<

INFECTION AND IMMUNITY. June 1982. p. 996-1005 Vol. 36. No. 30019-9567/82/060996-10502.00/0

Purification and Biological Characterization of Shiga Toxinfrom Shigella dysenteriae 1

js J. EDWARD BROWN,* DARRELL E. GRIFFIN, SARA W. ROTHMAN. AND B. P. DOCTOR

Department of Biological ChemistrY, Walter Reed Army Institute of Research, Washington, D.C. 20012

Received 14 December 1981/Accepted 16 February 1982

Shiga toxin has been purified in milligram quantities to near homogeneity fromIS .cell lysates of Shigella dysenteriae I strain 3818-0. Purification involved an initial

ultracentrifugation, ammonium sulfate fractionation, chromatography on DEAE-cellulose and carboxymethyl cellulose, gel filtration, and preparative isoelectricfocusing in sucrose gradients. The purified toxin was resolved by discontinuouspolyacrylamide gel electrophoresis into a major cytotoxic protein band and aclosely migrating, cytotoxic protease-nicked minor band. Antiserum generated byimmunization with glutaraldehyde-inactivated toxin was shown to be monospeci-fic against S. dysenteriae cell lysates. This highly purified toxin was cytotoxic to"HeLa cells, enterotoxic in rabbit ileal loops, and lethal to mice. Monospecificantiserum to the toxin neutralized completely these toxin activities in bothpurified toxin preparations and crude shigella cell lysates.

Shigella dysenteriae 1, P causative agent of column chromatography (18). These attemptsshigellosis, has been known for about 80 years to provide microgram quantities of toxin, physical-produce a protein toxin (6). The role of Shiga ly detectable only after radioiodination duringtoxin in pathogenesis is still not understood. the purification procedure.Nonetheless, involvement of toxin has been We have purified Shiga toxin to near homoge-proposed both in the diarrheal (13) and in the neity in milligram amounts from cell lysates of S.invasive aspects of this disease (11). Bacterial dysenteriae 1 by using a combination of ion-invasion of the colonic epithelium is a recog- exchange chromatography, gel filtration chro-nized feature in the pathogenesis of shigellosis matography, and preparative isoelectric focus-(15). We considered that further understanding ing. A preliminary report of this purificationof the role of toxin in disease required its purifi- procedure has been made (J. E. Brown. M. K.cation and characterization. Gentry, D. E. Griffin, S. W. Rothman, W. J.

Several biological activities have been associ- Cahillane, B. P. Doctor, and M. R. Thompson.ated with Shiga toxin. It is: (i) lethal to rabbits Abstr. Annu. Meet. Am. Soc. Microbiol. 1979,(20) and mice (25), (ii) cytotoxic to several cell B42, p. 22). The results clearly demonstrate thatlines (9, 12, 14, 26), (iii) enterotoxic in that it all of the biological activities associated withcauses fluid accumulation in rabbit ileal loops Shiga toxin are present in the purified protein.(13), and (iv) able to inhibit cell-free proteinsynthesis in both mammalian (24) and bacterial(19) systems. Inhibition of protein synthesisappears to be the primary cytotoxic event in Prparation of Inittal cell lysate. S. dysenteriae 1

strain 3818-0 (11) was grown and processed at the Newwhole cells (2). Such inhibition of cell-free pro- England Enzyme Center to yield a lysate which served> tein synthesis is apparently catalytic and in- as the initial toxin extract of our purification proce-0... volves an inhibition of the elongation phase of dure. The lysate was prepared as follows. A 500-liter

0 translation (3). fermentor containing modified syncase medium (2)C.) Olitsky and Kligler first demonstrated Shiga was inoculated with a 1% inoculum. and bacterial

toxin to be distinct from endotoxin in culture growth was allowed to continue for 18 h at 37°C withLii filtrates of shigellae (20). Early attempts at isola- maximum aeration. After the cells were harvested in a

I : . tion of Shiga toxin involved either extraction at Sharpies centrifuge (size 16), the cell pellet was sus-

pH 11 from heat-killed bacteria (25) or enrich- pended at 4"C in 1.5 parts of lysis buffer (50 mM Tris-Sment from 24-h culture filtrates by means of hydrochloride-5O mM KCI-10 mM magnesium acetate

ultrafiltration and isoelectric focusing on poly- IpH 81 containing 10 pg of phenylmethylsulfonyl fluo-lraflrlami ande ge s (1. either proceusingeon poyi ride per ml). The suspension was passed twice through

m acrylamide gels (17). Neither procedure yielded aMnt-aunmoe MTAlbrtryhog•pure toxin. More recent reports on purification a Manton-Gauhin model I5M8TBA laboratory homog-enizer. Between pasa.2ges the suspension was cooled

"have described procedures using adsorption to on ice to 15"C. This suspension was centrifuged againacid-treated chitin columns (21) or antitoxinity - the Sharpies centrifuge, and the resulting superna- Y

J A482 07 22 034

St . -.. - .. . .. - .. -

VOL. 36, 1982 SHIGA TOXIN PURIFICATION AND CHARACTERIZATION 997

TABLE 1. Purification of Shiga toxin from cell lysates

Procedure Vol (nl) Cytotoxicity" Protein Sp act Purification YieldPCD,Wml) (mg/ml) CD),4mg) (fold)

