Interaction of Bordetella pertussis Adenylate Cyclase with Calmodulin IDENTIFICATION OF TWO SEPARATED CALMODULIN-BINDING

The structural organization of Bordetella pertussis adenylate cyclase was examined by limited proteolysis with trypsin and/or cross-linking with azido-calmod-ulin a pbotoactivable derivative of its activator, calmodulin (CaM). Adenylate cyclase (which consists of three structurally related peptides of 50, 45, and 43 kDa as judged by sodium dodecyl sulfate-polyacryl-amide gel electrophoresis) formed a 1:l complex with CaM or azido-CaM. CaM-bound adenylate cyclase was cleaved by trypsin into two separate trypsin-resistant fragments of 25 and 18 kDa which both interacted with CaM as judged by their ability to be cross-linked with azido-CaM. These two fragments remained asso- ciated with CaM in a catalytically active conformation resembling that of the undigested complex. When pro- teolysis was carried out in the absence of CaM, the adenylate cyclase was completely inactivated in less than 3 min. Sodium dodecyl sulfate-polyacrylamide gel revealed a single 24-kDa trypsin-resistant fragment. Since this fragment cannot be cross-linked with azido-CaM we suggest that the CaM-binding site on the 25- kDa moiety of the adenylate cyclase is located on a short segment of 1 kDa.

Bordetellapertussis and Bacillus anthracis, causative agents of whooping cough and anthrax, respectively, are the only two prokaryotic organisms which are so far known to secrete an adenylate cyclase (1-3). B. anthracis adenylate cyclase, identified as the edema factor (3, 4), enters animal cells by means of protective antigen and elevates intracellular adenosine cAMP levels. Genetic and biochemical evidence incriminate B. pertussis enzyme as playing a role in the pathogenesis of whooping cough (5)(6)(7)(8)(9). Adenylate cyclase from both organisms exhibits a striking property, namely activation by calmodulin (CaM)' ( 3 , 10). Interaction of CaM with secreted adenylate cyclase of B. pertussis has been investigated mainly from a kinetic standpoint using crude or partially purified enzyme preparations (10)(11)(12)(13)(14)(15). Since CaM was shown to prevent entry of the bacterial enzyme into target cells (9), we were interested to learn more about the mechanism of interaction between B. pertussis adenylate cyclase and CaM. This task was considerably facilitated by our recent purification to homogeneity of adenylate cyclase from culture supernatants of B. pertussis (16).
In this paper I examined the CaM-adenylate cyclase interaction using structural probes, such as a photoactivable derivative of CaM and limited proteolysis by trypsin.
CaM was azidated essentially as described by Zurini et al. (17). To 1 mg of CaM dissolved in 0.5 ml of 60 mM sodium borate, pH 9.8, 0.2 mM CaCl,, 0.1 M NaC1, 0.5 mg of methyl-4-azido-benzimidate dissolved in 0.5 ml of the same buffer were added. After 2 h of stirring at room temperature in the dark, the reaction mixture was desalted on a 10-ml Sephadex G.25 column equilibrated with 25 mM Tris-HC1, pH 7.4,O.l M NaCl, 0.2 mM CaCl,. Incorporation of azido groups into CaM was estimated spectrophotometrically: a value of 1.3 mol of azido groups/mol of CaM was found.
Azido-CaM was stored at -20 "C protected from light. Purification and Assay of Adenylate Cyclase-Concentrated culture supernatant of B. pertussis, 18323, phase I (type strain ATCC 9797) was obtained as described previously (16). The specific activity of concentrated culture supernatant was between 100 and 120 units/mg of protein. Adenylate cyclase was purified in a single step by chromatography on Affi-Gel-CaM (16). Purified enzyme in buffer A (50 mM Tris-HC1, pH 8,0.1% Nonidet P-40,O.l mM CaC1,) with a specific activity of 1600 units/mg of protein could be stored a t -80 "C for several weeks with no loss of activity. SDS-polyacrylamide gel revealed three structurally related bands (see "Results") corresponding to 50,45, and 43 kDa. Adenylate cyclase activity was measured using the procedure of White (18), as modified by Hanoune et al. (19). The reaction was performed at 30 'C in 100 pl of a medium containing 50 mM Tris-HCI, pH 8, 1 mM [w3'P]ATP (5 X IO5 cpm/assay), 6 mM MgCl,, 100 pg of bovine serum albumin, 0.13 mM [3H]cAMP (1.5 X 10' cpm/ assay), 0.12 mM CaCl,, and 0.1 p~ CaM (when added). One unit of adenylate cyclase corresponds to 1 pmol of cAMP formed in 1 min at 30 "C at pH 8.
