THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc

Vol. 263, No. 14, Issue of May 15, pp. 6884-6892, 1988 Printed in U. S. A.

Superoxide Dismutase and Catalase Conjugated to Polyethylene Glycol Increases Endothelial Enzyme Activity and Oxidant Resistance*

(Received for publication, September 3, 1987)

Joseph S. Beckman$, Robert L. Minor, Jr.5, Carl W. Whiten, John E. Repinell, Gerald M. Rosenil, and Bruce A. Freeman$ From the $Departments of Anesthesiology and Biochemistry, University of Alabama at Birmingham, University Station, Birmingham, Alabama 35233, the §School of Medicine, University of Iowa, Iowa City, Iowa 52242, the ll Webb- Waring Lung Institute, University of Colorado Health Sources Center, Denver, Colorado 80302, and the 11 Department of Pharmacology, Duke University Medical Center, Durham, North Carolina 27710

Covalent conjugation of superoxide dismutase and catalase with polyethylene glycol (PEG) increases the circulatory half-lives of these enzymes from <lo min to 40 h, reduces immunogenicity, and decreases sensi- tivity to proteolysis. Because PEG has surface active properties and can induce cell fusion, we hypothesized that PEG conjugation could enhance cell binding and association of normally membrane-impermeable en- zymes. Incubation of cultured porcine aortic endothe- lial cells with lZ6I-PEG-catalase or ‘261-PEG-superox- ide dismutase produced a linear, concentration-de- pendent increase in cellular enzyme activity and radioactivity. Fluorescently labeled PEG-superoxide dismutase incubated with endothelial cells showed a vesicular localization. Mechanical injury to cell mono- layers, which is known to stimulate endocytosis, fur- ther increased the uptake of fluorescent PEG-super- oxide dismutase. Endothelial cell cultures incubated with PEG-superoxide dismutase and PEG-catalase for 24 h and then extensively washed were protected from the damaging effects of reactive oxygen species de- rived from exogenous xanthine oxidase as judged by two criteria: decreased release of intracellular W r - labeled proteins and free radical-induced changes in membrane fluidity, measured by electron paramag- netic resonance spectroscopy of endothelial membrane proteins covalently labeled with 4-maleimido-2,2,6,6- tetramethylpiperidinooxyl. Addition of PEG and PEG- conjugated enzymes perturbed the spin-label binding environment, indicative of producing an increase in plasma membrane fluidity. Thus, PEG conjugation to superoxide dismutase and catalase enhances cell asso- ciation of these enzymes in a manner which increases cellular enzyme activities and provides prolonged pro- tection from partially reduced oxygen species.

Controlled manipulation of cellular superoxide dismutase- and catalase-specific activities can help define mechanisms of tissue injury mediated by superoxide and hydrogen peroxide (1). The toxicity of reactive oxygen species, whose rate of production can be amplified by pathological events including neutrophil activation, hyperoxia, metabolism of redox-active

* This work was supported by National Institutes of Health Grants NS-23700 and NS-24275 (to B. A. F.), HL-28182 (to J. E. R.), and HL-33550 (to G. M. R.) and by grants from the Health Effects Institute (to B. A. F.) and the Alabama Affiliate of the American Heart Association (to J. S. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

drugs, radiation exposure, and ischemia (2, 3), has also stim- ulated the use of antioxidant enzymes as therapeutic agents. However, the experimental and therapeutic potentials of su- peroxide dismutase and catalase are limited by two factors. First, both proteins are rapidly cleared by kidneys (4, 5 ) , leading to circulatory half-lives of only 6-10 min following intravenous injection. Second, reactive oxygen species diffuse very short distances before reacting with cellular components (6), and neither superoxide dismutase nor catalase can pene- trate across cell membranes. Thus, antioxidant enzymes can not gain access to intracellular sites of free radical generation to achieve effective pharmacological value.

The renal clearance of superoxide dismutase and catalase can be prevented by increasing their molecular weight through covalent attachment of the inert linear polymer, monome- thoxy-polyethylene glycol (PEG;’ Refs. 7-9). The structure of PEG is H-0-(CH,-CH,-0)”-CH, with n = 150 for PEG weigh- ing 5000 Da. The free hydroxyl of PEG is conjugated to t-

amino groups of lysine with a bifunctional reagent such as cyanuric chloride. Typically, from 10 to 15 of the 20 available amino groups on superoxide dismutase are modified with PEG monomers of an average molecular mass of 5000 Da each, increasing the molecular mass of superoxide dismutase from 32 kDa to about 100 kDa for the modified protein. Similarly, the molecular weight of catalase may be increased by 300% following PEG conjugation. Modification by PEG blocks renal clearance and increases the circulating enzyme half-life from 6 min to 30-40 h in the rat (9). The inert nature of PEG also reduces the antigenicity of the native protein and inhibits the hydrolysis of protease-sensitive proteins such as catalase (8). Additionally, the long term stability of enzymes in aqueous solution is frequently increased by PEG conjugation.

Polyethylene glycol is a surface-active molecule that is used extensively to induce fusion for cell hybridization (10). Mem- branes can absorb substantial quantities of PEG, binding 1 PEG molecule/l2 dipalmitoyl-phosphatidylcholine molecules in liposomes with a dissociation constant of 6 pM (11). Be- cause PEG constitutes up to two-thirds of the total molecular weight of PEG-conjugated enzymes, we have investigated the possible membrane association of PEG-conjugated enzymes with cultured endothelial cells. Endothelial cells were used as target cells in this study, because the vascular endothelium is a significant site of oxidant injury (12) and because the anatomical localization of the vascular endothelium provides direct contact with circulating antioxidant enzymes. We re-

’ The abbreviations used are: PEG, polyethylene glycol; FITC, fluorescein isothiocyanate; HBSS, Hank’s balanced salt solution; Mal-6-TEMPO, 3-maleimido-2,2,6,6,-tetramethylpiperidinooxy~ W/ S ratio, the weakly to strongly immobilized peak ratio.

