Biochem. J. (1982) 201, 505-5 13 505Printed in Great Britain

A new method for the extraction and purification ofK99 pili fromenterotoxigenic Escherichia coli and their characterization

K. ALTMANN,* N. A. PYLIOTISt and T. K. S. MUKKURt§*Department of Veterinary Pathology, University ofSydney, Glebe, N.S. W. 2037, tPrince Henry's Hospital,

Melbourne, Vic. 3004, and tC.S.I.R.O. Division ofAnimal Health, McMasterLaboratory, Glebe,N.S. W. 203 7, Australia

(Received 28 September 1981/Accepted 26 October 1981)

It was found that K99 pili from enterotoxigenic Escherichia coli (of bovine origin)could be extracted by treatment with 3 M-KSCN solution. The K99 pili were purified bypreparative isoelectric focusing to apparent homogeneity as judged by the presence of asingle band on sodium dodecyl sulphate/polyacrylamide-gel electrophoresis; themolecular weight of this component was calculated to be 12 600 ± 300. This indicatedthat the K99 pili were composed of a single subunit. On analytical ultracentrifugation, asingle boundary with an s2O,w of 12.2S at a concentration of 0.42mg/ml was observed.The average length of purified pili at zero concentration was appprox. 160nm and thediameter was 7.4+0.6nm. Amino acid analysis of the purified K99 pili revealed thatsulphur-containing amino acids, cysteine and methionine, were absent. Aromatic aminoacids, phenylalanine and tyrosine, previously reported to be absent [Isaacson (1977)Infect. Immun. 15, 272-2791, constituted 7.14% of the total amino acid residues present.On immunoelectrophoresis, purified K99 pili migrated towards the cathode and causedmannose-resistant haemagglutination of horse, but not of sheep or guinea-pig, red bloodcells. Pili from enterotoxigenic E. coli of porcine and human origin and from anotherbacterial species, namely Fusiformis nodosus, could also be extracted by the treatmentof respective micro-organisms with 3 M-KSCN.

Enterotoxigenic Escherichia coli cause an acutecholera-like diarrhoea in both animals and humans.The disease process initially involves the coloniza-tion of the mucosal surface of the small intestine,followed by the elaboration of a heat-labile entero-toxin and/or heat-stable enterotoxin. Intestinalcolonization by enterotoxigenic E. coli is generallymediated by a specific surface-associated pilus orfimbrial antigen that endows the bacteria with thecapacity to adhere to epithelial-cell surface (Smith &Linggood, 1972). Such pili, which are plasmid-mediated, are the K88 antigens of swine-specific(Jones & Rutter, 1974), the K99 antigen of calf- andsheep-specific (0rskov et al., 1975) and the coloniza-tion-factor antigens of human-associated enterotoxi-genic E. coli (Evans et al., 1975, 1977; Evans &Evans, 1978). K88, K99 and colonization-factorantigens are antigenically unrelated to the type 1 piligenerally associated with E. coli and are dis-

Abbreviations used: SDS, sodium dodecyl sulphate;RNAase, ribonuclease.

§ To whom requests for reprints should be addressed.

Vol. 201

tinguishable from the latter by their mannose-resistant haemagglutinating activity (Burrows et al.,1976; 0rskov et al., 1977; Evans et al., 1979).K99 has been reported to be associated with a

majority of enterotoxigenic E. coli isolated fromcases of diarrhoeal disease in neonatal calves andlambs (Guinee et al., 1976; Moon et al., 1976;Myers & Guin&e, 1976; Sivaswamy & Gyles, 1976).K99 was previously reported to have been purified

and composed primarily of protein subunits of22 500 and 29 500 daltons with a pl in excess of 10(Isaacson, 1977). Such purified pili did not haem-agglutinate guinea-pig erythrocytes. On the otherhand, Morris et al. (1977, 1978) reported that K99pili isolated in their laboratory haemagglutinatedguinea-pig erythrocytes and had a pl of 4.2.Isaacson (1978) later reported that if K99 pili wereprepared by acid precipitation, they caused guinea-pig erythrocytes to agglutinate, presumably becauseof the presence of a second antigen with a pl of 4.2.However, if the K99 pili were prepared by ion-exchange chromatography on DEAE-Sephadex,guinea-pig erythrocytes were not agglutinated. More

0306-3275/82/030505-10$01.50/1 (© 1982 The Biochemical Society

K. Altmann, N. A. Pyliotis and T. K. S. Mukkur

recently, Morris et al. (1980) presented evidencesuggesting the existence of two types of adhesiveantigens, namely cationic and anionic haemagglu-tinins, in their K99-pili preparations.

