Vol. 173, No. 17JOURNAL OF BACTERIOLOGY, Sept. 1991, p. 5494-55010021-9193/91/175494-08$02.00/0Copyright © 1991, American Society for Microbiology

Purification and Characterization of Benzoate-Coenzyme A Ligaseand 2-Aminobenzoate-Coenzyme A Ligases from

a Denitrifying Pseudomonas sp.UWE ALTENSCHMIDT, BRIGITTE OSWALD, AND GEORG FUCHS*

Angewandte Mikrobiologie, University of Ulm, P.O. Box 4066,W-7900 Ulm, Germany

Received 22 April 1991/Accepted 25 June 1991

The enzymes catalyzing the formation of coenzyme A (CoA) thioesters of benzoate and 2-aminobenzoate werestudied in a denitrifying Pseudomonas sp. anaerobically grown with these aromatic acids and nitrate as solecarbon and energy sources. Three different rather specific aromatic acyl-CoA ligases, El, E2, and E3, werefound which catalyze the formation of CoA thioesters of benzoate, fluorobenzoates, and 2-aminobenzoate. ATPis cleaved into AMP and pyrophosphate. The enzymes were purified, their N-terminal amino acid sequenceswere determined, and their catalytic and molecular properties were studied. Cells anaerobically grown onbenzoate and nitrate contain one CoA ligase (AMP forming) for benzoic acid (E1). It is a homodimer of Mr120,000 which prefers benzoate as a substrate but shows some activity also with 2-aminobenzoate andfluorobenzoates, although with lower Km. Cells anaerobically grown on 2-aminobenzoate and nitrate containthree different CoA ligases for aromatic acids. The first one is identical with benzoate-CoA ligase (El). Thesecond enzyme is a 2-aminobenzoate-CoA ligase (E2). It is a monomer of Mr 60,000 which prefers 2-amino-benzoate but also activates benzoate, fluorobenzoates and, less effectively, 2-methylbenzoate, with loweraffinities to the latter substrates. The enzymes El and E2 have similar activity levels; a third minor CoA ligaseactivity is due to a different 2-aminobenzoate-CoA ligase. This enzyme (E3) is a monomer ofMr 65,000 whichis identical to an isoenzyme 2-aminobenzoate-CoA ligase, operating in a new plasmid-encoded aerobic2-aminobenzoate pathway (U. Altenschmidt, C. Eckerskorn, and G. Fuchs, Eur. J. Biochem. 194:647-653,1990); apparently, it is not completely repressed under anaerobic conditions and therefore also is induced to asmall extent by 2-aminobenzoate under anaerobic growth conditions.

Aromatic compounds are metabolized by microorganismsby two fundamentally different methods. Under aerobicconditions, aromatic compounds are transformed by mo-nooxygenases and dioxygenases into a few central interme-diates such as catechol, protocatechuate, and gentisate.These compounds are suitable for an oxidative chemicalattack. Accordingly, the aromatic ring structures are cleavedenzymatically by dioxygenases (for a review, see reference11).Under anaerobic conditions, aromatic compounds have to

be transformed by other means than by oxygenases. Figure1 gives an outline of the initial reactions in the anaerobicmetabolism of some aromatic compounds leading to themost important central intermediate, benzoyl-coenzyme A(CoA), as studied in denitrifying Pseudomonas species.These reactions have three functions. The first function is toactivate chemically inert compounds such as phenol ortoluene and others; CoA thioester formation of aromaticacids is one notable form of activation. CoA ligases for thearomatic acids benzoate (10) and 4-hydroxybenzoate (9a)from anaerobically grown Rhodopseudomonas palustris andfor phenylacetate (22) from aerobically grown Pseudomonasputida have been purified before. The second function is toreduce the enormous variety of natural and synthetic aro-matic compounds, channelizing them into a few centralintermediates. The third function is to direct to those inter-mediates compounds such as benzoyl-CoA (rather thanbenzoate), resorcinol, and phloroglucinol which are suitable

* Corresponding author.

for a reductive attack of the aromatic nucleus. In keepingwith this, these central aromatic compounds appear to beattacked enzymatically by reductases, and the resultingalicyclic compounds have been shown or postulated tobecome hydrolytically cleaved (7, 9, 14, 17-19).The present work aimed at studying the initial reactions

and enzymes in the anaerobic metabolism of benzoate and2-aminobenzoate. These aromatic acids are of biologicalimportance and in addition are formed secondarily from avariety of aromatic precursors by microbial activity (13, 16).The bacterium studied, Pseudomonas strain KB 740- (3),was anaerobically grown with nitrate as an electron acceptorand benzoate or 2-aminobenzoate as the sole source of cellcarbon and electrons. Different CoA ligases for aromaticacids (acyl-CoA synthetases) such as benzoate, 2-aminoben-zoate, 4-hydroxybenzoate, and (4-hydroxy)phenylacetatehave been detected in this bacterium when anaerobicallygrown on the respective acids (12, 28, 30, 35). We haverecently disclosed a new aerobic, plasmid-encoded pathwayof 2-aminobenzoate metabolism in the same organism which,unexpectedly, also proceeds via 2-aminobenzoyl-CoA but isunder aerobic control; therefore, it involves a 2-aminoben-zoate-CoA ligase even under aerobic growth conditions (la).

