Classical MHC proteins come in two flavors: class I and class II. In some ways, it can seem that class I MHC molecules have lots of advantages over class II: class I molecules are expressed on almost all cells instead of just a subset of immune cells and are recognized by exciting and dangerous “killer” T cells, whereas class II molecules are targeted by seemingly more benign “helper” T cells. However, class II MHC molecules have their own story to tell, and the first crystal structure of a class II MHC protein (1), solved in 1993, was no exception.
In the decades before structural studies were applied to MHC proteins, two very well-studied and critically important immune responses were mapped to the genetic locus known as the MHC: 1) to the class II region, the degree of immune responsiveness to certain simple Ags, and 2) to the class I region, alloreactivity (the large immune reaction to MHC differences that leads to graft rejection). Both involved T cells and MHC molecules, but the role of MHC proteins in immunology was cemented by Zinkernagel and Doherty’s 1974 discovery of MHC-restricted T cell recognition of Ag (2), and especially by their altered-self model to explain MHC involvement in T cell responses. The altered-self model proposed that T cells recognize self MHC proteins that have been changed somehow by Ag. In other words, a TCR would recognize a complex of Ag plus an MHC molecule so that neither the Ag nor the MHC molecule alone would trigger a T cell response.
Initial progress toward defining what altered self meant in molecular terms mainly involved experiments with class II MHC molecules. In 1983, Pippa Marrack, John Kappler, Howard Gray, and colleagues established that degradation of chicken OVA into peptides was both necessary and sufficient to trigger a Th cell response to that Ag, and in 1985, Emil Unanue and colleagues demonstrated that short synthetic peptides bound to purified MHC class II proteins. After crystal structures of class I MHC proteins were solved in the late 1980s, the peptide-binding groove with an extended peptide nestled between two long α-helices provided an explanation for the dual recognition (i.e., Ag plus MHC) properties of TCRs. Perhaps the most surprising aspect of the first MHC structure was the finding of electron density for what turned out to be a mixture of endogenous (i.e., cell-derived) peptides; the protein had been purified by papain cleavage from a B cell line, which was not expected to be presenting Ag, so why would a peptide or peptides be occupying the so-called “foreign Ag” binding site?
By 1993, the immunological community had come to terms with the idea that it was up to the TCR to distinguish self from non-self because MHC molecules present self, as well as foreign, peptide Ags. The dichotomy between presentation of peptides from exogenous Ags (class II MHC) versus Ags synthesized by the cell (class I MHC) was known, and a mixture of cell-derived (self) peptides had been extracted from purified class II MHC molecules (3). So when Jerry Brown, Ted Jardetzky, Joan Gorga, Larry Stern, and Robert Urban, graduate students and postdoctoral fellows in Don Wiley’s and Jack Strominger’s laboratories at Harvard University, set out to solve a human class II MHC structure, they suspected that they would find a structure with a class I MHC–like peptide-binding groove occupied by cell-derived peptides. Indeed, Jerry Brown had used the class I MHC structure solved in 1987 to model the class II MHC peptide-binding groove, publishing a paper in 1988 (4) that was widely cited until he published the crystal structure in 1993.
That structure of a class II MHC molecule, human HLA-DR1, was actually solved using data from three different forms of crystallized protein: DR1 purified from B cells (and therefore containing cell-derived peptides), DR1 expressed in baculovirus-infected insect cells and reconstituted with a defined peptide, and DR1 bound to staphylococcal enterotoxin B, a Staphylococcus aureus superantigen. Taken together, the three DR1 structure papers (1, 5, 6) defined how class II MHC molecules recognize peptide Ags and revealed important differences from class I MHC peptide recognition.
First the similarities: what the class II MHC structure immediately confirmed was a close overall structural similarity to class I MHC proteins. The two membrane-distal domains of class II MHC (α1 and β1) were paired to form an eight-stranded β-pleated sheet topped by two α-helices, analogous to the α1 and α2 domains of the class I MHC H chain, and the two membrane-proximal class II domains (α2 and β2) were folded into Ig-like domains that were paired asymmetrically to more closely resemble the class I α3 and β2-microglobulin pairing than the pairing of Ab domains. These similarities were present despite the fact that the four extracellular domains of class II MHC molecules are arranged as two domains per polypeptide chain in a heterodimer, whereas the class I domains are distributed as three domains on an H chain and a single domain on a L chain. A common domain organization for class I and class II had actually been suggested prior to any MHC crystal structures based on similarities between class I and class II molecules (7). So, there were no real surprises yet.
The beauty of the class II structures, which was apparent even in the first reported structure containing a mixture of cell-derived peptides (1), was that class II MHC molecules bound peptides in a different way than did their class I MHC relatives. For example, peptides bound to class II MHC molecules could extend out of both sides of the peptide-binding groove. They were not constrained by the physical features of the class I peptide-binding groove that created barriers at both ends; instead, the ends were open, explaining why class II MHC proteins bound 14-mer and longer peptides, whereas class I MHC was limited to relative short (7–10 residue) peptides. The electron density for the mixture of cell-derived peptides in the first DR1 structure was remarkably clear: what was revealed was a long, extended polypeptide that was anchored at multiple places within the groove rather than at its ends (as found for class I MHC–bound peptides). This suggested that even longer peptides, which were known to bind to class II MHC, could be accommodated by protruding out of one or both ends of the groove, explaining why peptides eluted from a particular class II MHC molecule did not exhibit common, or anchor, residues when aligned by their N termini; although anchor residues could be found when aligning peptides, the N termini would not be in register as would be found for peptides eluted from a class I MHC protein. Later analyses of DR1 bound to a defined 13-mer peptide from influenza hemagglutinin (5) would reveal a common theme for class II MHC interactions with bound peptides: conserved residues within the class II peptide-binding groove made hydrogen bonds with main chain atoms within the peptide, forcing bound peptides into a twisted conformation called a polyproline II helix. Importantly, note that unlike the main chain atoms in an α-helix, main chain atoms within a polyproline II helix (which is neither really a helix nor composed exclusively of prolines) are not occupied with internal hydrogen bonding and are therefore free for external interactions. These interactions with peptide main chain atoms explained the ability of each individual class II allele to bind many different peptides, whereas allele specificity for peptides was explained by side chains at anchor positions of the peptides reaching into pockets within the class II groove.
The fundamentally different ways that class I and class II MHC proteins associate with peptides were later rationalized by different pathways for acquiring and associating with peptides, but I doubt anyone would have anticipated these differences without the crystal structures that allowed direct visualization of class II MHC/peptide interactions.
The author has no financial conflicts of interest.