The focus of this work has been to understand the molecular details that control initial steps in the recognition of cells infected with pathogens such as viruses by cells of the innate and adaptive immune systems. In particular, we study the large family of major histocompatibility complex (MHC)-encoded molecules from a biophysical and structural perspective. Thus, we are interested in how MHC-I molecules interact with receptors on natural killer (NK) cells or on T lymphocytes through their NK and T cell receptors, respectively. These studies are dependent upon structural, functional, and biophysical analysis of the interaction of the molecules in question, and attempts are made to correlate binding properties with function and structure. Large DNA viruses of the herpesvirus family produce proteins that mimic host MHC-I molecules as part of their immunoevasive strategy, and we have directed our effort to understand the function, cellular expression, and structure of a set of these MHC-I molecules encoded by the mouse cytomegalovirus (mCMV). We have analyzed the expression of several of these genes by transfection in different cell types, and have established that, unlike the classical MHC-I molecules, the viral MHC-I molecules (referred to as MHC-Iv) do not require either beta-2 microglobulin or self-peptide for expression. In addition, MHC-Iv molecules m152 and m155 are not expressed at the cell surface under any circumstances. By engineering the m144 molecule for expression in E. coli, we produced X-ray diffraction quality crystals of m144 in complex with its beta-2 microglobulin light chain, and published the structure of this first viral MHC-I-like molecule. This study revealed, for the first time, that one of the mCMV MHC-Iv molecules indeed preserves the fundamental molecular fold characteristic of classical MHC-I molecules. The structure reveals a molecule that lacks bound peptide, and has unique structural features that contribute to thermal stability. Mutagenesis experiments confirm the importance of a unique disulfide bond of this molecule. In addition, we have examined the cellular expression of additional cytomegalovirus MHC-Iv immunoevasins, M37, m151, and m153. Each of these has unique cell biological features. In collaboration with Stipan Jonjic, University of Rijeka, we have generated monoclonal antibodies to m153, and have recently demonstrated that m153 forms stable non-covalent homodimers. Diffraction quality crystals of selenomethionyl m153, expressed in Drosophila S2 cells, were obtained. Using single anomalous dispersion (SAD) methodology, we have solved the structure of m153, and refined the structure to 2.4 Angstrom resolution. This crystallographic model, coordinates and structure factors, have been deposited in the Protein Data Bank under accession number 2O5N, and a manuscript describing the structure in detail has been published. The most striking feature of this completely new MHC-like structure is that the molecule forms a stable head to tail homodimer. To confirm the dimerization interface observed in the crystal structure of m153 we have introduced alanine mutations at four residues involved in hydrogen bonds at the interface. The mutant m153 and wild type m153 was expressed transiently in baculovirus insect cells and purified. Analysis of the wild type and mutant m153 proteins by size exclusion chromatography and analytical ultracentrifugation experiments indicated that the mutations destabilized the dimer and that the mutant protein existed as a monomer. The biological function of m153 is unknown. To gain insight into its role during viral infection we have developed several strategies to identify the putative ligand of m153. Initial immunoprecipitation experiments using metabolically labelled mouse fibroblasts, infected with MCMV or transfected with m153, failed to identify any binding partners in the mouse fibroblasts. However, more recent efforts to identify a ligand by pull-down experiments and mass spectrometry analysis of the coprecipitated product suggest that a ligand might be one of several cell surface molecules expressed on dendritic cells. In a second, complementary, approach, we constructed an m153 reporter cell as a tool to screen cell lines and primary cells for ligands. The indicator cell line 43.1 is a T cell hybridoma in which an NFAT-driven GFP construct was stably introduced. Upon triggering of the TCR by antibody cross-linking these cells produce GFP. To generate an m153-specific GFP reporter cell line a fusion construct consisting of the extracellular domain of m153 and the transmembrane and intracellular portions of the human zeta protein was generated. The m153-Zeta construct was introduced in the 43.1 cell line by retroviral infection. The resulting m153-Zeta/43.1 cells produce GFP after overnight culture on plates coated with the anti-m153 monoclonal antibody m153.16 and thus provide a valuable ligand-screening tool. A number of murine cell lines from various origins were screened with the m153-Zeta reporter cells, but none stimulated the production of GFP. Recently we have found that freshly isolated spleen cells from several different mouse strains stimulate the m153-Zeta reporter cells to produce GFP, whereas the parental 43.1 cells and a control cell line, which expresses another viral glycoprotein fused to human zeta, were unaffected by coculture with the splenocytes. Further fractionation of the spleen cell populations indicate that CD11c+ dendritic cells (DC) are the most potent in activating the indicator cells. Expression cDNA libraries from DCs have been made, and we are in the process of screening this in an effort to identify this ligand. As a complementary study of mCMV-encoded immunoevasins, we have examined the direct interaction of m152, a known mCMV immunoevasin, with the stress-induced MHC-I-like molecule Rae-1beta. We have expressed the extracellular domains of RAE-1beta as inclusion bodies in bacteria, and refolded them in vitro. A soluble recombinant form of the extracellular domain of m152 has been purified from insect cells. Results from native gel shift assay and size exclusion chromatography suggested a direct interaction between m152 and RAE-1 beta. Both sedimentation equilibrium and sedimentation velocity analysis as well as isothermal titration calorimetry revealed that m152 binds RAE-1beta tightly (KD<5 microM) and at a 1:1 ratio. These studies clearly indicate a direct interaction of m152 with RAE-1βand lead to further experiments to define the molecular details of this interaction. We have also examined a set of different isoforms of RAE-1, RAE-1 gamma and delta, and reciprocal mutations of each of the three RAE-1 forms, in binding studies to understand better the nature of this interaction. A manuscript describing the detail characterization of these binding interactions has been submitted for publication. Preliminary efforts to understand in detail the structural basis of the RAE-1/m152 interaction by X-ray crystallography have been encouraging: several conditions for crystallization of the complex have been identified. Additional biochemical and structural studies of the mCMV immunovasin m04, which can block NK and T cell recognition of cytomegalovirus-infected cells, are underway. In particular, m04 has been studied by multidimensional NMR, and a preliminary structure of its basic fold, unique in the structural database has been determined. Further refinement is underway.
