NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Frank SA. Immunology and Evolution of Infectious Disease. Princeton (NJ): Princeton University Press; 2002.

Cover of Immunology and Evolution of Infectious Disease

Immunology and Evolution of Infectious Disease.

Show details

Chapter 6Immunodominance within Hosts

Each parasite presents a large number of epitopes to the host's immune system. The immune response focuses on only a few of the many potential epitopes, a process called immunodominance. Immunodominant focus determines which epitopes are favored to vary antigenically to escape immune pressure. In this chapter, I describe how immunodominance develops by competition among B and T cell lineages with different specificities.

The first section reviews antibody immunodominance. The diverse, naive B cells secrete IgM antibodies that bind to nearly any epitope. On initial infection, B cells that bind epitopes with relatively high equilibrium affinity divide rapidly and dominate the early phase of the immune response by outcompeting other B cells. However, antibodies that bind too strongly clear the matching antigens quickly and prevent feedback stimulation to their B cells. The later phases of B cell competition and maturation of IgG favor antibodies with increased on-rates of association to epitopes rather than increased equilibrium binding affinity.

The second section discusses cytotoxic T lymphocyte (CTL) immunodominance. Aspects of specificity such as MHC binding and avoidance of self-recognition determine which epitopes could potentially be recognized. Among this potential set, some epitopes dominate others in stimulating a CTL response. Earlier stimulation of T cell lineages in response to infection rather than more rapid T cell division seems to determine the dominance of lineages. Dominant lineages may repress subdominant lineages by pushing the abundance of pathogens below the threshold needed to trigger weaker, subdominant responses.

The third section describes original antigenic sin, in which the specificity of the immune response depends on the sequence of exposure to antigenic variants. If a host first encounters a variant A and then a later variant A′, the second variant will sometimes restimulate the initial response against A rather than a new, specific response against A′. In this case, A′ recalls the memory against an earlier cross-reacting epitope rather than generating a primary, specific response against itself. Sometimes the cross-reaction is rather weak, causing the host to respond weakly to the second antigen because of interference by its memory against the first variant. Original antigenic sin has been observed in both antibody and CTL responses.

The final section takes up promising issues for future research.

6.1. Antibody Immunodominance

I emphasize Rao's (1999) studies of B cell competition, one of the few empirical analyses of the dynamics of antibody immunodominance.

Broad IgM and Narrow IgG Response

Kumar et al. (1992) engineered a recombinant polypeptide with 100 amino acids that contained several epitopes from the envelope of hepatitis B virus. They called this polypeptide MEP-1. The initial antibody response, detected one week after injection into a mouse, contained heterogeneous IgM against several epitopes that collectively spanned the entire 100-amino-acid sequence. By contrast, the IgG response four weeks after injection was highly specific for a single epitope. These observations support the idea that the naive antibody repertoire can bind almost any epitope, but that only a subset of the initially binding antibodies stimulate their B cell clones to expand significantly and make the transition to IgG production.

Review of Processes by Which Antibody Response Develops

Major expansion of a B cell clone and transition to IgG production typically depend on stimulation from helper T cells, although some nonprotein antigens can stimulate IgM response without T cell help (Janeway et al. 1999). The interaction between B cells and T cells happens roughly as follows. The B cell receptor (BCR) is an attached form of antibody, which has specificity for particular epitopes. Each B cell expresses many BCRs on its surface, each with the same specificity. When a BCR binds antigen, it may pull the antigen into the cell. If the antigen is a protein, the B cell processes the antigen into smaller peptides, binds some of those peptides to MHC class II molecules, and presents the peptide–class II complexes on the cell surface.

If a helper (CD4+) T cell has a T cell receptor (TCR) that binds the peptide–class II complex, then the T cell sends a stimulatory signal to the B cell. Thus, B cell stimulation requires binding to an epitope of an antigen, processing the antigen, and finding a helper T cell that can bind an epitope of the same antigen. The epitopes recognized by the BCR and TCR may differ, but must be linked on the same antigen molecule to provide matches to both the BCR and TCR (Shirai et al. 1999). T cell stimulation causes B cells to divide more rapidly, to undergo somatic hypermutation, and to switch from IgM to IgG production. Immunodominance arises when some B cells receive relatively greater stimulation from helper T cells. Signal strength depends on the dynamics of antigen binding for BCRs and TCRs.

The vertebrate host has specialized organs to facilitate interaction between B and T cells. The initial interaction occurs when antigen-binding B cells are trapped in a zone of lymphoid tissue that has a high density of T cells. Some of the stimulated B cells differentiate into antibody factories, whereas others migrate along with matching T cells to primary follicles of the lymphoid tissue. There, if the B cells receive sufficient stimulation from T cells, they undergo rapid division to form germinal centers. At these centers, the B cells hypermutate and proceed through affinity maturation.