Here, we report on the demonstration and purification ofthree different CoA ligases, those for anaerobic benzoateand 2-aminobenzoate metabolism and for aerobic 2-ami-nobenzoate metabolism. The expressions of these enzymeactivities are controlled by different means. The knowledgeof the N-terminal amino acid sequences of these enzymeswill enable us to study some molecular biological aspects of

5494

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

AROMATIC ACID-COENZYME A LIGASES 5495

,SCoAAMP SCO COOH

ATP PP CH2 H20 CoA r0

COA PyPh >nyl-I ~~~~~glyoxylic 2H ,SCoA

COOH acid ScCOAC02 | Co

k ' t ~~TP AAMP H20OH +PP OH CA

4-OH Benzoic MATP NH3acidH

/SCoA

H0CH20H

enocC-0

acid NH2

HCOOH CoA

H20 2PH

ATP

Anthranilicacid

100,000 x gSupernatant

30 % Ammonium SulfatePrecipitate, Supernatant

Sephadex G-25Chromatography

150 mM KCI

DEAE-CelluloseChromatography

100 mM KCI

HydroxylapatiteChromatography

5 mM HP042- 5-20 mM HP042-H20 0

Toluene

FIG. 1. Outline of reactions and intermediates involved in theinitial steps of the anaerobic metabolism of various aromatic com-

pounds via benzoyl-CoA as central aromatic intermediate. Thereactions and enzymes and their regulation patterns have beenstudied in the denitrifying Pseudomonas strains KB 740- and K 172.They involve CoA thioester formation (references 5 and 35 and thispaper), carboxylations (20, 31), oxidative decarboxylations (5, 30),anaerobic hydroxylations with water acting as the source of oxygen(dehydrogenases) (references lb and 30 and unpublished data), as

well as reductive eliminations of aromatic hydroxyl (12) and possi-bly of amino substituents.

the metabolism of aromatic acids under anaerobic and aer-

obic conditions.

MATERIALS AND METHODSMaterials. Chemicals and medium components were ob-

tained from Merck (Darmstadt, Germany), Roth (Karlsruhe,Germany), Difco (Hamburg, Germany), Fluka (Neu-Ulm,Germany), or Sigma (Heidelberg, Germany). Materials forcolumn chromatography were obtained from Pharmacia(Freiburg, Germany), Fluka, Bio-Rad (Munich, Germany),and Sigma. Fast protein liquid chromatography (FPLC)equipment was from Pharmacia. Biochemicals were ob-tained from Boehringer (Mannheim, Germany), Fluka, Phar-macia, or Sigma. Gases were from Linde (Hollriegelskreut,Germany). Pseudomonas strain KB 740- was a kind giftfrom Konstantin Braun.Growth of bacteria. Pseudomonas strain KB 740- was

grown aerobically or anaerobically (plus nitrate), with ben-zoate or 2-aminobenzoate as sole carbon and energy

sources, essentially as described previously (3). Growthdetermination, growth yield determination, cell harvest, andstorage were described previously (35).

Assays of CoA ligase activity. The determinations of ben-zoate-CoA ligase activity and of 2-aminobenzoate-CoA li-gase activity, as well as the determination of the reactionstoichiometry, were accomplished with spectrophotometricassays as described previously (35).

Determinations. Benzoate and 2-aminobenzoate as well astheir CoA thioesters were separated by high-pressure liquidchromatography (HPLC) (34). 2-Aminobenzoyl-CoA was

characterized by UV and nuclear magnetic resonance spec-troscopy as described previously (4), and benzoyl-CoA was

Q-SepharoseChromatography230-340 mM KCI

Q-SepharoseChromatography125-175 mM KCI

Q-SepharoseChromatography125-175 mM KCI

Reactive Green-Agarose Reactive Green-Agarose Reactive Green-AgaroseChromatography Chromatography Chromatography0.5MKCI 1 MKCI 10mMATP

Benzoyl-CoASynthetase (E1)

2-Aminobenzoyl-CoASynthetase (E2)

2 Aminobenzoyl-CoASynthetaae (E3)

FIG. 2. Scheme of purification protocols for the three CoAligases (AMP forming) El, E2, and E3 acting on the aromatic acidsbenzoate and 2-aminobenzoate.

characterized by UV spectroscopy (32). CoA thioesterswere hydrolyzed (6).

Protein determination. Protein was determined by themethod of Bradford (2), using crystalline bovine serumalbumin as the standard.SDS-PAGE. Sodium dodecyl sulfate-polyacrylamide gel

electrophoresis (SDS-PAGE; 12.5% polyacrylamide) was

carried out at 4°C with the discontinuous buffer system ofLaemmli (21). Polypeptides were visualized by silver stain-ing (24).

Preparation of cell extract. All subsequent procedureswere carried out at 0 to 4°C. About 10 g (wet weight) ofharvested cells was suspended in 20 ml of a 100 mMtris(hydroxymethyl)aminomethane-HCl (Tris-HCl) buffer(pH 7.8) containing 2 mM dithioerythritol (DTE) and 2 mMMgCl2 and passed twice through a French pressure cell(American Instruments Company) at a pressure of 137 MPa.Unbroken cells and cell debris were removed by ultracen-trifugation at 100,000 x g for 1 h. The resulting supematant(about 26 ml) was termed the soluble protein fraction.

Purification of CoA ligases. The purification scheme issummarized in Fig. 2. All steps were carried out at 4°C.

(i) Ammonium sulfate precipitation and desalting with a

Sephadex-G 25 column. The soluble protein fraction was

treated with ammonium sulfate to 30% of saturation andstirred for 30 min. After centrifugation at 12,000 x g for 15min, the supernatant containing CoA ligase activity was

passed through a Sephadex-G 25 column (diameter, 3.2 cm;volume, 230 ml) and equilibrated with Tris-HCl buffer (10

,OOHCH2

Phenyl-acetic acid

OH

Phenol

CH

OH

p-Cresol

VOL. 173, 1991

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

5496 ALTENSCHMIDT ET AL.

mM Tris-HCl [pH 7.8], 2 mM DTE, 2 mM MgCl2) toeliminate ammonium sulfate.

(ii) DEAE-cellulose chromatography. The fractions withligase activity (110 ml) were applied at a flow rate of 2 mlmin-' to a DEAE-Sephacel column (diameter, 3.2 cm;volume, 64 ml) that had been equilibrated with Tris-HClbuffer (10 mM Tris-HCl [pH 7.8], 2 mM DTE, 2 mM MgCl2,50 mM KCI). The column was washed with 150 ml of thesame buffer and eluted with a stepwise KCI gradient (step 1,150 ml of 100 mM KCI; step 2, 150 ml of 150 mM KCI).Fractions of 10 ml were collected. The 2-aminobenzoate-CoA ligase (E2) eluted with 75 mM KCI, and the benzoate-CoA ligase (E1) eluted with 150 mM KCI.

