ARCHIVES

OF BIOCHEMISTRY

AND

BIOPHYSICS

Vol. 278, No. 1, April, pp. 26-34, 1990

Metal-Catalyzed Oxidation of Escherichia co/i Glutamine Synthetase: Correlation of Structural and Functional Changes A. Jennifer Rivett’ Laboratory

and Rodney L. Levine2

of Biochemistry,

National

Heart, Lung, and Blood Institute,

Received June 29,1989, and in revised form November

Institutes

of Health, Bethesda, Maryland

20892

1, 1989

Metal-catalyzed oxidation of proteins has been implicated in a variety of biological processes, particularly in the marking of proteins for subsequent proteolytic degradation. The metal-catalyzed oxidation of bacterial glutamine synthetase causes conformational, covalent, and functional alterations in the protein. To understand the structural basis of the functional changes, the time course of oxidative modification of glutamine synthetase was studied utilizing a nonenzymic model oxidation system consisting of ascorbate, oxygen, and iron. The structural modifications induced included: decreased thermal stability; weakening of subunit interactions; decrease in isoelectric point; introduction of carbonyl groups into amino acid side chains; and loss of two histidine residues. These changes did not denature the protein, but instead induced relatively subtle changes. Indeed, even the most extensively modified protein had a sedimentation velocity which was identical to that of the native enzyme. Comparison of the time courses of the structural and functional changes established that: (i) Loss of the metal binding site and of catalytic activity occurred with loss of one histidine per subunit; (ii) increased susceptibility to proteolysis occurred with loss of two histidine residues per subunit. Thus, oxidation at one site suffices to inactivate the enzyme, but two sites must be modified to induce susceptibility to proteolysis. The limited and specific changes induced by metal-catalyzed oxidation are consistent with a site-specific free radical mechanism.

Metal-catalyzed oxidation3 of proteins is implicated in a number of physiologic and pathologic processes, inr A.J.R. received a travel grant from The Royal Society. Present address: Department of Biochemistry, University of Leicester, University Road, Leicester LEl 7RH, United Kingdom. ’ To whom correspondence should be addressed. ’ These systems are correctly referred to as mixed-function oxidation systems because they require oxygen and reducing equivalents. However, mixed-function oxidation is often confused with mixed26

National

eluding intracellular protein turnover, host defense activities, oxygen toxicity, and the aging process (1). Studies of the regulation of turnover of glutamine synthetase (GS)4 from Escherichia coli suggested that the process occurs in two steps. In the first step, the protein is covalently modified, thereby “marking” it for subsequent proteolytic degradation. In the second step, the modified protein is degraded by a proteinase which attacks the marked protein, but not the native form (l-3). Metalcatalyzed oxidation is thus one of several covalent modifications which may serve to mark proteins for intracellular degradation (1, 3). The metal-catalyzed oxidation of GS has been extensively studied, in part of its relationship to protein turnover (4). The enzyme is oxidized and inactivated by several metal-catalyzed oxidation systems (58), and the oxidized enzyme is preferentially degraded by a variety of different types of purified proteinases (9, 10). Many other enzymes are also inactivated by metal-catalyzed oxidation systems (3, 5), and some of these also acquire an enhanced proteolytic susceptibility (11). A model metal-catalyzed oxidation system composed of ascorbate, oxygen, and iron has been used previously to characterize some of the changes occurring during oxidative inactivation of glutamine synthetase (6). These changes were specific and were not simply due to denaturation of the protein by generalized oxidative attack. The specificity is consistent with a site-specific free radical reaction (1,12-14). Two covalent modifications were initially described in GS preparations which were oxidized to the point of almost complete inactivation: the loss of a single histidine residue per subunit was detected by amino acid analysis, and the introduction of carbonyl function oxidases. To avoid that confusion, our laboratory now refers to these as metal-catalyzed oxidations. 4 Abbreviations: DTNB, 5,5’-dithiobis-(2.nitrobenzoic acid), Ellman’s reagent; GS, glutamine synthetase; Hepes, 4-@hydroxyethyl)I-piperazine-ethanesulfonic acid; SDS, sodium dodecyl sulfate. 0003.9861/90

