Expression and Purification of Human γ-Glutamylcysteine Synthetase

PROTEIN EXPRESSION AND PURIFICATION ARTICLE NO.

13, 268 –276 (1998)

PT980897

Expression and Purification of Human g-Glutamylcysteine Synthetase1 Ila Misra and Owen W. Griffith2 Department of Biochemistry, Medical College of Wisconsin, Milwaukee, Wisconsin 53226

Received January 22, 1998, and in revised form April 2, 1998

INTRODUCTION

g-Glutamylcysteine synthetase (g-GCS) catalyzes the ATP-dependent ligation of L-glutamate and L-cysteine to form L-g-glutamyl-L-cysteine; this is the first and rate-limiting step in glutathione biosynthesis. Inhibitors of g-GCS such as buthionine sulfoximine are widely used as tools for elucidating glutathione metabolism in vivo and as pharmacological agents for reversing glutathione-based resistance to chemotherapy and radiation therapy in certain cancers. Although g-GCS is readily isolated from rat kidneys, future drug design efforts are better based on structure– activity relationships established with the human enzyme. We report here the coexpression in Escherichia coli BL21(DE3) of the human g-GCS catalytic (heavy) subunit and regulatory (light) subunit using pET-3d and pET-9d vectors, respectively. Intracellular assembly of the holoenzyme occurred without difficulty, and levels of expression were acceptable (;32 mg holoenzyme/100 g cells). Recombinant human g-GCS was purified to homogeneity in an overall yield of 45% by ammonium sulfate fractionation followed by sequential chromatography on Q-Sepharose ion-exchange, Superdex 200 gel filtration and ATP-affinity resins. Trace amounts of E. coli g-GCS were removed by immunoaffinity chromatography. The specific activity of the isolated enzyme was >1500 units/mg, comparable to the best preparations from rat kidney. The Km values for L-glutamate, L-cysteine, L-g-aminobutyrate (an L-cysteine surrogate), and ATP are 1.8, 0.1, 1.3, and 0.4 mM, respectively. Recombinant human g-GCS, like native rat g-GCS, is feedback inhibited by glutathione and is potently inhibited by buthionine sulfoximine and cystamine. © 1998 Academic Press

1 These studies were supported in part by NIH Grant CA77233 (OWG) and by an MCW Cancer Center Community Advisory Board Fellowship (IM). 2 To whom correspondence should be addressed. Fax: (414) 4566510. E-mail: [email protected].

268

g-Glutamylcysteine synthetase (g-GCS)3 catalyzes the reaction L-Glutamate

1 L-Cysteine 1 ATP 3

L-g-Glutamyl-L-Cysteine

1 ADP 1 Pi [Reaction 1]

the first and rate-limiting step in glutathione (L-gglutamyl-L-cysteinylglycine, GSH) biosynthesis (1, 2). The mammalian enzyme is a heterodimer composed of a catalytic (heavy) subunit, Mr ; 73,000, and a regulatory (light) subunit, Mr ; 31,000. Association of the subunits is maintained by at least one disulfide bond in addition to noncovalent forces (3). Studies with both native and recombinant rat g-GCS demonstrate that the isolated heavy subunit is competent to carry out Reaction 1 as well as all known partial reactions associated with the g-GCS holoenzyme (2– 4). However, when the light subunit is present, the Km for L-glutamate is reportedly reduced from 18 to 1.4 mM and the Ki for nonallosteric inhibition by GSH is increased from 1.8 to 8.2 mM (4, 5). Because intracellular levels of L-glutamate and GSH are 1– 3 and 1– 10 mM, respectively, it was concluded that presence of the light subunit is necessary for full activity of g-GCS in vivo (4, 5). Consistent with the central role played by GSH in cellular defenses against reactive oxygen species and electrophiles, GSH and the enzymes for its synthesis have been found in every mammalian cell line and tissue examined (2). Intracellular levels of GSH are regulated in part by the amount of g-GCS present; feedback inhibition of g-GCS by GSH, L-cysteine availability, and the rate at which GSH is transported out of the cell also contribute to GSH homeostasis (1, 2). In 1987 Vistica and colleagues (6) reported that a subAbbreviations used: g-GCS, g-glutamylcysteine synthetase; GSH, glutathione; BSO, L-buthionine-SR-sulfoximine; IPTG, isopropyl-b-Dthiogalactopyranoside. 3

1046-5928/98 $25.00 Copyright © 1998 by Academic Press All rights of reproduction in any form reserved.

