ATP-dependent and NAD-dependent modification of glutamine synthetase from Rhodospirillum rubrum in vitro.

Glutamine synthetase from the photosynthetic bacterium Rhodospirillum rubrum is the target of both ATP- and NAD-dependent modification. Incubation of R. rubrum cell supernatant with [alpha-32P]NAD results in the labeling of glutamine synthetase and two other unidentified proteins. Dinitrogenase reductase ADP-ribosyltransferase does not appear to be responsible for the modification of glutamine synthetase or the unidentified proteins. The [alpha-32P]ATP- and [alpha-32P] NAD-dependent modifications of R. rubrum glutamine synthetase appear to be exclusive and the two forms of modified glutamine synthetase are separable on two-dimensional gels. Loss of enzymatic activity by glutamine synthetase did not correlate with [alpha-32P]NAD labeling. This is in contrast to inactivation by nonphysiological ADP-ribosylation of other glutamine synthetases by an NAD:arginine ADP-ribosyltransferase from turkey erythrocytes (Moss, J., Watkins, P.A., Stanley, S.J., Purnell, M.R., and Kidwell, W.R. (1984) J. Biol. Chem. 259, 5100-5104). A 32P-labeled protein spot comigrates with the NAD-treated glutamine synthetase spot when glutamine synthetase purified from H3 32PO4-grown cells is analyzed on two-dimensional gels. The adenylylation site of R. rubrum glutamine synthetase has been determined to be Leu-(Asp)-Tyr-Leu-Pro-Pro-Glu-Glu-Leu-Met; the tyrosine residue is the site of modification.

A 32P-labeled protein spot comigrates with the NAD-treated glutamine synthetase spot when glutamine synthetase purified from H332P04-grown cells is analyzed on twodimensional gels. The adenylylation site of R. rubrum glutamine synthetase has been determined to be Leu-(Asp)-Tyr-Leu-Pro-Pro-Glu-Glu-Leu-Met; the tyrosine residue is the site of modification.
Under nitrogen-fixing conditions in photosynthetic bacteria, the glutamine synthetase (Equation l)-glutamate synthase (Equation 2) pathway carries out the central reactions of ammonia assimilation: ATP + NH,' + glutamate Mg*+ -glutamine + ADP + P, (1) NADPH + glutamine + OL -ketoglutarate (2) -+ 2 glutamate + NADP+ The regulation of glutamine synthetase by gene expression, feedback inhibition, and covalent modification has been studied extensively in enteric bacteria (l-3). In Escherichiu coli, glutamine synthetase exists as a dodecamer of 12 identical subunits, each of which can be regulated independently by the reversible adenylylation of a specific tyrosine. Both the * This work was supported in part by National Science Foundation Grant DCB-8821820.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Adenylyltransferase in its turn is regulated by the PI~ protein, which itself is regulated through the reversible uridylylation of a specific tyrosine, carried out by uridylyltransferase.
Uridylyltransferase responds to the ratio of glutamine to a-ketoglutarate, shifting glutamine synthetase to the more adenylylated state as the ratio of glutamine to a-ketoglutarate increases, and vice versa. This bicyclic cascade provides both signal amplification and the ability to fine-tune the activity of glutamine synthetase in response to the organism's changing nitrogen needs.
