|Home | About | Journals | Submit | Contact Us | Français|
We describe a novel hybrid anthrax toxin approach that incorporates multiple components into a single vaccine product. The key domains of protective antigen (PA) and lethal factor (LF) that may be critical for inducing protective immunity are combined into one recombinant molecule. Two LF N-terminal domain-PA hybrids, one with wild-type PA and another with furin cleavage-minus PA, were expressed in E. coli and purified in a native form. Both the hybrids bind to the extracellular domain of the host receptor, CMG2; the wild-type hybrid can be cleaved by furin exposing the LF interacting domain, allowing it to oligomerize into lethal toxin as well as translocation pore-like complexes. The hybrid antigens are immunogenic in Dutch-belted rabbits, eliciting strong PA-specific and LF-specific antibodies. However, the lethal toxin neutralizing antibody titers are 3-7 times lower than those elicited by PA-alum. The hybrid antigens conferred 100% (6/6) protection in rabbits challenged intranasally with a 100 LD50 dose of Bacillus anthracis Ames strain spores.
Bacillus anthracis, the etiologic agent of anthrax, has been high on the list of potential biological warfare and bioterrorist agents due to the ease of preparation and dissemination of its spores. Of the two major virulence factors, the poly-gamma-D-glutamic acid capsule and the tripartite anthrax toxin, the toxin has been widely targeted for development of anthrax vaccines [1, 2].
The anthrax toxin is composed of three components, protective antigen (PA, 83 kDa), lethal factor (LF, 90 kDa) and edema factor (EF, 89 kDa). These proteins are nontoxic as monomers but assemble into potent toxin complexes through a series of protein-protein interactions . PA binds to the host cell receptors, either the capillary morphogenesis protein 2 (CMG2) or the tumor endothelial marker 8 (TEM8) [3, 4], and the N-terminal 20 kDa fragment (PA20) is cleaved off by the furin protease [5,6]. The receptor-bound PA63 (63 kDa) assembles into a heptameric pre-pore ring structure. Up to 3 molecules of LF and/or EF then bind to the heptamer through the N-terminal domain (LFn or EFn) to form lethal toxin (LTx) and edema toxin (ETx). The toxin complexes are internalized via receptor-mediated endocytosis where the pre-pore is converted to pore in the acidic environment of the endocytic vesicle. Translocation of LF (metalloprotease specific for mitogen-activated protein kinase) and EF (adenylate cyclase) through the pore into the cytosol triggers a cascade of cellular metabolic changes leading to cytotoxicity and potentially cell death [7-9].
As protein-protein interactions between PA and LF are central to anthrax toxicity, antigen formulations that induce high titers of antibody (Ab) that can block these interactions are essential for an effective anthrax vaccine. The human anthrax vaccines licensed in the USA (AVA) and the UK (AVP) consist of PA as the principal component with varying amounts of LF and EF [10, 11]. While PA was well documented to induce LTx neutralizing Abs and protective immunity in animal models [12-14], the contributions by LF (or EF) have been sparsely investigated. Recent reports suggest that inclusion of LF in vaccine formulations might provide additional protective benefit [15-19].
We have been exploring novel approaches such as bacteriophage T4 display and transcutaneous immunization to develop multicomponent single product anthrax vaccines [20, 21]. Here we describe the construction of LFn-PA, a hybrid recombinant in which the key domains of PA and LF are engineered into one antigen. The hybrid domains fold into the respective functional states and assemble into oligomeric intermediates of the anthrax toxin pathway. In a Dutch-belted rabbit model, the hybrid antigens elicit high titers of PA- and LF-specific IgG Abs, LTx neutralization titers, and provide 100% protection against intranasal challenge with Ames strain spores.
