JOURNAL OF BACTERIOLOGY, Oct. 1967, p. 949-957 Vol. 94, No. 4Copyright © 1967 American Society for Microbiology Printed in U.S.A.

Regulation of Glutamine SynthetaseX. Effect of Growth Conditions on the Susceptibility of Escherichia coli

Glutamine Synthetase to Feedback Inhibition

HENRY S. KINGDON' AND E. R. STADTMAN

Laboratory of Biochemistry, Section on Enzymes, National Heart Institute, Nationial Institutes of Health,Bethesda, Maryland 20014

Received for publication 20 July 1967

The kinetic properties of Escherichia coli glutamine synthetase are markedlyinfluenced by the manner in which the organism is grown. Enzyme obtained fromstationary-phase cells grown on glycerol and glutamate is strongely inhibited byeach of the eight feedback effectors known to influence this enzyme; however, theenzyme from log-phase cells grown on glucose and growth-limiting concentrationsof NH4C1 is stimulated by some of these effectors. Of the growth variables examined,nitrogen source and time of harvest were the most important; carbon source andaeration seemed to have no effect. Two purified enzyme preparations have been ob-tained from cells grown under two different conditions, designated enzymes I andII for convenience. Enzyme I is stimulated by adenosine 5'-monophosphate, histi-dine, and tryptophan in the transfer assay, whereas enzyme II is strongly inhibitedby all effectors tested. Enzyme I has a higher specific activity in the forward assay inthe presence of Mg++ or Co++, whereas enzyme II is more active in the presenceof Mn §.

Cumulative feedback inhibition of Eschcerichiacoli glutamine synthetase by eight potential endproducts of glutamine metabolism has been de-scribed (8). In the course of more detailed kineticstudies, it was found that a particular crystallinehomogeneous enzyme preparation (7) was moresusceptible to most of the inhibitors than was thepartially purified enzyme orginally described (8).Efforts to reconcile the difference in behavior ofthe two enzyme preparations led to the discoverythat E. coli produces two forms of glutaminesynthetase (synthetases I and II) that exhibitmarkedly different kinetic characteristics. Asshown in this paper, the glutamyl transferaseactivity of synthetase II is strongly inhibited byhigh concentrations of any one of the eight endproducts of glutamine metabolism, whereas thetransferase activity of synthetase I is either un-affected or is actually stimulated by high concen-trations of some of these compounds. Data arepresented showing that the relative concentrationof the two enzymes is markedly influenced by theconditions of growth. Finally, a procedure for theisolation of apparently homogeneous prepara-tions of the two enzyme forms is described and

1 Present address: Department of Medicine, Sectionof Hematology, University of Chicago, Chicago, Ill.60637.

kinetic properties of each are reported. A prelim-inary report of this work has been published (3).

MATERIALS AND METHODS

Chemicals. Adenosine 5'-triphosphate, disodium(ATP), and L-glutamic acid -y-monohydroxamatewere obtained from Sigma Chemical Co., St. Louis,Mo. L-Glutamine (commercial) and monosodiumL-glutamate were obtained from Nutritional Bio-chemicals Corp., Cleveland, Ohio. Adenosine 5'-diphosphate, sodium salt (ADP), and adenosine5'-monophosphate (AMP) were obtained from PabstLaboratories, Milwaukee, Wis. Carbamyl phosphate,dilithium salt, B grade, and cytidine 5'-triphosphate,tripotassium, tetrahydrate (CTP) were obtained fromCalbiochem, Los Angeles, Calif. All other chemicalswere reagent grade.

Organism. E. coli strain W was originally obtainedfrom G. N. Cohen.

Growth coniditionis. In early experiments, growthconditions were varied, as explained in Results.Except for variations in carbon and nitrogen sources,all media were identical to the one given below forthe production of synthetase II. Except where noted,all preparations were grown in air at 37 C withvigorous shaking. Growth was followed by increasein optical density as measured with a Klett colorimeter(no. 66 filter).

Preparation of crude extracts. Cells were suspendedin 10 mM imidazole buffer, pH 7.1, containing 10 mMMnCl2. Sonic disruption was carried out for 2 min in

949

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

KINGDON AND STADTMAN

30-sec bursts with a Sonifier, model S75 (BransonInstruments Corp., Stamford, Conn.). Debris wasremoved by centrifugation at 12,100 X g for 10 min.

