Inactivation of Glutamine Synthetases by an NAD:Arginine ADP-Ri~osyltransf~rase*

Glutamine synthetase from ovine brain has a critical arginine residue at the catalytic site (Powers, S. G., and Riordan, J. F. (1975) Proc. Natl. Acad. Sci. U. S. A. 72,2616-2620). This enzyme is now shown to be a substrate for a purified NAD:arginine ADP-ribosyl-transferase from turkey erythrocyte cytosol that cat- alyzes the transfer of ADP-ribose from NAD to arginine and purified proteins. The transferase catalyzed the inactivation of the synthetase in an NAD-depend-ent reaction; ADP-ribose and nicotinamide did not sub- stitute for NAD. Agmatine, an alternate ADP-ribose acceptor in the transferase-catalyzed reaction, pre- vented inactivation of glutamine synthetase. MgATP, a substrate for the synthetase which was previously shown to protect that enzyme from chemical inactiva- tion, also decreased the rate of inactivation in the presence of NAD and ADP-ribosyltransferase. Using [32P]NAD, it was observed that approximately 90% inactivation occurred following the transfer of 0.89 mol of [32P]ADP-ribose/mol of synthetase. The erythrocyte transferase also catalyzed the NAD-dependent inactivation of glutamine

The catalytic activity of enzymes located at critical points in metabolic pathways is, in many cases, regulated by covalent modification. Certain bacterial toxins perturb cellular metabolism by catalyzing the transfer of ADP-ribose from NAD to a key enzyme, thereby altering its activity (1-4). Choleragen and Escherichia coli heat-labile enterotoxin catalyze the ADPribosylation of a regulatory component of the hormone-sensitive adenylate cyclase system (1,5-7); evidence from studies with model substrates is consistent with the hypothesis that * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisetnent" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
"-the target amino acid is an arginine (8,9). Diphtheria toxin and Pseudomoms exotoxin A inhibit protein synthesis in susceptible animal cells by ADP-ribosylating elongation factor I1 (2,3); in this case, the target amino acid is a modified histidine residue (10).
Animal cells contain endogenous enzymes that catalyze the mono-~P-ribosylation of proteins in a reaction analogous to that catalyzed by the bacterial toxins (11)(12)(13). The role of the mono-ADP-ribosyltransferases is not clear. Two transferases that have been purified to homogeneity from turkey erythrocytes catalyze the ADP-ribosylation of proteins, arginine, and low molecular weight ~a n i d i n o compounds; amino acids, such as lysine and histidine, are inactive as ADP-ribose acceptors (11,14).
A number of enzymes have critical arginine residues that participate in catalysis. These have been identified by the use of chemical reagents that specifically react with arginine f15). We used glutamine synthetase, which contains an active site arginine that is reactive with such reagents (15), to determine whether this arginine might also be preferentially modified by an ADP-ribosyltransferase.

Inactivation of Glutamine Synthetase by ADP-ribosylation 5101
Protein was determined by a modification of the method of Lowry et al. (19) using bovine serum albumin as a standard.
Purification of the NAD:Arginine ADP-Ribosyltransferase-The NAD:arginine ADP-ribosyltransferase purified from turkey erythrocytes by sequential chromatography on phenyl-Sepharose, carboxymethylcellulose, NAD-agarose, and concanavalin A-agarose as described (12) exhibited one major protein band by sodium dodecyl sulfate-polyacrylamide gel electrophoresis; 1 unit of transferase activity equals 1 pmol of ADP-ribose transferred from NAD to agmatine/ min at 30 "C.

RF
FIG. 1. Sodium dodecyl sulfate-gel electrophoresis of purified chicken heart glutamine synthetase. Left, glutamine synthetase was subjected to electrophoresis on a 10% polyacrylamide gel in the presence of 0.1% sodium dodecyl sulfate. Right, the log of the molecular weight of standard protein (0) was plotted as a function of RF to obtain an approximate molecular weight of 41,000 for glutamine synthetase (0). Standard proteins were phosphorylase b (94,000), albumin (67,000), ovalbumin (43,000). and carbonic anhydrase (30,000). Arrow specifies dye front.