$3.00

CHANGES

DURING

OXIDATION

OF Escherichia

groups was demonstrated by reaction with 2,4-dinitrophenylhydrazine. Other changes detected in the oxidized enzyme included decreased subunit interactions (15) and altered dye-binding characteristics (16). When the enhanced proteolytic susceptibility of oxidized glutamine synthetase was studied in detail using a high-molecular-weight liver proteinase which does not degrade the native enzyme, it was found that complete oxidative inactivation of glutamine synthetase molecules was insufficient to render them susceptible to proteolytic attack. Increased susceptibility to proteolysis requires continued exposure well after loss of glutamine synthetase activity (3, 17, 18). After this longer oxidation, glutamine synthetase is extensively degraded by the proteinase. This paper describes the time courses of structural and functional changes which take place in the glutamine synthetase molecule during oxidation of the enzyme by the ascorbate system. Analysis of these time courses clarified the relationship of the structural alterations to the functional changes of inactivation, loss of divalent cation binding, and increased proteolytic susceptibility. EXPERIMENTAL

PROCEDURES

(Zutamine svnthetase. E. coli glutamine synthetase was purified from a strain (YMClO/pglnG) which overproduces the enzyme. The enzyme was purified as described previously (6,9) and assayed by the p-glutamyltransferase method at pH 7.57 (19). The specific activity was always >105 units/mg and was usually 130 units/mg. Protein concentration and the average state of adenylylation of the enzyme (e modification. GS was oxidized by an ascorbate system (4,6) as described previously (9). Enzyme samples (l-2 ml of a 5 mg/ ml solution) were placed in 75,000 molecular weight cut-off collodion bags (Schleicher & Schuell) and dialyzed against 50 mM Hepes/KOH buffer, pH 7.2, containing 100 mM KC1 and 10 mM MgCl,. For the oxidation, 25 mM ascorbate and 0.1 mM FeCl:< were added to the dialysis buffer (30-50 ml/ml GS solution) from freshly prepared stock solutions. GS samples were incubated with shaking at 37°C in the collodion bags, allowing continued dialysis against the inactivation buffer. At desired times, the oxidation process was stopped by addition of I mM EDTA followed by dialysis at 4°C against buffer without ascorbate or iron. Control preparations were incubated at 37°C in the absence of ascorbate and iron. Following oxidation, GS samples were typically dialyzed over a 2.day period with several buffer changes. EDTA (1 mM) was included in the first two buffer changes. All measurements were made on these freshly prepared, dialyzed samples. Samples in the collodion bags were also transferred to dialyze at 37°C against buffer without ascorbate and iron. Transfers were made at 2, 4, or 6 h, and dialysis was then continued for 6, 4, or 2 h to give a total incubation time of 8 h. For each assay or measurement reported here, these samples had values identical to samples taken directly to dialysis at, 4°C. For clarity in the figures, these results are not shown, but they establish that the observed time courses require continued exposure to the oxidizing system. They do not result from conformational changes or spontaneous reactions which might occur after oxidative modification.

