MHC Antigen Presentation

How MHC presentation works

Every nucleated cell in your body is constantly broadcasting a sample of its protein contents at the surface. Not the whole proteins — just snippets, eight to ten amino acids each, displayed in tens of thousands of molecular cradles called MHC class I. Walk down the body's hallway and every cell is wearing thousands of name tags advertising what it's currently translating. A patrolling CD8 T cell reads those tags one by one, looking for something that doesn't belong.

The pipeline starts in the cytosol. Proteins that fold incorrectly, get tagged for turnover, or simply finish their job are marked with ubiquitin chains and fed to the proteasome — a barrel-shaped protease complex that chops them into peptides averaging 8-10 amino acids with a hydrophobic C-terminus. Those peptides are then grabbed by TAP (transporter associated with antigen processing), an ATP-powered pump in the endoplasmic reticulum membrane that flips them from cytosol into ER lumen.

Inside the ER, newly synthesized MHC class I heavy chains pair with β2-microglobulin and wait in a peptide-loading complex with calreticulin, ERp57, and tapasin (which is tethered to TAP). Peptides arriving from TAP get screened: weak binders fall out, strong binders stabilize the MHC I structure, and the loaded molecule is released and shipped through the Golgi to the cell surface. Total wall-clock time from translation to surface display: 30 to 90 minutes.

Class II works in parallel but on a different substrate. Dendritic cells, macrophages, and B cells endocytose extracellular material into phagosomes that progressively acidify and merge with lysosomes loaded with cathepsins — proteases active at low pH. Engulfed proteins get hydrolyzed into longer peptides (13-25 amino acids). Meanwhile, MHC class II α/β heterodimers form in the ER but their groove is blocked by the invariant chain CLIP fragment to prevent them from grabbing the ER's class-I-bound peptides. The invariant chain also routes the complex to a late endosomal compartment called the MIIC, where CLIP is cleaved out by cathepsin S and the chaperone HLA-DM catalyzes loading of the high-affinity processed peptide. Then to the surface.

Worked clinical example: dendritic cell priming a CD8 response to influenza

A dendritic cell in the respiratory mucosa engulfs flu virions and infected epithelial debris. Within hours it's expressing thousands of MHC class I molecules per cell — typical surface density runs 10⁴ to 10⁵ — and a small fraction (perhaps 100-1000 per cell) carry peptides derived from viral nucleoprotein, matrix protein M1, or polymerase fragments. The viral peptides outcompete self peptides because they tend to have higher affinity for the local HLA allele, and because the cell is making a lot of them.

That dendritic cell migrates via afferent lymphatics to the draining lymph node — typically the mediastinal node for a respiratory infection. Inside the T-cell zone, naive CD8 T cells with random T-cell receptors scan the dendritic cell's surface, dwelling for seconds to minutes at each contact. The frequency of CD8 T cells specific for any given influenza epitope is tiny — perhaps one in 10⁵ to 10⁶ — but the lymph node concentrates them into a few cubic millimeters where the chance of encounter becomes meaningful. A productive contact extends the dwell time to hours and triggers activation.

Within 24-48 hours of that activation, the responsive CD8 T cell has proliferated through 8-10 division cycles, generating roughly 10³ daughter cells. Within a week the expansion can reach 10⁵-fold — a million-strong clone all bearing the same TCR, ready to inspect every cell in the lung for the matching peptide-MHC complex and kill those that display it. The entire chain depends on that initial display of viral peptide on the dendritic cell's class I.

Class I vs class II compared

	MHC Class I	MHC Class II
Expression	All nucleated cells	Dendritic, macrophage, B cell, thymic epithelium
Chains	Heavy chain + β2-microglobulin	α chain + β chain heterodimer
Peptide source	Cytosolic proteins (intracellular)	Endocytosed antigens (extracellular)
Processing	Proteasome → TAP → ER loading	Endosomal cathepsins → MIIC loading
Peptide length	8-10 amino acids (closed groove)	13-25 aa (open-ended, peptide can dangle)
Co-receptor	CD8 binds α3 domain	CD4 binds β2 domain
T cell function	Cytotoxic killing	Helper cytokines, B-cell licensing
HLA loci	HLA-A, HLA-B, HLA-C	HLA-DR, HLA-DQ, HLA-DP

The class I groove is closed at both ends, so peptides have to fit exactly — usually 9 amino acids — anchored by hydrophobic residues at positions 2 and 9. The class II groove is open at the ends, so longer peptides can dangle out and the same core sequence can be presented as a 14-mer or an 18-mer with different flanking residues. That structural difference is what makes class II permissive enough to handle the messier processing in lysosomes.

Molecular variants and special cases

• Non-classical class I. HLA-E, F, G have restricted polymorphism and specialized roles. HLA-E presents conserved class I leader peptides to NK cells via the CD94/NKG2A inhibitory receptor — a signal saying "I'm a healthy normal cell." HLA-G is expressed on trophoblast and protects fetal tissue from maternal NK attack.
• CD1. Structurally similar to class I but presents lipid antigens (mycobacterial mycolic acids, glycolipids) to invariant NKT cells and CD1-restricted T cells — important for tuberculosis immunity.
• MR1. Presents bacterial vitamin B metabolites to mucosal-associated invariant T cells (MAIT cells), which are abundant in liver and gut.
• Cross-presentation by cDC1. Conventional type 1 dendritic cells can route engulfed proteins onto class I, allowing CD8 priming against viruses that don't infect dendritic cells directly. Defects in cDC1 (BATF3 knockout in mice) cripple tumor immunity.
• Tapasin-independent alleles. Some HLA-B alleles (B*44:02) require tapasin for proper loading; others (B*44:05) load tapasin-independently. The distinction matters in transplant outcomes and in viral immune escape.
• Class II on activated T cells. Human CD4 and CD8 T cells upregulate HLA-DR upon activation, which can serve as a flow-cytometry marker but doesn't make them efficient antigen-presenters.

