Antibodies and T Cell Recognition

Introduction

The adaptive immune system achieves its extraordinary specificity through two classes of antigen receptors: antibodies produced by B cells and T cell receptors (TCRs) on T cells. Antibodies can recognize virtually any molecular surface — proteins, carbohydrates, lipids, even small synthetic molecules — and do so with affinities that rival the tightest enzyme-substrate interactions. T cell receptors, by contrast, recognize a composite surface: a short peptide fragment displayed in the groove of a major histocompatibility complex (MHC) molecule on a cell surface. Together, these two recognition systems provide complementary surveillance: antibodies patrol the extracellular space, neutralizing toxins and tagging pathogens for destruction, while T cells inspect the intracellular contents of every nucleated cell in the body.

This lesson explores the molecular architecture of antibodies and the genetic mechanisms that generate their vast diversity, the germinal center reaction that refines antibody quality, and the therapeutic exploitation of monoclonal antibodies. We then turn to MHC molecules and T cell recognition — how peptide fragments are displayed on cell surfaces, why MHC genes are the most polymorphic in the genome, and how different T cell subsets coordinate the immune response. Finally, we survey the computational tools that have become indispensable for antibody engineering, HLA typing, epitope prediction, and immune repertoire analysis.

24.3 — B Cells and Antibodies

Antibody Structure: Heavy Chains and Light Chains

An antibody (immunoglobulin) molecule is a Y-shaped glycoprotein composed of four polypeptide chains: two identical heavy chains (H) and two identical light chains (L), linked by disulfide bonds. Each chain consists of repeating immunoglobulin domains of roughly 110 amino acids, folded into a characteristic β-sandwich structure called the immunoglobulin fold. The heavy chain contains one variable domain (V_H) and three or four constant domains (C_H1, C_H2, C_H3, and sometimes C_H4), depending on the antibody class. The light chain contains one variable domain (V_L) and one constant domain (C_L). Light chains come in two types — κ and λ — which are functionally equivalent.

Two Identical Antigen-Binding Sites

Because each antibody molecule contains two identical heavy-light chain pairs, it possesses two identical antigen-binding sites — one at the tip of each arm of the Y. Each binding site is formed by the juxtaposition of the V_H and V_L domains, specifically by six hypervariable loops called complementarity-determining regions (CDRs): CDR1, CDR2, and CDR3 of each chain. The CDR3 loop of the heavy chain is the most variable and typically makes the largest contribution to antigen binding specificity. The intervening, less-variable segments that form the structural scaffolding of the variable domain are called framework regions (FRs).

The bivalent nature of antibodies enables them to cross-link antigens, forming large aggregates that are efficiently cleared by phagocytes. IgM, which assembles as a pentamer, has ten binding sites and is especially effective at agglutination during early immune responses.

Antibody Gene Assembly from Separate Gene Segments

The genome does not contain a separate gene for every possible antibody. Instead, antibody genes are assembled by V(D)J recombination during B cell development in the bone marrow. The immunoglobulin heavy-chain locus contains arrays of V (variable), D (diversity), and J (joining) gene segments, while light-chain loci contain V and J segments only.

The recombination process proceeds in defined steps:

1. A D segment joins to a J segment (heavy chain only)
2. A V segment joins to the DJ (or VJ for light chains)
3. The recombined VDJ exon is then transcribed and spliced to a constant-region exon

Recombination is catalyzed by the RAG1/RAG2 recombinase, which recognizes conserved recombination signal sequences (RSSs) flanking each gene segment. The process introduces additional diversity through imprecise joining: N-nucleotides are added by the enzyme terminal deoxynucleotidyl transferase (TdT), and P-nucleotides are generated by asymmetric hairpin opening. These junctional modifications ensure that even B cells using the same V, D, and J segments will produce different CDR3 sequences.

The combinatorial and junctional diversity of V(D)J recombination, combined with the pairing of different heavy and light chains, generates an estimated 10¹¹ or more distinct antibody specificities — enough to recognize virtually any antigen.

