Potent neutralization of anthrax edema toxin by a humanized monoclonal antibody that competes with calmodulin for edema factor binding

Abstract

This study describes the isolation and characterization of a neutralizing monoclonal antibody (mAb) against anthrax edema factor, EF13D. EF13D neutralized edema toxin (ET)-mediated cyclic AMP (cAMP) responses in cells and protected mice from both ET-induced footpad edema and systemic ET-mediated lethality. The antibody epitope was mapped to domain IV of EF. The mAb was able to compete with calmodulin (CaM) for EF binding and displaced CaM from EF-CaM complexes. EF-mAb binding affinity (0.05–0.12 nM) was 50- to 130-fold higher than that reported for EF-CaM. This anti-EF neutralizing mAb could potentially be used alone or with an anti-PA mAb in the emergency prophylaxis and treatment of anthrax infection.

Infection by inhalational anthrax is often fatal if treatment is delayed. Anthrax bacteria can be killed by vigorous treatment with antibiotics, but patients may still die because the lethality of anthrax is largely because of the action of toxins (1). Anti-toxin neutralizing monoclonal antibodies (mAbs) are the only viable choice for immediate neutralization of toxin and they could augment the effectiveness of antibiotics.

Anthrax bacteria produce 3 toxin components: Protective antigen (PA), lethal factor (LF), and edema factor (EF) (2, 3). PA binds to cellular receptors and acts as a vehicle to deliver LF or EF into the cytosol where they exert their enzymatic activities (4–8). LF is a zinc-dependent protease that cleaves mitogen-activated protein kinase kinases and causes lysis of macrophages (9, 10). EF is a calcium-calmodulin (CaM)-dependent adenylate cyclase and causes local inflammation and edema (11). The combination of PA with LF results in lethal toxin (LT). LT can replicate symptoms of anthrax disease when injected into animals (12). PA together with EF forms edema toxin (ET) and ET can produce a range of toxic effects in the host (11, 13).

PA has been regarded as the most important target for prophylaxis and therapy of anthrax, because PA is common to both LTs and ETs, initiates the toxic process via receptor binding, and is highly immunogenic. In fact, PA is the major component in the current anthrax vaccine and the target for most of the available human or human-like neutralizing mAbs that have been shown to be very effective in protection against anthrax toxin or spore challenge (14–19). However, there is evidence that LF and EF may play important roles in providing protective immunity (20–22). Furthermore, concerns that PA could potentially be manipulated, such that it would no longer be neutralized by current anti-PA neutralizing mAbs have led to interest in therapeutics against the other 2 toxin components. A mixture of mAbs that recognize distinct epitopes on multiple toxin components (PA, LF, or EF) would not only enhance the protective efficacy but also broaden the spectrum of protection. Thus, in recent years, several anti-LF mAbs have also been reported (23–27). However, no anti-EF neutralizing mAbs have been reported to date. A previous report had indicated that immunization with the PA-binding N-terminal domain of EF (amino acids 1–254) resulted in polyclonal sera containing both EF and LF neutralizing activities (28).

The purpose of this study was to determine (i) if anti-EF neutralizing mAbs could be isolated; (ii) the effectiveness of such antibodies against anthrax ET effects; and (iii) the neutralization mechanism of these antibodies. We have made a Fab combinatorial phage display library from chimpanzees that were immunized with anthrax toxins (17). From the library, 4 EF-specific Fabs were recovered, and 1 of them had potent neutralizing activity independent of the homologous PA-binding N-terminal (1–254) domain of LF. In this report, we describe the detailed characterization of these anti-EF clones.

Results

Isolation and Characterization of EF-Specific Fabs.

A phage library expressing chimpanzee γ1/κ antibody genes was constructed after immunization of chimpanzees with full-length EF. After 3 rounds of panning against EF, 96 individual EF-specific clones were identified by phage ELISA. Sequence analysis identified 4 unique Fab clones with distinct VH and VK sequences. These were designated EF12A, EF13D, EF14H, and EF15A. The closest human V-gene germline origins of the 4 clones were assigned by conducting sequence similarity searches of all of the known human Ig genes. The findings are shown in Table 1.