Growth conditions for the production of synthetaseII. Cells were grown aerobically at 37 C in a 300-literfermentor containing 21 mm glycerol, 19 mm gluta-mate, 1.7 mM MgSO4, 14.3 mr K2SO4, 43 mi NaCl,and 0.1 M mixed potassium phosphates to give a finalpH of 7.1. The bulk of the volume of all media wastap water. The fermentor was inoculated with 30 litersof cells grown in carboys on the same medium at 37 Cfor 24 hr. The cell growth was allowed to proceedinto the stationary phase, with harvest 20 to 24 hrafter inoculation. The conditions were the same asthose which had been used to obtain the crystallineenzyme previously described (7), and which had beenchosen for maximal derepression. The usual yield ofcells was 1.5 to 1.7 kg (wet weight) per 300-literfermentor.

Growth conditions for the productionz of syntthetase I.Cells were grown aerobically at 37 C in a 300-literfermentor on a medium differing from the above onlyin carbon and nitrogen sources; the other componentswere identical. The carbon source was 11 mm glucose,and the nitrogen source was 4 mm NH4Cl. Thefermentor was inoculated with 15 liters of cells grownin a carboy on the same medium at 37 C for 8 to 10 hr,and the optical density of the medium was followed byusing the no. 66 filter in the Klett colorimeter. Thefermentor was quickly cooled and the cells wereharvested as soon as the Klett reading reached 140,which was the value at the end of the log phase. Har-vesting usually began 4 to 5 hr after inoculation. Theseconditions were achieved in experiments describedunder Results. The usual yield of cells was 0.40 to0.45 kg (wet weight) per 300-liter fermentor.

Protein. Concentration in crude extracts was de-termined by using the Biuret reagent of Gornall et al.(2), with protein samples which had been freed ofMn++ by precipitation in 5% cold trichloroacetic acidand resuspension in water. Protein concentration ofpurified proteins was determined from the opticaldensity at 280 my, by using the extinction coefficientpreviously determined for pure glutamine synthetase(synthetase II) (7).

Glutamine synthetase activity. Two assay methodswere used. The first was a modification of the transferassay described by Levintow (4). The major modifica-tion was reduction of the concentrations of glutamineand hydroxylamine to permit detection of inhibitorscompetitive with ammonia and glutamate (Shapiroand Stadtman, unipublished data). The assays werecarried out at 37 C for 15 min in 1.0-ml volumes con-taining 20 mM imidazole buffer, pH 7.0, 20 mm hy-droxylamine, 3 mM MnCl2, 0.4 mm ADP, 20 mMglutamine, and 20 mm arsenate. The amount ofenzyme was adjusted to yield up to 1.5 ,umoles ofy-glutamyl hydroxamate per tube. The reactions werestopped by adding 2.0 ml of a solution containing0.5 M HCl, 2% trichloroacetic acid, and 0.83%c FeCI3.The intensity of color was read in the Klett colorimeterwith a no. 54 filter, and compared to a standard curveemploying genuine y-glutamyl hydroxamate. Specific

activities were expressed as micromoles of hydroxa-mate formed per minute per milligram of enzyme.The second assay of glutamine synthetase was the

forward assay which measures the release of phos-phate from ATP in the presence of glutamate andammonia (7). Imidazole buffer was reduced from 50to 10 mm, and 10 mm tris(hydroxymethyl)amino-methane (Tris), 5 mm acetate, and 5 mm borate wereincluded to permit examination of a broad pH range.Assays were run for 10 mmn at 37 C. Specific activitieswere expressed as micromoles of phosphate producedper minute per milligram of enzyme.

Disc-gel electrophoresis. Disc-gel electrophoresiswas performed according to Davis (1), using thestandard 7.5% gel and Tris-glycine buffer.

RESULTS

Effect of growth time, nitrogen source, and car-bon source on susceptibility ofglutamine synihetaseto effectors. A 25-ml inoculum of E. coli strainW was grown overnight (or in the case of nitritefor several days) on the five pairs of carbon andnitrogen sources listed in Table 1. In all cases theconcentrations of ingredients other than carbonand nitrogen sources were identical to those de-scribed for large-scale preparations under Mate-rials and Methods. For each pair of carbon andnitrogen sources three flasks were prepared forinoculation; two contained 250 ml medium in a2-liter flask; one contained 50 ml of medium in a300-ml flask with a test tube side arm. The 2-literflasks were inoculated with 10 ml of the appro-priate inoculum, and the 300-ml flasks with 2 mlof inoculum. In these experiments, nitrogensource determined the growth rate. The timesrequired to double optical density were: onNH4CI, 50 min; on glutamate, 140 min; and onnitrite, 10 hr.

In each case, one of the 2-liter flasks was har-vested at or near the end of the log phase, andthe other flask after considerable time in thestationary phase. Glutamine synthetase wasassayed in the crude extracts by the transfer assayas described in Materials and Methods, in thepresence and absence of effectors. The results arepresented in Table 1. Numbers over 100 representstimulation by the effector, and numbers less than100, inhibition.

