What else can the immune system recognize?

One of the earliest challenges to our understanding of the molecular basis of the immune system was the system’s apparent unlimited capacity to bind essentially any chemical species, natural or synthetic, that it encounters (1–3). Ideas of an “instructional” mechanism of attaining this broad spectrum of binding specificity (1, 2) have later on been excluded and replaced by an evolutionary selection process. More than 50 years ago, Karl Landsteiner (3), the pioneer of immunochemistry, already had recognized this feature. This observation later led to the conclusion that the immune system evolved the capacity to generate an adoptive repertoire of binding sites from which exposure to a given antigen will select the specific ones. Recent years of progress in chemistry now have presented the recognition spectrum of the immune system with both novel and more esoteric chemical entities. Specifically, can one raise antibodies that bind elementary carbon in its recently discovered new allotropic form, fullerene? As detailed in a previous issue of the Proceedings, Chen et al. (4) have managed to raise IgG class polyclonal antibodies specific to the 60-carbon-atom compound named Buckminsterfullerene. This achievement of immunochemists, though perhaps not surprising, is still interesting and may underscore some less-attended facets of the interactions that take place within the antigen combining sites of antibodies.

Fullerenes have generated considerable interest for their unique structural and chemical properties. Their possible applications in biology and medicine (5) require methods for specific assays of their concentrations, which was a major motivation for Chen et al.’s study (4). One of the first problems encountered en route to raising antibodies to this relatively inert compound was the preparation of its covalent conjugates that suit the requirements of animal immunization. During the past decade, many different approaches have been used for preparation of water-soluble fullerene derivatives. These include fullerene complexes with cyclodextrin (6), liposomes (7), and, most relevant, protein-conjugated-fullerene (8) (fullereyl protein?), all of which were shown to be water soluble. Chen et al. successfully prepared water-soluble fullerene derivatives of several proteins and peptides that were used for the immunization as well as for the isolation and characterization of the antibodies. It therefore would be interesting to test the interaction of the currently prepared antibodies with the other water-soluble fullerene derivatives, because they may have different modes of presenting the fullerene molecule. Of particular interest, are fullerene-enriched liposomes that could be useful for raising mAbs. Indeed, Alving et al. (9, 10) already have developed and used such complexes to raise cholesterol-specific mAbs. Cholesterol, also a highly hydrophobic molecule with very low water solubility, was presented in liposomes without being derivatized to a protein carrier.

Fullerenes are highly ordered and symmetric molecules, with a structure known at the atomic resolution (11). It therefore is worthwhile to compare the capacity of vertebrate immune system to respond to other water-insoluble, ordered antigens. Water-insoluble crystals were found to be treated as antigens and when introduced into experimental animals are inducing specific antibodies. This property of the immune system has been recently illustrated by studies of Kessler et al. (12), who raised and selected mAbs specific for crystals with well-defined structures at the molecular level of 1,4-dinitrobenzene. Significantly, none of these antibodies bound to the 1,4-dinitrophenyl hapten when it was conjugated to a protein carrier. An antibody interacting with a crystal may recognize a particular set of molecular components exposed on a specific crystal surface. From the structure of various antibody-antigen complexes that have been determined by x-ray crystallography, the antibody binding site typically spans an area of 600–900 Ų (13). For example, the influenza virus neuraminidase epitope that is recognized by mAb NC41 (14) has an antigen contact area of 855 Ų. An antibody therefore can bind to arrays of 5–20 molecules exposed at the crystal surface. Like antigens that carry ordered multiple epitopes, crystals expose chemically and geometrically distinct surfaces. Thus, it is conceivable that different antibodies may recognize distinct faces of a crystal, and this type of interaction is expected to be similar to that of antibodies for repetitive epitopes present on protein surfaces. Among several mAbs that have been selected against crystals of 1,4-dinitrobenzene, one was shown to specifically interact with the molecularly flat, aromatic, and polar {101̄} face of these crystals, rather than other faces of the same crystal. The high preference for a given crystal face was revealed by immunolabeling techniques and analyzed by molecular modeling of the antibody binding site (15). The model predicted for the above binding site showed striking complementarity between the amino acid side chains forming the contact region and the structure of the crystal surface that is recognized.

Analogously, monoclonal antibodies also have been elicited against cholesterol monohydrate crystals (16). One of these antibodies was shown to specifically recognize the stepped {301} face of the cholesterol monohydrate crystals. Interestingly, at the surface of this face, the hydrophobic cholesterol hydrocarbon backbone is exposed on one side of the molecular steps, whereas hydroxyl residues and attached water molecules are exclusively exposed on the other side of the steps. These studies further illustrated the capacity of antibodies to recognize three-dimensional repetitive patterns probably like those present for example on viral coats. It is noteworthy that all the crystal-specific antibodies that were raised so far to crystals were found to belong to the IgM class. This finding is in accord with the assumption that, unlike most commonly used antigens, crystals cannot be processed by the antigen presenting cells and that the antibodies therefore are induced through a T cell-independent path (17). In contrast, the observation that the fullerene-specific antibody population raised by Chen et al. (4) is of the IgG isotype is compatible with the fact that processing-susceptible protein-hapten conjugates were used to induce the immune response. It is likely that immunization with fullerene complexes with liposomes or cyclodextrin also would yield IgM class mAbs, as indicated by Alving et al.’s experience with cholesterol immunization (9, 10).

Though Chen et al. (4) indicate that the potential practical applications of fullerenes as biological or pharmaceutical agents motivated their attempts to raise specific antibodies, intellectual curiosity to interrogate how far the repertoire of antigen binding sites can go also was a driving force. Having raised the specific antibodies to fullerenes, a list of rationales for the mode of binding of the fullerenes to the antibodies is presented. Starting with the first and obvious binding energy, the hydrophobic fullerene composed of only carbon atoms may evade the aqueous phase, displace water from and fit a similarly hydrophobic amino acids-lined surface of the antigen combining site. Are there also contributions to the binding energy from π-orbitals stacking and charge-transfer complex formation? These problems will have to be investigated when fullerene-specific mAbs are produced. It also will be interesting to examine the possible operation of conformational changes in the binding site leading to an optimal induced fit to the rather rigid epitope (18, 19). A point of general interest and perhaps relevant to the ongoing efforts to raise antibodies to crystals is the fact that the actual size of the C₆₀ fullerene is just 7.2 Å in diameter. It is a relatively small molecule comparable in size to other, more conventional hydrophobic haptens. Chen et al. correctly point out that one therefore can expect a similarity between binding sites of fullerenes and the very well-characterized mAbs raised to hydrophobic haptens like progesterone. Still, the shape of a high-affinity binding site may be very different from the putative scheme drawn by Chen et al. To attain maximal contact and thereby optimize binding, the complementarity-determining residues (CDR) may fold to form a deeper cavity that surrounds the spherically shaped fullerenes. A flat binding site where the CDRs provide only superficial contacts, similar to that assumed for the crystal-specific antibody binding site, is less probable as it may yield rather low affinity. Finally, a fundamental difference between the two haptens has to be stressed; namely, while progesterone and analogous haptens present faces of hydrocarbons, the former is an all carbon one. Determining the actual contact residues forming such a binding site therefore may be an illuminating step toward understanding in atomic details the way immunorecognition is achieved.