T-cell and B-cell activation, class switching and anergy

T-cell and B-cell activation are best understood as a problem of proof. The immune system must decide whether an antigen encounter is dangerous enough to justify clonal expansion, antibody production and tissue inflammation. It solves this problem by requiring several independent signals rather than one. Signal one is recognition of antigen: peptide displayed on MHC for a T cell, or intact antigen bound by the B-cell receptor for a B cell. Signal two is a co-stimulatory or innate context signal that says the antigen is associated with infection or tissue damage. Additional cytokine signals then instruct the type of response. Without signal two, the lymphocyte does not simply remain idle; it often becomes anergic, meaning it survives but responds poorly on later encounter.

B-cell activation begins with the B-cell receptor, a membrane immunoglobulin that recognises native antigen. During development in the bone marrow, B cells rearrange their immunoglobulin genes to assemble a functional B-cell receptor through site-specific V(D)J recombination^[1]. Upon antigen encounter in secondary lymphoid organs, the antibody repertoire is further diversified by somatic hypermutation and class switch recombination^[1]. This sequence matters: the initial receptor is assembled before antigen is seen, and the fine-tuning happens after antigen is seen. The first encounter selects clones that can bind; the germinal centre reaction then improves them.

T-cell activation follows the same logic but with an important difference. The T-cell receptor cannot see intact antigen. It sees short peptides bound to MHC molecules on the surface of an antigen-presenting cell. Signal one is therefore peptide-MHC engagement. Signal two is delivered by co-stimulatory molecules such as CD80 or CD86 binding CD28. Without this second signal, the T cell may become anergic or die. With it, the T cell receives permission to proliferate and differentiate. This two-signal model is the conceptual foundation for understanding adjuvants, checkpoint inhibitors and transplant immunosuppression.

Class switch recombination changes the constant region of the immunoglobulin heavy chain without changing antigen specificity. The B cell keeps the same variable region, so it still binds the same antigen, but it replaces IgM or IgD with IgG, IgA or IgE. This allows the antibody to acquire new effector functions: better placental transfer, mucosal secretion, complement activation or mast-cell sensitisation. Somatic hypermutation changes the variable region and therefore the affinity of the receptor. The two processes are linked by a requirement for activation-induced cytidine deaminase, which is believed to initiate these DNA modification reactions by deaminating cytidine residues in DNA^[1]. Transcription facilitates targeting of both reactions by exposing single-stranded DNA, which is the substrate for the enzyme, and the efficiency of both correlates directly with the rate of transcription^[1].

The germinal centre is the anatomical site where these events are coordinated. B cells divide rapidly, mutate their antibody genes and compete for limiting help from T follicular helper cells. The outcome is selection of higher-affinity clones and, in some cases, switching to other antibody classes. The germinal centre is therefore not a static structure but a competition arena. Its output depends on the quality of the antigen, the number and affinity of T follicular helper cells, and the balance between proliferation and death.

A final definition is useful before the details. Class switch recombination is a DNA recombination event between switch regions upstream of constant-region genes. Somatic hypermutation is a targeted mutational process in the variable region. Both are initiated by the same enzyme but resolved by different repair pathways. When these pathways go wrong, the consequences include immunodeficiency, autoantibody production and B-cell lymphoma. The rest of this page explains how the signals, the germinal centre and the mutational machinery fit together.

Signal one for a T cell is the engagement of the T-cell receptor by peptide bound to an MHC molecule. The receptor itself has no intrinsic enzymatic activity; it signals through associated CD3 chains. The key point is specificity: only a T cell whose receptor fits that particular peptide-MHC combination will receive signal one. This is why T-cell responses are described as antigen-specific and MHC-restricted. The same peptide presented by a different MHC allele may not be recognised at all, which is the cellular basis of HLA restriction and of transplant rejection.

Signal one for a B cell is the engagement of the B-cell receptor by antigen. Because the B-cell receptor is membrane immunoglobulin, it can bind native protein, carbohydrate or lipid epitopes without processing. This allows B cells to recognise pathogens directly, but it also means that signal one alone is not enough to decide whether a response should proceed. B cells need a second signal, which can come from complement receptors, Toll-like receptors, or helper T cells through CD40 and cytokines. The requirement for a second signal is a safety mechanism: it prevents antibody production against harmless self-antigens that happen to bind the receptor.