The T7 expression plasmid pET-28b (Kanr) (EMD Biosciences, Inc) was used for construction of recombinant PA and LFn hybrid clones. Plasmids pPA26 , pYS5-PAU7 (the cleavage site 164RKKR167 of PA was replaced by PGG, which cannot be cleaved by furin or other proteases)  and pLF7 (LFE687C)  were used as templates for PCR amplification of the PA, PA cleavage-minus (PACM), and LFn, respectively. Escherichia coli strains, XL10-Gold and codon-plus BL21 (DE3) RIPL (Stratagene), were used as host strains for initial transformation and subsequent IPTG-induced over-expression of hybrid proteins, respectively.
PCR-based splicing-by-overlap-extension (SOE) was used to construct gene fusions . As shown in Fig. (1A), LFn, the N-terminal PA binding domain of LF [amino acids (aa) 1-262;  was fused to the N-terminus of PA or PACM with a four aa linker, SASA, in the middle. The primers, pl that corresponds to nucleotides (nt) 100-120 of LF gene [5′- CCATCAGCTAGCGCGGGCGGTCATGGTGATGTA-3′] and p2 that corresponds to the LFn-PA junction sequence (5′- AATAACCGGTTCTCCTGTTTAACTTAGCACTTGCTGACCGTTGATCTTTAAGTTCTTCCAAG-3′; of this sequence, nt 1-26 are complementary to nt 88-113 of the PA gene and nt 27-38 correspond to the SASA linker, and nt 39-63 are complementary to nt 864-888 of LF gene) were used to amplify the 837-bp LFn fragment plus the 26 nt 3′-end corresponding to the N-terminus of PA gene. A second PCR was performed using oligonucleotides p3 that corresponds to the LFn-PA junction sequence (5′- CTTGGAAGAACTTAAAGATCAACGGTCAGCAAGTGCTGAAGTTAAACAGGAGAACCGGTTATT -3′; of this sequence, nt 1-25 correspond to nt 864-888 of the LF gene, nt 26-37 corresponding to the linker and nt 38-63 correspond to nt 88-113 of the PA gene) and p4 that corresponds to nt 2268-2292 of the PA gene [5′- CGCGGATCCTTATCCTATCTCATAGCCTTTTTTAGAA -3′) to amplify the 2251-bp PA or PACM gene plus the 25-bp 5′-end corresponding to the 3′-end of LFn gene (underlined sequences correspond to Nhe I or BamH I restriction sites; bolded sequences correspond to SASA linker; italicized sequences correspond to 5′-tags that allow restriction enzyme digestion at the ends of the PCR amplified DNA). The two PCR products were mixed and a third PCR was performed to “stitch” the fragments through overlap extension. The LFn-PA and LFn-PACM fusion products were then amplified using the end primers, p1 and p4, gel-purified, digested with NheI and BamHI, and inserted into the pET28b vector. This results in in-frame fusion of the 18 aa hexa-histidine sequence at the N-terminus of LFn-PA hybrids. The recombinants were transformed into the non-expression strain, E. coli XL10-Gold (Stratagene), and the sequence of the hybrid clones was ascertained by DNA sequencing. The clones were then transformed into the expression strain, E. coli codon-plus BL21(DE3) RIPL.
Highly purified recombinant proteins were prepared in our laboratories for binding assays as well as immunizations, as described below. For ELISA and LTx neutralization assays, purified PA and LF were purchased from List Biologicals Laboratories, Inc., Campbell, CA.
PA and LF (the mutant LFE687C was used in all binding assays) were purified from B. anthracis strain BH445, as described previously [27, 28]. Briefly, the culture filtrate was loaded onto the Phenyl Sepharose column and eluted with a linear gradient from 1.5 M to 0 M ammonium sulfate. The proteins were then loaded onto Q Sepharose Fast Flow anion exchange column and eluted with a linear gradient of 0 to 0.5 M NaCl. The pooled fractions were further purified by Superdex 75 gel filtration.
The PA63 heptamer was prepared by trypsin cleavage of PA and purified as per the procedures described previously . The purified soluble human furin (aa 1 - 604)  was a gift from Dr. Iris Lindberg, University of Maryland Medical School, Baltimore, MD. The purified extracellular domain of the anthrax toxin receptor CMG2 (aa 40 - 218) was a gift from Robert Liddington, Burnham Institute, LaJolla, CA .