Culture broth 250,000 5.7 x 10' 0.46 1.2 X l0`Cell lysate 3,800 170 x 105 52 3.3 x 105 1 100Centrifugation. 2.650 33 x 10' 28 1.2 x 10' 0.36 13

type 21 rotor(NH 4 )zSO4. 28 to 1.225 190 x 101 56 3.4 x 10' 1.0 35

50% saturationDEAE-Sephacel 3,230 60 x 105 2.45 32 x 10' 9.7 39CM-52 355 220 x 10' 1.0 110 x t0o 33.3 12(NH 4 )2SO4, 70C91 18.4 4.600 x 105 31.3 147 x 105 44.5 13

saturationSephacryl S-200 59 9"0 x 10' 2.81 352 x 10' 107 8.9Isofocusing, pH 5 to 8 24 980 X 10' 1.2 820 x 10' 250 3.6

"CDo is Lhe amount of toxin activity required to reduce .ye staining to half the control values, as described inthe text.

tant fluid was centrifuged at 84,000 x g twice through plot of dye absorbance versus the dilution of toxinan ElectroNucleonics model R-K continuous flow allowed determination of the dilution yielding 50% cellcentrifuge. The first passage was at a flow rate of 30 detachment. With this assay, a unit cytotoxic dose isliters per h, and the second was at a flow rate of 10 defined as that amount of toxin which causes a 50%liters per h. The clarified supernatant lysate was reduction of the dye retained in a microtiter well (12).frozen at -70'C until processed. Protein determinations. Protein concentrations were

Buffers. Various buffers were used during purifica- determined by the method of Lowry et al. (16) withtion and characterization. These common buffers are bovine serum albumin (BSA) as a standard.referred to as follows. Buffer 20T8/100 consisted of Preparation of antitoxin. Rabbit antisera to Shiga0.020 M Tns-hydrochloride (pH 8.0) containing 0.10M toxin were prepared as described below. Purified toxin

NaCI; buffer 20T8/250 consisted of 0.020 M Tris- (55 lxg/ml) was incubated at 37°C in 0.11% glutaralde-hydrochloride (pH 8.0) containing 0.25 M NaCI: buffer hyde in 0.1 M sodium phosphate buffer (pH 8) for 3010P6/50 consisted of 0.010 M sodium phosphate (pH min at 37"C. Lysine was added to a twofold molarS6.0) containing 0.050 M NaCI; buffer 10P6/350 consist- excess over glutaraldehyde. The resulting toxoid solu-

ed of 0.010 M sodium phosphate (pH 6.0) containing tion was divided into fractions (0.5 ml) containing 100.35 M NaCt; buffer 130G8 consisted of 0.13 M gly- lig of toxin protein and frozen until used. Initialcine-NaOH (pH 8.0). injections were made intradermally at eight different

Cytotoxieity assays. Cytotoxic activity was used to sites on the backs of rabbits by using 10 Kg of toxoidmonitor Shiga toxin enrichment during purification. emulsified in an equal volume of Freund completeThe extent of HeLa cell detachment was employed as adjuvant. At 2-week intervals, animals were giventhe indicator ofcytotoxicity for Shiga toxin (12). HeLa subcutaneous booster injections of 10 lig of toxoidcells (line CCL2: Flow Laboratories, Rockville. Md.) without adjuvant.were maintained at 35°C in growth medium consisting PAGE. Nondenaturing polyacrylamide gel electro-of Eagle minimum essential medium with Earle salts phoresis (PAGE) was performed in 8% rod or slab gelssupplemented with 10% heat-inactivated fetal bovine by the method of Davis (7). Gels were fixed in 10%serum. 2 mM glutamine. 180 U of penicillin per ml. and sulfosalicylic acid, stained with 0.02% Coomassie bril-0.18 mg of streptomycin per ml in a 5% CO, atmo- liant blue R250 in 7% acetic acid, and destained withsphere (HEM Research, Rockville, Md.). To establish ethanol-acetic acid-water (25:8:67).monolayers, freshly trypsinized cells were suspended 1mmunoblotting of PAGE slabs. Immunological anal-at a concentration of 1.6 x 10- cells per ml in growth ysis was performed by the Western blot method ofmedium, and 0.10-ml samples were dispensed into 96- Burnette (4). Electrophoresis was performed as abovewell microtiter plates (Costar, Cambridge, Mass.). in nondenaturing polyacrylamide slab gels. Proteins inCells were allowed to attach for 18 to 20 h before the gel were transferred electrophoretically to nitrocel-

experimental use. Serial dilutions of samples were lulose (BA 85, 0.45 (l.m; Schleicher & Schuell. Keene.added (0.10 ml), and plates were incubated for an N.H.) by using a gel destainer (E-C Apparatus Corp..additional 18 h. The endpoint of toxin activity was Jacksonville, Fla.) with a palladium anode. The elec-determined by fixing and staining with crystal violet- trophoretic transfer was performed at a constant volt-formaldehyde solution. Stained cell monolayers were age of 16 V for 14 to 16 h. Briefly, the nitrocellulosedissolved in 50% ethanol containing 1% sodium dode- was incubated sequentially in buffer containing BSA.cyl sulfate. The absorbance (595 nm) of the extracts buffer containing BSA and antiserum (1:50), severalwas determined either with a spectrophotometer or washes to remove unbound antibody, buffer contain-with a microtiter plate colorimeter (Division of Instru- ing BSA and lI-labeled protein A (4 x 10' cpm/ml),

mentation. Walter Reed Army Institute of Research: and final ,.sli.s to remuv unbound ' 25 1-protein A.from a design provided by Robert Yolken, Johns Autoradiography was performed at -70"C with X-

, 'Hopkins University, Baltimore, Md.). A logarithmic Omat AR film (Eastman Kodak Co.. Rochester, N.Y.)