(iii) Hydroxylapatite chromatography. The fractions (100ml) with 2-aminobenzoate-CoA ligase activity were applieddirectly to a hydroxylapatite column (diameter, 2.4 cm;volume, 25 ml) at a flow rate of 1 ml min-1. The column wasequilibrated beforehand with a potassium phosphate buffer(5 mM, pH 7.8) containing 2 mM DTE and 2 mM MgCl2. Thecolumn was washed with 100 ml of the same buffer. One2-aminobenzoate-CoA ligase (E2) was eluted with buffer. Asecond 2-aminobenzoate-CoA ligase (E3) was eluted in a200-ml linear potassium phosphate gradient (5 to 20 mM).Fractions of 4 ml were collected.

(iv) Q-Sepharose chromatography. The fractions contain-ing benzoate-CoA ligase (E1) from the DEAE chromatogra-phy and the two 2-aminobenzoate-CoA ligases (E2 and E3)from the hydroxylapatite chromatography were applied to aQ-Sepharose column (diameter, 1.6 cm; volume, 20 ml) at aflow rate of 3 ml min-'. To elute the benzoate-CoA ligase,the Q-Sepharose column was equilibrated and washed with aTris-HCl buffer (10 mM Tris-HCI [pH 7.8], 2 mM DTE, 2mM MgCl2, 230 mM KCl). The ligase was eluted in a 200-mllinear KCI gradient (230 to 340 mM KCl). To elute both2-aminobenzoate-CoA ligases, the column was equilibratedand washed with a Tris-HCI buffer (10 mM Tris-HCI [pH7.8], 2 mM DTE, 2 mM MgCl2, 125 mM KCl). The enzymeswere eluted in a 200-ml linear KCl gradient (125 to 175 mMKCl). Three-milliliter fractions containing ligase activitywere collected and pooled.

(v) Affinity chromatography. Q-Sepharose-purified CoAligases were applied at a flow rate of 0.5 ml min-' to a"reactive green" cross-linked agarose column (diameter, 4.4cm; volume, 23 ml) equilibrated with a buffer of 10 mMTris-HCl (pH 7.8) containing 2 mM DTE and 2 mM MgCl2.The column was washed with the same buffer. Benzoate-CoA ligase (E1) was eluted with 0.5 M KCI; one of the2-aminobenzoate-CoA ligases (E2) was eluted with 1 M KCI,and the second 2-aminobenzoate-CoA ligase (E3) was elutedwith 10 mM ATP. Fractions of 2 ml were collected.

(vi) Determination of the native molecular weight by gelfiltration. An FPLC "Superdex 200" column (diameter, 1.6cm; volume, 112 ml) was equilibrated with a 10 mM Tris-HClbuffer (pH 7.8) containing 2 mM DTE and 2 mM MgCl2.Fractions (5 ml) containing ligase activity were applied to thecolumn. The column was run at a flow rate of 0.8 ml min-',and 1-ml fractions were assayed for enzyme activity. Thecolumn was calibrated with the following Mr markers: fer-ritin, 440,000; catalase, 232,000; bovine serum albumin,67,000; ovalbumin, 43,000.

Determination of N-terminal amino acid sequence. Fordetermination of the N-terminal amino acid sequence, theCoA ligases were separated on an SDS-polyacrylamide gel.The proteins were stained with Coomassie blue and trans-ferred to a siliconized glass fiber foil (Clossybond; Biometra,Gottingen, Germany), and the protein bands were cut off.

The sequences were determined in an automatized gas-phasesequencer (model 470 A; Applied BioSystems, Weiterstadt,Germany) based on Edman degradation. The phenylthiohy-dantoin derivatives of the released amino acids were deter-mined by C8 reversed-phase HPLC (8).

RESULTS

Enzyme activities in cell extracts. Cell extracts (100,000 xg supernatant) of Pseudomonas strain KB 740- grownanaerobically with benzoate or 2-aminobenzoate plus nitrateas sole carbon and energy sources were tested for ATP-dependent CoA ligase activities acting on the aromatic acidssupplied as the substrates. Cells from both cultures con-tained CoA ligase activity for benzoate and 2-aminoben-zoate. The highest amount of 2-aminobenzoate-CoA ligaseactivity (0.14 ,umol of 2-aminobenzoyl-CoA formed min-'mg of protein-') was obtained from 2-aminobenzoate-growncells growing in the middle of the exponential growth phase(A578 = 0.6); similarly, the highest benzoate-CoA ligaseactivity was in benzoate-grown cells from the logarithmicgrowth phase (0.24 ,umol of benzoyl-CoA formed min-'mg 1). The activity decreased to one-fourth in the stationarygrowth phase (0.03 to 0.06 ,umol min-' mg-'). Hereafter,one enzyme unit (U) refers to 1 ,umol of acyl-CoA formedmin- 1. It has to be taken into account that the benzoate- and2-aminobenzoate-CoA ligase activities were due to threedifferent enzymes (see below). From these data alone, thespecific activity of the individual enzymes in cell extractscannot be estimated.

Purification of benzoate-CoA ligase (E1) and 2-amino-ben-zoate-CoA ligases (E2 and E3). Three different soluble aro-matic acid-CoA ligases were present in cells anaerobicallygrown on 2-aminobenzoate and one was present in cellsanaerobically grown on benzoate. Cells aerobically grownon the aromatic acids also contained CoA ligases for therespective substrates. This will not be followed up here. Theprocedure described above (Fig. 2) resulted in a 640-foldpurification of the benzoate-CoA ligase (E1), <1430-foldpurification of one 2-aminobenzoate-CoA ligase (E2), and>120-fold purification of a second 2-aminobenzoate-CoAligase (E3) when cells anaerobically grown on 2-aminoben-zoate were analyzed (for purification factor, see Discussion)(Table 1). E2 and E3 could only be separated during the latersteps of the purification procedure. The enzyme E3 was alsopurified from cells aerobically grown on 2-aminobenzoate(>550-fold purification); these data and the localization,cloning, and sequencing of the gene will be presentedelsewhere. Fifty-nine percent of total E1, 87% of total E2,and 10% of E3 were recovered. The highest content ofenzyme activity was found in fractions after 30% ammoniumsulfate treatment, because the ligases are inhibited in cellextracts by an unknown component which is lost during thepurification (35). Final specific activities in the eluate fromthe affinity chromatography column ranged from 152 U permg of protein for E1 and 169 U per mg for E2 to 16.5 U permg for E3. During the exploration stage of the purification, anumber of affinity chromatography matrices were tested.Except for reactive green-agarose, none of the matrices wereable to bind or to release the enzymes totally. The enzymeswere found to elute with 1 M KCl (E1), 0.5 M KCl (E2), or 10mM ATP (E3). The kinetic properties of the enzymes aresummarized in Table 2.Nature of the products and stoichiometry of the reactions.

