PMC

Mice produce antibodies binding the M1H5 and M2B10 neutralizing epitopes whether immunized with ACT (A) or RTX₉₈₅ (B). Sera at a 1:200 dilution were incubated on ELISA plates coated with ACT, before addition of 0.1 nm M2B10, M1H5, or M1F11 as a competitor. After incubation, immobilized monoclonal antibody was detected with anti-human Fc-HRP, with absorbance normalized to wells without sera. Lower absorbance indicates a higher concentration of epitope-specific murine antibodies.

ACT and RTX domains bind purified α_Mβ₂ receptor. Soluble murine α_Mβ₂ receptor was coated onto ELISA plates and blocked, and ACT or domains were serially diluted in M-PBST. Bound protein was detected with polyclonal rabbit anti-ACT antibody and goat anti-rabbit HRP. To assess nonspecific binding, control wells were not coated with α_Mβ₂ receptor but blocked with M-PBST only. A, RTX₇₅₁*, and B, acylated ACT* showed receptor-dependent binding, although ACT also exhibited significant nonspecific binding. All other domains showed no specific or nonspecific binding.

RTX dominates the human immune response to ACT. Nine serum samples from humans exposed to B. pertussis were tested for reactivity to the catalytic domain, RTX₇₅₁, or intact ACT by ELISA. Absorbance values at a 100-fold dilution of the sera were normalized to that of ACT at the same dilution. A paired t test was used to determine the statistical significance between signals for CAT and RTX binding domains. ***, p ≤ 0.001.

M2B10 and M1H5 antibodies block the ACT-α_Mβ₂ integrin interaction. A, antibody neutralization of cAMP intoxication using J774A.1 presenting the α_Mβ₂ receptor and CHO-K1 cells lacking the receptor. Both assays were performed with a 160-fold molar excess of scAb over ACT and analyzed as in . B, antibody blockade of ACT binding to J774A.1 cells assessed by FACS. Biotinylated ACT was incubated with a 300-fold molar excess of scAb and added to 4 × 10⁵ J774A.1 suspension cells on ice. After washing, bound biotinylated ACT was detected with streptavidin-PE and analyzed by FACS (mean fluorescence noted next to each peak). Controls include untreated cells (Cells only) and cells treated with nonbiotinylated ACT followed by streptavidin-PE (ACT). C, antibody blockade of ACT binding to soluble α_Mβ₂ integrin by ELISA. ACT (0.5 μg/ml) was incubated with serial dilutions of M2B10, M1H5, 3D1 and 2A12 antibodies at 5-, 1.6-, and 0.5-fold molar excess, before transfer to an ELISA plate coated with murine α_Mβ₂ integrin. Bound ACT was detected with rabbit anti-ACT polyclonal antibody followed by HRP-conjugated goat anti-rabbit IgG antibody. D, antibody neutralization of ACT secreted by B. pertussis. Antibodies at 10, 2.5, 0.63, and 0.16 μg/ml were incubated with live B. pertussis (A₆₀₀ = 0.2) before adding to adherent J774A.1 cells. The resulting intracellular cAMP concentrations were measured, normalized to total protein concentration, and expressed as % relative cAMP.

RTX domain is immunodominant and elicits neutralizing antibodies. A, immunization with ACT yields sera preferentially recognizing the RTX domain. Purified domains were coated at equal moles on microtiter plates, with sera serially diluted starting at 1:200. The average EC₅₀ for individual domains is shown. Error bars are the standard deviations of the EC₅₀ value among four mice. B, immunogenicity of purified domains. Mice were immunized with intact ACT or individual domains, with the serum EC₅₀ for ACT measured by ELISA after the first boost (6 weeks) or second boost (8 weeks). C, sera from mice immunized with the ACT and RTX₉₈₅ domain neutralize cAMP intoxication similarly. Sera from each immunization group at a 1:400 dilution were incubated with 125 ng/ml ACT in DMEM before adding to J774A.1 cells. Intracellular cAMP levels were measured by cAMP ELISA, divided by the total protein concentrations, and normalized to cells treated with ACT alone, as in . pre indicates baseline sera collected prior to immunization. Statistical significance was determined by one-way analysis of variance with Tukey's test; **, p ≤ 0.01; ***, p ≤ 0.001. For all panels, * indicates an acylated domain. n.s. indicates not significant (p > 0.05).

Expression and purification of intact ACT and domains. A, adenylate cyclase toxin domain architecture. ACT is a 177-kDa protein toxin, consisting of five sequential domains as follows: the catalytically active N-terminal adenylate cyclase (CAT) domain, the central hydrophobic (HP) domain, modification (Mod) region carrying two acylation sites, Lys-860 and Lys-983, the RTX domain, and finally a C-terminal secretion signal (Sec). The five shaded blocks in the RTX region represent tandem Gly-Asp-rich repeats. B, ACT forms high molecular mass species (>600 kDa) when urea is removed by dialysis or dilution. Size exclusion chromatograms of ∼120 μg of ACT using a Superdex 200 column were collected directly after a 1:10 dilution (final urea concentration 0.8 m) or overnight dialysis into HBSC with 1 m urea. Arrow indicates the size expected for monomer, 177 kDa. C, SDS-polyacrylamide gel comparing full-length ACT with different domain constructs after purification from E. coli. Three versions of the CAT domain (residues 1–373 (CAT₃₇₃), 1–385 (CAT₃₈₅), and 1–400 (CAT₄₀₀)) and three versions of the RTX domain (residues 985–1706 (RTX₉₈₅), 751–1706 (RTX₇₅₁), and acylated 751–1706 (RTX₇₅₁*)) were expressed from plasmid pET28a with an N-terminal His₆ tag. The HP domains (residues 399–1096 (HP₁₀₉₆) and acylated 399–1096 (HP₁₀₉₆*)) were expressed from plasmid pMalc-5x with an N-terminal MBP fusion for enhanced solubility and C-terminal His₆ tag.