6884

Augmenting Endothelial Antioxidant Enzymes 6885

port that PEG conjugation of superoxide dismutase and cat- alase enhances cell association of these antioxidant enzymes, alters membrane fluidity, and confers resistance to oxidant stress.

EXPERIMENTAL PROCEDURES

Materials-Pharmaceutical grade, bovine Cu,Zn superoxide dis- mutase was generously provided by Diagnostic Data Inc. (Mountain View, CA) and by Grunenthal GmbH (Aachen, Germany). The su- peroxide dismutase obtained from Diagnostic Data Inc. was dialyzed overnight against phosphate-buffered saline to remove sucrose added as a preservative. Bovine catalase was obtained from Cooper- Biomedical, Inc. Bovine milk xanthine oxidase was purchased from Behring Diagnostics. Tissue culture media were obtained from Gibco, and fetal calf serum was obtained from HyClone (Logan, UT). All other reagents were obtained from Sigma.

Enzyme Assays-Superoxide dismutase was assayed by inhibition of cytochrome c reduction, and units of activity are defined as given by McCord and Fridovich (13). Catalase activity was determined by the disappearance of 10 mM hydrogen peroxide in 50 mM K' phos- phate, pH 7.0, monitored at 240 nm ( A c Z 4 ~ = 43.6 M" cm"; Ref. 14). Xanthine oxidase was assayed by monitoring the production of uric acid at 295 nm (AcZs5 = 1.1 X lo' M" cm"; Ref. 15) at 25 "C. The reaction contained 50 p~ xanthine, 100 p~ EDTA, and 50 mM K' phosphate, pH 7.8. Units of activity for catalase and xanthine oxidase are defined as 1 pmol product produced per min.

Preparation of PEG-Superoxide Dismutase and PEG-Catalase- Cyanuric chloride (10 g) was dissolved in 100 ml of chloroform, filtered through Whatman No. 1 paper and evaporated to dryness immediately before use (16). Monomethoxy-polyethylene glycol (5000 Da) was activated with cyanuric chloride and coupled to superoxide dismutase and catalase by standard methods (7-9). The extent of conjugation, measured by the number of reactive amino groups, was determined by titration with trinitrobenzene sulfonic acid (17), and protein concentrations were measured by the Biuret method (Sigma technical procedure 540). Typically, about 60-70% of the available amino groups on superoxide dismutase and 60-80% of the amino groups on catalase were conjugated to PEG, and there was no loss of enzyme activity. Superoxide dismutase assay showed 3300-3500 units/mg protein both before and after PEG conjugation. Previously, a 50% loss of superoxide dismutase activity was reported when 95% of the available amino groups were modified with PEG by the cyanuric acid method (9). Using the same protocol, we found slightly less modification (60-70%) without loss of superoxide dismutase activity while retaining their long circulatory half-lives in rodents.' The commercial preparation of PEG-superoxide dismutase, made with a succinylated PEG, is also only 50% modified and retains full activity.3

Fluorescent and Rudwuctive Labeling of Proteins-Proteins were labeled with lZ5I via a solid phase oxidative reaction using 1,3,4,6- tetrachloro-3a,6a-diphenylglycoluril (Pierce Chemical Co.; Ref. 18) and then exhaustively dialyzed against phosphate-buffered saline to remove free lz5I. Fluorescence labeling of Cu,Zn superoxide dismutase was accomplished by addition of 0.3 mg of fluorescein isothiocyanate (FITC) to 1.0 ml of 10 mg/ml protein in 50 mM sodium carbonate, pH 9.5, and stirred for 1 h at 4 "C while maintaining pH at 9.5 by addition of Na2C0, (19). Unbound FITC was removed by gel filtration chromatography on Sephadex G-25 eluted with phosphate-buffered saline (150 mM NaCl, 10 mM K' phosphate, pH 7.4).

Cell Culture-Porcine thoracic aortic endothelial cells were cul- tured as described previously (20). Third or fourth passage cells were grown to confluence in 25-cm2 flasks (Costar; Cambridge, MA) in M199 medium with 10% fetal calf serum at 37 "C before use. Cells were passaged in a 1:3 split ratio with 0.05% trypsin and used by the fourth passage to avoid changes in endogenous antioxidant enzyme activities (21).

Sample Preparation-Cells were washed three times with Hank's balanced salt solution (HBSS) before being removed from culture flasks with a cell scraper (Costar), washed again in 5 ml of HBSS, and sonicated in 1 ml of 50 mM K+ phosphate, 0.1% Triton X-100, pH 7.8. The sample was centrifuged at 15,000 X g for 30 min and the supernatant assayed for either catalase or superoxide dismutase ac- tivity. The amount of '"I-label uptake was determined from 150-pl

* J. S. Beckman, R. L. Minor, Jr., C. W. White, J. E. Repine, and G. M. Rosen, unpublished data.

A. Abuchowski, personal communication.

samples of cell homogenate by y-counting. Protein concentrations of cell homogenates were measured by the method of Lowry et al. (22).