In each case, in this and in many similar experi-ments, there was a difference between log-phaseand stationary-phase extracts with respect to sen-sitivity to every effector tested. With the exceptionof glycine and L-alanine, the change in going fromlog-phase to stationary-phase cells is in the direc-tion of more inhibition (the change in CTP inhi-bition in column A has not been confirmed).Glycine and L-alanine consistently show less inhi-bition of stationary-phase extracts than of log-phase extracts, except in cells grown on NH4Cl,

950 J. BACTERIOL.

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INHIBITION OF GLUTAMINE SYNTHETASE

where there was little change. The magnitude ofthe change going from log phase to stationaryphase seems to be unaffected by the carbonsource. In contrast, the magnitude of the changeis significantly less in the NH4Cl-grown cells whencompared to the other two nitrogen sources, inspite of the fact that the NH4Cl cells stayed in thestationary phase the longest period of time.

Effect of concentration of NH4Cl on suscepti-bility of glutamine synthetase to effectors. Anexperiment similar to the preceding one wasperformed to test the effect of NH4Cl concen-tration on the type of glutamine synthetase pro-duced. In this experiment only one 250-ml flaskwas run at each concentration, and all werestarted from the same inoculum, which had been

grown overnight on 6 mm NH4Cl. All flasks wereharvested 5 hr after inoculation, very close to theend of log phase in all cases. Except for the flaskwith 4 mm NH4Cl, all flasks had reached 140 to150 Klett units at harvest. Extracts were preparedand transfer assays were performed as described inMaterials and Methods. The results are presentedin Table 2.

It can be seen that even over this narrow range,the concentration ofNH4CI in the growth mediumhas a marked effect on the responsiveness of theglutamine synthetase to various effectors. At thelower concentrations of NH4Cl, enzyme is pro-duced which is stimulated by AMP, histidine, andtryptophan. At the higher concentrations ofNH4Cl, enzyme is produced which does not show

TABLE 1. Effect ofgrowth time, nitrogen source, and carbon source onsusceptibility ofglutamine synthetase to effectors'

Growth corditons

(A) Glucose- (B) Glycerol- (C) Glycerol- (D) Glucose- (E) Glucose-Effector ammonium ammonium glutamate glutamate n itrite

Log ogationao- Log Station-Log Station-ph tion-pas ary Logs ary phase ary' phase aryphase |phase ps phase phase pnase phase

AMP, 40 mM ........... 131 115 141 122 124 20 131 20 118 19Glycine, 40 mM ......... 36 34 35 34 33 48 32 47 31 41L-Alanine, 40 mM ....... 10 9 8 8 7 15 7 15 7 11Tryptophan, 20 mM..... 101 92 108 97 97 43 102 50 103 51Histidine, 40 mM.115 107 121 110 108 52 113 53 114 47CTP,8 mM ............. 83 97 87 80 99 31 105 51 -

D-Alanine, 40 mm....... 60 54 61 54 53 22 55 21D-Norleucine, 20 mM.... 95 93 103 97 97 53 97 53

The results are expressed as relative specific activities, compared to duplicate controls withouteffector arbitrarily set at 100. Glucose concentrations were 11 mm; glycerol, 21 mM; NH4Cl, 6 mM; gluta-mate, 19 mM; and NaNO2, 14 mm. Times of harvest after inoculation were: (A) log phase, 5.5; (A) sta-tionary phase, 24 hr; (B) log phase, 6.25 hr; (B) stationary phase, 24 hr; (C) log phase, 10.25 hr; (C) sta-tionary phase, 24 hr; (D) log phase, 11 hr; (D) stationary phase, 24 hr; (E) log phase, 40.25 hr; (E)stationary phase, 71.5 hr.

TABLE 2. Effect of concentration of ammonium chloride on susceptibility ofglutamine synthetase to effectorsa

EffectorConcn of NH4CI in Speciic.ativi y.

culture media wSpecfic activity(mm) AMP (40 mm) Glycine L-Alanine Tryptophan Histidline withnut effectorb

(40 mm) (40 mM) (20 mm) (40 mM)

4 141 37 8 108 123 0.145 137 36 8 107 120 0.116 138 36 7 108 120 0.0848 113 39 12 94 104 0.045

10 107 40 12 90 98 0.04512 100 41 12 84 96 0.044

a Carbon source was 11 mM glucose. Results are expressed as in Table 1.b Micromoles of hydroxamate formed per minute per milligram of enzyme.

951VOL. 94, 1967

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

KINGDON AND STADTMAN

these stimulations, is inhibited by histidine andtryptophan, and is less strongly inhibited byglycine and L-alanine.