Two novel neutralizing epitopes are present in the RTX domain. A, two representative neutralizing scAbs were converted to chimeric IgG1/κ antibodies, in which the murine variable regions are appended by human constant domains. These were transiently expressed in CHO-K1 cells and purified by (NH₄)₂SO₄ precipitation and protein A affinity chromatography. SDS-PAGE (4–20% gradient gel, 2 μg loaded) shows high purity and expected size of the recombinant IgG as follows: 25 kDa (light chain) and 50 kDa (heavy chain) when reduced and 150 kDa under nonreducing conditions. ELISA demonstrates the ACT domain specificity of the M2B10 (B) and M1H5 (C) IgG antibodies. Microtiter plates were coated with ACT or ACT domains at equimolar concentrations, followed by serial dilution of antibody from 1 nm, and followed by detection with anti-human Fc antibody-HRP conjugate. D, competition ELISA determined that the M2B10 and M1H5 antibodies bind novel nonoverlapping epitopes. A 200-fold molar excess (20 nm) of previously described murine mAbs (3D1, 2A12, 10A1, 2B12, 9D4, 6E1, 7C7, and 1H6) () or scAb versions of M1H5 and M2B10 were mixed with M2B10 and M1H5 IgG (0.1 nm) and incubated on ACT-coated ELISA plate, with bound M2B10 or M1H5 detected as above. The absorbance was normalized to that of M2B10 or M1H5 with no competitor; absorbance significantly <1.0 indicates competition between the antibody pair.

ACT domain oligomeric state and secondary structure. Purified domains were separated by size exclusion chromatography (Superdex200 column, except Superdex75 for CAT₄₀₀), with far UV circular dichroism spectra (Jasco J-815) used to assess secondary structure in the presence of 2 mm CaCl₂ or the absence of calcium ions. A, catalytic domain, spanning residues 1–400 (CAT₄₀₀), eluted as a single peak of expected size. The secondary structure is similar to that observed in the CAT₃₇₃ crystal structure (56% helix, 14% strands, 13% turns, and 17% unordered). B, HP domain, spanning residues 399–1096, with an N-terminal MBP fusion protein eluted off SEC as high molecular weight aggregates whether acylated (*) or nonacylated. C, RTX₉₈₅ formed high molecular weight aggregates in the absence of calcium ions but eluted as two peaks, one corresponding to the expected molecular mass of 78 kDa in the presence of calcium. Circular dichroism revealed significant conformational change upon addition of 2 mm CaCl₂ corresponding to an increase in β-strand content. D, RTX₇₅₁ domain exhibits a similar calcium-dependent delay in elution volume, which is more pronounced when the protein is acylated (*), and under these conditions it yields a single monomer peak. Inset, SDS-polyacrylamide gels show proteins present in the indicated peaks, with arrows indicating the expected monomer size.

ACT immunization induces a diverse antibody response. A, phylogenetic tree depicting antibody sequence relatedness was generated using the light and heavy variable region amino acid sequences. Neutralizing scAbs are colored gray, with unique shapes denoting recognition of distinct epitopes among this antibody group as determined by competition ELISA. Open symbols denote antibodies whose binding does not depend on the presence of calcium. Antibodies competing with previously characterized monoclonal antibodies are indicated; all antibodies bind RTX except M1C5, M1F11, and M2G5, which bind CAT₄₀₀. B, representative SDS-PAGE of scAbs after purification by IMAC and Superdex S200. Arrow indicates expected size of ∼40 kDa, 2 μg each of M1F11, M1C12, M1H5, and M2B10 scAbs were loaded. C, 21 unique scAbs identified from the immune phage libraries were tested for the ability to neutralize ACT-mediated increases in intracellular cAMP concentration. ACT was incubated with a 160-fold molar excess of scAb protein before adding to J774A.1 cells. Data are reported as the percent relative cAMP, calculated from the total cAMP concentration in the cellular lysate as determined by cAMP ELISA, divided by the protein concentration of the lysate, and normalized to control cells treated only with ACT (open bar). Error bars indicate range of duplicate assays. D, 21 scAbs were evaluated for their ability to rescue J774A.1 macrophages from ACT-induced lysis, using a similar protocol as for cAMP neutralization. Cell lysis was measured via lactate dehydrogenase release using the Cytotox 96 kit (Promega), normalized to control cells treated only with ACT (empty bar), and reported as the percent relative lysis. Error bars indicate standard deviation of triplicate assays.