Introduction

Summary Introduction Procedures Results & Discussion References

Over the past decade, the importance of the role of polyglutamylation in the cellular retention of both folic acid derivatives and antifolates has become more widely recognized. As polyglutamyl derivatives, these metabolites are usually the preferred substrates for folate-dependent enzymes, are far less permeable to the plasma membrane, and may also determine subcellular localization (1-3). Prolonged retention of one antifolate compound commonly used as an antineoplastic agent, MTX (4-NH₂-10-CH₃-Pte_glu1), results in more effective inhibition of dihydrofolate reductase and is thought to provide the basis for tumor selectivity (1). The polyglutamylation process is dependent on at least two cellular enzyme activities: folylpolyglutamate synthetase (EC 6.3.2.17) which adds successive glutamate groups to the -glutamyl side chain of folates/antifolates, and GGH (EC 3.4.22.12), which removes (hydrolyzes) the polyglutamate side chain via either an endopeptidase or exopeptidase activity. In many cases, the second enzyme is sequestered into lysosomal bodies, and the rate of transport of folylpolyglutamate into and out of this organelle may also regulate the degree of polyglutamylation (4).

A considerable amount of information on folylpolyglutamate synthetase has been elucidated, including its catalytic properties (2, 5), and after isolation and sequencing of the human cDNA (6), work has progressed to recombinantly express the enzyme and determine features of transcriptional control (7). By contrast, relatively little is known about GGH, although the cDNA sequence for the rat enzyme was recently reported (8), and some information on the properties of this enzyme from various species is known. At the enzymological level, available information and its relevance to human GGH are confused by the fact that enzyme content and specificities (endopeptidase or exopeptidase activity) not only vary among species but also cell and tissue type (2, 9-11). It seems that this group of enzymes falls into one of two classes; one that is lysosomal in origin (10, 12-15), with an acid pH optimum, and another, smaller group that has a neutral or alkaline pH optimum. It is the enzyme from the former group that seems to be of importance in regulating folate and antifolate polyglutamate chain lengths (1, 2, 16, 17). Of this class, the bovine hepatic enzyme and the human intracellular jejunal enzyme have been studied most extensively. Bovine enzyme was purified to (apparent) homogeneity, confirmed to be a glycoprotein with highly reactive sulfydryl groups, and displayed an apparent molecular mass of 108 kDa (12). Although the partially pure human intracellular jejunal enzyme displayed a much lower apparent molecular mass on gel filtration (75 kDa), many of its other properties were similar to those of the bovine enzyme, including the ability to cleave both terminal and internal linkages of folylpolyglutamates (10). This activity preference differs from the strict exopeptidase action found for GGH from human jejunal brush border cells, which probably plays a role in the digestion of dietary folate (14), and very early work with human liver (lysosomal) GGH (18).

Other studies from the laboratories of Galivan and Sirotnak have focused on rodent forms of GGH and, in particular, the secreted forms harvested from cultured cell lines (11, 19). Interestingly, the enzymological properties of the secreted GGH are not different from the cellular (lysosomal) form. Purified GGH secreted from rat H35 hepatoma cells was shown to have extensive carbohydrate modification, and the deduced amino acid sequence from the cDNA sequence indicates a molecular mass of 33.4 kDa for the peptide component of this enzyme species (8).

Clearly more information on the human enzyme or enzymes is necessary to better define the role of GGH in the regulation of intracellular polyglutamate levels (and therefore cellular retention) of folates and antifolates in human cells. Here, we report the partial purification and characterization of GGH from the human fibrosarcoma cell line HT-1080. Purified GGH preparations were also used to test a broad range of potential inhibitor compounds and certain glutamine antagonists demonstrated inhibitory activity. One of these agents, DON, was found to enhance the cellular retention and cytotoxicity of MTX.