coli GLUTAMINE

SYNTHETASE

27

Heat stability. The heat stability of residual catalytic activity of partially oxidized GS preparations was determined by incubation at 65°C. GS activity was assayed during incubation of a 1 mg/ml solution in 50 mM Hepes/KOH, 100 mM KCl, 1 mM MgC&, pH 7.2. The rate of t.hermal inactivation was calculated from first order plots. Proteolysis. Susceptibility to proteolysis was measured by incubation of GS samples (So-100 Kg) with trypsin (l-2.5 fig) or with a purified high-molecular-weight liver proteinase (30-40 Kg) (9) in 200 ~1 of 50 mM Hepes, 1 mM MgCl,, 100 mM KC1 buffer, pH 7.5, for 30 min at 37°C. Bovine pancreatic trypsin (Type III) was purchased from Sigma Chemical Co. GS degradation was determined by fluorescamine assay of acid soluble products (ll), and this assay gave results which agreed with a competition assay utilized earlier (11). Activity determined by the fluorescamine assay is usually expressed as microgram glycine equivalents released per hour. These units were converted to nanomoles of peptide bonds hydrolyzed per hour by assuming that each glycine equivalent represented one peptide bond hydrolyzed. Thus, 1 pg glycine equivalent represented hydrolysis of 13.3 nmols. Metal binding studies. Metal ions were removed from native and oxidized GS samples (5 mg/ml) by passing them down Chelex columns (15 ml bed volume for 4 mg GS) equilibrated in Chelex-treated 20 mM Hepes, 100 mM KC1 buffer, pH 7.2 (21). A spectrophotometric assay (22) confirmed that the enzyme samples were fully relaxed (i.e., that all divalent metal ions were removed). Metal ion-free enzyme samples (0.5 mg) were then dialyzed overnight in 75,000 molecular weight cutoff collodion bags against 54Mn solutions of known concentration, using 54Mn (>40 Ci/g, MnCl,) purchased from New England Nuclear. The Mn2+ concentration of protein and buffer samples was determined using a Beckman gamma counter. The decrease in absorbance at 290 nm upon addition of EDTA (22) was measured in a Hewlett-Packard model 845OA diode-array spectrophotometer. As expected from the studies of Shapiro and Ginsburg, the decrease was fully reversible upon addition of a 1 mM excess of MgCl,. Sedimentation uelocity. Sedimentation velocity measurements were made at 20°C using a Beckman analytical ultracentrifuge (model E) with Schlieren optics, a two place AnD rotor, and two cells, as described by Shrake et al. (23). One cell contained the native enzyme and the other held the oxidized sample. The GS was dialyzed against a buffer of 20 mM Hepes/KOH, 100 mM KCl, 1 mM MnClz, pH 7.2; protein concentration was 2.5 mg/ml. Within a run, the difference in sedimentation coefficients (As) was computed by the procedure of Howlett and Schachman (24) with a precision of at least ?O.l% in As/s. That computation generated the difference sedimentation coefficient relative to 19.5 S for the native enzyme. Amino acid analysis. Amino acid analysis was carried out as described previously (6, 10) with hydrolysis in a Waters Workstation at 155°C for 45 min. Statistically significant differences were assessed after normalization (6). Aromatic amino acids were also determined spectrophotometrically by multicomponent analysis (25). Determination of carbonyl content. Carbonyl content of oxidized proteins was determined by the tritiated borohydride method, without acid hydrolysis (26). Protein concentration after labeling was determined by multicomponent analysis (25). Gel filtration. The extent of aggregation/dissociation of GS resulting from the oxidation was investigated by HPLC gel filtration using a Pharmacia Superose 6 column (25.ml bed volume) equilibrated in 50 mM potassium phosphate buffer, pH 7.0, containing 100 mM KC1 and 1 mM MgCl,. Approximately 200 pg of GS was applied to the column at a flow rate of 0.5 ml/min. Absorbance of the eluate was monitored at 215 nm and relative amounts of dissociated, aggregated, and dodecameric forms of GS were determined by integration of the area under the peaks. The following standards were used to calibrate the column: thyroglobulin, ferritin, catalase, IgG, ovalbumin, myoglobin, and vitamin B12. Molecular weights of the aggregated GS species were esti-

28

RIVETT

AND

LEVINE

0 2

0

6

4

ASCORBATE

8

INCUBATION,

2

4

6

8

10

10

ASCORBATE

HOURS

FIG. 1. Time course of the loss of enzymatic activity and increase in proteolytic susceptibility resulting from oxidation of glutamine synthetase. Inactivation (m); susceptibility to the liver proteinase (0). Inactivation followed first order kinetics with a rate constant of 0.024 mini’. The pattern of susceptibility to trypsin was the same as for the liver proteinase except that native enzyme was digested at 9% of the rate for the 9.5h-oxidized GS. Native GS had a specific activity of 134 units/mg protein. The maximal rate of proteolysis was 42.3 nmol peptide bonds cleaved/h.

