JOURNAL OF BACTERIOLOGY, Mar., 1967, p. 10451055 Vol. 93, No. 3Copyright © 1967 American Society for Microbiology Printed n U.S.A.

Regulation of Glutamine SynthetaseII. Patterns of Feedback Inhibition in Microorganisms'

JERRY S. HUBBARD2 AND E. R. STADTMANLaboratory ofBiochemistry, National Heart Institute, National Institutes ofHealth, Bethesda, Maryland

Received for publication 19 November 1966

The feedback inhibition of glutamine synthetase was investigated -by use ofpartially purified enzyme preparations from Salmonella typhimuriwn, Micrococcussodonensis, Pseudomonas fluorescens, Bacillus cereus, Bacillus licheniformis, Clostri-dium pasteurianum, Rhodospirillum rubrum, Neurospora crassa, Cauddautilis, andChlorella pyrenoidosa. Inhibition analyses indicated that the enzyme of -eachorganism can be effectively regulated with mixtures of end products from the diversepathways ofglune metabolism. When tested individually, tryptophan, histidine,alanine, glycine, glutamine, '-adenylate (AMP), cytidine-5'-triphosphate, carbamylphosphate, and glucosamine-6phosphate gave limited inhibition. In most cases, theinhibitors were independent in their action, and cumulative degrees of inhibition wereobtained with mixtures of these end products. In contrast, with the glutamine syn-thetases of the two Bacillus species, the simultaneous presence ofAMP and histidine(or AMP and glutamine) gave inhibition greater than the sum of te amounts ofinhibition caused by either inhibitor alone. Also, alanine and carbamyl phosphateacted synergistically to inhibit the enzyme from N. crassa. The remarkable siityin the overall patterns of end-product inhibition observed with the enzymes fromdifferent sources indicates that these diverse organisms have evolved comparablemechanisms for the regulation of glutamine metabolism. Nevertheless, the enzymesfrom different sources do differ significantly in their physical and catalytic properties,as was demonstrated by dissimilarities in their purification behaviors, specificity fornucleotide substrate, ability to catalyze the glutamyl transfer reaction, and abilityto utilize Mn+ and Mg++ as activators for the biosynthetic reaction.

A variety of mechanisms for feedback controlof enzymes catalyzing branched biosyntheticpathways have been recognized (for review see27). These include (i) the elaboration of iso-enzymes which individually are inhibited by dif-ferent end products and (ii) several mechanismsin which a single form of the enzyme is effectivelyinhibited only when more than one of the endproducts are simultaneously present. Our presentknowledge does not permit the conclusion thatdifferent microorganisms utilize a single mech-anism for the control of a particular enzyme. Insome cases (5), closely related species may exhibitmarked differences in the patterns of feedbackcontrol of an enzyme which presumably occupiesthe same relative position in their metabolism.

1 A portion of this paper was presented at the 66thAnnual Meeting of the American Society for Micro-biology, Los Angeles, Calif., May 1966.

' Recipient of a Public Health Service PostdoctoralFellowship. Present address: Bioscience Section, JetPropulsion Laboratory, Califomia Institute of Tech-nology, Pasadena.

Glutamine synthetase represents a complexmetabolic branch-point and thus imposesspecial requirements for an effective means ofcontrol. The amide nitrogen of glutamine may beutilized in diverse pathways leading to the forma-tion of purines (2), cytidine triphosphate (Hulbertand Chakrobarty, Federation Proc. 20:361,1961), tryptophan (8), histidine (18), carbamylphosphate (23), glucosamine-6-phosphate (12),pyridine nucleotides (24), and p-aminobenzoicacid (29). In addition, the a-amino group ofglutamine is a potential source of nitrogen for thetransaminative production f several a-aminoacids (15).From studies with the glutamine synthetase of

Escherichia coli (30), evidence was presented for aunique type of multiple end-product control, viz.,cumulative feedback inhibition. When testedindividually, alanine, glycine, histidine, trypto-phan, adenosine-5'-monophosphate (AMP), cyti-dine-5'-triphosphate (CfP), carbamyl phosphate,or glucosamine-6-phosphate caused limiteddegrees of inhibition. Cumulative degrees of

1045

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

.org/D

ownloaded from

HUBBARD AND STADTMAN

inhibition were obtained when various mixturesof these end products were tested. Further studies(Woolfolk and Stadtman, in press; Shapiro andStadtman, unpublished data) have indicated thatthese eight (inhibitors) react at separate sites onthe E. coli enzyme.Somewhat different patterns of control have

been suggested in preliminary studies with theglutamine synthetases of other microorganisms.For example, Kohlhaw et al. (10) reported thatnicotinamide adenine dinucleotide (NAD), CTP,and guanosine-5'-monophosphate (GMP) were

feedback inhibitors for the enzyme from Sac-charomyces cerevisiae. For the glutamine syn-thetase of Lactobacillus arabinosus (25), theproduct of the reaction in the synthetic direction,glutamine, serves as a potent inhibitor. To deter-mine to what extent these various regulatorymechanisms are general in nature, the patterns offeedback control of the glutamine synthetase of10 diverse microbial types were determined andare reported in the present study. The organismsselected for our investigation include aerobic,anaerobic, and photosynthetic bacteria, a mold, a

yeast, and a green alga.

MATERIALS AND METHODS

Organisms and culture media. The conditions forgrowth of the aerobic organisms are given in Table 1.Clostridium pasteurianum ATCC 6013 was grown

anaerobically at 30 C under conditions described byCarnahan and Castle (3) with 0.5% sodium glutamateas the nitrogen source. Rhodospirillum rubrum ATCC11170 was grown photosynthetically with malate at30 C under conditions described by Omerod et al. (20)with 0.5% sodium glutamate as the nitrogen source.