	Experimental Procedures

Summary Introduction Procedures Results & Discussion References

Materials. Polyglutamylated derivatives of pteroyl (Pte), 4-NH₂-10-CH₃-Pte, and pAB poly--L-glutamates were purchased from Dr. B. Schirks Laboratories (Jona, Switzerland). The only exception to this was MTX, which was from Lederle Laboratories (Pearle River, NY). [³H]MTX (20 Ci/mmol) was purchased from Moravek Biochemicals (Brea, CA). Reagents for enzymatic assays of -hexosaminidase and dihydrofolate reductase were from Sigma Chemical (St. Louis, MO) or as described (20). Affinity media Reactive Green 5, organomercurial agarose, gel filtration molecular mass standards, and reagents for cell cytotoxicity measurement were also from Sigma. Sephacryl S-200 was from Pharmacia (Piscataway, NJ). Several compounds used in the inhibition screen were kindly donated: 2-mercaptomethylglutaric acid was from Dr. T. I. Kalman (State University of New York, Buffalo, NY), ICI 198583--D-glu was from Dr. A. L. Jackman (Institute of Cancer Research, Surrey, UK). All other chemicals were of the highest purity available.

Cell culture and preparation of extracts. Various cancer cell lines, including the soft tissue sarcoma line HT-1080, were obtained from the American Type Culture Collection (Rockville, MD) and cultured in RPMI-1640 containing 2 mM glutamine, 10% fetal bovine serum, and antibiotics. The cytotoxic effect of various agents on the HT-1080 line, using variable concentrations and exposure times, was measured by cell counting or the 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino)carbonyl]-2H-tetrazolium hydroxide assay (21).

Activity assays. GGH activity was measured at 37°, pH 4.5, using 50 µM 4-NH₂-10-CH₃-Pte_glu5 as substrate (unless otherwise stated) and standard assay buffer consisting of the MTEN buffer system (22) with 1 mM ZnCl₂ and 2 mM DTT. Portions of the assay mix were sampled at various time points, placed in a boiling water bath for 5 min, and then centrifuged at 14,000 × g. Supernatants were used directly in HPLC analysis to resolve and quantify products as previously described (11). During the course of this work, a capillary electrophoresis procedure was developed for use in quantifying substrate and reaction products of GGH assays (23) and used primarily in inhibitor screening experiments. GGH activity was calculated from the percentage of substrate degraded per unit time (23). One unit is defined as the amount required to convert 1 µmol of substrate/min. Specific activity values (units/mg) were calculated from assay time points where < 20% of substrate was consumed.

Secretion of GGH in cultured HT-1080. Seeded HT-1080 cells were grown to early log phase; then, a portion of medium was taken and snap-frozen in liquid N₂. The remaining culture was spiked with 10 mM NH₄Cl, and additional samples of media were taken at days 1, 2, and 5. Medium for a duplicate flask of cells (without added NH₄Cl) was also sampled at these times. Total protein and enzymatic activities for GGH, -hexosaminidase, and dihydrofolate reductase were measured in all samples. The background activities contributed from media alone (close to negligible) was subtracted from all measurements.

Purification of GGH from HT-1080. Three to 4 ml of clarified crude HT-1080 lysate was dialyzed against buffer A (25 mM Tris·HCl, pH 7.5) and then applied to a 6-ml column of Reactive Green 5 gel previously equilibrated with identical buffer. Throughout the run, flow rate was maintained at 0.8 ml/min, eluate A₂₈₀ was continuously monitored, and 3-ml fraction volumes were collected. NaCl solutions prepared in buffer A were used to desorb enzyme, and approaches were used in different runs: either (a) a single linear gradient of 0-1.5 M NaCl (100 ml total volume) or (b) multiple step gradients and washing phases. In the second approach, two linear gradients were used (40 ml total volume each) with an intervening wash step of salt strength equivalent to the maximum achieved in the first gradient.

Characterization of GGH. Properties of GGH were investigated using preparations of purity of >3 units/mg (i.e., purified 1500-fold from crude extract); this included experiments examining the effect of anions, detergents, and urea on GGH activity. The S-200 column and conditions used for estimation of native molecular mass of GGH was the same as that for purification as described above and in the figure legend. Isoelectric focusing was performed using a Hoefer gel unit (GT 1) and operated, essentially, under conditions recommended by the manufacturer (Hoefer Scientific Instruments, San Francisco, CA).

Substrate kinetic analysis. Enzyme was incubated in standard GGH assay buffer (37°) with the different substrates at concentrations (final) of 0.2-250 µM. Incubation times varied from 6.5 min (for 0.2 µM) to 150 min (for >100 µM). Reactions were stopped by boiling, and polyglutamate reaction products were resolved and quantified as described above. Kinetic parameters were subsequently determined by fitting to data the Michaelis-Menten equation using nonlinear, least-squares regression analysis (25).

