Structural Studies of Immune Recognition
Summary: Pamela Björkman is studying proteins and cells in the immune system, using x-ray crystallography, electron tomography, and biophysical methods.
We are interested in structural mechanisms of recognition in the immune system, specifically in homologs and viral mimics of class I major histocompatibility complex (MHC) proteins, and in the structure, function, and therapeutic uses of antibodies and their receptors. In addition to using x-ray crystallography and biophysical techniques to analyze protein-protein interactions in solution, we also use electron tomography and confocal microscopy to image interactions in cells, examining, for example, transport pathways mediated by the class I MHC-related neonatal Fc receptor (FcRn), a receptor for immunoglobulin G (IgG). We also are applying our antibody structure expertise to a new area for us: "Engineering Immunity" against HIV.
Classical class I MHC proteins present peptide antigens derived from self- and nonself proteins to T cells during immune surveillance. The MHC structure seems ideally suited for its antigen presentation function, in that it includes a groove that is perfectly shaped to accommodate short peptide antigens. MHC homologs share similar three-dimensional structures with classical MHC molecules but have different functions, including immune functions (antibody transport by FcRn; evasion of host immune responses by viral MHC mimics) and nonimmune functions (regulation of iron or lipid metabolism by HFE and ZAG; chaperoning pheromone receptors by M10 proteins).
Our crystal structures revealed that FcRn, HFE, ZAG, M10, and the poxvirus protein 2L do not present peptides and therefore play no role in conventional adaptive immune responses. The FcRn, HFE, and 2L grooves are collapsed, and cocrystal structures with their respective protein ligands show that each uses a protein-protein interaction mode different from MHC interactions with peptides or T cell and other receptors. Our recent structure of 2L complexed with the host inflammatory cytokine tumor necrosis factor-α (TNFα) revealed the structural basis for its picomolar affinity for TNFα and may facilitate design of anti-inflammatory protein drugs. By contrast, the grooves of ZAG, M10, and the human cytomegalovirus MHC mimic UL18 are open and theoretically capable of antigen binding, but only UL18 associates with peptides. These results raise questions about the primordial function of the MHC fold: Did it originally arise for peptide presentation/T cell interactions as part of the adaptive immune response (a relatively recent acquisition of the vertebrate immune system), or did it arise for the seemingly more ancient functions of protein transport or metabolite regulation? Perhaps surprisingly, our results suggest the former. Our studies of MHC homologs provide striking examples that structure does not always dictate function: similar structures can adopt different functions, and conversely, similar functions can be accomplished by very different structures.
We extended our characterizations of FcRn, an MHC-related receptor for IgG antibodies, to include cell biological studies of intracellular trafficking. FcRn is the receptor that transfers maternal IgG to the bloodstream of fetal and newborn mammals, thereby passively immunizing the neonate against pathogens likely to be encountered prior to development of its own fully functional immune system. Transfer of IgG involves trafficking of FcRn-IgG complexes in acidic intracellular vesicles across an epithelial cell barrier in the placenta (for prenatal transfer) or the intestine (for postnatal transfer). A general question exemplified by FcRn trafficking is how cargo-containing intracellular vesicles are transported to their correct ultimate locations—for example, how does the cell know that FcRn-IgG complexes should be transported across the cell for eventual release of IgG into the blood, whereas other receptor-ligand pairs should be transferred to degradative compartments?
To study the process by which FcRn-IgG complexes are correctly trafficked across cells, we use electron tomography, a form of electron microscopy, to derive three-dimensional maps of transport vesicles in neonatal rat intestinal epithelial cells at resolutions of 4–6 nm. To facilitate these studies, we developed gold-labeling and enhancement methods to locate individual IgG fragments bound to FcRn inside intracellular vesicles. Our three-dimensional images of IgG transport reveal tangled webs of interlocking IgG-containing transport vesicles, some of which are associated with microtubule tracks to allow movement via motor proteins. Other IgG-containing vesicles include multivesicular bodies, normally associated with degradative functions but apparently functioning in IgG transport in the specialized proximal small intestinal cells of a neonate.
To complement these high-resolution, but static, studies, we do fluorescence imaging in live cells, which allows tracking of labeled vesicles and quantification of the velocities and directions of FcRn-positive vesicles. We have used fluorescent imaging to characterize the intracellular trafficking pathways of two other Fc receptors: the polymeric immunoglobulin receptor (pIgR), which transports polymeric IgA antibodies into secretions, and gE-gI, a viral Fc receptor for IgG. We discovered that gE-gI exhibits a pH-dependent affinity transition for binding IgG that is opposite that of FcRn: FcRn binds tightly to IgG at acidic, but not basic, pH, so as to bind IgG inside acidic vesicles during transport and to release IgG upon encountering the slightly basic pH of blood; by contrast, gE-gI binds IgG at the pH of blood but not at the pH of intracellular vesicles. We are testing the hypothesis that circulating IgG taken up by gE-gI by receptor-mediated endocytosis is destined for degradation after dissociating from gE-gI in acidic intracellular vesicles, which could form part of a viral mechanism to escape from antibody-mediated host immune responses.
In addition to studying antibody receptors, we began a new project to improve upon the binding and neutralization properties of antibodies themselves. This work is part of a collaboration with David Baltimore's laboratory (California Institute of Technology) to "Engineer Immunity" against HIV. The idea is to direct lifelong production of specified antibodies or antibody-like proteins with desired properties; for example, neutralizing antibodies or designed antibodies engineered to bind more tightly to a pathogen or to recruit immune effector cells. The antibodies would be produced in vivo by infecting autologous hematopoietic stem cells with lentiviral vectors bearing specific antibody genes, thus allowing lifelong production of anti-HIV proteins.
Our portion of the project involves designing, producing, and testing novel anti-HIV protein reagents in an effort to find proteins with increased efficacy in HIV neutralization. Although HIV has evolved to evade most or all antibodies (hence the difficulty of finding an immunogen capable of eliciting a strong neutralizing antibody response in vaccine development efforts), an attractive feature of the Engineering Immunity approach is that we are not limited to the traditional architecture of an antibody. Thus we can produce and express antibody-like proteins of different sizes (to facilitate access to hidden epitopes) and valencies (i.e., with different numbers of combining sites) and/or link antibodies to HIV-binding proteins such as the host receptor CD4.
In our initial efforts, we developed CD4-antibody fusion proteins that cross-react to neutralize a broad range of HIV strains, and characterized a dimeric form of an anticarbohydrate antibody, 2G12, that displays a 50- to 80-fold increased potency in the neutralization of clade B HIV strains. We also proposed a previously unappreciated general mechanism that HIV uses to evade antibodies. Our hypothesis states that an anti-HIV antibody fails to potently neutralize because it can only bind using one of its two antigen-binding sites. Simultaneous engagement of both antigen-binding sites leads to a synergistic effect called avidity, in which the antibody-antigen interaction can become nearly irreversible. With most viruses, antibodies bind with avidity because the antigenic spikes are present on the viral surfaces at high densities, a feature that is absent on HIV. The small number of antigenic spikes on the surface of HIV are mostly separated by distances that are too large to allow simultaneous engagement of both antibody-combining sites. In addition, the structure of the HIV spike trimer prohibits simultaneous binding of both combining sites to a single spike. We are constructing hinge-extended antibodies that would enable simultaneous binding by both antigen-binding sites, either within a spike or between spikes, thereby reducing the concentration of antibody required for sterilizing immunization to realistic levels.
The studies of Fc receptors are funded by the National Institutes of Health. Studies of anti-HIV antibodies are funded by the Bill and Melinda Gates Foundation.
As of May 30, 2012