7.2 Antigens and Epitopes

An antigen is whatever the immune system, or your purchased antibody, can recognize. An epitope (antigenic determinant) is the actual patch of that molecule the paratope touches—often a stretch of amino acids, sometimes a sugar, sometimes a folded surface made of residues that are far apart in the sequence. Staining fails when the epitope is gone, hidden, or never where you thought the biology lived.

This OpenExamPrep section is independent teaching on antigens and epitopes covering published QIHC staining topic areas. It does not claim ASCP or manufacturer approval. Epitope retrieval chemistry is Chapter 6. Here the question is what the antibody is looking for, what formalin does to that site, and where a true-positive stain is allowed to sit.

Linear versus conformational epitopes

A linear (continuous) epitope is a short stretch of primary sequence. Once the protein is unfolded, the same amino acids can still be recognized. Many FFPE-successful monoclonals were selected or screened on denatured protein or on peptide immunogens for this reason. Heat retrieval that unfolds formalin-fixed protein can help a linear epitope by unmasking the peptide.

A conformational (discontinuous) epitope is a three-dimensional patch. Residues that sit together in the native fold may be fifty amino acids apart in the gene. If formalin, heat, or reduction destroys that fold, the antibody has nothing to bind even though the protein is still in the cell. Retrieval cannot rebuild a conformation that fixation permanently wrecked.

Worked contrast. Two antibodies claim the same cytokeratin. Clone A was raised to a denatured peptide and works on routine 10% NBF after high-pH HIER (linear epitope, masked then unmasked). Clone B was raised to native filament protein and is brilliant on acetone-fixed frozen sections and nearly negative on FFPE (conformational epitope). They are not interchangeable, and a failed FFPE run of clone B is not proof the laboratory cannot do cytokeratin.

Think of the immunogen as what the animal saw and the epitope as what the bottle still sees. A whole native protein can raise both linear and conformational antibodies. A short peptide immunogen is biased toward linear epitopes. Screening on FFPE during clone selection is how vendors enrich for reagents that survive methylene bridges. An old research antibody that was never screened on paraffin is not a retrieval problem; it is an epitope-type problem.

Formalin destroys or hides conformational epitopes

Routine IHC uses 10% neutral buffered formalin (about 4% formaldehyde). Formaldehyde builds methylene bridges (Chapter 6). Those bridges do three different jobs that students mix up:

1. Mask epitopes that are still chemically present (the usual HIER story).
2. Chemically modify side chains (lysine-rich patches) so a paratope that needed a free amino group no longer fits.
3. Lock or collapse conformational epitopes so the native surface never exists again on that block.

The third point is why some antibodies are sold as frozen-section only or unfixed-cytology only. You cannot retrieve a conformation that fixation destroyed. Switching to a longer HIER clock on that clone produces holes and background, not a miracle native fold.

Under-fixation is the opposite trap: antigen diffuses, membranes leak, and a cytoplasmic protein appears in the wrong compartment or washes out. Over-fixation (days in NBF, delayed penetration of a large fatty specimen) deepens the mask and can destroy labile modifications. Breast predictive markers use published ASCO/CAP cold-ischemia and fixation windows (cold ischemia of 1 hour or less; formalin 6 to 72 hours) because both under- and over-fixation change the epitope the scorer will read. Those numbers are breast-biomarker guidance, not a claim that every antigen in the laboratory shares the same clock.

Acetone, methanol, and alcohol-based cytology fixatives preserve some conformational epitopes that formalin wrecks, and they can destroy others (especially lipid-associated or alcohol-labile membrane sites). A clone that is "negative on the cell block but positive on the air-dried smear" is often an epitope-plus-fixative story, not a staining-machine story. Validate the fixative as part of the antigen, not as decoration on the requisition.

Glycosylation

Many membrane and secreted proteins are glycoproteins. Antibodies may see:

• A peptide epitope that happens to sit near a glycan. Deglycosylation or a change in sugar processing (tumor versus normal, necrosis) can hide or expose that peptide.
• A carbohydrate epitope itself (blood-group-related chains, some CD markers, some microbial capsules). Formalin is not the main enemy; periodate, harsh oxidation, or an enzyme that strips sugars will erase the target. Conversely, some peptide clones become stronger after deglycosylation because the sugar was sterically blocking the peptide.

