MHC and Antigen Presentation

The major histocompatibility complex is a cluster of genes — called HLA (human leukocyte antigen) in humans — encoding cell-surface glycoproteins whose job is to bind short peptides and display them for inspection by T-cell receptors (TCRs). This process is called antigen presentation, and it converts intracellular or extracellular protein content into a surface-readable code. Critically, T cells do not recognize free-floating antigen the way antibodies do (see Antigens and Epitopes); a T cell's TCR only binds a composite surface formed by the peptide and the MHC molecule together. This dual requirement is called MHC restriction. There are two structurally and functionally distinct classes. MHC class I molecules are expressed on essentially all nucleated cells and present peptides derived from proteins synthesized inside that very cell — a running report of the cytosolic protein pool, useful for detecting viral infection or malignant transformation from within. MHC class II molecules are expressed mainly on professional antigen-presenting cells (APCs) — dendritic cells, macrophages, and B cells — and present peptides derived from material taken up from outside the cell, such as engulfed bacteria or extracellular debris. Class I peptides are typically 8–10 residues long and bind in a closed-ended groove; class II peptides are typically 13–25 residues and bind in an open-ended groove that allows the ends to hang out. The pairing of MHC class with T-cell type is not incidental but mechanistically enforced: cytotoxic CD8+ T cells recognize peptide–MHC I complexes because the CD8 coreceptor binds a conserved region on the MHC I heavy chain (the α3 domain), while helper CD4+ T cells recognize peptide–MHC II because CD4 binds a conserved region on the MHC II β2 domain. This coreceptor logic, refined during T-Cell Development and Selection in the thymus, guarantees that CD8 cells become killers targeting infected cells directly and CD4 cells become coordinators activating other immune cells — a division of labor elaborated further under Helper and Cytotoxic T Cells.

MHC molecular architecture. MHC class I is a heterodimer: a polymorphic heavy chain (α1, α2, α3 domains) non-covalently associated with a small invariant protein, β2-microglobulin. The peptide-binding groove is formed by the α1 and α2 domains folded into a floor of β-pleated sheet topped by two α-helices, with the peptide's N- and C-termini anchored by conserved pockets at each end of the groove — hence the fixed 8–10 residue window. MHC class II is also a heterodimer, but here both chains (α and β) are polymorphic and each contributes one domain (α1, β1) to the peptide-binding groove; because the groove is open at both ends, longer peptides can protrude past the binding site.

The class I (endogenous/cytosolic) pathway. 1. Cytosolic proteins — a cell's own proteins, viral proteins made by an infected cell, or tumor proteins — are tagged with ubiquitin and degraded by the proteasome, a large multi-catalytic protease complex, into peptide fragments. 2. These peptides are transported from the cytosol into the endoplasmic reticulum (ER) lumen by TAP (Transporter associated with Antigen Processing), an ATP-dependent heterodimeric transporter (TAP1/TAP2) embedded in the ER membrane. 3. Inside the ER, newly synthesized MHC class I heavy chains are stabilized by the chaperone calnexin, then associate with β2-microglobulin and join the peptide-loading complex, which includes tapasin (bridging MHC I to TAP), calreticulin, and ERp57 (a thiol oxidoreductase). 4. Peptides delivered by TAP are trimmed to final length by ER aminopeptidases (ERAAP/ERAP1) and loaded into the class I groove; a properly loaded, stable MHC I–peptide complex is released from the loading complex. 5. The stable peptide–MHC I complex traffics through the Golgi to the plasma membrane, where it is surveyed by circulating CD8+ T cells. If no peptide loads (e.g., in TAP-deficient cells), MHC I is unstable and largely fails to reach the surface — a fact exploited experimentally and by viruses that inhibit TAP as an immune-evasion strategy.