Preparation of cell-free extracts. Cell pastes afterharvest by centrifugation were washed once with 0.01M imidazole chloride buffer (pH 7) and were resus-

pended in 0.01 M imidazole buffer containing 0.01 M

MnCl2 and 0.001 M ,-mercaptoethanol. Micrococcussodonensis cells were disrupted by 15-min treatmentwith a sonic oscillator (Branson Instruments, Inc.,Stamford, Conn.). The other organisms were rupturedby passage through a French pressure cell. After eithermethod of rupture, the cell debris was removed bycentrifugation at 20,000 X g for 20 min. The resultingsupernatant solutions were subjected to the purifica-tion procedures described in Table 2. The three basicfractionation steps employed were precipitation byaddition of solid ammonium sulfate, by dropwiseaddition of 1 M acetic acid, and by dropwise additionof acetone. Precipitates were collected by centrifuga-tion at 20,000 X g for 20 min and were immediately

TABLE 1. Growth conditions for the aerobic organisms

IncubationOrganism Source Additions to basal mediuma

Time Temp

hr C

Salmonella typhimurium Bruce Ames, National Institute Glucose, 0.4% 18 3701202 of Arthritis and Metabolic

DiseasesMicrococcus sodenensis J. J. Perry, Department of Na acetate, 0.5%; bio- 18 37

Bacteriology, North Carolina tin, 60 m/hg/mlState University, Raleigh

Bacillus licheniformis ATCC 9945a Glucose, 0.4% 16 37B. cereus 14B22 Department of Microbiology, Glucose, 0.5%; biotin, 20 37

University of Texas, Austin 60 mjg/mlPseudomonas fluorescens N. N. Durham, Department of Glucose, 0.4% 15 37

Microbiology, OklahomaState University, Stillwater

Candida utilis ATCC 9950 Glucose, 1.2% 24 30Neurospora crassa EM 5297 Martin Flavin, National Heart Sucrose, 0.4%; biotin, 48 30

Institute 60 m,g/mlChlorella pyrenoidosa G. R. Vela, Department of Glucose, 0.4%; yeast 96 37TX71105 Microbiology, North Texas extract, 0.0025%; de-

State University, Denton hydrated nutrientbroth, 0.16%

Basal medium: K2HPO4 3H20, 13.9 g; KH2PO4, 5.3 g; NaCl, 2.5 g; K2SO4, 2.5 g; MgSO4-7H2O, 1 g;monosodium glutamate, 5 g; and tap water, 1,000 ml. The solutions of the phosphates, glutamate, andcarbon sources were sterilized separately. Uninoculated medium was pH 7. The C. pyrenoidosa cultureswere shaken in the dark in 4-liter flasks containing 600 ml of medium. The other organisms were grownin 20-liter carboys containing 10 liters of medium with forced aeration. The inocula were grown in themedia described above.

1046 J. BACTERIOL.

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

.org/D

ownloaded from

VOL. 93, 1967 FEEDBACK INHIBITION OF GLUTAMATE SYNTHETASE

TABLE 2. Partial purifications and the activity measurements of the microbial glutamine synthetases

Organism

Salmonella typhimurium

Micrococcus sodonensis

Bacillus licheniformisB. cereusPseudomonas flu-orescens

Clostridium pasteuria-num

Rhodospirillum rubrum

Candida utilis

Neurospora crassa

Chlorella pyrenoidosa

Fractionation of extract(precipitated at)

pH 5.1-4.4; (NH4)2SO4,33-55% saturation

pH 4.9-3.9; acetone,0-45% at pH 5.7

pH 5.1-4.4, twicepH 5.1-4.4, twicepH 5.1-4.4; acetone, 0-45% at pH 5.7;(NH4)2SO4, 50-70%saturation

(NH4)2SO4, 50-70%saturation; acetone,0-45% at pH 7

pH 5.7-4.7; acetone,0-45% at pH 5.7

pH 5.1-4.4; acetone,0-45% at pH 5.7

(NH4)2SO4, 30-50%saturation;(NH4)2SO4, 0-40%saturation

pH 7-4.4

Glutamine synthetase activity

Biosynthetic assaya

(A) WithWith MnC12 MWith

0.91

0.33

0.0850.150.66

0.14

0.26

0.13

0.13

0.05

1.08

0.09

0.391.751.92

0.81

0.75

0.81

0.94

0.34

(C)Glutamyl

transfer assay'

6.29

3.56

0.056.08

21.1

18.8

39.8

2.3

2.45

Activity ratio

B/A

1.2

0.27

4.611.72.9

5.8

2.9

6.1

7.2

6.8

C/A

6.9

10.8

0.640.5

150

72

30.6

17.7

49

a Results expressed as micromoles of Pi per minute per milligram of protein. Reaction mixture: cell-free extract, 0.0075 M ATP, 0.1 M glutamate, 0.05 M NH4Cl, 0.02 M imidazole, and 0.005 M MnCl2 (for A)or 0.05 M MgCl2 (for B).

b Results expressed as micromoles of glutamylhydroxamate per minute per milligram of protein.Reaction mixture: cell-free extract, 0.1 M glutamine, 0.3 M NH20H, 0.02 M arsenate, 0.0004 M ADP,0.003 M MnCI2, and 0.02 M imidazole.

dissolved in the imidazole-MnCl2 buffer describedabove. The fractionated extracts were stored at 4 C.Under these conditions (manganous ion present), theglutamine synthetase preparations showed no appreci-able losses in activity during 7 days of storage.

Biosynthetic assay of enzyme activity. Glutaminesynthetase catalyzes the amidation of glutamate,according to equation 1.

glutamate + NH4+ + adenosine-5'-triphosphate(ATP) Mn++(Mg++)) glutamine + adenosine- (1)

5'-diphosphate (ADP) + Pi

Enzyme activity is conveniently estimated by meas-uring the amount of glutamate-dependent ortho-phosphate (Pi) liberated (1). The complete (assay)mixture contained cell extract, L-glutamic acid (mono-sodium), NH4Cl, ATP, MnCl2, and imidazole in atotal volume of 0.3 or 0.6 ml at a final pH of 7.0. Theconcentrations of reactants are given in the appro-priate tables. Incubations were carried out in opentubes at 30 or 37 C, coinciding with the growth tem-perature used for the organism from which the extract

was prepared. The reactions were initiated by theaddition of enzymes. Proportionality of Pi liberationto enzyme concentration is observed up to 0.25 ,moleof Pi per 15 min of incubation under the assay condi-tions. Controls lacking glutamate were included ineach incubation employed, and corrections were madefor the small amounts of Pi liberated in the absence ofsubstrate.

