Characterization of the major hydroperoxide-reducing activity of human plasma. Purification and properties of a selenium-dependent glutathione peroxidase.

We have recently characterized the major hydroperoxide-reducing enzyme of human plasma as a glutathione peroxidase (Maddipati, K. R., Gasparski, C., and Marnett, L. J. (1987) Arch. Biochem. Biophys. 254, 9-17). We now report the purification and kinetic characterization of this enzyme. The purification steps involved ammonium sulfate precipitation, hydrophobic interaction chromatography on phenyl-Sepharose, anion exchange chromatography, and gel filtration. The purified peroxidase has a specific activity of 26-29 mumol/min/mg with hydrogen peroxide as substrate. The human plasma glutathione peroxidase is a tetramer of identical subunits of 21.5 kDa molecular mass as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and is different from human erythrocyte glutathione peroxidase. The plasma peroxidase is a selenoprotein containing one selenium per subunit. Unlike several other glutathione peroxidases this enzyme exhibits saturation kinetics with respect to glutathione (Km for glutathione = 4.3 mM). The peroxidase exhibits high affinity for hydroperoxides with Km values ranging from 2.3 microM for 13-hydroperoxy-9,11-octadecadienoic acid to 13.3 microM for hydrogen peroxide at saturating glutathione concentration. These kinetic parameters are suggestive of the potential of human plasma glutathione peroxidase as an important regulator of plasma hydroperoxide levels.

the development of several vascular diseases (2). Prostacyclin and thromboxane synthases that are responsible for the synthesis of these important physiological substances are differentially inactivated by fatty acid hydroperoxides. Whereas thromboxane synthase is relatively insensitive to fatty acid hydroperoxides, prostacyclin synthase is inactivated at hydroperoxide concentrations as low as 0.5 I.LM (3,4). The emergence of the concept of "peroxide tone" of tissues (5) as a result of this selectivity in inactivation gained support in recent years from the observation that atherosclerotic lesions contain elevated levels of lipid hydroperoxides (6, 7).
The major metabolic fate of fatty acid hydroperoxides, apart from their conversion to physiological end products, is reduction by peroxidases to the corresponding alcohols. We recently reported the characterization of the major hydroperoxidereducing activity of human plasma, showed that it is dependent on reduced glutathione, and demonstrated that it is a glutathione peroxidase (8). This glutathione peroxidase activity appears to account for all the hydroperoxide-reducing activity of human plasma. Considering the potential deleterious effects of fatty acid hydroperoxides, this peroxidase activity may play an important protective role against hydroperoxide pathology. We have undertaken the purification of this glutathione peroxidase from human plasma to study its physical and kinetic properties and as first step in elucidation of its involvement in the control of circulating fatty acid hydroperoxide levels. Our results are described herein.

MATERIALS AND METHODS
Fresh frozen human plasma was obtained from a local American Red Cross center. Glutathione, NADPH, glutathione reductase from bakers' yeast, phenyl-Sepharose CL-4B, DEAE-Sephadex A-50, protein standards for SDS-PAGE' and gel filtration chromatography, and human erythrocyte glutathione peroxidase were purchased from Sigma. Sephadex G-200 was obtained from Pharmacia LKB Biotechnology Inc., and 2,3-diaminonaphthalene hydrochloride was from Aldrich. Selenious acid was purchased from Alfa Products. 13-00H-182 and 15-00H-204 were prepared from linoleic and arachidonic acids, respectively, using soybean lipoxygenase according to the method of Funk et al. (9). PPHP was prepared according to the method of Weller et al. (10). Protein estimations were carried out by the Bio-Rad protein assay according to manufacturer's suggestions using bovine serum albumin as standard.