"Cr Release Studies-Endothelial cells were grown in 2.2-cm' well plates (Costar) and prelabeled with 3 pCi/ml 'lCr added to the culture medium for 4 h, washed 3 times, and then incubated for 24 h with either native or PEG-catalase. The cells were then washed five times and maintained in HBSS for the duration of xanthine oxidase expo- sure. Xanthine oxidase (10 milliunits/ml) and xanthine (100 p M ) were added to HBSS, and the release of 'lCr from cells into the medium was measured 4 h later. The percentage of 51Cr release for each treatment was calculated by subtracting the amount of "Cr released by untreated cells (not exposed to xanthine oxidase) and dividing by the additional amount of 'lCr released when untreated cells were exposed to xanthine plus xanthine oxidase.

Spin-lubeling Measurements-Endothelial cell monolayers were first washed free of serum-containing medium with HBSS containing 0.7 mM L-glutamine and 5 mM glucose, pH 7.4, and labeled for 3 h at 37 "C in the same buffer with 100 p~ 4-maleimido-2,2,6,6-tetra- methylpiperidinooxyl (Mal-6-TEMPO) as reported previously (23). The labeled cells were washed free of unreacted labeling reagent, removed from flasks by scraping, and resuspended to a density of 400,000 cells/ml, which corresponds to about 0.4 mg of protein. The labeling procedure did not affect cell viability or the rate of cell respiration. Immunoblot analysis using rabbit anti-Mal-6-TEMPO IgG following sodium dodecyl sulfate-polyacrylamide gel electropho- resis of solubilized endothelial cell membranes showed predominant labeling of two major membrane proteins (23). The EPR spectra were obtained from a Varian Associates E9 spectrometer (Palo Alto, CA) operated at 9.5 GHz with 100-kHz modulation. Spectra from 0.5-ml samples were recorded at room temperature in a flat quartz cuvette with a microwave power of 14 milliwatts and a modulation amplitude of 0.63 gauss.

Statistical Analysis-The rate of PEG-conjugated enzyme uptake by endothelial cells was determined by multiple regression. The rate of change in the W/S ratio during exposure to xanthine oxidase was determined by linear regression with the difference between slopes from the regression lines determined by Tukey's post hoc procedure (24). Changes in the extent of 'lCr release from endothelial cells by catalase and PEG-catalase were determined by analysis of variance and Scheffe's post hoc test (25).

RESULTS

Association of PEG-Catalase with Endothelium-Incubation of cultured porcine aortic endothelial cells with lZ5I-labeled PEG-catalase produced a linear, dose-dependent increase of cell-associated enzyme activity with time (Fig. 1). There was an initial rapid cell association of PEG-catalase within the first 4 h which was followed by a slower, continuous uptake over the next 24 h. This pattern of PEG-catalase uptake was described by the following equation:

Units = a,t[PEG-CAT] + a2[PEG-CAT] + a3 (1)

where units equal the catalase activity/mg cell protein meas- ured at time t following the addition of PEG-catalase, [PEG- CAT] is the concentration of PEG-catalase added to the medium, a, is the slow, continuous secondary rate of PEG- catalase uptake/h, a2 is the initial rapid phase of PEG-catalase uptake, and u3 is the endogenous catalase activity of control cells. Multiple regression analysis of the data shown in Fig. 1 yielded a correlation coefficient of 0.93, and the values deter- mined for a,, a2, and a3 are given in Table I. These values were used to calculate the fitted lines shown in Fig. 1 for each concentration of PEG-catalase. Thus, addition of 1 mg/ml PEG-catalase to cell monolayers doubled endogenous catalase activity within 4 h and resulted in an additional increase in activity of l l%/h over the next 20 h. This increase in both catalase activity and enzyme-associated radioactivity could not be removed after extensive washing of PEG-catalase- treated cells.

In the above experiment, the cell uptake of PEG-catalase measured by '''1 radioactivity closely paralleled the increase in catalase specific activity. The specific activity of PEG-

6886 12a

1W

20

0

Augmenting Endothelial Antioxidant Enzymes

INCUBATION TIME (HR)

FIG. 1. Polyethylene glycol-catalase uptake by endothe- lium. The increase in catalase specific activity/mg cell protein is shown for endothelial cell monolayers incubated with the following concentrations of PEG-catalase: 0, 1.0 mg/ml; A, 0.5 mg/ml; 0, 0.1 mg/ml; A, 0.05 mg/ml; and H, untreated cells. Regression lints were calculated from Equation 1 using the coefficients given in Table I for each concentration of PEG-catalase. Each point represents the mean f S.E. of three replicates.

TABLE I Analysis of the cell association of PEG-catalase by multiple regression

The data shown in Fig. 1 were fitted t o Equation 1 to determine the coefficients al, az, and a3. Each incubation time and PEG-catalase concentration contained three replicates with a total of 39 observa- tions used in the regression. The overall correlation coefficient was 0.93, and all coefficients were significant at the 0.0001 level.

Coefficient Estimate * S.E. Units

a1 2.5 _+ 0.24 Units (mg PEG-catalase)"

a2 21 f 4.2 Units (mg PEG-catalase)" h" (mg cell protein)"

(mg cell protein)" aa 22 * 1.5 Units (me cell motein)"

catalase added to culture medium was 6.5 cpm of "'I/unit catalase. When the radioactivity uptake by cells was plotted against the increase in cell catalase activity (Fig. 2), it was found that 7.6 f 0.6 cpm of lZ51/unit catalase was taken up by the endothelial cells. From these data, we estimate that at least 85% of PEG-catalase taken up by endothelial cells retained its activity, probably by electrostatic membrane- protein interactions (29).

