but contained high (249 Ci/mole) or low (24.9 Ci/mole) specific activity DDAH₄Pte[¹⁴C]Glu₁ for the substrate trapping and control experiments, respectively. A trapping solution was prepared that lacked only hFPGS, but contained ATP, glutamate, and either unlabeled DDAH₄PteGlu₁ or low specific activity DDAH₄Pte[¹⁴C]Glu₁ for the trapping and control experiments, respectively. Both the preincubation and trapping solutions were incubated at 37 °C for 5 min. hFPGS was then added to the preincubation solution yielding an enzyme concentration of 1.05 μM. After an additional 5 min at 37 °C, the reaction was initiated by the addition of the trapping solution to a preincubation solution containing hFPGS, resulting in a 10-fold dilution of FPGS. After the desired reaction time had elapsed, the reaction was stopped by addition of TCA (50% w/v to a final concentration of 5% w/v). To obtain a t = 0 min data point, TCA was added to the hFPGS-containing preincubation solution prior to the addition of the trapping solution. Following the TCA quench, reactions were chilled on ice. The reaction solutions were then clarified by centrifugation (16,000g, 10 min) and stored at −80 °C pending analysis by ion-pair HPLC.

Pulse-chase experiments. (Supporting Information, Figure S4)

For experiments where [DDAH₄PteGlu₁] = 25 μM during the chase phase, the following conditions were employed: 100 mM Tris (pH 8.80), 10 mM MgCl₂, 20 mM KCl, 10 mM NaHCO₃, 5 mM DTT, 0.5 mg/mL BSA, 10 mM ATP, 5 mM glutamate, 25 μM DDAH₄PteGlu₁ (pulse and chase phases), and 1.0 μM homogeneous hFPGS in a final volume of 80 μL. All reactions were assayed in triplicate at 37 °C and stopped by the addition of TCA (50% w/v to a final concentration of 5% w/v). A pulse solution was prepared that lacked only hFPGS, but contained high specific activity DDAH₄Pte[¹⁴C]Glu₁ (249 Ci/mole). A chase solution was prepared with unlabeled DDAH₄PteGlu₁. Both the pulse and chase solutions were preincubated at 37 °C for 5 min. To obtain the t = 0 min data points, TCA was added to the preincubated pulse solution followed by hFPGS, to 10 μM, and finally chase solution. For the t = 2 min pulse data points, the reaction was initiated by addition of hFPGS to 10 μM in the pulse solution. TCA was then added at t = 2 min to stop the reaction prior to the addition of chase solution. To obtain the chase data points, the reaction was initiated as above and the chase solution was added at t = 2 min. The reaction was stopped at the desired time by removing an aliquot from the common reaction solution and adding it to a tube containing TCA. In all cases, addition of the chase solution resulted in a ca. 10-fold dilution of FPGS. For all points, the tubes containing the quenched reactions were immediately placed on ice, then clarified by centrifugation (16,000g, 10 min), and stored at −80 °C pending analysis by ion-pair HPLC.

For experiments where [DDAH₄PteGlu₁] = 250 μM during the chase phase, the conditions were as described directly above, with the following exceptions. hFPGS was added to the pulse solution such that the concentration was 1.02 μM. DDAH₄PteGlu₁ was 25 μM in the pulse phase and 250 μM in the chase phase. The chase solution consisted of 9.01 mM unlabeled DDAH₄PteGlu₁. 1.0 μM was the final concentration of homogeneous hFPGS.