INCUBATION,

HOURS

FIG. 3. Spectral changes associated with the removal of metal ions by EDTA. The change in absorbance at 290 nm is shown as a percentage of the absorbance observed before addition of EDTA (22).

GS samples were made 8 M in urea by addition of solid urea. Samples were applied, focused, and stained with Coommassie blue under automated control of the PhastGel system. RESULTS

mated by extrapolation from the calibration curve and from the manufacturer’s stated exclusion limit (40,000,OOO M,). SDS-polyacrylamide geEelectrophoresis. Subunit molecular weight was determined by SDS-polyacrylamide gel electrophoresis using commercial gradient gels (lo-20% acrylamide, Integrated Separation Systems) and the buffer system of Laemmli (27). GS samples were heated at 100°C for 2 min in the presence of 2% SDS and 2% 2-mercaptoethanol. Isoelectric focusing. A PhastCel system (Pharmacia) was used Gels with 8 M urea gels prepared as suggested by the manufacturer: were incubated for 20 min in 10 ml solution containing 8 M urea, 10% (v/v) Pharmalyte as ampholyte, and 0.5% NP-40 (Calbiochem). The

zoo;

Loss of activity and enhanced proteolytic susceptibility. Figure 1 demonstrates the time courses of inactivation of GS and of the increase in susceptibility to proteolytic attack during oxidative modification. Results with trypsin and with the liver proteinase were similar, although trypsin was able to attack the control enzyme while the liver proteinase was not (18). For both proteinases, there was a lag before the increase in proteolytic susceptibility (Fig. 1). Almost all GS catalytic activity was lost during that lag period. Heat stability. Oxidation of glutamine synthetase also altered the heat stability of the enzyme. The remaining catalytic activity was less stable to thermal inactivation (Fig. 2), as also shown by Stadtman and co-

1.4

0.6

101 0



0.5



1

ASCORBATE



1.5

-



2

INCUBATION,



2.5





3

.



3.5

HOURS

FIG. 2. Decrease in heat stability of residual glutamine synthetase activity following ascorbate oxidation. The stability at 65°C was determined after dialysis to remove the ascorbate and iron. The first order rate constant was 0.025 min-‘, the same as that for inactivation (Fig. 1).

0

2

ASCORBATE

4

I

L

6

a

INCUBATION,

10

HOURS

FIG. 4. Decrease in manganese binding capacity occurring during the oxidation of glutamine synthetase. Equilibrium dialysis against 30 FM Mn2+ (A) and 2 /.LM Mn2+ (0).

CHANGES

DURING

OXIDATION

. 14.0 0

I

I

2

4

ASCORBATE

.

.

.

.

I

6

INCUBATION,

OF Escherichia

8

10

HOURS

FIG. 5. Loss of histidine residues in GS during metal-catalyzed oxidation. Dat,a from the amino acid analyses were normalized as described in Table I. Native GS had 16 histidine residues, as expected from the sequence (50). The rate constant was 0.0284 h-’ with 2.4 h required for the loss of one residue.

workers for oxidative modification of glucose 6-phosphate dehydrogenase and phosphoglycerate kinase (28). Site-specific oxidation of even a single amino acid could account for the changes in thermal stability observed (29, 30). In any case, partial oxidative modification of GS yielded a preparation of decreased specific activity and with decreased stability to thermal inactivation. These are the biochemical hallmarks of enzymes purified from older animals when compared to enzyme purified from younger animals (31). The hypothesis that metal-catalyzed oxidations cause these biochemical changes of aging has been discussed recently (28,32). In native GS, changes in the uv specMetal binding. trum occur upon removal of metal ions from the enzyme by treatment with EDTA [“relaxation” (22, 33)]. As shown in Fig. 3, there was a decrease in the magnitude of spectral changes with samples taken at increasing times of exposure to the ascorbate system. This decrease suggested the loss of a metal binding site in the oxidized enzyme. The ability of oxidized GS to bind divalent metal ions was therefore measured directly with “4Mn (Fig. 4). The oxidized enzyme shows a loss of almost one divalent cation binding site per subunit. It was reported previously that Amino acid analysis. oxidative inactivation of GS was due to the loss of a single histidine residue per subunit, specifically HisZe9 (6, 14). No other amino acid changes could be detected. The proteins in those earlier studies were exposed to the ascorbate system just long enough to inactivate -95% of the enzyme. This corresponds to 2.0-2.5 h in the current study, and we confirmed the loss of one histidine residue at this point in the time course. However, after longer oxidation, two histidine residues per subunit are lost (Fig. 5). No other statistically significant changes were

