Adenylate Cyclase Toxin from Bordetella pertussis - The Journal of ...

THEJOURNAL OF BIOLOGICAL

Vol. 266, No . 26, Issue of September 15,PP 17503-17508,1991 Printed in U.S.A.

CHEMISTRY

0 1991by The American Society for Biochemistry and Molecular Biology, Inc.

Adenylate CyclaseToxin from Bordetella pertussis CONFORMATIONALCHANGEASSOCIATEDWITHTOXINACTIVITY* (Received for publication, February 15, 1991)

Erik L. HewlettSQIl, Lloyd Gray 11, Margaretta Alliettall, Ingrid Ehrmann$§,Valery M. Gordon$§**, and Mary C. Gray$ From the Departments of $Medicine, §Pharmacology, and IIPathology, University of Virginia School of Medicine, Charlottesuille, Virginia 22908

Adenylate cyclase (AC) toxin from Bordetella pertussis interacts with and enters eukaryotic cells to

catalyze the production of supraphysiologic levels of cyclic AMP, Although the calmodulin-activated enzymatic activity (ability to convert ATP to cyclic AMP in a cell-free assay) of this molecule is calcium independent, its toxin activity (ability to increase cyclic AMP levels in intact target cells) requires extracellular calcium. Toxin activity as a function of calcium concentration is biphasic, with no intoxication occurring in the absence of calcium, low level intoxication (200300 pmolof cyclic AMP/mg of Jurkat cell protein) occurring with free calcium concentrations between 100 nM and 100 p~ and a 10-fold increase in AC toxin activity at free calcium concentrations above 300 pM. The molecule exhibits a conformational change when free calcium concentrations exceed 100 p~ as demonstrated by shift in intrinsic tryptophan fluorescence, an alteration in binding of one anti-AC monoclonal antibody, protection of a fragment from trypsin-mediated proteolysis, and a structural modification as illustrated by electron microscopy. Thus, it appears that an increase in the ambient calcium concentration to a critical point and the ensuing interaction of the toxin with calcium induces a conformational change which is necessary for its insertion into the target cell and for delivery of its catalytic domain to the cell interior.

target cells to catalyze the production of supraphysiologic levels of intracellular cyclic AMP is calcium dependent, as noted by a number of investigators (6-8). Little is known about therole of calcium inAC toxin action, except by way of analogy. Glaser et al. (9) have cloned and sequenced the genefor AC toxin, designatedcyaA. The protein has a molecular mass of 177 kDa as calculatedfrom the DNA sequence (9) and determined by equilibrium centrifugation (8), but an apparentsize of 200-216 kDa by SDS-PAGE (1014). The amino-terminal 400 amino acids comprise the catalytic domain, which includes a calmodulin-binding site and a n enzyme active site (9, 15). The carboxyl-terminal region (1300 amino acids) contains numerous copies of a glycinerich repeatconsisting of Leu-X-Gly-Gly-X-Gly-Asn-Asp, which is also represented by 13 copies inEscherichia coli hemolysin, HlyA (9, 16). This sequencewas recognized by Welch et al. (16) as potentiallyinvolved in the interaction of HlyA with calcium. Elimination of 11 of these 13 repeats in shown to abolish E. coli by site-directed mutagenesis has been calcium binding and hemolytic activity of the molecule (17). We have recently reported that the AC toxin is a calciumbinding protein(18). In the present study, the role of calcium in AC toxin structure and function has been addressed. The AC toxin interaction with calcium elicits a conformational change which is associated with expressionof maximal toxin activity. EXPERIMENTALPROCEDURES