A

m~ ~ ~ ~ ~ ~ ~ ~~1 W altlN am mm-

998 BROWN ET AL. INFECT. IMMUN.

4

ECo __

N3

Co

-- '-•;_[FIG. 1. DEAE-cellulose chromatography of toxin-containing ammonium sulfate fraction. Four columns wererun simultaneously, each loaded with 17 g of protein. Elution was carried out as described in the text. The eluantwas monitored for absorbance at 280 nm. Every sixth fraction was assayed for cytotoxicity in the HeLa cell

microtiter assay (fivefold serial dilutions). Toxin-containing fractions, as indicated by the bar, were collectedfrom each column and pooled for further processing. Fraction size was 12 ml.

*! and Cronex Lightning Plus intensifying screens (Du neutralizing activity was determined by measurement .

Pont Co.. Wilmington, Del.). of that dilution of antiserum which would completely :!Animal amasys. Enterotoxic activity was demonstrat- neutralize the cytotoxicity of a constant amount ofS~ed by using the rabbi• ileal loop model (10). Toxin Shipa toxin. Serial dilutions of antiserum were pre-

• lethality to mice was assayed by intraperitoneal injec- pared and mixed with an equal volume of toxin solu-•= ~tion of various concentrations (at least five) of toxin tion. After incubation for 1 h at 370C, 0.1 ml of eachi- into 15- to 20-g WR strain mice (four or more mice for mixture was added to duplicate HeLa cell monolayers.

each amount of toxin). From cumulative deaths for 10 After overnight incubation, the monolayers were fixedi• days, 50% lethal doses were calculated by the method and stained as described abOve. Endpoints were deter-

of Reed and Muench (22). For both assays. toxin was mined visually and by measurement with the micro- _serially diluted with 0.9% saline containing I mg of titer plate colorimeter. The endpoint was defined as .

iBSA per ml. that dilution of antibody which completely neutralized ,Nesitrullzatlon of bIohlqoeal aetlvity. The cytotoxin- the toxin.. -

M 2'

VOL. 36, 1982 SHIGA TOXIN PURIFICATION AND CHARACTERIZATION 999

2.4 H a flow rate of 3.5 ml/min. The dialyzed solutionwas divided into four equal portions of 300 ml.

_2.0 and each was applied simultaneously to one of* four separate DEAE-cellulose columns (DEAE-o 6.- Sephacel; Pharmacia Fine Chemicals, Inc.. Pis-4 cataway. N.J.) equilibrated with 20T8/100 buff-

1 .2 er. The columns were eluted with 20T8/100buffer. A typical elution profile is shown in Fig.

c 0.8- 1, with cytotoxic activity shown below the ab-0 sorbance profile. Cytotoxicity was not detected

S0.4in the initial absorbance peak, but was retained0.4 I slightly and eluted as a broad band after the

initially eluted protein. After elution of 6 column20 60 100 40 18 340 30 420 volumes, the buffer was changed to 0.25 M NaCI

FRACTION in 20T8 buffer. No further cytotoxin was detect-able in this eluate. The toxin-containing frac-

FIG. 2. Carboxymethyl cellulose chromatography tions were pooled and dialyzed overnight againstof toxin-containing DEAE-cellulose fractions. A totalof 7.8 g of protein was applied to a single column, and 10P6/50 buffer.elution was carried out as described in the text. The Carboxymethyl cellulose chromatography.arrow indicates the shift to elution with 10P6/350 Chromatography was performed using carboxy-buffer. Absorbance (280 nm) of fractions was mea- methyl cellulose (CM-52; Whatman Ltd., Clif-sured. Cytotoxicity was assayed as in Fig. 1. The bar ton, N.J.) equilibrated with 10P6/50 buffer in aindicates the fractions which contained cytotoxicity. 9.6- by 5-cm column at a flow rate of 10 ml/min.Fraction size was 20 ml. The toxin-containing fraction recovered from

DEAE-cellulose chromatography (>3 liters) wasNeutralization of enterotoxic activity was tested by pooled, dialyzed, and applied to a single column.

inoculation of toxin-antiserum mixtures into ligated All cytotoxic activity was found to be retainedsections of rabbit ileum. Toxin-containing solutions by the column, whereas most (90%) of thewere mixed with antiserum and incubated at 0°C for 30 protein was eluted (Fig. 2). After the columnmin. Ileal loops were injected with 1-ml volumes, was washed with one sample volume of 10P6/50Toxin controls were diluted with normal rabbit serum buffer, the toxin was eluted with 10P6/350 buff-or 20T8/100 buffer.

Neutralization of mouse lethality was tested by er. The toxin-containing eluant was pooled, andintraperitoneal injection of toxin-antiserum mixtures. ammonium sulfate was added to 70% saturation.Toxin-containing solutions were mixed with antiserum After 4 h at 4°C, the precipitate was collected byand incubated at 0°C for 30 min. Mice were injected centrifugation in a GSA rotor at 10,000 rpm forwith 0.2-ml volumes. Diluted antiserum and 20T8/100 60 min. The pellet was suspended and dialyzedbuffer alone were used as controls.

RESULTS

Initial enrichment. Cell lysates served as thestarting material in our purification, since these -

lysates contained more total cytotoxic activity _and more activity per milligram of protein than 2 2

did culture filtrates (Table 1). Approximately 4liters of thawed lysate was subjected to centrifu- •

.4 gation at 20,000 rpm overnight in Beckman type21 rotors. The supernatant fluid was collected,and ammonium sulfate (Ultrapure; Schwartz/Mann, Orangeburg, N.J.) was added to 28%saturation. After 2 h at 4"C, the suspension was &centrifuged in a GSA rotor (Sorvall) at 10,000 0 15 2o is 3o 3 40 45 50

rpm for 30 min, and ammonium sulfate was FRACTIONadded to the supernatant fluid to achieve 50% FIG. 3. Gel filtration chromatography of toxin-con-saturation. After 4 h, the precipitate was collect- taining carboxymethyl cellulose fractions. A total of

ed by centrifugation as above, dissolved in 20T8/ 550m ng was applied to the Sephacryl S-200 column (3.3by 95 cm), and elution was performed as described in100 buffer, and dialyzed for 36 to 48 h at 4*C with the text. The eluant was monitored for absorbance at

multiple changes of buffer. All further steps 280 nm. Cytotoxicity of each tube was assayed as inwere at 4"C. Fig. 1. The bar indicates toxin-containing fractions

DEAE-eUllulome ehromatography. Chromatog- collected for further processing. Fraction size was 15raphy was performed in 15- by 5-cm columns at ml.