RESULTS AND DISCUSSION

Steady-state kinetics data obtained previously at a single time point (28) provided the basis for the more extensive time course experiments described herein. Investigation of possible processivity in the reaction catalyzed by hFPGS was initiated by determination of the time course (0 – 120 min) of the reaction utilizing DDAH₄Pte[¹⁴C]Glu₁ at two different concentrations, 2.5 μM (~ K_M, Figure 4A) and 25 μM (~ 10 K_M, Figure 4B). These data clearly show that the degree of multiple glutamate ligations to DDAH₄PteGlu₁ is concentration-dependent. The results of this experiment are very similar to those observed for partially purified rat liver FPGS with H₄PteGlu₁ as the substrate (20); multiple glutamate ligations to H₄PteGlu₁ at low concentrations (5 μM, ~K_M) yielded H₄PteGlu_4-5 as the predominant products following initial formation and consumption of H₄PteGlu₂ and H₄PteGlu₃. In contrast, H₄PteGlu₁ at a higher concentration (35 μM, ~10 K_M) yielded predominately H₄PteGlu₂ and H₄PteGlu₃ as the initial products, but lower relative amounts of H₄PteGlu_4-5 products were observed at longer reaction times. The time course data for the reaction containing 2.5 μM DDAH₄Pte[¹⁴C]Glu₁ (Figure 4A) agree well with what was observed in whole CCRF-CEM cells (12) and in the livers of mice (13) that were treated with DDAH₄PteGlu₁, thus providing a strong correlation between data obtained in cell-free experiments and those obtained in intact cells or in vivo.

Figure 4.

Time course of the reaction of DDAH₄Pte[¹⁴C]Glu₁ with partially purified hFPGS expressed in baculovirus-infected SF9 insect cells. The concentration of FPGS was approximately 14 nM. The concentrations of the glutamate and ATP substrates were saturating at 5 mM and 10 mM respectively. Product formation, DDAH4PteGlu_n,was analyzed as described in the Experimental Section and Supporting Information: n = 1 (□), n = 2 (□), n = 3 (◇), n = 4 (◻), n = 5 (◻), n = 6 (◻). A. Reaction with 2.5 μM (~K_M) DDAH₄Pte[¹⁴C]Glu₁. B. Reaction with 25 μM (~10 K_M) DDAH₄Pte[¹⁴C]Glu₁. The data for n = 1 are not shown for clarity.

These data (Figure 4) are strongly suggestive of a processive reaction. When considered together with a simplified kinetic scheme (Figure 5), the time course data for 2.5 μM DDAH₄PteGlu₁ (Figure 4A) indicate that the rates of glutamate ligation, k₂[E·S₂], k₃[E·S₃], and k₄[E·S₄], are much greater than the corresponding rates of dissociation, k₆[E·S₂], k₇[E·S₃], and k₈[E·S₄]. Since the DDAH₄PteGlu₅ product (S₅) accumulates in this reaction, the rate of dissociation, k₉[E·S₅], is greater than the rate of further ligation, k₅[E·S₅]. However, the rate, k₅[E·S₅], is apparently not zero because small amounts of the DDAH₄PteGlu₆ product (S₆) are observed at longer reaction times (Figure 4A). Similarly, inspection of the 25 μM DDAH₄PteGlu₁ time course data (Figure 4B) shows continued accumulation of the triglutamate, DDAH₄PteGlu₃ (S₃), in the presence of a large amount of the monoglutamate, DDAH₄PteGlu₁ (S₁). This demonstrates that the rate of ligation, k₂[E·S₂], is greater than dissociation, k₆[E·S₂]. The fact that tetra- and pentaglutamates, DDAH₄PteGlu₄ (S₄) and DDAH₄PteGlu₅ (S₅), are detected at longer times indicates that the ligation rates k₃[E·S₃] and k₄[E·S₄] can ultimately prevail over the dissociation rates, k₇[E·S₃] and k₈[E·S₄]. This behavior is also consistent with a processive mechanism. Finally, the time course data shown in Figure 4 were subjected to a kinetics modeling analysis using DynaFit (v.3.28.046). Nine variants of the scheme shown in Figure 5 were evaluated (Supporting Information, Figure S1), each of which allow for either processive or distributive modes of multiple glutamate additions. Although excellent visual fits to the experimental data were obtained, error analysis of the rate constants derived from the fits led mainly to large uncertainties in the modeled rate constants. Similarly, values for the processivity factors, P = k_cat/(k_cat + k_off), could not be obtained with certainty by this analysis.

Figure 5.