Materials-Hanks’ balancedsaltsolution,fetal bovine serum, RPMI 1640, penicillin-streptomycin,Bordet-Genou,andStainerAdenylate cyclase toxin is a novel virulence factor for the Scholte medium were fromGIBCO. Phenyl-Sepharose CL-4B and causative agent of whooping cough, Bordetellu pertussis (1). calmodulin-Sepharose 4B were obtained from Pharmacia LKB BioIts enzymatic activity, as measured by the conversion of ATP technology Inc. Polyvinylidene difluoride (PVDF) membranes were into cyclic AMP, has been shown to be increased 1000-fold obtainedfrom Millipore Corp. (Bedford, MA). 45Ca was from Du by the eukaryotic regulatory protein, calmodulin (2,3). UnlikePont-New England Nuclear. Peroxidase-conjugated goat anti-mouse IgG heavy and light chains were obtained from Jackson Immunoother activitiesof calmodulin, however, this activationof AC’ Research (West Grove, PA). Monoclonal antibodies (9D4, 1H6, and toxin can occur in the absenceof calcium (4, 5). In contrast, 2F5) were prepared as described previously (10). An irrelevant monothe ability of AC toxin to interact with and enter eukaryotic clonal (MHS-5) was used as a control as previously described (10). All other reagents were obtained from Sigma, unless otherwise indi* This work was supported National Institutes of Health Grants cated. R 0 1 A118000 (to E. L. H.), R29 CA47401 (to L. S. G . ) ,and DK38942 Culture of Organisms and PurificationofAdenylate Cyclase Toxcn(University of Virginia Diabetes Center). The costs of publication of B. pertussis (Strain BP338) was grown as described previously (10). this article were defrayed in part by the payment of page charges. AC toxin was purified by urea extraction, phenyl-Sepharose chroThis articlemustthereforebe hereby marked“advertisement”in matography, sucrose density gradient centrifugation, and calmodulinaccordance with 18 U.S.C. Section 1734 solely to indicate this fact. Sepharose affinity chromatographyas described (10) with thefollow7 T o whom correspondence should be addressed Box 419, Health ing modifications. Pooled fractions from the sucrose gradient with Sciences Center, University of Virginia School of Medicine, Char- added CaCI2 (2 mM) was passed overa calmodulin-Sepharose4B lottesville, VA 22908. column, washed with 10 column volumes of 10 mM Tricine, pH 8, 1 ** Presentaddress:Dept. of Microbiology, F. EdwardHerbert mM CaCl,, 0.5 M NaCl; then 3 column volumes of 10 mM Tricine, pH School of Medicine, 4301 Jones Bridge Rd., Bethesda, MD 20814. 8,0.5 mM EGTA, 0.5 mM EDTA (Buffer A), and eluted with 8 M urea ’ The abbreviations used are: AC, adenylate cyclase; SDS-PAGE, in Buffer A. sodium dodecyl sulfate-polyacrylamide gel electrophoresis;PVDF, Sample Preparation and Determinationof Free Calcium-Purified polyvinylidenedifluoride; EGTA, [ethylenebis(oxyethylenenitrilo)] AC toxin was dialyzed against 10 mM Tricine, pH 8, 2 mM EDTA, 3 tetraacetic acid. mMMgC1, (Buffer B) prior to addition into assays. Free calcium

17503

Adenylate Cyclase Toxin from B. pertussis

17504

concentrations were calculated using the EGTA computer program (19) with substitution of equilibrium constants for EDTA found in Ref. 20. Adenylate Cyclase Enzymatic Activity-Enzymatic activity was measured by conversion of [cI-~'P]ATP to [32P]cAMPin a cell-free assay as described previously (10,21). Briefly, the reaction was carried out at30 "C and continued for 10 min in a final volume of 60 pl.Each sample contained 60 mM Tricine, pH 8.0, 10 mM MgCl', 1 mM ATP (with 2 X lo5 to 5 X lo6 cpm of [w3'P]ATP), and 1 p M calmodulin. The reaction was terminated by the addition of 100 p1 of a solution mM ATP, and 6.25 mM cAMP (including containing 1%SDS, 20 15,000-20,000cpmof [3H]cAMP/tube for calculation of recovery). The cAMP formed was isolated by the double column method of Salomon et al. (22). Adenylate Cyclase Toxin Activity-Adenylate cyclase toxin activity was determined by quantitation of intracellular cAMP accumulation in Jurkat cells, a human T helper cell line. Jurkat cells were maintained in RPMI 1640 medium supplemented with 15% fetal bovine serum, 2 mM glutamine, 50 units/ml penicillin and 50 pg/ml streptomycin in 5% CO,. AC toxin was added to 1 X lo6 cells/ml in Hepes buffer, pH 7.4 (140 mM NaCl, 5 mM KCl, 1%glucose, 3 mM MgC12, 2 mM EDTA) and incubated for 30 min a t 37 "C. Intracellular cAMP was extracted with 0.1 N HC1 at 25 "C for 30 min and measured by radioimmunoassay (23). Proteinwas extracted with 0.2 N NaOH and quantitated using the method of Lowry et al. (24). Intrinsic Tryptophan Fluorescence-The fluorescence emission spectrum ofAC toxin (8 pg/ml) was measured in an SLM 8000 spectrofluorimeter (SLM/Aminco, Urbana,IL). The sample was placed in a quartz cuvette with the sample chamber maintained at 37 "C.Excitation wavelength was 280 nm and theemission spectrum over the range of310-380 nm in 1-nm increments was recorded. Bandpass was set a t 4 nm for excitation and emission monochrometers. From the emission spectrum, the maximum emission wavelength and the ratio of emission intensities (330350 nm) was determined. Monoclonal Antibody Reactivity-Purified AC toxin in Buffer B with or without calcium (1 mM)was dottedontoa nitrocellulose membrane using a minifold apparatus (Schleicher & Schuell). The membranes were allowed to airdry, then incubated overnight at 4 "C with ascites of monoclonal antibodies at a 1:500 dilution in Buffer B containing 1%bovine serum albumin, 140 mM NaCl, with or without 1mM calcium. The nitrocellulose membranes were then washed three times in Buffer B containing 140 mM NaCl with or without 1 mM calcium and incubated with peroxidase conjugated IgG for 1 h at room temperature. The membranes were then washed three times as above, and incubated with 0.5 mg/ml chloronapthol in 50 mM Tris, pH 7.5, 200 mM NaCl, 0.01% hydrogen peroxide for 5 min at room temperature (25). For quantitation ofAC toxin interaction, 96-well flatbottom microtiter plates were coated with monoclonal antibodies at 10 pg/ well as previously described (26). AC toxin was then added in the absence or presence of calcium at the indicated concentrations and incubated overnight at 4 "C. Microtiter plates were then washed three times with phosphate-buffered saline, and theadenylate cyclase assay (as described above) was performed in the microtiter plate well. Trypsin Treatment-Purified AC toxin was treated with 400 pg/ ml acetylated trypsin for 10 min at 10 "C, stopped with 800 pg/ml lima bean trypsin inhibitor, and separated by SDS-PAGE on a 520% gradient gel (27). The separated proteins were then transferred to PVDF membrane for 1.5 h a t 0.6 A and incubated with a anti-AC monoclonal antibody (9D4) as described above with the following modifications: 1st and2nd antibodies were added a t a 1:lOOO dilution in 50 mM Tris, pH 7.5, 200 mM NaCl containing 1%bovine serum albumin, washes were done with 50 mM Tris, pH 7.5, 200 mM NaCl. Electron Microscopy-Affinity purified AC toxin was negatively stained by floating a thin carbon film onto a droplet of AC toxin for 1 min in the presence of indicated calcium concentrations, then transferred to a droplet of 0.5% aqueous uranyl formate for 30 s. The carbon film was picked up on a 300-mesh nickel grid and allowed to air dry for viewing with a Zeiss 902 electron microscope.