Zodinatiom-Five ml of concentrated culture supernatant containing about 30 units of adenylate cyclase were mixed with 70 pl of packed Affi-Gel-CaM and shaken overnight at 4 "C. The gel which times with 0.5 M NaCl in buffer A and further with 50 mM Tris-HC1, retained more than 75% of enzyme activity was then washed several pH 8. Iodination of adenylate cyclase bound to Affi-Gel-CaM was performed with chloramine T at room temperature for 5 min with occasional shaking. The reaction was quenched by addition of sodium metabisulfite, followed by addition of 0.7 ml of 8.8 M urea in buffer A. After 30 min of stirring at room temperature, the mixture was 2612 loaded onto a 10-ml Sephadex G-25 column equilibrated in buffer A, to remove both urea and free iodine. The iodinated adenylate cyclase (0.3-1 X lo' cpm/pg of protein corresponding to 0.08-0.25 mol of iodine/mol of adenylate cyclase) was fully active and stimulated by CaM 20-50-fold. Its specific activity was expressed in units per counts per minute of lZ5I.
CaM and azido-CaM were iodinated by the chloramine-T method at room temperature to a specific activity of about 20 Ci/mmol (0.02 mol of iodine/mol of CaM) (20).
Binding of Adenylate Cyclase to Blue-Sephurose-Adenylate cyclase was diluted in buffer B (50 mM Tris-HC1, pH 8, 0.1 mM Ca2+, 0.1% Nonidet P-40, 20% glycerol) to 0.1 unitlml. Two-hun~ed pi of this solution were gently shaken at 4 "C in an Eppendorf tube with 10 pl of a Blue-Sepharose suspension corresponding to 2.5 pl of packed resin and with different concentrations of CaM or azido-CaM. After different times of incubation (between 30 min and 24 h), tubes were centrifuged and the activity of adenylate cyclase in the supernatant was measured. The percentage of enzyme bound to Blue-Sepharose was calculated by subtracting the activity remaining in the supernatant from the initial activity.
Appropriate runs were made in the absence of Blue-Sepharose to account for enzyme inactivation or adsorption to Eppendorf tubes. It should be noted that in the presence of 20% glycerol adsorption was less than 2% in 3 h.
Isolation of the 43-, 45; and 50-kDa Polypeptides on SDS-Polyacrylamide Gel ~~c t r o~~r e s~-T h e iodinated adenylate cyclase preparation was run on a 7.5% SDS-polyacrylamide gel (21), and the proteins were fixed in 10% acetic acid; the gel was further washed with 25% isopropyl alcohol, 10% acetic acid and then exposed for autoradiography at 4 "C. The bands corresponding to the 50-, 45-, loo lr"-- and 43-kDa peptides were sliced from the gel with a razor blade. The slices were placed in siliconized tubes and washed extensively with 25% isopropyl alcohol, then with 10% methanol to remove SDS, and at last dried under a heat lamp. The slices, in Eppendorf tubes, were then soaked in 0.8 ml of 8 M urea in 50 mM Tris-HC1, pH 8, containing 0.1 mM Ca2+ and 1% Nonidet P-40, for 18 h at 37 "C. More than 70% of the iodinated peptides were extracted from the gel slices by this procedure. The urea solutions containing the iodinated peptides were supplemented with CaM (final concentration, 3 pM) and dialyzed extensively against buffer A. Each peptide was assayed for adenylate cyclase activity as described above.
~w f f -~~m~~~f f n a l Tryptic Peptide Maps of Adenylate Cyclase Fragments-The iodinated tryptic peptides T25, T18, and T24 were separated on a 12.5% SDS-polyacrylamide gel and then treated as described above. Iodinated gel slices were digested with 50 pg of TPCK-trypsin in 0.6 ml of 50 mM ammonium bicarbonate, pH 8.0, for 20 b at 37 "C. T h e s u~r n a t a n t solutions containing the solubilized peptides were lyophilized. The lyophilized samples were dissolved in 10 p1 o f buffer and spotted on cellulose thin-layer precoated plates (20 X 20 cm). The first dimension was el~trophoresis at pH 3.7 in pyridine/acetic acid/water (3:30867), run at 400 V for 1 h in a thin layer chromatography chamber (Desaga). The plates were dried at room temperature, and the chromatograms were developed for the second dimension in a system of pyridine/n-butyl alcohol/water/ acetic acid (2030:240.6). The plates were dried and exposed for autoradiography on X-Omat AR films at -70 'C.