In vitro inhibitor studies. Inhibitor screening was performed using the standard assay (50 µM 4-NH₂-10-CH₃-Pte_glu5) and ~33 µunits of GGH/assay tube. Agents were tested at several concentrations of 1 mM when solubility permitted. More formal inhibition analyses were undertaken for the glutamine antagonists using at least seven concentrations of each compound and a 30-min preincubation of inhibitor with enzyme before commencing the assay by adding substrate. K_{i app} values were obtained by fitting (25) to data the following equation: Inhibited Velocity = V_max/{1 + ([I]/K_{i app})}.

Analysis of polyglutamate chain formation. HT-1080 cells (2-5 × 10⁷) were incubated in complete medium containing 10 µM [³H]MTX at 37° for 24 hr or as described in text and figure legends. After the specified treatments, cells were harvested and suspended in 500 µl of boiling 50 µM Na phosphate, pH 5.5, and then placed in a boiling water bath for 5 min. Extracts were centrifuged at 20,000 × g for 10 min, and supernatants were analyzed by HPLC as previously described (26).

	Results and Discussion

Summary Introduction Procedures Results & Discussion References

Activity and secretion of GGH from cultured cell lines. In all instances, the HPLC profile of GGH reaction products for assays (4-NH₂-10-CH₃-Pte_glu5 as substrate) was consistent with an exopeptidase mode of action (Fig. 1A). This mode of activity is characterized by successive and sequential cleavage of end -glutamyl residues, with cleavage stopping at the one remaining glu residue (i.e., 4-NH₄-10-CH₃-Pte_glu1 when standard substrate is used). We (28) and, more recently, others (29) have reported this activity for human GGH from cultured cell lines, although the possibility that other activity forms exist for other human cell types or tissues (e.g., tissues involved in absorptive processes) cannot be ruled out at this stage. In this study, a strict exopeptidase mode of action was also found for HT-1080 GGH at various levels of purity and for other poly--glutamate substrates [i.e., Pte_glu4 and pAB_glu5 (vida infra)]. This clearly contrasts with data obtained from studies of many rodent tissues and rodent cultured cell lines. For these species, the enzyme has displayed either a mixed or an endopeptidase mode of cleavage for folate and antifolate poly--glutamates (11, 19). Human GGH is therefore distinct from rodent GGH in this respect.

View larger version (44K): [in this window] [in a new window]   Fig. 1.   Profile of GGH reaction products and secretion of enzyme from HT-1080 cells. A, Substrate 4-NH4-10-CH3-Pteglu5 (i.e., MTXglu4) was incubated with HT-1080 GGH and then the assay mix was periodically sampled and products were resolved by reverse-phase liquid chromatography. Glu5-Glu1, derivatives of 4-NH4-10-CH3-Pte. This time course for substrate/product profile is consistent with an exopeptidase mode of action. B, Medium from cultured HT-1080 cells at early log growth was periodically assayed for GGH activity over 7 days either () with or () without the addition of 10 mM NH4Cl at day 2 (arrow).

Purification of GGH from HT-1080 cells. GGH bound avidly to an agarose-immobilized dye matrix (Reactive Green 5) at pH 7.5 and could be eluted as a sharp band of activity by application of a salt gradient. Multiple protein components were detected on SDS-PAGE analysis of the eluted fractions with highest specific activity, nevertheless, this proved to be an ideal first step in the purification of GGH with good yield (51-82%, n = 4) and substantial purification (110-386-fold, n = 4). Further purification of HT-1080 GGH was achieved by passage on a organomercurial column (thiol elution) and Sephacryl-200 chromatography as summarized in Table 1. For both of these steps, activity behaved homogeneously (i.e., eluted as a single peak); however, omission of thiols from the S-200 equilibration buffer (in initial runs) resulted in an extremely broadly eluting activity peak. In addition, total activity recovered from the gel filtration step (see Table 1), and stability of the resulting preparation was low even when thiols and Zn²⁺ were present.