Somatic Hypermutation and Class Switching

After a B cell encounters its antigen and becomes activated, two further diversification mechanisms come into play:

Somatic hypermutation (SHM) introduces point mutations into the rearranged V-region genes at a rate roughly one million times higher than normal background mutation. The enzyme responsible is activation-induced cytidine deaminase (AID), which converts cytosine to uracil in the V-region DNA. Error-prone repair of these lesions by base-excision and mismatch-repair pathways generates the full spectrum of mutations. SHM predominantly targets the CDR loops, though mutations also occur in framework regions. B cells whose mutations improve antigen binding are positively selected; those with reduced affinity die by apoptosis. This iterative process of mutation and selection is called affinity maturation.

Class switch recombination (CSR) changes the antibody’s constant region without altering its antigen specificity. AID also initiates CSR by targeting switch regions upstream of each C_H gene. Recombination between switch regions deletes the intervening DNA, joining the rearranged VDJ to a downstream constant region. This allows a B cell initially producing IgM to switch to IgG, IgA, or IgE, each of which has distinct effector functions:

B Cell Activation and the Germinal Center Reaction

When a naive B cell encounters antigen that binds its surface immunoglobulin (the B cell receptor), it becomes activated — typically with the help of a T follicular helper (T_FH) cell that provides co-stimulatory signals (CD40 ligand) and cytokines. The activated B cell migrates to a germinal center (GC) within a lymph node or spleen follicle.

The germinal center is divided into two zones:

• The dark zone, where B cells (called centroblasts) proliferate rapidly and undergo somatic hypermutation
• The light zone, where B cells (called centrocytes) test their mutated receptors against antigen displayed on follicular dendritic cells (FDCs). Centrocytes compete for limiting amounts of antigen; those with higher-affinity receptors capture more antigen, process it, and present peptides on MHC class II to T_FH cells. Only B cells that receive survival signals from T_FH cells are selected — the rest undergo apoptosis.

B cells cycle between the dark and light zones multiple times, accumulating beneficial mutations over days to weeks. The output of the germinal center reaction is twofold: long-lived plasma cells that secrete high-affinity, class-switched antibodies and memory B cells that provide rapid responses upon re-exposure to the same antigen.

Monoclonal Antibodies and Their Therapeutic Uses

A normal immune response produces a polyclonal mixture of antibodies, each secreted by a different B cell clone. In 1975, Georges Köhler and César Milstein developed a technique to produce monoclonal antibodies — identical antibodies from a single B cell clone. They fused an antibody-producing B cell with an immortal myeloma cell to create a hybridoma, which could be propagated indefinitely and would secrete a single antibody species.

Monoclonal antibodies have become one of the most important classes of therapeutic drugs:

Early monoclonal antibodies were derived from mouse hybridomas and were immunogenic in human patients. Modern approaches use chimerization (mouse V regions grafted onto human C regions), humanization (only CDR loops from the mouse are grafted onto a human framework), and fully human antibodies generated from transgenic mice carrying human immunoglobulin loci or from phage and yeast display libraries.

Antibody Bioinformatics

The computational analysis and engineering of antibodies has become a thriving field:

Antibody sequence analysis and germline assignment. Tools such as IgBLAST and ANARCI align antibody sequences against germline V, D, and J databases (from IMGT, the international ImMunoGeneTics information system) to identify which gene segments were used, delineate CDR and framework boundaries, and quantify somatic hypermutation. ANARCI applies standardized numbering schemes (IMGT, Kabat, Chothia) that are essential for comparing antibodies across studies.

Antibody structure prediction. Predicting the three-dimensional structure of an antibody from its sequence is critical for understanding antigen recognition. ABodyBuilder and IgFold use deep learning to model antibody structures, with particular emphasis on the challenging CDR-H3 loop — the most diverse and structurally variable region. Accurate CDR loop modeling enables paratope prediction (identifying which residues contact the antigen).

Antibody humanization design. Computational tools guide the humanization of non-human antibodies by identifying the minimal set of CDR residues that must be retained for binding while replacing the rest of the variable region with human framework sequences. This reduces immunogenicity while preserving function.

Therapeutic antibody databases. Thera-SAbDab (Therapeutic Structural Antibody Database) catalogs the sequences and structures of all WHO-recognized therapeutic antibodies. abYsis provides tools for analyzing antibody sequences, comparing them to known antibodies, and identifying unusual features.

Antibody-antigen docking. Computational docking tools such as ClusPro and HADDOCK predict how an antibody binds to its antigen. These predictions are valuable for understanding binding mechanisms, designing improved variants, and identifying potential epitopes.