HUMAN g-GLUTAMYLCYSTEINE SYNTHETASE

population of L1210 leukemia cells maintained a 1.5to 2-fold higher than normal GSH level as a result of g-GCS overexpression and that the elevated GSH level accounted for the resistance of those cells to melphalan, a DNA cross-linking agent used in cancer chemotherapy. Similar observations were subsequently made with other tumor cell lines and with fresh tissue samples obtained from tumor-bearing animals and patients (7, 8). It is now clear that overexpression of g-GCS can play an important role in the resistance of tumors to radiation therapy and to chemotherapy with redox cycling drugs (e.g., doxorubicin) and bifunctional alkylating agents (e.g., cis-platin, melphalan, or cyclophosphamide) (7, 8). Animal studies and preliminary clinical trial results suggest further that administration of g-GCS inhibitors such as buthionine sulfoximine (BSO) can in some systems reverse resistance to therapy based on overexpression of g-GCS (1, 2, 7–9). Although BSO is currently in clinical trial for malignant melanoma and ovarian cancer as an adjuvant to melphalan (8, 10 –12), it is not sufficiently potent to allow complete g-GCS inhibition to be easily maintained in vivo. In addition, BSO is rapidly excreted in the urine, necessitating administration of large amounts of drug (12–14). In order to develop and adequately test novel, improved g-GCS inhibitors, we required a source of the relevant target enzyme, human g-GCS. Because R. T. Mulcahy and co-workers had previously reported the cloning of the human liver heavy (15) and light (16) g-GCS subunits and were willing to provide cDNA samples to us, we have undertaken the expression and purification of human g-GCS. We report here the results of those studies. MATERIALS AND METHODS

Materials E. coli BL21(DE3) and the expression vectors pET-3d and pET-9d were purchased from Novagen (Madison, WI). E. coli strain DH5a was obtained from Bethesda Research Laboratory (Gaithersburg, MD). Restriction enzymes were purchased from New England Biolabs (Beverly, MA), and Pfu DNA polymerase and deoxynucleotides were purchased from Stratagene (La Jolla, CA). Deoxyoligonucleotides were synthesized by Operon Technologies, Inc. (Alameda, CA). Ampicillin, kanamycin, and isopropyl-b-D-thiogalactopyranoside (IPTG) were obtained from United States Biochemical (Cleveland, OH). L-Buthionine-S,R-sulfoximine (BSO) was prepared as described previously (17). All other biochemical reagents were purchased from Sigma (St. Louis, MO). The cDNA for the heavy and the light subunits cloned in pSK-Bluescript vectors were a generous gift from R. T. Mulcahy (Department of Human Oncology, University of Wisconsin Medical School, Madison, WI). Rabbit antibodies to rat g-GCS were prepared by Dr.

269

Jihong Han of this laboratory using enzyme prepared essentially as described by Seelig and Meister (18). Methods Qiagen (Chatsworth, CA) plasmid mini kits were used to isolate plasmid DNA from bacterial cultures. Qiaex reagents and protocols, also from Qiagen, were used for the extraction of nucleic acid fragments from agarose gels. DNA sequencing was performed on an ALF automated sequencer using autoread kits and protocols provided by Pharmacia (Milwaukee, WI). Amino acid composition and N-terminal sequence analyses were performed by the Protein and Nucleic Acid Shared Facility of the Medical College of Wisconsin. The ratio of light to heavy g-GCS subunits in the final preparation was calculated from quantitative amino acid data obtained following hydrolysis of subunits extracted from an SDS–PAGE gel. The amount of subunit was calculated from the most reliably determined amino acids (.87% agreement with cDNAbased expectation); these were mainly the highly stable aliphatic amino acids and, for the heavy subunit, Tyr and Phe. The latter were too rare in the light subunit to accurately quantitate. Construction of the Expression Plasmids The pSK-GCS50 (15) and pSK-sGCS-7 (16) plasmids provided by Dr. Mulcahy were used to construct expression plasmids for the heavy and light subunits of human g-GCS, respectively. Inserts for expression of nonfusion proteins using the pET-3d and pET-9d expression vectors (19) were designed using a naturally occurring NcoI site at the 59 terminus of the coding sequence of each subunit and a BamHI restriction site engineered onto the 39 end of each cDNA. Since these restriction sites were not unique in either cDNA, a multiple fragment ligation strategy was devised (Fig. 1). Although most of the necessary fragments could be directly obtained by appropriate restriction digestion of the original plasmids, the 39 portion of each cDNA had to be generated by PCR in order to include the BamHI site. Amplification of the 39 fragment needed for the heavy subunit cDNA was achieved using a complementary upstream primer (59-CTGATGAAATGAATTATAGCCTTA-39) and a downstream primer containing the two substitutions in the noncoding region, indicated in boldface type, necessary to create the BamHI site (59GAGAGGCATGGTACTGGATCCAGTTCGTCAATA39). Similarly, amplification of the 39 fragment needed for the light subunit was achieved using a complementary 59 primer (59-AAACTTAGTTCAGAGCAAAAAGA39) and a downstream primer containing two substitutions in the noncoding region (59-ATTACAGGTAAGTTAGGATCCTAAGTCAGTTAAGA-39). Amplification of the 39 fragments of each cDNA was carried out using Pfu