Affinity Window for Epitope-Paratope Binding

The naive B cell repertoire binds with varying affinity to different epitopes of an antigen. The relative stimulation of different B cell clones by an antigen determines progression to the next steps in B cell response. Stimulation depends on the affinity of the BCR paratopes (binding sites) for their particular epitopes.

Rao (1999) found an affinity window for stimulation of B cells. Very strong epitope-paratope binding prevents stimulation; weakly binding B cells are outcompeted for stimulatory signals. I discuss in turn these upper and lower thresholds.

Vijayakrishnan et al. (1994) discovered an upper affinity threshold by the study of two epitopes of the MEP-1 peptide described above. One of these epitopes stimulated the immunodominant IgG response; the other was at the opposite end of the peptide. I refer to the immunodominant epitope as D and the subdominant epitope as S. Vijayakrishnan et al. (1994) followed antibody levels against these two epitopes in immunized mice. Surprisingly, the early antibody response was stronger against S than D. However, secreted antibodies against S bound so efficiently to S that they outcompeted the matching BCR and prevented stimulation of the B cell lineage. By contrast, anti-D antibodies bound with lower affinity and did not outcompete the matching BCR, allowing that B cell lineage to receive strong stimulation from the antigen.

Agarwal et al. (1996) found a lower affinity threshold determined by competition between BCRs for stimulation by helper T cells. They began by constructing a peptide that had on one side a known B cell antigen of hepatitis B virus and on the other side a known T cell epitope from the malaria parasite Plasmodium falciparum. They injected this chimeric peptide into mice and followed the antibody response. The early IgM response had specificities that spanned the entire hepatitis B segment. By contrast, the later IgG response focused on a single epitope in the hepatitis B segment composed of the four amino acids DPAF.

Single amino acid changes in DPAF destroyed immunodominance by this epitope, causing nearby epitopes to dominate the IgG response. The anti-DPAF antibodies had affinities between 8- and 60-fold higher than antibodies against neighboring epitopes. Agarwal et al. (1996) showed that the high-affinity response against DPAF suppressed B cells initially activated against neighboring epitopes. Immunodominance depended on competition for antigen-specific helper T cells, which are limiting during the initial stages of an immune response. The stronger-binding BCRs take up antigen and present to T cells more efficiently than do the weaker-binding competitors. Insufficient T cell stimulation leads to suppression of B cell clones.

In later experiments, Agarwal and Rao (1997) manipulated the size of the helper T cell pool. Reduced numbers of T cells allowed IgM response but prevented the switch from the IgM stage to the IgG stage. This supports the hypothesis that competition for T cell help is the rate-limiting step in the transition from the broad IgM response to the narrow IgG response.

Equilibrium Binding Affinity May Determine Early Response

Antibody affinity for epitopes influences initial IgM stimulation and subsequent competition for immunodominance during the switch to IgG. What determines antibody affinity to individual epitopes during these early phases of B cell competition? Rao (1999) summarizes studies that rule out mouse MHC genotype and various physical properties of the epitope such as accessibility within the overall peptide structure. This led to the hypothesis that the Gibbs free-energy of binding between epitope and paratope determines antibody affinity, and that the amino acid sequence of the epitope influences the potential free-energy of the bond.

Nakra et al. (2000) used various model peptides to show that the affinity of an epitope-specific response depends on the amino acid composition of the epitope. They suggested that the relative ordering of affinities for particular epitopes could be predicted by the amino acid sequence of the epitope. In particular, the amino acid side chains of an epitope sequence determine the potential free-energy of binding to an antibody paratope.

Chemical determination of free-energy seems particularly important in the early phases of antibody response, when the antibodies have not yet been optimized for binding by affinity maturation. Unoptimized antibodies do not have strong spatial complementarity of binding; thus there is less steric and greater chemical constraint on binding at this stage. After optimization, it may be that greater steric complementarity of antibody-epitope binding places more emphasis on spatial fit and reduces the predictability of binding energy based solely on chemical composition of amino acid side chains.

Kinetic Binding On-Rates May Determine Affinity Maturation

So far, I have summarized the first stage of antibody selection: IgM-producing B cells from the naive repertoire compete for T cell help, with the winner(s) dividing more rapidly and starting on the path to IgG production. Equilibrium binding affinity drives this first stage of antibody competition.

I now turn to the next stage, called affinity maturation (Janeway et al. 1999). During this stage, B cells congregate in germinal centers of the lymphoid tissue and mutate their antibody paratopes at a high rate. A selection process favors those mutated paratopes that bind relatively strongly to antigen, driving affinity maturation of antibodies for the particular epitopes.