Because of fairly recent optimistic reports regard-ing the feasibility of using pili preparations aspossible vaccines for the control of scours(diarrhoea) in calves, lambs and piglets (Isaacsonet al., 1977; Acres et al., 1978; Sojka et al., 1978),it was considered important to develop an efficientmethod for the isolation and purification of K99 pilito facilitate the evaluation of purified pili as apossible vaccine for the control of neonatal-calfscours, and to study the physicochemical propertiesof purified K99 pili with the ultimate aim ofachieving maximum immunopotentiation via anti-genic modification, with or without the assistance ofappropriate adjuvants. In the present paper, data areprovided on: (i) the comparative evaluation ofmethods developed in this laboratory for theisolation of purified K99 pili, and (ii) some of thephysicochemical properties of thus-purified K99 pili.

Materials and methodsBacteriaTwo strains of enterotoxigenic E. coli were used in

the present investigation. One, C 1443, was kindlyprovided by Dr. J. Craven of the VeterinaryResearch Laboratory, 'Attwood', Westmeadows,Vic., Australia, and had been isolated from anoutbreak of calf scours. Its serotype was 020:KX106:K99+, H- [I. Links (Regional VeterinaryLaboratory, Wagga Wagga, N.S.W., Australia),personal communication]. The organism producedboth heat-stable and heat-labile enterotoxins. Theother E. coli strain used was the K99 reference strain(B4 1) and was obtained through the courtesy of Dr.H. W. Smith.

Growth conditionsFor the extraction of K99, E. coli was grown at

370C in a batch fermenter (New Brunswick)containing 10 or 18 litres of Minca medium (Guineeet al., 1976), unless otherwise indicated. Thebacterial culture was aerated at the rate of 31itres/min.When micro-organisms were to be used for absorp-

tion purposes for the production of specific K99antiserum, they were grown at 180C in stationaryflasks. Under these conditions, the E. coli strain(C 1443) was found to be devoid of pili as judged bytransmission electron microscopy of negativelystained preparations, thus confirming previous ob-servations (0rskov et al., 1975), although occa-sionally cells possessing a long, single pilus (resembl-ing a sex pilus) were encountered. Whereas the cellsgrown as such caused neither mannose-sensitive nor

mannose-resistant haemagglutination, K99+ cellscaused a mannose-resistant haemagglutination ofhorse red blood cells.Extraction procedures

(i) Ultrasonication. Details of the procedure usedhave been described previously (Clements & Finkel-stein, 1979).

(ii) Homogenization. The conditions used wereessentially those described by Isaacson (1977).

(iii) Heating of broth cultures of E. coli. Thismethod of extraction of K99 pili has been describedpreviously (Morris et al., 1977).

(iv) Extraction with KSCN. As revealed by thekinetics of the release of K99 activity as a functionof total protein released, enterotoxigenic E. coli wereextracted with a final concentration of 3 M-KSCN,pH 7.1, for 1 h at 220 C, followed by centrifugationat 12 350g in a Sorvall centrifuge (RC-2B) for30min in order to sediment the extracted cells, whichwere discarded. The supernatant (referred to here-after as the 'crude KSCN extract') was dialysedexhaustively against phosphate-buffered saline(Morris et al., 1977) and stored frozen at -200C.Assay ofK99 activityK99 activity was detected by the following

methods.(i) Ouchterlony double-diffusion technique in

0.85% agarose C. Wells punched in the agar (5mmapart) were filled with 20,1 of either sample orspecific K99 antiserum, which was prepared inrabbits as described below. K99 activity wasquantified by determination of the most dilutesample that resulted in the development of aprecipitin arc when made to react with specific K99antiserum for up to 96h at 40C. The reciprocal ofthat dilution was designated as the K99 activity/20,u1.

(ii) Haemagglutination of horse red blood cells.Haemagglutination of horse red blood cells wascarried out as described by Morris et al. (1980) inround-bottomed microtitre plates. Serial 2-fold dilu-tions of samples (25,ul) to be assayed for K99activity were made in phosphate-buffered saline.K99 activity was quantified by determining the lastdilution of the sample that resulted in haem-agglutination. The reciprocal of that dilution wasused to designate the K99 activity/25,ul.

Haemagglutination was also carried out withguinea-pig and sheep red blood cells, but only withpurified K99 preparations.Preparation oJ K99-specific antisera

Antisera directed against 0 antigens (0: 20),polysaccharide capsule (KX 106) and K99 pili (OKantisera) were prepared essentially as described bySojka (1965) (hereafter referred to as 'unabsorbedC 1443 OK antisera') absorbed repeatedly with live

1982

506

K99 pili from enterotoxigenic E. coli

cells grown at 18°C until no agglutination of thelatter was encountered. Such antisera agglutinatedthe Minca-grown E. coli strain C 1443 (K99+) to atitre of 8-16. However, no agglutination wasobserved with (a) enterotoxigenic E. coli (C1443)grown at 180C or (b) enterotoxigenic E. coli(C1443) grown in trypticase soy broth at 370C,heated for 1 h at 1000C, with the agglutinationreaction being carried out at 560C.The K99 antisera were further judged to be

specific by the following additional criteria.(i) When immunoelectrophoresis of the crude

KSCN extract was carried out with K99 antiserum,only a single arc migrating towards the cathode wasobserved (Guinee et al., 1976).