The three enzymes were absolutely dependent on the aro-matic acids, ATP, Mg2+, and CoA. The products of the

J. BACTERIOL.

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

AROMATIC ACID-COENZYME A LIGASES 5497

TABLE 1. Purification protocol for the three CoA ligases E1, E2, and E3 acting on the aromatic acids benzoate and 2-aminobenzoatea

Purification step Vol (ml) Total enzyme Protein concn Total protein Sp act (U/mg Yield of Purificationactivity (U) (mg/ml) (mg) of protein) recovery (%) (fold)

1. Purification of E1100,000 x g supernatant

Elb 26 150 24 624 0.236 60 1E2 + E3c 26 85 24 624 0.137 42 1

Ammonium sulfate precipitateEl 35 250 11.5 402 0.614 100 2.6E2 + E3 37 216 11.5 425 0.505 100 3.7

Sephadex G-25E1 70 242 5.8 406 0.6 97 2.5E2 + E3 70 190 5.8 406 0.47 93 3.4

DEAE-celluloseEl 60 246 1 60 4.1 98 17.4E2 + E3 100 210 0.3 30 7 99 51

Q-Sepharose, E1 50 147 0.43 21.5 7.0 60 30Reactive green-agarose, E1 20 145 0.048 0.96 151.7 59 643

2. Purification of E2Hydroxylapatite, E2 180 160 0.2 36 4.4 78 32Q-Sepharose, E2 20 207 0.27 5.4 42.1 98 307Reactive green-agarose, E2 15 190 0.072 1.08 169.2 87 1,432

3. Purification of E3Hydroxylapatite, E3 30 22 0.1 3 7.2 10.7 53Q-Sepharose, E3 20 22 0.13 2.4 9.1 10.4 66.4Reactive green-agarose, E3 30 21 0.044 1.32 16.5 10 121

a The yield and purification factor for E2 and E3 cannot be given precisely. They refer to E2 plus E3 activity. If it is taken into account that E3 activity wasapproximately 10%o of E2 activity, the values for E3 will be up to 10 times higher and the values for E2 will be 210% lower. Cells anaerobically grown on2-aminobenzoate were used for purification of the enzymes. Similar specific activities of the purified proteins E1 and E3 were obtained when cells anaerobicallygrown on benzoate (E1) or aerobically grown on 2-aminobenzoate (E3) were used.

b E1 was measured with benzoate as the substrate by using the spectrophotometric assay described in Materials and Methods.c E2 and E3 were measured with 2-aminobenzoate as the substrate by using the spectrophotometric assay described in Materials and Methods.

reactions were the CoA thioesters of the respective acidsand were characterized as follows: by UV spectroscopiccharacterization after HPLC purification, by nuclear mag-netic resonance spectroscopy of 2-aminobenzoyl-CoA, bycochromatography with authentic benzoyl-CoA and 2-ami-nobenzoyl-CoA on HPLC, and by detection of the free acidsbenzoate and 2-aminobenzoate by HPLC upon alkalinehydrolysis of the assay mixture after the assay had run tocompletion. The product ofATP hydrolysis was AMP ratherthan ADP in all cases. One mole ofAMP was formed per molof added ATP when the cosubstrates benzoate and CoA andthe cocatalyst Mg2+ were present in excess; 1 mol of AMPwas formed per mol of added benzoate when ATP, Mg2+,and CoA were present in excess (Fig. 3).

Substrate specificity. All three enzymes acted on benzoate,2-aminobenzoate, and the three fluorobenzoate isomers; thespecific rates with the different aromatic acids differed onlyby a factor of two to four. However, the differences in theirsubstrate affinities, as indicated by the apparent Km values(Table 2) (see below) were more pronounced. The kinetic

and regulatory properties suggest that E1 is a benzoate-CoAligase and E2 and E3 are 2-aminobenzoate-CoA ligases. Allthree ligases used ATP preferentially; no other nucleotidetriphosphate (UTP, GTP) was used. ADP and AMP were

completely ineffective.Acetate, cyclohexanecarboxylate, and a selection of aro-

matic acids were tested as potential substrates (Table 3).Benzoate-CoA ligase (E1) activated only benzoate and ana-

logs with close steric resemblances, such as 2-fluoroben-zoate and 4-fluorobenzoate. With 3-fluorobenzoate and2-aminobenzoate, even less activity was observed. The2-aminobenzoate-CoA ligases (E2 and E3) had very similarsubstrate specificities. Benzoate, 2-fluorobenzoate, and4-fluorobenzoate were activated with the same reaction rateas was 2-aminobenzoate. With 3-fluorobenzoate and 2-me-thylbenzoate, there was half and less than one-fifth, respec-tively, of the activity found with 2-aminobenzoate. The othersubstrates were not activated (<1%).

Physicochemical properties of the ligases. In SDS-poly-acrylamide gels (Fig. 4), the purified enzymes migrated as

TABLE 2. Catalytic properties of the CoA ligases El, E2, and E3 acting on the aromatic acids benzoate and 2-aminobenzoatea

CoA ligaseTurnover numberb K__ __M_ pH optimum

(substrate) Benzoate 2-Aminobenzoate 2-Fluorobenzoate ATP CoA

El 26,000 (benzoate) 11 250 100 130C 1ooc 9.3CE2 16,000 (2-aminobenzoate) 35 13 50 83d 20d 8.5-9.2dE3 1,300 (2-aminobenzoate) 75 15 70 40d 26d 8.6d

a The standard assay described in Materials and Methods was used.bk,e in min-' at 28°C.

c Determined with benzoate as cosubstrate.d Determined with 2-aminobenzoate as cosubstrate.

