INFECTION AND IMMUNITY, OCt. 1976, p. 1077-1086Copyright C 1976 American Society for Microbiology

Vol. 14, No. 4Printed in U.S.A.

Large-Scale Purification and Characterization of theExotoxin ofPseudomonas aeruginosa

STEPHEN H. LEPPLAPathology Division, United States Army Medical Research Institute of Infectious Diseases, Fort Detrick,

Frederick, Maryland 21701

Received for publication 26 April 1976

The exotoxin (PE) of Pseudomonas aeruginosa was purified from 50-litercultures by a simple three-step procedure, yielding 135 mg of essentially homo-geneous protein. In Ouchterlony gel diffusion, PE produces a single line whichdoes not interact with a diphtheria toxin-antitoxin precipitin line. The proteinhas a molecular weight of 66,000, an isoelectric point of 5.1, N-terminal arginine,and four disulfide bridges. The amino acid composition shows no apparentsimilarity to that of diphtheria toxin. The median lethal dose of this PEpreparation in mice weighing 20 g is 0.1 ,g. The median lethal dose in 350-g ratsis 20 ,ug. The cytotoxicity of PE for mouse L929 fibroblasts is completelyneutralized by small amounts of specific pony antitoxin. The exotoxin possessesadenosine diphosphate-ribosylation activity. Both cytotoxic and adenosinediphosphate-ribosylation activities are shown to be properties of the intact66,000-dalton protein.

Infection by Pseudomonas aeruginosa is afrequent and serious complication in debili-tated patients, particularly those whose phago-cytic functions are impaired (29). The cause(s)of the special virulence of Pseudomonas inthese situations is unknown. However, one pos-sible source of the pathogenicity of this bacte-rium is a potent Pseudomonas exotoxin (PE)described by Liu (14). This protein, which wasnamed exotoxin A, was first recognized by itslethality for mice. Subsequent studies showedPE to be cytopathic for a number of culturedcell lines and to inhibit amino acid incorpora-tion and ribonucleic acid synthesis (26). Re-cently, Iglewski and Kabat (9) presented evi-dence that PE, like diphtheria exotoxin (DE),catalyzes transfer of the adenosine diphosphate(ADP) ribose portion of nicotinamide adeninedinucleotide to mammalian elongation factor 2(EF-2). The "ADP-ribosylation" (4) of EF-2 ren-ders it unable to cause translocation of ribo-somes along messenger ribonucleic acid duringprotein synthesis.A study of the role of the exotoxin in infec-

tions showed that passive immunization withpony antitoxin protected mice from a lethalchallenge with P. aeruginosa (16). This result,and the analogy to diphtheria, suggests thatPE may be an important virulence factor andthat passive or active immunization against PEmight aid in control of human infections.Further studies of the role of this exotoxin in

pathogenesis would be aided by the availability

of substantial amounts of purified PE andgreater knowledge of its chemical properties.Methods have been described for productionand purification ofPE, and a few of its chemicalproperties were reported (3, 17). However, onlymodest amounts of protein were obtained, andlittle evidence of the purity of the preparationswas provided. Reported here are an improvedmethod for the large-scale purification of PEand some properties of the purified exotoxin.

MATERIALS AND METHODSBacterial strain. P. aeruginosa strain PA103 was

obtained from P. V. Liu. This strain produces verylow levels of extracellular protease.

Biochemicals. Diphtheria toxin and diphtheriaantitoxin were gifts of A. M. Pappenheimer, Jr.Antitoxin to purified PE was prepared in a pony,using the immunization schedule of Liu and Hsieh(16). Nicotinamide adenine dinucleotide labeleduniformly with 14C in the adenosine moiety at aspecific activity of271 MCi/,umol and [1-_4C]iodoacet-amide, 60 ACi/,Amol, were purchased from Amer-sham/Searle (Arlington Heights, Ill).

Fermentation conditions. P. aeruginosa wasgrown using minor modifications of the culture con-ditions of Liu (15). The medium was the dialyzablefraction of Trypticase soy broth (Baltimore Biologi-cal Laboratories, Cockeysville, Md.) supplementedby addition of 50% (wt/vol) sterile glycerol and 2.0 Mmonosodium glutamate (filter sterilized) to finalconcentrations of 1% and 0.05 M, respectively. Aninoculum of 100 ml was grown to late log phase inshake flasks and added to 50 liters of medium in a70-liter fermenter (Fermentation Design, New

1077

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

1078 LEPPLA

Brunswick, N.J.). The culture was aerated at 24liters/min, stirred at 400 rpm, and maintained at32°C. Foaming was controlled by automatic additionof Antifoam 60 (Hartwick, Trenton, N.J.). The pH,which was not regulated, rose from an initial valueof 7.4 to a final value of 8.2. After 18 h of growth, theculture was cooled to 5 to 10°C; bacteria were re-moved by centrifugation with a continuous-flow ro-tor (Lourdes, Old Bethpage, N.Y.).