Regulation of glutamine synthetase activity in phototrophic bacteria is similar to that in the enteric bacteria. There is evidence of adenylylation of glutamine synthetase in Rhodopseudomonas palustris (4), Rhodobacter sphueroides (5), and Rhodobacter capsulatus (6, 7). In the case of Rhodospirillum rubrum, adenylylation of glutamine synthetase has also been shown, with some differences from the E. coli enzyme. Adenylylated glutamine synthetase from R. rubrum shows loss of y-glutamyltransferase activity in both the presence and absence of 60 mM M%' (8) (contrary to the case in E. coli, in which activity is lost only in the presence of 60 mM Mg2+) (9). Treatment of R. rubrum glutamine synthetase with snake venom phosphodiesterase fails to remove AMP (even from a glutamine synthetase peptide containing bound nucleotide) or reactivate glutamine synthetase (8) 13741 either in N-limited cultures (16). Methionine sulfoximine, a glutamine synthetase inhibitor, prevents ammonium switchoff of nitrogenase and slows dark switch-off. However, there is no consistent correlation of intracellular glutamine concentration with nitrogenase switch-off (16,18). Studies of site-specific ADP-ribosylation of eukaryotic proteins by bacterial toxins, for example of guanine nucleotidebinding proteins by cholera toxin (19) and pertussis toxin (20) and of elongation factor 2 by diphtheria toxin (21) and Pseudomonas exotoxin A (22), gave rise to the hypothesis that these toxins are interfering in an endogenous regulation. Eukaryotic mono-ADP-ribosyltransferases have since been purified by different groups and their in uiuo roles are being studied (23, 24). ADP-ribosylation in prokaryotes as diverse as R. capsulatus (25) and Pseudomonas maltophilia (26) have also been reported recently. ADP-ribosylation may turn out to be as widespread and important a mode of regulation as phosphorylation.
In this paper, further evidence for the adenylylation of R. rubrum glutamine synthetase is given, and the discovery of an ADP-ribosyltransferase activity capable of modifying glutamine synthetase is reported. In vitro ADP-ribosylation of glutamine synthetase and two other proteins, none of which appear to be substrates for DRAT, is also shown.

Purification of Glutamine
Synthetase-The procedure combines steps from the procedures of Streicher and Tyler (28) and Soliman et al. (29). All steps were carried out at 4-8 "C unless stated otherwise. R. rubrum ATCC 11170 was grown in glutamate/malate medium under photoheterotrophic, anaerobic conditions as described previously (8,30). Harvested cells were frozen and stored in liquid nitrogen until used. Cell paste (469 g) was thawed in 500 ml of 100 mM imidazole buffer, pH 7.5, and incubated for 1 h with 150 mg of lysozyme, 30 mg of DNase, and 30 mg of RNase. The cells were then broken in a bead beater, and the extract was centrifuged at 4400 x g for 5 min. The supernatant solution was then centrifuged at 50,000 x g for 3 h. The supernatant solution (440 ml) was diluted with 440 ml of 20 mM imidazole buffer, pH 7.5, containing 40 mM MgC12 and 32.6% (w/v) polyethylene glycol and centrifuged at 50,000 X g for 1 h to precipitate glutamine synthetase. The pellet from this centrifugation was resuspended in 190 ml of 4 mM MnCl*, 50 mM imidazole buffer, pH 7.5, using a Wheaton homogenizer, and centrifuged at 40,000 x g for 1 h. The final supernatant solution was frozen as pellets in liquid nitrogen and stored at -80 "C. Portions were thawed as needed in 4 volumes of 0.125 M KCl, 4 mM MnC12, 50 mM imidazole buffer. DH 7.5. and aDDlied to a 4-ml column of agarose-hexaneadenosine 5'-diphosphaie (AGADP Type 2, P-L Biochemicals) equilibrated with 0.1 M KCl. 4 mM MnCl*. 50 mM imidazole buffer. DH 7.5. The column was washed with 15 ml of equilibration buffer, &en with 200 ml of 0.5 M KCl, 4 mM MnClz, 50 mM imidazole buffer, pH 7.5, and finally with 15 ml of 4 mM MnCl*, 50 mM imidazole buffer, pH 7.5. Glutamine synthetase was eluted with 4 mM ADP in 4 mM MnC$, 50 mM imidazole buffer, pH 7.5. Glutamine synthetase-containing fractions were pooled and applied to a 2.5 x 34-cm Pharmacia Sephacryl S-300 superfine column. The column buffer was 1 mM ADP, 1 mM dithiothreitol, 4 mM MnCl*, 50 mM imidazole, pH 7.5. Glutamine synthetase activity was determined by y-glutamyltransferase assay (8).
Glutamine synthetase from cells grown in KzH3'P01-containing medium was purified essentially as above, following breakage by osmotic shock (31). One-and two-dimensional gel electrophoresis and gel autoradiography of AGADP-purified glutamine synthetase showed that the only 32P-labeled protein present was glutamine synthetase, so the Sephacryl S-300 gel filtration step was not carried out.