Phage display and yeast display data analysis. High-throughput display technologies generate massive libraries of antibody variants and select binders through iterative panning. Sequencing the selected clones across rounds of selection produces rich datasets that require bioinformatics pipelines to track enrichment, identify consensus sequences, and cluster families of related binders.

Let’s examine how the CDR3 region — the most variable part of antibodies — contributes to repertoire diversity by analyzing CDR3 length distributions and clonotype frequencies.

let cdr3_lengths = '[12, 14, 11, 15, 13, 10, 16, 12, 14, 13, 11, 15, 12, 14, 10]'
print("CDR3 length distribution:")
print(Stats.describe(cdr3_lengths))
let cdr3_diversity = '[8, 7, 6, 5, 4, 4, 3, 3, 2, 2, 2, 1, 1, 1, 1]'
print("CDR3 clonotype frequency distribution:")
print("Shannon diversity: " + Stats.shannon(cdr3_diversity))
print("Simpson dominance: " + Stats.simpson(cdr3_diversity))

A high Shannon diversity index indicates a broad repertoire with many distinct clonotypes, while a low Simpson dominance index confirms that no single clone dominates. These metrics are essential for assessing immune repertoire health and detecting clonal expansions in disease.

Exercise: Measure CDR3 Repertoire Diversity

The CDR3 region is the most variable part of antibodies and TCRs, generated by V(D)J recombination and junctional diversity. Compare the CDR3 diversity of a healthy individual with a patient who has a clonal B cell expansion:

let healthy = '[8, 7, 6, 5, 4, 4, 3, 3, 2, 2, 2, 1, 1, 1, 1]'
let clonal = '[65, 5, 4, 3, 3, 2, 2, 2, 1, 1, 1, 1]'
print("Healthy repertoire:")
print("Shannon: " + Stats.shannon(healthy))
print("Simpson: " + Stats.simpson(healthy))
print("Clonal expansion:")
print("Shannon: " + Stats.shannon(clonal))
print("Simpson: " + Stats.simpson(clonal))
// Which repertoire is more diverse?
let answer = "healthy"
print(answer)

Diversity indices are widely used in clinical immunology to monitor B cell lymphomas, assess immune reconstitution after transplantation, and evaluate vaccine responses.

24.4 — T Cells and MHC Proteins

MHC Class I and Class II Present Peptide Fragments on Cell Surfaces

While antibodies recognize intact molecular surfaces, T cells recognize short peptide fragments displayed in the groove of MHC molecules (called HLA — human leukocyte antigens — in humans). This system allows T cells to survey the internal protein contents of cells, detecting viral infection, intracellular bacteria, or aberrant self-proteins (such as those produced by cancer cells).

MHC class I molecules are expressed on virtually all nucleated cells. They present peptides derived from cytoplasmic proteins — proteins that have been degraded by the proteasome in the cytosol. The resulting peptide fragments (typically 8–10 amino acids long) are transported into the endoplasmic reticulum by the TAP transporter, loaded onto newly synthesized MHC class I molecules, and carried to the cell surface. This pathway provides a continuous window into the cell’s intracellular protein landscape. If the cell is infected by a virus, viral peptides will be displayed on MHC class I, flagging the cell for destruction.

MHC class II molecules have a more restricted expression pattern: they are found primarily on professional antigen-presenting cells (APCs) — dendritic cells, macrophages, and B cells. MHC class II presents peptides derived from extracellular proteins that have been endocytosed and degraded in lysosomes. The peptides are typically 13–25 amino acids long. This pathway reports on what is happening outside the cell — extracellular pathogens, toxins, and debris.

MHC Genes Are Highly Polymorphic

The MHC locus (chromosome 6 in humans) is the most polymorphic region in the human genome. The classical HLA genes — HLA-A, HLA-B, and HLA-C (class I) and HLA-DR, HLA-DQ, and HLA-DP (class II) — each have thousands of known allelic variants. Most polymorphisms map to the peptide-binding groove, meaning that different HLA alleles present different sets of peptides.

This polymorphism is maintained by balancing selection: pathogens that evolve to escape presentation by common HLA alleles still face presentation by rare alleles, so rare alleles enjoy a selective advantage and are maintained in the population. The result is that virtually every individual in the population has a unique combination of HLA alleles (the HLA haplotype), and at the population level there will always be some individuals capable of mounting an immune response against any given pathogen.