Net synthesis of glutamine was established by paperchromatography of deproteinized incubation mix-tures using 1-butanol-water-acetic acid (4:4:1) asthe developing solvent. The Rp values of glutamineand glutamic acid were 0.18 and 0.25, respectively.

Glutamyl transfer assay of enzyme activity. Gluta-mine synthetase also catalyzes a transfer of the -y-glutamyl moiety of glutamine to hydroxylamineaccording to equation 2.

glutamine + hydroxylamine ADP,ln4,As'

glutamylhydroxamate + NH4+(2)

For some studies, the amounts of -y-glutamylhy-droxamate formed in this reaction were used to esti-

1047

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

.org/D

ownloaded from

HUBBARD AND STADTMAN

mate glutamine synthetase activity. The y-glutamylhy-droxamate after reaction with ferric chloride (14) wasmeasured with a Klett-Summerson colorimeterequipped with a 540-mju filter, with authentic y-glutamylhydroxamate as the standard. The completeassay mixture contained cell extract, L-glutamine,neutralized hydroxylamine hydrochloride, arsenate,ADP, MnCl2, and imidazole, in a final volume of 1ml at pH 7.0. The concentrations of reactants aredescribed in the appropriate tables. Incubations wereperformed in open tubes at 30 or 37 C, the optimalgrowth temperature for the particular organism. Thereactions were initiated by the addition of enzyme.Glutamylhydroxamate formation is proportional toenzyme concentration in the range of 0 to 3 ,molesper 15 min of incubation. Controls lacking glutamineor ADP and arsenate were included in all assays.

In experiments wherein magnesium was employedas metal activator, the cell extracts were freed frommanganese either by exhaustive dialysis against 200to 500 volumes of 0.01 M imidazole buffer or by pre-cipitating the glutamine synthetase by one of thepreviously described methods and redissolving thecarefully drained pellet in the imidazole buffer. Pro-tein concentrations were estimated from absorbancyat 260 and 280 m,u (11).

Inhibition studies. Compounds tested as inhibitorswere added to the reaction mixtures prior to the intro-duction of enzyme. Controls for each inhibitor wereincluded in the experiments with both the biosyntheticand glutamyl transfer assays. With the biosyntheticassay, the control tube contained the inhibitor butlacked glutamate. With the transfer assay, the in-hibitor was added and glutamine was omitted. Theresults of inhibition analyses are expressed as per centactivity, i.e., 100 times the fractional activity observedin the presence of inhibitor. The reported valuesrepresent the average of two or more determinations.

Chemicals. ATP (disodium), CTP (disodium), L-glutamine (grade III), and L-glutamic acid-mono-hydroxamate (y) were obtained from Sigma ChemicalCo., St. Louis, Mo. ADP (sodium salt) and 5'-adenylic acid were obtained from Pabst Laboratories,Milwaukee, Wis. Carbamyl phosphate (dilithiumsalt, B grade) was obtained from Calbiochem, LosAngeles, Calif. All other chemicals were reagent grade.

REsuLTs

It is obvious that the presence of multiple formsof glutamine synthetase in a given organism couldbe most decisively demonstrated by examinationof crude unfractionated extracts, since anyfractionation procedure applied to cell-freeextracts might conceivably result in the selectivedestruction of one or more isoenzymes. Un-fortunately, the unambiguous demonstration ofglutamine synthetase activity in crude extractsof some organisms is not possible, owing to thepresence of phosphatase activities that catalyzethe liberation of Pi from ATP and thereby causeinterference in the biosynthetic assay procedure.Moreover, the possible presence of amidohydro-

lase, which leads to the formation of glutamyl-hydroxamate from glutamine and hydroxyl-amine as well as other hydroxamate-formingsystems, precludes the use of the transfer assay asa reliable measure of glutamine synthetaseactivity in crude extracts. Therefore, it wasgenerally necessary to subject crude extracts tosome fractionation steps to eliminate interferingactivities. Table 2 summarizes the minimaltreatment used to separate most of the interferingenzymes from the glutamine synthetases in thevarious cell-free extracts. The purification be-haviors serve also to illustrate certain differencesin the glutamine synthetases from the varioussources. For example, different levels of acid andammonium sulfate were required for the precipita-tion of enzymes from some of the microorgan-isms. Also, the glutamine synthetases of C.pasteurianum and Neurospora crassa wereparticularly acid-sensitive; these were almostcompletely inactivated at pH 5.7.

Variation in response to metal activation. Theability to catalyze the biosynthetic reaction withMnCl2 as the metal activator offers a means forcomparing the microbial glutamine synthetases.The wide range in activity ratios (Table 2)exhibited by the enzymes from the various micro-organisms demonstrates differences in theircatalytic properties. The enzymes from all of thesources except M. sodonensis showed higheractivities with Mge4 as the activator (B/A ratio> 1). Nevertheless, the assay system containingMnCI2 was used for most of the subsequentanalyses, since certain of the glutamine syn-thetases were unstable in the absence of man-ganous ion. The enzymes from Bacillus cereusand B. licheniformis were particularly unstable inthis regard. This is similar to the reportedstabilizing effect of manganous ion on the glu-tamine synthetases of E. coli and L. arabinosus(25, 30).

Variations in relative rates of transfer andsynthetic reaction. The microbial glutaminesynthetases also exhibited marked differences inthe relative rates in which the glutamyl transferand biosynthetic reactions were catalyzed (C/Aratios, Table 2). B. licheniformis was of particularinterest, since it had previously been reported(13) that its glutamine synthetase does notcatalyze the ADP-dependent glutamyl transferreaction. In the present study, a reproducibly lowlevel of glutamotransferase activity was detectedin the B. licheniformis extracts. Additional experi-ments (Hubbard and Stadtman, unpublished data)have indicated that the low activity is the con-sequence of a pronounced inhibition caused byglutamine, a substrate of the reaction.