7__ _ _

it

1000 BROWN ET AL. INFECT. IMMUN.

80

"r 120 0

o 7"5C 0

"090

- 060 - 70

0 65

030 00 60

16 20 24 28 32 36 40 44 48

FRACTIONFIG. 4. Preparative sucrose gradient isoelectric focusing of toxin-containing S-200 fractions. Isoelectric

focusing was carried out as described in the text. Fractions were measured for pH and assayed for cytotoxicity.and absorbance (280 nm) was measured. The bar indicates the toxin-containing portion which was applied toSephadex G-25 to remove sucrose and ampholytes. Fraction size was 2 ml.

with 130G8 buffer overnight and centrifuged at ampholytes and sucrose Purified toxin was fro-10,000 rpm for 30 min in a Sorvall SS-34 rotor to zen at -70°C.remove insoluble material. Purity of the isolated protein. The procedure

Gel filtration. Sephacryl S-200 equilibrated described here provided 25 mg of purified Shigawith 130G8 buffer was poured in a column (3.3 toxin from approximately 4 liters of cell lysate.by 95 cm). The flow rate was adjusted to 2 ml/ Cytotoxic activity was enriched about 250-foldcm 2 per h, and the concentrated toxin sample at an overall recovery of about 3% (Table 1).was applied. The elution profile (Fig. 3) indi- Recovery and purification factor can vary con-cates two major protein peaks. The cytotoxic siderably between batches, depending on start-activity eluted as a narrow band with the second ing toxin concentrations and on timely process-absorbance peak. Calibration of this column ing during the purification. To ascertain thewith aldolase, hexokinase, BSA, ovalbumin, homogeneity of the isolated toxin, discontinuousand myoglobin gave ar Mr of 70,000 for the buffer PAGE was carried out in 5, 8, and 10% gelpartially purified toxin. Those fractions contain- rods (Fig. 5A). In each case Coomassie blueing cytotoxic activity were pooled for subse- staining showed two closely migrating proteinquent isoelectric focusing. bands. Densitometric analysis of the stained gel

Preparative isoelectric focusing. Since glycine indicated that 82% of the protein was in the slowis isoelectric and does not disturb the pH gradi- band and 17% was in the faster-migrating band.ent, the pooled sample (60 ml) from gel filtration To determine which band possessed cytotoxicwas applied directly to an electrofocusing col- activity, companion gels were sliced into 1-mmumn. The toxin sample in 130G8 buffer was slices which were eluted overnight for cytotoxic-mixed into a 0 to 20% sucrose gradient (110 ml) ity assays. Each of the bands detected by pro-containing 1.0% Ampholine (pH 5.5 to 8.5; tein staining showed cytotoxic activity. A slightLKB) in an LKB 8100 Ampholine column. lso- contaminant, reflecting less than 1% of the stain-electric focusing was carried out with constant ing intensity (Rf = 0.51, 8% gel), did not containvoltage for 24 h at 10°C, after which current had cytotoxic activity. Other preparations of Shigadropped to 2 mA. Collected samples (2 ml) were toxin could be separated by discontinuousassayed for cytotoxicity, absorbance (280 nm). PAGE into three or four closely migratingand pH (Fig. 4). The toxin activity was found as bands, suggesting that limited proteolysis of aa distinct peak at pH 7.2 in the pH gradient. No single native toxin molecule might cause thecytotoxic activity was observed elsewhere in the multiple banding pattern observed. This possi-pH gradient. Toxin-containing fractions were bility was tested by incubation of purified toxinapplied to a Sephadex G-25 column (2.5 by 30 with Streptococcus griseus protease (Fig. 5B).cm) equilibrated with 20T8/100 buffer to remove Clearly, the toxin is susceptible to proteolysis

.4 r

VOL. 36, 1982 SHIGA TOXIN PURIFICATION AND CHARACTERIZATION 1001

B3 dysenteriae cell lysates. When assayed againstpurified toxin, the antiserum (0.1 ml) at a 1:8dilution completely neutralized 780 ng of toxinprotein. The preimmune serum did not neutral-ize cytotoxicity (data not shown). To assess thespecificity of the antiserum, immunochemicalanalysis was performed using a Western blotimmunoassay technique (4). Toxin samples wereseparated by PAGE, transferred to nitrocellu-lose, and analyzed with serum collected at 0 and6 weeks (Fig. 7A). As early as 2 weeks post-immunization, specific antibody was detected(data not shown). At 6 weeks post-immuniza-tion, the antibody response had increased dra-matically (lanes 4 and 8): both of the closelymigrating toxin bands observed by Coomassie .

blue staining are readily detectable. The slightcontaminant in the preparation can be detectedwhen the autoradiograph is overexposed (lane8). To determine whether this antiserum couldbe employed as a monospecific reagent, thebanding pattern detected for pure toxin wascompared with that of S. dysenteriae cell lysates(lanes 3 and 7). The two closely migrating toxin

1 2 3 4 5 6 1 2 3 4 bands were readily detectable in cell lysates.However, several other protein bands reacted