Limited Proteolysis with Trypsin-Free or CaM-bound 1251-adenylate cyclase in buffer A was mixed with TPCK-trypsin (w/w ratio; 12). After different times of digestion at 4 "C, aliquots were withdrawn and diluted in buffer A containing soybean trypsin inhibitor in a 20-fold molar excess over TPCK-trypsin. Adenylate cyclase activity was determined in the presence of 0.1 p~ CaM, and the remaining sample was run on a 12.5% SDS-polyacrylamide gel, as described by Laemmli (21). The gels were then dried and exposed on X-Omat AR films at -70 "C for autoradiography.
Photoaffinity Labeling-Photoaffinity labeling experiments were performed at 4 ' C for 1.5 min. The reaction mixture containing either native or proteolyzed adenylate cyclase-azido-CaM complex was irradiated with a "long-wave'' mercury lamp (mineral. light UVSL 58 without screen) positioned at 5 cm from the samples.

Interaction of B. pertussis Adenylate Cyclase with CaM and
A z~-C a M -P r e l i m i n a~ experiments showed that free adenylate cyclase, but not the CaM-complexed enzyme, bound reversibly to Blue-Sepharose at neutral pH. Blue-Sepharosebound adenylate cyclase was released into the medium by excess CaM, with a half-time varying between several minutes at 30 "C and several hours at 4 "C ( Fig. 1).
These differences in affinity for Blue-Sepharose between free and CaM-bound adenylate cyclase prompted me to investigate the CaM-adenylate cyclase interaction by a gel competition method. This method is similar in many respects to that reported by Schubert (22) for determination of the metal ions-nucleotide affinity constant.   If we take into consideration the following two equilibria: adenylate cyclase + Blue-Sepharose (1) e (adenylate cyclase) (Blue-Sepharose) adenylate cyclase + n(CaM) e (adenylate cyclase) (CaM), (2) the partition coefficient, i.e. the ratio between enzyme in solution and enzyme bound to the gel, in the presence (kd) or absence ( k s ) of CaM will be defined by the following equation: KD being the dissociation constant of the adenylate cyclase- "The three polypeptides (50, 45, and 43 kDa) were separated on SDS-polyacrylamide gel and renatured as described under "Experimental Procedures." " Iodinated adenylate cyclase preparation before SDS-polyacrylamide gel.
CaM complex. As shown in Fig. 2, kd is a linear function of CaM concentration, either in the presence or absence of calcium ions, which is consistent with a stoichiometry of 1:1 for adenylate cyclase-CaM interaction. Dissociation constants calculated from data obtained in several experiments range from 0.09 to 0.17 nM in the presence of Ca2+ and from 13 to 23 nM in the presence of a large excess of EGTA. When dissociation constants have been calculated from CaM-adenylate cyclase dose-response curves, the values obtained were similar (not shown). Despite these differences in affinity of B. pertussis adenylate cyclase for CaM, once the complex was formed in the presence of Ca2+, addition of excess EGTA did not promote its dissociation (Fig. 3). This is consistent with our previous observation that adenylate cyclase bound to CaM-agarose cannot be eluted by EGTA.
Azido-CaM, a photoactivable derivative of CaM (23), behaves almost identically to the parent compound, both in activating the bacterial adenylate cyclase (not shown) and binding to the enzyme (Fig. 2). Thus, the half-maximum activating concentrations of azido-CaM are the same as those of CaM either in the presence of Caz+ or EGTA. KO values calculated from data shown in Fig. 2 indicated a similar affinity of CaM and azido-CaM for adenylate cyclase.
Solid-phase Iodination of Pure Adenylate Cyclase-Solidphase iodination of pure adenylate cyclase bound to Affi-Gel-CaM yielded an active 'ZsII-adenylate cyclase preparation which was activated up to 50-fold by 100 nM CaM. Autoradiography after SDS-polyacrylamide gel of the iodinated enzyme revealed the same three peptides of 50, 45, and 43 kDa (Fig. 4) which were also detected by Coomassie Blue staining of the CaM-Affi-Gel-purified enzyme (16). Peptide mapping of these three bands gave similar patterns (not shown).
In order to determine whether all three peptides were endowed with adenylate cyclase activity, they were separated by SDS-polyacrylamide gel and the corresponding bands, revealed by autoradiography, excised; the iodinated polypeptides were extracted with 8 M urea, 1% Nonidet P-40 (see "Experimental Procedures"). Upon dialysis in the presence of CaM, each polypeptide recovered adenylate cyclase activity (Table I).