Assay of Glutathione Peroxidase-Glutathione peroxidase activity was assayed by the coupled assay described by Paglia and Valentine (11) using hydrogen peroxide as substrate. In brief, the enzyme was incubated in 0.1 M Tris-HC1 containing 5 mM EDTA, 1 mM glutathione, 0.2 mM NADPH, 1 mM sodium azide, and 1 IU of glutathione The abbreviations used are: SDS-PAGE, sodium dodecyl sulfatepolyacrylamide gel electrophoresis; 13-OOH-182, 13-hydroperoxy-9,ll-octadecadienoic acid; 15-OOH-204, 15-hydroperoxy-5,8,11,13eicosatetraenoic acid; PPHP, 5-phenyl-4-pentenyl hydroperoxide; HPLC, high performance liquid chromatography. reductase in a total volume of 3.0 ml at 37 "C for 10 min before the initiation of reaction with hydrogen peroxide (100 pM final). Organic hydroperoxides were added in methanol, and the total methanol concentration was kept constant at 1%. For kinetic studies the reaction mixture also contained 200 p~ Tween 20. The rate of decrease of absorbance at 340 nm was followed on a Cary 210 spectrophotometer equipped with a thermostated cuvette holder. Kinetic measurements using various hydroperoxides were carried out on an Aminco DW-2a spectrophotometer in the dual wavelength mode. The reference beam was 410 nm and the sample beam was 340 nm. One unit of enzyme activity was defined as the amount of enzyme required for the oxidation of 2 pmol of glutathione per min at 37 "C using HzOZ as substrate at 1 mM glutathione. All graphs were plotted using Cricket Graph plotting software (Cricket Software Inc., Philadelphia) on a Macintosh Plus microcomputer (Apple Computer Inc., Cupertino, CA). Data were subjected to regression analysis using the curve fitting routines available in the program. The correlation coefficients were always greater than 0.99.
Purification of Plasma Glutathione Peroxidase-The entire purification was carried out at 0-4 'C. Frozen human plasma (approximately 1,400 ml) was thawed overnight and centrifuged at 13,000 X g to remove precipitated lipids. To the clear supernatant, solid ammonium sulfate was slowly added over a period of 2 h to 20% saturation (158 g) (12). The plasma was centrifuged at 13,000 X g, the precipitate was discarded, and the supernatant's ammonium sulfate concentration was raised to 30% saturation (83 g). The plasma was again centrifuged at 13,000 X g and the supernatant was discarded. The precipitate was dissolved in 0.1 M Tris-HC1, pH 7.2 (75 ml), containing 0.2 M ammonium sulfate, 5 mM EDTA, and 1 mM glutathione (buffer A). This translucent protein solution was added to phenyl-Sepharose CL-4B (100 ml settled gel) that was preequilibrated with buffer A. The gel/protein mixture was allowed to stand for 2 h with occasional stirring and then loaded onto a column. The column was washed extensively with the equilibrating buffer until the 280-nm absorbance of the eluate was equal to that of the buffer. The column was subjected to a linear gradient elution between 0.1-0.01 M Tris-HC1, pH 7.2, both containing 5 mM EDTA and 1 mM glutathione (250 ml each). The perioxidase started eluting very early in the gradient but tailed off slowly. Active fractions were pooled and concentrated using a YM-10 membrane. The buffer was changed to 0.2 M Tris-HC1, pH 8.0, containing 5 mM EDTA and 1 mM glutathione (buffer B), while the enzyme was in the concentrator and concentrated to a final volume of 50 ml. This enzyme was then applied to a DEAE-Sephadex A-50 column (100 ml), preequilibrated with buffer B, and washed with about 200 ml of the equilibrating buffer. The enzyme was eluted from the column with a linear gradient between 0.2-0.4 M Tris-HC1, pH 8.0, both containing 5 mM EDTA and 1 mM glutathione (250 ml each). Peroxidase active fractions were pooled and concentrated to about 5 ml using a YM-30 membrane. The concentrated peroxidase was subjected to gel filtration on a Sephadex G-200 column preequilibrated with 0.1 M Tris-HC1, pH 7.2, containing 5 mM EDTA and 1 mM glutathione (buffer C). The most active fractions of the eluent were pooled separately from less active fractested for purity on a TSK G4000SW HPLC gel filtration column at tions. The concentrated enzyme from the highly active pool was a flow rate of 1 ml/min using buffer C. The effluent was followed by the absorbance at 280 nm using a variable wavelength UV detector (Varian Instruments, Model 2050).