Rao's (1999) group studied affinity maturation by continued analysis of the DPAF epitope in the chimeric hepatitis B and Plasmodium antigen mentioned above. Agarwal et al. (1998) found that the equilibrium binding affinity of antibodies did not increase over time, supporting observations in two earlier studies on other systems (Newman et al. 1992; Roost et al. 1995).

Rao's group modified their model antigen by substituting cysteine amino acid residues in the two sites flanking the DPAF epitope (Nayak et al. 1998). This changes the conformation of the DPAF peptide and influences the antibody-epitope binding reaction. Nayak et al. (1998) raised antibodies through affinity maturation against DPAF in the native antigen and in the cysteine-modified antigen. They then compared binding of each of the two antibody types against the native and modified antigen.

Antibodies raised against the native antigen bound with approximately equal equilibrium affinity to native and modified antigen. Antibodies raised against the modified antigen also bound at equilibrium approximately equally against the two antigens. By contrast, the kinetic on-rates of binding were 50-fold higher for native antibody to native antigen than for native antibody to modified antigen. Kinetic on-rates were 14- to 25-fold higher for modified antibody to modified antigen than for modified antibody to native antigen.

Kinetic on-rates measure rates at which bonds form, whereas equilibrium affinity measures the ratio of on-rates to off-rates. Selection during affinity maturation apparently favors faster rates of interaction with increases in both on-rates and off-rates: the on-rates rise, but the equilibrium affinity does not change.

In this model system, it appears that B cells compete by rate of antigen acquisition during affinity maturation. B cells with paratopes that bind more quickly to antigen receive stronger stimulatory signals to divide and to dominate the population in the germinal centers. Thus, the optimized antibodies bind more quickly to antigen than unoptimized precursors, but optimized antibodies do not necessarily increase their equilibrium binding affinity.

In summary, Rao proposed an integrated, dynamic view of how the specificity of an antibody response develops. The particular details may turn out to vary in different cases. However, in all cases, progress will require study of the interactions between molecular structure, the kinetics of binding, regulatory control of cellular competition, and immunodominance.

6.2. CTL Immunodominance

CTL immunodominance is not well understood, because it has been difficult to measure the abundance of CTLs specific for particular epitopes. The technical limitations for quantitative assay of specific T cells may soon be overcome with recently developed methods (Yewdell and Bennink 1999; Doherty and Christensen 2000).

I mentioned in chapter 4 several factors of antigen processing that affect CTL immunodominance. These factors include CTL repertoire shaped by selection against MHC and self-peptide complexes, timing and quantity of intracellular antigen production by pathogens, uptake of extracellular antigen by antigen-presenting cells, proteolytic cleavage of antigens, intracellular transport of peptides, binding to MHC, and specificity of T cell receptors. In this section, I focus on the relative abundance of T cell populations with different recognition specificities.

Breadth and Specificity of CTL Response

The CTL response can be described by breadth, the number of different CTL clones expanded, and by specificity, the number of pathogen epitopes recognized by the expanded CTL clones (Gianfrani et al. 2000). The current literature provides varying conclusions about CTL breadth and specificity. This partly reflects the technical difficulties mentioned above, but may also occur because the CTL response is variable.

The timing and methods of measurement may influence five aspects of the observed CTL response (Gianfrani et al. 2000). First, some studies measure primary CTL response, whereas other studies measure memory CTLs stimulated by secondary challenge. W. Chen et al. (2000) found that mouse primary response against influenza is more highly focused on a few epitopes than is the secondary response. By contrast, Busch et al. (1998) observed similar kinetics of primary and secondary responses against four epitopes of Listeria.

Second, persistent viral infections may evolve within a host, causing the host to develop a sequence of focused CTL responses.

Third, some methods measure relatively rare CTL-epitope combinations better than other methods. Relatively insensitive measurement leads to observations of narrow response. Relatively sensitive methods may pick up relatively weak CTL responses. The existence of a response does not mean that the response was a significant fraction of the total CTL expansion.

Fourth, it is often necessary to choose a priori a relatively small panel of epitopes as probes for the presence of matching CTLs. As the methods improve to predict CTL epitopes, the number of epitopes observed to stimulate CTL response will rise.

Fifth, some studies measure CTL response aggregated over several hosts. Each host may have a relatively narrow response, but hosts may differ in their choice of epitopes.

With these issues in mind, we can make some sense of the contrasting reports on the diversity CTL response. On the one hand, studies of influenza (Bednarek et al. 1991; Morrison et al. 1992) and Epstein-Barr virus (Callan et al. 1998) report a large fraction of CTLs focused on a single epitope. These studies emphasize the dominance of certain CTL clones at a particular time during infection within a single host (Murali-Krishna et al. 1998; Sourdive et al. 1998; Callan et al. 1998). On the other hand, observed human CTL responses were broad and multispecific against hepatitis B and C viruses and against HIV (Chisari 1995; McMichael and Phillips 1997; Rehermann et al. 1996). These pathogens tend to be genetically heterogeneous within a single host and may evolve by escape mutants in dominant epitopes. Thus, CTL focus may change over the course of infection within a single host.