270

MISRA AND GRIFFITH

DNA polymerase and 25 cycles of denaturation (94°C for 1 min), annealing (52°C for 50 s), and extension (72°C for 50 s), followed by a cycle of denaturation (94°C for 1 min), annealing (52°C for 50 s), and extension (72°C for 5 min). The fragments thus produced were separated by agarose gel electrophoresis, excised from the gel, and isolated. The PCR-amplified 192-bp fragment corresponding to the 39 portion of the heavy subunit was cleaved with BsmI and BamHI restriction endonucleases to generate the required 106-bp fragment; the 402-bp fragment corresponding to the light subunit was cleaved with BglI and BamHI restriction endonucleases to produce the necessary 99-bp fragment. Ligation of the gel-purified fragments produced the required expression vectors. Thus, the 493-bp NcoI– BstXI, 1354-bp BstXI–BsmI, 106-bp BsmI–BamHI fragments were mixed with the 4601-bp BamHI–NcoI DNA fragment of pET-3d for a four-fragment ligation to construct the heavy subunit expression plasmid (pHGCSH-3d, Fig. 1A). Similarly, the light subunit expression plasmid (pHGCSL-9d, Fig. 1B) was constructed by ligation of the 736-bp NcoI–BglI and 99-bp BglI–BamHI coding fragments with the 4302-bp BamHI–NcoI fragment from pET-9d. Competent DH5a cells were transformed with the individual ligation mixtures, and ampicillin- or kanamycin-resistant cells were selected for the heavy or light subunit, respectively. Plasmids were purified from single colonies. Correct fragment assembly was established by restriction mapping, and the sequence of the PCR-amplified region of each plasmid was confirmed by chain-termination sequencing (20). The validated plasmids for the heavy and light subunits were cotransformed into BL21(DE3), and cells resistant to both ampicillin and kanamycin were selected. Expression of Human g-GCS Single isolated colonies of transformed BL21(DE3) cells were inoculated into 23 YT medium (21) containing 200 mg/ml ampicillin, 100 mg/ml kanamycin and grown with shaking at 30°C. Expression of human g-GCS was induced by addition of 2 mM IPTG when A600 was ;1.5. At that time, additional ampicillin (200 mg/ml) and kanamycin (100 mg/ml) were added to ensure growth of only plasmid-containing cells, and glycerol (2% final concentration) was added because it improved expression. The culture was allowed to grow until A600 was ;3.2, and the cells were then harvested by centrifugation at 5000g for 15 min. Purification of Human g-GCS The entire procedure was carried out at 4°C. Bacterial pellets were suspended in 345 ml of homogenization buffer (50 mM Tris–HCl, pH 7.4, 5 mM L-glutamate, 5 mM MgCl2, 10% glycerol, 5 mg/ml bovine

pancreas ribonuclease, 5 mg/ml bovine pancreas deoxyribonuclease, and a protease cocktail containing 1 mg/ml each of leupeptin, pepstatin A, antipain, chymostatin, bestatin, and trypsin inhibitor). The cells were broken using a French Press at 16,000 psi, and a supernatant fraction was obtained by centrifugation of the crude homogenate at 30,000g for 45 min. That solution was fractionated by adding solid ammonium sulfate to 40% saturation (226 g/liter solution), stirring for 30 min, and discarding the pellet obtained by centrifugation (30,000g for 40 min). The supernatant was brought to 60% saturation with ammonium sulfate by adding 120 g/liter, and after stirring 30 min, the pellet was collected by centrifugation and dissolved in ;200 ml of isolation buffer (50 mM Tris–HCl buffer, pH 7.4, containing 5 mM L-glutamate, and 5 mM MgCl2). That solution was dialyzed overnight against 8 liters of isolation buffer and was then applied to a 2.5 3 40 cm column of Q-Sepharose anion-exchange resin (Pharmacia) equilibrated with the same buffer. After the resin was washed with isolation buffer until the A280 was ,0.25, g-GCS was eluted using a linear gradient established between 0 and 0.4 M NaCl in isolation buffer. Activity-containing fractions from the Q-Sepharose chromatography were pooled and g-GCS was precipitated by addition of ammonium sulfate (500 g/liter). The pellet obtained by centrifugation was dissolved in 20 mM imidazole buffer, pH 7, containing 120 mM NaCl, 5 mM L-glutamate, and 5 mM MgCl2, and that solution was applied to a 1.6 3 85 cm Superdex 200 gel filtration column (Pharmacia) equilibrated and eluted with the same buffer. Fractions with the highest specific activities were pooled and dialyzed against 8 liters of 50 mM Na1 Mops buffer, pH 7.0, containing 5 mM L-glutamate. The dialyzed solution was made 1.25 mM in MnCl2 and immediately applied to a 1 3 8 cm column of ATP-agarose (C-8 attachment; Sigma) equilibrated with the same buffer. The column was washed in sequence with equilibration buffer (;25 ml), equilibration buffer containing 0.25 M NaCl and 10% glycerol (;20 ml), 50 mM Na1 Mops buffer, pH 7.0, containing 5 mM L-glutamate and 5 mM MgCl2 (;15 ml), and the same buffer containing 1 mM ATP and 0.2 M NaCl (;20 ml); the enzyme elutes in the last buffer. Fractions containing activity were immediately pooled and dialyzed against 50 mM Tris–HCl buffer, pH 7.4, containing 150 mM NaCl. Immunoaffinity Chromatography Polyclonal antibody against E. coli g-GCS was generated by immunizing a rabbit with E. coli enzyme purified to homogeneity essentially as described by Huang et al. (22). The immunoaffinity column used to remove residual E. coli g-GCS from the preparations of human g-GCS was constructed using an immobilized