(ii) Haemagglutination of horse red blood cells bythe crude KSCN extract containing cell-free K99antigen was inhibited (Morris et al., 1977).

Assayfor heat-labile enterotoxic activityThe heat-labile enterotoxic activity was assayed

by using y-1 adrenal tumour cells (Donta et al.,1974).

Preparative isoelectricfocusingPreparative isoelectric focusing was carried out

with Sephadex IEF with 2% Pharmalyte, pH3-10(Pharmacia Fine Chemicals, Sydney, N.S.W.,Australia). The power limit was set at 2000V, withthe current limit being set at 8W (LKB power supply2103) and the total isoelectric-focusing time wasapprox. 24h. A portion (30ml) of the sample wasdialysed against phosphate-buffered saline, pH7.4(0.05 M-K2HPO4/0.15 M-NaCI) overnight and di-luted to 100ml by addition of 6% D-Mannose indeionized water before mixing with Sephadex IEF.After isoelectric focusing, the Sephadex IEF wassliced into 30 equal segments and eluted with 5 ml ofphosphate-buffered saline. Removal of Pharmalytewas accomplished by the addition of (NH4)2SO4 to50% final saturation to precipitate proteins. Thelatter were removed by ultracentrifugation at105 000g for 60min and reconstituted in 1.0 ml ofphosphate-buffered saline.

Analyticalpolyacrylamide-gel electrophoresisPolyacrylamide-gel electrophoresis in SDS was

carried out in 10% gels as described by Weber &Osborn (1969). Electrophoresis was carried out atSmA/gel for approx. 4h. Staining of the gels wasaccomplished by using 0.25% Coomassie Blue inmethanol/acetic acid/water (5:1:5, by vol.). De-staining was achieved with 7.5% acetic acid con-taining 5% methanol. SDS/polyacrylamide-gelelectrophoresis was used for the determination of themolecular weight of purified K99 as well as toanalyse: (a) fractions with K99 activity at various

Vol. 201

steps of purification; (b) immunoprecipitates ob-tained by allowing specific K99 antiserum to reactwith crude K99 prepared from enterotoxigenic E.coli (C 1443) by other methods (Isaacson, 1977;Morris et al., 1977); (c) the crude K99 pili preparedby KSCN extraction of K99 reference strain (B4 1).

Chemical analysisProtein concentration was determined as des-

cribed by Bradford (1976) using the Bio-Radprotein kit. For the detection of DNA, the methodused was that of Ceriotti (1952). The method usedfor the detection of RNA was based on a standardcurve depicting change in A260 of yeast RNAcaused by hydrolysis with RNAase and has beendescribed previously (Mukkur & Pyliotis, 1981). Thelipopolysaccharide content was determined by using2-oxo-3-deoxyoctonate as a marker as described byWeissbach & Hurwitz (1959) with the modificationthat the hydrolysis was carried out at 1000C for20min, and the concentrations of periodic acid andsodium arsenite were those recommended by Osborn(1963). Total carbohydrate was determined by usingthe anthrone reagent as described by Mokrasch(1954); glucose was used as a standard.

ImmunoelectrophoresisImmunoelectrophoresis was carried out by the

method of Scheidegger (1955) with 0.85% agarose Cdissolved in barbitone/acetate buffer, pH 8.6 (Ox-oid). The antisera used for the development ofimmunoelectrophoretic patterns (40C) includedspecific K99, unabsorbed C 1443 OK and un-absorbed KSCN-extract antisera. These were usedas indicated in the Results and discussion section.

Electron microscopyWhereas the preparations of cells (untreated and

KSCN-extracted) suspended in saline were stainedin 0.5% sodium phosphotungstate, pH 6.0, those ofthe extracts were stained in 2% sodium phospho-tungstate, pH 7.0, with standard negative-stainingprocedures. The preparations were then examinedin a Philips EM301 electron microscope.

Determination ofsedimentation coefficientThe S20,w of purified K99 pili was determined by

analytical ultracentrifugation in a Beckman ana-lytical ultracentrifuge equipped with a multiplexscanner. Centrifugation was carried out at 37020rev./min in a titanium rotor (AnG-Ti) at 200C. Theconcentration of purified K99 pili employed was0.42 mg/ml.