Disease relevance

• Transplantation. HLA mismatch drives acute and chronic rejection. Kidney grafts are matched at HLA-A, B, DR; six-antigen matches survive 5 years at ~85% vs ~75% for full mismatches. Hematopoietic stem cell transplant requires 8/8 or 10/10 high-resolution matching at HLA-A, B, C, DRB1 (and DQB1) to prevent graft-versus-host disease.
• Autoimmunity. HLA alleles are the strongest genetic risk factors for many autoimmune diseases. HLA-B*27 → ankylosing spondylitis (90% of patients carry it); HLA-DRB1*04:01 + shared epitope → rheumatoid arthritis; HLA-DQ2/DQ8 → celiac disease (binds gluten peptides); HLA-DR3/DR4 → type 1 diabetes. The mechanism: certain alleles bind self peptides with the right affinity to escape thymic negative selection but trigger peripheral autoreactivity.
• Viral immune evasion. HIV epitopes mutate in HLA-restricted hotspots, evading dominant CD8 responses; the most-protective HLA alleles (B*57, B*27) restrict slow-mutating epitopes. Herpesviruses encode multiple MHC-I inhibitors. SARS-CoV-2 reduces MHC I expression via ORF6 and ORF8.
• Cancer. Tumors that downregulate class I escape CD8 surveillance but become NK targets — partial downregulation is the worst of both worlds and is selected for during cancer evolution. Loss of heterozygosity at HLA is observed in many tumors.
• Drug hypersensitivity. Some severe drug reactions are HLA-restricted: HLA-B*57:01 + abacavir = abacavir hypersensitivity syndrome; HLA-B*15:02 + carbamazepine = Stevens-Johnson syndrome in Han Chinese; HLA-B*58:01 + allopurinol. Pre-prescription HLA typing prevents these.
• Pregnancy. Trophoblast cells lack classical class I (no HLA-A or -B) but express HLA-C, E, and G to avoid both maternal CD8 and NK attack. Recurrent miscarriage has been linked to HLA-C / KIR mismatch.

Common pitfalls and misconceptions

• "MHC and HLA are different things." Same locus. MHC is the generic term across species; HLA is the human-specific name (human leukocyte antigen). Mouse MHC is called H-2.
• "Class II only shows pathogen peptides." Class II presents whatever lives in the endosomal compartment, including a lot of self protein from membrane recycling. Tolerance comes from thymic deletion of self-reactive CD4 T cells, not from selective loading.
• "More MHC expression is better immunity." Interferon-γ massively upregulates both classes during infection, but baseline expression suffices for surveillance. Excess MHC on antigen-presenting cells doesn't fix poor T-cell repertoire.
• "CD8 only sees class I." Yes — but cross-presentation lets CD8 T cells see extracellular antigen via class I on dendritic cells. The CD8/class I co-receptor pairing is fixed; the antigen source can be flexible.
• "HLA matching guarantees no rejection." Even fully matched siblings can reject because of minor histocompatibility antigens (peptides differing between donor and recipient that are presented by identical HLA). Immunosuppression is still required for kidney, liver, and heart.
• "Tumors are inherently visible to T cells." They're only visible if they present tumor neoantigens on intact MHC I. Many tumors evolve to lose class I or specific HLA alleles, becoming invisible — which is one reason checkpoint inhibitors work in some patients and not others.

Therapeutic applications

• Pre-transplant HLA typing. Sequence-based typing at high resolution (4-digit or 6-digit) of donor and recipient is standard for kidney, marrow, and lung transplants. Anti-HLA antibodies in the recipient (panel-reactive antibody screen) identify "sensitized" patients who reject more readily.
• Pharmacogenomic HLA screening. Routine HLA-B*57:01 testing before abacavir prevents fatal hypersensitivity. HLA-B*15:02 testing before carbamazepine in Asian populations prevents Stevens-Johnson syndrome.
• Cancer immunotherapy. Checkpoint inhibitors (anti-PD-1, anti-CTLA-4) unleash existing CD8 responses that depend on MHC I-presented neoantigens. Tumors with high mutational burden present more neoantigens and respond better. Neoantigen-targeted vaccines (mRNA, peptide) are designed to fit predicted HLA binders for the individual patient.
• Adoptive T-cell therapy. TCR-engineered T cells target HLA-restricted tumor epitopes — limited to patients with matching HLA. NY-ESO-1 / HLA-A*02:01 is a classic example.
• Vaccine epitope design. Computational tools (NetMHCpan, MHCflurry) predict which peptides will bind which HLA alleles, allowing rational selection of vaccine epitopes that cover population HLA diversity.
• Pregnancy-related immune therapy. Recurrent miscarriage protocols sometimes include lymphocyte immunization or IVIG; the evidence is mixed but the HLA/KIR rationale is clear.