Specific activities of the glutamine synthetaseproduced at varied NH4Cl levels are included inTable 2, and confirm earlier results indicatingthat ammonium ion represses synthesis of glu-tamine synthetase (7, 8). In addition, they dem-onstrate that repression and control of effectorsensitivity are not tightly linked. In this case,maximal derepression at low ammonium ionconcentration is accompanied by production ofenzyme that is stimulated by AMP, histidine, andtryptophan. In the case of growth on glutamate,conditions for maximal derepression lead to pro-duction of enzyme which is strongly inhibited bythese effectors.

Effect of aeration on susceptibility of glutaminesynthetase to effectors. Another experiment wasperformed to test the effects of aeration on thetype of enzyme obtained. Three 250-ml flaskswere inoculated with 10 ml of inoculum grownovernight. One of these was harvested at the endof log phase; the others were harvested the nextmorning. One of the two stationary-phase flaskswas on the shaker for the entire experiment(aerated), and one nonaerated was removed fromthe shaker at the end of the log phase and was al-lowed to stand for an additional 19 hr withoutshaking. Extracts were prepared and transfer as-says performed as described in Materials andMethods. The results are presented in Table 3. It isapparent that the presence or absence of vigorousaeration during the stationary phase did not in-fluence the responsiveness of the enzyme to theeffectors AMP, histidine, and tryptophan. How-ever, by comparing these results with those incolumn A, Table 1, it can be seen that the amountof change during comparable growth times variedfrom experiment to experiment, which indicates

TABLE 3. Effects ofaeration durinig stationary phaseon the susceptibility of glutamine

synthetase to effectorsa

Stationary phase

Effector Log phaseAerated Non-

aerated

AMP, 40mM .......... 135 68 64Glycine, 40 mM. 34 43 45L-Alanine, 40 mm. 7 12 13Tryptophan, 20 mM.... 103 69 70Histidine, 40 mm ...... 119 75 75

a Carbon source: 11 mm glucose; nitrogensource; 6 mM NH4Cl. Times of harvest after inocu-lation were: log phase, 4.5 hr; stationary phase,23.5 hr. Results are expressed as in Table 1.

that there are unknown and uncontrolled vari-ables operating to bring about the changes ob-served.

Purification of large amounts of synthetase Iand II. The form of glutamine synthetase that isstrongly inhibited by high concentrations of eachof the end-product effectors, is referred to assynthetase II. This form of enzyme had alreadybeen prepared in homogeneous, crystalline form(7) from cells grown on glycerol and glutamate asdescribed in Materials and Methods. The formof glutamine synthetase that is resistant to inhibi-tion by tryptophan, histidine, AMP, or CTP isreferred to as synthetase I. This form of the en-zyme was prepared from 2.5 kg of cells grown onglucose and NH4C1 as described in Materialsand Methods. It was found that the synthetase Icould be purified by exactly the same procedure asthat previously described for synthetase II (7).After the third acid-ammonium sulfate step, theenzyme had been purified 220-fold with an overallyield of 55 %, and contained a minor contaminantdisclosed by disc-gel electrophoresis. This con-taminant was removed by repeated reprecipita-tions with acetic acid at pH 4.4, some with 10%(NH4)2SO4, and some with 30% (NH4)2SO4. Therecovery from these reprecipitations was quantita-tive. The final preparation contained 500 mg ofapparently homogeneous protein as judged bydisc-gel electrophoresis. Crystallization was notattempted. A mixture of synthetases I and IIcould not be separated by standard disc-gelelectrophoresis in glycine-Tris buffer in either 7.5or 5.0% gels. Synthetases I and II have the samesedimentation coefficient, and appear to be identi-cal on electron microscopy (B. M. Shapiro, A.Ginsburg, and R. C. Valentine, unpublished data).

Response of enzyme I and II to effectors. Inspite of the marked physical similarities betweenthese purified enzymes, and in spite of the factthat both catalyze the synthesis of glutamine andthe transfer reaction, there are marked kineticdifferences. The differences in response to effectorsin the transfer assay are presented in Table 4.Although the extent of stimulation by AMP andhistidine is somewhat reduced, and there areother smaller differences, purified synthetase Ibehaves toward inhibitors in a manner similar tothat observed in the crude extracts of log-phaseNH4Cl-grown cells (Table 1, column A). Simi-larly, purified synthetase II behaves in a man-ner similar to the activity in the crude extracts ofstationary-phase glutamate-grown cells (Table 1,column C). As with the crude extracts from cellsgrown under similar conditions, the purified syn-thetase II from glutamate-grown cells is slightlyless inhibited by glycine and L-alanine than isthe purified synthetase I from NH4Cl-grown cells,

952 J. BACTERIOL.

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INHIBITION OF GLUTAMINE SYNTHETASE

and enzyme II is much more strongly inhibitedthan is enzyme I by all other effectors tested. Thespecific activity of synthetase I is about 60%o thatof synthetase II in the transfer assay; however,synthetase I has a specific activity three times thatof synthetase II, with regard to the glutaminebiosynthetic function as measured in a coupledspectrophotometric procedure which followsADP production, or in the standard forwardassay (see below).At no time during the purification of synthetase

I was there a marked change in sensitivity towardeffectors indicative of removal of another kind ofglutamine synthetase, or of an enzymaticallyinactive effector molecule.

