Cascade Control of Escherichia cdi Glutamine Synthetase PROPERTIES OF THE P,, REGULATORY PROTEIN AND THE URIDYLYLTRANSFERASE-URIDYLYL-REMOVING ENZYME

SUMMARY The PII regulatory protein of Escherichia coli glutamine synthetase exists in two interconvertible forms: a uridylylated form (PIID) which promotes the deadenylylation of glutamine synthetase and an unmodified form (PI,,) which promotes the adenylylation of glutamine synthetase (Mangum, J. H., Magni, G., and Stadtman, E. R. (1973) Arch. Biochem. Bio-phys. 158, 514525). P,, has been purified to homogeneity. Its molecular weight is 44,000. The protein is composed of four subunits, each with a molecular weight of approximately 11,000. The subunits are identical as judged by: (a) the homogeneity of the subunits in sodium dodecyl sulfate, 8 M urea, and 6 M guanidine HCl; (b) the minimal molecular weight calculated from the amino acid composition; and (c) the isolation of only two tryptic peptides containing tyrosine (there


158, 514525).
P,, has been purified to homogeneity. Its molecular weight is 44,000. The protein is composed of four subunits, each with a molecular weight of approximately 11,000. The subunits are identical as judged by: (a) the homogeneity of the subunits in sodium dodecyl sulfate, 8 M urea, and 6 M guanidine HCl; (b) the minimal molecular weight calculated from the amino acid composition; and (c) the isolation of only two tryptic peptides containing tyrosine (there are 8 tyrosyl residues per 44,000 molecular species). Following iodination of PIIA and PIID with 12jI in the presence of chloramine-T, tryptic digestion yields two radioactive peptides from PII, and only one from P,,,.
Since a tyrosine with a substituted hydroxyl group cannot be iodinated, this result indicates that 1 tyrosyl residue in each subunit is modified by the covalent attachment of UMP. This conclusion is supported also by the fact that treatment of PIID with snake venom phosphodiesterase results in the release of covalently bound UMP and the stoichiometric appearance of phenolate ion (pH 13) as measured by ultraviolet absorption spectros-

COPY.
The enzyme activities (uridylyl-removing) responsible for removal and (uridylytransferase) responsible for attachment of UMP to PII have been partially purified. These activities co-purify through a variety of procedures, including hydrophobic chromatography, and are stabilized by high ionic strength buffers. Whereas Mn*+ alone supports only uridylylremoving activity, ATP, cu-ketoglutarate, and Mg2+ support both uridylyl-removing and uridylyltransferase activities.
In addition to its control by cumulative feedback inhibition and repression, the glutamine synthetase activity of Escherichia coli is regulated by the covalent attachment and removal of AMP from a specific tyrosyl residue in each of the enzyme's 12 identical subunits (l-4). Adenylation of a subunit converts it to a less active form dependent upon SW+ (1). The enzyme's activity is thus controlled by the average number of adenylylated subunits per molecule which can vary from 0 to 12. Both adenylylation and deadenylylation of glutamine synthetase are catalyzed by a single enzyme, adenylyltransferase (ATase)' (5). Adenylylation involves a transfer of the AMP moiety of ATP into an AMP-0-tyrosyl linkage (4), whereas deadenylylation involves a phosphorolysis of the latter to yield ADP (6). Although adenylyltransferase catalyzes both reactions, its ability to adenylylate or deadenylylate glutamine synthetase (GS) is modulated by the regulatory protein PII and metabolic effecters including oc-ketoglutarate, ATP, glutamine, and inorganic phosphate (Pi).
ATase ATP + GS p AMP-GS + PP, IIA Tl AMP-GS + P, 'LID > ATase GS + ADP SCHEME I Recent work from this laboratory has demonstrated that the regulatory protein PII exists in two interconvertible forms (7). As is shown in Scheme I, one form, PIIA, stimulates the adenylyltransferase-catalyzed adenylylation of glutamine synthetase, whereas the other form, PIID, is required for the adenylyltransferase-catalyzed deadenylylation. When PII, is incubated in the presence of UTP, ATP, a-ketoglutarate (crKG), R/In*+ or Mg2+, and another enzyme, uridylyltransferase (UTase), it is converted to PIID. This conversion involves the covalent attachment of UMP to the protein (Reaction 1) (8).