Exotoxin purification. All purification steps wereperformed at 5°C. The bacteria-free culture superna-tant was diluted to 200 liters with deionized water todecrease the ionic strength. Two liters of diethyl-aminoethyl (DEAE)-cellulose (Cellex-D, 0.9 meq/g,Bio-Rad Laboratories, Richmond, Calif.) was added,and the suspension was stirred gently for 1 h. TheDEAE-cellulose was then allowed to settle, and thesupernatant was discarded. The DEAE-cellulosewas transferred to a 4.8- by 120-cm column, whichwas eluted with 3-liter portions of 0.01, 0.05, and0.10 M NaCl in buffer A [0.01 M tris(hydroxy-methyl)aminomethane (Tris)-hydrochloride, pH8.1]. Exotoxin was precipitated from the 0.10 MNaCl eluate by addition of solid (NH4)2SO4 to 70%saturation. The precipitate was redissolved anddialyzed against buffer A and then applied to a 1.6-by 35-cm (70 ml) column of DEAE-cellulose (What-man DE-52, Reeve Angel, Clifton, N.J.) equili-brated in buffer A. The column was eluted by a

linear gradient of NaCl, 0.01 to 0.33 M, in buffer A(total volume, 1,500 ml). The exotoxin-containingfractions, emerging at 0.12 to 0.14 M NaCl, were

adjusted to pH 7.2 with 1.0 M HCl and pumpeddirectly onto a 1.6- by 35-cm (70 ml) column of hy-droxylapatite (Bio-Rad Laboratories) equilibratedin 0.005 M sodium phosphate, 0.05 M NaCl, pH 7.0.The column was eluted by a linear gradient of so-

dium phosphate, 0.005 to 0.10 M in 0.05 M NaClpH 7.0 (total volume, 1,000 ml). The protein peakemerging at 0.04 to 0.06 M sodium phosphate was

concentrated by precipitation with ammonium sul-fate, dialyzed against buffer A, and frozen at -700Cin small portions.

Analytical methods. Protein was measured by a

procedure in which samples are precipitated on pa-

per strips and stained with xylene brilliant cyanin G(2). The E21% of PE was determined by measuringnitrogen with a micro-Kjeldahl technique and as-

suming a nitrogen content of 16%. Neutral sugars

were measured with the anthrone reagent (27).Analyses of carbohydrate content by gas chromatog-raphy were made by using a modification of themethod of Bhatti et al. (1). Amino acid analyseswere performed by a combination of methods cur-

rently in use in this laboratory (11). N-terminalamino acids were determined by a dansyl chloridemethod (30). Sulfhydryl groups and disulfides were

measured with [14C]iodoacetamide, using modifica-tions of the method of Steinert (28). The['4C]iodoacetamide was diluted approximately 35-fold with nonradioactive iodoacetamide, and thespecific activity was determined by alkylation ofreduced ribonuclease to be 2.0 x 106 cpm/,umol. Fordisulfide determinations, the proteins were first re-

duced by incubation for 2 h at 1 to 2 mg/ml in buffer

B (9 M urea, 0.05 M Tris-hydrochloride, 0.5 mMethylenediaminetetraacetic acid [EDTA], pH 8.3)containing 3 mM dithiothreitol. The solutions werethen diluted to 200 yg/ml with buffer B containing 1mM dithiothreitol; ['4C]iodoacetamide was added toa concentration of 2 mM. Samples of 50 gl wereremoved at intervals, mixed with 20 ,l of 0.10 Mcysteine hydrochloride, and transferred to paperdisks (Schleicher and Schuell no. 740-E, Keene,N.H.), which were washed as described by Mans andNovelli (20). Experiments using higher concentra-tions of dithiothreitol or iodoacetamide showed thatthe conditions described here led to complete reduc-tion and alkylation of ribonuclease and PE. Radio-activity of the disks was measured by liquid scintil-lation counting in 2.0 ml of Liquifluor-toluene (NewEngland Nuclear Corp., Boston, Mass.). Radial im-munodiffusion according- to Mancini et al. (19) wasperformed in 1.0-mm-thick layers of Noble agar(Difco Laboratories, Detroit, Mich.) containing 7 ,lof pony anti-PE serum per cm2. Rings were mea-sured after 24 h of incubation at 23°C. Double immu-nodiffusion was done according to the method ofOuchterlony (25).

Polyacrylamide gel electrophoresis. Electropho-resis in 8 and 5% polyacrylamide gels containingsodium dodecyl sulfate (SDS) was by a modificationof the procedure of Hosada and Cone (8). The gelsand lower reservoir contained 0.05 M Tris acetate,0.1% SDS, and 2 mM EDTA, pH 7.7. The upperreservoir contained the same buffer diluted fivefold.Samples of 40 gl were mixed with 10 ,ul of denatur-ant (3% SDS, 1% 2-mercaptoethanol, 50% glycerol,0.05 M Tris acetate, 2 mM EDTA, pH 7.7) and,unless otherwise stated, were heated at 90°C for 3min. Electrophoresis was performed at 3 mA/tubefor 2.5 h at 23°C. Gels were either cut into 2-mmsections or stained with Coomassie brilliant blue.