VOL. 173, 1991

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

5498 ALTENSCHMIDT ET AL.

z

-6 100 SE

50-

00 E 50 75 00

pmoL Benzoate

FIG. 3. Stoichiometry of benzoate-CoA ligase reaction. Theformation ofAMP was indirectly measured by coupling the reactionto myokinase, pyruvate kinase, and lactate dehydrogenase reaction.

single bands. The calculated molecular weights were 55,000for E1, 60,000 for E2, and 65,000 for E3. The apparentmolecular weights (Fig. 5) determined by gel filtration chro-matography were 120,000 for E1, 60,000 for E2, and 65,000for E3. Hence, E1 is a homodimer, whereas E2 and E3 areactive as monomers. The spectra of the purified enzymesexhibited no significant absorption above 300 nm.

N-terminal amino acid sequence of the enzymes and differ-ential expression of the aromatic acid-CoA ligase activities.The N-terminal amino acid sequences of the enzymes weredetermined; they are shown in Table 4. The N-terminalamino acid sequence of benzoate-CoA ligase (E1) was iden-tical irrespective of whether the protein was obtained fromcells anaerobically grown on benzoate or 2-aminobenzoate.

TABLE 3. Substrate specificity of the three CoA ligases E1,E2, and E3 acting on the aromatic acids benzoate

and 2-aminobenzoatea

Relative activity of:Substrate"

El E2 E3

Benzoate 100 100 912-Aminobenzoate 28 100 100Cyclohexanecarboxylate -2 <3 c<3-Aminobenzoate <3 c4 cS2-Fluorobenzoate 120 110 1003-Fluorobenzoate 76 56 504-Fluorobenzoate 100 100 902-Methylbenzoate <1 20 9

a The substrate specificity was determined with 0.5 mM aromatic acid and1 mM ATP, respectively.

b None of the three enzymes showed activity with the following organicacids: acetate, phenylacetate, 4-aminobenzoate, 4-methylbenzoate, 2-hydrox-ybenzoate, 3-hydroxybenzoate, 4-hydroxybenzoate, 2-chlorobenzoate, and4-chlorobenzoate. The detection limit of the activity assay was 51% of therate observed with the physiological substrate.

c The values for El are given as percentages of the values with benzoate asthe substrate, and the values for E2 and E3 are given as percentages of thevalues with 2-aminobenzoate as the substrate. The approximate specificactivities of the enzymes El, E2, and E3 at 28°C were 150 ,umol of benzoyl-CoA formed per min per mg of protein, 160 ,umol of 2-aminobenzoyl-CoAformed per min per mg of protein, and 16 p.mol of 2-aminobenzoyl-CoAformed per min per mg of protein, respectively.

1 2 3 4 5 6 7 8

94- 67o 43x

302 20.1

14L

1. ...

.m-

FIG. 4. SDS-PAGE analysis of fractions obtained during thepurification of benzoate-CoA ligase E1 and 2-aminobenzoate-CoAligases E2 and E3. E1 was purified from cells grown anaerobicallywith benzoate; the same results can be obtained with cells grownanaerobically with 2-aminobenzoate. E2 and E3 were purified fromcells grown anaerobically with 2-aminobenzoate; much higheramounts of E3 are obtained if E3 is purified from cells grownaerobically with 2-aminobenzoate. Lane 1, molecular mass stan-dards: phosphorylase b (94 kDa), bovine serum albumin (67 kDa),ovalbumin (43 kDa), carbonic anhydrase (30 kDa), trypsin inhibitor(20.1 kDa), and alpha-lactalbumin (14.4 kDa). Lane 2, cell extract(from cells grown anaerobically with 2-aminobenzoate). Lanes 3 to5, Q-Sepharose fractions of El, E2, and E3, respectively. Lanes 6 to8, reactive green-agarose fractions of E1, E2, and E3, respectively.

It differed from the sequences of the two 2-aminobenzoate-CoA ligases (E2 and E3) which, surprisingly, appear to bealmost identical, although the two proteins migrated atslightly different rates on the denaturing SDS-polyacryl-amide gel. In addition, the expressions of E2 and E3 appar-ently were differently regulated in that the level of E2 waslow or not detectable under aerobic conditions, whereas E3was expressed at an approximately 10-fold-higher levelaerobically than anaerobically.pH dependence. The rates of benzoyl-CoA and 2-ami-

nobenzoyl-CoA formation at different pH values of Tris-HClbuffer were measured. The benzoate-CoA ligase (E1) had abroad pH optimum between 8.5 and 9.2, with half-maximalactivity at pH 7.5 and 9.9. pH optima of 9.3 for E2, withhalf-maximal activity at pH 8.1 and 9.9, and of 8.5 for E3,with-half maximal activity at pH 7.4 and 9.3, were observed.

Kinetics. The values of the Michaelis-Menten saturationkinetics and the turnover rates for benzoate, 2-aminoben-zoate, ATP, and CoASH were obtained (Table 2). Theapparent kinetic constants for the purified enzymes weredetermined from linear Lineweaver-Burk plots. The appar-ent Km values were as follows: for the physiological aro-matic acids, 10 to 15 ,uM; for ATP, 40 to 130 ,uM; for CoA,20 to 100 ,uM. Benzoate and 2-aminobenzoate were thearomatic acids with the lowest apparent Kms for E1 and forE2 and E3, respectively. The turnover numbers of E1 and E2were similar (in the order of 104 min-1), whereas E3 had akcat in the order of 103 min-1.

Stability and activity inhibition. The purified enzymes werenot sensitive to oxygen. The enzymes could be stored frozenin liquid nitrogen for 3 months without remarkable loss ofactivity. When stored at 4°C at a protein concentration of 50,ug ml-' in Tris-HCl buffer, pH 7.8, half of the activity waslost in 4 days. Several agents that react with thiol groups[5,5'-dithiobis-(2-nitrobenzoic acid) and p-chloromer-curibenzoic acid] were tested at a concentration of 1 mM.p-Chloromercuribenzoic acid inhibited the benzoate-CoA

J. BACTERIOL.

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

AROMATIC ACID-COENZYME A LIGASES 5499

C4

Elution volume (ml)

50-

IF

xI10.