Purification of Human Erythrocyte Glutathione Peroxidase-Commercially available human erythrocyte glutathione peroxidase had considerable impurities. To compare human plasma glutathione peroxidase with erythrocyte enzyme, the commercial erythrocyte peroxidase was purified by HPLC gel filtration on a TSK G4000SW column using the conditions described above. Each peak eluted from the column was tested for peroxidase activity by the coupled assay described above.
Molecular Mass Determinatwn-The molecular mass of the purified human plasma peroxidase was determined by gel filtration. This was done using a conventional Sephacryl S-200 column and by HPLC gel filtration on a TSK G4000SW column. In both methods buffer C was used as the mobile phase. Gel filtration on Sephacryl S-200 was performed on a 1.6 X 90-cm column at a flow rate of 10 ml/h. Protein samples were applied to the column in 1 ml of buffer C containing 5% glycerol. The eluent was collected in 1.5-ml fractions and their 280 nm absorbances and peroxidase activities determined. The elution volume for each protein was determined separately. HPLC gel filtration on TSK G4000SW (7.5 X 300 mm) was carried out by connecting two columns in series on a Varian model 5000 liquid chromatograph. All protein samples were dissolved in buffer C and injected separately. The injection volume was 50 pl. The column was eluted at 1 ml/min flow rate and the eluent was monitored at 280 nm. In addition to UV absorbance, the peroxidase peaks were confirmed by activity measurement.
Subunit MolecLclar Mass Determination-SDS-PAGE was performed according to the procedure of Laemmli using a 12% acrylamide gel (13). The mixture of molecular mass standards in SDS-7 (Sigma) was used for molecular weight determination. All the samples were heat-denatured (5 min at 80 "C) in the presence of &mercaptoethanol (2.5%) and SDS (1%). The gel was stained by Coomassie Blue.
Selenium Estimation-The selenium content of human plasma peroxidase was determined by the method of Watkinson (14), incorporating the recent modifications by Bayfield and Romalis (15). Selenious acid and 2,3-diaminonaphthalene hydrochloride were used instead of the sodium salt and the free base, respectively.

RESULTS
Purification of Peroxidase from Hurnun Plasm- Table I presents a summary of the purification of glutathione peroxidase activity from human plasma. Precipitation of peroxidase was found to be optimal at 20-30% saturation of ammonium sulfate, although there was a loss of about 50% of the activity. Recovery of peroxidase was greater at higher ammonium sulfate concentrations, but precipitation of extraneous protein negated the advantage of higher recovery. Hydrophobic interaction chromatography on phenyl-Sepharose proved to be a  Extensive washing of the gel with buffer A before application of the gradient was essential for the level of purification achieved. The peroxidase activity did not elute as a sharp peak ( Fig. 1) because the gel was mixed with protein and then loaded on the column.The peroxidase eluted as a single peak on DEAE-Sephadex (Fig. 2) suggesting the absence of multiple forms of the enzyme contrary to human erythrocyte glutathione peroxidase (16). The final purification step of gel filtration on Sephadex G-200 eliminated the major high molecular weight impurity (Fig. 3), as well as some low molecular weight proteins, and yielded an electrophoretically homogeneous protein (Fig. 4A). The enzyme also yielded a single peak on a TSK G4000SW HPLC gel filtration column (data not shown). Partial Purification of Human Erythrocyte Glutathione Peroxidase-Commercially available erythrocyte glutathione peroxidase contained approximately five proteins as judged by both electrophoresis and HPLC gel filtration (data not shown). Only one of the peaks from HPLC had peroxidase activity. Unfortunately, one of the low molecular impurities eluted very closely to the peroxidase on HPLC and was difficult to eliminate completely. However, this impurity