In Pseudomonas fluorescens extracts, the rela-

1048 J. BACrERioL.

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

.org/D

ownloaded from

FEEDBACK INHIBITION OF GLUTAMATE SYNTHETASE

tive activities of the transfer and synthetic reac-tions could not be compared because of thepresence of an amidohydrolase which catalyzedthe deamidation of glutamine or the formation ofglutamylhydroxamate from glutamine and hy-droxylamine. The latter reaction, in contrast tothe glutamyl transfer reaction catalyzed byglutamine synthetase, is not dependent on ADPor arsenate. By means of an acid precipitationstep (pH 4.4 to 3.5) amidohydrolase preparationswere obtained which were free from glutaminesynthetase. However, the reciprocal removal ofamidohydrolase activity from glutamine syn-thetase preparations was not accomplished. Thus,this interfering activity made it impractical tomeasure the glutamotransferase activity (Tables2 and 7) or to test glutamine as a feedbackinhibitor of the biosynthetic reaction catalyzedby the glutamine synthetase of P. fluorescens(Table 3).Comparisons of potential end products as feed-

back inhibitors. Results of analyses in whichpotential end products were tested as feedbackinhibitors are presented in Table 3. This surveyincluded seven of the eight compounds pre-viously found to be inhibitory for the glutaminesynthetase of E. coli (30), as well as glutamine,which was reported to inhibit the enzyme fromL. arabinosus (25). The assay conditions em-ployed to demonstrate the inhibitions were either

limiting glutamate with excess NH4C1 or limitingNH4C1 with excess glutamate, as indicated. Asseen in Table 3, the glutamine synthetase of eachorganism is inhibited by more than one endproduct. Moreover, each enzyme exhibits aunique regulation pattern which differs from theothers in respect to the extent of inhibitioncaused by the end products. However, certainsimilarities are apparent in the inhibition patternsof the glutamine synthetases of closely relatedorganisms. This similarity is seen with the twoBacillus species, and the inhibitory response of theenzyme from S. typhimurium resembles thatpreviously reported for E. coli glutamine syn-thetase (30). In addition to the compoundsreported in Table 3, p-aminobenzoic acid,anthranilic acid, methionine, isoleucine, GMP,and NAD were either noninhibitory or onlyslightly inhibitory, causing less than 5% diminu-tion in activity.As is also shown in Table 3, different degrees of

inhibition were obtained depending on whetherMnCI2 or MgCl2 was used in the assay system.With many of the enzymes studied, inhibitionscaused by alanine and glycine were greater whenMn++ was the activator, whereas the conversewas often true of the AMP inhibition. The choiceof metal activator even more dramatically in-fluenced the manner in which CTP affected theactivities of the glutamine synthetases of the

TABLE 3. Influence of end products of glutamine metabolism on the activities of the microbialglutamine synthetasesa

Source of enzyme

Salmonella typhimurium..Micrococcus sodonensis..Bacillus licheniformis....B. cereus................Pseudomonas fluorescens.Clostridium pasteurianumRhodospirillum rubrum...Candida utilis...........Neurospora crassa.......Chlorella pyrenoidosa....

Per cent activity in the presence of

Alab(5mm)

65842834511949888771

(58)(-)(-)(53)(49)(60)(31)(99)(98)(90)

Glyb(5mM)

80 (68)85 (-)51 (-)51 (75)59 (50)38 (62)68 (42)83 (100)87 (90)72 (90)

Tryb(Sme)

77 (95)98 (-)95 (-)96 (95)88 (95)90 (79)90 (95)98 (95)100 (95)100 (99)

AMPb(5mM)

83 (67)75 (-)38 (-)36 (-)83 (88)93 (84)88 (45)79 (64)(7 )d

73 (65)

cTPb(2mM)

79 (86)79 (-)85 (-)78 (-)85 (-)87 (113)72 (86)

(92)d(86)d(87)d

GluNH2b(10m)

78 (94)96 (-)60 (-)67 (-)

(-)41 (15)77 (87)96 (98)96 (94)94 (88)

Hisc(15mm)

67589185868375695280

GA-6-PC

(5mm)

8580104101979094858880

AMP¢(5m)

8375687181908490±82

a Abbreviations: Ala = L-alanine; Gly = glycine; Try = L-tryptophan; AMP = 5'-adenylic acid;CTP = cytidine-5'-triphosphate; GIuNH2 = L-glutamine; His = L-histidine; GA-6-P = D-glucosamine-6-phosphate.

b Assay system: 0.0075 M glutamate, 0.05 M NH4Cl, 0.0075 M ATP, 0.02 M imidazole, and metal acti-vator (as specified below). The values to the left refer to activity measurements when 0.005 M MnC12was used on the activator. The values in the parentheses refer to activity measurements when 0.05 MMgCl2 was used on the activator.

c Assay system: 0.1 M glutamate, 0.001 M NH4Cl, 0.0075 M ATP, 0.02 M imidazole, and 0.005 M MnCl2.d CTP and AMP served as substrates for Pi production.

1049VOL. 93, 1967

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

.org/D

ownloaded from

HUBBARD AND STADTMAN

higher protists, N. crassa, Candida utilis, andChlorella pyrenoidosa. When the assay systemcontained MgCl2, CTP served as an inhibitor ofthe enzymes from these three organisms, whereas,with MnCl2, CTP stimnulated Pi production. Otherexperiments with N. crassa have shown that theapparent stimulation by CTP is due to theability of CTP to replace ATP as a substrate forglutamine synthetase. For example, in theabsence of ATP, Pi is released from CT? only inthe presence of glutamate and ammonia. More-over, net synthesis of glutamine (as determinedby paper chromatography) could be demon-strated in both the CTP- and ATP-dependentreactions. Table 4 gives results of additionalinvestigations on these roles of ATP and CTP.In the assay containing MnCl2, CIP was slightlymore effective than ATP as substrate. Theactivities were additive when a mixture of 7.5 mmATP and 2 mm CTP was used. However, with asaturating level of ATP (15 mM), additional CTP(6 mM) did not stimulate Pi production. Thus, itwould appear that the glutamine synthetase ofN. crassa is a single enzyme which, when acti-vated by Mn4 +, requires high levels of ATP orCTP for maximal activity. Further evidence tosupport the view that the nucleotides are sub-strates for the same enzyme was obtained inother experiments, which showed that histidinecaused comparable degrees of inhibition of theATP- and CTP-dependent activities. When thereaction mixture contained MgC12, CTP wasmuch less effective than ATP as substrate.Monder (16) has reported that only Mn++activates mammalian glutamine synthetase when

TABLE 4. Utilization of ATP and CTP by theglutamine synthetase of Neurospora crassa

Reaction mixture"Pi produced6

MnC12 MgCI2 ATP CTP

mm mm mm mm

5 0 7.5 0 0.485 0 15 0 0.875 0 22.5 0 0.745 0 0 2 0.245 0 0 4 0.425 0 0 6 0.855 0 7.5 2 0.645 0 15 6 0.850 50 7.5 0 3.870 50 0 4 0.12

a Plus 0.1 M glutamate, 0.05 Mimidazole, and enzyme.