results in calcium binding to theholotoxin molecule, defining it asa calcium-binding protein (18).When equimolar concentrations ofAC toxin and calmodulin were compared using that system, the AC toxin appeared to bind slightly less radioactivity, suggesting that itmay have fewer than thefour calcium-binding sites in calmodulin. The calcium concentration dependence of AC enzyme and toxin activities is shown in Fig. 1. Consistent with previous studies showing inhibition by calcium concentrations 2 1 mM (4, 5, 30), the enzyme activity of AC toxin in the presence of a maximal concentration (1p ~ of) calmodulin is moderately affected by calcium. From its maximum at 1p M free calcium, the enzyme activity is reduced by 33% at 1.0 mM calcium (Fig. 1) and by 69% at 10 mM calcium (data not shown). In contrast, the toxin activity is profoundly dependent on the presence of calcium and exhibits a biphasic calcium concentration curve (Fig. 1).Intracellular cAMP accumulation elicited by AC toxin in Jurkat cells is above basal by 125 nM free calcium and exhibits a plateau at 10% of maximal activity between 7 and 300 p~ free calcium. A further increase in free calcium from 300 p M to 1.0 mM results in a 10-fold increase in toxin activity (Fig. l ) , which remains unchanged at calcium concentrations up to 10 mM (data notshown). Since the toxin activity is increased in conjunction with the previously observed inhibition of enzyme activity at higher calcium concentrations, it appears that several calcium-dependent processes are occurring simultaneously (4, 30). The remainder of this study addresses the physical changes which are associated with thesharp increase in toxin activity at free calcium concentrations above 100 pM. The derived amino acid sequence for AC toxin indicates that the molecule contains sixteen tryptophans (9). Tryptophan residues in an aqueous environment fluoresce with an emission maximum of -350 nm, while those exposed to hydrophobic conditions exhibit maxima at wavelengths of 320335 nm. Thus, theratio of fluorescence intensities at 330 and 350 nm may be used to monitor changes in the conformation of a protein (31).At free calcium concentrations of 0-11.4 pM, the peak fluorescence intensity of purified AC toxin occurred at 348 (+ 1) nm. At 100 p~ free calcium, the peak exhibited a blue shift to 339 (+1)nm. When expressed as theflorescence ratios (Fig. 2) these measurements reveal an alteration in the environment of one or more tryptophan residues in the AC toxin, suggesting a conformational change induced by increasing the free calcium concentration. Other studies using intrinsic fluorescence spectroscopy have demonstrated that tryptophan at residue 242 of a fragment of AC toxin expressed in E. coli is affected by the interaction of calmodulin with the molecule (32).