Incubation of cell monolayers with 0.5 mg/ml native cata- lase increased catalase activity from 11 units/mg cell protein to 30 units/mg cell protein within the first 4 h (Fig. 3). Unlike PEG-catalase-treated cells, longer incubation with native cat- alase did not lead to further increases of cellular catalase activity. This difference was not due to the presence of PEG alone, because the addition of 1% PEG to native catalase did not increase endothelial catalase specific activities (Fig. 3). Thus, significant augmentation of cellular catalase activity depended upon PEG being covalently bound to catalase. The uptake of radiolabel by cells incubated with '2'I-catalase was far greater than that observed when cells were incubated with a similar isotope specific activity and protein concentration of '2'I-PEG-catalase (Fig. 4). Radiolabel incorporation of lZ5I-

catalase was 20 times greater than could be accounted for by the increase in catalase enzyme activity after 24 h of incuba- tion (Table 11). The most likely explanation for this is that native catalase was avidly bound to and degraded by endothe- lial cells and that radiolabeled amino acids or peptides re- mained associated with cell.

Association of PEG-Superoxide Dismutase with Endothe- lium-Incubation of cells with 0.5 mg/ml PEG-superoxide dismutase increased cellular superoxide dismutase-specific activity by about 17%/h (Fig. 5). During the first 4 h, incu- bation with 0.5 mgjml native bovine Cu,Zn superoxide dis- mutase nearly doubled endothelial superoxide dismutase ac- tivities from 4.4 f 0.8 to 8.2 f 0.8 units/mg cell protein. Thereafter, the activity for native superoxide dismutase- treated cells did not increase significantly, rising only to 9.0 f 0.6 units/mg cell protein after 24 h (Fig. 5 ) . As with native catalase, Cu,Zn superoxide dismutase was predominantly in- corporated as an inactive species, since there was a progressive increase in the ratio of cpm of "'I/unit superoxide dismutase in endothelial cells compared with medium counts/min of lZ5I/unit superoxide dismutase (Fig. 6). If cells incubated with native superoxide dismutase for 8 h were washed and placed in fresh medium, the counts/min of lz5I/unit superoxide dis- mutase ratio declined over the next 16 h to near control values. In contrast, cells incubated with PEG-superoxide dis- mutase showed a constant ratio of counts/min of '"I/unit superoxide dismutase over endogenous superoxide dismutase activity that was the same as the specific activity of "'I-PEG- superoxide dismutase added to the medium. Furthermore, if medium containing PEG-superoxide dismutase was replaced with fresh medium after 8 h incubation, cellular superoxide dismutase activities remained elevated and the ratio of counts/min of "'I/unit superoxide dismutase remained con- stant. When lysates of endothelial cells incubated for 24 h with '2'I-labeled PEG-superoxide dismutase were chromato- graphed over Sephadex G-25, all of the radioactivity from the PEG-superoxide dismutase-treated cell lysates eluted in the void volume. These results show that PEG-superoxide dis- mutase is resistant to degradation by endothelial cells to smaller peptides or amino acids.

The uptake of FITC-PEG-superoxide dismutase could also be observed by fluorescence microscopy, demonstrating a punctate fluorescence distribution pattern (Fig. 7). Mechani- cal injury to the endothelium, produced by scratching the monolayer with the sealed tip of a glass Pasteur pipette, greatly stimulates pinocytosis in the endothelial layer that grows into the "injured" area (26). Forty-eight h after injuring endothelium in the presence of FITC-PEG-superoxide dis- mutase, the injured area showed numerous fluorescent patches which, upon focusing through the cell, were uniformly distributed at all depths. This was consistent with the uptake of fluorescent protein into discrete subcellular vesicles which might be endosomes and lysosomes.

Resistance to Oxidant Stress-The degree of protection afforded by superoxide dismutase or catalase against an oxi- dant stress imposed by xanthine plus xanthine oxidase de- pends upon the method chosen for assaying cellular injury. We have made use of two markers of cellular injury: the xanthine plus xanthine oxidase-induced release of 'lCr-la- beled proteins from cells, and EPR detection of membrane organizational changes measured by changes in the peak amplitude of the two major binding environments represent- ing weakly and strongly immobilized proteins labeled with Mal-6-TEMPO. Our previous studies have shown that xan- thine plus xanthine oxidase-induced 51Cr release from endo- thelial cells is predominantly inhibitable by catalase (21),

FIG. 2. Comparison of endothelial cell-associated radioactivity after incubation with ""I-PEG-catalase compared to catalase specific activ- ity for each culture flask for all of the data shown in Fig. 1. The l i n e represents the best fit by linear regres- sion (rZ = 0.92).

0 2 4

TIME (HR)

Augmenting Endothelial Antioxidant Enzymes 6887

0 0 0

I I I I

0

t

2 1

FIG. 3. Augmentation of endothelial cell catalase specific activity. Cell incubations contained 0.5 mg/ml PEG-catalase (W), native catalase (O), or native catalase plus 1% PEG (e) (average M. 5000). Control cell activities are shown (A).

while xanthine oxidase-induced changes in endothelial mem- brane protein mobility are inhibitable by superoxide dismu- tase (23).

The release of 51Cr into culture medium from cells following injury is analogous to the release of endogenous lactate de- hydrogenase (27, 28). Control experiments showed that incu- bation of PEG-superoxide dismutase or PEG-catalase (1 mg/ ml) with endothelial cells for 24 h (without exposure to xanthine plus xanthine oxidase) did not enhance 51Cr release from endothelial cells. Under the experimental conditions we employed, exposure to 10 milliunits of xanthine oxidase/ml medium plus 0.1 mM xanthine induced lysis or detachment of 85% of the monolayer. Augmentation of endothelial cell cat- alase activity after 24 h preincubation with PEG-catalase increased resistance of endothelial cells to lysis induced by xanthine oxidase (Fig. 8). Endothelial cell cultures incubated

30

18

16

14

e 12

Y X

* t

60 90 120

UNlTS CATALASVMG PROTEIN

0 E 0 2 4 24

INCUBATION TIME (HR) FIG. 4. The radioactivity associated with endothelial cell

monolayers incubated with 0.5 mg/ml medium of 'aaI-labeled PEG-catalase (W) and lZsI-catalase (0). These values were meas- ured in cell incubations reported in Fig. 3.

with concentrations of 0.5 or 1.0 mg of PEG-catalase/ml medium for 24 h and washed free of enzyme-containing me- dium before xanthine oxidase exposure prevented 80% of the 51Cr release compared with untreated controls. Preincubation with 0.5 or 1.0 mg of native catalase/ml medium reduced 'lCr to a lesser extent. Addition of 1% PEG in the presence or absence of native catalase (0.5 mg/ml) did not affect the amount of 51Cr release compared with untreated cells.