Properties of HT-1080 GGH. Concentrated partially pure GGH was stable at 75° for a 6-month period and demonstrated only ~6% activity loss after five freeze (196°)/thaw (37°) cycles. Several general properties in relation to GGH activity were noted. (a) Activity was highly pH dependent with an acidic optimum in the range of 4- 6.5 (Fig. 2A). This is a similar finding to that reported for murine and bovine species of GGH (12, 19) and consistent with its assumed lysosomal localization. (b) Although activity was routinely assayed with buffer containing ZnCl₂ and DTT, omission of DTT often diminished activity by 1-15%. In no instance did omission of ZnCl₂ result in lower activity, even with the purist preparations and after exhaustive dialysis. When activation by DTT (2 mM) was apparent, an equal extent of activity enhancement could be obtained with -mercaptoethanol or cysteine but not cystine or glutathione (data not presented). When the MTEN buffering component was replaced by citrate, succinic, and 2-(N-morpholino)ethanesulfonic acid buffers, each at 50 mM, pH 4.5, and also containing DTT and ZnCl₂, identical catalytic rates were observed. (c) GGH was inhibited by anions or polyanions as previously noted for other species of enzyme (2), presumably by binding a putative polyglutamate anion receptor site on the enzyme (18). Activity was completely abolished by 2 mM and 1 mM Na acetate at pH 4.5 and 5.5, respectively.

View larger version (17K): [in this window] [in a new window]   Fig. 2.   A, pH dependency of purified GGH. Reaction rates were measured at 37° in MTEN buffer using the standard assay procedure. Individual data points represent the average activity (±1 standard error) for three independent measurements. Appropriate assay controls were included to ensure that the lower activities measured at pH extremes reflect genuine reduced catalytic ability rather than enzyme inactivation. B, Estimation of the molecular mass of native GGH by gel filtration. Partially pure GGH was applied to a column of S-200 gel previously equilibrated with a buffer consisting of 20 mM Tris·HCl, pH 7.0, 10 mM NaCl, 2 mM -mercaptoethanol, and 1 mM DTT. Enzyme activity was eluted with the same buffer. Standard proteins used to calibrate the column are depicted (), as is the activity elution position of GGH from several runs. The interpolated average molecular mass for GGH is 83.0 ± 9.6 kDa. C, Isoelectric focusing of GGH. Partially pure enzyme was focused in a tube gel (polyacrylamide; 4% T, 0.5% C) containing carrier ampholytes pH 3.5-10 (2%) and pH 5-7 (3.1%). Focusing conditions and activity determinations were carried out as described in Experimental Procedures. The pH gradient of the gel was calibrated by measuring pH of macerated gel segments (dissolved in minimal water) from a second blank gel ran in tandem. The three major activity regions correspond to pI values of 6.2-6.4, 5.6-5.8, and 7.3-7.8.

Physicochemical properties. The molecular mass of native HT-1080 GGH was estimated on four occasions by gel filtration chromatography on Sephacryl S-200 in the presence of 2 mM -mercaptoethanol and 1 mM DTT. Values of apparent molecular mass obtained by interpolation of the calibration plot (Fig. 2B) ranged from 81.0 to 93.9 kDa; the average was 83.0 ± 9.6 kDa (±1 standard deviation). Omission of thiols from buffer resulted in a spreading of activity elution on initial trials, a phenomenon that did not seem to be related to the purity of material analyzed. This molecular mass estimate for HT-1080 GGH differs considerably from those reported for the partially pure jejunal enzyme [75 kDa, (10)], bovine liver GGH [108 kDa, (12)], and the secreted rat hepatoma enzyme [120 kDa, (19)]; each determined by gel filtration methodologies. For the latter enzyme species, more recent studies (8) with pure preparations (without deglycosylation) indicated a diffuse band of 55 kDa on SDS-polyacrylamide gel electrophoresis analysis, which is suggestive of quaternary structure for at least this rodent GGH species. Quaternary structure, extent of glycosylation, and how these factors contribute to differing apparent molecular mass estimates for various eukaryotic sources of GGH have yet to be determined.