HUMAN g-GLUTAMYLCYSTEINE SYNTHETASE

protein A IgG purification kit from Pierce (Rockford, IL). Briefly, 5 ml of immune serum was diluted with 5 ml of ‘‘antibody-binding buffer’’ provided in the kit and mixed with 1 ml of protein A–Sepharose preequilibrated with binding buffer. After being shaken gently for 1 h, the resin was collected by centrifugation and transferred to small column. The resin was next washed with binding buffer until A280 reached baseline, and it was then equilibrated with cross-linking buffer (0.2 M triethanolamine HCl, pH 8.2). The equilibrated resin was transferred to a tube, mixed with 2 ml of dimethyl pimelimidate solution (6.6 mg/ml in cross-linking buffer), and rocked gently to allow covalent cross-linking of bound IgG to the resin. After 1 h, the resin was washed extensively and sequentially with water, 0.1 M ethanolamine HCl, pH 8.2, water, and 1 M NaCl. The completed resin was equilibrated with 50 mM Tris–HCl, pH 7.4, containing 150 mM NaCl and stored at 4°C until used. Human g-GCS (;12 mg) purified through the ATP affinity chromatography step was mixed with immunoaffinity resin (;1 ml) and allowed to stand for 30 min with occasional mixing. The resin was then returned to a column and washed with 50 mM Tris–HCl buffer, pH 7.4, containing 150 mM NaCl. Activity-containing fractions were pooled, concentrated using a Centricon 30 microconcentrator (Amicon, Beverly, MA), and dialyzed against 20 mM imidazole HCl buffer, pH 7.4, containing 1 mM EDTA. If storage of the enzyme for more than a few days was contemplated, glycerol was added to a final concentration of 25% to stabilize the activity. Enzyme Assays Human g-GCS activity was determined spectrophotometrically on the basis of ADP formation using a pyruvate kinase–lactate dehydrogenase-based coupled assay system essentially as described previously (17, 18). The reaction mixtures contained, in a final volume of 1.0 ml, 150 mM Tris–HCl buffer, pH 8.2, 100 mM KCl, 0.3 mM EDTA, 40 mM MgCl2, 5 mM ATP, 15 mM L-a-aminobutyrate, 20 mM L-glutamate, 10 mM phosphoenolpyruvate, 0.3 mM NADH, 14 IU of pyruvate kinase, and 38 IU of lactate dehydrogenase. Reaction was initiated by addition of g-GCS, and NADH oxidation, which is stoichiometrically equal to ADP formation, was monitored at 340 nm. One unit of g-GCS activity is the amount necessary to form 1 mmol of ADP per hour under the conditions of the assay. Specific activity is expressed as units per milligram of protein. Protein was determined using the Bradford assay (23) with bovine serum albumin as standard. Initial velocity measurements for determination of Km values for L-glutamate, L-cysteine, L-a-aminobutyrate, and ATP were carried out in similar reaction

271

mixtures except that concentrations of the nonvaried substrates were fixed at 10-fold of their respective Km values as estimated in preliminary studies. Concentration ranges for the varied substrates were as follows: 0.9 – 20 mM L-glutamate; 0.06 – 1 mM L-cysteine; 0.8 –20 mM L-a-aminobutyrate, 0.2– 4 mM ATP. Inhibition by GSH was determined by adding fixed amounts of GSH (0, 2, and 4 mM) to reaction mixtures in which L-glutamate was the varied substrate. Inhibition by BSO was determined by preincubating 10 –50 units of g-GCS with 100 mM Tris–HCl buffer, pH 8.2, containing 5 mM ATP, 20 mM MgCl2, and 100 mM BSO at 37°C for 5 min and then measuring residual activity using the standard spectrophotometric assay. Inhibition by cystamine was tested similarly by preincubating 10 –50 units of human g-GCS in 100 mM Tris–HCl buffer, pH 8.2, with 100 mM cystamine at 37°C for 10 min and measuring residual activity using the standard spectrophotometric assay. In both studies, control incubations lacked inhibitor. RESULTS AND DISCUSSION