NH4C1, 0.02 M

uridine, inosine, or guanidine triphosphates areused, whereas either Mn44 or Mg++ activates themammalian enzyme when ATP is the nucleotidesubstrate.As shown in Table 3, CTP also stimulated Pi

production by the extracts of C. pasteurianum.However, this activity does not appear to be aconsequence of CTP acting directly as a sub-strate for glutamine synthetase. Other experi-ments have shown that CTP is ineffective as asubstrate for Pi production when ATP is omittedfrom the reaction mixture containing the C.pasteurianum extract. N. crassa extracts catalyzedthe liberation of Pi from AMP. This latteractivity appeared to be unrelated to glutaminesynthetase, since it was not dependent on glu-tamate or ATP.

Effects of combinations of inhibitors. In anattempt to characterize the patterns of multipleend-product regulation for the glutamine syn-thetases, experiments were conducted withseveral pairs of inhibitors (Table 5). For purposesof comparison, values were calculated on theassumption that each of the two inhibitors acts ona different form of the enzyme, or that both actindependently on a single form of the enzyme.Certain pairs of inhibitors obviously do not acton separate enzymes, since the calculated sum ofthe individual inhibitions exceeds 100% (nega-tive values). With the majority of the other pairstested, the experimental values are in closeragreement with the values calculated by assumingthe two inhibitors interact with a single form ofthe enzyme. The data which more closely agreewith the enzyme multiplicity predictions were ob-tained when one inhibitor of the pair caused asmall degree of inhibition. In these situations, thedifferences between the two theoretical values aresmall; therefore, the margin of experimentalerror prevents one from distinguishing betweenthe assumptions. Histidine and AMP actedsynergistically on the glutamine synthetases of B.licheniformis and B. cereus in causing inhibitionswhich were decidedly greater than either of thecalculated values (Table 5). This synergistic ac-tion, wherein the simultaneous presence of twoend products results in greater inhibition than thefractional inhibitions caused by either endproduct alone, has been observed with enzymesother than glutamine synthetase (4, 21), and hasbeen termed cooperative feedback inhibition.However, to avoid confusion with the cooperativeeffects often seen in the homologous interactionsof feedback inhibitors, it would appear desirableto use another term; hence, we suggest the termsynergistic feedback inhibition. The glutaminesynthetases of the Bacillus species were alsosusceptible to synergistic feedback inhibition with

I Micromoles per 15 min per milligram of pro-tein. Calculated from determinations made withappropriate dilutions of enzyme preparation.

1050 J. BACrERIOL.

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

.org/D

ownloaded from

VOL. 93, 1967 FEEDBACK INHIBITION OF GLUTAMATE SYNTHETASE 1051

0v o 0%l, Nx, 0% 0% 2 o0) It, 1f) It, It, sC N It, It,_CtC) U0 -

+ NOO %%Oo

x .c, v Z ^ . It It N a)va)

I-n zCen00Nr-~o00 %o

c:£24>, 5^81£ rv S QQI

0 -N o N00 N N_ cc

t- -IIQ Q

W')0 00 %0 0. )-U %O 0%O en001-N -0 0 00

01 L + 5.s C~ Cl n 00 r- I I a ,_ a

L. e -n̂oa)m I'

+ I0 N CO CO N 00 N N N F > C

u en-1t) mm

In

+e C Ir en e IS

0 4+.=2£ t N o v m u i 2:;o~~~~ONO £ t cr)~~~~~~~e en tr )

un o

N o c N 00 en0 _at, 00 ~~~~. ~t. t~. 00 't 0s_Z'Ev _s

~2%O N ~a %0 Clt 4 1- "-Cq I NH -_N t_<52-'_0 -'N--1t 0 IU _ t, 'IC t CO Nd aen t net) dII+ t) so C4 C1 dq _t en

1- .- _- I-- _-o _o- I,

.0 00

PO tn ID cr t b(1 00 ON- 00 %O %

+ 10 - - - t

lotN00 b ^ 0%^ m t F N S t Wt00 _

ei00e~e~10 ~ Ne~a ~% 00N00 C1 en 00 l_

ell0%Ist0tz--s N-N I0

N

S- -E- i ,

Cl en¢ C

q6)~ ~ ~

- a) -0 W

oz CZ *

o o4- J c0=cis*E

W Cduz Q C

W W E

i a)C.0_

C0)2a

:> co(_

0co

') a.-r S.°0 = O)

W C :i4- >' -.

u W cdUnC;S

cc 0 0,,,

Cd_°O5

3= VX NeXX .

*.bs

1.2

C1.0 0

0(Q

1.20 >

Q .00a 0..0z0

I..

ce2

-0*0*_

0

I-

0_

$t--4

tc;wb4

E--(

0.:R

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

.org/D

ownloaded from

HUBBARD AND STADTMAN

a mixture of AMP and glutamine. This lattersynergism can be more lucidly demonstratedwhen lower levels of the co-effectors are employed(Hubbard and Stadtman, unpublished data).With several pairs of inhibitors (Table 5), the

experimental activities were considerably higherthan either of the calculated values. These antag-onisitic effects were most often observed whenalanine or glycine served as one of the inhibitors.This antagonism is most readily visualized byassuming that the two ligands act on a single formof the enzyme. One possible explanation is thatthe two end products are mutually competitivefor the same site. Alternatively, the occupation ofone site by its ligand could interfere with thebinding of the second ligand. Regardless of thereason, the varying degrees to which two inhib-itors act antagonistically for the various glut-amine synthetases illustrate differences in theinteraction of the inhibitors with the enzymesfrom different sources.