The protection provided by preincubation of cells with native catalase might be due to the apparent electrostatic

6888 Augmenting Endothelial Antioxidant Enzymes

TABLE I1 Percentages of enzyme added to the medium which was taken up by

endothelial cell monolayers after 24-h incubation The catalase data were derived from the experiment shown in Figs.

3 and 4, while the superoxide dismutase data came from Fig. 6. Percentage uptake

Enzyme Radio- activity activity

PEG-catalase 0.07 0.12 Catalase 0.01 0.20 PEG-superoxide dismutase 0.13 0.10 Superoxide dismutase 0.01 0.04

0 2 4

TIME (HR)

24

FIG. 5. The time course of increased cellular superoxide dismutase (SOD) specific activities after incubation with 0.5 mg/ml either native superoxide dismutase (a) or PEG-super- oxide dismutase (B). Control cell activities are shown (A).

attachment of native catalase to membranes (29). Addition of native catalase to endothelial cells followed by washing of the cells increased the catalase specific activity from 11 & 1.2 to 20 f 1.6 units/mg cell protein. We have found that addition of catalase to cationic liposomes leads to membrane aggrega- tion (29). Modification of the surface change of catalase with succinic anhydride greatly diminished the aggregation of li- posomes, suggesting that electrostatic charges on catalase mediate its association with membranes.

Preincubation of cells with PEG-superoxide dismutase did not protect endothelial cells from xanthine oxidase-induced 51Cr release, consistent with previous studies where that na- tive superoxide dismutase added to culture medium at the time of cell exposure to xanthine oxidase was not protective (28). Thus, formation of hydroxyl radical by the superoxide- driven, iron-catalyzed Fenton reaction can be ruled out in the present study, although trace amounts of iron were undoubt- ably present in the HBSS. Ferric chloride added to the xan- thine oxidase and xanthine reaction in HBSS only slightly increased 'lCr release (21), indicating either that sufficient iron is already present in the HBSS to mediate toxicity or that iron plays only a minor role.

Membrane Fluidity-Pretreatment of cells with PEG-su- peroxide dismutase confers endothelial oxidant resistance using another index of injury based upon oxidant-induced membrane fluidity changes. Membrane fluidity was measured by conjugating a maleimide derivative of the spin label TEMPO to sulfhydryl groups of endothelial membrane pro-

I 1 I

0 8 16 24

INCUBATION TIME (HR) FIG. 6. The ratio of cellular '''1 counts/min uersus units of

superoxide dismutase (SOD) activity taken up by endothelial cells incubated with 0.5 mg/ml '261-PEG-superoxide dismu- tase (W) or '"I-superoxide dismutase (a). After 8 h of incubation with '251-PEG-superoxide dismutase or "'1-superoxide dismutase, some flasks were washed three times and placed in fresh medium to observe the disappearance of cell associated enzyme activities (dashed line). The "'I/unit superoxide dismutase ratios shown at time 0 reflect the specific activities of the '251-labeled superoxide dismutase and PEG-superoxide dismutase added to the media.

teins. Part of the EPR spectrum of Mal-6-TEMPO was split into two peaks depending upon the binding environments of Mal-6-TEMPO-protein conjugates. These binding environ- ments were characterized as being either weakly ( W ) or strongly ( S ) immobilized (Fig. 9A). The ratio of the peak heights (the W/S ratio), a measure of membrane disorder, increases as membrane fluidity becomes greater. Under the reaction conditions employed here, the W/S ratio was 2.8 f 0.1 for all labeled control endothelial cells and stable for at least 15 min in the EPR cuvette. Exposure of the cells to either 1% PEG or 1 mg/ml PEG-superoxide dismutase in- creased the W/S ratio from 2.8 to 4.0 with the W/S ratio remaining at 4.0 (in the absence of xanthine oxidase) even after extensive washing of cells treated with either 1% PEG or PEG-superoxide dismutase. These results show a mem- brane association and fluidity change induced by PEG alone or protein derivatives of PEG.

The oxidant stress imposed on Mal-6-TEMPO-labeled endothelial cells by xanthine plus xanthine oxidase caused a steady increase in the W/S ratio of 0.83 f 0.02 min" (Fig. 9B). In all cases involving exposure to xanthine oxidase, the sum of the peak heights for the weak and strong peaks were unchanged during the experiment, indicating that the spin label was not being reduced or otherwise affected by the experimental treatment. Analysis of supernatants of centri- fuged cell preparations showed the increased mobility was not due to the release of proteins from the cell surface into the medium. Previously, we have shown that 20% of the rate of increase in the W/S ratio could be prevented by the addition of superoxide dismutase (100 units/ml) during exposure to