Simplified kinetic scheme of the FPGS reaction. hFPGS is represented by E. DDAH₄Pte[¹⁴C]Glu_n, n = 1-6, are represented by S₁, S_2, S₃, S₄, S₅, and S₆ respectively. The products of rate constants and enzyme-substrate complex concentrations, k₁[E·S₁] – k₅[E·S₅], represent rates of ligation while k₆[E·S₂] – k₁₀[E·S₆] represent rates of complex dissociation. The binding of ATP and glutamate and the release of P_i and ADP are excluded in this simplified scheme.

In order to assess directly a possible processive kinetic mechanism in the reaction catalyzed by FPGS, substrate trapping and pulse-chase experiments were pursued. A substrate trapping experiment was performed to assess the formation of a catalytically competent binary complex, hFPGS·DDAH₄Pte[¹⁴C]Glu₁ (“E-Folate”, Figure 2B). An experiment which results in increased incorporation of DDAH₄Pte[¹⁴C]Glu₁, into polyglutamates due to substrate trapping would indicate that the binary complex forms and is catalytically competent. Failure to observe substrate trapping would be consistent with one of three scenarios: 1) the binary complex does not form, 2) it does form but is rapidly exchanged with free substrate, or 3) it does form but is not catalytically competent. The substrate trapping experiment addresses directly the interpretation of the previously determined kinetic mechanism of FPGS (25-27) that precludes the kinetic competence of the hFPGS·DDAH₄Pte[¹⁴C]Glu₁ binary complex and therefore a processive mechanism. These previous studies, however, were carried out under conditions in which the question of a processive catalytic mechanism could not be addressed. Specifically, the Corynebacterium FPGS produces only the H₄PteGlu₂ product when H₄PteGlu is utilized as the substrate (25). FPGS from hog liver or L. casei catalyzes the ligation of glutamate to aminopterin (27) or 5,10-CH₂-H₄PteGlu₂ (26), respectively. However, the products derived from a single ligation reaction, aminopterin-Glu and 5,10-CH₂-H₄PteGlu₃, are very poor substrates for further catalysis by the respective FPGS enzymes. Therefore, in all of these previous studies, the reaction conditions, enzyme source, and substrate were chosen to prevent multiple turnovers of the folate or antifolate substrate. A substrate trapping experiment allows for a direct evaluation of the proposed Ordered Ter Ter kinetic mechanism with the enzyme-substrate pair, hFPGS·DDAH₄Pte[¹⁴C]Glu₁, under multiple ligation conditions. The design of the substrate trapping experiments is shown in the Supporting Information, Figure S3.

Preincubation of hFPGS (105 nM) and high specific activity DDAH₄Pte[¹⁴C]Glu₁ (25 μM) was carried out for 5 min prior to initiation of the reaction to allow the formation of the postulated E·S binary complex. The reaction was initiated by the addition of ATP and glutamate at saturating concentrations with concurrent trapping by addition of 10 volumes of 25 μM unlabeled DDAH₄PteGlu₁. This results in a 10-fold reduction of specific activity while maintaining constant substrate concentration. The reaction was allowed to proceed for a given time (0 – 6 min) prior to quenching by the addition of TCA to obtain a time course of the reaction. A control reaction involved the same procedure except DDAH₄Pte[¹⁴C]Glu₁ of a low specific activity, identical to the final specific activity achieved in the substrate trapping experiment, was used in both the preincubation and trapping phases. If the hFPGS·DDAH₄Pte[¹⁴C]Glu₁ binary complex forms during preincubation and is catalytically active, the reaction will proceed with the enzyme-bound, high specific activity substrate when initiated. When a polyglutamate product is released following ligation of Glu to DDAH₄Pte[¹⁴C]Glu₁, this high specific activity product would be diluted by the large excess of unlabeled DDAH₄PteGlu₁. In contrast, if the hFPGS·DDAH₄Pte[¹⁴C]Glu₁ binary complex does not form, is not catalytically competent, or exchanges rapidly with free substrate, the high specific activity substrate used in the preincubation would be immediately diluted with the unlabeled DDAH₄PteGlu₁, yielding a pool of low specific activity substrate that would then be available to FPGS for reaction. This scenario would result in the formation of polyglutamate products of low specific activity, equivalent to the control reaction, whether or not the hFPGS·DDAH₄Pte[¹⁴C]Glu₁ binary complex is competent. The experiment is illustrated schematically in the Supporting Information, Figure S2.