For purification of LFn-PA hybrids, a 2-liter bacterial culture expressing LFn-PA or LFn-PACM was induced with 1 mM IPTG for 3 h. The cells were harvested by low speed centrifugation (6,000 × g, 15 min, 4°C) and resuspended in 50 ml PBS buffer (pH 7.4) containing the protease inhibitor cocktail (Roche Diagnostics). The cells were lysed using a French-press (Aminco, Inc.), and the lysate was centrifuged at 10,000 × g for 15 min. The pellet was washed twice with HisTrap binding buffer (50 mM Tris-HCl, pH 8, 20 mM imidazole and 300 mM NaCl), and the protein was solubilized with denaturing buffer (8 M Urea, 50 mM Tris-HCl, pH 8, 20 mM imidazole and 300 mM NaCl). The protein was bound to the HisTrap column (AKTA-prime, GE Healthcare) and renatured by a decreasing urea gradient (8-0 M) and eluted with 20-500 mM linear imidazole gradient. The peak fractions were desalted by passing through the Hiprep10/26 column and concentrated by Amicon Ultra-4 centrifugal filtration (5 kDa cut-off; Millipore). The proteins were then purified by gel filtration on Hi-Load 16/60 Superdex 200 column (prep-grade) (AKTA-FPLC, GE Healthcare) equilibrated with 20 mM Tris-HCl, pH 8 and 100 mM NaCl. Any denatured and/or aggregated protein eluted in the void volume was discarded; only the peak LFn-PA monomer fractions that elute as a symmetrical peak were collected, concentrated, and stored at −80°C.
The purified PA or LFn-PA were treated with furin using a furin:PA ratio of 1:500 (w/w) in 20 μL buffer containing 50 mM HEPES pH 7.5, 2 mM CaCl2, 0.5 mM EDTA, and 0.2% β-octylglucoside. The reactions were performed at 37°C for 45 min and terminated by the addition of phenyl methyl sulfonyl fluoride to 1 mM. Cleavage of 83 kDa PA or 115 kDa LFn-PA to 63 kDa nicked PA (nPA) was quantified by SDS-PAGE and laser densitometry.
PA or LFn-PA was cleaved by furin as described above in the presence of equimolar amounts of LF. SDS sample buffer was then added and the samples were boiled for 5 min. Formation of SDS-resistant complexes was analyzed by SDS-PAGE.
The PA-binding activity of LFn-PA hybrids was determined by incubating 1.5 μg of LFn-PA hybrids (or LF control) with 2.5 μg of furin-cleaved nPA at room temperature for 30 min. The nPA oligomerizes into heptamers, more efficiently in the presence of LF or LFn .
The receptor-binding activity of LFn-PA was determined by incubating 1.2 μg of CMG2 with 1.3 μg of PA or 1.2 μg of LFn-PA hybrids in furin cleavage buffer at room temperature for 30 min. The molar ratios of CMG2 to PA or LFn-PA were 4:1 and 6:1 respectively.
The above samples were electrophoresed on a native 4-12% gradient polyacrylamide gel and stained with Coomassie blue G-250 (Bio-Rad Laboratories, Inc.).
Six to seven week old female Dutch-belted rabbits (Myrtle’s Rabbitry) were used to inject the antigens intramuscularly into the caudal thigh muscle using a 26 gauze needle. Alternate thighs were used for subsequent injections. Six rabbits /group were used for LFn-PA and LFn-PACM antigens and 5 rabbits/group were used for PA and naïve controls. The rabbits were immunized at weeks 0, 4, and 8 with equimolar amounts of either PA (10 μg), LFn-PA (15 μg), or LFn-PACM (15 μg), adsorbed to Alhydrogel® (0.3 mg alu minum/injection; Brenntag Biosector). The rabbits were bled at 2-week intervals and individual serum samples were analyzed for PA-specific and LF-specific IgG antibodies (Abs) and for LTx neutralizing Abs as described below.