Differences in metal ion sensitivities. The puri-fied synthetases I and II were examined in theTABLE 4. Differences in response to effectors of

purified glutamine synthetase I and purifiedglutamine synthetase HJa

Effector Synthetase I Svntbetase II

AMP, 40 mM. 110 26Glycine, 40 mM ........ 29 41L-Alanine, 40 mM ...... 4 9Tryptophan, 20mM.... 98 55Histidine, 40 mM ..... 107 50CTP, 8 mm............ 104 61D-Alanine, 40 mm ...... 49 22Norleucine, 20 mm..... 88 52

a Each tube contained 3 ,gg of purified synthetaseI or 1.7 ,g of purified synthetase II. Results areexpressed as in Table 1.

forward assay at a number of pH values, witheach of the four activating divalent cations. Asshown in Fig. 1, the two enzymes differ markedlyin their responsiveness to each of the four cations.Synthetase I is more active in the presence ofMg++ and Co++, whereas synthetase II is moreactive in the presence of Mn++.

Moreover, in the presence of Ca++ the activitypH profile is distinctly different for the two en-zymes; synthetase II has a pH optimum at about7.1, whereas synthetase I exhibits a broad maxi-mum in the range of 8 to 10, without the decreaseat high pH seen with the other divalent cations.The data summarized in the four left-hand

frames of Fig. 1 have been plotted in order toemphasize differences in the responsiveness of thetwo kinds of enzymes to various divalent cations. Itshould therefore be noted that the specific activityscale (i.e., the ordinate) in Fig. 1 for each separatecation is different. Thus, the effectiveness of thevarious cations for both enzymes is ordered inthe sequence Mg++ > Mn++ > Co++ > Ca++.To emphasize the marked difference in the re-sponses of the two enzymes to Mg++ and Mn++,the data for both of these cations are plotted onthe same scale in the right-hand frame of Fig. 1.It can be seen that the specific activity of synthe-tase I is 40 times greater in the presence of Mg++than in the presence of Mn++, whereas the specificactivity of synthetase II is essentially the same inthe presence of either cation. This figure alsoshows, as noted earlier (7), that with Mn++ thepH optimum is about 7.0, whereas with Mg++the pH optimum is about 8.0.

12

0cl9++

-j4,~~~~~~~

8

4

0

Co++

0.6

0.40.2

- S _ rIOIT-6.0 7.0 8 0 9.0 10.0

Co++

ffA00

40

30

20

10

0

6.0 7.0 8.0 9.0 10.0pH

6.0 7.0 8.0 9.0 10.0

FIG. 1. Differences in metal ion sensitivities between synthetases I and II. For four left-hand frames: *, synthe-tase. 1; 0 synthetase 11. For right-handframe: *, synthetase I with Mg++; 0, synthetase I with Mn++; A, syn-thetase II with Mg++; L., synthetase II with Mn++.

60

40

C)C)50~

20

0

8

4

2

0

953VOL. 94, 1967

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

KINGDON AND STADTMAN

Interconversion of the enzyme types in vivo. Anexperiment similar to the one in the first sectionwas performed to test the interconvertibility ofthe two types of enzyme. An inoculum was pre-

pared by growing E. coli W overnight in 250 mlof the glycerol-glutamate medium described forpreparations of synthetase II. The cells wereharvested by centrifugation; three-fourths of thecells (batch A) were frozen for subsequent en-

zyme assays; one-fourth were washed once withsterile 0.85% NaCl, and then resuspended in 60ml of saline for use as an inoculum for the fol-lowing experiment. Samples (10 ml) of the inocu-lum were transferred to three 300-ml side-armflasks containing 50 ml of the glucose-ammoniumchloride medium described for production ofsynthetase I. Growth was followed by increase inoptical density as measured with a Klett color-imeter (no. 66 filter). The growth curve thus ob-tained is shown in Fig. 2.As indicated in Fig. 2, three batches of cells

were harvested as follows: batch B, after 60 minwhen cell mass had not yet doubled; batch C, atthe end of the log phase (105 min) when cell masshad doubled; and batch D, during the stationaryphase, 45 min after the end of the log phase.Extracts of these three batches and of batch A