UTP + P,, UTase cuKG, ATP, Mg2+ p P,,-UMP + PP, We now report the purification of PII to homogeneity, demonstrate that the protein is a tetramer of identical subunits, and that, like glutamine synthetase, its activity is modulated by the covalent attachment of a nucleotidc to a specific tgrosyl residue in each of the subunits. In addition, the protein(s) exhibiting UR and uridylyltransferase activities has been partially purified and characterized, and evidence is presented suggesting that both activities may be a property of the same enzyme or enzyme complex. Preliminary reports of this work have appeared (9, 10). Step 1: E&action-Cell-free extracts were prepared by homogenization of the frozen cell paste in 20 mM 2-methylimidazole, pH 7.6, 10 mM 2-mercaptoethanol, and 0.1 mM K&g EDTA (standard buffer; 2 liters of buffer per kilo of cell paste) in a Waring Blendor for 90 s. The homogenized extract was filtered through cheesecloth and then passed two times through a French pressure cell at 12,000 p.a.i. Cell debris was removed by centrifugation for an hour at 12,000 x g. No PIID activity could be detected at this step.
Step 3: Streptomycin-NH4S04 Precipitation-A 10% solution of streptomycin sulfate was added slowly to the supernatant fluid from Step 1 to a final concentration of 1%. After 15 min of slow stirring, the solution was centrifuged as above. To the supernatant, solid lYH,S04 was added to a final concentration of 300 mg/ml (5070 saturated).
After 30 min of stirring the precipitate was removed by centrifugation (15,000 x g for 45 min). The pellet was either stored frozen (-80') or immediately resuspended in a minimal volume of standard buffer and dialyzed two times against 200 volumes of this buffer. PIID activity was first detectable at this stage. There was no loss of activity due to storage of the IXH,SO, pellet at -80".
Step 3: DEA E-cellulose Chromatography-The dialyzed extract from Step 2 was applied to a DEAE-cellulose column equilibrated with standard buffer containing 50 mM KCl. Approximately 33 mg of protein were applied per ml of bed volume. After removal of unabsorbed protein by washing with starting buffer, a linear KC1 gradient (50 to 500 mM) in standard buffer was applied. Fractions were collected and assayed for their glutamine synthetase, adenylyltransferase, Prr, and UR uridylyltransferase activities. Fig. 1 illustrates a typical elution profile from DEAEcellulose.
Glutamine synthetase and UR uridylyltransferase activities co-elute with the major protein peak at a salt concentration of approximately 75 to 100 mM KCl. This is the first step in the purification where UR uridylyltransferase activities are detectable.
Under these conditions PII and adenylyltransferase co-elute from DEAE-cellulose at a KC1 concentration of approximately 0.2 M. The fact that both adenylyltransferase and PII co-elute is probably not significant since these proteins can be readily separated by gel filtration, or by hydroxyapatite, or hydrophobic chromatography. The fractions containing the UR uridylyltransferase and I',,-adenyltransferase activity peaks were pooled and concentrated by addition of (NH&SOI to 50% saturation as described in Step 2. The NH,SO, precipitates were taken up in a minimal volume of standard buffer or standard buffer plus 100 mM KC1 for UR uridylyltransferase factor, and stored at -80". Step 4: Agarose Chromatography-Concentrated fractions (30 ml) containing PII and adenylyltransferase were applied to an agarose column (5 x 120 cm) equilibrated with standard buffer. The column was then washed with standard buffer (100 cm of hydrostatic pressure), and 20.ml fractions wTere collected. After a void volume of approximately 600 ml had washed through the column, adenylyltransferase emerged along with the bulk of the protein. Prr eluted just before the salt fraction. Only the highest specific activity tubes were pooled. A 30.fold purification of Prr could be obtained at this step.
Step 5 were pooled and dialyzed two times versus 100 volumes of standard buffer.
