The historical definition of an antigen, given for many years in immunology textbooks, is a rather circular one, because it describes the antigen as a foreign substance that induces upon penetration in an animal (or a human being) the production of an antibody with which it will combine specifically, in vivo or in vitro. This classical antibody response is a characteristic of vertebrates. Although not basically wrong, this definition necessitates some comments addressing three points: (1) antibody, (2) induction, and (3) foreignness.
1. Antibody
When this definition was proposed, the immune response was characterized solely by the production of antibodies. When it was realized, in the 1960, that the specific functions of the immune system relied on two distinct types of lymphocytes, B cells and T cells, it became clear that the immune response could no longer be limited to the production of antibodies. B and T cells "recognize" the antigen, but in very different ways. B cells express immunoglobulins at their surface, whereas T cells express the so-called T-cell receptor (TCR). By extension, surface immunoglobulins are also given the name of B-cell receptor (BCR), by analogy with the TCR. Once stimulated by an antigen, in combination with other signals (cytokines), the B cells differentiate into plasma cells that will express a soluble form of immunoglobulins, also known as antibodies or circulating antibodies, because they are free in the bloodstream. Immunoglobulins and TCR are constructed on the same general pattern, but are encoded by discrete sets of genes. They interact with the antigen in very different manners. The immunoglobulins—or antibodies—interact directly with a nativeform of the antigen, through a small portion of this antigen called the antigenic determinant or epitope. The TCR does not recognize the antigen as a whole, but interacts witha fragment of it, which results from antigen processing and presentation. This has been well studied for protein antigens, which are cleaved into peptides (the processing step) in the so-called antigen-presenting cells (macrophages, dendritic cells, etc.), which will then reexpress these peptides at their surface, in close association with molecules encoded by genes of the major histocompatibility complex (MHC) (presentation step). Finally, the TCR will bind to both the peptide and the MHC-presenting molecule. The peptide derived from the antigen and that interacts with the corresponding TCR is describedas the "T-epitope."
2. Induction
For a long period of time, in fact until it became possible to work on the chemistry of antibodies in the beginning of the 1960, immunochemists centered their interest on the most accessible partner of the antigen-antibody complex—that is, the antigen, for which simple models of well-defined structure could be worked out. Although somewhat frustrating for a physiologist (this would be like an enzymologist exclusively studying substrates), this approach shed light on the exquisite specificity of recognition by the immune system, thanks mostly to the pioneering work of Landsteiner (1) and the discovery of haptens. Haptens are small molecules that are unable to stimulate the immune system, but may bind to an antibody. To produce this antibody, the hapten had to be conjugated to a protein that served as a carrier, conferring the property of stimulating the immune system. This observation clearly indicated that induction and recognition were two separate properties of an antigen. To clarify this, it was proposed to distinguish antigenicity from immunogenicity and, consequently, an antigen from an immunogen. Antigenicity describes the chemical structures that condition interaction with an antibody or with a TCR, whereas immunogenicity defines the properties of a molecule to induce an immune response. This distinction is far from pure semantics. A protein, considered as a typical "natural" immunogen, has the dual characteristics of a carrier, which is involved in processing and presentation, and of a hapten (in fact a mosaic of haptens) represented by the B epitope. This duality is of central importance for the immune system, in that it implies a close cooperation between B, T, and antigen-presenting cells. In most cases, an antibody response requires cooperation with the T-cell compartment, leading to the distinction of T-dependent antigens, as opposed to T-independent ones, which can stimulate B cells directly without this T-cell "help."