RESULTS

FIG. 1. Effect of calcium concentration on AC enzyme and toxinactivities. AC toxin was dialyzed against Buffer B with calcium added to achieve indicated free calcium concentrations. AC enzyme activity (A)and AC toxin activity ( 0 )were measured as described under "Experimental Procedures." Lettered arrows (A-D) designate calcium concentrations at which samples were prepared for electron microscopy (see Fig. 6).

AC toxin requires calcium for disruption of multilamellar liposomes (28), lysis of sheep erythrocytes (29), and intoxication of eukaryotic target cells (6-8). Recently, we have shown that separation ofAC toxin by SDS-PAGE, followed by transfer to PVDF membrane, and incubation with 45Ca

7t


17505 TABLEI

0.90,

Effect of calcium concentrations on the capture of AC toxin by monoclonal antibodies in microtiter wells Monoclonal antibody

AC enzyme activity OcaZ+

100 p~ Ca2+

1.0 mM Caz+

pmolll0 minlwell

25,200 27,100 22,400 2F5 (21)"

".." .

,,

0

. 10.7

106

10-5

10.4

10.3

[caa+l (M)

FIG. 2. Effect of calcium on the intrinsic fluorescence of AC toxin. Samples containing 87 pg/ml of AC toxin in Buffer B with added calcium a t indicated concentration were prepared at room temperature and scanned a t 37 "C. Excitation was a t 280 nm and the ratio of emission intensity at 330 nm to that at350 nm plotted.

21,200 9D4 23,200 18,700 (13) 7,480 1H64,410 2,670 (180) 1,300 MHS-51,390 1,310

(15)

(24) (65)

(6) Percent increase relative to 0 calcium.

(0)

A 200-

~

~

Monoclonal Antibody 2F5

904

1H6

MHS-5

0

95

-

55

-

Ica9 (mM) 1.0

43-

FIG. 3. Monoclonal antibody reactivity of AC toxin in the presence or absence of calcium. Samples containing0.6 pg of AC toxin in Buffer B with and without 1 mM calcium were dotted onto a nitrocellulose membrane using a minifold apparatus. The membrane was then incubated with monoclonal antibodies to AC toxin (904, 2F5, IH6) or an irrelevant antibody ( M H S - 5 ) as described under "Experimental Procedures." Densitometry scanning values for the data shown are: 2F5 at 0 calcium, 142; 2F5 at 1 mM calcium, 124; 904 at 0 calcium, 57.9; 904 at 1 mM calcium, 53.6;IH6 at 0 calcium, 5.3;IH6 at 1 mM calcium 41.8; MHS-5 at 0 calcium, 1.5; MHS-5 at 1 mM calcium, 4.5.

Further evidence for a calcium-dependent conformational change is provided by monoclonal antibody reactivity of AC toxin in the absence and presence of calcium. Four monoclonal antibodies were used, three (9D4, 2F5, and 1H6) selected for their reactivity with AC toxin (10) and the fourth (MHS-5) an irrelevant monoclonal used here and previously (10)as a negative control. As shown in Fig. 3, the binding of 9D4 and 2F5 to affinity purified AC toxin on dot-blot under non-denaturing conditions was equivalent at 0 and 1mM free calcium. In contrast, recognition of AC toxin by 1H6 was dependent upon the presence of calcium. MHS-5 did not react with AC toxin under either condition. In order to provide quantitative data on the antibody reactivity demonstrated in Fig. 3,an assay was developed which takes advantage of the intrinsic enzymatic activity of the toxin. Microtiter wells werecoated with the individual monoclonal antibodies (none of which effect AC enzyme activity) and the ability of each to bind AC toxin at 0,100p ~ and , 1 mM free calcium wasmeasured by AC enzyme activity in the well (Table I). The enzyme activity bound by 9D4 and 2F5 varied less than 25% overthe range of calcium concentrations tested. Antibody 1H6, on the other hand, was associated with only low level activity in theabsence of calcium, but showed increases of 65% at 100 p M calcium and 180% at 1 mM calcium. Together, these data suggest that 1H6reacts with an epitope on AC toxin which develops or becomes exposed in the presence of calcium, with the majority of the change occurring between 100 p~ and 1mM. The addition of calcium to AC toxin also alters theprofile of proteolytic fragments in response to trypsin. The AC enzymatic activity has been shown previously to bevery sensitive to trypsin (21) and analysis of the protein sequence