Step 6: Hydrophobic Chromatography-Hydrophobic chromatography on Sepharose-(CHJ &Hz as described by Shaltiel (12) was used as a final purification step. It had previously been determined that (CHJ5-NH2 or (CHJ6-NHz linked to Sepharose would completely retain PI1 and that 80 to 90% of the activity could be recovered by elution with KC1.2 Shorter hydrocarbon chain lengths did not retain all of the activity, whereas longer hydrocarbon arms (> CH2)6 retained all of the activity but elution with simple salt solutions was not efficient. Thus, a column (1.5 x 8.5) of Sepharose-(CHB)6-NHz was prepared (see "Experimental Procedures") and equilibrated with standard buffer. The dialyzed material from hydroxyapatite (10 to 100 ml, 0.1 mg/ml) purification was stable for at least 6 months when stored at -80" in standard buffer. The same procedure was used to purify Prr from Pseudomonas putida except that Step 6 was not necessary to obtain a homogeneous preparation3 The Prr preparations from E. coli and P. putida were found to have identical physical and catalytical properties.

Molecular Weights of PIrA and PIID
The form of Pn isolated as described above was the unmodified or PrrA form. It migrated as a single prot.ein band when subjected to electrophoresis in 7.5% acrylamide gels ( Fig. 2A) or in gels containing 0.1% sodium dodecyl sulfate. Conversion of the PIIA to the fully uridylylated form by incubation in the presence of [aH]UTP and uridylyltransferase followed by repurification over hydroxyapatite changed its electrophoretic mobility in standard gels (Fig. 2C). A mixture of the two forms of Prr, the unmodified form (PII*) and the fully uridylylated form (PIID) was completely separated electrophoretically (Fig. ZB). To determine whether this difference in electrophoretic mobilities between the two forms was due primarily to size or charge differences, the molecular weights of both species were determined by electrophoresis in gels of varying porosities (Fig. 3). Ribonuclease, ovalbumin (monomer and dimer), and bovine serum albumin (monomer, dimer, and trimer) were used as reference compounds. Both forms of the Prr protein have identical molecular weights, 44,000 when determined by this method. The molecular weight of the native protein was 41,700 f 2,300 when determined by equilibrium sedimentation in the Spinco model E ultracentrifuge. Thus, the difference in electrophoretic mobility is due to a charge difference attributable to the phosphate of UMP linked to PIID. This is further indicated by the fact that PIID migrates faster than PrrA (Fig. 2), and that all of the [3H]UMP was associated with the PIID band when the gels were sliced and counted for radioactivity.

Subunit Molecular Weight
The subunit molecular weight of PI1 as determined electrophoretically in sodium dodecyl sulfate acrylamide gels is 11,000. A value of 11,600 was calculated from electrophoretic mobilities in the presence of 8 M urea in gels of varying porosity. From sedimentation equilibrium data (see "Experimental Procedures") in the presence of 6 M guanidine HCl the molecular weight was estimated to be 13,400 + 2,000. A minimal molecular weight of 11,350 was calculated from the amino acid content, Table II. Based upon these data the protein appears to be composed of four identical subunits. There are 2 tyrosyl residues (Table II) per minimal molecular weight species and if these 2 residues are separated by a trypsinsusceptible bond and the four subunits are identical, then tryptic digestion should yield equimolar amounts of two tyrosine-containing peptides. To facilitate detection of such peptides a PrrA preparation was iodinated with lzsI in the presence of 0.1% sodium dodecyl sulfate and chloramine-T prior to tryptic digestion as described by Bray and Brownlee (30). As shown in the upper part of Fig. 4 only two radioactive peptides were formed. When the radioactive spots containing these peptides were cut from the paper and counted in a liquid scintillation spectrometer, one contained 124,100 cpm and the other 123,590 cpm. The fact that two tryptic peptides containing tyrosine are present in equal molar amounts is further evidence that the subunits are identical.

Site of UMP Linkage
When fully uridylylated Prr (PIID) was iodinated, digested with trypsin, and t'he radioactive peptides separated by the identical procedure used for PIIA, only one of the two radioactive peptides produced from PIIA was detected (lower part of Fig. 4). This suggested that the uridylyl group was covalently attached to the hydroxylyl oxygen of one of the tyrosine residues since a free hydroxyl group is essential for the iodination of tyrosine under these conditions.