Signal two for T cells is co-stimulation. The best-characterised pathway is CD80 or CD86 on the antigen-presenting cell binding CD28 on the T cell. CD28 engagement recruits signalling molecules that amplify T-cell receptor signals and lower the threshold for activation. Without CD28 co-stimulation, T cells can become anergic, meaning they survive but fail to respond properly when they later encounter the same antigen. This is why dendritic cell maturation matters so much: maturation increases CD80 and CD86 expression and therefore increases the chance that a T cell will receive both signals.

Signal two for B cells is more varied. Toll-like receptors can provide a co-stimulatory signal when they recognise microbial molecules. In the context of autoantibody production, the TLR9-MyD88 pathway is required for the generation of pathogenic anti-DNA autoantibodies of particular IgG subclasses in vivo. The TLR9-MyD88 pathway provides a co-stimulatory signal for B-cell stimulation by DNA. This is a clear example of signal two shaping not just whether a response occurs but which class of antibody is produced. ^[2]

The distinction between signal one and signal two is clinically important. Checkpoint inhibitors used in cancer therapy block inhibitory receptors such as CTLA-4 or PD-1, which normally restrain co-stimulation. The result is enhanced T-cell activation against tumours, but also a risk of autoimmune toxicity. Conversely, drugs that block co-stimulation are used in transplantation and autoimmune disease to induce anergy or tolerance. The same two-signal logic explains why adjuvants are included in vaccines: they provide signal two in the form of innate immune activation, allowing a weak antigen to generate a strong adaptive response.

A useful way to remember the hierarchy is that signal one answers "what is it?" and signal two answers "is it dangerous?". Signal three, the cytokine environment, answers "what kind of response should it be?". A T cell that receives signal one and signal two but no instructive cytokines may still differentiate, but the resulting population may be inappropriate for the pathogen. This is why innate immune signals are so influential: they determine the cytokines produced by antigen-presenting cells and therefore the polarisation of the responding T cells.

Co-stimulation converts antigen recognition into a proliferative programme. Once a T cell receives signal one and signal two, it begins to produce interleukin-2 and to express the high-affinity interleukin-2 receptor. Interleukin-2 then acts in an autocrine and paracrine manner to drive clonal expansion. The result is a burst of antigen-specific cells that can traffic to sites of infection and provide help to other immune cells. Without co-stimulation, this programme is not initiated, and the T cell may instead become anergic or regulatory.

B-cell clonal expansion follows a similar logic but with different molecular details. Activated B cells enter follicles, form germinal centres and begin to divide. During this period, they mutate their immunoglobulin genes and compete for survival signals. The competition is not simply for antigen; it is for help from T follicular helper cells. Limiting numbers of T follicular helper cells interact with and select B cells based on the latter's ability to bind, internalise and present antigen as peptide complexed with MHC proteins^[1]. Thus, B-cell selection requires competition for limiting T follicular helper signals^[1].

The germinal centre reaction is dynamic. T follicular helper cells themselves undergo antigen-dependent selection throughout the germinal centre reaction, resulting in differential proliferative expansion and contraction^[1]. Increasing the amount of antigen presented in the germinal centre leads to increased T follicular helper cell division^[1]. Competition between T follicular helper cell clones is mediated by T-cell receptor affinity for peptide-MHC ligand^[1]. T cells expanding preferentially in the germinal centre show increased expression of genes downstream of the T-cell receptor, genes required for metabolic reprogramming, cell division and cytokine production^[1]. This means that the helper population is not fixed; it is selected alongside the B cells it helps.

Clonal expansion is also metabolically demanding. Activated lymphocytes switch to glycolysis and other biosynthetic pathways to support rapid division. This is why nutrients and oxygen supply can influence the outcome of an immune response. In the germinal centre, the rapid division of B cells creates a need for nucleotide synthesis and DNA repair, especially because somatic hypermutation introduces DNA lesions. Cells that cannot repair these lesions or cannot meet the metabolic demand are eliminated by apoptosis.