Gianfrani et al. (2000) found a broad and multispecific human CTL memory response against influenza A. However, their measurements were aggregated over several hosts. Each host tended to respond strongly to a dominant epitope associated with one of its class I MHC alleles and to have memory CTLs for a small number of other epitopes for that dominant class I allele and for another class I allele. It seems that a few CTL clones prevail numerically within each host, but other clones may be stimulated and hosts may vary in which clones react to a particular epitope.

Time of CTL Recruitment

Three factors influence the relative abundance of expanded T cell clones: frequency in the naive repertoire, rate of cell division, and time of initial expansion. Bousso et al. (1999) studied expansion of CTL clone abundances in mice against the human MHC allele HLA-Cw3. The response was dominated by a few clones. Those dominant clones were not particularly frequent in the naive repertoire. The relative abundances did not change between dominant and subdominant CTL clones that increased in abundance from the early to late stages of the T cell response, suggesting that expanding clones did not vary in their rate of cellular division.

The dominant CTL clones began their numerical expansion earlier than subordinate clones. CTL clones double every six to eight hours; thus a one-day advance in clonal activation causes an 8- to 16-fold difference in cellular abundance. The timing of initial clonal expansion appears to control immunodominance in this case. Bousso et al. (1999) did not determine whether the earlier proliferation of certain clones arose from binding properties to epitopes that triggered faster activation or from other causes.

Initial Stimulation by Endogenous Versus Exogenous Antigen

Naive CD8+ T cells must be activated to proliferate and to become armed with killer function as CTLs. Naive CD8+ cells are also confined to the blood and lymph systems and generally do not pass outside to most tissues, whereas the armed CTLs can exit to infected tissues.

The confinement of naive CD8+ cells raises a paradox (Reimann and Schirmbeck 1999). To be activated, the CD8+ cells must bind peptide-MHC class I complexes on the surface of cells with foreign antigen. But if the infection is not in the blood or lymph compartments, the naive T cells cannot reach the site of infection. Somehow, the naive CD8+ cells must encounter peptide-MHC class I complexes within the blood or lymph compartments even though the site of infection may be outside those compartments.

One possible solution depends on the distinction between endogenous and exogenous antigen (Schumacher 1999; Sigal et al. 1999). The CD8+ T cell is traditionally thought to bind primarily to pathogen antigens created endogenously within infected host cells. Those antigens are digested within the cell and transported to the endoplasmic reticulum, where they bind MHC class I molecules. The peptide-MHC complex is then transported to the cell surface, where it becomes available to roving T cells. Endogenously generated antigen is confined to the surface of the cell in which it was produced; thus endogenous antigen cannot be transported from infected tissues that exclude naive CD8+ cells into the lymph nodes, where CD8+ cells are common.

When an infected cell dies, pathogen antigens become liberated and exist exogenously. Sigal et al. (1999) showed that naive CD8+ stimulation against a viral infection of peripheral tissue required transport of exogenous antigen by macrophages or dendritic cells into the lymph nodes. Dendritic cells are known to take up exogenous antigen in peripheral tissues and then to move to lymph nodes (Banchereau et al. 2000; Watts and Amigorena 2000). In addition, dendritic cells can process exogenous antigen and present it bound to MHC class I alleles (Reimann and Schirmbeck 1999). Thus, dendritic cells may serve as scouts in the peripheral tissue, bringing exogenous antigen to lymph nodes when stimulated by signs of infection or tissue damage.

How does this scouting network influence antigenic variation? For pathogens that infect peripheral tissue, CTL stimulation may focus on those antigens likely to be released exogenously and taken up by dendritic cells (or perhaps macrophages). Thus, abundant and stable pathogen proteins may be particularly likely to stimulate CTL response. For example, the capsid proteins of viruses may be more abundant than replicase enzymes and therefore more likely to be taken up as exogenous antigen.

Competition and Interference Between CTL Clones

Immunodominant CTL clones suppress the abundance of subdominant clones, a phenomenon called immunodomination (see the excellent review by Yewdell and Bennink 1999). Subdominant CTL clones responded more strongly with alteration or deletion of either the epitope stimulating the dominant CTL clone or the class I MHC molecule that presents the dominant epitope, or with direct inhibition of the dominant CTL clone (Zinkernagel et al. 1978; Doherty et al. 1978).

Yewdell and Bennink (1999) compare two explanations for immunodomination. First, the dominant epitope may interfere with the subdominant epitope for binding and presentation by MHC class I molecules. Such competition within antigen-presenting cells seems unlikely because CTL recognition of subdominant epitopes is not affected by coexpression of dominant epitopes (Mylin et al. 1995; Weidt et al. 1998). The abundance of epitopes (pathogen peptides) is usually so low that competition for binding to MHC class I molecules seems very unlikely (Yewdell and Bennink 1999).