NAD-dependent inactivation of ovine brain glutamine synthetase by
ADP-ribosyltransferase Samples of ovine brain glutamine synthetase (1.34 pg) were incubated in a total volume of 0.1 ml containing 5 mM potassium phosphate (pH 7.0) and 5% propylene glycol for 24 h a t 30 "C with or without 1.3 milliunits of transferase and other additions as indicated. Glutamine synthetase activity was then determined as described under "Methods" after addition of assay components in 0.1 ml. Final concentrations of reagents are noted under "Methods," with other indicated additions for the first incubation being present a t twice the concentration in the second incubation.

Effect of MgATP and agmatine on inactivation of ovine brain glutamine synthetase in the presence of NAD by ADPribosyltransferase
Samples of ovine brain glutamine synthetase (1.34 pg) were incubated for 5 h a t 30 "C in 0.1 ml containing 5 mM potassium phosphate (pH 7.0) and 5% propylene glycol with or without 1.3 milliunits of transferase and other additions as indicated (MgCI2 (20 mM), ATP (15 mM)). The reaction was initiated by the addition of 0.1 ml of mixture to bring the reactants to the concentrations shown under "Methods;" except for MgCI2 and ATP, all other additions shown in the table were present a t ' /z the given concentration in the assay. All assays were run in quadruplicate. Purification of Glutamine Synthetase from Chicken Heart-Chicken hearts (330 g) were homogenized in 1 liter of 20 mM potassium phosphate (pH 7.0), 1 mM EDTA, 1 mM EGTA, 0.1 mM phenylmethylsulfonyl fluoride (buffer A). The homogenate was centrifuged (40 min, 37,000 X g), and the supernatant (950 ml) was applied to a column (2.5 X 25 cm) of phenyl-Sepharose which was then washed with 250 ml of 20 mM potassium phosphate (pH 7.0). The column was eluted with 100 ml of 50% propylene glycol in buffer A. The eluate was applied to a column (2.5 X 8.5 cm) of DE52 which was then washed with 300 ml of the same buffer and eluted with 25 ml of 0.1 M potassium phosphate (pH 7.0), 1 mM EDTA, 1 mM EGTA, 0.1 mM phenylmethylsulfonyl fluoride (buffer B). The DE52 eluate (20 ml) was chromatographed on a column (2.5 X 90 cm) of Ultrogel AcA 34 which was eluted with buffer B. Fractions containing glutamine synthetase activity were pooled and applied to a column (1.2 X 13 cm) of ATP-agarose. The column was washed with 30 ml of buffer B and eluted with 50% propylene glycol in buffer A. The enzyme could also be eluted with ATP, but ATP was not used to avoid contamination of the preparation with nucleotide. Fractions containing glutamine synthetase activity were pooled, adsorbed to DE52 (1.5 X 4 cm), and eluted with 10 ml of buffer B. The ATP-agarose chromatography step was not effective when used before the Ultrogel AcA 34 column.
The purification procedure is summarized in Table I. The specific activity of the purified enzyme was in the range of that observed with other glutamine synthetases (18,19). The purified enzyme exhibited one major protein band on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (see Fig. 1). The molecular weight of the protomeric unit, 41,000, was similar to that of glutamine synthetases from other sources (18,20,21) (Fig. 1). On gel filtration (Ultrogel AcA 34), glutamine synthetase activity co-chromatographed with protein and specific activity was constant across the protein peak (data not shown).