HLA polymorphism has profound medical consequences:

• Transplant rejection: T cells recognize foreign HLA molecules on transplanted tissue, triggering rejection. HLA matching between donor and recipient is critical for organ and bone marrow transplantation.
• Disease associations: certain HLA alleles are strongly associated with autoimmune diseases (e.g., HLA-B27 with ankylosing spondylitis, HLA-DRB1 with rheumatoid arthritis).
• Vaccine design: effective vaccines must generate epitopes presentable by diverse HLA alleles across the population.

Cytotoxic T Cells (CD8⁺) Kill Infected Cells

CD8⁺ cytotoxic T lymphocytes (CTLs) are the immune system’s assassins. When a CTL’s TCR recognizes a foreign peptide displayed on MHC class I on a target cell, the CTL delivers a lethal hit. The killing mechanism involves two pathways:

1. Perforin/granzyme pathway: the CTL releases perforin (which forms pores in the target cell membrane) and granzymes (serine proteases that enter through the pores and activate caspases, triggering apoptosis)
2. Fas/FasL pathway: the CTL expresses Fas ligand, which binds Fas on the target cell and activates the extrinsic apoptotic pathway

CTLs kill with remarkable precision — a single CTL can destroy many target cells sequentially, and the directional release of granules minimizes damage to neighboring healthy cells. CTLs are essential for clearing viral infections, controlling intracellular bacteria, and eliminating cancer cells that display mutant peptides (neoantigens) on MHC class I.

Helper T Cells (CD4⁺) Activate Other Immune Cells

CD4⁺ helper T cells (T_H cells) do not kill directly. Instead, they orchestrate the immune response by activating other immune cells through cytokine secretion and cell-contact signals. Different helper T cell subsets produce different cytokine profiles and coordinate different types of immune responses:

CD4⁺ T cells are critical for virtually every adaptive immune response. They help B cells undergo class switching and affinity maturation in germinal centers (via T_FH cells), activate macrophages to kill intracellular pathogens (via T_H1 cells), and promote antibody production. The devastating consequences of CD4⁺ T cell loss are starkly illustrated by HIV/AIDS, where the virus destroys CD4⁺ T cells, leading to collapse of the adaptive immune system.

T Cell Activation Requires Co-stimulation

Recognition of a peptide-MHC complex by the TCR is necessary but not sufficient for T cell activation. Full activation requires a second signal — co-stimulation — provided by the interaction of co-stimulatory receptors on the T cell with their ligands on the antigen-presenting cell. The best-characterized co-stimulatory pair is:

• CD28 (on the T cell) binding to B7 molecules (CD80/CD86 on the APC)

Without co-stimulation, a T cell that encounters its antigen becomes anergic (functionally unresponsive) rather than activated. This safeguard prevents T cells from being activated by self-peptides displayed on normal tissue cells that do not express B7.

CTLA-4 is a co-inhibitory receptor on T cells that competes with CD28 for B7 binding but delivers an inhibitory signal, dampening the immune response. Similarly, the PD-1/PD-L1 pathway inhibits T cell activity. Cancer cells frequently exploit these inhibitory pathways to evade immune destruction — the basis for immune checkpoint immunotherapy. Antibodies that block CTLA-4 (ipilimumab) or PD-1/PD-L1 (pembrolizumab, nivolumab) remove these brakes and unleash anti-tumor T cell responses.

Regulatory T Cells Suppress Immune Responses

Regulatory T cells (T_regs) are a specialized subset of CD4⁺ T cells that suppress immune responses rather than promote them. They are characterized by expression of the transcription factor Foxp3 and the surface markers CD25 (IL-2 receptor α chain) and CTLA-4. T_regs maintain immune homeostasis by:

• Preventing autoimmune reactions against self-antigens
• Limiting the magnitude of immune responses to prevent tissue damage
• Maintaining tolerance to commensal gut bacteria and food antigens

T_regs suppress other immune cells through multiple mechanisms: secretion of inhibitory cytokines (IL-10, TGF-β), consumption of IL-2 (starving effector T cells), and direct cell-contact inhibition via CTLA-4. Loss of T_reg function leads to devastating multi-organ autoimmunity, as demonstrated by the IPEX syndrome (immune dysregulation, polyendocrinopathy, enteropathy, X-linked) caused by mutations in the FOXP3 gene.