Isoelectric focusing was performed on polyacryl-amide gel slabs containing pH 3 to 10 ampholytes(PAG plate, LKB Instruments, Inc., Rockville,Md.), using the instructions provided. Sample appli-cation strips were removed after 30 min, and electro-phoresis was continued for an additional 60 min. Fordetermination of the isoelectric point (pI) of PE, thepH at intervals on the gel was measured with asurface electrode (Ingold Electrodes, Inc., Lexing-ton, Mass.) prior to staining with Coomassie bril-liant blue.Assay of ADP-ribosylation activity. For assay of

ADP-ribosylation activity, the general procedure ofCollier and Kandel was followed (5). A 35 to 50%ammonium sulfate fraction of rabbit reticulocytelysate was used as a source of EF-2. Rabbit reticulo-cyte lysate was prepared by F. B. Abeles. Assays(100-,l total volume) contained about 10 pmol ofrabbit reticulocyte EF-2, 37 pmol of ['4C]nicotin-amide adenine dinucleotide (0.01 ,uCi), 2 to 200 ngof PE, and 10 Al of a 0.50 M Tris-hydrochloride,1 mM EDTA, 0.40 M dithiothreitol, pH 8.2, buffer.After incubation for 60 min at 23°C, samples (90 ,l)were placed on paper disks, which were processedas described above for sulfhydryl determinations.For "activation" of PE, 10-gl samples at a,pproxi-mately 100 Ag/ml were mixed with 10 ,l of 10 M

INFECT. IMMUN.

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PSEUDOMONAS EXOTOXIN 1079

urea, 100 mM dithiothreitol, 0.03 M EDTA, 0.05 MTris-hydrochloride, pH 8.1. After 5 min at 23°C,the samples were diluted with 1.5 ml of buffer A andsamples were removed for assay.

Cell cytotoxicity tests. Tests of the cytotoxic ac-

tivity of exotoxin were performed in mouse L929 cellcultures using a method developed at this Instituteby J. L. Middlebrook and R. Dorland (manuscript inpreparation). Mixtures containing 100 1.l of serum(diluted 1:10, 1:30, 1:90, 1:270, etc.) and 2.5 .lI of PE(5 or 50 yg/ml) were incubated for 1 h and thenadded to wells of 24-well tissue culture plates, whichhad been seeded 24 h previously with 1.0 ml ofmedium 199 (Grand Island Biological Co., GrandIsland, N.Y.) supplemented with 2.2 g of bicarbon-ate per liter, 0.1 g of glutamine per liter, 100 U ofpenicillin per ml, 100 Ag of streptomycin per ml, and10% (vol/vol) fetal calfserum and containing 5 x 104mouse L929 fibroblast cells/ml. After incubation for48 h, the monolayers were washed with a balancedsalt solution to remove disattached cells. Remainingcells were solubilized by addition of 0.5 ml of 0.1 MNaOH per well. Portions were assayed for protein byan automated Lowry procedure.

RESULTS

Preparation of exotoxin. PE was producedby growing strain PA103 in a 70-liter fermenterusing the medium developed by Liu (15). Ini-tially, several 50-liter culture supernatantswere processed by combining the purificationprocedure of Liu et al. (17) with chromatogra-phy on hydroxylapatite, which was shown byCallahan (3) to be an effective purification step.This protocol produced good yields of pure PE,portions of which were used to prepare a high-titer pony antitoxin, by using the immuniza-tion schedule of Liu and Hsieh (16). However,this combination of steps included repeated cen-

trifugation of large volumes, which proved la-

borious. Furthermore, difficulty was often ex-

perienced in redissolving the precipitate ob-tained in the initial zinc acetate step describedby Liu et al. (17). Jackson and Matsueda (10)reported that the same problem frequently oc-

curs when zinc acetate is used to precipitate a

myxobacter protease. For these reasons otherprotocols more suitable for large-scale purifica-tions were tested. The low pl (5.0) of the protein(3) suggested that it could be adsorbed fromculture supernatants by anion-exchange resins.A protocol starting with adsorption to DEAE-cellulose proved successful. Analysis of such a

purification is shown in Table 1. Detection andquantitation of PE was conveniently achievedduring the purification by radial immunodiffu-sion in agar gels containing antitoxin. In thepreparation described in Table 1 the toxin-con-taining fractions obtained at each step were

also assayed for carbohydrate, protein, proteo-lytic activity, and ADP-ribosylation activity.The culture supernatant contained 5 to 10 ,ug

of PE per ml, equal to that reported for culturesgrown in shake flasks (17). This constitutes 5 to10% of the protein secreted by the bacterium. Aprevious report that culture supernatants con-

tain 30 mg of protein per ml (17) apparently wasbased on a protein assay that responded to var-ious low-molecular-weight materials. Adsorp-tion and elution of PE from DEAE-celluloseachieved a 25-fold reduction in volume and a 6-fold increase in specific activity. Chromatogra-phy on a DEAE-cellulose column eluted with a

linear gradient of NaCl produced two peaks ofultraviolet-absorbing material. Radial immu-nodiffusion showed all the toxin to be in thefirst peak, eluted at 0.12 to 0.14 M NaCl.The toxin-containing peak could be applied

TABLE 1. Purification of P. aeruginosa exotoxin

Purification Total pro- Total toxin by Total toxin Total carbohy-stage Total vol (ml) tein' (mg) radial diffu- by ribosyla- Sp actc drates (mg)stage telna (mg) ~~~siona (mg) tion" (mg)drts(g

Culture 50,000 5,500 <500" 350e 6 28,000super-natant

DEAE 2,000 500 190 180 35 60eluate

DEAE 150 166 150 140 85 2peak

Hydroxyl- 30 135 135 135 100 0.15dapatitepeaka Determined by reference to the standard curve obtained using the hydroxylapatite peak material,

assumed to be pure; E'* = 12, and optical density at 280 nm = 5.4.b Values are for "activated" toxin (see text).(Toxin [milligram] assayed by ribosylation x 100)/milligram of protein.

d Sample was below limit of detection.e Sample was concentrated 10-fold by ammonium sulfate precipitation prior to assay.