RESULTS
Ovine brain glutamine synthetase was inactivated by incubation with NAD and NAD:arginine ADP-ribosyltransferase (Table 11). ADP-ribose and nicotinamide, products of the enzyme-catalyzed hydrolysis of NAD by the transferase, were inactive (Table 11). NADP, which is utilized much less efficiently by the transferase, was not an effective substitute for NAD (Table 11). Agmatine, an alternate ADP-ribose acceptor in the transferase-catalyzed reaction, prevented the inactivation of glutamine synthetase ( Table 111). Addition of MgATP, previously shown to block chemical inactivation of synthetase by arginine-specific reagents (15), reduced the rate of inactivation of synthetase by transferase and NAD (Fig. 2 Table 111). The transferase-catalyzed inactivation of glutamine synthetase was associated with the transfer of ["PI ADP-ribose from 13*P]NAD to the enzyme. Maximal inhibition of glutamine synthetase activity of 90%, reached after incubation of this enzyme with transferase for -1 h, was associated with transfer of -0.89 z k 0.07 mol of ADP-ribose/ mol of glutamine synthetase. The ratio of moles of [32P]ADPribose/mol of glutamine synthetase to percentage inactivation was -0.99. Glutamine synthetase purified from chicken heart was also inactivated by the transferase in an NAD-dependent reaction (Fig. 3). ADP-ribose and nicotinamide could not replace NAD (Table IV). Inhibition of the chicken heart enzyme by transferase was maximal by 1 h (Fig. 4) and was dependent on that amount of transferase present (Fig. 5). As noted with the ovine enzyme, both agmatine and MgATP protected the synthetase from inactivation ( Table IV) and also decreased the extent of ADP-ribosylation (data not shown). In the presence of [32P)NAD, 0.60 k 0.03 mol of ADPribose was transferred per mol of glutamine synthetase, resulting in a 95% inactivation of the enzyme. The ratio of moles of ~32P]ADP-ribo~/mol of glutamine synthetase to percentage inactivation was 0.63. (15) that the ovine brain glutamine synthetase has a critical arginine residue based on its inactivation by arginine-specific reagents; in these studies, loss of enzymatic activity was associated with the modification of 3 out of a possible 25 arginine residues. By performing the investigations in the presence of MgATP, a substrate for the synthetase that protected -1.4 of the arginine residues from modification, it was concluded that a critical arginine residue was present at the active site. In the present study, using the arginine-specific NAD:arginine ADPribosyltransferase, it was observed that ADP-ribosylation of glutamine synthetases from ovine brain and chicken heart resulted in a loss of enzymatic activity. Both glutamine synthetases were protected from enzymatic inactivation by the addition of a synthetase substrate, MgATP. Saturating concentrations of agmatine, an alternative ADP-ribose acceptor  (14), also blocked inactivation of the synthetases. Specificity of the agmatine effect was verified by demonstrating the formation of ADP-ribose-agmatine, rather than ADP-ribose-glutamine synthetase (data not shown), consistent with the hypothesis that agmatine and glutamine synthetase compete for the active site on the transferase. With glutamine synthetases from both sources, it appeared that inactivation resulted from the transfer of -1 mol of ADP-ribose to 1 mol of enzyme; the reaction thus appeared to be specific.