Co-stimulation also determines the balance between effector and regulatory outcomes. Strong co-stimulation in the presence of inflammatory cytokines favours effector differentiation, whereas weak or altered co-stimulation can favour regulatory T-cell induction. This is one reason why the same antigen delivered with different adjuvants can produce opposite immunological effects. The immune system interprets the context of antigen presentation, not just the antigen itself.

In B cells, co-stimulation through CD40 is particularly important. CD40 ligand on activated T cells engages CD40 on B cells, promoting survival, proliferation and class switching. Without CD40 signalling, B cells cannot mount effective T-dependent antibody responses. This is the cellular basis of the hyper-IgM syndrome, in which CD40 ligand deficiency causes poor class switching and susceptibility to opportunistic infections. The clinical lesson is that co-stimulation is not optional decoration; it is a required part of the activation programme.

The germinal centre is a specialised microenvironment within secondary lymphoid organs where B cells proliferate, mutate their antibody genes and undergo affinity selection. It is not present at steady state; it forms after antigenic stimulation and resolves when the response is complete. The germinal centre contains several zones, including a dark zone where B cells divide and mutate, and a light zone where they encounter follicular dendritic cells and T follicular helper cells. The spatial organisation allows mutation and selection to occur in sequence rather than simultaneously.

The germinal centre reaction is regulated by T follicular helper cells that express Bcl6, the chemokine receptor CXCR5, high levels of PD-1, and B-cell trophic cytokines, each of which is required to orchestrate the germinal centre reaction^[1]. These cells interact with B cells and select them based on the B cell's ability to bind, internalise and present antigen as peptide complexed with MHC proteins^[1]. The selection is therefore a test of both receptor affinity and antigen processing. A B cell that binds antigen poorly will present few peptides and receive little help; a B cell that binds antigen well will present more peptides and receive more help.

The kinetics of the germinal centre reaction can be measured in animal models. T follicular helper cells were first detected four to five days after immunization and expanded exponentially thereafter, reaching a peak at day seven to ten, before contracting by day twenty-one^[1]. This time course is useful clinically because it explains why antibody responses take several days to become detectable and why booster immunizations are timed to coincide with the expansion or memory phase. It also explains why the germinal centre is a target for pathogens that wish to evade antibody responses.

The germinal centre is also a site of competition among T follicular helper cells. When the affinity of a T follicular helper clone for peptide-MHC was increased, forty-six per cent of OTII T cells became T follicular helper cells after seven days in response to the high-affinity ligand, as opposed to only twelve per cent following low-affinity immunization^[1]. This shows that the helper compartment is itself selected by antigen quality. It also shows that the germinal centre is a dynamic ecosystem in which both the B cells and the T cells that help them are subject to competition.

Class switching occurs within the germinal centre and is influenced by the cytokine environment and by innate signals. The TLR9-MyD88 pathway is required for the generation of pathogenic anti-DNA autoantibodies of particular IgG subclasses in vivo. In the absence of TLR9, these autoantibodies were nearly as strongly reduced as in MyD88-deficient mice, suggesting that TLR9 is the main MyD88-dependent inducer of those subclasses in the lupus-prone mice studied. This is an example of how an innate signal can bias the class of antibody produced, even within a T-dependent germinal centre response.

The germinal centre is also where sequential class switching can occur. High-affinity IgE can be generated through sequential switching involving an IgG intermediate. In mice lacking the ability to generate IgG1, IgE antibodies are exclusively generated through direct switching. This matters because the pathway of switching affects the degree of affinity maturation: sequential switching through an IgG intermediate allows the B cell to accumulate more mutations and therefore higher affinity. The germinal centre is therefore not just a factory for antibody; it is a place where the route of class switching determines the quality of the final product.

Somatic hypermutation introduces point mutations into the variable region of rearranged immunoglobulin genes. The purpose is to generate a family of related B-cell clones with slightly different receptors, some of which bind antigen better than the original. Antigen then selects the best binders in the germinal centre. This is Darwinian selection at the cellular level, and it is the reason why antibody affinity increases during the course of an immune response.