When compared with the control experiment, there is an increase in the amount of product formed at each time point in the substrate trapping experiment (Figure 6). This indicates that the high specific activity substrate used in the substrate trapping reaction forms a hFPGS·DDAH₄Pte[¹⁴C]Glu₁ binary complex that is kinetically competent. Extrapolation of the time course data to t = 0 min indicates that an additional 34.1 ± 5.2 pmoles of glutamate were incorporated into product in the substrate trapping experiment vs. that incorporated in the control experiment. The concentration of hFPGS·DDAH₄PteGlu₁ (E·S) complex present in this experiment was calculated to be 98.5 nM (Eq 1) based on a total hFPGS concentration and DDAH₄PteGlu₁ concentration of 105 nM and 25 μM, respectively. Under these conditions, the K_M value for DDAH₄PteGlu₁ was determined to be 1.65 μM (data not shown). Considering the final reaction volume of 75 μL and [E·S] = 98.5 nM, a total of 7.39 pmoles of E·S complex was present. Therefore, an average of 4.6 ± 0.7 moles of glutamate were added per mole of hFPGS·DDAH₄PteGlu₁ complex formed. This result is supported by determination of the amount of DDAH₄PteGlu_n products formed in the substrate trapping (Figure 7A) and control (Figure 7B) experiments, as well as the difference in oligo-γ-glutamate (n = 1-6) formation between the two experiments (Figure 7C) during the time period involved. These data demonstrate a much greater production of long-chain polyglutamate products (n = 3-6) under substrate trapping conditions than in the control experiment, thus confirming the conversion of enzyme-bound DDAH₄PteGlu₁ to long-chain products. They are consistent with the calculated stoichiometry of 4.6 moles of glutamate incorporated per mole of E·S complex, and support formation of kinetically competent hFPGS·DDAH₄PteGlu₁ binary complex. The only reasonable explanation for the observation that [P]:[E·S] > 1 is that the initial substrate in the binary complex undergoes multiple turnovers. These data are interpreted as direct evidence of a processive mechanism of catalysis under these conditions.

Figure 6.

Substrate trapping and control experiments (□) and (◻) respectively. The concentration of hFPGS was 105 nM and DDAH₄PteGlu₁ was 25 μM. The average of three experiments was plotted with the error bars representing the standard deviation. When extrapolated to t = 0 min, 34.1 additional pmoles of glutamate were incorporated into product in the substrate trapping reaction when compared to the control reaction.

Figure 7.

Time dependence of DDAH₄PteGlu_n formation. Biosynthetic DDAH4PteGlu_n,was analyzed as described in the Experimental Section and Supporting Information: n = 2 (□), n = 3 (◇), n = 4 (◻), n = 5 (◻), n = 6 (◻). The data for n = 1 are not shown in A or B for clarity. A. Product distribution under substrate trapping conditions (Figure 6, solid line). B. Product distribution under control conditions (Figure 6, dashed line). C. The difference in average product formation between the substrate trapping and control reactions by product chain length and time. At t = 0 min, there is no difference between the two reactions. t = 2 min (crossed), t = 4 min (solid), and t = 6 min (diagonal).

A notable feature of the substrate trapping data (Figure 6) is a non-zero y-intercept for the control reaction. Examination of product distribution data for the control reaction (Figure 7B) reveals that an apparent burst of product formation occurs primarily in the formation of DDAH₄PteGlu₂. Similar results were observed when a similar but more extended reaction time course (0 – 120 min) was examined, also at 25 μM DDAH₄PteGlu₁ (Figure 4B).