Expression of Human g-GCS In earlier studies by Huang et al. (4), rat g-GCS was expressed in E. coli and purified to homogeneity; those investigators were able to obtain expression of ;38 mg of holoenzyme/100 g cells using a pT7-7 coexpression plasmid containing two T7 promoters in opposite directions, each immediately followed by either the heavy or the light subunit coding sequence. We initially attempted a somewhat similar coexpression strategy using a pET-3d vector in which a single T7 promoter was followed sequentially by the human light and heavy subunit coding sequences. This bicistronic vector resulted in expression with good stoichiometry between light and heavy subunits, but overall expression levels were poor. We next constructed individual, separately selectable expression vectors for each subunit as described under Materials and Methods (Fig. 1). Cotransformation of these into E. coli BL21(DE3) gave acceptable levels of holoenzyme expression; in five preparations we have consistently obtained ;32 mg holoenzyme/100 g cells calculated on basis of total activity and a specific activity of 1500 units/mg for pure g-GCS. Critical factors in ensuring good results include the use of freshly transformed cells, attention to the cell density at which the culture is induced, the concentration of IPTG used to induce, the temperature of cell growth, and the inclusion of 2% glycerol in the growth medium; optimal values are given under Materials and Methods. We estimate that g-GCS constitutes ;1% of the soluble protein in the crude homogenate of the cells; solubilization of the cell debris pellet shows little g-GCS by Western blotting, indicating that the enzyme does not

272

MISRA AND GRIFFITH

FIG. 1. Schematic representation of the construction of expression plasmids pHGCSH-3d (A) and pHGCSL-9d (B) encoding the heavy and the light subunits of human g-GCS, respectively. Amp and Kan represent the ampicillin resistance and the kanamycin resistance genes, respectively. The DNA fragments are not drawn to scale.

form inclusion bodies. Importantly, under the conditions described, the expression of the light subunit exceeds that of the heavy subunit by ;20%, minimizing the accumulation of catalytically active and difficult to separate free heavy subunit. Purification of g-GCS Human g-GCS was purified by ammonium sulfate fractionation followed sequentially by chromatography on Q-Sepharose, Superdex 200, ATP-affinity resin, and an E. coli g-GCS immunoaffinity column. Details of the individual steps are given under Materials and Methods, and the results are summarized in Table 1. The protocol used was derived from that used routinely for isolation of the rat kidney enzyme (18), but included modifications intended to achieve separation of human

and E. coli g-GCS. Such separation proved difficult. In initial studies, the high-speed supernatant obtained from the crude homogenate was fractionated using ammonium sulfate added in 5% increments from 35 through 65% saturation; precipitates containing g-GCS activity were subjected to SDS–PAGE and probed with antibodies to E. coli and rat g-GCS. Separation of the two enzymes was not obtained. Similarly, as shown in Fig. 2A, E. coli and human g-GCS are not usefully separated by ion-exchange chromatography. This result was unexpected in view of an earlier report indicating that E. coli and rat g-GCS are separated by chromatography on a DEAE–Sephacel (4). Because our preliminary studies using gel isoelectric focusing showed that the pI values of the E. coli and human g-GCS are only slightly different (;4.8 and

HUMAN g-GLUTAMYLCYSTEINE SYNTHETASE

TABLE 1 Purification of Human g-GCS from Transformed E. coli BL21 (DE3)a

Purification step Crude homogenate Supernatant 40–60% (NH4)2SO4 precipitate Q-Sepharose pool Superdex 200 pool ATP affinity pool Immunoaffinity pool

Total protein (mg)

Total activity (units)

5850 5610

27,600 27,000

4.7 4.8

(100) 98

3465 262 88 12 11.9

20,120 40,870 21,740 18,600 18,500

5.8 156 247 1550 1555

73 148b 79 67c 67c

Specific activity (units/mg)

Yield (%)

a

From 100 g (wet wt.) cells. The higher than expected activity of this fraction is actually due to the underestimation of activity in earlier fractions. Because the crude homogenate and supernatant fractions showed substantial ATPase activity, g-GCS activity was estimated as L-a-amino butyrate-dependent ADP formation. The correction applied was large, apparently compromising accuracy and resulting in an underestimation of activity in those fractions. The 60% ammonium sulfate precipitate showed less ATPase activity, but the high levels of ammonium sulfate present caused some inhibition of enzyme activity, again causing the true activity to be underestimated. c If the activity of the Q-Sepharose pool is taken as the best estimate of the activity initially present, the overall yield is 45%. b

;5, respectively), we anticipated that E. coli g-GCS would elute near mammalian g-GCS. In fact, we find that the E. coli monomer elutes slightly ahead of human g-GCS (see shoulder on leading edge of the activity peak). Unfortunately, aggregates of E. coli g-GCS trail through the peak of human g-GCS. Nevertheless, a significant amount of E. coli activity is removed by not including the leading edge of the g-GCS activity peak in the pooled fractions carried forward. As noted