The molecular details are increasingly well understood. Somatic hypermutation introduces point mutations in the variable region of rearranged immunoglobulin V genes to generate families of related B-cell clones bearing receptors with different antigen-binding affinities, which are selected by antigen in germinal centres^[1]. Transcription-dependent mutations occur within a two-kilobase region downstream of the immunoglobulin promoter^[1]. Although a consensus for initiation of mutation is not defined, about fifty to sixty per cent of the mutations are found in RGYW hotspots^[1]. This tells us that the mutational machinery is not random; it prefers certain sequence contexts.

The enzyme that initiates somatic hypermutation is activation-induced cytidine deaminase. It deaminates cytidine residues in DNA, creating uracil, which is then processed by repair pathways to produce mutations. The same enzyme initiates class switch recombination, which is why the two processes are linked^[1]. Transcription facilitates targeting by exposing single-stranded DNA, which is the substrate for the enzyme, and the efficiency of both reactions correlates directly with the rate of transcription^[1]. This explains why active immunoglobulin loci are mutated while silent loci are not.

The role of chromatin and histone modifications has been studied using genetic models. Phosphorylation of histone H2AX is thought to alter chromatin structure, and this modification could alter transcription^[1]. However, H2AX deficiency does not have an impact on transcription of switch regions, or the accessibility of switch regions to activation-induced cytidine deaminase as measured by mutation, or intra-switch region recombination^[1]. Consistent with this, similar levels of switch-region mutation were found in H2AX-deficient and control B cells that had completed five cell divisions^[1]. Analysis of germinal centre B cells revealed no differences between H2AX-deficient and wild-type Peyer's patch germinal centre cells in mutation frequencies or in the proportion of mutated clones^[1]. ^[3]

The pattern of mutations can be used to infer the activity of the mutational machinery. In one study, more than ninety-five per cent of the mutations in switch regions accumulated in a one-thousand-base-pair region flanking the actual junction, five hundred base pairs on each side. There were significantly more mutations in sequentially switched junctions than in directly switched junctions, with values of five point seven versus twelve point four. Sixty-seven per cent of the sequentially switched clones had more than ten mutations, whereas only nineteen per cent of the directly switched clones had more than ten mutations. Conversely, two thirds of the directly switched junctions had zero to four mutations, whereas only seventeen per cent of the sequentially switched junctions had such low mutation levels.

The target motif of the enzyme is also informative. Over seventy per cent of the mutations had the activation-induced cytidine deaminase target motif WGCW. The difference in the number of mutations between direct and sequential switching increased as the number of immunizations increased, from five point seven in direct switching junctions to nine point eight in sequential switching junctions after four immunizations. The preferred motif WGCW is present four hundred and fifteen times in the mu switch region, at a density of one hundred and thirty-four point seven per kilobase, one hundred and twenty times in the epsilon switch region, at a density of forty-one point nine per kilobase, and four hundred and one times in the gamma-one switch region, at a density of thirty-four point nine per kilobase. These differences in motif density help explain why some switch regions are more mutable than others.

The consequence of somatic hypermutation is affinity maturation. In a T-dependent response, ten to thirty per cent of wild-type switch junctions had remnants of an IgG intermediate, depending on the mice. In mice unable to generate that intermediate, IgE antibodies were exclusively generated through direct switching. When a polyclonal stimulus was used, seventy-nine out of eighty clones lacked the intermediate remnant, and only about one per cent of the IgE was sequentially switched, compared with fourteen per cent in mice immunized once with antigen. This shows that the pathway of class switching is not uniform; it depends on the type of stimulus and on whether the germinal centre reaction is engaged.

At the level of individual cells, the germinal centre is a highly dynamic place. Twenty per cent of germinal centre T follicular helper cells were in the S or G2M phases of the cell cycle after labelling^[1]. Single-cell transcriptomics showed that three hundred and forty-seven of the eight hundred and fifteen differentially expressed genes between expanding and contracting clones changed in parallel with genes responsive to increased antigen presentation^[1]. Genes expressed in at least ten per cent of all cells belonging to expanding or contracting clones were selected as statistically significant differentially expressed genes^[1]. These data reinforce the view that somatic hypermutation and selection are not separate processes; they are coupled within a continuously remodelling germinal centre.