Moreover, as will be shown below, the 50-and 45-kDa peptides can be converted to the 43-kDa species by limited proteolysis. These results suggest that the three peptides are structurally related. Since adenylate cyclase is released extracellularly, it seems likely that the three peptides arose from a differential processing of a common precursor during secretion.
Limited Proteolysis of Purified '2sI-Adenylate Cyclase by Trypsin-Incubation of free I21-adenylate cyclase with trypsin at 4 "C (at a 1:l (w/w) ratio) resulted in complete inactivation of the enzyme within 3 min (Fig. 5A); the 50-, 45-, and 43-kDa polypeptides were converted to a 24-kDa fragment units/ml in buffer A containing 3 p~ of CaM) was digested with 0.5 pg/ml of trypsin a t 4 "C. At the indicated times a 100-excess of soybean trypsin inhibitor was added over trypsin and adenylate cyclase activity was assayed; then the corresponding samples were run on a 9% ( A ) or a 12.5% ( B ) SDS-polyacrylamide gel and autoradiographed. Lanes a-d: the 50-kDa polypeptide was digested for 0, 0.5,5, and 10 min, respectively (corresponding % of adenylate cyclase activity: 100, 94, 59, and 43, respectively). Lanes e-h: the 45-kDa polypeptide was digested for 0, 0.5,5, and 10 min, respectively (% of adenylate cyclase activity: 100, 86, 56, 46, respectively). Lanes i-1: the 43-kDa polypeptide was digested for 0,0.5,5, and 10 min, respectively (% of adenylate cyclase activity: 100, 92, 71, and 44, respectively). Note that in A the T18 fragment cannot be seen because it had migrated with the front of the gel. (T24) which was largely resistant to further proteolysis, suggesting a compact structure. When "'I-adenylate cyclase was complexed with CaM prior to exposure to trypsin, more than 50% of the enzymatic activity remained after 20 min of incubation. SDS-polyacrylamide gel analysis revealed a rapid conversion of the 50-and 45-kDa peptides to the 43-kDa species which was progressively cleaved to a major, trypsininsensitive fragment of 25 kDa (T25), and a minor one of about 18 kDa (T18) (Fig. 5A). The T24 fragment obtained in the absence of CaM appeared to be structurally related to T25 as shown by their two-dimensional tryptic peptide map (Fig. 6). The generation of the minor iodinated peptide T18, obtained only in the presence of CaM, was highly reproducible with the same '"I-adenylate cyclase preparation. The apparent yield of this peptide varied with different preparations of '"I-adenylate cyclase; for instance in a previous report it was not detected (16). It is likely that differences in intensity could be accounted for by differences in iodination of the regions that yield T25 and T18 due to the number of exposed tyrosines in each fragment.
T o elucidate the origin of T25 and T18, the bands corresponding to these peptides (Fig. 5A) have been excised for quantification. As shown in Fig. 5B, the kinetics of appearance of the two tryptic peptides was similar and in addition, their ratio remained constant. Thus, it is unlikely that T18 is derived from T25. Moreover, the two-dimensional tryptic peptide map of T25 was significantly different from that obtained for T18 (Fig. 6), suggesting that T25 is not related to T18. Thus, T25 and T18 might represent two different domains of adenylate cyclase.
To determine further whether all three native polypeptides (50, 45, and 43 kDa) contain both T25 and T18 domains, limited proteolysis was performed on the separated iodinated polypeptides complexed with CaM. As shown in Fig. 7, upon exposure to trypsin, the 50and 45-kDa bands were rapidly converted into a 43 kDa one, which was further cleaved into T25 and T18. These results demonstrate that each polypeptide of 50, 45, and 43 kDa contains a common region of 43 kDa which can be proteolytically cleaved into two peptides of 25 and 18 kDa.
Gel Filtration Chromatography of Trypsin-cleaved Adenylate Cyclase-CaM Complex-Analysis of the trypsin-cleaved ade- nylate cyclase-CaM complex on AcA-44 gel filtration chromatography revealed a unique peak of adenylate cyclase activity comigrating with the main peak of radioactivity (Fig.   8). Its apparent M, was identical to that of the undigested adenylate cyclase-CaM complex (60 kDa). SDS-polyacrylamide gel analysis of the fractions corresponding to the peak of activity revealed the same ratio between T18 and T25 as shown in Fig. 5A. This indicates that after proteolysis, T25, T18, and CaM remained associated by noncovalent interactions in an active, native-like structure. Identification of CaM-binding Sites in Adenylate Cyclase by Photoaffinity Labeling with Azido-CaM-Preliminary experiments revealed that photolysis of a mixture of azido-"'I-CaM and pure or crude preparations of adenylate cyclase resulted in one major cross-linked product exhibiting a M, of 70 kDa on SDS-polyacrylamide gel (Fig. 9) which was not observed when photolysis was carried out in the presence of a 25-fold molar excess of unmodified CaM. In the absence of photolysis no cross-linked product could be revealed. In a similar way, photolysis of a mixture of ""I-adenylate cyclase and azido-CaM resulted, again, in one major cross-linked product with a M , of 70 kDa on SDS-polyacrylamide gel (Fig. 9). Since the apparent M, of CaM or azido-CaM on SDS-polyacrylamide gel was 20 kDa, the 70-kDa cross-linked product corresponds most likely to covalent attachment of one adenylate cyclase polypeptide to one azido-CaM molecule.