Amino acid analysisPurified K99 pili were dialysed exhaustively

against water, freeze-dried and dissolved in 6 M-HCl.After hydrolysis of the sample at 110°C for 24h

507

K. Altmann, N. A. Pyliotis and T. K. S. Mukkur

under vacuum, the amino acid composition wasdetermined in an automated amino acid analyser(JLC-6AH). Tryptophan was determined spectro-scopically (Edelhoch, 1967). The results werecalculated as mol of amino acid residues recovered/mol of purified K99.

Results and discussion

Different methods were compared for their effi-ciency of extraction of K99 pili from an entero-toxigenic E. coli strain of bovine origin. The methodsevaluated included homogenization (Isaacson,1977), heating of an enterotoxigenic E. coli brothculture at 600C for 30min (Morris et al., 1977),ultrasonication [employed by Clements & Finkel-stein (1979) for the extraction of heat-labile toxinfrom enterotoxigenic E. colil], and extraction with achaotropic agent, namely KSCN. The latter hadbeen successfully employed previously to extractprotective antigens of Pasteurella multocida, whichwere protein in nature (Mukkur, 1979; Mukkur &Pyliotis, 1981). The assay systems used for moni-toring the presence of K99 pili were: (a) trans-mission electron microscopy of negatively stainedpreparations; (b) haemagglutination of horse redblood cells (Morris et al., 1980); and (c) pre-cipitation reactions in gel using specific K99 anti-sera (Isaacson, 1977). When the enterotoxigenic E.coli extracted by various methods were negativelystained and examined by transmission electronmicroscopy for the efficiency of removal of K99 pilifrom the bacterial surface, it was discovered thatwhereas the cells extracted with KSCN for I hwere completely devoid of pili (Plate 1), the surfaceof those prepared either by homogenization (Isaac-son, 1977) or heating (Morris et al., 1977) was stillassociated with significant numbers of pili. Also, thepercentage yield of K99 activity per mg of proteinwas the greatest when the cells were extracted withKSCN, the lowest yield being obtained by ultra-sonication (Table 1). It was decided therefore to

devise procedures for the purification of K99 piliextracted with KSCN in order to characterize theirphysicochemical and biological properties.At first the kinetics of the release of protein and

K99 activity from enterotoxigenic E. coli (C 1443)by the KSCN extraction procedure, as judged byhaemagglutination of horse red blood cells, wasdetermined. Fig. 1 shows that the total proteinreleased was essentially the same after ih and 1 h ofKSCN extraction. If the extraction was allowed toproceed for any further length of time, the totalamount of protein released from the bacterial cellswas increased, although the haemagglutinatingactivity actually decreased, possibly due to therelease of as-yet-unidentified haemagglutinationinhibitors. In this regard it would be interesting toexamine the possible correlation, if any, between therelease of carbohydrates, lipopolysaccharide, DNAand RNA with the increased release of protein after1 h. All available K99 activity had apparently been

60

loEo 40

0.

0F-

oI

-----O-,0 -----

0~~ ~ ~~~--0 0 --

20 [-

'f ., I I

0 1 2Period of extraction with KSCN (h)

-.J,3

cn60 r-

0CZCo._

0)to40 ZEcONON

20

II

Fig. 1. Kinetics of the release of total K99 activity (0)and proteins (o) as a function of time employedfor theextraction of enterotoxigenic E. coli (C1443) with

3 M-KSCNThe K99 activity was measured by haemagglu-tination of horse red blood cells.

Table 1. Comparative efficiency of various methods in affecting the release of K99 activity from enterotoxigenicEscherichia coli

Abbreviations used: H.A., haemagglutination units; P.U., precipitation units; N.D., not detectable.

Method of extraction(i) KSCN(ii) Ultrasonication(iii) Homogenization(iv) Heating broth culture

at600C

Totalprotein(mg)35.84

426.2100.6

1860

TotalH.U.*716801536064000240000

H.U. permg of protein

2000360636129

TotalP.U.*1040080005760N.D.

P.U. permg of protein

29018.757N.D.

ReferenceThe present paperThe present paper(Isaacson, 1977)(Morris et al., 1977)

* Based on activity released from either 10 litres of Minca-grown E. coli (5 x 109 bacteria/ml) (i-iii) or 6 litres of broth(TSB)-grown E. coli (1 x 1010 bacteria/ml).

1982

508

The Biochemical Journal, Vol. 20 1, No. 3 Plate 1

(b)

.. Wa 4 ............. .ii4-.r X.?>;'."w@,.,g'.48.>~~~~~11EXPLANATION OF PLATE 1

Electron micrographs ofnegatively stainedpreparations(a) Untreated enterotoxigenic E. coli (C1443) (magnification x35000); (b) enterotoxigenic E. coli (C1443) afterextraction with 3M-KSCN (x 35000); (c) crude KSCN extract rich in K99 pili (x 136000); (d) purified K99 pili(x 136000). P denotes pili and the bars equal lum (a, b) and 0.25,um (c, d).