In clinical terms, somatic hypermutation explains several important phenomena. It explains why antibody affinity increases after infection or vaccination. It explains why some B-cell lymphomas carry highly mutated immunoglobulin genes, which is a marker of germinal centre origin. It explains why immunodeficiency caused by defects in activation-induced cytidine deaminase or in DNA repair produces hyper-IgM syndrome with poor class switching and poor affinity maturation. And it explains why therapeutic antibodies that target germinal centre pathways can reduce autoantibody production in diseases such as lupus.

Class switch recombination is the process by which a B cell changes the constant region of its immunoglobulin heavy chain while keeping the same variable region. The consequence is that the antibody retains its antigen specificity but acquires new effector functions: the ability to activate complement, to be transported across mucosa, to cross the placenta or to bind mast cells. The reaction is a region-specific recombination event that replaces the IgM constant region with a downstream constant-region gene, thereby switching the antibody isotype expressed while retaining receptor specificity^[1]. This definition is worth learning precisely, because it separates class switching from affinity maturation. Class switching changes what the antibody does; somatic hypermutation changes how well it binds.

The molecular mechanics follow from the structure of the immunoglobulin heavy chain locus. Class switch recombination involves large repetitive switch region sequences located upstream of each constant-region gene^[1]. The reaction is deletional: the intervening DNA is excised as a circular episome, and the ends are joined by nonhomologous end joining^[1]. This is why the reaction depends on components of the nonhomologous end joining pathway of DNA repair^[1]. The requirement for this pathway is independent of any role in cell division, because Ku80-deficient B cells that divide in response to lipopolysaccharide and interleukin-4 are still unable to complete the reaction as measured by IgG1 expression or intra-switch region recombination^[1]. The interpretation is that B cells producing double-strand breaks without Ku80 cannot repair them and die by apoptosis^[1].

The reaction is initiated by activation-induced cytidine deaminase, which deaminates cytidine residues in DNA to create uracil^[1]. The resulting uracil-guanine mismatches are recognised and processed by uracil-DNA glycosylase and components of the mismatch repair pathway, including MSH2, PMS2 and MLH1^[1]. In the absence of any of these factors, class switch recombination is impaired^[1]. MLH1 or PMS2 deficiency increases the length of homology of recombination junctions, whereas MSH2 deficiency affects local sequence specificity and focusing of junctions to consensus motifs^[1]. These details matter because they show that the reaction is not a single enzyme but a coordinated programme of DNA damage and repair.

Transcription is the targeting mechanism. Class switch recombination is transcription dependent, and recombination is targeted to individual switch regions by T cell-derived cytokines that induce transcription from intronic promoters located upstream of each switch region^[1]. This explains why different cytokines direct switching to different isotypes. Interleukin-4 favours IgG1 and IgE, interferon-gamma favours IgG2a in mice, and transforming growth factor-beta favours IgA and IgG2b in mice. The cytokine signal is therefore the instruction that tells the B cell which constant region to use.

A useful experimental model illustrates the kinetics. When wild-type B cells were stimulated with lipopolysaccharide and interleukin-4, a significant increase in mutation frequency in the S mu region was found after 48 hours, and mutations accumulated at a rate of 1.6 times 10 to the minus 4 mutations per base pair per day^[1]. The proportion of mutated sequences reached a maximum of 23 per cent at 96 hours of stimulation^[1]. Mutation was specific for the S mu region, with no significant increase in the intronic enhancer, the variable region or the IgG3 switch region^[1]. These numbers are a useful reminder that class switching is a process measured in days, not minutes, and that it is targeted rather than genome-wide.