VOL. 14, 1976

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

1080 LEPPLA

directly to a hydroxylapatite column, since 0.15M NaCl does not prevent retention of protein onthis adsorbent. The column was eluted by alinear gradient of sodium phosphate. The exo-toxin emerged in a peak between 0.04 and 0.06M sodium phosphate, preceded by several verysmall peaks of impurities. Only a 15% increasein specific activity is achieved by this step be-cause the DEAE-cellulose peak is already nearhomogeneity. This step does substantially de-crease the carbohydrate content of the prepara-tion.The demonstration that PE possessed ADP-

ribosylation activity (9) provides an alternatemethod of quantitating the exotoxin protein. Ihave confirmed that PE does possess this enzy-matic activity (details presented below) andhave found, in addition, that the specific activ-ity (ADP-ribosylation activity per mole of pro-tein) can be increased 20- to 50-fold by simulta-neous exposure to a denaturing agent and areducing agent (manuscript in preparation).This activation phenomenon shows that PE ex-ists largely in a proenzyme form. The activityseen in untreated samples is probably due to asmall fraction of molecules which have becomeactivated (through unknown mechanisms) dur-ing production or purification. Since this frac-tion can be expected to vary, ADP-ribosylationassays of such samples would not provide avalid measure of the total mass of exotoxinprotein. Treatment with reducing and denatur-ing agents is believed to lead to full expressionof the potential enzymatic activity, so that allsamples achieve a maximum, uniform specificactivity. If this is true, then enyzmatic activitywould become a valid measure of exotoxin pro-tein. These considerations suggested that ADP-ribosylation assays would provide an accuratemeasure of the mass of exotoxin protein onlywhen performed on reduced and denaturedsamples. An analogous situation exists in diph-theria toxin (4), where ADP-ribosylation can beused as a measure of the total amount of toxinprotein, provided that samples are first acti-vated by treatment with trypsin and a reducingagent. The toxin-containing fractions obtainedin the purification were therefore assayed forADP-ribosylation after activation (Table 1).The values obtained correlated well with theamounts of PE as determined by radial immu-nodiffusion, an indication that both methodsare reliable measures of PE.Assays for proteolytic activity were per-

formed using a '4C-labeled, denatured hemoglo-bin substrate (data not shown). Activity wasdetected only in the culture supernatant and inthe supernatant of the initial adsorption toDEAE-cellulose.

The materials obtained at each step were alsoexamined by electrophoresis in polyacrylamidegels containing SDS (Fig. 1). The exotoxin bandis a major component of the culture superna-tant, consistent with the data of Table 1. Eachof the three purification steps removes certainof the contaminating proteins. The final prod-uct shows a single band, indicative of a highdegree of purity. The yield is 30 to 40% and isequal to that obtained by the more laboriousmethod referred to above. These PE prepara-tions have been stored at - 70°C for 6 months orat SoC for 5 to 10 days without apparent loss ofeither lethal or enzymatic activity.As an additional test of the purity of the

preparation, samples were examined by isoelec-tric focusing on polyacrylamide gel slabs; adja-cent sample positions received 0.4, 4, and 40 ,tgof PE. Minor bands, seen only in the largestsample, were estimated to comprise less than5% of the total mass.

Double diffusion in agarose gels of PE versusthe specific pony antitoxin yielded a single pre-

AS CD

Uq so

..

FIG. 1. SDS gels of materials from exotoxin puri-fication. Samples run on 8% SDS gels were: (A) 40pil of culture supernatant, previously concentrated10-fold by (NH4)2S04 precipitation (40 ,ug of pro-tein), (B) 80 piu of eluate from DEAE-cellulose batch-wise adsorption (20 pg ofprotein), (C) 10 p1 ofpeakfrom DEAE-cellulose column (10 pAg ofprotein), and(D) 4 p1 ofpeak from hydroxylapatite column (15 pgofprotein).

INFECT. IMMUN.

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PSEUDOMONAS EXOTOXIN 1081

cipitin line, which did not interact with thediphtheria toxin-antitoxin reaction (Fig. 2).This further demonstrates the purity of thepreparation and is consistent with the failure ofthe heterologous antitoxin to inhibit the ADP-ribosylation activities of DE and PE.Chemical properties of purified exotoxin. In

a 0.05 M Tris-hydrochloride buffer at pH 8.1,the purified exotoxin had an El"0 12, a maxi-mum absorbance at 280 nm, and a ratio ofabsorbance at 280 nm to that at 260 nm of 1.92.The pI was determined by isoelectric focusingin a polyacrylamide gel slab of pH 3 to 10,operated at 5°C. The pH at intervals on the gelwas measured with a surface electrode at 23°C.The apparent pI was 5.1, in good agreementwith the value of 5.0 previously reported (3).The molecular weight of the exotoxin was de-termined by electrophoresis on 8% polyacryl-amide gels containing SDS (Fig. 3). Compari-son of the mobilities of PE and of severalmarker proteins showed that the molecularweight of PE is 66,000 + 1,000. This value issubstantially above the previously reportedvalues of 50,000 to 54,000 obtained by gel filtra-tion. The electrophoretic pattern of Fig. 3 alsodemonstrates that PE contains a single poly-peptide chain, rather than being composed ofseveral subunits.Amino acid analysis of the purified exotoxin

is presented in Table 2. The exotoxin has alarge number of acidic amino acids, consistentwith its low pL. The high arginine-to-lysine ra-

p

tio is very unusual, since in a recent compila-tion of 476 proteins (12, 13) only 10 (lysine-containing) proteins had a more extreme ratio.The N-terminal amino acid was found to bearginine.