Practical staining consequences:

• Do not assume every membrane marker is a peptide story. If a clone insert mentions a glycan or a blood-group-related antigen, treat oxidation and enzyme retrieval as epitope threats.
• Heavy glycosylation can change how densely a membrane stains. A thick glycocalyx can also hinder bulky detection reagents (a steric issue, not "no antigen").
• Plasma-membrane staining that disappears after a long proteinase K step may be shaved glycoprotein, not a true-negative tumor.
• Mucin-rich tumors and goblet-cell cytoplasm can hide or mimic carbohydrate epitopes; expected-pattern reading still applies.

Enzyme-induced epitope retrieval (Chapter 6) nicks protein that masks some epitopes and can also cleave a peptide epitope or release a glycan-rich ectodomain. That is why EIER time is a dose, not a proof that more digestion is more antigen.

Expected localization is part of the antigen definition

A true-positive IHC result has a legal address. The same brown DAB in the wrong compartment is not a weaker positive; it is a different interpretation or an artifact.

Nuclear. Expect the counterstain to be partly obscured by chromogen in positive nuclei. A nuclear marker that is only cytoplasmic on the control is the wrong clone, the wrong retrieval, a degraded phospho-nuclear target, or detection dirt.

Cytoplasmic. Filamentous cytokeratin is not the same as granular mitochondrial biotin. Name the texture. Diffuse enzyme stains are not Golgi dots.

Membranous. Complete, intense circumferential membrane staining is scoring language for some predictive assays (HER2-class reading appears in later staining-pattern sections). For method immunology, remember that a membrane antigen that looks cytoplasmic may have been internalized, poorly fixed, or over-retrieved. Apical-only versus circumferential membrane is also an expected-pattern fact for some antigens, not a weak stain.

Golgi. A compact paranuclear dot is a recognized true pattern for some antigens. Do not correct it to diffuse cytoplasm because a textbook drawing showed a filled cell.

Internal controls use the same map: residual normal epithelium should label where that antigen lives in normal tissue. If internal keratin is nuclear, the assay is not to be interpreted. Combined nuclear-plus-cytoplasmic labeling is allowed only when the biology of that antigen includes shuttling or when the validated pattern says so—not because the polymer is generous.

Phosphopeptides are labile antigens

A phospho-specific antibody recognizes a peptide only when a defined serine, threonine, or tyrosine is phosphorylated. The epitope is the sequence plus the phosphate. Lose the phosphate and the antibody should go negative even though the protein is still there.

Why phosphopeptides fail on real slides:

• Cold ischemia. Tissue phosphatases remain active after the specimen leaves the patient. Long unfixed time erases phospho-epitopes. This is a specimen-handling problem (Chapter 4) that shows up as a staining-pattern lie.
• Fixation delay or weak formalin penetration. The center of a large specimen can dephosphorylate while the rim fixes. A rim-positive, center-negative phospho stain is a fixation map, not two clones.
• Over-fixation and harsh retrieval. Some phospho-epitopes are also just fragile peptides; aggressive HIER can destroy them.
• The antibody was never fully phospho-specific. Some phospho reagents cross-react with the non-phospho peptide. A phosphatase-treated control section, where the laboratory has validated that control, should drop true phospho signal.

Do not treat a negative phospho-STAT or phospho-ERK as proof the pathway is off until you have a fixation history and a control that was handled the same way. Do not rescue a lost phosphate with more HIER as if it were a masked linear keratin epitope. Do not call a dirty cytoplasmic blush a phospho-nuclear result.

In practice

A conformational antibody that the insert limits to frozen tissue is pale on FFPE after every retrieval you know. Believe the epitope type. Order a linear-epitope clone for paraffin or cut a frozen section if the validation exists.

A membrane glycoprotein is gone after 15 minutes of proteinase K and intact after citrate HIER. That is enzyme shaving of a glycan-rich surface, not proof the block lacks antigen.

A phospho-marker is strong on a promptly fixed needle core and negative in the center of a long-ischemia resection. The antigen was labile. Retrieval will not put the phosphate back.

Expected-pattern reading is how you tell a real epitope from a dirty method: nucleus, cytoplasm, membrane, or Golgi—then texture—then the negative reagent control.