Augmenting Endothelial Antioxidant Enzymes 6889

i

FIG. 7. Cell uptake of FITC-labeled PEG-superoxide dis- mutase. Endothelial monolayers were incubated with 1 mg/ml FITC- PEG-superoxide dismutase in MI99 medium plus 10% fetal calf serum for 48 h. Immediately before adding FITC-PEG-superoxide dismutase, the monolayer was scored with the sealed end of a sterile Pasteur pipette (upper panel). This removed a narrow strip of cells which was completely recolonized 48 h later (middle panel). The middle panel was photographed from a different location than the upper panel but is superimposable with the lower panel. After the 48- h incubation, the Petri plates were washed with HBSS three times. The lower panel shows the punctate distribution of fluorescence within the endothelial cells. The region with intense fluorescence dots in the lower panel corresponds to part of the area scraped 48 h before.

xanthine oxidase (23). In the present study, we found that pretreating endothelial cells with 1 mg/ml PEG-superoxide dismutase for 24 h, followed by washing, was as protective as that afforded by a much greater concentration of superoxide dismutase present during exposure to xanthine oxidase. Ad-

120

100

80 W v)

w -I

E 60

a

6 8

Lo

40

20

a 0

T

A

.5

[CATALASE]

1

FIG. 8. Increased endothelial cell resistance to oxidant stress following treatment with PEG-catalase. Endothelial monolayers were prelabeled with %r and then incubated for 24 h with 0.5 mg/ml either native (0) or PEG-catalase (M). The cell monolayers were then washed five times and placed in HBSS. Xan- thine oxidase (10 milliunits/ml) and xanthine (100 PM) were added to the medium and the release of 51Cr by cells was measured 4 h later. Each point represents the mean f S.E. of eight replicates. At concen- trations at and above 0.4 mg/ml, the differences in 51Cr release between PEG-catalase- and catalase-treated cells were significantly different at the 0.05 level.

dition of superoxide dismutase to suspensions of PEG-super- oxide dismutase-pretreated cells during xanthine oxidase ex- posure did not enhance protection against membrane fluidity changes provided by PEG-superoxide dismutase pretreatment alone (Fig. 10). Although the initial W/S ratio of labeled cells was increased by 1% PEG, preincubation with 1% PEG did not affect the rate of change in the W/S ratio caused by xanthine oxidase-derived free radicals or the protection af- forded by native superoxide dismutase present during the xanthine oxidase exposure. Because catalase previously did not attenuate xanthine oxidase-induced membrane fluidity changes when added either alone or with superoxide dismu- tase (23), no EPR experiments were performed with PEG- catalase.

DISCUSSION

We have shown that conjugation of superoxide dismutase and catalase to PEG enhanced uptake of the active form of these enzymes by cultured endothelial cells and that prior incubation of cells with these enzyme conjugates provided greater resistance to oxidant stress than that afforded by native enzymes. Three possibilities could account for the augmentation of endothelial cell enzyme activity following incubation with PEG-conjugated enzymes: (a) direct penetra- tion of membranes by PEG-enzymes, ( b ) binding of PEG- enzymes to membrane surfaces, and (c) uptake of PEG- enzymes by endocytosis (Fig. 11). Polyethylene glycol conju- gation of proteins probably does not mediate migration di- rectly through cell membranes. Horseradish peroxidase be- comes soluble in benzene when conjugated to PEG, suggesting

6890 Augmenting Endothelial Antioxidant Enzymes

15

12.5

10

0 c- a 7.5 U cn 5

5

2.5

0

r

1 1 I 1 I I

0 2 4 6 8 10 12

TIME (MIN) FIG. 9. A, the EPR spectrum of Mal-6-TEMPO-labeled endothe-

lial cells is shown. B, the time-dependent changes in the W/S ratio are shown for control cells (O), cells exposed to xanthine oxidase (10 milliunits/ml) plus 100 pM xanthine (+), cells pretreated with 1% PEG and exposed to xanthine oxidase plus xanthine (A), and cells incubated in the presence of 300 units/ml superoxide dismutase and xanthine plus xanthine oxidase (m). The lines were fit by linear regression.

that enough hydrophilic residues are masked by PEG to permit an otherwise hydrophilic protein to partition into hydrophobic solvents (30). While PEG itself is quite soluble in aromatic solvents such as benzene, it precipitates rapidly in alkane solvents, a fact made use of in the synthesis of activated PEG for coupling to proteins (7). Hence, PEG- derivatized enzymes would not be expected to be soluble within the aliphatic core of phospholipid membranes and therefore are unlikely to traverse the membrane core.

Polyethylene glycol avidly associates with membranes in the phospholipid headgroup region ( l l ) , which might allow anchoring of PEG-conjugated proteins to exofacial plasma membrane surfaces. Evidence in favor of this concept is the increased W/S ratio of Mal-6-TEMPO-treated endothelial membranes secondary to incubation with PEG-superoxide dismutase, suggesting some influence is exerted upon cell membranes by superoxide dismutase after PEG conjugation.

t 1% PEG + PEG-SOD

FIG. 10. The effect of pretreatment of endothelial cells with 1 mg/ml PEG-superoxide dismutase (SOD) or with 1% PEG on the rate of change in the W/S ratio following exposure to xanthine oxidase plus xanthine. Cell exposures also had native superoxide dismutase (300 units/ml) added directly to the cuvette (open bars) to establish the maximum protection that could be pro- vided by superoxide dismutase. Reaction conditions for exposure to xanthine oxidase are described in Fig. 9. The rates of change in the W/S ratio were determined by linear regression, and the standard errors of the slopes are shown.

The binding of PEG-enzymes to cell membranes does not explain the continuous uptake of enzyme conjugates by endo- thelial cells over 24 h because the binding of PEG to mem- branes should be rapidly saturated (11).