Further investigations showed that the micro-bial glutamine synthetases could be inhibited 53to 94% with a mixture of six end products ofglutamine metabolism (Table 6). These data can-not be directly compared with those presented inTables 3 and 5, since different assay conditionswere used. However, it would seem likely thatcertain of these enzymes would have been moreeffectively inhibited had other end products ofglutamine metabolism been included in theinhibitor mixtures. In any event, these data andthe data of Table 5 indicate that the simultaneouspresence of a physiological excess of several endproducts causes greater inhibition than when onlyone or two are present in excess. It therefore ap-pears evident that mechanisms involving some

degree of cumulative feedback inhibition areoperative in most of the organisms examined.End products of glutamine metabolism also

inhibit the glutamyl transfer reaction catalyzed bythe microbial glutamine synthetases. The data inTable 7 show that these activities are inhibited tovarying degrees by alanine, AMP, and carbamylphosphate. Again, the results from assays withpairs of inhibitors generally favor the assumptionthat with each organism inhibitors are acting ona single form of the enzyme. N. crassa representsthe exceptional case, in that alanine and carbamylphosphate acted synergistically in inhibiting theglutamotransferase activity. Additional experi-ments have shown that the biosynthetic reaction

TABLE 6. Effects of a mixture of six end productson the activities of the glutamine synthetasesa

Per cent activity inthe presence of

Source of glutamine synthetase prepn Ala + Gly + Try+ His + AMP +GA-6-P (5 mm each)

Salmonella typhimurium ......... 32Micrococcus sodonensis.......... 47Bacillus licheniformis............ 6B. cereus....................... 13Pseudomonas fluorescens ....... 37Clostridium pasteurianum ........ 30Rhodospirillum rubrum.......... 30Neurospora crassa....... 20

20bCandida utilis................... 29Chlorella pyrenoidosa ........... 9

a Reaction mixture: 0.0075 M glutamate, 0.001 MNH4CI, 0.0075 M ATP, 0.005 M MnC12, 0.02 Mimidazole, and enzyme.bAMP omitted from this inhibitor mixture.

TABLE 7. Effects of end products of glutamine metabolism on the glutamotransferase activitiesa

Per cent activity in the presence ofSource of glutamotransferase prepn

Ala AMP CP Ala + CP Ala + AMP AMP + CP

Salmonella typhimurium 37 55 51 18 (19)b 22 (20) b 31 (28)bMicrococcus sodonensis 70 81 68 44 (48) 56 (57) 53 (55)Bacillus cereus.32 70 85 30 (27) 20 (22) 56 (60)Clostridium pasteurianum 9 98 59 6 (5) 9 (9) 58 (58)Rhodospirillum rubrum 6 98 61 6 (4) 5 (6) 64 (60)Candida utilis.52 95 47 33 (24) 52 (49) 47 (45)Neurospora crassa. 64 99 71 24 (46) _Chlorella pyrenoidosa 62 97 61 30 (37) 61 (60) 61 (59)

a CP, carbamyl phosphate. Inhibitor concentrations were 5 mm. Reaction mixture: 0.03 M glutamine,0.06 M NH20H, 0.02 M arsenate, 0.004 M ADP, 0.003 M MnC12, 0.02 M imidazole, and cell extracts.

I The numbers to the left are the per cent activities observed in the presence of the pair of inhibitors.The numbers in parentheses are values calculated with the assumption that the two inhibitors act inde-pendently on a single form of the enzyme, i.e., 100 times the product of the fractional activities observedwhen the inhibitors were tested individually.

1052 J. BACTERIOL.

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

.org/D

ownloaded from

VOL. 93, 1967 FEEDBACK INHIBITION OF GLUTAMATE SYNTHETASE

catalyzed by the N. crassa enzyme is also sus-ceptible to this synergistic feedback inhibitioncaused by the alanine and carbamyl phosphatemixture.

DISCUSSION

The formation of glutamine may be regardedas the first enzymatic step of a highly branchedpathway that leads ultimately to the biosynthesisof a variety of end products. Therefore, byanalogy to the regulation of other branchedpathways, it is not surprising that glutaminesynthetase is subject to feedback inhibition by theultimate end products of glutamine metabolism.

Previous studies have shown that the regulationof a given biosynthetic branched pathway may beachieved by distinctly different mechanisms indifferent organisms (5, 27). For example, theregulation of aspartokinase, which is the firstcommon step in the biosynthesis of lysine,methionine, threonine, and isoleucine, involvesthe elaboration of multiple enzymes in E. coli (28),concerted feedback inhibition in Rhodopseudo-monas capsulatus (5) and in B. polymyxa (22), asequential feedback mechanism in R. spheroides(Datta, personal communication), and still dif-ferent mechanisms in Rhodospirillum rubrum (5)and in S. cerevisiae (26). It is, therefore, some-what surprising that the feedback controlpatterns of the glutamine synthetases from widelydifferent organisms appear to be qualitativelyvery similar. In general, the patterns of inhibitionare similar to the cumulative inhibition establishedin E. coli (30) Thus, each of the various endproducts is able by itself to cause only a limiteddegree of inhibition; however, as the number ofend products that are present simultaneously isincreased, there is a progressive increase in theextent of inhibition. Such a mechanism ofinhibition constitutes an elegant means of meta-bolic control of those enzymes that catalyze thesynthesis of an early precursor that is a commonintermediate in the biosynthesis of several endproducts. For, as each of the various end productsbecomes present in excess, it causes a curtail-ment in the biosynthesis of the common precursorby an amount presumably required for the syn-thesis of the particular end product.The drastic consequences that would be

expected (28) if a physiological excess of one endmetabolite could cause total inhibition of thecommon precursor are thus avoided. Yet, whenall end products are present in more than ade-quate amounts, the synthesis of the commonprecursor is effectively stopped.