B.

a

b \ E1

E3E2

40 50 60 70

Elution volume (ml)

FIG. 5. Determination of molecular masses of native CoA ligasesE1, E2, and E3 by gel filtration chromatography on FPLC Superdex200. (A) Elution profiles of E1 (solid line), E2 (dots), and E3 (dashes).(B) Calibration curve with molecular mass standards as follows: a,ferritin (440 kDa); b, catalase (232 kDa); c, bovine serum albumin(67 kDa); d, ovalbumin (43 kDa).

ligase (E1) and the 2-aminobenzoate-CoA ligase (E3) virtu-ally completely but had no effect on 2-aminobenzoate-CoAligase (E2). E1 and E2 were totally inhibited by 5,5'-dithiobis-(2-nitrobenzoic acid), whereas E3 was not inhibited. Theseresults might suggest that the presence of SH groups whichare accessible to different degrees to these inhibitors areessential for the catalytic activity.

Inhibition effects of several univalent (1 mM) cations (K+,Na+, Li', and Rb+) on the ligase activities were notdetectable. Some divalent cations (1 mM) (Zn2+ and Cu2+)did cause strong inhibition; Cu2+ totally inhibited all en-zymes, probably because of reaction with SH groups. Zn2+inhibited E1 strongly, whereas E2 and E were less sensitive.No inhibition was observed with Mn3+ ions. Mg2+ ions

TABLE 4. N-terminal amino acid sequences of the CoA ligasesE1, E2, and E3 acting on benzoate and 2-aminobenzoate

CoA ligase N-terminal amino acid sequence

E1 Ala Glu Leu Ser Val Ala Asp (His) (Ser) Val (X)Pro Pro

E2 Thr Ser His Val Asp Thr Phe Ala (Arg) Asp (Arg)(X)' Pro Pro (Thr) Glu Gln Gln (Thr) Glu (Ser)Leu

E3 Thr Ser His Val Asp Thr Phe Ala Arg Asp (X)Leu Pro Pro (X) (Glu) Gln Gln

a This residue appears not to be Leu.

could be fully replaced by Mn2" ions (5 mM) in the test.Detergents (1 mM) also affected the ligase activity. SDSinactivated the enzyme. Other molecules such as Tween 100and N-octylglucoside inhibited the activities to a much lesserextent.

DISCUSSION

In the denitrifying Pseudomonas strain KB 740- and instrain K 172 (5), at least six different aromatic acid-CoAligases are present (30, 35), three of which have been purifiedin this investigation. E1 is induced anaerobically, probablyby benzoate; E2 is induced anaerobically by 2-aminoben-zoate; and E3 is induced aerobically by 2-aminobenzoate.Enzymes which play a role in anaerobic metabolism arebenzoate-CoA ligase E1 (previously referred to as synthetase1 [35]), 2-aminobenzoate-CoA ligase E2 (previously referredto as synthetase 2 [35]), 4-hydroxybenzoate-CoA ligase, andphenylacetate-CoA ligase. In aerobic cells, a benzoate-CoAligase of unknown function which may be slightly differentfrom E1 (unpublished results) and a 2-aminobenzoate-CoAligase (la) were found. E3 is identical with the aerobic2-aminobenzoate-CoA ligase isoenzyme and is probablycoded on an 8.1-kbp plasmid which also carries the gene for2-aminobenzoyl-CoA monooxygenase/reductase, a key en-zyme of this aerobic pathway (1, la). The total E3 synthetaseactivity of cells grown anaerobically with 2-aminobenzoatewas only 10% of the total 2-aminobenzoate-CoA ligaseactivity of cells grown aerobically with 2-aminobenzoate.This shows that the two 2-aminobenzoate-CoA ligases E2and E3 are differently regulated. E2 and E3 differ also withrespect to other properties; e.g., they exhibit different chro-matographic behavior, turnover numbers, and molecularweights. In spite of these differences, E2 and E3 have similarKm values, pH optima, and substrate specificities, and theN-terminal amino acid sequences were identical except forone amino acid position. Since these two enzymes could notbe differentiated by simple activity measurement, the yieldand purification factors given are only approximate values.The relation of the two enzymes to each other and thelocalization of the two genes remain to be studied.

In several bacteria able to be grown on aromatic com-pounds in the absence of molecular oxygen, similar enzymeswere reported; they were proposed to be involved in theanaerobic metabolism of the following compounds (in pa-rentheses): benzoate-CoA ligase (benzoate, benzaldehyde,benzyl alcohol, and toluene) (lb, 5, 10, 15, 17, 25, 28,35), 4-hydroxybenzoate-CoA ligase (phenol, p-cresol, and4-hydroxybenzoate) (5, 9a, 12, 15, 23, 27, 35), 3-methylben-zoate-CoA ligase (3-methylbenzoate) (27), 4-hydroxy-3-me-thylbenzoate-CoA ligase (4-hydroxy-3-methylbenzoate, o-cresol, and 2,4-dimethylphenol) (27), 4-aminobenzoate-CoAligase (aniline and 4-aminobenzoate) (29), phenylacetate-CoA ligase (phenylacetate and 4-hydroxyphenylacetate) (5,30), phthalate-CoA ligase (o-, m-, and p-phthalate) (25), and2-hydroxybenzoate-CoA ligase (salicylate) (15, 25). The listis certainly not complete. Of these enzymes, only benzoate-CoA ligase and 4-hydroxybenzoate-CoA ligase from R.palustris (9a, 10) have been purified. Furthermore, phenyl-propionic acids and compounds transformed to C6-C3 com-pounds (33) may be metabolized via their CoA thioestersthrough reactions of beta-oxidation. This may be true forboth anaerobic and aerobic conditions.Under aerobic conditions, the metabolism of 2-aminoben-

zoate (la) and 4-chlorobenzoate (24a) was shown to proceedvia CoA thioesters. The role of phenylacetate-CoA ligase