T Cell and MHC Bioinformatics

Computational immunology has become essential for translating immunological knowledge into clinical applications:

MHC/HLA typing from sequencing data. Determining a patient’s HLA alleles is critical for transplantation, disease-association studies, and neoantigen prediction. Computational tools perform HLA typing from next-generation sequencing data: OptiType uses an integer linear programming approach optimized for HLA class I typing from RNA-seq or whole-exome data; HLA-LA performs graph-based typing from whole-genome sequencing; arcasHLA provides fast, accurate typing from RNA-seq. These tools have largely replaced serological typing in research settings.

Peptide-MHC binding prediction. Perhaps the most impactful computational tools in immunology are those that predict which peptides will bind to specific HLA alleles. NetMHCpan (now in its fourth generation) uses neural networks trained on binding affinity and eluted ligand mass spectrometry data to predict peptide-MHC class I binding for any HLA allele, even those with limited experimental data. MHCflurry is an open-source alternative using deep neural networks. For MHC class II, NetMHCIIpan provides analogous predictions. These tools are foundational for epitope discovery, vaccine design, and neoantigen prediction.

Neoantigen prediction pipelines. In cancer immunotherapy, identifying which tumor-specific mutations generate peptides that bind the patient’s HLA alleles is critical for personalized treatment. Pipelines such as pVACseq (part of the pVACtools suite) and MuPeXI integrate somatic mutation calls, HLA typing, and peptide-MHC binding predictions to prioritize candidate neoantigens for vaccine or adoptive T cell therapy.

T cell epitope databases. The Immune Epitope Database (IEDB) is the largest public repository of experimentally verified immune epitopes, cataloging peptide sequences, their MHC restrictions, and the assay data supporting their immunogenicity. IEDB also hosts a suite of prediction tools for binding, processing, and immunogenicity.

TCR-pMHC interaction prediction. Understanding which TCR sequences recognize which peptide-MHC complexes is a grand challenge. TCRdist quantifies sequence similarity between TCRs and identifies clusters of TCRs with shared specificity. ERGO uses deep learning to predict TCR-epitope binding from sequence alone. These tools are valuable for interpreting TCR repertoire data and predicting immune responses.

Single-cell TCR-seq integration with transcriptomics. Technologies such as 10x Genomics Chromium enable simultaneous profiling of TCR sequences and gene expression in individual T cells. Computational frameworks integrate these data to connect clonotype identity with functional state — for example, identifying which T cell clones in a tumor are exhausted, which are cytotoxic, and which are regulatory.

Let’s explore how antibodies recognize antigens by using local alignment to model the complementarity between CDR3 loops and epitope surfaces.

let antibody_cdr = "ARDYGDYWYFDV"
let epitope1 = "NITNLCPFGEVFNATR"
let epitope2 = "KSNLKPFERDISTEIY"
let match1 = Align.local(antibody_cdr, epitope1)
let match2 = Align.local(antibody_cdr, epitope2)
print("CDR3 vs Epitope 1:")
print("Score: " + match1.score)
print(match1.alignment)
print("CDR3 vs Epitope 2:")
print("Score: " + match2.score)
print(match2.alignment)

Local alignment identifies the best-matching subsequences between the antibody CDR3 and each epitope, modeling the molecular complementarity that drives binding specificity. Higher alignment scores suggest greater structural compatibility.

let vgene_usage = '[{"label": "IGHV1-69", "value": 18}, {"label": "IGHV3-30", "value": 15}, {"label": "IGHV4-34", "value": 12}, {"label": "IGHV3-23", "value": 10}, {"label": "IGHV1-2", "value": 8}, {"label": "Other", "value": 37}]'
let chart = Viz.bar(vgene_usage, '{"title": "V-Gene Usage in Antibody Repertoire (%)", "color": "#3B82F6"}')
print(chart)

V-gene usage profiling reveals which germline gene segments are most frequently used in an antibody repertoire. Certain V-genes are preferentially recruited during specific immune responses — for example, IGHV1-69 is heavily used in antibodies against influenza hemagglutinin.