A

C

B

D

B 200-

e5

-

5s43

-

A

B

C

D

E

F

FIG.4. Differential sensitivityof AC toxin to trypsin treatment. Purified AC toxin in Buffer B with the indicated free calcium concentration was treated with 400 pg/ml trypsin for 10 min at 10 "C. The reaction was stopped by the addition of 800 pg/ml trypsin inhibitor and proteins separated on a5-20% gradient gel. The separated proteinswere then transferred to PVDF membrane and probed with an anti-AC monoclonal antibody, 9D4.A, purified AC toxin (2 pg) a t 0 calcium, lane A; AC toxin at 0 calcium with trypsin treatment, lane B; AC toxin at 1 mM calcium, lane C; AC toxin at 1 mM calcium with trypsin treatment, lane 0. B, purified AC toxin (6 pg) was treated with trypsin as described above. Lune A, AC toxin with 0 calcium; lane B, AC toxin with 100 nM calcium; lane C,AC toxin with 1 p M calcium; lane D,AC toxin with 10 pM calcium; lane E , AC toxin with 100 p M calcium; lane F, AC toxin with 1 mM calcium.

using the GCG programs (34) reveals 139 potential trypsin cleavage sites. Fig. 4A demonstrates that incubation of purified AC toxin with trypsin (400 pg/mlfor 10 min at 10 "C) in the absence of calcium results in loss of virtually all the AC toxin present. When this procedure is carried out with 1mM calcium several major peptide fragments are protected. To determine the critical calcium concentration for this effect, a higher concentration of toxin was exposedto trypsin at several levels of calcium (Fig. 4B). Under these conditions, not all the toxin is degraded but the increased protection of specific fragments is firstdemonstrable at 100 p~ calcium, consistent with the other indicators of a conformational change. The proteolytic activity of trypsin using the artificial substrate NO-benzyloxycarbonyl-L-lysinethiobenzyl ester hydrochloride was unaffected by calcium over the range of 0-1.0 mM (data not shown). The most prominent peptide (60 kDa) which is protected by the presence of calcium (Fig. 4-4)was isolated and anNHp-terminal sequence determined by Edman degradation. The 20 amino acid sequence corresponded exactly with the predicted sequence from amino acid 1368-1387

17506


FIG. 5. Calcium-induced conformational change of AC toxin and its reversibility as detected by electron microscopy. Purified AC toxin (90 pg/ml) in Buffer B without added calcium (panel A ) or with 1 mM freecalcium (panel B ) were negatively stained for electron microscopy as describedunder “Experimental Procedures.”A portion of the sample shown in panel B was then extensively dialyzed against Buffer B to remove calcium (panel C). Magnification was X592,000, bar = 17.0 nm.

(data not shown), indicating that, based on its apparent size, p~ calcium the concentration that elicits a pronounced the fragment probably includes the entire carboxyl terminus change in the ultrastructureof the molecule, AC toxin activity of the protein. Together these data illustrate further that theis half-maximal (arrow C, Fig. 1). Finally, complete opening AC toxin, perhapsin large part thecarboxyl-terminal portion, of the molecule at 1 mM calcium is associated with maximal toxin activity (arrow D,Fig. 1). The ultrastructural datawith is altered in structure by its interaction with calcium. In order to visualize the structuralchange in theAC toxin the intrinsic fluorescence, monoclonal antibody reactivity, molecule suggested by the biophysical and immunological and change in sensitivity to trypsin-mediated proteolysis indata, negative staining electronmicroscopy (EM) was used. dicate that the binding of calcium by AC toxin results in a The electron photomicrographs presented in Fig. 5 provide a major conformational change which may be essential for full field view of the dramatic conformationalmodification in the toxin activity. AC toxin molecule which occurs in the presence of calcium. DISCUSSION Panel A shows holotoxin in the absence of calcium as a tightly closed spherical structure possessing an off-centered “dimpleThe dependence of AC toxin activity upon calcium has been like” feature seen in most molecules. Panel B illustrates the noted by several investigators (6-8). The basis for the calcium open, linear form which is seen after the addition of 1 mM effect, however, has not been understood. Recognition of the free calcium. This change is reversible; dialysis against Buffer sequence homology between AC toxin and E. coli hemolysin B to remove the added calcium results in reduction of toxin by Glaser et al. (9) provides an important perspective. The activity to calcium-free level (data notshown) and closure of deduced amino acid sequences of both toxins contain nonathe toxin structure (panel C). Aggregated material is seen meric repeats which Welch et al. (16) recognized to contain occasionally in the electron micrographs, but it appears that features present in calcium-binding proteins. There are 13 of the forms demonstrated here represent AC toxin monomers. these glycine-rich repeats inE. coli hemolysin (HlyA) and 18To relate the conformationalchange with the different 43 in B. pertussis AC toxin depending upon the stringency of levels of AC toxin activity, samples of toxin were prepared at the criteria used to define the motif. Ludwig et al. (35) calcium concentrations of 0, 100 pM, 424 pM, and 1 mM and demonstrated that deletion of one or two of the repeats from examined by EM. As shown in Fig. 6, theotoxin is a closed, HlyA resulted in a molecule which retained hemolytic funcspherical structure with a diameter of 116 A in theabsence of tion only at higher calcium concentrations. Those authors a structure calcium (panel A ) . No ultrastructural change was observed a t proposed that therepeats participate in finger-like calcium concentrations below 100 pM, but at that level of with an octahedral calcium-binding site and possibly projeccalcium, a modest unfolding or uncoiling of the molecule has tions involved in binding to thetarget cell surface. Evaluation occurred (panel B ) . The opening is more pronounced at 424 by amino acid sequence analysis, however, reveals that neither E. coli hemolysin nor AC toxin possess EF hand-like strucpM free calcium (panel C) and is essentially complete by 1 tures (36).* More recently, Boehm et al. (17) showed that mM free calcium (panel D). When these electron micrographs are compared with the HlyA binds calcium and that calcium is necessary for intercalcium concentration curve for intoxication, it is apparent action of the hemolysin with the erythrocyte membrane. In that theminimal change a t 100 p~ calcium is associated with low level toxin activity (arrow B, Fig. 1). In contrast, a t 424 R. H. Kretsinger, personal communication.

