The repertoire of antibodies to a single antigenic determinant

Structural biology of moonlighting: lessons from antibodies

Protein moonlighting is the property of a number of proteins to have more than one function. However, the definition of moonlighting is somewhat imprecise with different interpretations of the phenomenon. True moonlighting occurs when an individual evolutionary protein domain has one well-accepted role and a secondary unrelated function. The 'function' of a protein domain can be defined at different levels. For example, although the function of an antibody variable fragment (Fv) could be described as 'binding', a more detailed definition would also specify the molecule to which the Fv region binds. Using this detailed definition, antibodies as a family are consummate moonlighters. However, individual antibodies do not moonlight; the multiple functions they exhibit (first binding a molecule and second triggering the immune response) are encoded in different domains and, in any case, are related in the sense that they are a part of what an antibody needs to do. Nonetheless, antibodies provide interesting lessons on the ability of proteins to evolve binding functions. Remarkably similar antibody sequences can bind completely different antigens, suggesting that evolving the ability to bind a protein can result from very subtle sequence changes.

N-glycan Cryptic Antigens as Active Immunological Targets in Prostate Cancer Patients

Although tumor-associated abnormal glycosylation has been recognized for decades, information regarding host recognition of the evolving tumor glycome remains elusive. We report here a carbohydrate microarray analysis of a number of tumor-associated carbohydrates for their serum antibody reactivities and potential immunogenicity in humans. These are the precursors, cores and internal sequences of N-glycans. They are usually masked by other sugar moieties and belong to a class of glyco-antigens that are normally “cryptic”. However, viral expression of these carbohydrates may trigger host immune responses. For examples, HIV-1 and SARS-CoV display Man9 clusters and tri- or multi-antennary type II (Galβ1→4GlcNAc) chains (Tri/m-II), respectively; viral neutralizing antibodies often target these sugar moieties. We asked, therefore, whether prostate tumor expression of corresponding carbohydrates triggers antibody responses in vivo. Using carbohydrate microarrays, we analyzed a panel of human sera, including 17 samples from prostate cancer patients and 12 from men with Benign Prostatic Hyperplasia (BPH). We observed that IgG antibodies targeting the Man9- or Tri-/m-II-autoantigens are readily detectable in the sera of men with BPH, as well as those with cancer. Importantly, these antibody activities were selectively increased in prostate cancer patients. Thus, human immune systems actively recognize these N-glycan cryptic carbohydrates and produce targeting antibodies. This finding shads a light on a class of previously less studied immunological targets of human cancers. Identifying the diagnostic, prognostic and therapeutic values of these targets will require further investigation.

Evidence of immune tolerance to blood group antigens in a case of ABO-incompatible pediatric liver transplantation

In a 12-year-old patient with blood group O, who had received a partial liver graft 10 years ago from his father with blood group A, the levels of anti-A-specific antibodies (Abs) were persistently undetectable after the transplantation, while the levels of anti-B-specific Abs gradually increased and attained a plateau. Peripheral blood mononuclear cells (PBMCs) from this patient were engrafted into NOD/SCID mouse in order to investigate the immune response to donor-type blood group antigens. Even after sensitization with blood group-A erythrocytes, no anti-A Abs were detected in the serum samples of the mouse that received PBMCs from the blood group-O recipient of group-A liver allograft, however, immunoglobulins specific for antigens other than the A antigens were produced. Thus, we provide a possible evidence of immune tolerance to blood group antigens in this ABO-incompatible pediatric liver transplantation.

High-Throughput Carbohydrate Microarray Technology

One of our long-term interests is to explore the immunogenic sugar moieties that are important for “self-” and “nonself” discrimination and host immune responses. We have established a highthroughput platform of carbohydrate microarrays to facilitate these investigations. Using this technology, carbohydrate-containing macromolecules of distinct structural configurations, including polysaccharides, natural glycoconjugates, and mono- and oligosaccharides coupled to lipid, polyacrylamide, and protein carriers, have been tested for microarray construction without further chemical modification. Here, we discuss issues related to the establishment of this technology and areas that are highly promising for its application. We also provide an example to illustrate that the carbohydrate microarray is a discovery tool; it is particularly useful for identifying immunological sugar moieties, including differentially expressed complex carbohydrates of cancer cells and stem cells as well as sugar signatures of previously unrecognized microbial pathogens.