160

140

120

to

z

I.-I1-

-i

If

100

80

60

0

B C D

I

2 3HOURS

FIG. 2. Growth curve for Escherichia coli W, withheavy glycerol-glutamate grown inoculum in a glucose-NH4CI medium. Points B, C, and D are the points intime at which the three flasks (discussed in the text andin Table 5) were harvested. With the particular Klettcolorimeter and no. 660 filter used, 100 Klett units isequal to 1.11 absorbancy at 660 m,A as measured in a

Zeiss spectrophotometer (1-cm light path).

were then prepared as described in Materials andMethods, and the effector responses of each weredetermined in the glutamyl transferase assay. Theresults are summarized in Table 5. As expected,the enzyme from batch A (i.e., glutamate-growncells) was strongly inhibited by all effectors as istypical of synthetase II (see Table 4). A similareffector response was observed also with extractsof batch B cells. However, as is characteristic ofsynthetase I, the activity in extracts of cell batchesC and D was not inhibited by AMP, tryptophan,or histidine but was slightly more sensitive toglycine and alanine inhibition. It is therefore:obvious that a more or less complete conversionof synthetase HI-type activity to synthetase I-typeactivity occurred during a short interval of time-(ca. 45 min) during the last phases of the doubling.of cell mass which followed transfer of the gluta-mate-grown cells to the NH4C1 media. The factthat an average of one cell division in the NH4C1media was required to effect this change may be-fortuitous, since this occurred at the end of thelog phase and presumably coincides with deple-tion of the limited nitrogen supply. Nevertheless,it is significant from the mechanistic point ofview that the transition from synthetase II- tosynthetase I-type activity was achieved duringthe brief period of time between harvesting batchB and batch C cells, when the total mass ofcellular material was increased by only 40%. Thisfact, and the further observation that the specificactivity of the enzyme in the transferase assaydid not change appreciably during this transitionperiod, support the conclusion that in this experi-ment synthetase I was formed by modification ofpreformed synthetase II. The alternative explana-tion, that there was a rapid degradation of syn-thetase II with concomitant rapid synthesis ofsynthetase I, is unlikely in view of the low turn-over rates normally observed for bacterial pro-teins.

TABLE 5. In vivo conversion of synthetase II tosynthetase I

Effector Batch Batch Batch BatchEffector A B C D

AMP, 40mM ......... 16 21 118 101Glycine, 40 mM ....... 48 43 31 32L-Alanine, 40 mM..... 14 12 7 8Tryptophan, 20 mM. 49 55 105 97Histidine, 40 mM. 51 50 116 103

Results are expressed as in Table 1. Batches Athrough D are batches of cells harvested at thetimes shown in Fig. 2.

954 J. BACTERIOL. .

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

INHIBITION OF GLUTAMINE SYNTHETASE

DISCUSSIONDuring a search for growth conditions giving

maximal derepression of glutamine synthetaseand thus higher yields of enzyme for purification,the methods for growing E. coli in this laboratorywere changed between the report of cumulativefeedback inhibition of the glutamine synthetase(8) and the report of some of the properties ofthe crystalline enzyme (7). An investigation ofdiscrepancies between current work and theearlier work led to the discovery of the two en-zyme forms reported here. The phenomenon, if ageneral one, emphasizes the need to specifyrigorously the growth conditions under whichlarge batches of organisms are obtained for thepurposes of enzyme purification. In addition, it isevident that experiments, in which crude extractsare examined for amount of enzyme to deter-mine the level of derepression, may not revealthe possibility that the enzyme being examinedis changing not only in amount but also in type.The isolation, in apparently homogeneous

form, of two glutamine synthetases with similarphysical properties but different kinetic proper-ties, adds a new dimension to the biological con-trol of this enzyme, which is of such centralimportance in the nitrogen metabolism of E. coli.In contrast to the aspartokinase system (6), theglutamine synthetase seems to be changed withrespect to all effectors, rather than existing inseparate forms each of which is sensitive to oneeffector. The last experiment presented suggestsstrongly that the two forms of enzyme are met-abolically interconvertible. The rapidity withwhich the interconversion took place suggeststhat a mechanism exists for the control of theproportion of the two forms which would pro-vide for rapid feedback control of glutaminemetabolism.

Preliminary experiments (B. M. Shapiro, H. S.Kingdon, and E. R. Stadtman, unpublished data)indicate that synthetases I and H have the sameamino acid compolition, and that synthetase IIcontains covalently bound AMP which synthetaseI lacks. If this chemical difference is the basis forthe kinetic differences observed, it seems likelythat an enzyme exists which introduces nucleo-tide into the glutamine synthetase, and thatanother enzyme exists which removes it. It wouldthen be important to identify these enzymes anddetermine what factors in the growth of theorganisms determine their levels. Such a system ofcontrol would be analogous to the system whichcontrols the level of active phosphorylase.The fact that synthetases I and II are purified