For purification of more than 5 mg of glutamine synthetase at a time, ion-exchange and Reactive Blue 2 chromatography (32,33) were substituted for the agarose-hexane-adenosine 5'-diphosphate chromatographv step. The polyethylene glycol-precipitated glutamine syntl&&e-from 200 g df &tamate/malate--&own cells was resusoended in 0.1 M KCl. 4 mM MnCl?. 50 mM imidazole. DH 7.5. and centrifuged as descrided above. The-supernatant was liaied on's 2.6 x 34-cm bed of Whatman DE52 anion-exchange resin which was equilibrated with the same buffer. One bed volume of 0.1 M KC1 biffer was passed through the column after loading. The glutamine synthetase was eluted with a 5-bed volume linear gradient of 0.1-0.5 M KC1 in 4 mM MnClg, 50 mM imidazole, pH 7.5 buffer. The glutamine synthetase-containing fractions were pobled and diluted 1:i with a 2 M NaCl. 10 mM MnCl?. 0.1 mM EDTA. 1 mM dithiothreitol. 50 mM imidazoie, pH 7.15 buffer and loaded & a column of Bio-kad Affi-Gel Blue equilibrated with 1.2 M NaCl in this pH 7.15 buffer. The column was washed with 8 volumes of 1.2 M NaCl buffer, then 3 volumes of the buffer without NaCl. The glutamine synthetase was eluted with 5 mM ADP, 10 mM MnCl*, 0.1 mM EDTA, 1 mM dithiothreitol, 50 mM imidazole, pH 7.15. On two-dimensional gels, glutamine synthetase purified in this way appeared just the same as glutamine synthetase from the first procedure. To our knowledge, the agarose-hexane-ADP matrix is no longer available, and the Affi-Gel Blue method is now used exclusively.
Glutamine synthetase was purified from cells grown on limited ammonium medium, using the anion-exchange/Affi-Gel Blue procedure. All buffers were evacuated and flushed with nitrogen and contained 1 mM dithiothreitol. Buffers used in the cell breakage and polyethylene glycol precipitation steps also contained 5 mM dithionite. The first steps of the purification through the resuspension of the polyethylene glycol-precipitated glutamine synthetase were carried out in an anaerobic glove box. The chromatography steps were carried out as normal.
In some experiments, purified glutamine synthetase was concentrated to 5-8 mg/ml in an Amicon stirred cell ultrafiltration unit, using a YM-10 membrane.

Preparation of Crude Extracts
for Assay of ADP-Ribosyltransferae Actiuity-All steps were carried out anaerobically at 4-8 "C unless otherwise noted. One to five grams of frozen cells grown either in limited ammonium (2 mM), high ammonium (20 mM), or glutamate/ malate medium were thawed in equal volumes of 0.1 mMdithionite, 1 mM dithiothreitol, 1 mM ADP, 50 rr~ EDTA. 100 mM MOPS buffer, pH 7.0 (cell-breaking buffer), together with 011 mg of DNase, 0.1 mg of RNase, and 0.5 mg of lysozyme. The cells were then broken either by osmotic shock or by sonication under nitrogen using a Heat Systems-Ultrasonics, Inc. model 350 Sonicator and microprobe. A portion of broken cells was frozen as pellets in liquid nitrogen and the rest centrifuged at 40,000 x g for 90 min. The supernatant solution was removed by syringe and frozen as pellets in liquid nitrogen. The pellet was resuspended in cell-breaking buffer and frozen as pellets in liquid nitrogen. Crude supernatant solutions were desalted on Pharmacia PD-10 columns equilibrated with anaerobic 0.1 mM dithionite. 100 mM MOPS buffer. DH 7.0. Desalting was carried out at Incubations were carried out either aerobically in 1.5-ml Eppendorf tubes or anaerobically in cut-off Eppendorf tubes surrounded by 0.5 ml of 100 mM dithionite in rubber-stoppered, nitrogen-filled 5-ml flint glass vials. Reagents were made anaerobic by repeated evacuation and flushing with nitrogen. The reaction mixtures were incubated at 30 "C fir 20 min and stopped by the addition of 0.5 ml of 5% trichloroacetic acid. They were then centrifuged for 5 min, and the supernatant solution was discarded. The pellets were dissolved by sonication in 50 ~1 of two-dimensional gel sample buffer.