Sulfhydryl groups and disulfides were mea-sured by reaction with ["4C]iodoacetamide (28).Figure 4 shows that the protein contains no freesulfhydryl groups and four disulfide bonds. Theanalyses for carbohydrate using anthrone (Ta-ble 1) show that the purified exotoxin containsless than one residue of hexose per molecule.Carbohydrate analysis by gas chromatographywas also negative.

Biological activity. PE was first recognizedby its lethality for mice. When tested by intra-peritoneal injection into mice weighing 20 g,the purified material described above had amedian lethal dose of 0.1 ,ug. This is in goodagreement with the previously reported valueof 0.12 ,ug (17). The median lethal dose of toxinin 300-g Sprague-Dawley rats was also deter-mined. The value obtained, 20 ,Lg, indicatesthat rats are about eightfold less sensitive thanmice on a body weight basis.PE has been shown to be cytotoxic to a num-

ber of cell lines. A convenient assay of cytotox-icity using mouse L929 cells, available in thislaboratory (see Materials and Methods), can beused to measure small amounts of either theexotoxin or antibodies to exotoxin. The experi-ment described in Table 3 shows that the cyto-toxic action of purified exotoxin is completely

D

o~P D -P

ANTISERA

FIG. 2. Ouchterlony analysis ofPE and DE. Toxin wells P and D received 1 pg ofPE and 2 pg of DE,respectively. Antisera wells P and D received 2 p1 of pony anti-PE and p1 of diphtheria antitoxin,respectively.

TOXINS-

Di P

.... >-~~~ ..,>X

VOL. 14, 1976

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

1082 LEPPLA

A B*rn

C D E

70,000 Aa57,000 B48,000

m. imAgoa* BSA 68,000

' DE 63,000

FIG. 3. Molecular weight of PE. Samples subjected to electrophoresis on 8% polyacrylamide-SDS gelswere: (A) a mixture of cold-insoluble globulin (200,000 molecular weight), f3galactosidase (130,000),phosphorylase a (100,000), and the fibrinogen peptides Aa (70,000), B,8 (57,000), and y (48,000); (B) thesame sample as in (A) plus 2 pg ofPE; (C) 2 pg ofPE; (D) the same sample as in (E) plus 2 pg ofPE; (E) 4 pgeach of diphtheria toxin (63,000) and bovine plasma albumin (68,000).

TABLE 2. Amino acid composition ofPE

Amino acid Residues/Aminoacidmoeuemoleculea

Asp ................................ 57Thr ............................... 28Ser ................................. 39Glu ................................ 68Pro ................................. 38Gly ................................. 56Ala ................................. 67Half-Cys ................ 7Val ................................. 33Met ................................ 5Ile ................................. 24Leu ................................ 67Tyr ................................. 18Phe ................................ 14Lys ................................. 15His ................................. 14Arg ................................ 45Trp ................................. 11

Total residues ............ ........... 606

a Calculated for a molecular weight of 66,000.

neutralized by the specific pony antitoxin. Itcan be calculated from these data that 1 gld ofantitoxin neutralizes 1 gg of exotoxin.

Iglewski and Kabat (9) presented evidencethat PE acts by blocking protein synthesis in arabbit reticulocyte lysate. I have found (in ex-periments performed with F. B. Abeles) thatexotoxin also blocks amino acid incorporationin a wheat germ cell-free system. This is con-sistent with the observation that diphtheriatoxin fragment A can inactivate EF-2 from avariety of eukaryotic cells.Enzymatic activity. The purified exotoxin

and the crude materials from which it is ob-tained (Table 1) possess ADP-ribosylation ac-tivity. At concentrations of less than 100 ng ofactivated PE per ml of assay solution, the rateof incorporation of ['4C]ADP-ribose into the tri-chloroacetic acid-insoluble fraction was con-stant over 60 min. Blanks lacking exotoxin had50 cpm, whereas high concentrations of exo-toxin saturated the EF-2, about 3,000 cpm.Some assays used a wheat germ extract as thesource of EF-2. This material contained abouttwo to five times as much EF-2 as the reticulo-cyte preparation, but blanks were higher, prob-ably due to a PE-independent incorporation orpolymerization of nicotinamide adenine dinu-cleotide (24). Electrophoresis of a reaction mix-ture on a 5% polyacrylamide-SDS gel confirmed

INFECT. IMMUN.

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PSEUDOMONAS EXOTOXIN 1083

zEJi08O

a.

w

0

26-l

w

4-

$

-J0

30 60

TIME (MIN)

FIG. 4. Determination of cysteine and cystine. PEdissolved in buffer B (Ol) or pretreated with dithio-threitol in buffer B (0O) was reacted with['4C]iodoacetamide. Samples were spotted on paperdisks, which were washed to remove unreacted iodo-acetamide.