FIG. 5. Discontinuous PAGE of purified Shiga tox- with the immune serum. With the exception ofin. (A) PAGE of two different preparations of toxin one band, these proteins in the cell lysate werewas performed in gel rods of either 5% (lanes 1 and 4), also detectable with preimmune serum Oane 5).8% (lanes 2 and 5), or 10% (lanes 3 and 6) polyacryl- Se cta ru m uro serum(lne5amide. The purified toxin described in Table I wasanalyzed at 5 ptg per gel (lanes 1, 2, and 3). A separate rabbits was examined by this immunoblot tech-preparation was analyzed at 15 j'g per gel (lanes 4. 5, nique. In each serum, antibodies against variousand 6). The dye front was marked by injection of bacterial cell lysate proteins could be detectedinsoluble dye. (B) PAGE of purified toxin in 8% gel (data not shown). Therefore, the antibodies toslab after treatment with S. griseus protease. Toxin these additional proteins were not generated bywas incubated with protease (1 Rg/ml) for 1, 5, and 30 immunization with purified toxin. The one un-min at 37°C, and phenylmethylsulfonyl fluoride was identified protein (Rf = 0.32) detected in celladded (60 ng/ml) to inactivate the protease. Each lane lysates by immune serum migrated to the samecontained 5 pkg of toxin. The dye front is shown byinjected insoluble dye. Lanes: 1. untreated: 2, 1 min: position as the fastest moving of the three toxin3, 5 min: 4, 30 min. species generated by treatment of purified toxin

by S. griseus protease (Fig. 7B). Therefore, thisband appears to be a proteolytic product of thenative toxin present in the cell lysate. Thus.

such that the predominant slowly migrating pro- when tested against S. dysenteriae cell lysates,tein band is readily converted into more rapidly this antiserum generated by immunization withmigrating species. However, incubation of toxin purified toxin behaved as a monospecific rea-with protease at 10 P.g/ml for 60 min did not gent.produce further detectable proteolysis (data not Toxicity of purified Shiga toxin. Purified toxinshown), suggesting that sites for proteolytic was assayed for enterotoxic activity and mouse"nicking on the native toxin are limited, lethality. The results in Table 2 demonstrate that

Production of monospeciflc antiserum with pu- purified Shiga toxin possesses all three biologi-rifled toxin. Antibody was developed in rabbits cal toxicities of shigella lysates. In the cytotoxic-by immunization with glutaraldehyde-inactivat- ity assay, toxin caused 50% cell detachment ated Shiga toxin. Antibody response during hyper- doses in the range of I pg per well. Fluidimmunization was monitored by using the neu- accumulation in rabbit ileal loops was observedtralization of cytotoxicity as the index of consistently at a dose of I Ag of toxin per loop.antibody titer. Neutralization was detectable at To obtain a value indicating a 50% dose level,2 weeks after immunization and remained con- we defined a 50% enterotoxic dose so that thestant after 4 weeks. Figure 6 demonstrates the data could be analyzed by the method of Reedneutralization both of purified toxin and of S. and Muench (22). A response was considered

r ! "4

1002 BROWN ET AL. INFECT. IMMUN.

2.0

•. .0 4C4

.01 L_ L

-9 -8 -7 -6 -5 -4 -3 -2 -I 0

LOG5 OF DILUTION

FIG. 6. Cytotoxicity neutralization by rabbit antiserum. The curves on the left demonstrate cytotoxicityassays of purified toxin (0) or crude cell lysates (0). Purified toxin was adjusted to 7.8 ixglml, and cell lysateswere adjusted to 720 lig/ml. Each was serially diluted (fivefold) with medium, and 0.05-ml volumes were appliedto cell monolayers. The curves on the right are neutralization assays with rabbit antiserum collected 6 weekspost-immunization. In these assays, 0.13 ml of serially diluted (twofold) serum was mixed with an equal volumeof purified toxin (0) or crude lysate (0): 0.1 ml of this mixture (390 ng of pure toxin or 36 gig of lysate protein)was applied to each HeLa cell monolayer. Values are the mean of three determinations with the standarddeviations indicated by the bars.

positive if fluid accumulation was ->0.5 ml/cm of (0.20 ml) prevented lethality of 0.90 gig of puri-the loop. By this procedure, purified toxin had a fled toxin injected intraperitoneaily in five of five50% enterotoxic dose value of 0.02 gig. For cell animals. For cell lysates, 0.15 ml of antiserumlysates, the 50% enterotoxic dose value was 2 neutralized lethality of 10 gLig of crude lysategig. In addition to its enterotoxicity, purified (five of five mice). In the absence of antiserumtoxin was lethal to mice when injected intraperi- or in the presence of normal rabbit serum, thetoneally. The 50% lethal dose was 0.2 gig, amounts of either toxin or cell lysate used werewhereas heat-inactivated toxin (100°C, 30 min) strongly toxic in both the ileal loop and thewas not lethal as high as 0.9 gig. In contrast, mouse lethality assay.crude cell lysates had a 50% lethal dose of 2 gig,whereas heat-inactivated lysate (0.6 mg) was not DISCUSSIONlethal. DSUSO