In control experiments free tyrosine was shown to be readily iodinated by the chloramine-T method, whereas 0-phosphotyrosine was not iodinated. Thus, the fact that only one of the two tyrosyl groups of a uridylylated PI1 subunit can be iodinated indicates that the UMP is covalently attached to the hydroxyl group of the other tyrosyl residue in each of the sub- FIG. 4. Audioradiograph of a two-dimensional separation of rz51-labeled tryptic peptides of PIIA and PIID were subjected to electrophoresis on acrylamide gels as described in the text. Stained protein bands were cut from the gels and the protein was eluted, destained, iodinated with I251 in the presence of chloramine-T and 0.1% sodium dodecyl sulfate, and digested with trypsin as described by Bray and Brownlee (30). The pept.ide digest was first subjected to electrophoresis on Whatman No. 3MM paper (33 X 32 cm) at pH 1.8 in 47, formic acid (3009 volts for 3 hours at 25") followed by descending chromatography in l-butanol/glacial acetic/H*0 (4/l/5). The paper was dried and exposed to standard x-ray film for 24 to 72 hours.

units. Iodination
and tryptic digestion of partially uridylylated Prr yields varying amounts of the iodinated tyrosyl peptide that is not detected in digests of fully uridylylated Prr (data not shown).
With the exception of the amino acid analyses, which show minor differences in composition between the E. coli Prr and the P. putida Prr, the PI1 preparations from the two microorganisms are otherwise indistinguishable by the methods described in the present study. Hydrolysis of the P. putida Prr yielded two tyrosyl peptides with map positions identical with those obtained from the E. coli protein.
Independent evidence supporting the conclusion that UMP is linked to Prr through the hydroxyl group of tyrosine is shown in Fig. 5. When a difference spectrum at pH 13 was taken between PIID (in the reference cuvette) and PIID treated with snake venom phosphodiesterase to remove the covalently bound UMP (in the sample cuvette), a 290-nm absorption peak characteristic of an ionized tyrosyl hydroxyl group was observed. In a comparable experiment, incubation of [3H]UMP-uridylylated Prr (112,000 cpm/nmol) with phosphodiesterase resulted in the release of 5.1 PM [3H]UMP and the concomitant appearance of 5.5~~ ionizable tyrosyl groups (calculated from a molar extinction coefficient of 2,330, pH 13). Thus, there is an approximately 1: 1 molar relationship between the amount of UMP cleaved by phosphodiesterase and ionizable tyrosyl groups exposed.
A summary of the physical properties of Prr described above is shown in Table III been determined.
As illustrated in Fig. 1, the uridylyltransferase and uridylyl-removing enzyme activities co-elute from DEAEcellulose.
Both the UR and uridylyltransferase activities copurify through a variety of subsequent procedures including: gel filtration, hydroxyapatite chromatography, isoelectric focusing, sucrose density gradient centrifugation, hydrophobic chromatography and polyacrylamide gel electrophoresis.4 An example of their co-purification is illustrated in Fig. 6. When the DEAEcellulose fractions containing UR and uridylyltransferase activities from the first passage on Sepharose- (CHJ5-?;H2 were pooled, concentrated, and reapplied to the same column, elution with a shallower KC1 gradient again failed to resolve the two activities and a second minor peak of both activities appeared (Fig. 6B). Whereas the UR uridylyltransferase activities have been extensively purified, a homogeneous preparation has not been obtained.
Both the UR and uridylyltransferase activities are heat labile in the absence of high ionic strength buffers. Incubation at 37" for 30 min in low ionic strength buffers destroys both activities but addition of KC1 (100 mM) completely protects against this inactivation.
Routinely, 100 mM KC1 is used in all buffers and assay mixtures, although other salts except phosphate (which inhibits both activities) of equal ionic strength can replace KCl. These include NaCl, NatSO+ K804, 1\IgC12, CaCL, and ;Ilg,SOa. Under the usual assay conditions ("Experimental Procedures") uridylyltransferase had a pH optimum of 7.6, whereas UR activity was greatest between pH 8.6 and 9.0.