The third explanation entails the uptake of PEG-conju- gated enzymes by endocytosis from the medium. In confluent stationary phase bovine aortic endothelial cells, the rate of medium uptake by endocytosis is reported to be 50 nl h"/106 cells (26). If we assume that this rate of endocytosis is similar in cultured porcine endothelium and take into account cell number and medium volume, between 0.06 and 0.09% of the medium would be taken up by cells by endocytosis in 24 h. The fraction of cellular uptake of both PEG-superoxide dis- mutase and PEG-catalase present in culture medium was 0.07-0.13% in 24 h (Table 11), consistent with the calculated rate of uptake by endocytosis. The punctate distribution of fluorescence observed in endothelial cells incubated with FITC-labeled PEG-superoxide dismutase also suggests that a vesicular uptake process plays a major role in the uptake of PEG-conjugated proteins. Mechanical injury to the mono- layer, known to stimulate endocytosis (26), considerably aug- mented the uptake of FITC-PEG-superoxide dismutase, sup- porting endocytosis as a contributing factor in uptake of PEG- conjugated enzymes.

The difference in the rate of native and PEG-conjugated enzyme uptake by endocytosis can be explained by the in- creased resistance of PEG-conjugated enzymes to proteolysis. Because PEG-protein conjugates are resistant to proteases (7), they can survive within lysosomes, yielding higher cell- associated specific activities. Native catalase added to either crude endothelial cell lysates or delivered to endothelial cells via liposomes is degraded with a half-life of 2-3 h (31). On the other hand, superoxide dismutase is more resistant to proteolysis and liposome-delivered superoxide dismutase was far more stable in endothelial cells than catalase (20). How- ever, Cu,Zn superoxide dismutase loses its zinc cofactor when exposed to an acidic pH and becomes susceptible to protease degradation (32). Thus, native superoxide dismutase present in acidic endosomal and lysosomal compartments should be degraded. In the present study, degradation of both cell- associated native superoxide dismutase and catalase was ob-

Augmenting Endothelial Antioxidant Enzymes 6891

FIG. 11. Three possible mecha- nisms for PEG-conjugated antioxi- dant enzyme uptake by endothelial cells. Direct membrane penetration of PEG-enzymes due to the amphiphilic nature of PEG is considered to be un- likely because PEG would be insoluble in the alphatic core of the membrane. However, evidence suggests that PEG- conjugated proteins may bind to the po- lar head groups of phospholipid mem- branes as well as being taken up by en- docytosis. The same mechanisms apply equally to PEG-catalase. SOD, superox- ide dismutase.

served, as evidenced by the continuous increase in the ratio of counts/min of Iz5I/unit enzyme in cells incubated with lZ5I- labeled proteins (Figs. 4 and 6). This ratio for PEG-superoxide dismutase- and catalase-treated cells remained almost the same as the specific activity of the PEG-superoxide dismutase and PEG-catalase added to the medium, even after 24 h of incubation with the endothelial cells.

On balance, there is evidence to support cell uptake of PEG-enzymes by both membrane binding and endocytosis. These two explanations are not mutually exclusive and could even complement each other. For example, the binding of PEG-enzymes to cell surfaces may enhance endocytosis, thereby increasing rates of endocytic uptake. Further studies are required to differentiate between PEG-enzyme uptake by endocytic bulk phase transport of protein conjugates in solu- tion, or the internalization of PEG-enzymes bound to plasma membrane, which would then be distributed in endocytic vacuoles and throughout the cell by membrane flow (33).

Can PEG-conjugated antioxidant enzymes entrapped within endosomes or lysosomes also provide protection against intracellularly generated superoxide and hydrogen peroxide, since these enzymes are separated from the cyto- plasm by a membrane barrier? Endosomes and lysosomes become strongly acidic by an active proton pumping mecha- nism that requires the presence of an anion channel which allows migration of chloride ions to counterbalance proton accumulation within the endosome (34). Because anion chan- nels permit migration of superoxide across red blood cell membranes (35), it is possible that PEG-superoxide dismutase entrapped within lysosomes or endosomes could act as a sump for intracellular superoxide radicals which traverse the chlo- ride anion channel. Recently, Saez et al. (36) have shown that native superoxide dismutase added to the medium of cultured superior cervical ganglion neurons is rapidly taken up by endocytosis after the cells are transiently depolarized with potassium chloride. These neurons, which had superoxide dismutase entrapped within an endocytic compartment, were more resistant to a 40-min starvation stress, whereas starved neurons that were not depolarized but had superoxide dis- mutase added to the medium were not protected.

The increased activity of PEG-conjugated enzymes ob- served in cultured endothelial cells suggests that prolonged administration of PEG-conjugated enzymes may gradually augment vesicular-associated cell antioxidant enzyme activi- ties in uiuo, implying a more complex pharmacokinetics than reflected by plasma clearance. Although the uptake of PEG- superoxide dismutase and PEG-catalase by endothelium re- ported here is a slow process, the continual exposure of endothelium to PEG-conjugated enzymes having long circu- latory half-lives and stability may lead to significant augmen-

tation of cellular antioxidant activities. Furthermore, areas of reversibly injured endothelium in vivo may concentrate even greater levels of PEG-conjugated enzymes due to injury- enhanced endocytic processes (26).

Our results suggest that an additional consequence of PEG- conjugation to proteins is the enhancement of protein asso- ciation with cells. Uptake of PEG-conjugated enzymes by other cell types almost certainly will occur in uiuo as well. The prolonged treatment with PEG-conjugated adenosine deaminase in children afflicted with severe combined immu- nodeficiency disease due to adenosine deaminase deficiency led to restoration of immune function after 6 months (37). Conceivably, uptake of PEG-adenosine deaminase by imma- ture lymphocytes may have contributed to the intracellular removal of inhibitory metabolites in addition to those secreted into plasma, enabling these cells to differentiate and restore immune function.