273

in Table 1, Q-Sepharose chromatography also removes most of the nonspecific ATPase activity present in the initial fractions; the fractions pooled from the Q-Sepharose column are thus the first in which total g-GCS activity can be accurately estimated by the ADP formation assay used. In the standard protocol for purifying rat g-GCS the pool obtained from ion-exchange chromatography is dialyzed to remove salt prior to ATP affinity chromatography (18, 24). Because E. coli and human g-GCS differ significantly in Mr, we chose instead to remove salt by gel filtration on Superdex 200, a resin that might also separate the bacterial and mammalian enzymes. As shown in Fig. 2B, E. coli and human g-GCS are not cleanly separated. Western blotting of the Superdex 200 fractions displayed on SDS–PAGE without b-mercaptoethanol again demonstrated that it is aggregates of E. coli g-GCS that are present in fractions containing human g-GCS; monomeric E. coli g-GCS eluted just after human g-GCS (fractions 68 – 88). Although chromatography on Superdex 200 does not remove all of the E. coli g-GCS, we continue to include this step because it significantly reduces the amount of the E. coli enzyme and it fully removes other proteins (e.g., unidentified proteins of ;25 and ;45 kDa) that are not eliminated by subsequent purification steps. Where overall yield is of major concern, the gel filtration step can be omitted. Conditions for the ATP affinity chromatography step are similar to those used for the rat kidney enzyme. It is important to use C8-linked ATP affinity resin and to prepare the Mn21-containing buffer correctly and immediately before use (see Materials and Methods); in addition, the enzyme should be dialyzed into storage buffer as soon as possible after chromatography. Following the ATP affinity chromatography step, the

FIG. 2. Elution profile for anion-exchange chromatography of human g-GCS on Q-Sepharose (A) and for gel filtration chromatography on Superdex 200 (B). The shaded bar represents E. coli g-GCS as detected by Western blotting. Flow rates for Q-Sepharose and Superdex 200 were 3.5 and 1.0 ml/min, respectively. Fractions 151–165 (;0.2 M NaCl) of the Q-Sepharose chromatography and fractions 62–72 of the Superdex 200 chromatography were pooled.

274

MISRA AND GRIFFITH

Characterization of Human g-GCS

TABLE 2 Kinetic Constants for Rat and Human g-GCS Native rat kidney g-GCS Kinetic parameter Vmax(mmol/h/mg) KmL-glutamate (mM) KmL-cysteine (mM) KmL-aminobutyrate (mM) Km ATP (mM)

From Recombinant Ref. (18) This study a human g-GCS 1580b 1.6 0.3 1.0 0.2

1675 1.5 6 0.4 0.10 6 0.01 1.1 6 0.2 0.34 6 0.06

1650 6 100 1.8 6 0.2 0.10 6 0.02 1.3 6 0.2 0.40 6 0.04

a Kinetic constants for native rat kidney g-GCS, isolated as described (18) and having a specific activity of 1550 units/mg, were determined under the same assay conditions as used for recombinant human g-GCS. b Reported specific activity.

preparation of human g-GCS typically contains 0.5 to 1% E. coli g-GCS as judged by Western blotting or, more quantitatively, by cystamine inhibition (i.e., E. coli g-GCS is not inhibited by cystamine, whereas rat or pure human g-GCS is 100% inhibited). Where such levels of contamination are unimportant, no further purification is necessary. We attempted to fully eliminate contaminating E. coli g-GCS by altering the gradient used with QSepharose, by using other ion-exchange or gel filtration media (e.g. Mono-Q, DEAE–Sephacel, Superdex 75, AcA 34), by using chromatofocusing or hydrophobic chromatography (e.g., phenyl-agarose), by pooling activity fractions more narrowly, and by repeating steps that offered partial resolution. None of these approaches were successful in reducing the level of contamination below ;0.5%. We therefore constructed a column containing bound rabbit polyclonal antibodies to E. coli g-GCS and passed the 99% pure human g-GCS obtained following ATP affinity chromatography through that column (see Materials and Methods). The eluent contained no detectable E. coli g-GCS as judged by Western blotting or cystamine inhibition. In studies completed to date, immunoaffinity columns have been washed and reused twice without loss of effectiveness. Purified human g-GCS is stable for at least 7 days when stored at 4°C in 20 mM imidazole HCl buffer, pH 7.4, containing 1 mM EDTA. Addition of 25% glycerol allows the enzyme to be stored at 220°C without freezing; under those conditions the activity is stable for at least 12 months. As has been observed with the rat kidney enzyme, freezing solutions of human g-GCS results in substantial, irreversible inactivation.

After ATP affinity chromatography or immunoaffinity chromatography human g-GCS exhibits a specific activity of ;1500 units/mg; this value is comparable to the best values reported previously (18, 24) or obtained by us (Table 2) for rat kidney g-GCS. Sriram and AliOsman have previously reported the isolation of 0.28 mg of g-GCS with a specific activity of 1725 units/mg from a human malignant astrocytoma cell line (25); no other highly purified preparation of the human enzyme has been reported. Figure 3 shows Coomassie bluestained SDS–PAGE gels for each of the key fractions obtained during the purification of the enzyme. As shown, the final preparation contains only bands corresponding to the light and heavy subunits of human g-GCS. Identity of the bands was confirmed by Nterminal amino acid sequencing. The observed heavy subunit sequence was GLLSQGSPLSWEETK and the light subunit sequence was GTDSRAAKALLARAR. As expected for proteins expressed in E. coli, both subunits are missing the N-terminally coded Met residue, but the sequences are otherwise identical to those predicted by the published cDNA sequences (15, 16). Furthermore, quantitative amino acid analysis of the two SDS–PAGE bands indicates that the heavy and light subunits are present in essentially equal molar amounts (39.3 6 3.4 pmol heavy subunit and 38.7 6 3.2 pmol light subunit). This result establishes that holoenzyme rather than isolated heavy subunit has been isolated. Seelig et al. (3) report that the subunits of rat kidney g-GCS can be linked by a disulfide bond; in their preparations the bond was largely intact in the isolated enzyme as estimated by SDS–PAGE carried out in the absence of g-mercaptoethanol (3). Similar studies with our preparations of native rat kidney and recombinant human g-GCS show that the sub-