The effects of divalent cations and other effecters on the UR and uridylyltransferase activities are shown in Table IV. Mn2+ by itself can support maximal UR activity, whereas Mg2+ cannot support this activity except in the presence of cr-ketoglutarate and ATP; even then the activity with NIg2+ is less than with MI?+. In contrast, MI?+ and Mg2+ support equal uridylyltransferase activity in the presence of ATP and a-ketoglutarate but neither divalent cation is able to support activity in the absence of these effecters. it seemed likely that UMP was the primary product since a rapid hydrolysis of UMP to uridine was catalyzed by the UR enzyme preparation.
Using a more highly purified UR enzyme prepared as described in Fig. 7, UMP was identified as the primary cleavage product either in the presence of Mn2+ or with Mg2+, ATP, and cu-ketoglutarate ( Table V). Addition of UMP (Line 2 of Table V) had no effect upon this reaction.   (20) 3543 (74) 240 (6)  1206 (23) 3703 (73) 183 (4)  639 (12) 4402 (84) 211 (4)  7407 (97) 109 (1) 149 (2) a These reaction mixtures also contain 2 mM KzMg EDTA to chelate traces of Mn2+ that were present in the enzyme preparation. Fig. 8 summarizes current knowledge of the complex system that regulates glutamine synthetase activity in E. coli. It consists essentially of two enzyme-catalyzed interconversion systems that are interconnected by the Prr regulatory protein. The uridylylation system involves covalent attachment and detachment of UMP to and from the regulatory protein; this in turn regulates the adenylylation and deadenylylation of glutamine synthetase and thereby determines its catalytic potential. Viewed in a more conventional manner, the covalent modification systems constitute two opposing cascade systems that lead either to the activation or inactivation of glutamine synthet.ase. As shown in Fig. 9B, inactivation of glutamine synthetase activity is initiated by the activation of the UR enzyme which catalyzes the conversion (diuridylylation) of PIID to PIIA. The latter, presumably by direct action, stimulates the capacity of adenylyltransferase to catalyze the adenylylation of glutamine synthetase, thus converting it from a Mg2+-dependent form with a pH optimum of 8.0 to the less active Mn2f-dependent form having a pH optimum of 6.9. A similar cascade leading to the activation of glutamine synthetase is depicted in Fig. 9A. This cascade is initiated by the action of uridylyltransferase which catalyzes the uridylylation of PIIA, converting it to the modified form, PIID, whose interaction with adenylyltransferase stimulates the deadenylylation of glutamine synthetase, which is then converted back to the more active Mg2+-dependent form. As shown in both Figs. 8 and 9, the activities of the uridylplation and adenylylation systems and thus glutamine synthetase activity are finely modulated by the concentrations of various metabolites including UTP, ATP, a-ketoglutarate, Pi, glutamine, and probably other compounds as yet unidentified.
The Prr regulatory protein has a molecular weight of approximately 44,000 and has four identical subunits, each of which contains a specific tyrosyl residue that is covalently modified by the attachment of UMP in a phosphoryl-O-tyrosyl linkage. The role 9. Cascades involved in the regulation of glutamine synthetase activity: A, activation (deadenylylation) of glutamine synthetase; B, inactivation (adenylylation) of glutamine synthetase. EP refers to end products of glutamine metabolism.
of this uridylylation in the conversion of PIrD to PIIA has been clearly demonstrated.
The fully uridylylated protein (Prrn) can be separated from the unmodified form (I'& by polyacrylamide gel electrophoresis on the basis of charge, presumably due to the phosphoryl groups of PIID. Since each subunit can be uridylylated, there are at least five species of PII that differ in the number (0 to 4) of moles of UMP bound per mole of protein. Preliminary results indicate that at least four and perhaps five species are resolved by gel electrophoresis.
Since the relative mobilities of these species are altered by treatment with either snake venom phosphodiesterase (which deuridylylates 1 'rrn) or with uridylyltransferase, it is probable that they differ from one another by the also by endogenous inhibitors of deadenylylation activity known to be present (31). In these respects it should be noted that PIID activity in crude extracts of P. putida is at least 40-fold higher than in E. coli grown under identical conditions and homogeneous Prr preparations are obtained after only 120-fold purification. Since the P. putida PI1 appears to have the same regulatory and physical properties as the E. coli protein, this may be a more suitable source of Prr for further detailed studies.