The observation of a burst in formation of DDAH₄PteGlu₂, the product of the first ligation, followed by a slower, steady-state ligation of additional glutamates in both the control and substrate trapping reactions (Figure 7), is also consistent with a processive mechanism. In a processive polymerization reaction, it is necessary for the product of the reaction to reorient itself in the active site prior to the next round of catalysis. For the FPGS reaction, the newly formed C-terminal glutamate of DDAH₄PteGlu₂, assumed to form rapidly from the substrate-based quaternary complex, hFPGS·DDAH₄Pte[¹⁴C]Glu₁·ATP·Glu, occupies the binding site of the incoming glutamate substrate immediately following catalysis. For another round of catalysis to occur, several events need to occur: 1) the C-terminal glutamate of the polyglutamate product must be released from the incoming glutamate binding site, 2) two of the products, ADP and P_i, must be released, 3) ATP and glutamate must bind to FPGS, and 4) the C-terminal glutamate of the polyglutamate product must be repositioned such that the γ-carboxylic acid is adjacent to the γ-phosphorus of ATP (18). Together, these steps require a rearrangement of the active site and suggest a mechanism in which the chemical step of peptide bond formation is followed by a slower, rate-limiting rearrangement of the newly formed hFPGS·DDAH₄Pte[¹⁴C]Glu-γ-Glu_n binary complex, thus giving rise to a burst. Rearrangement of the FPGS active site during catalysis was suggested previously, based on poor substrate activity of a α-descarboxy analog of methotrexate (37).

The proposed processive mechanism of hFPGS-catalyzed ligation of multiple glutamate residues to DDAH₄PteGlu₁ was further investigated by pulse-chase experiments.³ The design of the pulse-chase experiment is shown in the Supporting Information, Figure S5. Experiments of this type are generally used in conjunction with kinetic isotope effect studies to determine the commitment to catalysis (39) and are conceptually related to the substrate trapping experiments described above. The major difference between pulse-chase experiments and substrate trapping experiments is that, in the former, the reaction is already proceeding with the high specific activity substrate (pulse) when the unlabeled substrate (chase) is added. The time course of the reaction after addition of the chase is determined and then extrapolated back to t₁, the time when the chase was initiated. The point at which the extrapolated line intersects t₁, the chase initiation time, is compared with a time point obtained when an identical reaction is quenched at t₁ instead of adding the chase. This experiment is illustrated schematically in the Supporting Information, Figure S4.

If the enzyme-bound, high specific activity substrate is able to exchange with free, unlabeled substrate in solution at a much faster rate than catalysis, k_off ⪢ k_cat, then the extrapolated line based on the time-dependent chase data will intersect the data point obtained by quenching the pulse at t = t₁, hereafter described as the “quench point.” This result would indicate that a distributive mechanism is operative because catalysis is slow compared to dissociation. However, if the chase data extrapolate to a point at t₁ that is greater than the quench point, catalysis is faster than dissociation, k_cat ⪢ k_off. This would provide strong evidence for a processive mechanism. The difference between the extrapolated chase data point and the quench point at t₁ is directly proportional to the amount of enzyme-bound high specific activity substrate that is converted to product.

Two pulse-chase experiments were performed, each of which involved a pulse of 25 μM DDAH₄Pte[¹⁴C]Glu₁.⁴ The first experiment utilized 25 μM DDAH₄PteGlu₁ as the chase (Figure 8A), while the second experiment utilized 250 μM DDAH₄PteGlu₁ as the chase (Figure 8B). The concentration of 25 μM DDAH₄Pte[¹⁴C]Glu₁ (K_M = 1.65 μM, vide supra) in the pulse was chosen to ensure near saturation of the enzyme while maintaining the production of long-chain polyglutamate products (Figure 3B). The concentrations of DDAH₄PteGlu₁ present in the chase were chosen to enhance (25 μM, Figure 4B) or preclude (250 μM) (28) the formation of long-chain polyglutamates. Upon examination of the results of these experiments, it is immediately apparent that in the 25 μM chase experiment (Figure 8A) the enzyme-bound labeled substrate is converted to product after addition of the chase since the chase data extrapolate to t = t₁, 2 min) at a point that is greater than the quench point. This is not the case in the experiment with the 250 μM chase (Figure 8B) where the chase data extrapolate to t = t₁ at a point that is equal to the quench point. These data are consistent with the concentration-dependent distribution of polyglutamate products previously demonstrated (20, 28).

Figure 8.