Adenylate Cyclase from Toxin

A

B. pertussis

17507

B

VI(;. 6. Effect of calcium concentration on the conformation of AC toxin. l'uril'icd )I t o( x i' n ( 9 0 jrg:/mll without calcium (panel A ) or with 100 p M calcium (panel H ) , 421 pivl calcium (pclnd c),and 1 mM calcium (pat7c~l D )was negatively stained for electronmicroscopy. Magnification was ~1,200,000.Relow each elect,ronphotomicrograph is a proposed model of the toxin structure under the given condition to aid in interpretation. Values for corresponding AC enzyme and toxin activities are shown in Fig. 1, where lettered arrou)s indicate the calcium concentrations at which the samples were prepared for EM.

that study, a mutant HlyA from which 11 of 13 repeats were deleted did not bind calcium, interact with target cell membranes, or elicit hemolysis. Because of the sequence homology between AC toxin and HlyA, the recent data from this laboratory showing that AC toxin is also a calcium-binding protein (18) was not unexpected. However, the effects of calcium on intoxication, as demonstrated in this report, are striking. First, the calcium dependence ofAC toxin activity is biphasic, with low level intoxication occurring up to300 p M free calcium. Becauselow level intoxication is absolutely calcium-dependent but occurs without evidence of conformational change in the molecule, it may be that calcium at those concentrations is acting to bind the toxin to the target cell. The presence of toxin may then allow low level entry by an uptake mechanism unrelated to thatby which fully functional toxin acts. The increase in calcium above 300 PM results in a sharp, 10-foldrisein toxinactivity which is associatedwith an apparent conformational changein the AC toxin. The change in intrinsicfluorescence ratio is nearly maximal at thecalcium concentration which is half-maximal for intoxication resulting in a difference in half-maximal concentrations for these activities of slightly more than 10-fold (Figs. 1 and 2). The differential sensitivities for these two effects may be interpreted in several ways. First, they could reflect separate and unrelatedphenomena.Alternatively,theintrinsic fluorescence measurement may be a more sensitive indicator of a calcium-induced modification which must be essentially complete before there is an alteration in function. Finally, the shift in intrinsic fluorescence may represent a small, initial

change in the proteinin response to calcium, which is necessary, but not sufficient for the increased toxin activity observed at higher calcium concentrations. In any case, these data aswell as those concerning altered monoclonal antibody reactivity and proteolytic enzyme sensitivity support the hypothesis that one ormore calcium-dependent conformational changes occur in the AC toxin molecule in conjunction with the acquisition of full toxin activity. As noted above, the proteolytic sensitivity ofAC toxin is well recognized (21) and is supported by the identification of 139 potential trypsin sites in the aminoacid sequence. Modification of thetrypsinsensitivity of I?. pertussis AC by calmodulin has been described in several studies previously. Ladant (37) demonstrated thatproteolysis of a 43-kDa fragment corresponding to the catalytic domain is impaired in the presence of calmodulin and that two separate peptides of 18 and 25 kDa appear to interact simultaneouslywith calmodulin. More recently, Raptis et al. (38)showed that the toxin activity ofAC toxinis more sensitivetotreatment with trypsin or chymotrypsin than is the enzymatic activity. Furthermore, under conditionsof controlled proteolysis, calmodulin was abletoprotectcatalyticactivity,butnot toxin activity, implying that domains involved in enzyme activity are modified in the presence of calmodulin. Those investigators did not address theissue of a possible role of calcium in the protection fromproteolysis and the experiments were apparently carried outin the presence of calcium. Therefore, it appears that the alteration in sensitivity induced by calcium (as demonstratedin the present study) is separate and distinct from the calmodulin effects observed.