Acknowledgments-Both Diagnostic Data Inc. (Mountain View, CA) and Grunenthal GmbH (Aachen, Germany) generously provided Cu,Zn superoxide dismutase for these studies. We thank Zermeena Mirza and M. Kathy Cunningham for their expert assistance in the culture of endothelial cells.

REFERENCES 1. Beckman, J. S., and Freeman, B. A. (1986) in Physiology of Oxygen

Radicals (Taylor, A., and Matalon, S., eds) pp. 39-53, American Physiological Society, Bethesda, MD

2. Freeman, B. A., and Crapo, J. D. (1982) Lab. Znuest. 47,412-426 3. Taylor, A. E., and Martin, D. J. (1983) Physiologist 26, 152-155 4. Petkau, A., Chelack, W. S., Kelly, K., Barefoot, C., and Monas-

terski, L. (1976) Res. Commun. Chem. Pathol. Phnrmacol. 15,

5. Turrens, J. F., Crapo, J. D., and Freeman, B. A. (1984) J. Clin.

6. Hutchinson, F. (1957) Radiat. Res. 7,473-483 7. Abuchowski, A., van Es, T., Palczuk, N. C., and Davis, F. F.

(1977) J. Biol. Chem. 252,3578-3581 8. Abuchowski, A., McCoy, J. R., Palczuk, N., Van Es, T., and

Davis, F. F. (1977) J. Bid. Chem. 252,3582-3586 9. Pyatak, P. S., Abuchowski, A,, and Davis, F. F. (1980) Res.

Commun. Chem. Pathol. Phnrmacol. 29, 113-127 10. Wojcieszyn, J. W., Schlegel, R. A., Lumley-Sapanski, K., and

Jacobson, K. A. (1983) J. Cell Biol. 96, 151-159 11. Boni, L. T., Hah, J. S., Hui, S. W., Mukherjee, P., Ho, J. T., and

Jung, C. Y. (1984) Biochem. Biophys. Acta 775,409-418 12. Taylor, A. E., and Martin, D. J. (1983) Physiologist 26, 152-155 13. McCord, J. M., and Fridovich, I. (1969) J. Biol. Chem. 244,6049-

14. Bergmeyer, H. R. (1963) Methods of Enzymatic Analysis, pp. 885-

15. Fridovich, I. (1970) J. Biol. Chem. 245,4053-4057 16. Biagioni, S. (1978) Anal. Biochem. 89, 616-622 17. Habeeb, A. F. S. A. (1966) Anal. Biochem. 14, 328-336 18. Fraker, P. J., and Speck, J. C., Jr. (1978) Biochem. Biophys. Res.

641-657

Invest. 73,87-95

6055

888, Academic Press, NY

Commun. 80,849-857

6892 Augmenting Endothelial Antioxidant Enzymes 19. Rinderknecht, H. (1962) Nature 93, 167-169 20. Freeman, B. A., Young, S. L., and Crapo, J. D. (1983) J. Biol.

21. Bishop, C. T., Mirza, Z., Crapo, J. D., and Freeman, B. A. (1985)

22. Lowry, 0. H., Rosebrough, N. J., Farr, A. J., and Randall, R. J.

23. Freeman, B. A,, Rosen, G. M.. and Barber, M. J. (1986) J. Biol.

24. Sokal, R. R., and Rohlf, F. J. (1981) Biometry, 2nd Ed. pp. 507-

25. Wallerstein, S., Zucker, C. L., and Fleiss, J. L. (1980) Circ. Res.

26. Davies, P. F., Selden, S. C., and Schwartz, S. M. (1980) J. Cell.

27. Andreoi, S. P., Baehner, R. L., and Bergstein, J. M. (1985) J.

28. Freeman, B. A., Turrens, J. F., Mirza, Z., Crapo, J. D., and Young,

Chem. 258, 12534-12542

Zn Vitro 21,229-236

(1951) J. Biol. Chern. 193,265-275

Chem. 26 1,6590-6593

509, W. H. Freeman Co., NY

47, l -9

Physiol. 102,119-127

Clin. Invest. 106, 253-261

S. L. (1985) Fed. Proc. 44, 2591-2595

29. Tanswell, A. K., and Freeman, B. A. (1987) J. Appl. PhysioL 63,

30. Takahashi, K., Nishimura, H., Yoshimoto, T., Saito, Y., and Inada, Y. (1984) Biochem. Biophys. Res. Commun. 121, 261- 265

347-352

31. Freeman, B. A. (1988) Biochem. Pharrnacol., in press 32. Forman, H. J., and Fridovich, I. (1973) J. Biol. Chern. 248,2645-

33. Steinman, R. M., Mellman, I. S., Muller, W. A., and Cohn, Z. A.

34. Stahl, P., and Schwartz, A. L. (1986) J. Clin. Znuest. 77,657-662 35. Kellogg, E. W., 111, and Fridovich, I. (1977) J. Biol. Chem. 252,

36. Saez, J. C., Kessler, J. A., Bennett, M. V. L., and Spray, D. C. (1987) Proc. Natl. Acad. Sci. U. S. A . 84, 3057-3059

37. Herschfield, M. S., Buckley, R. H., Greenberg, M. L., Melton, A. L., Schiff, R., Hatem, C., Kurtzberg, J., Markert, M. L., Ko- bayashi, R. H., Kobayashi, A. L., and Abuchowski, A. (1987) N . Engl. J. Med. 316, 589-596

2649

(1983) J. Cell. Biol. 96, 1-27

6721-6728