We have recently found that viral antigens expressed in tandem with HBsAg will spontaneously assemble into 22-30 nm virus-like particles which contain about 20% lipid, based on the density of the particles. The particles are lipid-protein micelles, with proteins displayed on the surface of lipid droplets. This milieu resembles the surface of an enveloped virus like HIV. It also resembles a bacterial toxin bound to the surface of a cell . Transmembrane glycoprotein gp41 is an important target for HIV neutralizing antibodies. Human monoclonals such as 2F5 bind a site on gp41 close to the lipid bilayer. They can neutralize a broad range of HIV isolates, including fresh isolates and regardless of T cell or macrophage trophism. So far, vaccines have been unable to elicit similar antibodies. We hypothesize that they failed because they lack lipid, and cannot fold or display the 2F5 site in an immunogenic form. We have engineered constructs in which gp41 is expressed in tandem with HBsAg and have created the first gp41-rich particles. HBsAg is an intrinsic membrane protein and spans the membrane four times. By linking gp41 to it, the transmembrane domain of gp41 can span the membrane as well, allowing lipid-depend epitopes to form. We have expressed three hybrids. In the first, the ectodomain of gp41 extends into the aqueous phase. The second construct includes the membrane-spanning domain of gp41, so the hybrid may span the membrane a fifth time. This would create an ideal environment for the 2F5 site to assume its native conformation, just above the lipid layer. The third construct is the same as the second, except that a furin cleavable site is provided between HBsAg and gp41. After cleavage in the mature particle, gp41 may gain the rotational freedom needed to form trimers, just as it does in forming spikes on the surface of HIV virus. Each of these proteins was expressed in baculovirus, and each assembled particles in preliminary experiments. We are scaling up production, to analyze the particles by EM and study their immunogenicity. In addition, we will use monoclonals specific for the trimer to detect spike formation on the surface of the particles and determine whether this depends on cleavage from HBsAg. Particulate gp41 binds 2F5, and we will study its affinity, as a measure of lipid-dependent folding. By providing a membrane-spanning form of transmembrane protein gp41, these particles may be the first immunogen to recreate gp41 as it exists on the virus itself. By combining the native structure with a highly-potent particulate immunogen, we may create a greatly improved vaccine based on gp41. Since the neutralizing sites on gp41 are conserved among a great diversity of HIV isolates, antibodies elicited by these antigens have the potential to be broadly neutralizing. Similarly, we have created a series of three constructs containing protective antigen PA of anthrax. Antibodies to this protein are known to protect against anthrax. We anticipate that the particulate form of PA may greatly increase immunogenicity, allowing us to immunize with fewer doses and to maintain immunity with less frequent boosting than the current anthrax vaccine. In addition, by providing PA on a lipid surface, we may elicit qualitatively better antibodies to sites involved in subunit binding and translocation events that depend on PA activation. The three constructs will be tested for immunogenicity, and the resulting antibodies will be tested for the ability to block toxin-mediated killing of cell lines in culture. If successful, we will immunize mice and challenge them with anthrax spores, in collaboration with NIH scientists.