The Fabs were converted to full-length chimeric IgG molecules with human γ1 heavy chain constant regions, and the binding specificity of all 4 clones for EF was verified by ELISA (Fig. 1A). No binding was observed for negative controls LF, PA, BSA, thyroglobulin, lysozyme, or phosphorylase b.

Fig. 1.

Antibody binding and in vitro neutralization assays. (A) ELISA titration of 4 anti-EF mAbs: Recombinant EF was used to coat the wells of an ELISA plate. Wells were then incubated with various dilutions of mAbs, and bound IgGs were detected by the addition of peroxidase-conjugated anti-human Fc antibody followed by TMB substrate. (B) EF13D inhibits ET-mediated cAMP production in cells. ET was first incubated with serial dilutions of antibody. The ET-mAb mixture was then incubated with RAW264.7 macrophage cells as described in the Methods. The cAMP production levels were assessed with the BioTRAK cAMP enzyme immunoassay from Amersham Pharmacia Biotech according to the manufacturer's protocol. Results were plotted and analyzed with Prism software (version 5, Graphpad Software).

mAb EF13D Potently Neutralizes ET In Vitro and In Vivo.

Of all of the anti-EF mAbs tested for neutralization of ET activity, only a single antibody, EF13D, was able to inhibit ET-mediated cyclic AMP (cAMP) production in RAW264.7 cells and did so with an EC₅₀ of 10–118 ng/mL. Thus, the potency of EF13D was comparable to the potent anti-PA mAb 14B7 in a similar assay (29), but lower than that of anti-PA mAb W1 (17) (Fig. 1B). We tested the in vivo efficacy of mAb EF13D in 2 different models. The antibody significantly inhibited ET-mediated edema in the mouse footpad model when premixed with toxin before administration in the footpad (Fig. 2 A–C) as well as when given systemically (IV) before footpad administration of toxin (Fig. 2D). The antibody was able to prevent ET-induced footpad edema almost completely when compared with untreated or PBS-treated mice after administration of 3 different doses of toxin. More significantly, systemic preadministration of mAb EF13D (50 μg per mouse) also provided significant protection against ET-mediated death in the mouse model after challenge with a lethal dose of ET in C57BL/6J mice (Fig. 3).

Fig. 2.

Anti-EF mAb EF13D protects mice against ET-mediated footpad edema. (A) Comparison of left footpad treated with 0.5 μg ET (Top), 0.5 μg ET premixed with 5 μg anti-EF mAb EF13D (Middle), and ET premixed with 5 μg anti-PA mAb W1 (Lower). (B and C) mAb EF13D (5 μg) and 5 μg W1 mediated inhibition of footpad edema induced by ET (panel B, 0.5 μg; panel C, 1.0 μg). Averages and standard errors presented are based on n = 3 mice per treatment. (D) mAb EF13D (50 and 100 μg) injected IV 1 h before toxin administration inhibited footpad edema induced by 0.25 μg ET. Footpad experiment results shown are for the Balb/cJ strain.

Fig. 3.

Anti-EF mAb EF13D protects mice against ET-mediated death. C57BL/6J mice received a single IV dose of 50 μg mAb EF13D or PBS 1 h before a lethal dose of ET (25 μg, IV), and survival was monitored for 100 h.

mAb EF13D Recognizes EF Domain IV.

We next wanted to establish the mechanism for the potent neutralizing activity of EF13D. Full-length mature EF protein consists of 776 aa and is structurally organized into 4 domains (Fig. 4A). Domain I (amino acid 1–259) binds to PA, the interface of domain II and III (amino acid 260–588) forms the catalytic center, and domain IV (amino acid 589–766) is connected to domain IIb by a linker (30, 31). Binding of CaM induces a large reorganization of the structure that creates the catalytic center. We initially tested whether EF13D could recognize LF or FP119 (a fusion protein of EF domain I and diphtheria toxin) in Western blots, as domain I (amino acid 1–259) of EF has been previously shown to contain PA-binding epitopes shared by EF and LF (28). We found that EF13D did not react with LF, PA, or FP119 in Western blots, suggesting the epitope was not in the N terminus. Next we constructed and purified truncated EF peptides representing different domains of the toxin (Fig. 4B). The binding assay (ELISA) showed that, of all of the anti-EF antibodies, only EF13D IgG bound to domain IV (Fig. 4 C and D). This suggested that the difference between neutralizing and nonneutralizing mAbs we isolated could be because of their different binding sites. As domain IV is a helical domain that facilitates CaM-mediated activation of EF, binding to this region by antibody could interrupt this process and thereby block the toxin's enzymatic activity.