In some experiments, as noted, the [n-32P]NAD incubation was stopped by the addition of 0.5 ml of a stop mixture containing 30 ~1 of anti-GS antiserum, 50 ~1 of 10% (v/v) insoluble protein A (Sigma), and 420 ~1 of 100 mM MOPS, pH 7.5. After 60 min on ice, the samples were centrifuged for 30 s, and the supernatants were removed. The pellets were resuspended and centrifuged three times in 1 ml of 2 mM NAD, 0.5% (v/v) Nonidet P-40, 100 mM NaCl, 25 mM KzHP04, pH 7.6 buffer.
The on a DHB-Bio-Rex 70 column to remove any ADP-ribosylated peptides, and further purified by reverse-phase and ion-exchange HPLC as described under "Methods and Materials." The amino acid sequence of the 32P-labeled peptide is Leu-Asp-Tyr-Leu-Pro-Pro-Glu-Glu-Leu-Met, although the identification of the aspartate is uncertain due to interference by ammonium in the sample buffer. That the tyrosine residue is the site of modification was deduced from the observation that position 3 of the sequence was blank and the sequence lacked a tyrosine residue even though one tyrosine was detected unambiguously in the amino acid analysis. The sequence of the adenylylation site of E. coli glutamine synthetase is Asn-Leu-Tyr-Asp-Leu-Pro-Pro-Glu-Glu-Ala-Lys (residues 395-405); the tyrosine is adenylylated (45). An absorbtion spectrum of the purified peptide from R. rubrum glutamine synthetase shows an absorption maximum at 260 nm (Fig. 2), as seen previously in a partially purified 32P-labeled glutamine synthetase peptide (8).
Incubation of crude extracts of R. rubrum (broken cells, supernatant, and pellet) with [w~*P]NAD resulted in labeling of three proteins besides dinitrogenase reductase, as observed by autoradiography of two-dimensional gels (Fig. 3). These proteins were labeled most strongly in the supernatant fraction, so all further experiments were carried out using the supernatant fraction. No labeling unique to the membrane fraction (pellet) was observed under the conditions tested. Spot A appears on the autoradiogram over a protein that is present in very low quantity, but it is labeled relatively strongly. Spot B is not labeled very strongly. The conditions under which labeling of Spot B increases roughly parallel those for the labeling of glutamine synthetase, although the conditions for Spot A labeling are generally the opposite of those for glutamine synthetase labeling (Tables I and III). In extracts of cells grown on glutamate/malate and high ammonium media, glutamine synthetase was labeled most strongly when ADP and ATP were omitted from the incubation. In extracts of N-starved cells, glutamine synthetase was not labeled strongly, even when ADP and ATP were omitted. Thus, either some factor that is required for [w~~P]NAD labeling of glutamine synthetase is lacking in extracts of Nstarved cells, or some inhibitor is present. Dinitrogenase reductase labeling was as expected in these experiments. In limited ammonium cell supernatant, dinitrogenase reductase is labeled strongly when ADP and MgC12 (required for dinitrogenase reductase modification by DRAT) are added to the incubation. In glutamate/malate cell supernatants, much of the dinitrogenase reductase is already modified, so the dinitrogenase reductase is labeled only faintly. Additional DRAT increases the labeling of dinitrogenase reductase in limited ammonium cell supernatants. Aerobic incubation decreases the amount of labeled dinitrogenase reductase because Ozdenatured dinitrogenase reductase is no longer a substrate for DRAT (46). In high ammonium-grown cells, dinitrogenase reductase is not synthesized.
To test for nonenzymatic addition of [ol-32P]adenosine diphosphoribose (ADP-ribose) to proteins, extracts from glutamate/malate-grown cells were incubated with [cx-~~P]NAD together with unlabeled ADP-ribose (Table II). The labeling of glutamine synthetase was unaffected. The labeling of protein Spot B was decreased, either by nonenzymatic addition of ADP-ribose or possibly by inhibition of a transferase activity. The labeling of protein Spot A was increased by the addition of ADP-ribose, perhaps by the same mechanism by which ADP and ATP enhance the labeling of protein Spot A.