Exercise: Antibody-Epitope Binding

Antibodies bind antigens through complementarity between CDR loops and epitope surfaces. Use local alignment to find the best matching epitope for a given antibody CDR3:

let cdr3 = "ARDYGDYWYFDV"
let epitope_a = "YGDYWGQGTLVTV"
let epitope_b = "KSNLKPFERDISTEIY"
let match_a = Align.local(cdr3, epitope_a)
let match_b = Align.local(cdr3, epitope_b)
print("CDR3 vs Epitope A:")
print("Score: " + match_a.score)
print(match_a.alignment)
print("CDR3 vs Epitope B:")
print("Score: " + match_b.score)
print(match_b.alignment)
// Which epitope is the better match?
let answer = "epitope_A"
print(answer)

Local alignment scores model the binding complementarity between antibody CDR3 regions and candidate epitopes. In real-world antibody engineering, structural docking and binding affinity assays complement sequence-based approaches.

Exercise: V-Gene Usage Bias

Different immune responses use different V-gene segments preferentially. Compare V-gene usage between a normal repertoire and one responding to a specific pathogen:

let normal = '[{"label": "IGHV1-69", "value": 8}, {"label": "IGHV3-30", "value": 12}, {"label": "IGHV4-34", "value": 10}, {"label": "IGHV3-23", "value": 11}, {"label": "Other", "value": 59}]'
let flu_response = '[{"label": "IGHV1-69", "value": 45}, {"label": "IGHV3-30", "value": 15}, {"label": "IGHV4-34", "value": 5}, {"label": "IGHV3-23", "value": 5}, {"label": "Other", "value": 30}]'
print("Normal repertoire:")
print(Viz.bar(normal, '{"title": "Normal V-Gene Usage (%)", "color": "#3B82F6"}'))
print("Influenza response:")
print(Viz.bar(flu_response, '{"title": "Flu Response V-Gene Usage (%)", "color": "#EF4444"}'))
// Which V-gene is most enriched during the flu response?
let answer = "IGHV1-69"
print(answer)

V-gene usage bias during immune responses reflects the structural compatibility between certain germline-encoded antibody frameworks and specific pathogen antigens. IGHV1-69 is a well-documented example: its germline-encoded CDR-H2 loop provides a hydrophobic surface that is pre-adapted for binding the stem region of influenza hemagglutinin.

Knowledge Check

Summary

In this lesson you learned:

• Antibodies are Y-shaped proteins composed of two heavy chains and two light chains, with two identical antigen-binding sites formed by six CDR loops (three from V_H, three from V_L)
• Antibody genes are assembled by V(D)J recombination during B cell development, generating enormous diversity through combinatorial and junctional mechanisms
• Somatic hypermutation (driven by AID) diversifies antibody variable regions, and selection in germinal centers progressively improves affinity (affinity maturation)
• Class switch recombination changes the antibody constant region (IgM → IgG, IgA, or IgE), altering effector function without changing antigen specificity
• The germinal center reaction produces high-affinity, class-switched plasma cells and memory B cells
• Monoclonal antibodies are among the most important therapeutic drugs, with applications in cancer, autoimmunity, and infectious disease
• Antibody bioinformatics tools (IgBLAST, ANARCI, ABodyBuilder, IgFold, Thera-SAbDab) enable sequence analysis, structure prediction, humanization design, and database mining
• MHC class I presents intracellular peptides to CD8⁺ cytotoxic T cells; MHC class II presents extracellular peptides to CD4⁺ helper T cells
• MHC genes are the most polymorphic in the genome, maintained by balancing selection to ensure population-level pathogen resistance
• CD8⁺ CTLs kill infected or cancerous cells via perforin/granzyme and Fas/FasL pathways
• CD4⁺ helper T cell subsets (T_H1, T_H2, T_H17, T_FH) coordinate diverse immune responses through distinct cytokine programs
• T cell activation requires both TCR engagement and co-stimulation (CD28-B7); co-inhibitory receptors (CTLA-4, PD-1) are targets for cancer immunotherapy
• Regulatory T cells (Foxp3⁺) suppress immune responses and maintain self-tolerance
• Computational tools for HLA typing (OptiType, HLA-LA, arcasHLA), peptide-MHC binding prediction (NetMHCpan, MHCflurry), neoantigen prediction (pVACseq, MuPeXI), and TCR analysis (TCRdist, ERGO, IEDB) are now central to immunology and immunotherapy