It was shown by Powers and Riordan
The NAD:arginine ADP-ribosyltransferase may de used as a reagent to catalyze the covalent modification in both pure protein and tissue homogenate of arginine residues. The ADPribosyl(arginine) protein bond is relatively stable in acid and at physiological pH (22). In tissue homogenates, enzyme(s) responsible for degradation of the ADP-ribosyl(arginine) protein bond, if they exist, are relatively inactive under standard assay conditions, and thus unlikefy to present a threat to the stability of the ribosyl(arginine) protein linkage? Reversal of transferase-like reactions requires low pH (5.5-6.0) and high concentrations of nicotinamide (5); it is thus unlikely to proceed under physiological conditions. Since phosphodiesterases that catalyze the degradation of the ADP moiety are common (23), to tag arginine residues in crude extracts it would be preferable to use NAD labeled in the nicotinamide ribose; phosphodiesterase and phosphatase action on ADPribose(arginine) protein would result in the formation of ribose(arginine) protein as a radiolabeled end product.
Indirect evidence that ADP-ribosylation might be involved in the regulation of glutamine synthetase in cells was obtained by assessing the effects of lowering cellular NAD levels on glutamine synthetase activity in Chinese hamster ovary cells. Nicotinamide omission from the growth medium produced a 78 and 109% increase in glutamine synthetase activity in two separate experiments coincident with a lowering of the cellular NAD levels by 90%. Nicotinamide deprivation had no effect on cell growth rate over the 12-h treatment period. NAD levels were also reduced by 75% by exposing Chinese hamster ovary cells to 2 mM 6-aminonicotinamide for 24 h. With cells in stationary phase, the compound increased glutamine synthetase levels by 99 f 15% compared to untreated control^.^ In prior studies, it was shown that ADP-ribosylation was affected by nucleotides such as GTP or ATP which, depending on the protein, either increased, decreased, or had no effect on the rate of modification (24). These experiments did not demonstrate an effect of ADP-ribosylation on function. In the present report, it is clear that MgATP blocks the inactivation of glutamine synthetase. Although it was uncertain from the previous studies whether the modification of the proteins had any selectivity other than the presence of a "readily accessible" arginine, the present investi~ation demonstrates that the tr~sferase-ca~lyzed reaction can be specific for certain arginine residues. Of the 25 arginine residues in ovine brain glutamine synthetase, the erythrocyte transferase and phenylglyoxal selectively modified that residue critical for enzymatic activity. It is clear from model studies on the transferase-catalyzed ADP-ribosylation of arginine and other low molecular weight guanidino compounds that the environment of the guanidino is a critical determinant of its ability to serve as an ADP-ribose acceptor (14). The presence of negatively charged residues in the vicinity of the guanidino moiety decreased its reactivity in the transferase-catalyzed reaction; agmatine and arginine methyl ester were more ef-  3 (left). Effect of NAD on the inactivation of chicken heart glutamine synthetase by the ADPribosyltransferase. Chicken heart glutamine synthetase (12.2 milliunits) was incubated for 1 h at 30 "C in the presence of 10 mM Tricine (pH 7.6), 5% propylene glycol, 10 mM potassium phosphate, 200 mM NaC1, ADPribosyltransferase (8.19 milliunits), and the indicated concentrations of NAD (final volume, 0.1 ml). The synthetase reaction was initiated by the addition of 0.1 ml of mixture to bring the concentrations of reagents to that noted under "Methods;" control assays run with NAD but without transferase indicated that the presence of NAD in the assay did not significantly change synthetase activity. All assays were run in quadruplicate. FIG. 4 (center). Time course of inhibition of chicken heart glutamine synthetase by transferase. Glutamine synthetase purified from chicken heart (6.74 fig) was incubated at 30 "C with 8.19 milliunits of transferase, 0.2 M NaC1, and 1 mM NAD in a total volume of 0.1 ml. At the indicated times, the entire sample was assayed for glutamine synthetase activity as described under "Experimental Procedures." Results are the mean of duplicate incubations expressed as percentage of zero time.
FIG . 5 (right). Effect of NAD:arginine ADP-ribosyltransferase on the NAD-dependent inactivation of chicken heart glutamine synthetase. Chicken heart glutamine synthetase (12.2 milliunits) was incubated for 1 h at 30 "C in 10 mM Tricine (pH 7.6), 5% propylene glycol, 10 mM potassium phosphate, 200 mM NaC1, 1 mM NAD, and the indicated milliunits of ADP-ribosyltransferase (final volume, 0.1 ml). The synthetase reaction was initiated with 0.1 ml of mixture to bring the concentration of reagents to that noted under "Methods." Reaction was run for 30 min at 30 "C; all assays were done in quadruplicate.

TABLE IV
Inactivation of chicken heart glutamine synthetase by ADPribosyltransferase Chicken heart glutamine synthetase (12.2 milliunits) was incubated in the presence or absence of ADP-ribosyltransferase (8.19 milliunits) for 1 h at 30 "C in a mixture containing the indicated additions (MgC12 (20 mM), ATP (15 mM)) and 5% propylene glycol, 200 mM NaC1,lO mM Tricine (pH 7.6), 10 mM potassium phosphate in a final volume of 0.2 ml. Following the first incubation, 0.1 ml of a reaction mixture was added to bring the final concentration of reagents to 50 mM Tricine (pH 7.6), 4 mM NH4C1, 20 mM sodium glutamate (126,000 cpm), 20 mM MgCl,, 15 mM ATP. After 30 min at 30 "C, the assay was terminated as described under "Methods." All assays were run in ouadruulicate.

Additions MgATP
Glutamine synthetase activity -Transferase + Transferase 7.5 1.5 8.0 5.2 6.5 6.9 7.0 6.7 6.6 6.8 7.4 7.4 7.4 7.4 8.0 8.4 7.4 7.4 7.6 7.6 fective substrates than were arginine or guanidinopropionate (14). In addition, the reactivity of an arginine in a protein is in part determined by the nucleophilicity of the guanidino group and thus its ability to displace nicotinamide from NAD'. The secondary and tertiary structures of glutamine synthetase serve as determinants of the pK of the guanidino moiety. A decrease in pK would enhance the reactivity of the guanidino group with phenylglyoxal and at the catalytic site of the transferase. In the case of the transferase-catalyzed reaction as opposed to chemical modification, however, the picture is complicated by the fact that the substrate for the transferase is another protein; in order for the critical arginine to be modified, the catalytic site of glutamine synthetase must be accessible to the active site on the transferase. It is thus appealing to speculate that the specificity reflects an in uiuo significance for this reaction and a function for the NAD:arginine ADP-ribosyltransferase.