in exactly the same manner, and that they are

inseparable by techniques thus far employed,suggests that the present preparations might becontaminated one with the other. This is furthersuggested by several findings presented in thispaper. Crude extracts of log-phase glucose-ammo-nium grown cells routinely showed 30 to 40%stimulation by 40 mm AMP (see Table 1, columnA). In contrast, the purified synthetase I showedonly 10% stimulation by 40 mm AMP (see Table4), suggesting that a small amount of conversionof synthetase I to synthetase II might have takenplace during the early stages of purification. Simi-larly, the purified synthetase II is less sensitiveto inhibition by AMP (Table 4) than are crudeextracts of stationary-phase glycerol-glutamatecells (Table 1, column C), suggesting that syn-thetase II is contaminated in some degree withsynthetase I.

It is evident from the data in Fig. 1 that theconversion of synthetase I to synthetase II isassociated with a large decrease in specific ac-tivity as measured in the forward assay in thepresence of Mg++. The conversion of synthetaseI to synthetase II is therefore reminiscent of theglutamine synthetase inactivation system de-scribed by Mecke et al. (5). These workers havedemonstrated the presence of an enzyme inE. coli B that catalyzes an ATP, glutamine-dependent inactivation of glutamine synthetase,as measured in the biosynthetic assay, in thepresence of Mg++. They report that this inactiva-tion proceeds almost to completion (90%).However, the lower specific activity of theirinactivated enzyme (glutamine synthetase b)compared to that of our synthetase II could beexplained by slight contamination of the latterwith synthetase I, as was suggested above. Ittherefore seems possible that the inactivatingenzyme of Mecke et al. is identical with thatpostulated here for the conversion of synthetaseI to synthetase II. In this case, synthetases I andII would be identical with the glutamine synthe-tases a and b, respectively, of Mecke et al., andthe apparent inactivation reaction would involvethe covalent binding of AMP to synthetase I(i.e., to glutamine synthetase a). Moreover, theglutamine synthetase b of Mecke et al. shouldbe catalytically active in the bioynthetsic assaywhen Mn++ rather than Mg++ is supplied asthe divalent cation. This follows from the datain Fig. 1, which show that the conversion ofsynthetase I to synthetase II is associated with aloss in biosynthetic activity with Mg++ as theactivating cation, but is associated with an in-crease in catalytic activity when assayed inthe presence of Mn i. Thus, the so-called in-

VOL. 94, 1967 955

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

KINGDON AND STADTMAN

activation reaction of Mecke et al. would in-volve a change in specificity toward divalent cat-ion rather than absolute inactivation; i.e., pureglutamine synthetase a (synthetase I) would bespecific for Mg++ and pure glutamine synthetaseb (synthetase II) would be specific for MnThese considerations are under investigation.

In a preliminary account of this work (3), thepurified synthetases I and II were compared withrespect to their sensitivity to effectors in thetransfer assay, and the striking differences incrude extracts reported here were evident. Earlyattempts to perform similar experiments in theforward assay have been complicated by the dif-ferences in sensitivity to divalent cations reportedhere. Preliminary experiments indicate thatin the case of AMP the differences in sensitivityto effectors are not as great in the forward assayas in the transfer assay, but the magnitude of theresponses is dependent upon which divalent cat-ion (Mg++ or Mn++) is present. In view of thepossibility that pure synthetases I and II mayhave nearly absolute divalent cation specificitiesin the biosynthetic reaction, and that, as isolated,each synthetase is slightly contaminated with theother, it is not possible to evaluate the variationsin responses of the two preparations to differentfeedback inhibitors in the biosynthetic assay. Suchstudies must await the availability of preparationsof synthetases I and II of established purity.The differences in sensitivity to Mg++ and

Mn+ reported here for glutamine synthetasesI and II may be of significance in the regulationof the enzyme in vivo. First, variations in theratio of the two forms of enzyme could lead tovariations in the net catalytic activity present,assuming a fixed amount of Mg++ and Mn++ inthe cell. Second, the net catalytic activity seenwith any given ratio of the two synthetases couldbe controlled by the amounts of Mg++ and Mn++available. In the third place, the presence of aform of enzyme relatively specific for Mn raisesthe possibility of finer control of the glutaminesynthetase by nucleotides, in view of the recentfinding that the activity of the enzyme in thepresence of Mn++ is at a maximum only whenthe total trinucleotide concentration is equal tothe Mn++ concentration (Hubbard and Stadt-man., unpublished data).The evaluation of the above possibilities must

await methods for evaluation of the relativeamounts of synthetases I and II in mixtures andin crude extracts. It is possible that intermediatebehavior toward effectors (e.g., Table 2, right-hand column) represents a mixture of synthetasesI and II. It is equally possible, in view of thefact that the enzyme is made up of 12 identical

subunits (7) and that synthetase II containscovalently bound AMP, that intermediate be-havior with respect to metals or effectors repre-sents partial adenylation of the enzyme, e.g., 4 ofthe 12 subunits containing covalently boundAMP. This would lead to a number of possibleenzyme molecules which could be viewed ashybrids of adenylated and unadenylated sub-units; each of these enzymes might exhibit differ-ent allosteric behavior. Such a situation couldarise merely by controlling the extent of adenyla-tion, and need not involve dissociation and re-association of subunits.