3. Foreignness
At the end of the nineteenth century, the first antigens that were identified were pathogens: bacteria, viruses, parasites, or toxins of various origins. It was soon realized that pathogenicity and immunogenicity were not linked and that a huge variety of cells or molecules could induce an immune response, providing that they were foreign to the animal it was injected into. Furthermore, this requirement of foreignness was strengthened by the famous aphorism "horror autotoxicus" put forward in 1901 by Ehrlich, to indicate that the immune system could not be stimulated by self-components, leading to the concept that it was a master of self-nonself discrimination. Due to independent observations of Grubb, Oudin, and Kunkel (2-4), it became apparent that this distinction between self and nonself was not as clear as initially thought. Oudin described three types of antigenic specificities, termed isotypy, allotypy, and idiotypy. Isotypy refers to those specificities that are characteristic of one molecule of one given animal species, for instance the mouse serum albumin. An antibody raised against albumin of one given mouse will react with the albumin of anymouse. Allotypy defines antigenic specificities that are shared by a group of individuals within a given species, as initially shown by Oudin for the rabbit immunoglobulins. This is the case of the human blood groups, for which people of, say, group A share this specificity. Allotypy simply reflects the existence of epitopes (or allotopes) that are encoded by allelic variants (see Alloantibody, Alloantigen). The last type of specificity, idiotypy, is more subtle, and is inherent to the molecules of the immune system itself. It was initially described by Oudin as the characteristics of one antibody molecule, synthesized by one given animal and specific for one given antigen. So, if a rabbit is given antigen X, it will produce an anti-X antibody (or Ab1). If a second rabbit, expressing the same immunoglobulin allotypes, is immunized against the anti-X antibody, it will produce an anti-anti-X antibody (or Ab2). Ab1 is an idiotype, Ab2 an anti-idiotype, and the antigenic specificities recognized on Ab1 by Ab2 are called the idiotopes. Because it was later shown that Ab1 and Ab2 could be produced as discrete waves within the same animal, it was clearly realized that the self-nonself distinction was not so obvious. The horror autotoxicus dogma had also to be revisited when it was shown by Avrameas, and extended by Coutinho (5), that a low level of natural autoantibodies was consistently present in every individual, as well as autoreactive T-cell clones. Such natural antibodies have no pathogenic effect,as opposed to those identified in certain autoimmune diseases, for reasons that are still not completely understood.
The term antigen covers a huge number of structures, from cells and macromolecular complexes, to relatively small molecules, the lower limit in molecular weight being of the order of a few thousand. Ultimately, and whatever the source of the materials considered, proteins and, to a lesser extent, polysaccharides are the main antigens. Proteins are representative of the T-dependent antigens, which implies processing and presentation. Depending upon their size, the potential number of discrete epitopes of one given antigen may vary from a very small number to several tens, in roughly linear proportion with the exposed area of the molecule. Polysaccharides behave most frequently as T-independent antigens. They are frequently constituents of the bacterial surface and are thus important for the preparation of vaccines against these microorganisms. Another example of polysaccharide antigens is given by the major blood groups in humans. Some macromolecular complexes, such as bacterial lipopolysaccharides, have been studied extensively because they also behave as polyclonal potent mitogenic activators of B cells. Nucleic acids are considered poor immunogens, although they clearly constitute a target for autoantibodies in systemic lupus erythematosus, a severe autoimmune disease. More recently, DNA has been tentatively used as a vaccine, in the form of expression vectors encoding a protein that acted secondarily as an immunizing antigen.
Detailed analysis of the antigenic determinants or epitopes of natural antigens is a difficult task and has necessitated the use of models, among which the most extensively studied in the 1960s and 1970s were polysaccharides by the group of Kabat at Columbia University (6). Based on the binding inhibition of dextran (a polymer of glucose) to anti-dextran antibodies by oligosaccharides of various lengths, the range of size of an epitope, and hence that of the corresponding antibody combining site, was estimated to be between 3 and 6 monosaccharide units. A similar approach was taken by the group of Sela at the Weizmann Institute, with synthetic peptides that mimicked the structure of natural proteins. Direct analysis of natural peptide epitopes with the corresponding antibody was made much later by X-ray crystallography and gave a clear picture of the organization of the antibody combining site. The central message is that there is a huge diversity in size and form of natural epitopes. At the surface of a protein, the number of amino acid residues that interact with the antibody combining site varies from a few to about 20. The variation in both size and number of epitopes at the antigen surface contribute a large part of the heterogeneity of the immune response.