When pure 'TI-adenylate cyclase was complexed with azido-CaM, then submitted to trypsin proteolysis for 10 min, 69% of enzymatic activity was retained while the enzyme was entirely converted into T25 and "18 polypeptides. When the digested complex was subsequently photolyzed, three new cross-linked species of 63, 45, and 38 kDa were evidenced (Fig. 1OA, lune 2). The 45-and 38-kDa bands might correspond to covalent attachment of azido-CaM to T25 and T18, respectively, whereas the 63-kDa band might result from covalent attachment of azido-CaM to both T25 and T18. Digestion of '"I-adenylate cyclase with trypsin prior to photolysis in the presence of azido-CaM did not yield any crosslinked product (Fig. lOA, lane 3). Similar experiments using unlabeled adenylate cyclase complexed with azido-lZ5I-CaM gave essentially the same results: after 10 min of trypsin digestion, the same cross-linked peptides of 63, 45, and 38 kDa were detected upon photolysis (Fig. 10B, lane 2). Again, trypsin digestion of adenylate cyclase for 2 min prior to addition of azido-lZ5I-CaM did not lead to any cross-linked species upon photolysis (Fig. 10B, lane 3 ) . DISCUSSION CaM stimulation of B. pertussis adenylate cyclase has been widely documented (10)(11)(12)(13)(14)(15)24). The present results confirmed that CaM binds strongly to adenylate cyclase even in the absence of Ca2+ albeit with less affinity than in the presence of the divalent cation. The 1:l CaM-adenylate cyclase complex did not dissociate upon addition of EGTA or in media of high ionic strength. This is not an unprecedented case since it has been shown that CaM (the 6 subunit) remained associated with muscle phosphorylase kinase in the absence of Ca2+ (25).
Controlled proteolysis of adenylate cyclase by trypsin and/ or photoaffinity labeling with azido-CaM suggested that adenylate cyclase is composed of two separate domains of 25 kDa (T25) and 18 kDa (T18) which both interact with CaM. After exposure of the adenylate cyclase-CaM complex to trypsin, the CaM molecule would bridge the two cleaved domains in a structure resembling the native complex. T25 and T18 associated with CaM appeared as very resistant species toward further proteolysis; in contrast, when proteolysis was carried out on adenylate cyclase in the absence of CaM, only a 24-kDa trypsin-resistant fragment (T24) was detected. The specific radioactivity of this fragment as well as its tryptic map were similar to that of T25, suggesting that it corresponds to the same region of the native polypeptide. Since this 24-kDa fragment could not be cross-linked with azido-CaM, it is likely that the extra 1-kDa peptide present in T25 is involved in the binding of CaM. Edelman et al. (26) have already described differences in affinity for CaM of chymotryptic fragments of myosin light chain kinase differing by only 2 kDa at their C terminus.
These results suggest that adenylate cyclase contains a compact domain of about 24 kDa largely resistant to trypsin proteolysis whether CaM was present or not. In contrast the T18 fragment displays resistance toward protelysis only in the presence of CaM. This suggests that binding of CaM to adenylate cyclase could induce a conformational change in the T18 domain rendering it more resistant to proteolysis; alternatively, CaM could mask potential cleavage sites on T18 thus making them inaccessible to trypsin. In both cases, the protective effect involves a multisite interaction between CaM and T18 as depicted in the model presented in Fig. 11.
The structure proposed for B. pertussis adenylate cyclase differs significantly from those known for other CaM-dependent enzymes which can be modified by proteolytic cleavage to yield active CaM-independent forms (17,(26)(27)(28)(29)(30). Attempts to identify a CaM-independent catalytic domain of B. pertussis adenylate cyclase have failed thus far. It remains to determine whether this could be achieved using other approaches or if it represents a profound structural difference between B. pertussis adenylate cyclase and other CaM-dependent enzymes.