It is obvious that, if such a mechanism ofcontrol is to be effective, it must be supplemented

with secondary feedback controls by each of thediverse end metabolites of the first unique stepthat is specifically concerned with its biosynthesis.By means of these secondary controls, the reducedsupply of the common precursor that is availablewhen a given end product is present in excess isdiverted away from that particular end productand is thus made available only for the synthesesof those end products that are still required.

In the case of glutamine metabolism, theexistence of secondary controls in some bacteriahas been established for the first unique stepinvolved in the biosynthesis of CTP (31), AMP(4), histidine (17), and tryptophan (8). Whereasit is reasonable to assume that similar secondarycontrols occur also in the biosynthesis of the otherfeedback inhibitors of glutamine synthetase, theexistence of such controls remains to be demon-strated.Although the present data suggest that each of

the organisms studied elaborates only one gluta-mine synthetase that is susceptible to partialinhibition by the different end products, thepossibility is not excluded that multiple forms ofthe enzyme are produced in some organisms. Aparticularly labile isoenzyme could have beendestroyed during preparation of the cell-freeextracts, or by the minimal purification pro-cedure required to eliminate reactions that inter-fered with the enzyme assays. Also, the enzymeassay conditions employed might not be favor-able for the detection of isoenzymes with differentoptimal requirements. Moreover, in those situa-tions where the maximal inhibition obtained by agiven inhibitor is very small, it is not possiblewith the relatively insensitive assay proceduresemployed to distinguish between the possibilitythat they act independently on a common enzyme(cumulative feedback inhibition) or are specificinhibitors of isoenzymes that are present in smallquantities.

It should be emphasized that the significanceof the observed inhibition patterns for the gluta-mine synthetases from different organisms can beproperly evaluated only after the completeregulatory patterns for the biosynthesis of eachof the multiple inhibitors is also established inthese organisms. It remains to be determinedwhether the inhibition of glutamine synthetase bysome of the metabolites is due to end-productinhibition, or if instead it is the manifestation ofa modified sequential feedback mechanism inwhich the inhibitor is an intermediate rather thanan ultimate end product. For example if glutamineis the physiological nitrogen donor in the bio-synthesis of glucosamine-6-phosphate (GA-6-P),then would not N-acetyl-GA-6-P, UDP-Nacetyl-

1053

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

.org/D

ownloaded from

HUBBARD AND STADTMAN

glucosamine, or some other derivative be moreappropriately regarded as the ultimate endproduct? If so, then the inhibition of glutaminesynthetase by GA-6-P might be the expression ofa sequential feedback control mechanism (9, 19)in which the last common intermediate (GA-6-P)of a branched pathway inhibits the first bio-synthetic step leading to its formation. If this istrue, then it follows that the first steps involved inthe further metabolism of GA-6-P should beunder feedback control by UDP-N-acetylglucosa-mine or other terminal end products.The fact that glutamine is a strong inhibitor of

the glutamine synthetase from some organisms,might also be attributable to the operation of asequential feedback control system in these or-ganisms. In the presence of secondary controlsinvolving feedback inhibitors of the first stepsthat are uniquely concerned with the biosynthesisof the individual end products, it follows that theinhibitory action of an end product at its second-ary site would spare the utilization of glutamine.This would result sequentially in the accumulationof glutamine, which in turn would lead to inhibi-tion of glutamine synthetase, thereby controllingits own synthesis. Such a mechanism of sequentialfeedback inhibition of the glutamine synthetasesof B. licheniformis and B. cereus could be rein-forced by the synergistic inhibition caused by thesimultaneous presence of glutamine and its endproduct AMP, thus providing for an even moredelicately balanced system of control.Although there is a remarkable qualitative

similarity in the inhibition patterns of the gluta-mine synthetases from different organisms, it maybe significant that there is considerable variationin the absolute effectiveness of any given inhibitortoward the different enzymes (see Table 3). Itremains to be seen whether these differencesreflect differences in the relative demands forglutamine in the biosynthesis of the variousmetabolites, or whether they perhaps reflectfundamental variations in the overall regulatorypatterns of the different organisms with efficientmechanisms for cellular regulation. By the sametoken, metabolic economy would suffer if abiosynthetic enzyme is inhibited by a metabolitewhose metabolic fate is not in some way depend-ent upon that enzyme. Attention is thereforedirected to the fact that alanine and glycine areamong the most effective inhibitors of the gluta-mine synthetases from all of the organisms exam-ined. Whereas these results suggest that glutaminemay be intimately involved in the biosynthesis ofalanine and glycine, a definite role of glutamine inthe formation of these two amino acids has notbeen unequivocably established.

Another question is raised by the observationthat carbamyl phosphate is an inhibitor for theglutamine synthetase of N. crassa. Davis (6)recently reported that ammonia, and not gluta-mine, was the source of nitrogen for the carbamylphosphate synthetase which was demonstrable incell-free extracts of this organism. The carbamylphosphate produced by this enzyme was specifi-cally utilized in the synthesis of arginine. On theother hand, physiological and genetic experiments(7) pointed toward a second mechanism for theformation of the carbamyl phosphate specificallyutilized in pyrimidine biosynthesis. Since thepresent findings indicate a regulatory role forcarbamyl phosphate in the synthesis of glutamineby N. crassa, it might be well to re-examine therole of glutamine in the synthesis of carbamylphosphate in this organism. If it is not involvedin the synthesis of the carbamyl phosphate usedspecifically for the synthesis of arginine, perhapsit is used for the synthesis of the aspartylphos-phate that is specifically involved in the synthesisof the pyrimidine.

LITERATURE CITED

1. BOYER, P. D., R. C. MILLS, AND H. J. FROMM.1959. Hypotheses for and some kinetic studieswith glutamine synthetase and acetate thio-kinase. Arch. Biochem. Biophys. 81:249-263.

2. BUCHANAN, J. M. 1960. Biosynthesis of purinenucleotides, p. 303-322. In E. Chargaff andJ. N. Davidson [ed.]. The nucleic acids, vol. 3.Academic Press, Inc., New York.