VOL. 173, 1991

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

5500 ALTENSCHMIDT ET AL.

aerobically expressed in a P. putida strain is unknown (22);it has been hypothesized that it is involved in a novel aerobicphenylacetate metabolism, although its kinetic constants arenot in favor of such a function.The three enzymes reported here have properties similar

but not identical to those of the benzoate-CoA ligase and4-hydroxybenzoate-CoA ligase from R. palustris first iso-lated by Gibson and coworkers; therefore, many aspects ofthese enzymes have been discussed previously (10). Theenzymes belong to a group of carboxylate-CoASH ligases(AMP forming) (EC 6.2.1.-) with a relatively high specificityfor aromatic substrates and little or no activity towardnonaromatic acids. With respect to high nucleotide speci-ficity, E1, E2, and E3 are comparable to most other investi-gated CoA ligases, but not to phenylacetate-CoA ligase of P.putida, which reacts weakly also with ADP, CTP, or UDP(22), or acetate-CoA ligase of Bradyrhizobium japonicum,which was reported also to activate acetic acid with ADPand dATP (26).The benzoate-CoA ligase E1 activates only a small number

of structurally related aromatic acids. Significant activitieswere observed only with benzoate, 2-fluorobenzoate, and4-fluorobenzoate; little activity was found with 3-fluoroben-zoate and 2-aminobenzoate. The 2-aminobenzoate-CoA li-gases E2 and E3 activated 2-aminobenzoate, benzoate, 2-flu-orobenzoate, and 4-fluorobenzoate; with 3-fluorobenzoateand 2-methylbenzoate, little activity was observed. Of theenzyme substrates tested, benzoate, 2-aminobenzoate, 2-flu-orobenzoate, 3-hydroxybenzoate, 4-hydroxybenzoate, phe-nylacetate, cyclohexanecarboxylate, and acetate have beenreported to be growth substrates (3, 28, 34). It is likely thatmost if not all of these organic acids are metabolized via theirCoA thioesters. This again suggests that several other syn-thetases exist which may be induced by their specific sub-strates.The apparent molecular mass of the benzoate-CoA ligase

E1 as determined by gel filtration is 120 kDa; the enzyme isa homodimer with an estimated subunit mass of 55 kDa. Thisis approximately the molecular mass of the following mono-meric enzymes: benzoate-CoA ligase (60 kDa) ofR. palustris(10), the 2-aminobenzoate-CoA ligases E2 (60 kDa) and E3(65 kDa), and phenylacetate-CoA ligase in Pseudomonasstrain KB 740- (61 kDa) (22). The homodimeric acetate-CoAligase of B. japonicum (26) has a molecular mass of 150 kDa.The pH optima of all three CoA ligases range from pH 8.5

to 9.2, which are the same as the values reported for otherinvestigated CoA ligases (10). This may reflect the require-ment for the CoA thiolate anion (pK 8) as a nucleophile incatalysis. The enzymes were inhibited by agents that reactwith thiol groups. This may suggest that SH or thiolategroups are essential for the catalytic activity, e.g., in theformation of an S-acylated enzyme intermediate after theadenylic acid has been transferred to the aromatic acid.

ACKNOWLEDGMENTS

This investigation was financially supported by the DeutscheForschungsgemeinschaft and the Fonds der Chemischen Industrie.U.A. was supported by a graduate fellowship of the Land Baden-Wuerttemberg.Thanks are due to C. Eckerskorn, MPI fuer Biochemie, Martin-

sried, for determination of the N-terminal amino acid sequences, toHarry Bundschuh for technical assistance, and to Ingrid Koenig forsecretarial work.

REFERENCES1. Altenschmidt, U., M. Bokranz, and G. Fuchs. Unpublished data.la.Altenschmidt, U., C. Eckerskorn, and G. Fuchs. 1990. Evidence

that enzymes of a novel aerobic 2-aminobenzoate metabolism indenitrifying Pseudomonas are coded on a small plasmid. Eur. J.Biochem. 194:647-653.

lb.Altenschmidt, U., and G. Fuchs. Anaerobic degradation oftoluene in dentitrifying Pseudomonas sp.: indication for tolu-ene methylhydroxylation and benzoyl-CoA as central aromaticintermediate. Arch. Microbiol., in press.

2. Bradford, M. M. 1976. A rapid and sensitive method for thequantitation of microgram quantities of protein utilizing theprinciple of protein-dye binding. Anal. Biochem. 72:248-254.

3. Braun, K., and D. T. Gibson. 1984. Anaerobic degradation of2-aminobenzoate (anthranilic acid) by denitrifying bacteria.Appl. Environ. Microbiol. 48:102-107.

4. Buder, R., K. Ziegler, G. Fuchs, B. Langkau, and S. Ghisla.1989. 2-Aminobenzoyl-CoA monooxygenase/reductase, a noveltype of flavoenzyme. Studies on the stoichiometry and thecourse of the reaction. Eur. J. Biochem. 185:637-643.

5. Dangel, W., R. Brackmann, A. Lack, M. Mohamed, J. Koch, B.Oswald, B. Seyfried, A. Tschech, and G. Fuchs. 1991. Differen-tial expression of enzyme activities initiating anoxic metabolismof various aromatic compounds via benzoyl-CoA. Arch. Micro-biol. 155:256-262.

6. Decker, K. 1959. Die aktivierte Essigsaure. Enke Verlag, Stut-tgart, Germany.

7. Dutton, P. L., and W. C. Evans. 1969. The metabolism ofaromatic compounds by Rhodopseudomonas palustris: a newreductive method of aromatic ring metabolism. Biochem. J.113:525-536.

8. Eckerskorn, C., W. Mewes, H. Goretzki, and F. Lottspeich.1988. A new siliconized-glass fiber as support for protein-chemical analysis of electroblotted proteins. Eur. J. Biochem.176:509-519.

9. Evans, W. C. 1977. Biochemistry of the bacterial catabolism ofaromatic compounds in anaerobic environments. Nature (Lon-don) 270:17-22.

9a.Fogg, G. C., and J. Gibson. 1990. Abstr. Annu. Meet. Am. Soc.Microbiol. 1990, K137, p. 242.

10. Geissler, J. F., C. S. Harwood, and J. Gibson. 1988. Purificationand properties of benzoate-coenzyme A ligase, a Rhodopseudo-monas palustris enzyme involved in the anaerobic degradationof benzoate. J. Bacteriol. 170:1709-1714.