FIG. 3. Coomassie-stained SDS–PAGE showing human g-GCS at different stages of purification. Lane 1, molecular mass markers; lane 2, rat g-GCS. Lanes 3–9 display human g-GCS containing fractions as follows: lane 3, crude extract; lane 4, supernatant; lane 5, ammonium sulfate dialysate; lane 6, Q-Sepharose pool; lane 7, Superdex 200 pool; lane 8, ATP-affinity pool; and lane 9, immunoaffinity pool. Lanes 2, 8, and 9 contain 3 mg of total protein and lanes 3–7 contain 12 mg of total protein.

HUMAN g-GLUTAMYLCYSTEINE SYNTHETASE

units are 25–50% cross-linked by disulfide bonds. Although these values are lower than that reported by Seelig et al., the recombinant human g-GCS appears very similar to native rat enzyme prepared and analyzed in this laboratory. Kinetic constants for the purified recombinant human g-GCS were determined as described under Materials and Methods (Table 2). As shown, Vmax and substrate Km values are essentially identical to those obtained with the native rat kidney enzymes assayed under the same conditions. Literature values for the rat enzyme differ modestly, probably reflecting the fact that they were generally not determined under conditions in which the nonvaried substrates were truly saturating. The values shown in Table 2 differ significantly from those reported for the human malignant astrocytoma enzyme (25). In that study, Km values for L-glutamate and L-g-aminobutyrate were reportedly 0.03 and 0.14 mM, respectively. An explanation for the difference between those results and ours is not evident. Tested as described under Materials and Methods, recombinant human g-GCS was fully inhibited by BSO and by cystamine. Glutathione inhibits the enzyme competitively with respect to L-glutamate; the Ki value is 3.3 mM (data not shown). In summary, we have developed an expression system that allows growth of cells and isolation of 10 – 12 mg of homogeneous recombinant human g-GCS in ;7 days at an overall yield of ;45%. The isolated enzyme is stable, free of E. coli g-GCS, and composed of equal amounts of heavy and light subunits. Human g-GCS is shown to have substrate Km values essentially identical to those of native rat kidney g-GCS and to be inhibitable by L-SR-BSO, cystamine, and GSH. In studies to be reported elsewhere, we have shown that the expression plasmids described can also be used to obtain the heavy and light human g-GCS subunits separately. ACKNOWLEDGMENTS We thank Mr. Michael Hayward for purification of the E. coli and rat kidney g-GCS used in these studies and for excellent assistance with the studies reported. We thank Dr. Liane MendeMuller, Director of the MCW Protein and Nucleic Acid Shared Facility, for help with the amino acid analysis and peptide sequencing. We thank Dr. R. Timothy Mulcahy, University of Wisconsin, Madison, for generously providing the cDNA needed to construct the expression plasmids.

REFERENCES 1. Meister, A. (1991) Glutathione deficiency produced by inhibition of its synthesis, and its reversal: Applications in research and therapy. Pharmacol. Ther. 51, 155–194. 2. Griffith, O. W., and Mulcahy, R. T. (1998) The enzymes of glutathione biosynthesis: g-Glutamylcysteine synthetase. Adv. Enzymol. Rel. Areas Mol. Biol.,[cf1] in press. 3. Seelig, G. F., Simondsen, R. P., and Meister, A. (1984) Reversible