of approximately 59-kDa size was much larger than the peroxidase subunit (approximately 21 kDa) ( The RF values for the peroxidases in the above gel are 0.828 and 0.848, respectively, for plasma and erythrocyte peroxidases. Six such graphs were generated from six different gels. The average molecular masses with standard deviations are as follows: plasma peroxidase: 21.5 * 0.2 kDa. Erythrocyte peroxidase: 20.6 f 0.2 kDa. The molecular mass difference between these two peroxidases from each experiment was calculated and the average difference was 800 f 31 Da. Molecular Mass and Subunit Composition-The molecular mass of the purified plasma glutathione peroxidase was determined by gel filtration on a Sephacryl s-200 column and by HPLC on a TSK G4000SW column. Using standards ranging from 150 to 29 kDa, the molecular mass of the plasma peroxidase was estimated to be about 73 kDa by conventional gel filtration, but 94 kDa by HPLC gel filtration (Fig. 5). The subunit molecular mass was estimated to be 21.5 kDa by SDS-PAGE on a 12% acrylamide gel. The relative mobilities of the standards on 12% acrylamide gel were plotted, and the molecular masses of the subunits of human erythrocyte and plasma peroxidases were calculated from the graph (Fig. 4B). The subunit of human erythrocyte glutathione peroxidase has about 800 Da lower molecular mass (Fig. 5A). This difference in the molecular mass, calculated based on their difference in electrophoretic mobility, was highly consistent (800 f 31 Da, n = 6). Based on eletrophoretic and chromatographic behavior under denaturing and nondenaturing conditions, respectively, human plasma glutathione peroxidase appears to be a tetramer of identical subunits of about 21.5 kDa molecular mass.
Selenium Determination of Plasma Peroxidase-The selenium content of the peroxidase was estimated fluorimetrically, at three different protein concentrations simultaneously with a set of standards ranging from 0 to 150 ng of selenium.
The selenium content was 3.7 f 0.2 ng/pg protein (1.03 nmol of selenium/nmol of subunit) and was linear with protein concentration ( r = 1.00, Fig. 6). The enzyme is competitively inhibited (with respect to glutathione) by 8-mercaptosuccinic acid (Iso = 33 p~) , which is a specific inhibitor for seleniumdependent glutathione peroxidases (17).
Kinetics of Plasma Glutathione Peroxidase-The purified peroxidase exhibited a specific activity of 26-29 units/mg ( n = 8). Saturation of the peroxidase with glutathione was attempted using 13-OOH-182 (4 p~) as the hydroperoxide substrate ( Fig. 7). At concentrations of glutathione above 5 mM the peroxidase appeared to be inhibited slightly. Data points below 5 mM glutathione were used to construct a double-reciprocal plot (inset of Fig. 7). The K , value for glutathione calculated from the double-reciprocal plot is 4.3 mM. Substrate saturation curves were obtained for hydroperoxides, uiz. 13-OOH-182, 15-OOH-20:4, PPHP, and HzOz at constant glutathione concentrations of 1 and 5 mM using the coupled assay of Paglia and Valentine (11) as described under "Materials and Methods." Methanol at 1% (vehicle for organic hydroperoxides) did not affect the rate of peroxidase activity when HzOz was used as a substrate (data not shown).
Also, 200 p~ Tween 20, used to maintain homogeneous assay mixtures in the presence of organic hydroperoxides, did not affect the rate of glutathione oxidation with HzOZ as substrate. Lineweaver-Burk plots of the data from the linear portion of the substrate saturation curves are given in Fig. 8. Kinetic constants derived from the double-reciprocal plots are given in Table 11. At 1 mM concentration of glutathione, commercially available human erythrocyte glutathione peroxidase gave K , values of 12 and 10 p~ for hydrogen peroxide and PPHP, respectively. "The kinetic constants kcat and kat/Km are calculated from the The bimolecular association rate constant haJKm is given as per data obtained with 5 mM glutathione. molar subunit per s.