TABLE 3. Antibody neutralization of PE-inducedcell cytotoxicity"

Serum Cell protein (mg/100 ml) inwells containing:

12 ng of 125 ng ofSource Dilution No toxin toxin/ toxin/

well well

None 30 3 3Horse, normal 100 31 3 3Pony, immune 100 33 35

300 32 35900 28 30

2,700 30 308,100 29 27

24,000 30 573,000 30 4

216,000 7 4a Mixtures of PE and the serum to be tested were

incubated for 1 h and added to established L929 cellmonolayers to give the listed final serum dilutionsand PE amounts. After 48 h, wells were washed andassayed for residual cell protein.

that the '4C-labeled product had a molecularweight of about 100,000, which is the reportedsize of rat liver EF-2 (6).

Structure-function studies of diphtheriatoxin (5, 7) revealed a proteolytic fragment of24,000 molecular weight, called fragment A,

which possesses ADP-ribosylation activity butis biologically nontoxic. Conversely, intact DEis toxic but enzymatically inactive, apparentlybecause the active site on fragment A isblocked. To examine whether a similar situa-tion exists in PE, samples pretreated with 2-mercaptoethanol were fractionated on SDSgels. The gels were cut into 2-mm slices, whichwere placed in 1.0-ml samples of buffer. Afterovernight incubation to allow diffusion of thepeptides out of the gel slices, samples wereassayed for cytotoxic and enzymatic activity.Gill and Pappenheimer have previously shownthat the enzymatic activity of diphtheria toxincan be recovered from SDS gels (7) if dilutionlowers the SDS concentration below 0.02%. Inour hands, the ADP-ribosylation activity of PEwas inhibited by 0.005% SDS, but not by 0.002%SDS. The assay of gel slices was designed sothat SDS concentrations were below 0.001%.The profiles seen in Fig. 5 demonstrate thatboth the cytotoxic and enzymatic activities co-incide with the protein band. It appears thatthe intact PE molecule possesses enzymatic ac-tivity, unlike the situation with diphtheriatoxin. Two peaks of lower molecular weight,which possess ADP-ribosylation activity butare not cytotoxic, also appear on the gel. Thesemay be proteolytic fragments analogous to frag-ment A of DE or simply breakdown products ofthe exotoxin with a small residual activity andno biological significance.

I,0

E A

I-

U4

z0

D4

2 3 4DISTANCE (cm)

80 6

60

:E0B

40 -J

20

5 6 7

FIG. 5. Molecular sizes ofcytotoxic and ADP-ribo-sylating activities of PE. Samples of 2.5 Mg of PEwere treated with denaturant at 23°C for 30 min andrun on 8% polyacrylamide-SDS gels. One gel wasstained. The solid vertical line indicates the positionof the stained band. Another gel was cut into 2-mmslices; each slice was incubated overnight at 4°C in1.0 ml of 5 mM Tris-hydrochloride, 0.1% 2-mercap-toethanol, pH 7.5. Ribosylation assays (a) were per-formed on 20-,pl samples of the eluate. Cytotoxicity(0) was measured in mouse L929 cell monolayercultures (1.0 ml), which received 20 pl of a 1:10dilution of the eluate.

VOL. 14, 1976

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

1084 LEPPLA

DISCUSSION

The purification described here provides a

simple and reliable method for preparation ofsubstantial amounts of pure PE. A previousmethod used growth on solid media and afreeze-thaw step which apparently releasedcellular proteins, complicating the purification.It is shown here that PE is a major componentof the culture supernatant. Therefore, it is notsurprising that a relatively simple purificationis effective. The use of radial immunodiffusionto detect and quantitate PE during the puri-fication proved more convenient, accurate,and economical than injecting samples intomice. The PE preparation is at least 95% pure,as measured by several electrophoretic tech-niques. The few trace components remainingdo not cause any apparent interference duringstudies of exotoxin now underway in thislaboratory. The minor bands of lower pI seen inisoelectric focusing may be toxin molecules thathave suffered amide hydrolysis. Microhetero-geneity in staphylococcal enterotoxins has beenattributed to this cause (21).PE was reported to be readily destroyed by

the protease(s) of P. aeruginosa (17). For thisreason, special attention was directed duringdesign of this purification procedure to theearly separation or inhibition of the Pseudomo-nas proteases. Morihara et al. (23) describedthree extracellular proteases made by moststrains of P. aeruginosa. Strain PA103 wasoriginally selected as being protease deficient,but is known to produce small amounts of pro-

tease. By use of a sensitive protease assay usinga "C-labeled, denatured hemoglobin substrate,it was found (unpublished data) that the resid-ual proteolytic activity of PA103 is not adsorbedto DEAE-cellulose. It appears therefore thatPA103 retains principally the "neutral protein-ase" (23). This fact encouraged selection ofbatch adsorption to DEAE-cellulose as the firststep in purification, since this operationachieves both concentration of PE and removalof protease.