3. CARNAHAN, J. E., AND J. E. CASTLE. 1958. Somerequirements of biological nitrogen fixation. J.Bacteriol. 75:121-124.

4. CASKY, C. T., D. M. ASHTON, AND J. B. WYN-GAARDEN. 1964. The enzymology of feedbackinhibition of glutamine phosphoribosylpyro-phosphate amidotransferase by purine ribo-nucleotides. J. Biol. Chem. 239:2570-2579.

5. DATTA, P., AND H. GEST. 1964. Alternativepatterns of end-product control in biosynthesisof amino-acids of the aspartic family. Nature203:1259-1261.

6. DAVIS, R. H. 1965. Carbamyl phosphate synthesisin Neurospora crassa. I. Preliminary charac-terization of arginine-specific carbamyl phos-phokinase. Biochim. Biophys. Acta 107:44-53.

7. DAVIS, R. H. 1965. Carbamyl phosphate synthesisin Neurospora crassa. II. Genetics, metabolicposition, and regulation of arginine-specificcarbamyl phosphokinase. Biochim. Biophys.Acta 107:54-68.

8. EDWARDS, J. M., F. GIBSON, L. M. JACKMAN, ANDJ. S. SHANNON. 1964. The source of the nitrogenatom for the biosynthesis of anthranilic acid.Biochim. Biophys. Acta 93:78-84.

9. JENSEN, R. A., AND E. W. NESTER. 1965. Theregulatory significance of intermediary metabo-

1054 J. BACTRIOL.

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

.org/D

ownloaded from

FEEDBACK INHIBITION OF GLUTAMATE SYNTHETASE

lites: control of aromatic acid biosynthesis byfeedback inhibition in Bacillus subtilis. J. Mol.Biol. 12:468-481.

10. KOHLHAW, G., W. DRAGERT, AND H. HOLTZER.1965. Parallel-Repression der Synthese vonGlutamin-Synthetase und DPN-abhangigerGlutamatdehydrogenase in Hefe. Biochem. Z.341:224-238.

11. LAYNE, E. 1957. Spectrophotometric and turbidi-metric methods for measuring proteins, p.447-454. In S. P. Colowick and N. 0. Kaplan[ed.], Methods in enzymology, vol. 3. AcademicPress, Inc., New York.

12. LELOIR, L. F., AND C. E. CARDJNI. 1953. The bio-synthesis of glucosamine. Biochim. Biophys.Acta 12:15-22.

13. LEONARD, C. G., R. D. HOUSEWR1GHT, AND C. B.THORNE. 1962. Effects of metal ions on theoptical specificity of glutamine synthetase andglutamyl transferase of Bacillus licheniformis.Biochim. Biophys. Acta 62:432-434.

14. LEVINTOW, L. 1954. The glutamyl-transferaseactivity of normal and neoplastic tissues. J.Natl. Cancer Inst. 15:347-352.

15. MEISTER, A. 1962. Amino group transfer(survey), p. 193-217. In P. D. Boyer, H. Lardy,and K. Myrback [ed.], The enzymes, vol. 6.Academic Press, Inc., New York.

16. MONDER, C. 1965. Metal ion interactions andglutamine synthetase activity. Biochemistry4:2677-2686.

17. MOYED, H. S., AND B. MAGASANIK. 1960. Thebiosynthesis of the imidazole ring of histidine.J. Biol. Chem. 235:149-153.

18. NEIDLE, A., AND H. WAELSCH. 1959. The origin ofthe imidazole ring of histidine in Escherichiacoli. J. Biol. Chem. 234:586-591.

19. NESTER, E. W., AND R. A. JENSEN. 1966. Controlof aromatic acid biosynthesis in Bacillussubtilis: sequential feedback inhibition. J.Bacteriol. 91:1594-1598.

20. OMEROD, J. G., K. S. OMEROD, AND H. GEST. 1961.Light-dependent utilization of organic com-pounds and photoproduction of molecularhydrogen by photosynthetic bacteria; relation-ships with nitrogen metabolism. Arch. Bio-chem. Biophys. 94:449-463.

21. PArTE, J. C., T. LOVING, AND G. N. COHEN. 1965.Effects inhibiteurs cooperatifs de la L-lysineavec d'autres amino acids sur une aspartokinasede Escherichia coli. Biochim. Biophys. Acta99:523-530.

22. PAULUs, H., AND E. GRAY. 1964. Multivalentfeedback inhibition of aspartokinase in Bacilluspolymyxa. J. Biol. Chem. 239:PC4008-PC4009.

23. PIERARD, A., AND J. M. WIAME. 1964. Regulationand mutation affecting a glutamine dependentformation of carbamyl phosphate in Escherichiacoli. Biochem. Biophys. Res. Commun. 15:76-81.

24. PREISS, J., AND P. HANDLER. 1958. Biosynthesis ofdiphosphopyridine nucleotide. II. Enzymaticaspects. J. Biol. Chem. 233:493-500.

25. RAVEL, J. M., J. S. HUMPHREYS, AND W. SHIVE.1965. Control of glutamine synthesis in Lacto-bacillus arabinosus. Arch. Biochem. Biophys.111:720-726.

26. ROBICHON-SZULMAJSTER, H. DE, AND D. CORRI-vAUX. 1963. Regulations metaboliques de labiosynthesede la methionine et de la threoninechez Saccharomyces cerevisiae. I. Repression etretroinhibition de l'aspartokinase. Biochim.Biophys. Acta 73:248-256.

27. STADTMAN, E. R. 1966. Allosteric regulation ofenzyme activity. Advan. Enzymol. 28:41-154.

28. STADTMAN, E. R., G. N. COHEN, G. LE BRAs, ANDH. ROBICHON-SZULMAJSTER. 1961. Feedbackinhibition and repression of aspartokinaseactivity in Escherichia coli and Saccharomycescervisiae. J. Biol. Chem. 236:2033-2038.

29. WEIss, B., AND P. R. SRINVASON. 1959. The bio-synthesis of p-aminobenzoic acid. Proc. Nat].Acad. Sci. U.S. 45:1491-1494.

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

31. YATs, R. A., AND A. B. PARDEE. 1956. Control ofpyrimidine biosynthesis in Escherichia coli by afeedback mechanism. J. Biol. Chem. 221:757-770.

1055VOL. 93, 1967

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

.org/D

ownloaded from