coli GLUTAMINE

SYNTHETASE

29

detected with the analytical methods used here (Tables I and II)5. In particular, there was no change in content of cysteine residues or of aromatic amino acids. Most of the oxidative modification of amino acids which have been described to date would yield products which are more acidic than the original amino acids (34). Isoelectric focusing of the oxidized series of GS revealed the expected decrease in apparent p1 (Fig. 6). The p1 decreased and the number of bands increased with continuing exposure to the oxidation system, implying that multiple oxidations occurred. Correlation of carbonyl content and histidine oxidation. Metal-catalyzed oxidation of proteins introduces carbonyl groups into the amino acid side chains of the protein, providing a convenient assay of oxidative modification (6,26). During exposure to the oxidation system, carbonyl content increased to a plateau of -0.7 carbonyl per subunit, and this time course matched that of the loss of histidine residues (Fig. 7). While such a correlation suggests that the carbonyl groups derive from the histidine residues, other residues certainly contribute to carbonyl production (34, 35). Thus, the increase in carbony1 content could result either from a continued increase in oxidation of certain residues or from later oxidation of different residues which react more slowly. These two possibilities can be distinguished by examination of the distribution of radioactivity in the amino acid chromatogram after acid hydrolysis (34). Figure 8 shows that the pattern of labeling is very similar for the early and late times of oxidation. Hence, the same amino acids are likely oxidized throughout the time course. These amino acids have not yet been firmly identified. Interrelationship between loss of histidine residues and functional changes. Analysis of the time course of histidine loss permits calculation of the fraction of GS which is unaltered, the fraction which has lost one histidine residue per subunit, and the fraction which has lost two histidine residues per subunit. This analysis generates the time courses for the loss of the first and second histidine residues which can then be compared to the time courses of the other changes (Fig. 9). Loss of enzymatic activity was similar to, but clearly faster than the loss of the first histidine residue. The rat,e at which activity was lost was about twice as fast as the oxidation of histidine. The decrease in manganese-binding capacity paralleled the loss of the first histidine residue per subunit. However, the increased susceptibility to proteolysis does not correlate with the loss of the first histidine. Instead, ins Proline is not detected by the o-phthaldialdehyde method. Methionine shows greater variability than other residues, and methionine sulfoxide formation could be missed due to variable conversion back to methionine (49). However, no methionine sulfoxide formation was detected for inactivated GS (6). One should also note that fractional losses of most residues could not be detected by amino acid analysis.

30

RIVETT

AND

LEVINE

TABLE Amino Hours oxidation:

I

Acid Analyses-Acid

Hydrolysatesa

0

0.33

0.67

1.00

1.50

2.00

2.50

3.00

4.00

5.00

6.00

7.00

8.00

9.50

2.00*

4.00’

Asp Glu Ser His Glr Thr ‘4% Ala TY~ Met Val Phe Ile Leu Lys

1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000

0.978 0.945 0.993 0.993 0.982 0.998 0.983 0.964 0.978 0.991 0.991 1.201 1.002 0.993 1.195

0.983 0.997 0.998 0.988 0.990 1.019 0.985 0.986 0.985 0.995 1.014 0.994 1.035 1.019 1.466

0.942 0.962 0.938 0.885 0.998 0.974 0.999 0.992 0.984 1.020 1.015 1.125 1.022 1.031 1.251

0.987 1.011 0.979 0.959 0.969 0.971 0.977 0.979 1.011 1.034 1.018 0.976 1.060 1.028 1.426

0.999 0.976 0.980 0.944 0.978 0.979 0.988 0.986 0.988 1.012 1.020 1.059 1.026 1.009 1.323