Second, dominant CTL clones may directly or indirectly suppress subdominant clones. This may occur because dominant CTLs clear infected cells and associated subdominant epitopes too quickly for the weak CTL stimulation by subdominant epitopes to generate a strong CTL response (Nowak et al. 1995; Nowak and May 2000). Alternatively, the CTLs may compete directly at antigen-presenting cells for stimulation. Finally, the dominant CTLs may be able to suppress subdominant clones by competition for resources or by expressing suppressive cytokines.

On the whole, the evidence supports the second explanation, in which dominant clones suppress subdominant clones. Weidt et al.'s (1998) detailed study of lymphocytic choriomeningitis virus (LCMV) showed that immunodomination in that system arose from CTLs against dominant epitopes suppressing the viral population to a low level such that the suppressed viral population in the host stimulates only a weak CTL response against subdominant epitopes. This supports Nowak et al.'s (1995) model, in which immunodomination arises by the population dynamic consequences of birth and death rates for specific CTL clones and for viral populations that express matching epitopes.

In particular, Weidt et al. (1998) analyzed CTL response against two viral strains. The strain WE stimulated a dominant response against the epitope NP118–126, whereas the strain ESC lacked this dominant epitope and stimulated response against various minor epitopes including GP283–291. Class I MHC presents the minor epitopes in WE-infected cells, but does not stimulate significant CTL response. Importantly, CTLs specific for the subdominant epitope GP283–291 lyse WE and ESC target cells to the same extent, suggesting that the subdominant epitope is presented effectively equally on the surfaces of WE and ESC cells. Thus, the strength of the CTL response is not caused by numerical differences in epitope presentation on cell surfaces.

The NP118–126-specific CTLs do not directly suppress CTLs against minor epitopes, because coinfection by WE and ESC produces a significant CTL response against both NP118–126 and GP283–291, suggesting that ESC generates a CTL response against GP283–291 without interference by the WE-induced CTLs against NP118–126.

Although the dominant CTLs did not directly interfere with the subdominant CTL population, further evidence suggests indirect competition. Expansion of the dominant CTL clone against NP118–126 and clearance of WE infection occurred more rapidly than did expansion of the subdominant CTL clone against GP283–291 and clearance of ESC infection. Either WE or ESC infection activated CD8+ T cells against the minor epitope, but in WE infection those minor-epitope T cells did not expand into a significant CTL response with lytic activity. It appears that, in WE infection, the fast development of CTLs against NP118–126 suppressed the viral load quickly enough that the weaker-stimulated CD8+ T cells against GP283–291 did not have time to develop into a primary CTL response.

These kinetic processes lead to indirect competition. Kinetic control suggests that immunodomination should be a quantitative phenomenon ordering epitopes into a hierarchy. An immunodomination hierarchy has been demonstrated by Wettstein (1986). In addition, factors that alter the rate of CTL expansion against particular epitopes should be able to change the dominance hierarchy. Such changes in the hierarchy occur when the immune system has previously experienced an epitope. For example, if epitope A dominates epitope B in a naive host, then prior exposure only to B can reverse the dominance ranking and cause B to dominate A (Bennink and Doherty 1981; Jamieson and Ahmed 1989; Cole et al. 1997). This switch apparently occurs because secondary challenge causes a more rapid CTL response, allowing CTLs against B to reduce antigen load quickly enough to suppress a CTL response against A.

CTL Repertoire

Why are CTL responses stronger against some epitopes than others? It could simply be that the immunodominant epitopes are expressed more commonly on cell surfaces than subdominant epitopes. However, Yewdell and Bennink (1999) summarize various lines of evidence arguing against a simple correlation between the abundance of presented epitopes and immunodominance, for example, the study by Weidt et al. (1998) described above. Thus, immunodomination of CTL clones appears to be influenced by biases in the CD8+ repertoire.

Three important questions arise concerning CD8+ biases in the naive repertoire (Yewdell and Bennink 1999). First, does immunodomination arise because more CD8+ cells respond to an immunodominant epitope or because CD8+ clones for immunodominant epitopes divide more quickly upon initial stimulation? The available data cannot distinguish between these alternatives.

Second, how does variation between individuals in naive CD8+ repertoires influence the hierarchical ordering of epitopes? Individual variation can occur in self-peptides, TCR genes, and MHC genes. Negative selection shapes the TCR repertoire to avoid matching to self-peptides. TCR genes form the building blocks for combinatorial generation of TCR variability in each individual. MHC genes influence the presentation of peptides.