17508

The ultrastructural studies of AC toxin reveal that there is little or no gross change in themolecule at theplateau of low level intoxication, but notable opening which has occurred at 424 p~ free calcium (Fig. 6). This apparentopening is reversible as is the associated functional change; that is, dialysis against EDTA to remove free calcium results in reversion to closed form (Fig. 5) and in elimination of toxin activity. The open form of the toxin would be expected to behave as a larger molecule bygel chromatography. This is consistent with data from Masure et al. (39) who found that an AC fraction with enzyme and toxin activity (designated Peak 1) had greater molecular mass and Stokes radius inthe presence of calcium. These authorsalso noted calcium-dependent shifts inmobility on SDS-PAGE and in heat stability. In contrast, Hanski and Farfel (7) observed that on fractionation of AC toxin by Ultrogel ACA 34, the peak with toxin activity had an apparent size of 190 kDa in the presence of calcium and 340 kDa in the absence of calcium. The cause of the apparent discrepancy between the results of these investigators and the current work is unclear. However, substantial methodological differences make a direct comparison difficult. It is unlikely that the calcium-induced conformational change accounts for the very large forms (>600 kDa) described previously (40,41) which probably represent aggregates of inactive toxin (11). In summary, calcium alters the conformation of B. pertussis AC toxin. This change in conformation, demonstrated by biophysical, biological, immunological, and microscopic techniques, appears to be necessary for maximal intoxication of target cells. The nature of the conformational change and its significance to binding and/or entry of AC toxin will be the subject of subsequent studies. Acknowledgments-We acknowledge the help of the following individuals: Peter Fremgen for laboratory work, Elizabeth Robinson, Jennifer Saines, and Kathryn May for clerical and administrative assistance, Margaret McLeod, Sharon Snider, and George Vandenhoff for assays and reagents, and Drs. Rodney Biltonen and Steve Gonias for helpful suggestions and comments. Thanks are also given to Professor Alastair Wardlaw and the staff of the Department of Microbiology, University of Glasgow, fortheir assistance and support during the preparation of this manuscript. REFERENCES 1. Weiss, A.A., Hewlett, E. L. (1986) Annu. Rev. Microbiol. 40, 661-686 2. Wolff, J., Cook, G. H., Goldhammer, A. R., and Berkowitz, S. A. (1980) Proc. Natl. Acad. Sci. U. S. A. 7 7 , 3841-3844 3. Goldhammer, R., Wolff, J., Cook, G. H., Berkowitz, S. A., Klee, C. B., Manclark, C. R., and Hewlett, E. L. (1981) Eur. J. Biochem. 115, 605-609 4. Greenlee, D.V., Andreasen, T. J.,and Storm, D.R. (1982) Biochemistry 21,2759-2764 5. Kilhoffer, M.D., Cook, G. H., and Wolff, J. (1983) Eur. J. Biochem. 1 3 3 , 11-15 6. Confer, D. L., Slungaard, A. S., Graf, E., Panter, S. S., and Eaton, J. W. (1984) in Adv. Cyclic Nucleotide Protein Phosphorylation Res. 1 7 , 183-187 7. Hanski, E. and Farfel, Z. (1985) J. Biol. Chem. 260,5526-5532 8. Gentile, F., Knipling, L. G., Sackett, D. L., and Wolff, J . (1990) J. Bid. Chem. 265,10686-10692 9. Glaser, P., Ladant, D., Sezer, O., Pichot, F., Ullmann, A., and Danchin, A. (1988) Mol. Microbiol. 2, 19-30 10. Hewlett, E. L., Gordon, V. M., McCaffery J. D., Sutherland, W. M., and Gray, M. C. (1989) J. Biol. Chem. 264, 19379-19384