The identification of glutamine synthetase as one of the proteins labeled by [c~-~~P]NAD was carried out as follows.  Extracts of glutamate/malate-grown cells were incubated with [cx-~'P]NAD and electrophoresed on two-dimensional gels as described under "Materials and Methods." The proteins were transferred to nitrocellulose and antigens detected by the appropriate antibodies. The developed blots were then autoradiographed. Comparison of the proteins recognized by the various antisera with the location of the [cu-32P]NAD-labeled proteins gave negative results for anti-R. rubrum dinitrogenase and anti-R. rubrum ribulose bisphosphate carboxylase oxygenase and positive results for anti-R. rubrum glutamine synthetase and anti-R. rubrum dinitrogenase reductase (Fig.  4). In the case of glutamine synthetase, the radiolabeled spot corresponded to the most acidic portion of the spot recognized by the anti-glutamine synthetase serum.
When an extract from glutamate/malate-grown cells is incubated with 0.2 mM NAD and [a-32P]ATP (5 x lo6 cpm) instead of [a-32P]NAD, neither the A protein spot nor glutamine synthetase are labeled (Fig. 5)  labeled glutamine synthetase. The [cu-32P]ATP-labeled glutamine synthetase (presumably adenylylated glutamine synthetase) is the major protein spot between unmodified glutamine synthetase and the [Cu-32P]NAD-labeled glutamine synthetase. The hazy streak in the autoradiogram in Fig. 5 is also seen in some autoradiograms of [w~*P]NAD reactions with cell supernatant, and it can be reduced or eliminated by the addition of RNase and DNase either before or after the reaction.
32P-Labeled glutamine synthetase purified from cells grown in glutamate/malate medium appears as three spots in a twodimensional gel (Fig. 6). The center and rightmost spots are radiolabeled. In a two-dimensional gel, in uiuo 32P-labeled glutamine synthetase comigrates with the proteins identified as [c+"P]ATP-and [cu-32P]NAD-labeled glutamine synthe-tase in glutamate/malate-grown cell supernatant (data not shown).
Glutamine synthetase purified from glutamate/malategrown cells migrates as two bands on SDS-PAGE, while glutamine synthetase purified from limited ammonium-grown cells migrates as one band, as was previously shown by Soliman and Nordlund (47) (Fig. 7). In a two-dimensional gel, the glutamine synthetase purified from limited ammonium grown cells migrates as two spots (Fig. 7). It is possible, but not yet shown, that these two spots correspond to unmodified and NAD-modified glutamine synthetase.
To investigate whether glutamine synthetase, protein Spot A, or protein Spot B were substrates for DRAT, inhibition of modification by antibody against DRAT was tested. Treatment of DRAT with a 1:lO dilution of anti-DRAT antiserum for 15 min at 30 "C (or 4 h on ice) reduced the ADP-ribosylation of dinitrogenase reductase by 89-95%, as compared to control DRAT incubated in buffer alone. DRAT incubated with a 1:lO dilution of normal rabbit serum showed activities of 121-143% of the control DRAT. This increase in activity may be due to stabilization of the DRAT activity by serum proteins, because anti-DRAT and normal rabbit serum incubated with dinitrogenase reductase without DRAT had background activities of only l-2% of the control DRAT activity.
Anaerobic treatment of crude supernatants from limited ammonium-and high ammonium-grown cells with anti-DRAT before incubation with [w~*P]NAD strongly inhibited modification of dinitrogenase reductase in the limited ammonium supernatant, as expected. Labeling of the A and B proteins and of glutamine synthetase, however, was not affected in either supernatant (data not shown).
DRAT did not ADP-ribosylate glutamine synthetase under conditions in which glutamate/malate cell supernatant modifies glutamine synthetase in vitro (Table III)  MgC& or ATP/MgC& to the incubation is most likely due to adenylylation.