The role of chromatin and histone modification has been dissected genetically. Phosphorylated histone H2AX and the Nijmegen breakage syndrome protein form nuclear foci at the immunoglobulin heavy chain locus in the G1 phase of the cell cycle, and these foci are activation-induced cytidine deaminase dependent^[1]. H2AX-deficient lymphocytes show impaired class switch recombination^[1]. However, the defect is not in targeting. Induction of mu and gamma-one sterile transcripts was similar in H2AX-deficient and control B cells, but the gamma-one sterile transcription remained elevated in H2AX−/− B cells while the gamma-one circle transcript was only modestly increased, consistent with decreased switching in the absence of H2AX^[1]. Neither the rate nor the distribution of somatic hypermutation is affected by the absence of H2AX, so H2AX is entirely dispensable for the targeting stage and initial lesion formation in both reactions^[1]. The conclusion is that H2AX alters the efficiency of class switch recombination but does not have a role in processing DNA ends before joining^[1].

Class switching can occur by direct or sequential routes. In direct switching, the B cell recombines from IgM directly to the final isotype. In sequential switching, it passes through an intermediate isotype. The route matters because it affects the opportunity for affinity maturation. High-affinity IgE can be generated through sequential switching involving an IgG intermediate, and in mice unable to generate IgG1, IgE antibodies are exclusively generated through direct switching. In wild-type mice, ten to thirty per cent of switch junctions had remnants of the IgG1 intermediate, depending on the mouse. When a polyclonal stimulus was used, seventy-nine out of eighty clones lacked the intermediate remnant, and only about one per cent of the IgE was sequentially switched, compared with fourteen per cent in mice immunized once with antigen. The clinical implication is that the pathway of switching is not uniform; it depends on the type of stimulus and on whether a germinal centre reaction is engaged.

Tolerance is the set of mechanisms that prevent the immune system from attacking self. It operates at two levels. Central tolerance removes strongly self-reactive lymphocytes during their development in the bone marrow or thymus. Peripheral tolerance controls self-reactive cells that escape into the circulation. Anergy is one of the principal mechanisms of peripheral tolerance. It is a state of functional unresponsiveness induced when a lymphocyte recognises antigen without adequate co-stimulation. The cell survives but responds poorly on subsequent encounter.

The two-signal model explains why anergy exists. Signal one is antigen recognition. Signal two is co-stimulation, delivered by CD80 or CD86 binding CD28 on T cells, or by innate receptors and CD40 on B cells. If a T cell receives signal one without signal two, it does not simply ignore the antigen. It enters a state in which it cannot be fully activated later. This is a safety mechanism: self-antigens are usually presented without co-stimulation, so self-reactive T cells that escape central tolerance are silenced in the periphery. The same logic applies to B cells, where antigen recognition without a second signal can lead to tolerance rather than antibody production.

Central tolerance is not perfect, and its failures are informative. In the anti-DNA knockin model, the mice already develop autoantibodies after five weeks. The increased escape of polyreactive B cells from central tolerance in Fc gamma RIIB-deficient mice resembles the phenotype seen in paediatric systemic lupus erythematosus patients. This is a useful clinical bridge: the same checkpoint that fails in a mouse model is implicated in human disease. It also shows that central tolerance is a quantitative barrier rather than an absolute one.

Peripheral tolerance uses several mechanisms. Anergy is one. Deletion is another. Regulatory T cells are a third. The balance between these mechanisms determines whether a self-reactive clone is silenced, removed or actively suppressed. When the balance fails, autoimmunity follows. The TLR9-MyD88 pathway provides a good example of how an innate signal can break tolerance. The pathway is required for the generation of pathogenic anti-DNA autoantibodies of particular IgG subclasses in vivo. In the absence of TLR9, these autoantibodies were nearly as strongly reduced as in MyD88-deficient mice, suggesting that TLR9 is the main MyD88-dependent inducer of those subclasses in the lupus-prone mice studied. The mechanism is that TLR9 provides a co-stimulatory signal for B cell stimulation by DNA.

The regulation of antibody class by innate signals is a clear illustration of how tolerance and class switching intersect. Switching to the IgG2a and 2b subclasses was restricted to the endogenous alleles in the MyD88-deficient strain. TLR9-deficient B cells were specifically defective in their ability to secrete IgG2a, 2b and 3 antibodies in response to CpG, anti-CD40 and interleukin-4, whereas IgM secretion was unchanged and IgG1 secretion increased. This is a cell-autonomous defect, not a failure of T cell help. The transcription factor T-bet is part of the mechanism: T-bet expression was ninefold higher in B cells from the TLR9-sufficient strain compared with the TLR9-deficient strain. T-bet deficiency produces a similar pattern of decreased IgG2a and 2b with increased IgG1. The pathway therefore links innate sensing, transcription factor expression and class switching.