DISCUSSION
In a recent report, we published the characterization of a glutathione-dependent peroxidase activity in human plasma that appeared to be different from the well-characterized and purified human erythrocyte glutathione peroxidase (8). To complete the characterization of the human plasma glutathione peroxidase and to study its physical and kinetic properties, we undertook the purification of the peroxidase. Purification of the enzyme from human plasma was achieved in four steps with an overall yield of about 19%. Precipitation of the peroxidase at relatively low salt concentration in the first step prompted us to employ hydrophobic interaction chromatography. This step eliminated most of the impurities and dramatically increased the specific activity in the initial stage of the purification. Another important aspect of the purification was careful fraction collection in the gel filtration step. The purified peroxidase was electrophoretically homogeneous. Using the recovery and purification data in Table I, one can calculate that this glutathione peroxidase accounts for about 0.007% of the total human plasma protein.
Glutathione peroxidase from human tissues such as erythrocytes and placenta have been purified and characterized (16, 18). When we recently characterized the glutathione peroxidase activity as the only peroxidase of human plasma (8), a major concern was the possible contamination of plasma with glutathione peroxidase from erythrocytes. Although several biochemical and immunochemical differences were observed between these two peroxidase activities, the argument that the two activities are due to different proteins was equivocal. Like glutathione peroxidases from all other sources, the human plasma glutathione peroxidase appears to be a tetramer of identical subunits of about 21.5 kDa subunits. It is difficult to assign an exact molecular mass to the native enzyme because of different results obtained with conventional and HPLC gel filtration methods. The molecular mass calculated for a tetrameric structure should be 86 kDa, which lies in between the molecular masses obtained by conventional gel filtration on Sephacryl S-200 column (73 kDa) and HPLC gel filtration on TSK G4000SW column (94 kDa). From SDS-PAGE it is clear that the subunits of human plasma and erythrocyte peroxidases are electrophoretically different by about 800 Da (k30 Da, n = 6) (Fig. 4). The molecular mass of the human erythrocyte peroxidase subunit is reported to be 23 kDa based on SDS-urea-polyacrylamide disc gel electrophoresis (16); however, the molecular mass calculated from our electrophoretic experiments is 20.6 kDa. The difference in the electrophoretic mobility of the subunits of plasma and erythrocyte peroxidases supports the suggestion that these two peroxidases are different.
Human erythrocyte and plasma glutathione peroxidases are also kinetically different. The K, values for Hz02 and PPHP are 3.3 and 2.6 pM, respectively, for plasma peroxidase, and 12 and 10 p M , respectively, for erythrocyte peroxidase at 1 mM glutathione. In our earlier experiments with whole plasma, we obtained a much higher K,,, value for PPHP (54 p M ) at the same concentration of glutathione (8). This is probably due to the presence of large amounts of extraneous proteins in plasma (e.g. albumin) that bind to organic hydroperoxides. However, the K, value for the same substrate with erythrocyte peroxidase was also higher in our earlier experiments (24 p M ) than in the present experiments (10 p~) . This difference is probably due to a limitation of the direct HPLC assay, employed in our earlier studies, where the initial rate was estimated based on the product formed at the end of 1 rnin of the reaction. It is difficult to follow the initial rates continuously in the HPLC assay, unlike the coupled assay presently employed. That the two peroxidases are different is also substantiated by immunochemical experiments of Takahashi et aZ. (19) who showed that rabbit anti-human erythrocyte glutathione peroxidase does not cross-react with purified human plasma glutathione peroxidase. It appears there is no significant lysis of erythrocytes during plasma preparation because only a single peak of peroxidase activity is obtained at every step of the purification protocol. Furthermore, none of the erythrocyte peroxidase is detected when fractions are analyzed by SDS-PAGE. Also, this suggests that there are probably no other isozymes present in plasma in contrast to erythrocyte peroxidase (16). The plasma glutathione peroxidase is a selenoprotein containing one selenium per subunit. This suggests that the peroxidase is a selenium-dependent glutathione peroxidase. P-mercaptocarboxylic acids are well known specific inhibitors of seleniumdependent glutathione peroxidases (17). The involvement of selenium in the peroxidase activity, probably through a selenocysteine, is substantiated by its inhibition with P-mercaptosuccinic acid (17). Glutathione peroxidases from bovine and ovine erythrocytes and rat and hamster livers show no saturation with respect to glutathione, so true K,,, values cannot be obtained for glutathione (20,21). For human erythrocyte glutathione peroxidase, a K, value of 4.3 mM was reported for glutathione, but no data were presented showing the saturation of the peroxidase with glutathione. The plasma glutathione peroxidase definitely shows saturation with respect to glutathione.