Enzyme kinetic parameters and substrate specificity. The catalytic activity of HT-1080 GGH was measured with varied concentrations of Pte_glu2, 4-NH₂-10-CH₃-Pte_glu2, pAB_glu2, and the corresponding glu₅ derivatives. The amount of enzyme and the time points selected in assays were chosen to avoid substantial substrate depletion and to permit the generation of predominantly glu₄ product derivatives only in the case of glu₅ substrates. All substrates conformed to Michaelis kinetics, yielding rectangular hyperbolic curves when reaction velocities were plotted against substrate concentration. Iterative fitting of the Michaelis-Menten equation to data enabled estimation of K_m and V_max parameters, as listed in Table 2. Most notably, the enzyme showed preference toward the longer-chain derivatives with 15-42-fold lower K_m values for the glu₅ versus glu₂ forms for each of the substrate types analyzed. A similar observation of distinct preference for the longer-chain -glutamate substrates has been noted in other studies (some are reviewed in Ref. 2), including the early work with partially pure human liver enzyme [although only in a qualitative fashion (18)] and more recent work by Sirotnak et al. (11) with murine GGH. In the latter study, analyses of GGH activity in crude preparations of L1210 and mouse small intestine revealed respective 3- and 7.7-fold decreases in K_m values for the glu₄ derivatives of 4-NH₂-10-CH₃-Pte versus the glu₂ derivative (11). It is also worthy of note that in this same study (11), enriched GGH from mouse sarcoma 180 cells and mouse intestine extract (through ammonium sulfate fractionation) displayed K_m values of 19.7 and 86.4 µM, respectively, for 4-NH₂-10-CH_3-Pte_glu2. These values encompass the value obtained in the current study with human enzyme using identical substrate (52.6 µM; Table 2).

Inhibition of HT-1080 GGH. 2-Mercaptomethylglutaric acid was reported to be inhibitory toward partially pure chicken pancreas GGH but not lysosomal hog kidney GGH (17, 35). Experiments performed in the current study with HT-1080 GGH failed to show any in vitro inhibition of activity by this agent at 250 µM, at least at pH 4.5. To identify inhibitors, we used initially the quantitative HPLC assay and later the high through-put capillary electrophoresis assay to screen a series of potential inhibitor compounds selected on the basis of structural considerations. Many of these compounds were folate or antifolate derivatives with unusual glutamate modifications or substitutions, although none selected displayed significant inhibition,¹ including ICI 198583--D-glu. However, the "glutamine antagonists azaserine, acivicin, and DON, all of which are well-characterized inhibitors of -glutamyltranspeptidase [EC 3.3.2.2 (36)], were found to be inhibitory toward the human GGH. Without regard to the mode of inhibition, or the presence of substrate, and using inhibition reaction rates after a 30-min preincubation with enzyme, respective K_{i app} values of 5.7, 0.85, and 0.05 mM were derived for these agents (Fig. 3A).

View larger version (17K): [in this window] [in a new window]   Fig. 3.   In vitro inhibition of purified GGH. A, Dixon plots of data obtained for () azaserine, () acivicin, and () DON inhibition of HT-1080 GGH. Different concentrations of agent were preincubated with enzyme for 30 min before assessment of activity using the standard GGH assay protocol (substrate 4-NH2-10-CH3-Pteglu5 at 50 µM). B, Time dependency for inhibition by DON. GGH (30 µunits) was incubated with 50 µM DON for varying times at 37° and then assayed as above (). Inhibition data are presented as the percentage value of control activity measurements for which preincubation was carried out without added agent ().

Cell culture studies of GGH inhibition by DON and its effect on MTX action. The cytotoxicity of DON toward the HT-1080 line by DON alone was determined before assessing the potential of this agent to enhance MTX retention and action. Cell survival after a 24-hr exposure to DON at concentrations of 500 µM was measured using the 2,3-bis(2-methoxy-4-nitro-5-sulfophenyl)-5-[(phenylamino)carbonyl]-2H-tetrazolium hydroxide cytotoxicity assay. For the HT-1080 cell line, < ~5% cell death was observed at 10 µM DON, so this was the maximal concentration adopted in subsequent studies.

Concluding comments. The properties of purified human GGH reported in the current study have similarities to the few characteristics noted for human liver GGH but differ in several respects to those reported for rodent, bovine liver, and human jejunal and brush border enzymes. GGH for cultured human cell lines (excluding those derived from tissues with absorptive functions) may be predominantly one species, an exopeptidase that is highly sequestered into the lysosome. The only indication of possible GGH variants identified in the current work could also be explained on the basis of variable post-translational modification of the enzyme (i.e., glycosylation).