Consistent with previous reports, exotoxinwas found to be a rather acidic polypeptide, freeof carbohydrate. The amino acid content is notunusual, except for a rather extreme arginine-to-lysine ratio. Comparison of the amino acidcontent of PE to that of diphtheria toxin (22)reveals no apparent similarities, consistentwith the absence of serological cross-reactivity.The molecular weight of 66,000 determined

by SDS-gel electrophoresis is substantiallyhigher than the previously reported values de-termined by gel filtration. Analyses of protein

molecular weights by gel filtration often yieldincorrect results, because dextran gels bindproteins nonspecifically and retard their elu-tion. Since artifacts of this type are rare inSDS-gel electrophoresis, it is probable that thevalue reported here is more accurate. An alter-nate explanation, that the material previouslydescribed is a degradation product of the 66,000-dalton species, appears unlikely, since that ma-terial was produced under equivalent cultureconditions and possessed the same biologicalactivity.Analyses using ['4Cliodoacetamide showed

that PE contains no free sulfhydryls and fourdisulfide bonds. This is in reasonable agree-ment with the value of seven half-cystinesfound by amino acid analysis. In comparison,diphtheria toxin possesses two disulfides, one ofwhich links fragments A and B. Reduction ofthis latter disulfide in nicked molecules re-leases fragment A. Studies are in progress todetermine whether the disulfides ofPseudomo-nas exotoxin play a similar role.

It is confirmed here that the median lethaldose of PE in mice weighing 20 g is about 0.1,tg. This is the amount of PE produced by 108bacteria in culture. Since P. aeruginosa canreadily grow to densities of 108/ml in burnedtissue, enough PE could be made in vivo tocontribute significantly to pathogenesis. Stud-ies of the role of PE in burn wound infectionmay most conveniently be performed in rats,since this animal model has been well charac-terized (18). However, it was found here thatrats are much less sensitive than mice to PE,both on a per animal and a body weight basis.Therefore the use of rats in such studies wouldrequire significantly greater amounts of PEthan if mice were used.The work of Iglewski and Kabat (9) clearly

demonstrated that PE acts enzymatically likeDE. However, no information was provided onthe relative specific activites of the two toxins.Preliminary experiments relating to the struc-ture-function relationship were describedwhich showed that trypsin treatment did notactivate PE and that prior reduction did notalter the mobility of PE on SDS gels. Theseresults led the authors to suggest that intact PEmight be enzymatically active. Their results,however, did not exclude the possibility thatthe enzymatic activity of PE preparations isdue to a very low level of a preexisting peptidefragment analogous to fragment A of DE. Infact, several results obtained here are entirelyconsistent with the existence of such a peptidefragment. Thus, the initial specific activity(ADP-ribosylation activity per mole of protein)

INFECT. IMMUN.

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PSEUDOMONAS EXOTOXIN 1085

of PE preparations is low compared to that offragment A of DE. Activation of PE by simulta-neous reduction and denaturation increases thespecific activity 20- to 50-fold, making it ap-

proximately equal to that of fragment A. Theseresults are exactly those expected for the assay

in low levels of reducing agent of a preparationof DE that is fully nicked and contains 2 to 5%of free fragment A. Therefore, it was not cer-

tain that PE had a structure-function relation-ship different from that of DE. However, theresults of Fig. 5 clearly demonstrate that intactPE is enzymatically active. Unlike DE, PE doesnot appear to require fragmentation for expres-

sion of enzymatic activity.Whereas intact PE can achieve ADP ribosy-

lation, no evidence exists to show that intactPE is the active form of toxin within sensitivecells. In fact, Fig. 5 and other experiments notdescribed here show that smaller, enzymati-cally active peptides do exist. Determiningwhich forms of PE are actually present in thecytoplasm of sensitive cells may be achieved byanalysis of intoxicated cell cultures.

ACKNOWLEDGMENTS

I thank John Middlebrook, Leonard Spero, Fred Abeles,Joseph Metzger, and Anna Johnson for advice and assist-ance in some experiments. I am grateful for the technicalassistance of Theodora Grayson, Violet Defonsey, LauraMuehl, and Ona Martin.

LITERATURE CITED

1. Bhatti, T., R. E. Chambers, and J. R. Clamp. 1970. Thegas chromatographic properties of biologically impor-tant N-acetylglucosamine derivatives, monosacchar-ides, disaccharides, trisaccharides, tetrasaccharidesand pentasaccharides. Biochim. Biophys. Acta222:339-347.

2. Bramhall, S., N. Noack, M. Wu, and J. R. Loewenberg.1969. A simple colorimetric method for determinationof protein. Anal. Biochem. 31:146-148.

3. Callahan, L. T., III. 1974. Purification and characteri-zation of Pseudomonas aeruginosa exotoxin. Infect.Immun. 9:113-118.

4. Collier, R. J. 1975. Diphtheria toxin: mode of action andstructure. Bacteriol. Rev. 39:54-85.

5. Collier, R. J., and J. Kandel. 1971. Structure and activ-ity of diphtheria toxin. I. Thiol-dependent dissocia-tion of a fraction of toxin into enzymatically activeand inactive fragments. J. Biol. Chem. 246:1496-1503.

6. Collins, J. F., S. Raeburn, and E. S. Maxwell. 1971.Aminoacyltransferase II from rat liver. II. Somephysical and chemical properties of the purified en-

zyme and its adenosine diphosphate ribose deriva-tive. J. Biol. Chem. 246:1049-1054.

7. Gill, D. M., and A. M. Pappenheimer, Jr. 1971. Struc-ture-activity relationships in diphtheria toxin. J.Biol. Chem. 246:1492-1495.