Sera from individual rabbits were assayed by ELISA for the presence of PA-specific and LF-specific IgG. Briefly, 96-well flat-bottomed Nunc Maxisorp plates (VWR International, Inc.) were coated with either 0.1 μg/well of PA or LF (List Biological Laboratories, Inc; wild-type LF was used in all ELISA and LTx neutralization assays), blocked with 0.5% casein in PBS, washed and incubated for 1 h at 37°C with the test serum followed by incubation with horseradish peroxidase-labeled, affinity-purified goat anti-rabbit IgG (The Binding Site) and the substrate (2,2′-Azinobis [3-ethylbenzothiazoline-6-sulfonic acid]-diammonium salt; KPL). Plates were read at 405 nm on a SpectraMax 250 plate reader (Molecular Devices). The data are expressed as end point titers defined as the highest dilution that yielded an optical density reading greater than or equal to twice that of the background values. The titers were calculated after subtracting the mean absorbance of duplicate wells lacking antigen from wells containing antigen.
LTx neutralizing Ab titers were determined by the ability of the sera to neutralize the cytotoxicity of LTx in a J774A.1 macrophage cell line as described previously [33,34]. A 4-parametric sigmoid regression curve was used to determine the dilution of antisera that resulted in 50% reduction in toxicity (ED50) of anthrax LTx. Each plate contained control wells that had toxin only (no antisera) and no toxin and no antisera. Polyclonal rabbit anti-PA Ab was used as a positive control.
Rabbits were challenged under BSL-3 conditions at week 10 with 100 LD50 (1 × 107 CFU) of B. anthracis Ames strain spores by the intranasal route as described previously . Briefly, rabbits were anesthetized and suspended vertically by the upper incisors. They were instilled with spore solutions (50 μl/ naris) over 2-3 minutes followed by 50 μl PBS to wash any non-adherent spores from the nasal cavity into the lungs. Rabbits were monitored twice daily for 14 days for signs of disease.
The LFn domain (aa 1-262) of LF , which interacts with the nPA heptamer, was fused to either the full-length wild-type PA (LFn-PA) or the furin cleavage-minus mutant, PACM (LFn-PACM) (Fig. 1A). The recombinant hybrid proteins were over-expressed in E. coli. Appearance of a new band after IPTG induction corresponding to the molecular weight (mol. wt.) of the fusion protein (115 kDa) demonstrated the production of the recombinant LFn-PA hybrid proteins (Fig. 1B, lanes 2 and 4 labeled with an arrow). Most of the over-expressed protein was partitioned into the insoluble fraction.
The LFn-PA hybrid proteins were purified by the denaturation-renaturation method (see Materials and Methods). The purified proteins migrated as a sharp band upon native-PAGE, suggesting proper folding (Fig. 1C, lanes 2 and 3). The purified proteins also migrated as one predominant species upon SDS-PAGE, corresponding to the expected mol. wt. of the hybrid protein (Fig. 1D, lanes 3 and 5). The LFn-PA was cleaved by furin into 2 fragments, a 50 kDa N-terminal LFn-PA20 and a 63 kDa C-terminal PA63 (Fig. 1D, lane 4), whereas no cleavage was evident with the LFn-PACM (Fig. 1D, lane 6).
The in vitro binding experiments show that the LFn domain of the hybrid proteins forms high mol. wt. LTx-hybrid complexes with PA63 heptamer. These complexes (Fig. 2A, lanes 4 and 5) migrate slower than the LTx complexes formed with LF (lane 3), which are reported to have a stoichiometry of (PA63)7-(LF)1-3 [36,37]. This is consistent with the presence of additional PA mass attached to each LFn domain interacting with the PA63 heptamer.
The LFn-PA20 and PA63 fragments produced by furin cleavage of LFn-PA self-assembled (SA) into a LTx-SA complex through interactions between the exposed LFn binding surface of PA63 and LFn-PA20 (Fig. 2B, lane 2), suggesting that both the PA63 and LFn domains of LFn-PA are functional. Considering that all the PA63 was in the complex (note that no PA63 band is seen in lane 2), it should have a stoichiometry of (PA63)7-(LFn-PA20)1-3.