Fig. 4.

mAb EF13D recognizes EF domain IV. (A) Schematic diagram of full-length EF showing the domain structure. Numbers refer to the amino acid residues at the beginning of the each domain or the beginning and end of the protein itself. (B) A diagram of 7 constructs generated in this study. Numbers refer to the amino acid residues at the beginning and end of each peptide. (C) Reactivity of various peptides with anti-EF mAb EF13D or anti-His mAb. (D) Reactivity of full-length EF (amino acid 1–766) and truncated EF (amino acid 589–766) with 4 EF-specific Fabs. Expressed peptides were fused to his tag and were detected by 1 μg/mL anti-EF mAbs or anti-His mAb (1:3,000). Each reaction was done in triplicate. The average binding signal with standard deviation for each reaction upon binding by anti-EF or anti-His was plotted.

mAb EF13D Prevents CaM Binding to EF and Can Also Displace Prebound CaM from EF.

To determine if the mAb competes with CaM for binding to EF, we performed native gel electrophoresis and analytical ultracentrifugation (AUC) assays. Because the mobility of proteins or protein complexes on native gel is dependent on their size, conformation, and charge, and the Phast native gel system allows visualization of sharp bands, we assumed that the mobility of a complex made of 3 components (Fab·EF·CaM) would be easily distinguished from that of the complex made of 2 components (Fab·EF or EF·CaM). When EF was mixed with 4-fold molar excess of Fab, the EF·Fab complex and free Fab were seen as distinct bands on the native gel (Fig. 5A, lane 3). Interestingly, the mixture of 3 components (EF, Fab, and CaM) also always resulted in the same shifted band as the binary complex (EF·Fab), in a manner independent of whether CaM was prebound to EF (EF-CaM) or added after EF-Fab complexes were formed (Fig. 5A, lane 4 and 5). Therefore, it appeared that CaM could not bind or remain bound to EF when Fab was present. We hypothesized that despite the clear charge and conformation-based shift difference CaM induced in EF mobility after binding to EF (causing the EF·CaM band to migrate much faster than EF) (Fig. 5A, lane 2), perhaps the ternary complex Fab·EF·CaM was simply indistinguishable from the Fab·EF complex because of CaM-mediated conformational changes. Thus, to make sure that the CaM was not present in the shifted complexes formed from ternary reaction of EF, Fab, and CaM, Phast-Western blots with anti-CaM antibody were performed. While reactivity with the free CaM and the EF·CaM bands were strong, CaM was not detected in the expected ternary complexes. Next, to rule out the possibility that the inability to detect CaM by Western blot in the complex was because of insufficient transfer of the large complex (158 kDa) to nitrocellulose membranes or other epitope masking issues, we labeled CaM with an IR-dye. This allowed direct detection of CaM on gels with high sensitivity in any complex in which it was present. Although labeling resulted in only partial conversion of EF to EF·CaM, it was clear that the trace amounts of Fab·EF·CaM complex (green bands) was formed, which migrated faster than the Fab·EF complex (red bands) (Fig. 5B). Therefore, in all of the ternary mixtures (different premix configurations), Fab·EF complex was the predominant species, although with high sensitivity labeling, trace amounts of the ternary complex can be detected. The same results were obtained in 8 experiments performed with both IgG and Fab. Thus, it appeared that anti-EF neutralizing antibody was not only able to prevent CaM from binding to EF but could even displace CaM that was prebound to EF.

Fig. 5.