Third, independently of self-peptides and TCR genes, do immunodominant epitopes stimulate TCRs more strongly? Favorable structural attributes of immunodominant epitopes could interact with relatively constant features of the TCR.

A few studies have compared immunodominance in humans with that in transgenic mice expressing the same human MHC alleles. Both humans and transgenic mice recognized the same immunodominant epitopes when injected with viruses (Engelhard et al. 1991; Man et al. 1995; Shirai et al. 1995). Humans and mice that recognized the same peptide-MHC class I complex used TCRs composed of different Vα and Vβ germ-line components (Man et al. 1994). Thus, different self-peptides (mouse versus human) or variable TCR genes do not necessarily influence immunodominance, although this is a rather weak conclusion. Instead, immunodominance may depend on interactions between structural properties of the epitopes and relatively constant features of TCRs (Yewdell and Bennink 1999).

Negative selection against self-peptides can influence CD8+ response to particular epitopes. This was shown in a study of human infection by Epstein-Barr virus (Burrows et al. 1994, 1995). Individuals with the MHC class I allele B8 typically have immunodominant responses against EBNA3A325–333. The CTLs in this immunodominant response have a high proportion of the same germline TCR genes; that is, the response is composed of a very narrow set of CD8+ clones.

CTL response against EBNA3A325–333 cross-reacts to the MHC class I allele HLA B*4402 when the host lacks this HLA allele. Individuals with both B8 and B*4402 produce a relatively weak CTL response against EBNA3A325–333. This weaker CTL response does not cross-react with B*4402 and has a TCR gene composition that differs from the TCRs in the strong CTL response of the B8-positive and B*4402-negative individuals. Self-reactivity apparently reduced the CTLs that cross-react with B*4402. Reduction of those dominant CTLs allows other CD8+ clones to expand, suggesting that the dominant CTLs in the B*4402-negative individuals suppress those CD8+ clones that expand in B*4402-positive individuals.

Altered Peptide Ligands

A CTL response depends on binding of the TCR to a peptide-MHC complex. An altered parasite peptide sometimes interferes with or enhances a CTL response against the original peptide, a phenomenon known as altered peptide ligand (APL) (reviewed by Sette et al. 1994; Jameson and Bevan 1995; Franco et al. 1995; Moskophidis and Zinkernagel 1995; Davis et al. 1998; Price et al. 1998; Germain and Štefanová 1999; Madrenas 1999; Abrams and Schlom 2000).

Disruption of the lytic activity in an active CTL response provides one example of APL antagonism. In this case, the CTL response against the original epitope can be influenced by the presence of an altered epitope (peptide) with a small number of amino acid substitutions. The altered peptide prevents CTL lytic activity against cells expressing the original epitope. Reduced lytic activity has been observed in HIV (Klenerman et al. 1994) and hepatitis B virus (Bertoletti et al. 1994). Antagonism by APLs lowers the clearance rate of viruses carrying the original peptide, potentially allowing longer survival and greater success of the protected genotypes within infected individuals.

Gilbert et al. (1998) studied the effects of APLs on the distribution of strains in a population of Plasmodium falciparum infecting humans in The Gambia. They examined four variant peptides of the circumsporozoite protein. The peptides cp26 and cp29 have CTL epitopes that bind the MHC class I molecule HLA-B35, the most frequent MHC class I molecule in their study area. The other two peptides (cp27 and cp28) do not bind HLA-B35.

The octamers cp26 and cp29 differ only in a single amino acid substitution. Peptide cp26 antagonizes cp29-specific CTLs and cp29 antagonizes cp26-specific CTLs. Interference occurs even when the antagonist occurs in relatively low concentration and is presented on different cells from the partner epitope. In addition, in vitro studies of T cells from hosts unexposed to malaria show that cp26 and cp29 mutually interfere with induction of primary CTL responses (Gilbert et al. 1998). In vivo studies in mice also show the same mutual interference of primary CTL induction (Plebanski et al. 1999).

Mutual interference suggests that hosts jointly infected with cp26 and cp29 will be less effective in clearing parasites than singly infected hosts or hosts with other combinations of strains. Gilbert et al. (1998) found hosts jointly infected with cp26 and cp29 more frequently than expected from population frequencies, suggesting that mutual interference of CTL response favors joint survival and transmission of these strains.

It is clear that APLs sometimes reduce CTL response. In the case of Plasmodium falciparum, it appears that variant peptides interfere with immunity sufficiently to influence the distribution of antigenic variation in the parasite population. It is not clear at present whether APLs have a significant influence on the antigenic diversity of other parasites.

6.3. Sequence of Exposure to Antigens: Original Antigenic Sin

The host amplifies specific B and T cell lineages in response to challenge by foreign antigen. Often the host has several B and T cell specificities that match the various antigens of a parasite, but the host amplifies only a subset of matching specificities. I discussed in earlier sections various factors that influence immunodominance—the particular subset of antigens that stimulate an immune reaction from among the broader set of antigens that could potentially stimulate a response.