1.012 1.028 0.968 0.939 1.001 0.979 0.965 0.981 0.970 0.991 1.003 1.110 1.000 0.993 1.237

1.031 0.975 1.048 0.920 1.009 1.057 0.938 1.017 0.953 0.979 0.968 1.062 0.989 0.974 1.110

1.048 1.023 1.019 0.943 1.026 1.032 0.973 1.039 0.973 0.927 0.977 0.965 1.010 0.988 1.357

1.021 0.982 1.043 0.883 1.029 1.062 0.939 1.017 0.924 0.919 0.955 1.197 0.961 0.951 1.190

1.042 1.009 1.022 0.882 1.015 1.039 0.968 1.026 0.955 0.978 0.971 0.980 0.995 0.999 1.574

1.057 1.030 1.038 0.890 1.046 1.047 0.944 1.039 0.924 0.939 0.980 0.954 1.016 0.986 0.643

1.033 0.990 1.036 0.890 1.019 1.056 0.963 1.021 0.959 0.962 0.961 1.057 0.973 0.970 1.379

1.019 0.977 0.997 0.843 1.125 0.991 0.936 1.000 0.902 0.911 0.931 1.317 0.955 0.939 0.973

1.042 1.026 1.032 0.975 1.021 1.038 0.972 1.031 0.977 0.963 0.973 0.978 0.986 0.961 1.399

1.031 0.977 1.056 0.895 1.015 1.056 0.959 1.014 0.963 0.949 0.961 1.039 0.996 0.983 1.391

Mean: SD: Mean ~ 2 SD: Mean + 2 SD:

1.000 0.000 1.000 1.000

1.000 0.060 0.880 1.120

1.000 0.016 0.968 1.032

1.000 0.046 0.908 1.092

1.000 0.028 0.944 1.056

1.000 0.024 0.953 1.047

1.000 0.036 0.928 1.072

1.000 0.039 0.922 1.078

1.000 0.034 0.931 1.069

1.000 0.073 0.855 1.145

1.000 0.027 0.946 1.054

1.000 0.046 0.909 1.091

1.000 0.037 0.927 1.073

1.000 0.107 0.786 1.214

1.000 0.030 0.940 1.060

1.000 0.037 0.927 1.073

Residue

n Two entirely separate series of oxidized enzyme were prepared and analyzed. Data were normalized to the results obtained for four separate control samples, as described (6). The mean ? 2 standard deviations defines the 95% statistical limits. Only histidine was reproducibly changed. In the other series (not shown), methionine values were more variable, presumably due to variable loss during hydrolysis in screw cap vials. The series shown here were hydrolyzed in the Waters Workstation after 3 cycles of vacuum/nitrogen purge. In this series, the lysine values are rather variable, but this was not observed in the first series. b This sample was exposed to the ascorbate oxidation system for 2 h, then incubation was continued for another 6 h in the absence of ascorbate and iron. ’ This sample was exposed to the ascorbate oxidation system for 4 h, then incubation was continued for another 4 h in the absence of ascorbate and iron

creased susceptibility parallels the slower loss of the second histidine residue per subunit. Conformational changes ocSedimentation velocity. curring during the oxidation were investigated by sedimentation velocity experiments. In the case of GS, sedimentation velocity is a more sensitive method than circular dichroism for detecting conformational changes (36). There was no significant difference in sedimentation coefficient for the major sedimenting peak of either the just-inactivated or the g-h-oxidized dodecamer when compared to the native enzyme. In the course of the sedimentation Molecular weight. velocity experiments, a small amount of aggregation and of nonsedimenting material was detected in the 8-h-oxidized GS. This was further examined by gel filtration. In native and control enzyme preparations, GS was present entirely as the dodecamer. After 10 h oxidation, 88% of the oxidatively modified enzyme was still present as the 600,000 M, dodecamer. A small amount of dissociation (

Metal-catalyzed oxidation of Escherichia coli glutamine synthetase: correlation of structural and functional changes.

Metal-catalyzed oxidation of proteins has been implicated in a variety of biological processes, particularly in the marking of proteins for subsequent...
1MB Sizes 0 Downloads 0 Views