A small fraction of the LTx complexes formed in solution are SDS-resistant. These are thought to mimic the intermediates of acid-mediated pre-pore to pore conversion in the anthrax toxin translocation pathway [1, 38, 39]. To determine whether LFn-PA can form such complexes, furin-cleaved LFn-PA or the PA control was incubated with LF. Both formed high mol. wt. SDS-resistant complexes. Unlike the single complex band visualized in the PA sample (Fig. 2C, lane 1), multiple bands were observed in the LFn-PA sample (lane 2). This is consistent with the expectation that a combination of LFn-PA20, LF, and LFn-PA (uncleaved) would be part of these complexes.
The extracellular soluble domain of the cellular receptor, CMG2, was incubated with LFn-PA, LFn-PACM, or LTx-SA complexes, all having the intact PA domain-4, and the complexes formed were analyzed by native PAGE. The affinity for CMG2 binding to PA is very high (Kd: ~2nM) ; consequently, at the molar ratios used, essentially all the PA complexed with CMG2 (Fig. 3, lane 3). All three hybrids, like the control PA and PA63 (Fig. 3, compare lanes 3 and 5 to lanes 2 and 4), bind to the CMG2 receptor, as evident by the disappearance of the hybrid band(s) and appearance of higher mol. wt. complex bands (Fig. 3, compare lanes 7, 9, and 11 with lanes 6, 8, and 10, respectively).
The immunogenicity of LFn-PA hybrids was tested in our recently established Dutch-belted rabbit model . The rabbits were immunized by intramuscular route with equimolar amounts of PA, LFn-PA, or LFn-PACM absorbed to Alhydrogel® and the PA- and LF-specific Ab responses as well as the LTx neutralization titers were determined. All the groups elicited high levels of PA-specific end point Ab titers (Fig. 4); the LFn-PA hybrids (E. coli-produced) induced titers of 3.8 million (LFn-PACM; p = 0.016 compared to PA) and 4.6 million (LFn-PA), whereas the PA (B. anthracis-produced) induced titers of 9 million (black bars, left axis). The LFn-PA hybrids also induced high LF-specific Ab titers in the range of 0.4 to 1 million (open bars, left axis). In addition, the LFn-PA constructs induced LTx neutralization titers (ED50) of 500-1300 at week-10, whereas PA gave a titer of about 3500 (gray bars, right axis).
The LFn-PA and LFn-PACM immunized Dutch-belted rabbits were challenged intranasally with a 100 LD50 dose of B. anthracis Ames strain spores at week-10 post-immunization. All naïve control rabbits died by day 3 whereas both the LFn-PA and LFn-PACM immunized rabbits, as well as the control PA-immunized rabbits, showed 100% survival (Fig. 5).
The anthrax vaccines licensed in the US (AVA) and UK (AVP) are cell-free culture supernatants of B. anthracis strains V770-NP1-R and Sterne 34F2 adsorbed to either pre-formed aluminum hydroxide or alum precipitate, respectively [41,42]. The principal component of the vaccines that is thought to provide protective immunity against anthrax is the PA [12, 13, 43-45], although contribution by the variable amounts of LF and EF present in these vaccines is possible [10, 11]. More recent efforts have been directed towards the development of pure recombinant PA-alum vaccines that can confer equally good protection. It is however unclear whether the anthrax vaccines would benefit from the addition of certain LF epitopes that could elicit additional protective responses. We have explored a novel approach in which the key domains of the anthrax lethal toxin are combined into a single product vaccine.