The K,,, value obtained from the double-reciprocal plot is 4.3 mM, which is identical to the reported value for human erythrocyte glutathione peroxidase. The activity of plasma glutathione peroxidase appears to decrease with increasing concentrations of glutathione beyond the saturation point of 5 mM. Because of this, all the kinetic experiments with various hydroperoxides were carried out at 5 mM glutathione. However, to compare the kinetic constants of plasma peroxidase with those of glutathione peroxidases from other sources reported in literature, we also determined the "K,,," values at 1 mM glutathione. The apparent K , values for human plasma glutathione peroxidase are approximately 10-fold lower than the corresponding values for bovine, ovine, and human erythrocyte peroxidases with hydrogen peroxide as substrate under similar conditions (20). With organic hydroperoxides, the apparent K,,, values for plasma peroxidase are as much as 30fold lower than other peroxidases. The plasma glutathione peroxidase appears to be a very efficient enzyme for the reduction of organic hydroperoxides. In fact, the values of k,,,/K, are very close to the diffusion-controlled limit for a bimolecular reaction. At 5 mM concentrations of glutathione, the K , values for all the hydroperoxides tested are in the low micromolar range (Table 11). This makes the human plasma glutathione peroxidase a very effective hydroperoxide scav-

Human Plasma
Glutathione Peroxidase 17403 enger even at relatively low concentrations of glutathione. Recent estimates of organic hydroperoxide concentrations of human plasma are about 0.5 p~ and H202 is estimated to be approximately 5 p~ (22,23). The low K, values for organic hydroperoxides, especially that for 13-hydroperoxyoctadecadienoic acid (2.3 p~) , may indicate a possible physiological role for this enzyme in the control of circulating fatty acid hydroperoxides. It is also interesting to note that the cholesterol ester of 13-hydroperoxyoctadecadienoic acid was recently identified as a major hydroperoxide of atherosclerotic lesions (7). Furthermore, esters of hydroperoxy fatty acids to cholesterol appear to be the major circulating forms of organic hydroperoxides (23). Esterified fatty acid hydroperoxides are very poor substrates for glutathione peroxidases (24). However, Ursini et al. (25) have purified a selenium-dependent glutathione peroxidase from pig heart that accepts both esterified and non-esterified fatty acid hydroperoxides as substrates. Studies are in progress to determine the versatility of plasma glutathione peroxidase in the reduction of various hydroperoxides of physiological relevance.
It is known that the steady-state levels of glutathione in human plasma are in the range of 0.3 p~ (26). Although it is difficult to assess the physiological significance of the plasma peroxidase activity at this low concentration of glutathione, inter-organ transport of glutathione may play an important role in the modulation of the peroxidase activity. y-Glutamyl transpeptidase has been shown to actively degrade extracellular glutathione, whereas liver constantly releases this ubiquitous tripeptide into plasma (27). It has been demonstrated that liver glutathione is transported in substantial amounts to hepatic vein plasma in rats, where the level of glutathione is found to be as high as 26 p~ (28). Arterial glutathione levels can be increased severalfold by selectively inhibiting yglutamyl transpeptidase. This suggests that y-glutamyl transpeptidase is an important regulator of plasma glutathione levels, thereby indirectly regulating the plasma peroxidase activity. Thus, a detailed knowledge of glutathione transport and its regulation appears to be essential to ascertain the physiological function of the plasma glutathione peroxidase.