275

dissociation of g-glutamylcysteine synthetase into two subunits. J. Biol. Chem. 259, 9345–9347. 4. Huang, C. S., Chang, L. S., Anderson, M. E., and Meister, A. (1993) Catalytic and regulatory properties of the heavy subunit of rat kidney g-glutamylcysteine synthetase. J. Biol. Chem. 268, 19675–19680. 5. Huang, C. S., Anderson, M. E., and Meister, A.(1993) Amino acid sequence and function of the light subunit of rat kidney g-glutamylcysteine synthetase. J. Biol. Chem. 268, 20578 –20583. 6. Ahmad, S., Okine, L., Le, B., Najarian, P., and Vistica, D. T. (1987) Elevation of glutathione in phenylalanine mustard-resistant murine L1210 leukemia cells. J. Biol. Chem. 262, 15048 – 15053. 7. Griffith, O. W., and Friedman, H. S. (1991) Inhibition of metabolic drug inactivation: Modulation of drug activity and toxicity by perturbation of glutathione metabolism. in ‘‘Synergism and Antagonism in Chemotherapy’’ (Chou, T. C., and . Rideout, D. C., Eds.), pp. 245–284, Academic Press, New York. 8. Schro¨der, C. P., Godwin, A. K., O’Dwyer, P. J., Tew, K. D., Hamilton, T. C., and Ozols, R. F. (1996) Glutathione and drug resistance. Cancer Invest. 14, 158 –168. 9. Meister, A., and Griffith, O. W. (1979) Effects of methionine sulfoximine analogs on the synthesis of glutamine and glutathione: Possible chemotherapeutic implications. Cancer Treat.Rep. 63, 1115–1121. 10. Bailey, H. H., Mulcahy, R. T., Tutsch, K. D., Arzoomanian, R. Z., Alberti, D., Tombes, M. B., Wilding, G., Pomplun, M., and Spriggs, D. R. (1994) Phase I clinical trial of intravenous L-buthionine sulfoximine and melphalan: An attempt at modulation of glutathione. J. Clin. Oncol. 12, 194 –205. 11. O’Dwyer, P. J., Hamilton, T. C., LaCreta, F. P., Gallo, J. M., Kilpatrick, D., Halbherr, T., Brennan, J., Bookman, M. A., Hoffman, J., Young, R. C., Comis, R. L., and Ozols, R. F. (1996) Phase I trial of buthionine sulfoximine in combination with melphalan in patients with cancer. J. Clin. Oncol. 14, 249 –256. 12. Bailey, H. H., Ripple, G., Tutsch, K. D., Arzoomanian, R. Z., Alberti, D., Feierabend, C., Mahvi, D., Schink, J., Pomplun, M., Mulcahy, R. T., and Wilding, G. (1997) Phase I study of continuous infusion L-S,R-buthionine sulfoximine with intraveneous melphalan J. Natl. Cancer Inst. 89, 1789 –1796. 13. Gallo, J. M., Brennan, J., Hamilton, T. C., Halbherr, T., Laub, P. B., Ozols, R. F., and O’Dwyer, P. J. (1995) Time-dependent pharmacodynamic models in cancer chemotherapy: Population pharmacodynammic model for glutathione depletion following modulation by buthionine sulfoximine (BSO) in a phase I trial of melphalan and BSO. Cancer Res. 55, 4507– 4511. 14. Malaker, K., Hurwitz, S. J., Bump, E. A., Griffith, O. W., Lai, L. L., Riese, N., and Coleman, C. N. (1994) Pharmacodynamics of prolonged treatment with L-S-buthionine sulfoximine. Int. J. Radiat. Oncol.Biol. Phys. 29, 407– 412. 15. Gipp, J. J., Chang, C., and Mulcahy, R. T. (1992) Cloning and nucleotide sequence of a full-length cDNA for human liver g-glutamylcysteine synthetase. Biochem. Biophys. Res. Commun. 185, 29 –35. 16. Gipp, J. J., Bailey, H. H., and Mulcahy, R. T. (1995) Cloning and sequencing of the cDNA for the light subunit of human liver g-glutamylcysteine synthetase and relative mRNA levels for the heavy and light subunits in normal human tissues. Biochem. Biophys. Res. Commun. 206, 584 –589. 17. Griffith, O. W. (1982) Mechanism of action, metabolism, and toxicity of buthionine sulfoximine and its higher homologs, potent inhibitors of glutathione synthesis. J. Biol. Chem. 257, 13704 –13712. 18. Seelig, G. F., and Meister, A. (1985) Glutathione biosynthesis:

276

19.

20.

21.

22.

MISRA AND GRIFFITH

g-Glutamylcysteine synthetase from rat kidney. Methods Enzymol. 113, 379 –391. Studier, F. W., Rosenberg, A. H., Dunn, J. J., and Dubendoroff, J. W. (1990) Use of T7 RNA polymerase to direct expression of cloned genes. Methods Enzymol. 185, 60 – 89. Sanger, F., Nicklen, S., and Coulson, A. R. (1977) DNA sequencing with chain-terminating inhibitors. Proc. Natl. Acad. Sci. USA 74, 5463–5467. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) ‘‘Molecular Cloning: A Laboratory Manual,’’ 2nd ed., Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Huang, C. S., Moore, W. R., and Meister, A. (1988) On the active site thiol of g-glutamylcysteine synthetase: Relationship to ca-

talysis, inhibition, and regulation. Proc. Natl. Acad. Sci. USA 85, 2464 –2468. 23. Bradford, M. M. (1976) A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal. Biochem. 72, 248 – 254. 24. Sekura, R., and Meister, A. (1977) g-Glutamylcysteine synthetase: Further purification, ‘‘half of the sites’’ reactivity, subunits, and specificity. J. Biol. Chem. 252, 2599 –2605. 25. Sriram, R., and Ali-Osman, F. (1993) Purification and biochemical characterization of g-glutamylcysteine synthetase from a human malignant astrocytoma cell line. Biochem. Mol. Biol. Int. 30, 1053–1060.