Analysis of complex formation among CaM, EF, and Fab on the native gel and by AUC. (A) Phast-gel electrophoresis of single, binary, and ternary complex of EF, Fab, and CaM followed by Coomasie blue staining. Lane 1, EF; lane 2, EF + CaM; lane 3, EF + Fab; lane 4, EF-CaM + Fab; lane 5, EF-Fab + CaM. (B) Phast-gel electrophoresis of protein complex with IR dye-labeled CaM. Gels were scanned at 800 nm for labeled CaM (green) and at 700 nm for Coomasie blue-stained proteins (red). Lane 1, EF; lane 2, EF + CaM; lane 3, EF + Fab; lane 4, EF-CaM + Fab; lane 5, EF-Fab + CaM; lane 6, Fab; lane 7, CaM. The shifted bands of Fab·EF binary complex were indicated by arrows (A and B). (C) The sedimentation coefficient distributions c^(P)(s) obtained from the Bayesian analysis of sedimentation velocity data of individual proteins, binary mixture and ternary mixture. The sedimentation profile of the ternary mixture is shown in the Inset. c^(P)(s) curves: CaM (blue), Fab (yellow), EF (green), mixture of CaM and EF (cyan), mixture of Fab and EF (red), ternary mixture of 3 proteins (black).

We next confirmed the results from the native gel analyses by sedimentation velocity analytical ultracentrifugation (SV-AUC), taking advantage of the size-dependent migration difference in the centrifugal field. Sedimentation coefficient distributions were calculated from the measured concentration profiles (Fig. 5C). For solutions of single-protein components, the sedimentation data yielded major peaks for CaM, Fab, and EF with sedimentation coefficients of 1.87 S, 3.62 S, and 4.92 S, respectively. Because of the high sensitivity of SV-AUC for trace populations, minor peaks from low level impurities and aggregates were also observed. The binary mixtures showed close to stoichiometric binding, with only minor populations of unreacted free species visible at the s-values of the individual components. For the mixture of CaM and EF, as indicated in Fig. 5C, the complex sedimented as a major peak at 5.46 S, whereas for the Fab and EF mixture, a major peak at 6.16 S confirmed formation of Fab·EF. For the ternary mixture, the major peak appeared at an s-value of 6.17 S, mutually identical to the value for the binary complex Fab·EF. A peak corresponding to CaM·EF was not detected. This suggested that a ternary complex was not present. Theoretically, assuming the common (2/3)-power scaling law of sedimentation coefficients relative to molar masses of globular proteins, one would expect a ternary complex to sediment with s-values between 7.2–7.7 S. There is only a minor peak at ∼7.1 S, which we believe to represent low level aggregates in this experiment. In fact, in a separate experiment at higher (equimolar) protein concentrations, this peak could not be detected. That the major species in the ternary mixture is the binary Fab·EF complex is supported by the fact that, in this mixture, the peak of free CaM reappears with an area quantitatively corresponding to the loading concentration. Further, the virtually exclusive presence of Fab·EF complexes in the ternary mixture is supported independently by the analysis of the signal amplitudes from the absorbance relative to the interference data, taking advantage of the much lower extinction coefficient of CaM compared with EF and Fab .

These results unambiguously show that under the SV-AUC conditions, Fab and CaM bind with high negative cooperativity to EF. In a separate SV-AUC experiment, we observed the same Fab·EF complex formation when Fab was added to a CaM-EF mixture preincubated for 2 h, confirming that Fab can displace CaM from EF.

mAb EF13D Binds to EF with High Affinity.

Our results that the antibody can displace prebound CaM from EF suggest that EF13D binds to EF more tightly than CaM does. Surface plasmon resonance (SPR) analyses showed that the dissociation of EF from the Fab·EF complex was extremely slow. Using the binding affinity distribution model, the k_on was determined as 10⁵ M⁻¹s⁻¹. Because of the extremely slow dissociation, K_d and k_off could not be directly measured. However, the lower and upper limit of k_off can be estimated. Consequently, based on the relationship between K_d, k_off, and k_on (K_d = k_off /k_on), the upper limits of k_off and K_d are 1.2 × 10⁻⁵ s⁻¹ and 0.12 nM, respectively. For the lower limit of k_off at 10⁻⁶ s⁻¹, the estimated K_d from the distribution analysis is 0.05 nM. Thus, the affinity for Fab-EF interaction is 50- to 130-fold higher than that of EF-CaM (32).