Neutralization of biological activity. Monospe- Through the use of gentle biochemical tech-cific antibody for purified Shiga toxin was used niques, we have developed a procedure whichto determine whether the toxic activities ob- allows recovery of milligram amounts of purifiedserved in cell lysates of S. dysenteriae were due Shiga toxin from bacterial lysates. As shown insolely to the presence of the protein purified Fig. 5, purified toxin behaved as a near-homoge-here. This antiserum (13 gil) neutralized the neous preparation, containing 1% contaminant.cytotoxic activity of cell lysates (36 gil) as shown Although our procedure should minimize chemi-in Fig. 6. Furthermore, the antiserum neutral- cal or biological insults to the purified protein.ized enterotoxic activity and mouse lethality of the toxin occasionally displayed multiple bandsboth purified toxin and cell lysates. Antiserum after PAGE. Limited proteolysis during purifica-(0.75 ml) prevented fluid accumulation in rabbit tion is the most likely explanation of this behav-ileal loop segments treated with 5.0 gig of purl- ior since the treatment of purified toxin withfled toxin; for cell lysates, 0.20 ml of antiserum proteases also yielded multiple species. Addi-prevented fluid accumulation in loops injected tional support for the purity of the toxin waswith 60 gig of lysate protein. In mice, antiserum obtained by the observation that monospecific 7-

V.4 +s,,I - ° ...- >vt

-

VOL. 36, 1982 SHIGA TOXIN PURIFICATION AND CHARACTERIZATION 1003

A Bl

N

1 2 3 4 5 6 7 8 1234FIG. 7. Immunoblot analysis of rabbit antiserum produced against purified toxin. Discontinuous electropho-

resis was performed in 8% polyacrylamide gel slabs until the dye front was within 1 cm of the bottom of the gel.After PAGE, electrophoretic transfer onto nitrocellulose was carried out for 11.5 h. (A) Autoradiographscomparing immune and preimmune serum. Lanes I through 4 show a 16-h autoradiograph. and lanes 5 through 8show a 72-h autoradiograph. The nitrocellulose was divided into strips and incubated with preimmune serum(lanes 1. 2, 5, and 6) or 6-week antiserum (lanes 3.4, 7. and 8). The dye front is shown by the arrow. Lanes 1, 3. 5.and 7 contained 15 ýLg each of crude lysate. Lanes 2, 4. 6, and 8 contaii'ed 0.25 ý.g each of purified toxin. I B)Immunological detection of toxin after treatment with S. griseus protease. Toxin was incubated with protease (1ý±g/ml) for 1, 5, and 30 min at 37°C. and phenylmethylsulfonyl fluoride was added (60 ng/ml) to inactivate theprotease. One microgram of toxin was applied to each lane. The dye front is shown by the arrow. Lanes: I.untreated; 2, 1 min; 3, 5 min; 4, 30 min.

antiserum was produced when toxin was used to fore that such cell lysates contain other toxinsimmunize rabbits. possessing the classic toxicities observed in

Previous reports have suggested that the toxic Shiga toxin extracts. Our conclusion is support-activities of Shiga toxin extracts can be resolved ed by the finding that the protein purified byinto different proteins. On the basis of our O'Brien et al. also contained all of the toxicfindings, this conclusion is unlikely. As shown in activities of Shiga toxin (18).Table 2, the purified protein possessed all of the Our observations indicate that the Shiga toxinbiological activities attributed to Shiga toxin, purified here differs in some respects from toxinThe protein displayed cytotoxicity at picogram isolated by others. When subjected to prepara-levels, enterotoxicity at nanogram levels, and tive isoelectric focusing during purification, thewas lethal to mice in submicrogram quantities toxin focused as a sharp peak at pH 7.2. Inwhen injected intraperitoneally. Fui .hermore, contrast, the material obtained from culture fil-all three activities were completely neutralized trates by Mclver et al. contained two toxicby monospecific antiserum. Cell-free protein moieties that were resolved by preparative iso-synthesis inhibition was also neutralized by electric focusing (17). Both of these were cyto-immunoglobulin isolated from the antiserum by toxic, but onc had no enterotoxicity or mouseprotein A-Sepharose chromatography (unpub- lethality. The purity of their preparation, howev-lished observation). In addition, monospecific er, was not established. The cytotoxin purifiedantibody was able to neutralize all toxic activi- by Olsnes and Eiklid appeared to be homoge-ties of crude cell lysates, including inhibition of neous, but showed a rather broad isoelectriccell-free protein synthesis. It is unlikely there- distribution, focusing between pH 6 and 7 when • .- ,

7W •iI I a

1004 BROWN ET AL. INFECT. IMMUN.

TABLE 2. Biological properties of Shiga toxin ed. Furthermore, in light of our findings, previ-

"Cyto- Entero- Mouse ous reports of electrolyte secretion (8). nonelec-Prepn toxicity" toxicity' lethality' trolyte secretion (1), and intestinal adenylcyclase

(CDýVmg) (ED,o/mg) ILD ,/mg) activation (5) require reevaluation with purifiedCell lysate 4 × 104 1 × l0- 5x10- toxin. With the ready availability of purified

Purified toxin 2 x 101 5 x 104 6 x 10- Shiga toxin, it will be possible to define carefullyits biochemical properties and relate these to itsFivefold serial dilutions of toxin were assayed for function in pathogenesis.

cytotoxicity in triplicate HeLa cell monolayers. Afterthe cells were fixed and stained the dilution causing ACKNOWLEDGMENTS50% detachment (CD,() was determined from a loga- We thank P. Gemski for critical advice, continued interest.nithmic plot of the mean of absorbances of retained and extensive discussions during the course of this work. Wedye versus dilution of toxin, thank H. Collins for skilled animal surgery. Excellent techni-

b Fluid accumulation of at least 0.5 ml per cm of cal assistance was provided by M. K. Gcntrv. M. Barreiro. A.loop was considered a positive enterotoxic response. Stone. M. Alvarado. and W. Cahillane.Each amount of toxin was assayed in at least three LITERATURE CITEDloops. The dose causing a positive response in 507 ofloops tested (ED,,o) was calculated by the method of I1. Binder, H. J., and D. S. Whiting. 1977. Inhibition of small

intestinal sugar and amino acid transport by the enterotox-Reed and Muench (22). in of Shigella dysenteriae 1. Infect. Immun. 16:510-512.