The sequence in which the host encounters antigenic variants influences the specificity of the immune response. The first observations of sequential effects were made on influenza infections (Francis 1953; Fazekas de St. Groth and Webster 1966a, 1966b). These authors called sequential effects original antigenic sin because the first antigenic exposure influenced response to later antigens.

Sequential effects in B and T cell responses can occur in various ways. Figure 6.1a shows the most commonly described pattern. Consider two variant epitopes, A and A′, at the same antigenic site. The specific immune response a against the original epitope A cross-reacts with the antigenic variant A′. If the host encounters A first, then secondary infection with A′ stimulates a secondary immune response, a, with relatively higher specificity for A and weaker specificity for A′ (original order). If the host encounters A′ first, then secondary infection with A stimulates a secondary response, a′, with higher specificity for A′ and weaker specificity for A (reversed order). This form of cross-reactive interference occurs in CTLs (Good et al. 1993; Klenerman and Zinkernagel 1998) and antibodies (Fazekas de St. Groth and Webster 1966a; Barry et al. 1999).

Figure 6.1. Three different patterns of original antigenic sin, as described in the text.

Figure 6.1

Three different patterns of original antigenic sin, as described in the text.

Figure 6.1b shows a second pattern of cross-reactive effects between variants at the same epitope. This case is similar to the first, in which sequential stimulation by A and then A′ causes a cross-reactive response a against secondary challenge by A′ (original order). However, primary stimulation by A′ does not elicit a response (reversed order). Thus, initial priming of cross-reactive memory cells by first exposure to A is required to generate a response to secondary challenge by A′. Fish et al. (1989) have demonstrated this pattern for antibody response.

The third pattern of sequential effects occurs when parasite challenge raises a specific immune response against several epitopes (fig. 6.1c). In this example, the first challenge with a pair of different epitopes, AB, raises a specific response ab against both epitopes. Secondary challenge by AB, in which epitope A is altered, yields a robust immune response b against the original variant B but only a weak or absent response against the modified epitope A′ (original order). Thus, a strong response against a constant epitope represses the response against the changed epitope. This pattern has been observed in sequential influenza infections (Janeway et al. 1999, p. 411).

It is not known how memory B or T cells reduce stimulation of naive clones during a secondary challenge. The rapid response from memory cells may keep parasite density below the threshold required to stimulate naive B or T cells. This would be a form of indirect repression mediated by the population dynamics of the parasite and the specific immune cells.

Alternatively, the memory cells may exert a more direct form of repression (Janeway et al. 1999, p. 410). For example, antigen bound to BCRs on naive B cells stimulates the B cells. But if the bound antigen also has a free antibody attached to it, the antibody interacts with the surface receptor FcγRIIB-1 on the naive B cell to repress activity of that naive B cell. By contrast, antibody bound to antigen-BCR complexes does not repress memory B cells.

6.4. Problems for Future Research

1. Molecular structure, binding kinetics, and competition between cellular lineages

Binding kinetics determine winners and losers in the competition between B cell lineages with different antibody specificities (Rao 1999). Equilibrium affinity dominates early in the competition, whereas on-rates dominate later during affinity maturation.

How can one study the biochemical and structural attributes that determine the binding kinetics of antibodies and epitopes? With regard to equilibrium affinity, one can compare structurally the different antibodies from the naive repertoire in relation to their success in binding a particular epitope and stimulating its B cell lineage. With regard to the shaping of on-rates, the hypermutation and selection during affinity maturation produce a lineal sequence of substitutions that enhances on-rates and perhaps also increases off-rates.

The contrast between the early selection of equilibrium affinity (on:off ratio) and the later selection of on-rate may provide insight into the structural features of binding that separately control on-rates and off-rates. This is a superb opportunity to relate structure to function via the kinetic processes that regulate the immune response.

2. Mathematical models and experimental perturbations

Immunodominance results from the kinetics of cellular lineages. Quantitative models help to develop hypotheses that can be tested by experimental perturbation. For example, Rao (1999) suggested that competition for helper T cells determines the expansion of B cell lineages. He tested this idea by manipulating the pool of helper T cells, and found that reducing the helper T cell pool did lower stimulation to B cell lineages.

Rao's quantitative model could be expanded into a mathematical analysis, with interactions between binding rates, pool sizes for different B and T cell lineages, and the rules of competition that determine which lineages succeed. Such a model presents clear hypotheses about the quantitative interactions that regulate immunity. Those hypotheses call attention to the sorts of experimental perturbations that should be informative.