Discussion

We report here the isolation and characterization not only of an EF neutralizing mAb, but one with high binding affinity (K_d of 0.12 to 0.05 nM). The neutralizing mAb, EF13D, was as potent as anti-PA mAb 14B7 in preventing cAMP induction by ET in vitro. More impressively, this antibody protected mice from ET-mediated edema formation and death. The differential binding to truncated EF peptides identified domain IV of EF as the target of the antibody. Crystallographic analysis has demonstrated that domain IV is not part of the catalytic core, but contains the major sites for CaM binding and undergoes dramatic conformational changes upon CaM binding (30). Native gel and AUC analyses indicated that EF13D was not only able to prevent CaM from binding to EF, but also able to displace CaM from EF. This suggested to us that the antibody had a higher binding affinity for EF than CaM did. Indeed, SPR experiments yielded an estimate for the K_d of 0.12 to 0.05 nM for Fab-EF interaction, compared with 6.7 nM reported for EF-CaM (32). Of course, the binding affinity of EF-CaM is highly dependent on the concentration of calcium (32), but the binding affinity of EF13D for EF appears to be at least 50-fold higher than that of CaM for the toxin even under the highest affinity EF-CaM interaction conditions. The mAb epitope appears to be conformational because the antibody could not interact with the denatured form of EF on Western blots.

The presence of trace amounts of the ternary complex, which ran faster than the binary complex of Fab·EF, was detected in the gel assay when CaM was labeled with IR-dye. This is somewhat different from results obtained by AUC-SV; the latter did not reveal any presence of the ternary complex. The discrepancy might be because of variation in the experimental conditions (such as salt, pH, protein concentration, difference in reactivity between labeled and unlabeled CaM, and sensitivity of detection of the complex). However, if there was a ternary complex of Fab·EF·CaM formed, the amount was very small.

The binding of CaM is essential for the activity of EF, because the EF·CaM complex is 1,000-fold more active than is EF alone (30, 33). Thus, inhibition of the formation of the EF·CaM complex could be an important goal of passive immunoprophylaxis and reversal of complex formation an important goal of antibody therapy. Indeed, a search for small molecules that can inhibit interaction between EF and CaM has been pursued, and a molecule with a low inhibitory activity (IC₅₀ of 10–15 μM) has been found (34). Here, we report a naturally occurred EF-neutralizing mAb that is not only able to inhibit the interaction between CaM and EF, but able to displace prebound CaM from the EF·CaM complex. While the ability of EF13D to displace CaM from EF is certainly fascinating and could be 1 of the neutralizing mechanisms, it is unlikely to account for all of the potency with which this mAb neutralizes EF. Because antibody is not expected to enter the cytoplasm with EF, where CaM is readily available, it is far more likely that this antibody blocks unfolding and translocation of EF through the PA heptamer channel in a manner similar to what has been described for antibodies that block botulinum toxin action (35). The use of this mAb alone or in combination with anti-PA mAbs may not only improve the efficacy, but also broaden the spectrum of protection against anthrax infection.

Methods

Toxins.

PA was made from Bacillus anthracis in our laboratory as previously described (36). EF and FP119 (a fusion protein of the N terminus of EF and diphtheria toxin) were made from Escherichia coli as described previously (37).

Construction and Selection of Phage Library.

The combinatorial cDNA library of chimpanzee γ1/κ antibody genes was constructed by inserting the heavy and light chains at XhoI/SpeI and SacI/XbaI sites, respectively, in pComb3H vector as described previously (38). The library was panned against recombinant EF protein immobilized on ELISA wells for 3 rounds, and EF-specific clones were selected by 96-well phage ELISA (39).

Production and Purification of Fab and IgG.

Removal of the phage coat protein III-encoding region from phagemid DNA by digestion with NheI/SpeI and religation resulted in phagemid encoding soluble Fab. Soluble Fab was expressed and purified on a nickel-charged column as described (38). The conversion of Fab to IgG and the expression of IgG were carried out as described previously (38). The purity of Fab and IgG was determined by SDS polyacrylamide gel electrophoresis (NuPAGE MOP; Invitrogen). Protein concentrations were determined both by BCA assay (Pierce) and by measurement of OD₂₈₀, assuming that 1.35 OD₂₈₀ is equivalent to 1 mg/mL.