11. Gibson, I. T., and V. Subramiaian. 1984. Microbial degrada-tion of aromatic hydrocarbons, p. 181-252. In D. T. Gibson(ed.), Microbial degradation of organic compounds. MarcelDekker, Inc., New York.

12. Glockler, R., A. Tschech, and G. Fuchs. 1989. Reductive dehy-droxylation of 4-hydroxybenzoyl-CoA to benzoyl-CoA in adenitrifying, phenol degrading Pseudomonas species. FEBSLett. 251:237-240.

13. Gross, G. G. 1985. Biosynthesis and metabolism of phenolicacids and monolignols, p. 229-271. In T. Higuchi (ed.), Biosyn-thesis and biodegradation of wood components. AcademicPress, Inc., New York.

14. Haddock, J. D., and J. G. Ferry. 1989. Purification and proper-ties of phloroglucinol reductase from Eubacterium oxi-doreducens G-41. J. Biol. Chem. 264:4423-4427.

15. Harwood, C. S., and J. Gibson. 1986. Uptake of benzoate byRhodopseudomonas palustris grown anaerobically in light. J.Bacteriol. 165:504-509.

16. Higuchi, T. 1985. Biosynthesis of lignin, p. 141-160. In T.Higuchi (ed.), Biosynthesis and biodegradation of wood com-ponents. Academic Press, Inc., New York.

17. Hutber, G. N., and D. W. Ribbons. 1983. Involvement ofcoenzyme A esters in metabolism of benzoate and cyclohexan-ecarboxylate by Rhodopseudomonas palustris. J. Gen. Micro-biol. 129:2413-2420.

18. Kluge, C., A. Tschech, and G. Fuchs. 1990. Anaerobic metabo-lism of resorcylic acids (m-dihydroxybenzoic acids) and resor-cinol (1,3-benzenediol) in a fermenting and in a denitrifyingbacterium. Arch. Microbiol. 155:68-74.

J. BACTERIOL.

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

AROMATIC ACID-COENZYME A LIGASES 5501

19. Krumholz, L. R., R. L. Crawford, M. E. Hemnlig, and M. P.Bryant. 1987. Metabolism of gallate and phloroglucinol in Eu-bacterium oxidoreducens via 3-hydroxy-5-oxohexanoate. J.Bacteriol. 169:1886-1890.

20. Lack, A., I. Tommasi, M. Aresta, and G. Fuchs. 1990. Catalyticproperties of "phenol carboxylase": in vitro study of C02:4-hydroxybenzoate isotope exchange reaction. Eur. J. Biochem.197:473-479.

21. Laemmli, U. K. 1970. Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London)227:680-685.

22. Martinez-Blanco, H., A. Reglero, L. B. Rodriguez-Aparicio, andJ. M. Luengo. 1990. Purification and biochemical characteriza-tion of phenylacetyl-CoA ligase from Pseudomonas putida. Aspecific enzyme for the catabolism of phenylacetate acid. J.Biol. Chem. 265:7084-7090.

23. Merkel, S. M., A. E. Eberhard, J. Gibson, and C. S. Harwood.1989. Coenzyme A thioesters are involved in the anaerobicmetabolism of 4-hydroxybenzoate by Rhodopseudomonaspalustris. J. Bacteriol. 171:1-7.

24. Morrissey, J. H. 1981. Silver stain for proteins in polyacryl-amide gels: a modified procedure with enhanced uniform sensi-tivity. Anal. Biochem. 117:307-310.

24a.Muiller, R. (Hamburg, Germany). Personal communication.25. Nozawa, T., and Y. Maruyama. 1988. Anaerobic metabolism of

phthalate and other aromatic compounds by a denitrifyingbacterium. J. Bacteriol. 170:5778-5784.

26. Preston, G. G., J. D. Wall, and D. W. Emerich. 1990. Purifica-tion and properties of acetyl-CoA synthetase from Bradyrhizo-bium japonicum bacteroids. Biochem. J. 267:179-183.

27. Rudolphi, A., A. Tschech, and G. Fuchs. 1991. Anaerobicdegradation of cresols by denitrifying bacteria. Arch. Microbiol.155:238-248.

28. Schennen, U. K., K. Braun, and H.-J. Knackmuss. 1985. Anaer-obic degradation of 2-fluorobenzoate by benzoate-degradingdenitrifying bacteria. J. Bacteriol. 161:321-325.

29. SchneUl, S., and B. Schink. 1991. Anaerobic aniline degradationvia reductive deamination of 4-aminobenzoyl-CoA in Desulfo-bacterium anilini. Arch. Microbiol. 155:183-190.

30. Seyfried, B., M. Mohamed, A. Tschech, and G. Fuchs. Submit-ted for publication.

31. Tschech, A., and G. Fuchs. 1987. Anaerobic degradation ofphenol by pure cultures of newly isolated denitrifying Pseudo-monas. Arch. Microbiol. 148:213-217.

32. Webster, L. T., J. J. Mieyal, and U. A. Siddiqui. 1974. Benzoyland hydroxybenzoyl esters of coenzyme A. Ultraviolet charac-terization and reaction mechanisms. J. Biol. Chem. 249:2641-2645.

33. Zenk, M. H., B. Ulbrich, J. Brusse, and J. Stockigt. 1980.Procedure for the enzymatic synthesis and isolation of cin-namoyl-CoA thioesters using a bacterial system. Anal. Bio-chem. 101:182-187.

34. Ziegler, K., K. Braun, A. B6ckler, and G. Fuchs. 1987. Studieson the anaerobic degradation of benzoic acid and 2-aminoben-zoic acid by a denitrifying Pseudomonas strain. Arch. Micro-biol. 149:62-69.

35. Ziegler, K., R. Buder, J. Winter, and G. Fuchs. 1989. Activationof aromatic acids and aerobic 2-aminobenzoate metabolism in adenitrifying Pseudomonas strain. Arch. Microbiol. 151:171-176.

VOL. 173, 1991

on January 16, 2020 by guesthttp://jb.asm

.org/D

ownloaded from