In attempts to scale up the in vitro labeling of glutamine synthetase with [cx-~*P]NAD, it was found that when 100 pg of desalted glutamine synthetase was added to the normal 50 ~1 (total volume) incubation mixture, the amount of glutamine synthetase labeled was much less than when 10 rg of this glutamine synthetase was added to the mixture. This inhibition was removed if the glutamine synthetase was dialyzed against 100 mM MOPS, pH 7.0 buffer. We suspected that ADP or MnC12 from the purification procedure which was not removed by the desalting column might be responsible for the inhibition. To test this hypothesis, reactions were performed in the presence of 1 mM ADP, MnC12, MgC12, and EDTA (Fig.  8). One mM ADP, MnC12, and EDTA all inhibited the labeling of glutamine synthetase by extracts from glutamate/malategrown cells, while 1 mM MgC12 had no effect or enhanced the labeling slightly. One mM ADP together with 1 mM MgC12 was also inhibitory.
The time course of [a-32P]NAD labeling of glutamine synthetase in extract from glutamate/malate-grown cells was followed as shown in Fig. 9. Labeling was complete by 60 min as determined by densitometry of the autoradiogram. glutamine synthetase, with tyrosine as the modified residue in both cases. The [a-32P]NAD labeling studies were undertaken originally to discover if there were any other protein substrates for DRAT besides dinitrogenase reductase in R. rubrum. Unlike cholera toxin and turkey erythrocyte ADP-ribosyltransferase, which are also NAD:arginine transferases, DRAT is highly specific in the substrates it will modify. Thus far only native dinitrogenase reductase from R. rubrum, K. pneumoniae, Azotobacter vinelandii, and Clostridiumpasteurianum have been shown to be modified by DRAT (14). Three other proteins besides dinitrogenase reductase were labeled in crude extracts of R. rubrum incubated with [LU-~~P]NAD. Although it has not been demonstrated chemically, our working hypothesis is that these proteins are ADP-ribosylated.
Preincubation of the crude extract with antibody against DRAT greatly reduced labeling of dinitrogenase reductase, but did not affect labeling of the other three proteins. From this result it appears that these other three proteins are not substrates for DRAT and that there must be at least one more ADP-ribosyltransferase in R. rubrum.
Finding that R. rubrum glutamine synthetase could be ADP-ribosylated as well as adenylylated in vitro was surprising but not unprecedented. Moss et al. (48) have shown that ovine brain glutamine synthetase and chicken heart glutamine synthetase are inactivated by ADP-ribosylation carried out by a NAD:arginine ADP-ribosyltransferase from turkey erythrocytes. More recently, E. coli glutamine synthetase was also found to be a substrate for this NAD:arginine ADPribosyltransferase (49). ADP-ribosylation of a specific arginine in E. coli glutamine synthetase resulted in the loss of both biosynthetic and y-glutamyltransferase activities. In view of these results, it is interesting that we did not see a loss of R. rubrum glutamine synthetase activity together with NAD-dependent modification. There are two possibilities. One is that the effect of ADP-ribosylation on R. rubrum glutamine synthetase activity is more subtle, a small shift in pH optimum or affinity for substrates for example, than we could detect under our assay conditions. Also, the amino acid ADP-ribosylated in R. rubrum glutamine synthetase is not necessarily the same as that ADP-ribosylated in the other glutamine synthetases by the erythrocyte transferase. The other possibility is that ADP-ribosylation of R. rubrum glu-tamine synthetase does not affect the activity directly, but may be involved in coordination of nitrogenase and glutamine synthetase regulation. In limited ammonium cell supernatant, glutamine synthetase was not labeled by [cz-"'PINAD as strongly as in high ammonium and glutamate/malate cell supernatants.
The adenylylation and ADP-ribosylation of R. rubrum glutamine synthetase appear to be exclusive reactions. In the two-dimensional gel autoradiograms, there was no glutamine synthetase spot that could be labeled by both [a-32P]ATP and [a-32P]NAD. Also, adenylylation of glutamine synthetase required added MgC12 in the incubation, while ADP-ribosylation did not. The modification sites may be near enough to each other to prevent double modification by steric hindrance or a change in protein conformation after one site is modified may conceal the second site.
Work is underway to determine the site of NAD-dependent modification of R. rubrum glutamine synthetase and confirm the identity of the modifying group. The putative glutamine synthetase ADP-ribosyltransferase will also be purified and characterized.