Mouse lethality was determined by intraperitoneal 2. Brown. J. E.. S. W. Rothman, and B. P. Doctor. 1980.injection of each dose into five or more mice. The 50% Inhibition of protein synthesis in intact HeLa cells bylethal dose (LD.,) was calculated by the method of Shigella dysenteriae I toxin. Infect. Immun. 29:98-107.Reed and Muench (22). 3. Brown. J. E., M. A. Ussery, S. H. Lqppla. and S. W.

Rothnman. 1981. Inhibition of protein synthesis by Shigatoxin. Activation of the toxin and inhibition of peptideelongation. FEBS Lett. 117:84-88.

analyzed with sucrose gradient isoelectric focus- 4. Burnette, W. N. 1981. "'Western blotting." Electrophoret-ing (21). Enterotoxicity and mouse lethality ic transfer of proteins from sodium dodecyl sulfate-poly-were not tested. The tox--, !'nlated by O'Brien e acrylamide gels to unmodified nitrocellulose and radio-

grapnic dc;c.tion with antibody and radioiodinated .al. contained all three classical toxicities of protein A. Anal. Biochem. 112:195-203.Shiga toxin, but molecular weight estimates by 5. Charney, A. N., R. E. Gots, S. H. Formal, and R, A.Sephacryl S-200 gel filtration gave an Mr of Glannella. 1976. Activation of intestinal mucosal adenyl-33,000, suggesting to the authors that dissocia- ate cyclase by Shigella dssenteriae I enterotoxin. Gastro-

enterology 70:1085-1090.tion may have occurred during elution at the 6. Conradl, H. 1903. Ueber I1sliche durch aseptische auto-

affinity chromatography step (18). The toxin lyse Erhaltene Giftstoffe von Ruhr-und Typhusbazillen.described here gave an estimated Afr of 70,000 in Dtsch. Med. Wochenschr. 29:26-28.the process of purification on Sephacryl S-200. 7. Davts, B. J. 1964 Disc elec'rophoresis. II. Method andin prclse ag pureementwith tn ehat f lsn and00. application to human serum proteins. Ann. N.Y. Acad.in closer agreement with that of Olsnies and Sci. 121:404-427.Eiklid (21). In all cases, previous attempts have 8. Donowltz, M., G. T. Keuseh, and H.J. Binder. 1975.produced only microgram quantities of purified Effect of Shigetta enterotoxin on electrolyte transport intoxin, rabbit ileum. Gastroenterology 69:1230-1237.

9. Elkild, K., and S. Olmes. 1980. Interaction of ShigellaToxin purified as described here has been shigae cytotoxin with receptors on sensitive and insensi-

used to establish the biochemical basis of Shiga tive cells. J. Receptor Res. 1:199-214.toxin action. Our studies of toxin action in cell 10. Formal, S. B., D. Kundel, H. Schneider. N. Kuner. and H.culture have shown that inhibition of protein S'Inz. 1961. Studies with Vibrio cholerae in the ligatedion loop of the rabbit intestine. Br. J. Exp. Pathol. 421:504-synthesis is the primary event (2). Additional 510.studies with cell-free protein synthesis indicate it. Gemakl, P., A. Takeochil, o. Washington. and S. B. For-that Shiga toxin may function enzymatically nial. 1972. Shigellosis due to Shigella daseateriae 1:through an inhibition of peptidyl elongation, relative importance of mucosal invasion versus toxin

production in pathogenesis. J. Infect. Dis. 126:523-530.whereas no evidence of ADP-ribosyl transferase 12. Gentry, M. K., and J. M. Dalrymple. 1980. Quantitativeactivity was found (3). Preliminary data suggest microtiter cytotoxicity assay for Shigella toxin. J. Clin.that the toxin inhibits the amino-acyl tRNA Microbiol. 12:361-366.binding stage of peptidyl elongation (. E. 13. Kegach. G. T., G. F. Grady. L. J. Mata. and J. McIver.

1972. The pathogenesis of Shigella diarrhea. I. Entero-Brown and M. Ussery, Fed. Proc. 40:1580, toxin production by Shigella advsenteriae 1. J. Clin. In-1981). This apparently occurs by inactivation of vest. 51:1212-1218.the 60S ribosomal subunit (23). 14. Ketch, G. T., M. Jacewlcz. and S. Z. Hirschman. 1972.

The precise role of Shiga toxin in pathogenesis Quantitative microassay in cell culture for enterotoxin ofShigella dysenteriae 1. J. Infect. Dis. 125:539-541.

remains unclear. Inhibition of protein synthesis is. L , L. H., H. Schneider, T. J. Mngnati, and S. u.appears to be the basis of all recognized toxic Formal. 1964. Epithelial cell penetration as an essentialproperties of Shiga toxin. The sequence of step in the pathogenesis of bacillary dysentery. J. Bacteri-events leading to pathological sequelae of toxin 16o. ,8:1503-1518.

1.Lowry, 0. 11. N. J. Rosebroagh, A. L. Farr. and R. J.activity such as fluid secretion in rabbit ileal Randall. 1951. Protein measurement with the Folit phenolloops or lethality to mice remains to be elucidat- reagent. J. Biol. Chem. 193:265-275.

*$a

.4. . I

VOL.. 36. 1982 SHIGA TOXIN PURIFICATION AND CHARACTERIZATION 1005

17. Mclver. Jr., G. F. Grady. and G. T. Keuscb. 1975. Pro- estimating fifty per cent endpoints. An. J. Hyg. 27:493-

duction and characterization of exotoxin of Siuikella ci s- 497.

enteriae type 1. 1. Infect. Dis. 131:.559-566. 23. Rebig,~ R.. S. Oboes. and K. Eldild. 1981. The cytotoxic

18. O'Brien. A. D., G. D. LaVeck, D. E. Griffin, and M. R. activity of She gella toxin. Evidence for catalytic inactiva-

Thosipson. 1980. Characterization of ShigelIa dvisenteriae tion of the 60GS ribosornal subunit. 1. Biol. Chem.