Results of the pulse-chase experiments utilizing a pulse of 25 μM DDAH₄ Pte[¹⁴ C]Glu₁ in the presence of 80 pmoles hFPGS with a final concentration of 1.0 μM. A. Quench (□) or chase with 25 μM DDAH₄PteGlu₁ at t = 2 min followed by quenching at times indicated (◻). B. Quench (□) or chase with 250 μM DDAH₄PteGlu₁ at t = 2 min followed by quenching at times indicated (◻). The average of three experiments is plotted with the error bars representing the standard deviation. The difference in the value of the quench points between A and B arises from the experimental conditions, see Experimental, Pulse-chase experiments, 25 μM DDAH₄PteGlu₁ chase and 250 μM DDAH₄PteGlu₁ chase, respectively. The 25 μM DDAH₄PteGlu₁ chase experiment (A) utilized a 10-fold greater concentration, but equimolar amount, of hFPGS in the pulse than was used in the 250 μM DDAH₄PteGlu₁ chase experiment (B). The final concentration of hFPGS was identical during the chase phase in both experiments.

In a detailed analysis of the experimental results, the difference between the extrapolated chase data and the quench point at t₁ observed in Figure 8A was evaluated quantitatively. Additional product formation in the amount of 540 ± 186 pmoles over that predicted from the pulse data had occurred in the 25 μM chase experiment (Figure 8A). When divided by the amount of E·S complex present (75 pmoles, Eq 1), it was determined that 7.2 ± 2.5 glutamate additions per E·S complex occurred under the conditions employed in these experiments. The experimental design of the 25 μM pulse-chase experiment involved the use of a chase in which the concentration of unlabeled DDAH₄PteGlu₁ was equal to that utilized in the pulse solution, 25 μM. This required the use of a volume dilution to chase the labeled DDAH₄PteGlu₁ originally present in the pulse and therefore a concentration of hFPGS in the pulse phase that was 10-fold greater than in the final reaction mixture. The small volume of the pulse solution (8 μL) combined with the high hFPGS concentration (10 μM) resulted in the consumption of nearly 50% of the substrate present in the 25 μM pulse and may explain the larger than expected average ratio of product formation to E·S complex. In contrast, the 250 μM pulse-chase experiment (Figure 8B), utilized an approximately 10-fold lower concentration of hFPGS in the pulse phase. In this experiment, a small volume of a high concentration solution (9.01 mM) of unlabeled DDAH₄PteGlu₁ was added as the chase. This experimental design was predicated on the fact that the final concentration of the DDAH₄PteGlu₁ in the chase was 10-fold greater, 250 μM, than that used in the pulse, 25 μM. In both pulse-chase experiments, the final concentration of hFPGS was identical and the specific activity of DDAH₄Pte[¹⁴C]Glu₁ utilized in the pulse was diluted by 10-fold during the chase phase. The only plausible explanation for an observed [P]:[E·S] > 1 is that the initial substrate in the binary complex undergoes multiple turnovers. Thus, the results of the 25 μM pulse-chase experiment are consistent with a processive mechanism. However, a distributive mechanism is indicated in the experiment that utilized 250 μM DDAH₄PteGlu₁ in the chase. Together, these data provide direct evidence in support of a concentration-dependent processive mechanism for FPGS-catalyzed ligation of multiple glutamate residues to DDAH₄PteGlu₁.

The observed concentration-dependent processivity may be the result of folate heterocycle binding at two different sites. The possibility that FPGS may possess two folate binding sites was first suggested in a computational analysis of ligand binding based on the structure of the enzyme isolated from L. casei (40). This analysis identified a new site that binds monoglutamates with higher affinity than the site originally identified in the crystal structure (41), which is now postulated to bind primarily polyglutamate derivatives. Mathieu et al. provided independent support of this hypothesis based on the structure of a bifunctional protein, folC from E. coli, which harbors both dihydrofolate synthetase (DHFS) and FPGS activities (42). In the folC work, the two binding sites were associated with the binding of two different substrates, 7,8-dihydropteroic acid (DHFS) and 5,6,7,8-tetrahydrofolic acid (FPGS), for the bifunctional enzyme. The two binding site hypothesis is also supported by recent experiments involving “domain swap” chimeric proteins and selected site-directed mutants of both L. casei FPGS and folC (43).