The class II (exogenous/endosomal) pathway. 1. Extracellular material — bacteria, foreign proteins, cell debris — is internalized by professional APCs via phagocytosis, macropinocytosis, or receptor-mediated endocytosis, and delivered to progressively acidifying endosomal/lysosomal compartments where proteases (cathepsins) degrade it into peptides. 2. Meanwhile, in the ER, newly made MHC class II α/β dimers associate with a chaperone called the invariant chain (Ii, CD74), which occupies the peptide groove (as a fragment called CLIP) to prevent premature binding of ER-resident peptides — this would otherwise let class II poach peptides meant for class I. 3. The MHC II–invariant chain complex traffics to a specialized acidic endosomal compartment, the MIIC (MHC class II compartment), where proteases degrade the invariant chain, leaving only the short CLIP fragment in the groove. 4. HLA-DM (a non-classical, non-peptide-presenting MHC II-like molecule) catalyzes exchange: it destabilizes CLIP and facilitates loading of the antigenic peptides generated in the endosome, then edits toward higher-stability peptide–MHC combinations. 5. The stable peptide–MHC II complex is delivered to the cell surface for inspection by CD4+ T cells.

Cross-presentation — the exception that proves the rule. Some dendritic cells can route exogenous antigens (e.g., from dying infected cells or tumor cells) into the class I pathway, a phenomenon called cross-presentation, allowing CD8+ T-cell priming against antigens the dendritic cell never synthesized itself. This is essential for initiating cytotoxic responses against viruses that don't infect dendritic cells directly, and is a key design principle in cancer vaccine platforms.

The single most important experiment establishing how T cells "see" antigen is Rolf Zinkernagel and Peter Doherty's 1974 study of virus-specific cytotoxic T lymphocytes (CTLs) in mice, for which they received the 1996 Nobel Prize in Physiology or Medicine.

Setup. Zinkernagel and Doherty infected mice of a given MHC genotype (e.g., H-2^k, using inbred mouse strains that differ only at the MHC locus, called H-2 in mice) with lymphocytic choriomeningitis virus (LCMV). After the mice mounted a cytotoxic T-cell response, they harvested CTLs and tested them in vitro against target cells infected with the same virus — but the target cells came from mice of varying MHC genotypes, some matching the CTL donor's MHC and some not.

Observation. CTLs efficiently killed virus-infected target cells only when the target cells shared the same MHC haplotype as the mouse from which the CTLs originated. If the target cells were infected with the identical virus but expressed a different MHC haplotype, the CTLs failed to kill them — even though the virus (and its antigens) were identical. Killing required both viral antigen and self-MHC to be present together on the same target cell.

Interpretation. This showed T cells do not recognize antigen alone; they recognize antigen in the context of self-MHC. This was named MHC restriction, and it forced a conceptual break from the antibody paradigm (in which antibodies bind free antigen directly, independent of any presenting molecule — see Antigens and Epitopes). It explained why T-cell immunity is intrinsically tied to genetic identity, and it set up the subsequent recognition that CD8 and CD4 coreceptors specify restriction to class I versus class II respectively (established through later work identifying coreceptor-MHC binding, notably using coreceptor-blocking antibodies and gene-knockout mice in the 1980s–90s). A complementary line of evidence came from structural biology: Pamela Bjorkman, Don Wiley, and colleagues solved the first X-ray crystal structure of human MHC class I (HLA-A2) in 1987. The structure revealed the α1/α2 platform-and-groove architecture and, strikingly, showed extra electron density inside the groove in every crystal examined — density that did not belong to the MHC protein itself. This was the direct physical proof that MHC class I molecules are, in their native state, always occupied by a bound peptide, confirming biochemically what Zinkernagel and Doherty had inferred functionally. Later co-crystal structures of the TCR bound simultaneously to peptide and MHC (Garboczi, Garcia et al., mid-1990s) visualized the ternary complex directly, closing the loop between genetics, cell biology, and structure. A third strand of evidence came from genetics of TAP-deficiency and invariant-chain knockouts: mice lacking TAP1/TAP2 show a dramatic reduction in surface MHC class I (because unloaded MHC I is unstable and retained/degraded in the ER), while mice lacking the invariant chain show impaired MHC class II peptide loading and CD4+ T-cell development — confirming that each chaperone/transporter is non-redundantly required for its respective pathway, exactly as the biochemical model predicts.

Example 1 — Reasoning through MHC restriction in a mixed-infection experiment. Suppose you have two inbred mouse strains: strain A (H-2^b) and strain B (H-2^d). You infect strain A mice with influenza virus, harvest their CD8+ CTLs, and test killing against four target-cell populations:

Only target 1 satisfies both requirements simultaneously — viral peptide and matching self-MHC — reproducing the Zinkernagel–Doherty logic. If you instead used CTLs from an (A×B)F1 mouse (H-2^b/d heterozygote), which co-expresses both haplotypes, those CTLs would kill both target 1 and a version of target 3 infected with flu, because the F1 CTL repertoire was educated on both MHC haplotypes during thymic selection — foreshadowing the logic of T-Cell Development and Selection.