Determination of Binding Specificity by ELISA.

EF, LF, PA at 5 μg/mL and unrelated proteins (BSA, thyroglobulin, lysozyme, and phosphorylase b; Sigma) at 10 μg/mL in carbonate buffer (pH 9.5) were coated on ELISA microtiter plates. ELISAs were performed as described previously (17).

Nucleic Acid Sequence Analysis of EF-Specific Fab Clones.

The genes coding for the variable region of heavy (VH) and light (VL) chains of EF-specific clones were sequenced, and their corresponding amino acid sequences were determined. The presumed family usage and germ-line origin were determined for each VH and VL gene by search of V-Base, which is a compilation of all of the available human variable region IG germ line sequences (40).

Affinity Measurement by Biosensor Analysis.

A SPR biosensor (BIAcore 3000; GE Healthcare) was used to measure the binding kinetics and affinity of the interaction between EF and EF13D Fab. The experiments were performed at 25 °C in the running buffer HBS (10 mM HEPES, pH 7.4, 150 mM NaCl, 1 mM EDTA, 0.005% P20) using a CM3 sensor chip (GE Healthcare) with a carboxymethylated dextran matrix. The purified Fab was covalently attached to the chip surface by amine coupling. An immobilization level of 900 RU was reached after applying 10 μg/mL Fab in NaOAc buffer, pH 5.5, on the surface. Binding experiments were conducted by injecting 3-fold serial dilutions of recombinant EF ranging from 0.1 to 300 nM to the chip surface. After each round of injection, the bound EF was completely removed by 2 passages of 10 μL 2 M NaCl, 0.01% Triton X-100 followed by 10 μL 0.1% SDS. The kinetic traces were globally fitted with a model for continuous affinity and rate constant distributions (41).

ET In Vitro Neutralization.

ET (100 ng/mL = 100 ng/mL PA + 100 ng/mL EF) was prepared in DMEM in a 96-well plate. Antibodies were diluted serially directly into the toxin mixture and incubated for 1 h at 37 °C. Anti-PA antibody 14B7 (29) was used on each plate as a positive control. The ET-mAb mixtures were then transferred to RAW264.7 macrophage cells grown to 80–90% confluence in 96-well plates. Cells were incubated for 1 h and total cAMP levels were assessed using the BioTRAK cAMP enzyme immunoassay from Amersham Pharmacia Biotech according to the manufacturer's protocol.

Epitope Mapping.

Genes coding for full length mature EF (amino acids 1 to 766), domain I (amino acids 1 to 259), domain II to domain IV (amino acids 260 to 766), domain IIb, III and IV (amino acids 316 to 766), domain IIb and IV (amino acids 456 to 766), domain II and III (amino acids 260 to 588), and domain IV (amino acids 589 to 766) were amplified by PCR and cloned into pET-31b (Novagen) at NdeI and XhoI sites. The peptides were in frame at the C terminus with 6 histidines in the vector. Constructs were confirmed by DNA sequencing. Proteins were expressed from E. coli BL21(DE3)pLysS (Promega) and purified through ProBond nickel-chelating affinity column (Invitrogen).

The purity of the peptides was determined by SDS/PAGE and the identity by Western blots probed with anti-His antibody (Sigma). Each peptide was tested for its binding to mAb EF13D as well as to anti-His by ELISA. An anti-His reaction was used to quantify each coated peptide. In brief, 5–10 μg/mL of each peptide were coated onto the wells of a 96-well plate. The plate was then incubated with 1 μg/mL EF13D IgG or anti-His-HRP conjugate (1:3,000) for 2 h at room temperature (RT). Wells containing EF13D IgG were next incubated with anti-human IgG Fc-HRP conjugate (1:5,000; Jackson ImmunoResearch) for 1 h. HRP activity was assessed by adding tetramethylbenzidine substrate (TMB; Sigma) and measured at OD₄₅₀. Each reaction was performed in triplicate. The average binding signal with standard deviation for each reaction with anti-EF or anti-His was plotted.