K. ALTMANN, N. A. PYLIOTIS AND T. K. S. MUKKUR (facingp. 508)

K99 pili from enterotoxigenic E. coli

extracted within the first h. The amount of proteinreleased was much greater (ranging from 3- to1 2-fold) when other methods were used for thepreparation of pili (Table 1).When KSCN was previously used to extract

protective antigens from P. multocida type A(Mukkur, 1979; Mukkur & Pyliotis, 1981), it wasdiscovered that such extraction resulted not only inthe release of proteins, carbohydrates, lipopoly-saccharides but also DNA and RNA. Therefore wedecided to determine if, in addition to the protein,other macromolecules were also present in the crudeKSCN extract. All the above-mentioned macro-molecules were detectable in the crude KSCNextract prepared from enterotoxigenic E. coli(C 1443). These components were not quantifiedexcept where indicated. Since recently it wasreported (Clements & Finkelstein, 1979) that thebest source for the isolation of heat-labile entero-toxin was the whole-cell lysate, it was consideredimportant to determine its presence in the KSCNextract. By using the y- 1 adrenal-tumour-cell assay (Donta et al., 1974), the heat-labileenterotoxic activity ranging from 1/4 to 1/16 wasdetected.The initial method attempted for the purification

of K99 pili from the KSCN extract of entero-toxigenic E. coli (C 1443) was that reported byIsaacson (1977), which involved precipitation ofK99 pili with (NH4)2SO4 to a final saturation of14%, discarding the precipitate, followed by asecond precipitation of the supernatant with(NH4)2SO4 to a final saturation of 31%. However,when using this procedure we discovered thatapprox. 40% of the K99 activity was non-precipit-able, as judged by haemagglutination of horse redblood cells and by the Ouchterlony gel-diffusion testusing specific K99 antisera. However, using(NH4)2SO4 to a final saturation of 500/o, approx.90% of the available K99 activity was precipitated.In this regard, it is noteworthy that Wevers et al.(1980) recently reported precipitation of pili asso-ciated with enterotoxigenic E. coli of human origin(0: 18ac) with only 10% final saturation of(NH4)2SO4. This step of 'salting out' resulted in1.7-fold purification as judged by the increase in

specific activity per mg of protein (Table 2). Thelipopolysaccharide, DNA, RNA and the heat-labileenterotoxin were still detectable in the enrichedK99-pili preparations, however. In order to removethe heat-labile enterotoxic activity, the latter werenow subjected to batchwise absorption with Bio-GelA5m equilibrated with TEAN buffer (Clements &Finkelstein, 1979) until no such activity was detectedat a final concentration of 150,ug of protein/ml.After this step, the pili-containing preparation wassubjected to preparative isoelectric focusing usingPharmalyte (pH range 3-10). A typical isoelectric-focusing profile is shown (Fig. 2). After removalof the Pharmalyte and concentration of the frac-tions, they were analysed by the gel-diffusion testusing specific K99 antisera, and by immunoelec-trophoresis using unabsorbed C 1443 OK antisera.Specific K99 activity was found to be localized in thefirst seven fractions, the greatest activity beingassociated with fraction number 2. On immuno-electrophoresis, a single cationic arc was discernibleon development either with unabsorbed C 1443 OKor rabbit anti-(crude KSCN extract) antisera.However, on SDS/polyacrylamide-gel electro-phoresis, whereas fractions 1-3 showed a singleband, the molecular weight of which corresponded to12 600 + 300 (Figs. 3 and 4), fractions 4-7 showed aminimum of five additional bands with mol.wts.greater than 39 800 (result not shown).On immunoelectrophoresis of the crude KSCN

extract with unabsorbed C 1443 OK antisera, onefaint and two dense arcs were observed. Whereasone of the dense arcs was cationic, the second wasanionic. The faint arc was localized in line with thewell (Fig. 5). The anionic component was found tobe distributed in an immunoelectrophoretically purestate in fractions 8-29. Fraction 30 was found tocontain both the anionic as well as the thirdcomponent represented by the faint arc. Thepresence of the latter component in fraction 30 wassurprising in view of its position in the immuno-electrophoretic profile; the reason for this behaviouris not clear at present.

Anionic antigen titration was accomplished byimmunoelectrophoresis of serially diluted fractions8-30 with unabsorbed C 1443 OK antiserum. A

Table 2. Purification ofK99from enterotoxigenic E. coli

StepCrude KSCN extract(NH4)2SO4 precipitationBio-Gel A5m absorptionIsoelectric focusing (pure pili)

Total protein(mg)54.430.015.02.9

TotalP.U.*22 0002000016 00011200

Specific activity(P.U./mg)t

40466710673862

* P.U., precipitin units.t Data are based on purification from 18 litres of Minca-grown E. coli (5 x 109 bacteria/ml).