During processive catalysis, the DDAH₄PteGlu₁ substrate could bind at one of two folate binding sites and initial glutamate ligation would occur. Under conditions of low concentration of the folate substrate, the glutamate tail would remain bound while the heterocycle (i.e., pterin or analog) would migrate to the second binding site. Once the heterocycle is bound at the second site, the C-terminus of the DDAH₄PteGlu₂ product would rearrange to position the γ-carboxyl of the terminal glutamate in close proximity to ATP for a second glutamate ligation reaction via the acyl phosphate intermediate (18). Subsequent multiple glutamate ligation reactions would occur with the heterocycle bound at the second site which is postulated to bind preferentially the polyglutamate derivatives (40, 43). A similar translocation of the electrophilic substrate, glutathione (GSH), is postulated in the reaction catalyzed by glutathionylspermidine synthetase (GspS, EC 6.3.1.8) based on crystallographic evidence; i.e., the position of GSH in the dead-end complex, GspS · GSH · ADP vs. the GSH portion of a phosphorylated bi-substrate phosphinic acid-containing pseudopeptide inhibitor (I) in the complex, GspS · I · ADP (44).

This proposed sequence of events is consistent with the observed burst in formation of the first ligation product, DDAH₄PteGlu₂, followed by slow steady-state ligation of additional glutamates. However at high concentration of the folate substrate, it is possible that the monoglutamate substrate could bind to both sites thus preventing the postulated rearrangement just described. Under such conditions, the initial diglutamate product would dissociate without formation of higher polyglutamates, thus resulting in the observed distributive mechanism.

CONCLUSIONS

Three experimental approaches were used to probe the degree of processivity of the reaction catalyzed by the ATP-dependent ligase, hFPGS, involving the multiple glutamylation of DDAH₄Pte[¹⁴C]Glu₁. The formation of products of multiple turnovers in the presence of a large excess of monoglutamate substrate was observed. If the kinetic constants obtained for each polyglutamate, DDAH₄PteGlu_n (n = 2-6), as a substrate (28) are taken into account, it is apparent that continued addition of glutamate is preferred to dissociation. This result is consistent with a processive mechanism. The second experiment involved a substrate trapping approach and probed directly for the formation and kinetic competence of the hFPGS·DDAH₄PteGlu₁ (E·S) complex. A significant commitment to catalysis (39) was observed in the substrate trapping reaction, indicating that a catalytically competent E·S complex is formed and that multiple glutamate ligation reactions occur for each initial E·S complex formed without dissociation of intermediate products. Pulse-chase experiments were then used to confirm the apparent processivity of the hFPGS-catalyzed reaction. Reaction progress data obtained during the chase phase at a low but still saturating folate substrate concentration (25 μM, ~10 K_M), extrapolated to t = t₁ significantly above that of the corresponding quench point. These experiments confirmed that multiple moles of glutamate were added per mole of DDAH₄PteGlu₁ in an E·S complex under these conditions. Additionally, it was determined that the reaction catalyzed by hFPGS is not processive at a supersaturating concentration of folate substrate (250 μM, ~100 K_M) in the chase phase, consistent with the previously observed formation of DDAH₄PteGlu₂ as the major reaction product (78%) at this high concentration (28).

In summary, a combination of time course, substrate trapping, and pulse-chase experiments leads to the conclusion that hFPGS catalyzes multiple ligations of glutamic acid to DDAH₄PteGlu₁ in a processive fashion but the degree of processivity is dependent on the concentration of the DDAH₄PteGlu₁ substrate. If this finding extends to the natural folates as hFPGS substrates, the concentration dependence of processivity in folylpoly-γ-glutamate biosynthesis may have evolved as a mechanism for regulation of folate homeostasis in the cell. Thus, low intracellular concentrations of folate would encourage the formation of long-chain polyglutamate metabolites thereby increasing their retention with in the cell and efficacy as cofactors. However, at high concentrations of folate, cellular retention is not required and long-chain polyglutamate metabolite formation is curtailed, thus allowing folate efflux from the cell while minimizing unnecessary consumption of ATP.