Example 2 — HLA polymorphism and population peptide coverage. The human MHC (HLA) locus is the most polymorphic region of the genome: HLA-B alone has several thousand known alleles. Consider a simplified calculation of why this matters. Suppose a given viral peptide binds well to HLA-A02:01 (carried by roughly 40–50% of some populations) but binds poorly to HLA-A01:01. An individual homozygous for HLA-A*01:01 might fail to present this particular peptide on MHC I at all — creating a "hole" in that person's antigen coverage for this pathogen, even though their immune system is otherwise fully competent. Now scale this to a population: if each person expresses up to 6 different classical MHC class I alleles (2 each of HLA-A, -B, -C, assuming heterozygosity) and up to 6–8 classical class II alleles, and there are thousands of alleles segregating in the population, then across a population of unrelated individuals, a huge diversity of peptide-binding specificities exists. This is precisely why a pathogen that mutates to escape presentation by one common HLA allele may still be presented efficiently in hosts carrying a different allele — MHC polymorphism is a population-level bet-hedging strategy against pathogen evolution, distinct from any single individual's advantage. It's also the direct reason transplant matching is difficult: the more polymorphic the locus, the lower the a priori chance that an unrelated donor and recipient share all relevant HLA alleles, which is why sibling-matched or registry-based searches (matching HLA-A, -B, -C, -DR, -DQ) are central to transplant medicine.

Rolf Zinkernagel (Swiss immunologist, University of Zurich) and Peter Doherty (Australian immunologist/veterinary pathologist) worked together at the John Curtin School of Medical Research in Canberra in the early 1970s, studying how mice cleared LCMV infection. Neither set out to discover a new immunological principle — they were trying to understand why some inbred mouse strains cleared LCMV while others developed fatal choriomeningitis, a question rooted in classical virology and mouse genetics. Their systematic testing of CTL killing across mismatched MHC combinations turned an incidental technical observation (killing sometimes "just didn't work" with certain target cells) into one of immunology's foundational laws. Doherty has written that the finding was almost accidental — a control experiment that "shouldn't" have failed, and did.

Transplant medicine. HLA typing (matching HLA-A, -B, -C, -DR, -DQ between donor and recipient) is the single strongest predictor of organ and bone-marrow graft survival apart from ABO compatibility; mismatched MHC molecules are themselves recognized as foreign, driving allograft rejection and graft-versus-host disease.

Autoimmune disease risk. Specific HLA alleles are strongly associated with autoimmune conditions — HLA-B27 with ankylosing spondylitis, HLA-DQ2/DQ8 with celiac disease, certain HLA-DR alleles with type 1 diabetes and rheumatoid arthritis — reflecting how particular peptide-binding grooves shape which self- or dietary peptides get presented and potentially trigger autoreactive T cells.

Vaccine design. Effective vaccines must generate peptides that are efficiently processed and loaded onto common MHC alleles in the target population; epitope prediction algorithms model proteasomal cleavage, TAP transport efficiency, and MHC-binding affinity to select vaccine antigens likely to be well presented across diverse HLA genotypes.

Cancer immunotherapy. Tumor cells that accumulate somatic mutations generate "neoantigens" — peptides absent from normal self-protein but presented on MHC I because they arise from mutant cytosolic proteins. Personalized cancer vaccines and some checkpoint-inhibitor therapies work by boosting cytotoxic T-cell responses against these neoantigen–MHC I complexes; tumors that downregulate MHC I or TAP are a recognized mechanism of immune evasion and a target of ongoing therapeutic strategies (e.g., combining checkpoint blockade with agents that restore antigen-processing machinery expression).

Infectious disease and pathogen evolution. Viruses such as HIV, herpesviruses, and adenoviruses encode proteins that specifically block TAP, degrade MHC I, or interfere with the peptide-loading complex, directly targeting this pathway to escape cytotoxic T-cell surveillance — a direct, ongoing evolutionary arms race built on the mechanism described above.

Ask the tutor: How does thymic selection use these same peptide–MHC complexes to decide which T cells survive and become CD4 versus CD8 cells?