Purification(fold)

1.72.69.6

Yield(%)100917351

Vol. 201

509

K. Altmann, N. A. Pyliotis and T. K. S. Mukkur

21

15Fraction no.

30

6

0.6

4 r-

l._

2 -- 0.3

o

JO

Fig. 2. Isoelectric-focusing profile of (NH4)2S04 precipitate of crude KSCN extract of enterotoxigenic E. coli (C1443),obtained bv using Pharmalvte, pH3-10

A, A280; 0, log2 (precipitin titre) obtained by Ouchterlony gel-diffusion test using specific K99 antisera fordevelopment; 0 log2 (precipitin titre) of the anionic component obtained immunoelectrophoretically with unabsorbedenterotoxigenic E. coli (C 1443) antisera for development; O, pH gradient.

bimodal pattern of activity represented by peaks IIand III (Fig. 2) was revealed. However, the Ouchter-lony gel-diffusion test on peaks II and III revealedlines of complete identity on development withunabsorbed C 1443 OK antiserum (result not shown).The yield of K99 pili was 51% on the basis of

specific K99 activity/mg of protein (Table 2). Thisyield was significantly greater than that reported byIsaacson (1977). The total purification achieved was9.6-fold.

The introduction of gel filtration on Bio-Gel A-5mas an additional purification step after Bio-Gel A-5mabsorption resulted neither in any increased reso-lution nor in the purification of K99. In fact, anapprox. 4-fold decrease in the final yield of K99 piliwas the result (results not shown). Whether thelosses incurred were due to non-covalent interactionbetween the galactose residues of Bio-Gel A-5m orto non-specific absorption remains to be deter-mined.

Although the Pharmalyte used in the isoelectricfocusing of K99-pili-rich preparations was in the pHrange 3-10, a gradient shift was always recorded.That this shift did not result from the presence ofmannose or the buffer system used in the iso-electric-focusing run was revealed when a controlrun performed under the same conditions showed asimilar gradient shift. Regardless of differences in thepH gradient, maximum K99 activity always wasconfined to the fraction with a pH between 8 and 9(Fig. 2).The cationic characteristic of purified K99 pili

was consistent with a fairly recent report by Morriset al. (1980), who stated that the pili belonging togroup 0:20 were cationic. This was supported bythe finding that the intact enterotoxigenic E. coli

(C 1443) and purified pili caused a strong mannose-resistant haemagglutination of horse, but not ofsheep or guinea-pig, red blood cells.The s20 w of purified pili was calculated to be

12.2S at a concentration of 0.42mg/ml, the lattervalue being in good agreement with that reported byIsaacson (1977). Electron microscopy of purified pilirevealed that they were aggregated not only in aparallel arrangement but also in a linear fashion asjudged by an increase in the average pilus length as afunction of increased protein concentration (Fig. 6).Whereas the diameter of an individual pilus wasmeasured to be 7.4 + 0.6 nm, the length at zeroconcentration was estimated to be approx. 160nm(Fig. 6). These data are in good agreement withthose reported by Isaacson (1977).No lipopolysaccharide, DNA, RNA or heat-labile

enterotoxic activity were detectable in the purifiedK99-pili preparations. However, the amount of totalcarbohydrates was estimated at 51 ug/mg of pro-tein.On Ouchterlony gel diffusion, the purified K99 pili

showed a reaction of complete identity with acomponent in the KSCN extract of the K99reference strain (B41), on development both withspecific K99 as well as with the unabsorbed C 1443OK antisera (results not shown). This presumablyrepresented the presence of the cationic K99 pili(Morris et al., 1980) in the KSCN extract of strainB4 1.

Isaacson (1977) reported K99 pili to comprisetwo subunits: a major component with a mol.wt. of22500 and a minor component with a mol.wt. of29500. Because of the discrepancy, it was con-sidered important to ascertain that the lowermolecular weight of the K99 subunit obtained in this

1982

510

'o

K99 pili from enterotoxigenic E. coli

Bovine serum albumin monomer

OvalbuminiAdolaseXPepsin

i RibonucleaseAK99Cytochrome c

I I0 20 30 40 50 60 70 80 90 100

Mobility relative to Bromophenol Blue (%)

Fig. 4. Determination of the molecular weight ofpurifiedK99 pili by SDS/polyacrylamide-gel electrophoresisThe molecular weights of the marker proteins are:

bovine serum albumin, 68000; ovalbumin, 45000;aldolase subunit, 40000; pepsin, 35000; subtilisin,27500; chymotrypsin, 25000; trypsin, 23 800;ribonuclease A, 13 700; cytochrome c (horse heart),12100.