11. Rogel, A,, Schultz, J. E., Brownlie, R. M., Coote, J. G., Parton, R., and Hanski, E. (1989) EMBO J. 8,2755-2760 12. Masure, H. R., and Storm, D. R. (1989) Biochemistry 28, 438442 13. Leusch, M. S., Paulaitis, S., and Friedman, R. L. (1990) Infect. Immun. 58,3621-3626 14. Bellalou, J., Ladant, D., and Sukamoto, H. (1990) Infect. Immun. 58,1195-1200 15. Glaser, P., Elmaoglou-Lazaridou, A., Krin, E., Ladant,D., Barzu, O., and Danchin, A. (1989) EMBO J. 8,967-972 16. Welch, R. A., Felmlee, T., Pellett, S., and Chenoweth, D. (1986) in Protein-Carbohydrate Interactions in Biological Systems. (Lark, D. L., Normark, S., Uhlin, B. E., and Wolf-Watz, H., eds) pp. 433-438, Academic Press, Inc., New York 17. Boehm, D. F., Welch, R.A., and Snyder, I. S. (1990) Infect. Immun. 58, 1951-1958 18. Hewlett, E. L., Ehrmann, I., Gray, L., Fremgen, P., Allietta, M., Gordon, V. M., and Gray, M. (1990) in Proceeding of the Sixth International Symposium on Pertussis (Manclark, C. R., ed) DHHS Publication 90-1164, pp. 13-19, Department of Health and Human Services, United States Public Health Service, Bethesda, MD 19. Stockbridge, N. (1987) Comput. Biol. Med. 17, 299-304 20. Fabiato, A., and Fabiato, F. (1979) J. Physiol. (Paris) 7 5 , 463505 21. Hewlett, E. L., Urban, M.A., Manclark, C.R., and Wolff J. (1976) Proc. Natl. Acad. Sci. U. S. A. 73, 1926-1930 22. Salomon, Y., Londos C., and Rodbell, M. (1974) Anal. Biochem. 58,541-548 23, Brooker, G., Terasaki, A., and Price, M. G. (1970) Science 1 9 4 , 270-276 24. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 193, 265-275 25. Towbin, H., Staehelin, T., andGordon, J . (1979)Proc. Natl. Acad. Sci. U. S. A. 76,4350-4354 26. Engvall, E., and Perlman, P. (1971) Immun. Chem. 8 , 871-874 27. Laemmli, U. K. (1970) Nature 2 2 7 , 680-685 28. Gordon, V. M., Young, W. W., Jr., Lechler, S. M., Gray, M. C., Leppla, S. H., and Hewlett, E. L. (1989) J. Biol. Chem. 2 6 4 , 14792-14796 29. Ehrmann, I. E., Gray, M. C., Gordon, V. M., Gray L. S., and Hewlett, E. L. (1991) FEBS Lett. 278,79-83 30. Wolff, J., Cook, G. H., Goldhammer, A. R., Londos, C., Hewlett, E.L. (1984) Adv. Cyclic Nucleotide Protein Phosphorylation Res. 1 7 , 161-172 J. M., McKeever, B., Sarma, R., and 31. Blewitt, M.G.,Zhao, London, E. (1984) Biochem. Biophys. Res. Commun. 120,286290 32. Gilles, A.-M., Munier, H., Rose, T., Glaser, P., Krin, E., Danchin, A., Pellecuer, C., and Barzu, 0. (1990) Biochemistry 29,81268130 33. Lakowicz, J. R. (1983) in Principles of Fluorescence Spectroscopy, pp. 345-347, Plenum Publishing Co., New York 34. Devereux, J., Haeberli, P., and Smithies, 0. (1984) Nucleic Acids Res. 12,387-395 35. Ludwig, A., Jarchau, T., Benz, R., and Goebel, W. (1988) Mol. Gen. Genet. 214,553-561 36. Kretsinger, R. H., Tolbert, D., Nakayama, S., and Pearson, W. (1991) in Novel Calcium Binding Proteins (Heizmann, C., ed) pp. 17-38, Springer Verlag, New York 37. Ladant, D. (1988) J. Biol. Chem. 263,2612-2618 38. Raptis, A., Knipling, L., and Wolff, J. (1989) Infect. Immun. 57, 1725-1730 39. Masure, H. R., Oldenburg, D. J., Donovan, M. G., Shattuck, R. L., and Storm D. R. (1988) J. Biol. Chem. 263,6933-6940 40. Kessin, R. H., and Franke, J. (1986) J. Bacteriol. 1 6 6 , 290-296 41. Weiss, A. A., Myers, G. A., Crane, J. K., and Hewlett, E. L. (1986) in Microbiology 1986 (Leive, L., Bonventre, P. F., Morello, J. A., Silver, S. D., and Wu, H. C., eds) pp. 70-74, American Society for Microbiology,Washington, D.C.