8. Hosoda, J., and R. Cone. 1970. Analysis of T4 phageproteins. I. Conversion of precursor proteins intolower molecular weight peptides during normal cap-sid formation. Proc. Natl. Acad. Sci. U.S.A. 66:1275-1281.

9. Iglewski, B. H., and D. Kabat. 1975. NAD-dependentinhibition of protein synthesis by Pseudomonasaeruginosa toxin. Proc. Natl. Acad. Sci. U.S.A.72:2284-2288.

10. Jackson, R. L., and G. R. Matsueda. 1970. MyxobacterAL-1 protease, p. 591-599. In G. E. Perlmann and L.Lorand (ed.), Methods in enzymology, vol. 19. Aca-demic Press Inc., New York.

11. Johnson, A. D., J. F. Metzger, and L. Spero. 1975.Production, purification, and chemical characteriza-tion of Staphylococcus aureus exfoliative toxin. In-fect. Immun. 12:1206-1210.

12. Kirschenbaum, D. M. 1974. A compilation of amino acidanalyses of proteins. VI. Residues per molecule-IV.Anal. Biochem. 61:567-609.

13. Kirschenbaum, D. M. 1975. A compilation ofamino acidanalysis of proteins. VII. Residues per molecule-5.Anal. Biochem. 66:123-150.

14. Liu, P. V. 1966. The roles of various fractions of Pseu-domonas aeruginosa in its pathogenesis. III. Identityof the lethal toxins produced in vitro and in vivo. J.Infect. Dis. 116:481-489.

15. Liu, P. V. 1973. Exotoxins ofPseudomonas aeruginosa.I. Factors that influence the production of exotoxin A.J. Infect. Dis. 128:506-513.

16. Liu, P. V., and H. Hsieh. 1973. Exotoxins ofPseudomo-nas aeruginosa. III. Characteristics of antitoxin A. J.Infect. Dis. 128:520-526.

17. Liu, P. V., S. Yoshii, and H. Hsieh. 1973. Exotoxins ofPseudomonas aeruginosa. II. Concentration, purifica-tion, and characterization of exotoxin A. J. Infect.Dis. 128:514-519.

18. McRipley, R. J., and D. W. Garrison. 1964. Increasedsuceptibility of burned rats to Pseudomonas aerugi-nosa. Proc. Soc. Exp. Biol. Med. 115:336-338.

19. Mancini, G., J. P. Vaerman, A. 0. Carbonera, and J. F.Heremans. 1964. A single radial diffusion method forthe immunological quantitation of proteins, p. 370-373. In H. Peeters (ed.), Protides of the biologicalfluids. Elsevier, Amsterdam.

20. Mans, R. J., and G. D. Novelli. 1961. Measurement ofthe incorporation of radioactive amino acids into pro-tein by a filter-paper disk method. Arch. Biochem.Biophys. 94:48-53.

21. Metzger, J. F., A. D. Johnson, and L. Spero. 1975.Intrinsic and chemically produced microheterogene-ity of Staphylococcus aureus enterotoxin type C. In-fect. Immun. 12:93-97.

22. Michel, A., J. Zanen, C. Monier, C. Crispeels, and J.Dirkx. 1972. Partial characterization of diphtheriatoxin and its subunits. Biochim. Biophys. Acta257:249-256.

23. Morihara, K., H. Tsuzuki, T. Oka, H. Inoue, and M.Ebata. 1965. Pseudomonas aeruginosa elastase. Isola-tion, crystallization, and preliminary characteriza-tion. J. Biol. Chem. 240:3295-3304.

24. Nishizuka, Y., K. Ueda, K. Nakazawa, and 0. Hay-aishi. 1967. Studies of the polymer of adenosine di-phosphate ribose. I. Enzymic formation from nicotin-amide adenine dinucleotide in mammalian nuclei. J.Biol. Chem. 242:3164-3171.

25. Ouchterlony, 0. 1953. Antigen-antibody reactions ingels. IV. Types of reactions in coordinated systems ofdiffusion. Acta Pathol. Microbiol. Scand. 32:230-240.

26. Pavlovskis, 0. R., and F. B. Gordon. 1972. Pseudomo-nas aeruginosa exotoxin: effect on cell cultures. J.Infect. Dis. 125:631-636.

27. Spiro, R. G. 1966. Analysis of sugars found in glycopro-teins, p. 3-26. In E. F. Neufeld and V. Ginsburg (ed.),Methods in enzymology, vol. 8. Academic Press Inc.,New York.

28. Steinert, P. M. 1975. The extraction and characteriza-

VOL. 14, 1976

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

1086 LEPPLA INFECT. IMMUN.

tion of bovine epidermal a-keratin. Biochem. J.149:39-48.

29. Tapper, M. L., and D. Armstrong. 1974. Bacteremiadue to Pseudomonas aeruginosa complicating neo-plastic disease: a progress report. J. Infect. Dis.

130(Suppl.):S14-S23.30. Weiner, A. M., T. Platt, and K. Weber. 1972. Amino-

terminal sequence analysis of proteins purified on ananomole scale by gel electrophoresis. J. Biol. Chem.247:3242-3251.

on June 12, 2020 by guesthttp://iai.asm

.org/D

ownloaded from