(a) (b) (c)

Fig. 3. SDS/polyacrylamide-gel electrophoresis of (a)the crude KSCN extract (100ug), (b) (NH4)2SO4 pre-cipitate of the crude KSCN extract (100,ug) and (c)

purified K99 pili (100 pg)The letter D denotes the dye marker; the arrow

points to the K99 subunit.

investigation was not a consequence of the methodused for the extraction of K99 pili. Therefore, crudeextracts containing K99 pili prepared by the methodof Isaacson (1977) and Morris et al. (1977) andpurified K99 pili (the present investigation) were

subjected to immunoprecipitation using specific K99antiserum. A 0.5 ml portion of various crudepreparations was mixed with 0.5 ml of specific K99antisera and incubated at 40C for 48h. Theprecipitates were washed once with ice-cold saline(0.9% NaCI), dissolved in 1% SDS, and subjected toSDS/polyacrylamide-gel electrophoresis. In allcases, the molecular weight of the precipitated K99pili was found to be 12 600 (result not shown).Therefore it is probable that the K99 pili purified byIsaacson (1977) represented either a different type of

Vol. 201

(a)

(b)

Fig. 5. Immunoelectrophoretic patterns of the crudeKSCN extract of enterotoxigenic E. Coli (C1443) and

purified K99 piliThe antisera used for development were (a) specificK99 antiserum and (b) unabsorbed enterotoxigenicE. coli (C 1443) antiserum. Wells 1 and 4 containedpurified K99 pili, wells 2 and 3 contained the crudeKSCN extract.

pili or partially purified pili. In this regard it isnoteworthy that heterogeneity of molecular weightsin the case of colonization factor antigens I and IIwas recently reported (Wevers et al., 1980).The general similarity observed between the

amino acid composition of K99 pili reportedpreviously by Isaacson (1977) and that determinedin this laboratory (Table 3) was in the order in which

4.6 F:3

Co

C-

00

0.2

4.3 -

511

4.9r

512 K. Altmann, N. A. Pyliotis and T. K. S. Mukkur

1000

800

600

00400

200

0 20 40 60 80 100LProteinl (ug/mi)

Fig. 6. Determination of the length of K99 pili as afunction ofincreasing protein concentration

Table 3. Amino acid composition ofpurified K99 isolatedfrom enterotoxigenic Escherichia coli (C1443)

Content (mol of amino acid residue/Amino acid mol of K99 pili)

Lysine 6Histidine 2Arginine 3Aspartic acid 13Threonine 10Serine 12Glutamic acid 9Proline 4Glycine 16Alanine 10Cysteine NilValine 5Methionine NilIsoleucine 5Leucine 7Tyrosine 3Phenylalanine 5Tryptophan 2

the percentages of various groups of amino acidresidues were present, viz., aliphatic>dicarboxylicand hydroxy amino acids> basic amino acids.However, significant differences observed were thatwhereas hydroxylysine, cysteine and methioninewere absent, tyrosine and phenylalanine were pre-sent in our purified K99 pili preparations. Since ahigh proportion of acidic amino acid residues isuncharacteristic of proteins with cationic isoelectricpoints, it was assumed that most of the aspartic andglutamic acid residues were present as.asparagineand glutamine before acid hydrolysis. The molecularweight of purified K99 was calculated to be 13687from its amino acid composition, which appeared tobe in reasonable agreement with that obtained bySDS/polyacrylamide-gel electrophoresis (12 600 +

300). Whether more than one type of K99 pilusexists in Nature will require isolation and immuno-chemical characterization of K99 pili from additionalenterotoxigenic E. coli strains.Because of the ease with which K99 pili could be

extracted with KSCN, we decided to determinewhether pili from enterotoxigenic E. coli of porcine(e.g. K88) and human (e.g., colonization-factorantigens) origin could also be extracted using thesame procedure. Electron microscopy of oneKSCN-treated enterotoxigenic E. coli of porcine(P155) and human origin (H10407) revealed thatthese pili were extractable. Similarly, pili of Fusi-formis nodusus (the bacterium causing foot-rot insheep) was also found to be extractable with KSCN.Therefore it appears feasible that this method ofextraction of pili may be applicable to other piliatedmicro-organsms as well, e.g., Moraxella bovis,Neisseria gonorrhoea and some Salmonella species.