3. Nowak's predator-prey model of competition

Nowak et al. (1995) developed a mathematical model of immunodominance. Their model focused on competition between immune cell lineages for stimulation by epitopes. Immune cells receive relatively stronger stimulation as their matching epitopes increase in numbers. The strongest immune-epitope match leads to the largest, immunodominant population of immune cells. That immunodominant lineage expands until its killing effect reduces the parasite population within the host down to a point of balance.

At that balance point, the parasite population stimulates division of the immunodominant population of immune cells just enough to match the tendency of the immune cell population to die off. In turn, the immunodominant immune cells reduce the parasites just enough to balance their births and deaths and hold the parasite population at a constant level. Other immune cell lineages receive weaker stimulation by the parasites because of their weaker binding characteristics to epitopes. Those subdominant lineages decline because the dominant lineage pushes parasite abundance down to the point where the weaker stimulation received by the subdominant lineages cannot overcome their tendency to decline.

The bottom line from this mathematical analysis matches the simplest, standard theory of predator-prey population dynamics: the most efficient predator reduces the prey down to a level where less efficient predators cannot survive.

Can such idealized mathematical models capture the complex molecular and kinetic details of the immune response? The answer depends on one's point of view. On the one hand, immunodominance is shaped in part by competition between lineages of immune cells, and thus the population dynamics of competition contribute in some way to the patterns of immunodominance. On the other hand, the model abstracts away many aspects of regulatory control, such as the role of helper T cells, the distinction between equilibrium binding affinity and kinetic on-rates of binding in different phases of the immune response, and structural properties that govern affinity and cross-reactivity.

The mathematical abstraction pays off as long as one understands the goal: to bring into sharp focus a hypothesis about how essential processes shape immunodominance. For example, Nowak et al.'s (1995) model predicts that essential properties of population dynamics and stimulatory thresholds matter. Some of the studies described in the text support this prediction. If one suspects that the distinction between equilibrium affinity and kinetic on-rates matters in an essential way for immunodominance, then an extended mathematical model would provide testable predictions about that aspect of the system.

I emphasize these issues here because the dynamics of immune cells and parasite populations within each infected host provide one of the few subjects that has been developed mathematically (Nowak and May 2000). The simple principles from those models do seem to be important, if only because the rules of population dynamics must play a key role in shaping how populations of immune cells and parasites interact.

One can, of course, make more specific mathematical models to predict the dynamics of particular parasites or the role of particular molecular mechanisms. Those specific models require empirical study of their specialized predictions. And that is exactly what we want: tests of clearly and logically formulated quantitative predictions.

4. Helper T cell epitopes in an antibody response

Helper T cells provide an important stimulus in the development of an antibody response. As B cells bind antigen to their BCR, they often pull the antigen into the cell. The B cells process protein antigens into small peptides, bind those peptides to MHC class II molecules, and present the peptide-MHC complexes on their cell surfaces. Helper T cells with matching specificity in their TCRs bind the peptide-MHC complexes and stimulate the B cells. Thus, an antigen must have two epitopes to stimulate a robust B cell response with affinity maturation. One epitope binds the BCR, and a second must survive digestion and be presented on the B cell surface bound to a class II molecule for the TCR of a helper T cell.

Several factors likely affect the degree to which helper T cell epitopes modulate the immunodominance of B cell epitopes. These factors include the proximity of the two epitopes, the binding kinetics of the T cell epitope to the TCR, the nature of the helper T cell signal that provides stimulation to the B cell, and the population dynamics of the helper T cell lineages with different TCR specificities. For example, Shirai et al. (1999) showed that the proximity of a helper T cell epitope to a B cell epitope can influence development of the antibody response. In particular, a helper T cell epitope near the hypervariable region of the hepatitis C virus envelope gene aids in generation of antibodies to the hypervariable region.

5. CTL versus antibody response

Antibody attack favors antigenic variation in parasites' surface molecules. By contrast, CTLs favor variation in any parasite molecule that can be presented by the host's MHC system. The balance of antibody versus CTL defense affects the population dynamics of the parasites within the host, the time before clearance, and the memory properties of host immunity against reinfection (Seder and Hill 2000). The factors that tip an immune response toward antibody, CTL, or a mixture of the two are not fully understood (Constant and Bottomly 1997; Power 2000). Studies of model systems sometimes show a sharp dichotomy between CTL and antibody response controlled by a simple variable such as antigen dosage (Menon and Bretscher 1998). But the immune response to many viruses includes robust antibody and CTL attack (Knipe and Howley 2001).

As more parasite genomes are sequenced, it may be useful to look at which potential antigenic sites do in fact show significant variation. Those highly variable sites can be studied to determine if they are CTL or antibody epitopes, providing clues about which type of immunity imposes the strongest selective pressure on the parasite.

Copyright © 2002, Steven A Frank.
Bookshelf ID: NBK2386

Views

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...