I (Shigp) toxin purified by anti-Shigp toxin affinity chro- 256:8739-8744.mnatography. Infect. Immun. 30.170-179. 24. Thompson. M. IR.. M. S. Stekinberg., P. Gersnak. S. B.

19. Olenkk. J. G., and A. D. Wolfe. 1980. Shijge/Iu toxin Formal, and B. P. Doctor. 1976. Inhibition of in vimr

inhibition of the binding and translation of polyuridylic protein synthesis by Shigella dysenteriae I toxin- Bio-

acid by ribosomes of Escherichia to/i. J. Bacteriol. chem. Siophys. Res. Commun. 71:783-7,88.141:1246-1250. 25. van Iieyningen. W. E., and G. P. Gladstonae. 1953. The

20. Olitsky, P. K., and 1. J. Ktlgir. 1920. Toxins and antitox- neurotoxin ot Shigella shigae. 1. Production, purification

ins of Bacillus dvsenteriae .uhiga. J. Exp. Mled. 31:19-33. and properties of the toxin. Br. J. Exp. Pathol. 34:102-

21. Ohoes. S.. aod Ilk. Elildli. 1980. Isolation and characteriza- 216.tion of Shigella shigae cytotoxin. J. Biol. Chem. 255:284- 26. Vicail. G.. A. L. Olltzki. and Z. Ofitzid. 1960. The action

289. of therniolabile toxin of Shigella dysenferiae on cells

22. Reed. L. J., and Hi. Mueneb. 1938. A simple method of cultured in vgrro. Br. J. Exp. Pathol. 41:179-189.

Accl35iof For

NTIS GRA&I

DTIC TAB E)

Unannounced C3iY 1justfictioELEF T

DiStributiofl ___ U 2 9 2

Availability CodeS

Avail and/or

DiSt Special

DTI

REPORT DOCUMENTATION PAGE READ INSTRUCTIONSBEFORE COMPLETING FORMI. REPORT w MOER 12, GOVT ACCESSION NO. 3. RECiPIENTS CATALOG NUMBER

4. TITLE (and Subtitle) S. TYPE OF REPORT & PERIOD COVEREO

Purification and characterization of Shiga toxinfrom Shigella dysenterial 1

S. PERFORMING ORO. REPORT NUMBER

7. AUTHOR(a) S. CONTRACT OR GRANT NUMBER(@)

J. Edward Brown, D. E. Griffin, S. W. Rothman,and B. P. Doctor

__3/A 14/i .)/O4 I-3S. PERFORMING ORGANIZATION NAME AND AOORESS 10. PROGRAM ELEMENT. PROJECT. TASK

AREA & WORK UNIT NUMIMERSDept of Biological Chemistry, Div. Biochemistry Program element: 6111ilO2,WRAIR, Washington D. C. 20012 Project: 6,I11O2S1OAH

Task Area 00 Work Unit 124

11 CONTRO66IND OPFICE NAVE AND ADDRWS IS. REPORY DAYSU.S. Army Medical Rsh & Developement CommandFt. Detrick, Frederick, MD. 21701 13. NUMBEROF'PACES

1214. MONITORING AGENCY NAME A ADDRESS(i dlloerat from Controlling Oflfie) 15. SECURITY CLASS. (of this ,eport)

Walter Reed Army Institute of ResearchWashington, D. C. 20012 Unclassified

IS*. DECLASSIFICATION/DWNHGRADINGSCHEDULE

16. DISTRIBUTION STATEMENT (of &tie Report)

Approved for public release

17. OISTRIBUTION STATEMENT (of the ctlroat aengorredI Blo k Wea, 1idE ftrent fem Reprtf)

"1S. SUPPLEMENTARY NOTES

15. KEY WORDS (Continue on reverse side Ii nocoeary mad Idonftly by block number)

Toxin, Shigella, dysentery, purification, protein synthesis

20, ADmYRACT r~whu *a .. Veres eb ff ,nesoay seat Ide~ifr bp block nulmb.,)

Shiga toxin has been purified in milligram quantities to near homogeneityfrom cell lysates of Shigella dysenteriae I strain 3818-0. Purificationinvolved an initial ultracentrifugation, ammonium sulfate fractionation,chromatography on DEAE-cellulose and carboxymethyl cellulose, gel filtration,and preparative isoelectric focusing in sucrose gradients. The purifiedtoxin was resolved by discontinuous polyacrylamide gel electrophoresis

(over)

DO 143 OPTION OF IVESISO5SO awETR UNCLASSIFIED

* . UOIUTY C6ASSIPICATION OP TNII PA4l (She. Do. tnee0

...... .... ..............

SIECURITY CL I- .•l-A lON OF THIS PAGE(IWnhn Dala Entered)

into a major cytotoxic protein band and a closely migrating, proteolyticallynicked minor component. Antiserum generated by immunization with glutar-aldehvde-inactivated toxin was shown to be monospecific against S. dysenteriaecell lysates. This highly purified toxin wascytotoxic to HeLa cells,enterotoxic in rabbit ileal loops, and lethal to mice. Monospecific ahtiserumto the toxin neutralized completely these toxin activitiesin both purifiedtoxin preparations and crude shigella cell lysates.

CT

I-

j * mcumrrY CLASSIFgCATION OF THIS PAOtI(Wh. lD,, IRate.)