The role of inhibitory receptors adds another layer. Fc gamma RIIB is an inhibitory receptor that limits the expansion of plasma cells and secretion of autoreactive IgGs. The reduction of Fc gamma RIIB in other autoimmune-susceptible strains removes this checkpoint and allows for the expansion of plasma cells secreting pathogenic subclasses. This is a good example of a tolerance checkpoint operating at the level of the B cell itself, rather than through T cell help. When the checkpoint is lost, the consequence is not simply more antibody but a bias towards pathogenic classes.

Regulatory T cells are also part of the tolerance network. It has been suggested that at higher doses of immunization, regulatory T cells can be recruited to the response and suppress the IgE response. This is a useful reminder that tolerance is not only about avoiding autoimmunity; it also shapes the magnitude and class of responses to foreign antigens. The same regulatory mechanisms that prevent self-damage can limit vaccine responses, which is why adjuvants are designed to overcome them.

The clinical relevance of anergy is broad. In transplantation, the goal is to induce anergy or deletion of donor-reactive T cells while preserving protective immunity. In autoimmunity, the goal is often to restore tolerance by blocking co-stimulation or by expanding regulatory T cells. In cancer, the goal is the opposite: to break tolerance so that T cells can attack tumour cells. Checkpoint inhibitors block inhibitory receptors such as CTLA-4 or PD-1, which normally restrain co-stimulation, and the result is enhanced antitumour immunity at the cost of autoimmune toxicity. Understanding anergy is therefore not academic; it is the basis for predicting both the benefit and the harm of immunomodulatory drugs.

The first pitfall is to confuse class switching with affinity maturation. Class switching changes the constant region and therefore the effector function of the antibody. Somatic hypermutation changes the variable region and therefore the affinity. They share an initiator enzyme but are distinct processes with distinct outcomes. A B cell can switch class without improving affinity, and it can improve affinity without switching class. The distinction matters clinically because therapies that target class switching may not affect affinity maturation and vice versa.

The second pitfall is to assume that all somatic hypermutation happens in the germinal centre. It is true that germinal centres are the main site of affinity maturation, but the evidence suggests that IgE can be generated by sequential switching through an IgG intermediate, with the IgG1-positive intermediate undergoing the germinal centre phase. The observation that IgE cells are absent from germinal centres during some infections and immunizations is consistent with this model. The germinal centre is therefore not the only place where mutation and selection can occur, and the route of switching determines where the mutation happens.

The third pitfall is to treat H2AX as a general requirement for all immunoglobulin gene diversification. H2AX deficiency impairs class switch recombination but does not affect somatic hypermutation or switch region mutation^[1]. The defect is in the efficiency of the recombination reaction, not in the targeting or initial lesion formation^[1]. A student who remembers only that H2AX is important for class switching may wrongly predict that H2AX deficiency also causes a hypermutation defect. It does not.

The fourth pitfall is to assume that Ku80 is required for all somatic hypermutation. Ku80 deficiency abolishes S mu mutation in B cells stimulated with lipopolysaccharide and interleukin-4^[1]. However, Ku80-deficient mice lack T cells and therefore cannot produce germinal centres, which are required for efficient somatic hypermutation of immunoglobulin variable genes^[1]. It is therefore possible that Ku80 is required for switch region mutation but not for variable region somatic hypermutation^[1]. This is a good example of how a defect in one cell type can masquerade as a defect in another process.