Discrepancies remain between the current workand the original reports from this laboratory(7, 8). For example, Woolfolk et al. (7) in Fig. 4of their paper presented pH curves in the for-ward assay using the same buffer system usedhere, and three of the four metals used here. Thecurves have similar shapes, and they have thesame pH optima for Mg++, Mn++, and Co++.However, the specific activities reported by Wool-folk et al. at the pH optima were: in the presenceof Mg++, 125; in the presence of Mn++, 28; inthe presence of Co++, 27. All of these values arehigher than the values obtained with eitherpreparation reported here, and could thus not beexplained as resulting from a mixture of the twoforms reported here. Similarly, the lower levelsof inhibition by glycine and L-alanine reportedearlier (8, 9) cannot be explained by a mixtureof the two forms reported here. The differences inbehavior between different preparations of en-zyme may be caused by differences in in vitromanipulations which may affect the sensitivity ofthe enzyme to effectors (9).The present report shows, however, that the

conditions under which E. coli is grown have aprofound effect on the kinetic properties of theglutamine synthetase produced. Differences be-tween earlier enzyme preparations and thoseunder current study may be related to variationsin growth conditions. In this regard, it may besignificant that the earlier enzyme preparationswere obtained from cells grown under continuousculture in a Biogen under conditions of extremenitrogen starvation, whereas all later prepara-tions were obtained from cells grown in a batch-type fermentor. If these differences in nitrogennutrition are responsible for the differences be-tween the current work and the original reports,clarification must await better understanding ofthe factors controlling the modification of theglutamine synthetase molecule.

ACKNOWLEDGMENTSWe express our appreciation to Henry Lutterlough

and J. Maurice Miles for invaluable assistance in the

956 J. BACTERIOL.

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

VOL. 94, 1967 INHIBITION OF GLUTA

large-scale growth of E. coli, and in the early steps ofthe large-scale enzyme preparations. We also thankBennett Shapiro for generously providing the crystal-line synthetase II used in the experiments comparingthe purified enzymes.

LITERATURE CITED

1. DAVIS, B. J. 1964. Disc electrophoresis-IT. Methodand application to human serum proteins. Ann.N.Y. Acad. Sci. 121:404-427.

2. GORNALL, A. G., C. J. BARDAWILL, AND M. M.DAVID. 1949. Determination of serum proteinsby means of the biuret reaction. J. Biol. Chem.177:751-766.

3. KINGDON, H. S., AND E. R. STADTMAN, 1967. TwoE. coli glutamine synthetases with differentsensitivities to feedback effectors. Biochem.Biophys. Res. Commun. 27:470-473.

4. LEVINrOW, L. 1954. The glutamyltransferaseactivity of normal and neoplastic tissues. J.Natl. Cancer Inst. 15:347-352.

5. MECKE, D., K. WULFF, AND H. HOLZER. 1966.

kMINE SYNTHETASE 957

Metabolit-induzierte Inaktivierung von Gluta-minsynthetase aus Escherichia coli im zellfreienSystem. Biochim. Biophys. Acta 128:559-567.

6. STADTMAN, E. R., G. N. COHEN, G. LEBRAS, ANDH. DE ROBICHON-SZULMAJSTER. 1961. Feed-backinhibition and repression of aspartokinase activ-ity in Escherichia coli and Saccharomyces cere-visiae. J. Biol. Chem. 236:2033-2038.

7. WOOLFOLK, C. A., B. SHAPIRO, AND E. R. STADT-MAN. 1966. Regulation of glutamine synthetase.I. Purification and properties of glutaminesynthetase from Escherichia coli. Arch. Biochem.Biophys. 116:177-192.

8. WOOLFOLK, C. A., AND E. R. STADTMAN. 1964.Cumulative feedback inhibition in the multipleend product regulation of glutamine synthetaseactivity in Escherichia coli. Biochem. Biophys.Res. Commun. 17:313-319.

9. WOOLFOLK, C. A., AND E. R. STADTMAN. 1967.Regulation of glutamine synthetase. III. Cumula-tive feedback inhibition of glutamine synthetasefrom Escherichia coli. Arch. Biochem. Biophys.118:736-755.

on June 8, 2020 by guesthttp://jb.asm

.org/D

ownloaded from