Monoclonal Abs with LTx neutralization capacity have been helpful in defining the domains that may be critical for an effective anthrax vaccine [1, 2]. These domains include: i) PA domain-4, which is required for the attachment of PA to the host cell receptor(s); ii) PA domain-1b, which is necessary for interaction with LFn domain; iii) PA domain-2, which is involved in oligomerization of PA; and iv) LFn domain, which interacts with PA63 heptamer forming the LTx complex. We have constructed a single recombinant antigen, LFn-PA, which includes all four of these domains. Additionally, we have constructed a second antigen, LFn-PACM, which lacks the furin cleavage site and therefore could generate stable or alternative intermediates for antigen presentation. A series of in vitro binding assays demonstrated that the functionality of all four domains in the LFn-PA hybrids was comparable to that of their respective native domains. Specifically, the LFn-PA hybrid binds to the host receptor, exposes the LF interacting surface through furin cleavage, and oligomerizes into pore-like complexes. Additionally, the furin cleaved PA63 and LFn-PA20 fragments can self-assemble, as well as further interact with CMG2 receptor, thus mimicking the native LTx complexes.
Both the LFn-PA hybrid antigens generated robust PA-specific Ab titers that were comparable to that of PA-alum. In addition, the hybrids induced strong LF-specific Ab titers, which could be considered an advantage because there is evidence that LF Abs alone can confer protection against anthrax [15, 16]. While the hybrids also elicited LTx neutralization titers, these responses were 3-7 times lower than that of PA. Thus, the inclusion of LFn in the vaccine formulation as a fusion protein appears to be a disadvantage with respect to the abundance of LTx neutralization titers. However this is not to say that LFn does not contribute to the LTx neutralization titers as the assay cannot differentiate contributions of individual domains, nor do our results imply that immunization with LFn or LF would not enhance LTx neutralization titers in a different context. One possible explanation for the weaker LTx neutralization response by the LFn-PA hybrids may be that self-assembly of the furin-cleaved LFn-PA20 to PA63, or the presence of additional LFn-PA20 mass in the cleavage-minus mutant, interfered with the optimal exposure of the neutralization epitope(s) present in PA domain-1b. Alternatively, antigen processing of the hybrid complexes may be altered generating qualitatively different responses. Another contributing factor may be that potential differences exist between the E. coli-produced hybrid antigens vs. the naturally secreted B. anthracis PA; PA modification by deamidation and its conformational sensitivity have been previously reported [46, 47], although their effect on the quality of Ab responses is unknown.
Both the hybrid antigens conferred 100% protection against a high 100 LD50 intranasal challenge with Ames strain spores using our newly established Dutch-belted rabbit model. All control animals died within 3 days of the challenge, demonstrating that the hybrid vaccines are efficacious in providing protection against the most virulent form of anthrax. These data are consistent with the hypothesis that both the Ag-specific Ab titers, as well as the LTx-neutralization Abs, are likely correlates of protection in the Dutch-belted rabbits as was previously reported in the New Zealand white rabbits [12,13,48].
The key domains of the anthrax lethal toxin are combined into a single recombinant vaccine. The LFn-PA hybrid antigens thus constructed retain the biochemical functions of the respective domains, immunogenicity in rabbits, and provide complete protection against intranasal Ames strain spores challenge.
This research was supported by the NIAID grant U01-A1056443 from the National Institutes of Health (VBR). This research was also supported in part by the NIAID contract N01-AI-30065 (JWP) and by the Intramural Research Program (SHL) of the NIAID, NIH.
We gratefully acknowledge the assistance of Ms. Elaine Morrison for performing all of the animal immunization and Mrs. Sarah McCormack for performing the ELISAs.
#The views expressed in this article are those of the authors and do not reflect the official policy of the Department of the Army, the Department of Defense, or the U.S. Government.
DISCLOSURES The research was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals and adheres to principles stated in the Guide for the Care and Use of Laboratory Animals, NRC Publication, 1996 edition. The investigators used facilities fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care, International. All animal experimentation was approved by The University of Texas Medical Branch at Galveston Animal Care and Use Committee. All animals were provided food and water ad libitum.
This is an open access article licensed under the terms of the Creative Commons Attribution Non-Commercial License (http://creativecommons.org/licenses/by-nc/3.0/) which permits unrestricted, non-commercial use, distribution and reproduction in any medium, provided the work is properly cited.