This project was supported by a grant from theAustralian Dairying Research Committee made toT. K. S. M. The excellent technical assistance of Mrs. D.Stewart is gratefully acknowledged. Also acknowledged isthe part-time assistance of Ms. H. Thomas. We arethankful to Mr. Berto Reibero, Department of VeterinaryMedicine, University of Sydney, for donating piliatedcultures of Fusiformis nodosus. Finally, we thank Mr. L.Higginbottom for performing the amino acid analysis andDr. Alex Reisner (C.S.I.R.O. Cellular and MolecularBiology Unit, North Ryde, N.S.W., Australia) forpermitting us to use their Beckman analytical ultra-centrifuge.

ReferencesAcres, S. D., Isaacson, R. E., Khachatourians, G.,

Babiuk, L. & Kapitany, R. A. (1978) in Proceedings ofthe Second International Symposium on NeonatalDiarrhoea (Acres, S. D., ed.), pp. 433-455, Vido,Saskatoon

Bradford, M. M. (1976) Anal. Biochem. 72, 248-254Burrows, M. R., Sellwood, R. & Gibbons, R. A. (1976) J.

Gen. Microbiol. 96, 269-275Ceriotti, G. (1952) J. Biol. Chem. 198, 297-303Clements, J. D. & Finkelstein, R. A. (1979) Infect.

Immun. 24, 760-769Donta, S. T., Moon, H. W. & Whipp, S. C. (1974)

Science 183, 334-336Edelhoch, H. (1967) Biochemistry 6, 1948-1954Evans, D. G. & Evans, D. J., Jr. (1978) Infect. Immun.

21, 638-647Evans, D. G., Silver, R. P., Evans, D. J., Jr., Chase, D. G.& Gorbach, S L. (1975) Infect. Immun. 12, 656-667

Evans, D. G., Evans, D. J., Jr. & DuPont, H. L. (1977)J.Infect. Dis. 136, S 118-S 123

Evans, D. J., Jr., Evans, D. G. & DuPont, H. L. (1979)Infect. Immun. 23, 336-346

Guinee, P. A. M., Jansen, W. H. & Agterberg, C. M.(1976) Infect. Immun. 13, 1369-1377

Isaacson, R. E. (1977) Infect. Immun. 15, 272-279Isaacson, R. E. (1978) Infect. Immun. 22, 555-559

1982

K99 pili from enterotoxigenic E. coli 513

Isaacson, R. E., Morgan, R. L., Moon, H. W. & Brinton,C. C. (1977) in Proceedings of the 13th JointU.S.-Japan Conference on Cholera, Atlanta, GA, pp.285-293, National Institutes of Health, Bethesda

Jones, G. W. & Rutter, J. M. (1974) J. Gen. Microbiol.84, 135-144

Mokrasch, L. C. (1954) J. Biol. Chem. 208, 55-59Moon, H. W., Whipp, S. C. & Skartvedt, S. M. (1976)Am. J. Vet. Res. 37, 1025-1029

Morris, J. A., Stevens, A. E. & Sojka, W. J. (1977) J.Gen. Microbiol. 99, 353-357

Morris, J. A., Stevens, A. E. & Sojka, W. J. (1978) Infect.Immun. 19, 1097-1098

Morris, J. A., Thorns, C. J. & Sojka, W. J. (1980) J. Gen.Microbiol. 118, 107-113

Mukkur, T. K. S. (1979) J. Gen. Microbiol. 113, 37-43Mukkur, T. K. S. & Pyliotis, N. A. (1981) J. Comp.

Pathol. 91, in the pressMyers, L. L. & Guinee, P. A. M. (1976) Infect. Immun.

13, 1117-11190rskov, I., 0rskov, F., Smith, H. W. & Sojka, W. J.

(1975) Acta. Pathol. Microbiol. Scand. Sect. B 83,31-36

0rskov, I., 0rskov, F., Jann, B. & Jann, K. (1977)Bacteriol. Rev. 41, 667-7 10

Osborn, M. J. (1963) Proc. Natl. Acad. Sci. U.SA. 50,499-506

Scheidegger, J. J. (1955) Int. Arch. Allergy Appl.Immunol. 7, 103-109

Sivaswamy, G. & Gyles, C. L. (1976) Can. J. Comp.Med. 40, 247-251

Smith, H. W. & Linggood, M. A. (1972) J. Med.Microbiol. 5, 243-250

Sojka, W. J. (1965) Escherichia coli in Domestic Animalsand Poultry, Review Series no. 7, CommonwealthBureau of Animal Health, Weybridge, Surrey,U.K.

Sojka, W. J., Wray, C. & Morris, J. A. (1978) J. Med.Microbiol. 11, 493-499

Weber, K. & Osborn, M. (1969) J. Biol. Chem. 244,4406-4412

Weissbach, A. & Hurwitz, J. (1959) J. Biol. Chem. 234,705-709

Wevers, P., Picken, R., Schmidt, G., Jann, B., Jann, K.,Golecki, J. R. & Kist, M. (1980) Infect. Immun. 29,685-691

Vol. 201