A similar ELISA was used to determine if the other nonneutralizing anti-EF Fabs bound to the shortest peptide (amino acid 589–766, EF589–766) that was still reactive with EF13D.

Native Gel Analysis and Western Blotting.

Human, recombinant CaM was from BioMol. A portion was labeled with IRDye 800CW using a kit (Li-Cor Biosciences) per manufacturer's protocol. EF was prebound to CaM or IR-dye-labeled CaM by mixing 3 μM EF and 40 μM CaM for 1 h at RT. EF13D was then added to EF-CaM or EF (EF and antibody at final concentrations of 1.7 and 7.5 μM), and the mixtures were incubated for 1 h before running on native Phast gels (Amersham Pharmacia Biotech). In other experiments, 3 μM EF was prebound to 12 μM EF13D before the addition of 40 μM CaM. All binding experiments were performed with CaCl₂ present (0.35–0.45 μM). Phast Gel were scanned in both 800 and 700 nm channels using the Odyssey Infrared Imaging System (Li-Cor Biosciences), before and subsequent to Coomasie blue staining. For Western blotting, gels were removed from plastic backing and transferred to nitrocellulose according to protocols recommended by the manufacturer (Amersham Pharmacia Biotech) for the Phast Gel System. The blot was incubated with anti-CaM mAb (1:1,000; Upstate). In epitope mapping experiments, PA, LF, EF, or FP119 (4–10 ng of each) were subjected to western blotting after conventional SDS/PAGE on 4%–20% Tris-Glycine gels (Invitrogen), using EF mAbs (1:500) as primary antibody and IR-dye-conjugated secondary antibodies. Blots were scanned using the Odyssey Infrared Imaging System.

AUC Sedimentation Velocity.

Samples in 400 μL at 1 μM each in working buffer (5 mM HEPES, 50 mM NaCl, 100 μM CaCl₂, pH 7.5 at 20 °C) were loaded into 12-mm path length double sector cells and were centrifuged at 50,000 rpm at 20 °C in a ProteomeLab-XL-I analytical ultracentrifuge (Beckman Coulter) after the standard protocol (42). The evolution of the resulting concentration gradient was monitored by the interference optics and absorbance optics.

Sedimentation coefficients for the protein constructs were obtained from the analysis of the sedimentation velocity data using the software SEDFIT. Twenty-three scans were loaded and modeled with a continuous c(s) distribution model using a resolution of 200 s-values between 0.1 to 15 S and maximum entropy regularization with a 68% confidence interval. It should be noted that the width of the distribution is governed by the regularization and the noise in the data acquisition and that the area under the peaks, not the peak heights, corresponds to the species concentrations. The s-value of the protein species was determined as the weight-averaged sedimentation coefficient of the peak. After the analysis of the single protein samples, the obtained sedimentation coefficients for the 3 molecules of interest were used as a priori information for the Bayesian analysis (43) of the data of the mixture.

Animal Studies.

Chimpanzees 1603 and 1609 were immunized with recombinant PA, LF, and EF, and bone marrow aspirates were used for library construction as described previously (17). Female BALB/cJ or C57BL/6J mice were purchased from Jackson Laboratories and used at 8–12 weeks of age. For the footpad edema model, BALB/cJ mice were injected in 1 foot pad with 0.5 or 1.0 μg ET premixed with anti-EF mAb EF13D, anti-PA mAb W1 (17) or PBS. Antibodies were used at 5 μg. Footpad injection volumes were 20 μL. Footpad edema was monitored at 20 and 46 h after injection by measuring dorsal/planar and medial/lateral sizes using digital calipers. In alternative experiments, 50 or 100 μg per mouse mAb EF13D were preadministered to mice via i.v. (IV) injection 1 h before 0.25 μg ET injection into footpads, and footpad size was monitored. To test for efficacy against ET lethality, C57BL/6J mice received a single mAb EF13D injection (50 μg, IV) 1 h before administration of a lethal dose of ET (25 μg ET, IV), and survival was monitored for 100 h. All experiments involving animals were performed under protocols approved by the respective Institutes as well as by the Animal Care and Use Committee of the National Institute of Allergy and Infectious Diseases, National Institutes of Health.