The fifth pitfall is to assume that sequential switching is the only route to IgE. In mice unable to generate IgG1, IgE antibodies are exclusively generated through direct switching. Even in wild-type mice, the proportion of sequentially switched IgE varies with the stimulus: only about one per cent of IgE was sequentially switched after anti-IgD treatment, compared with fourteen per cent after a single protein immunization. The route of switching is therefore context-dependent, and a finding in one model cannot be generalised to all IgE responses. ^[3]

The sixth pitfall is to confuse tolerance with immunodeficiency. Anergy is an active state of unresponsiveness, not a failure of lymphocytes to develop. A patient with anergy may have normal numbers of T and B cells but poor responses to specific antigens. This distinction matters because the management of tolerance-related disease is different from the management of immunodeficiency. In autoimmunity, the goal is to restore tolerance selectively; in immunodeficiency, the goal is to replace or support the missing function.

The biology of class switching and anergy is used every day in allergy, autoimmunity, transplantation and vaccination. In allergy, the class of antibody determines the clinical syndrome. High-affinity IgE is the antibody responsible for immediate hypersensitivity, and its production requires sequential switching through an IgG intermediate, which allows affinity maturation. The consequence is that the potential for causing pathology of low-affinity and high-affinity IgE antibodies is very different. In experimental models, blocking or neutralising IgG antibodies can intercept antigen and prevent or reduce antigen-induced IgE-mediated anaphylaxis, whereas injecting high-affinity anti-peptide IgE would typically cause sensitisation and anaphylaxis upon antigen challenge rather than eliminate the reaction. This is why allergen immunotherapy aims to shift the response away from IgE and towards IgG, and why therapies that reduce IgE production can prevent anaphylaxis.

In autoimmunity, the class of autoantibody often predicts the organ damage. In the lupus model, the TLR9-MyD88 pathway is required for pathogenic anti-DNA autoantibodies of the IgG2a and 2b subclasses, which efficiently trigger inflammatory responses by engaging activation receptors on macrophages. The absence of TLR9 nearly abolished these subclasses, suggesting that TLR9 is the main MyD88-dependent inducer in this model. This provides a rationale for targeting the TLR9-MyD88 pathway in systemic lupus erythematosus. The same logic applies to other autoantibody-mediated diseases: the isotype and subclass of the autoantibody determine which effector mechanisms are engaged, and therefore which therapies are likely to work. ^[2]

In transplantation, the goal is to prevent the recipient from class switching to donor-specific IgG, which mediates chronic rejection. Co-stimulation blockade is one strategy: by preventing signal two, it induces anergy or deletion of donor-reactive T cells and reduces T-dependent B cell help. The same principle explains why belatacept, a CTLA-4-Ig fusion protein, is used in renal transplantation. It blocks CD80 and CD86, preventing co-stimulation and favouring tolerance. The cost is an increased risk of certain infections and post-transplant lymphoproliferative disorder, because the same pathway is needed for antiviral immunity.

In vaccination, the goal is to induce class switching to the appropriate isotype and to generate affinity-matured memory B cells. Protein vaccines require T cell help and therefore engage the germinal centre pathway, generating IgG with high affinity and long-lived plasma cells. Polysaccharide vaccines can induce T-independent responses, which are dominated by IgM and have poor memory. Conjugate vaccines solve this problem by linking polysaccharide to a carrier protein, recruiting T cell help and enabling class switching and affinity maturation. Adjuvants provide signal two in the form of innate activation, allowing weak antigens to generate strong and durable responses.

The clinical importance of the germinal centre is also evident in immunodeficiency. Patients with defects in CD40 ligand, activation-induced cytidine deaminase or uracil-DNA glycosylase have hyper-IgM syndrome, characterised by normal or elevated IgM with poor IgG, IgA and IgE. They suffer from recurrent infections with encapsulated organisms and from opportunistic infections. The defect is not in antigen recognition but in the ability to switch class and to mature affinity. This is why immunoglobulin replacement is the mainstay of treatment for many of these patients.

Finally, the biology of anergy is used in cancer immunotherapy. Checkpoint inhibitors block CTLA-4 or PD-1, releasing T cells from inhibitory signals and allowing them to attack tumour cells. The same release can cause immune-related adverse events that resemble autoimmune diseases, including colitis, thyroiditis and hepatitis. These toxicities are the clinical price of breaking tolerance, and they are managed by stopping the drug and using immunosuppression. Understanding the two-signal model helps predict which patients are at risk and which organs are most likely to be affected.