Role of immunogen design in induction of soman-specific monoclonal antibodies

The study of monoclonal antibodies raised against defined hapten epitopes has been a useful approach to understanding antibody repertoire. The situation in which antibodies are raised against different epitopes of the same hapten but have some common recognition or binding features has been less frequently examined. To explore the latter situation, we have characterized three monoclonal antibodies previously raised against two structurally different epitopes of the same organophosphorus nerve agent hapten, pinacolymethyl phosphonofluoridate (soman). Two antibodies, BE2-IA10 (BE2) and CC1-IIA4 (CC1), raised against the hydrophobic pinacolyl motif of soman, bind exclusively to soman and not to any other organophosphorus nerve agents. We determined that these antibodies have the same heavy chain sequence, which they share with the unrelated antibodies MOPC 21 and H17-L19. While all these antibodies share the same heavy chain sequence, they each possess different light chain sequences. Binding studies revealed that each of these antibodies has a unique reactivity with a panel of structurally related ligands, suggesting that the light chains are critically important in determining specificity in these antibodies. The third antibody, #2.ID8.2, raised against the methyl phosphoryl portion of soman, has unique heavy and light chain sequences. This antibody binds to all the currently identified chemical warfare agents. Given that the presenting epitope used to induce #2.ID8.2 is common to sarin, soman, tabun and VX, the ability of this antibody to recognize each of these haptens versus the inability of BE2 or CC1 to do so demonstrates the important role that immunogen design can play in the specificity of an antibody response.

NOD/SCID mice engrafted with human peripheral blood lymphocytes can be a model for investigating B cells responding to blood group A carbohydrate determinant

Human antibodies (Abs) against blood group A or B carbohydrate determinant are a major barrier to ABO-incompatible organ transplantation; however, the phenotype and other properties of B cell types responding to A or B carbohydrate epitopes have not been defined. Studies here, which use fluorescein-labeled synthetic A determinant (GalNAcalpha1-3Fucalpha1-2Gal), demonstrate that B cells bearing surface IgM (sIgM) receptors recognizing blood group A carbohydrate determinant are found exclusively in a small B cell subpopulation, i.e. sIgM+ CD11b+ CD5+ B1 cells, in blood group O human peripheral blood mononuclear cells (PBMC). In order to test anti-A Abs producing capacity of the human PBMC, nonobese diabetic (NOD)/severe combined immune-deficient (SCID) mice that have been treated with rabbit anti-asialo GM1 serum to deplete natural killer cells and with 3 Gy of whole body irradiation were engrafted with blood group O or A human PBMC, followed by sensitization of human blood group A red blood cells. Anti-A-specific human Abs were detected in the sera of the mice that received blood group O human PBMC, whereas they were not detected in the sera of the mice that received blood group A human PBMC, indicating profound tolerance of auto-reactive B cells. The human PBMC-NOD/SCID chimera developed by injection of blood group O human PBMC might be a useful in vivo model to test effects of immunosuppressants or other approaches on human B cells that respond to blood group A antigens.

Carbohydrate microarrays for the recognition of cross-reactive molecular markers of microbes and host cells

We describe here the development of a carbohydrate-based microarray to extend the scope of biomedical research on carbohydrate-mediated molecular recognition and anti-infection responses. We have demonstrated that microbial polysaccharides can be immobilized on a surface-modified glass slide without chemical conjugation. With this procedure, a large repertoire of microbial antigens (approximately 20,000 spots) can be patterned on a single micro-glass slide, reaching the capacity to include most common pathogens. Glycoconjugates of different structural characteristics are shown here to be applicable for microarray fabrication, extending the repertoires of diversity and complexity of carbohydrate microarrays. The printed microarrays can be air-dried and stably stored at room temperature for long periods of time. In addition, the system is highly sensitive, allowing simultaneous detection of a broad spectrum of antibody specificities with as little as a few microliters of serum specimen. Finally, the potential of carbohydrate microarrays is demonstrated by the discovery of previously undescribed cellular markers, Dex-Ids.

Molecular characterization of a monoclonal murine anti-blood group A antibody

A mouse monoclonal anti-human blood group A antigen (AC12, mu, kappa) has been generated and sequenced in order to analyze the immunoglobulin genes used to generate antibodies with anti-human blood group A specificity. Mice were immunized with human type A RBC. Anti-A producing hybridomas were detected by agglutination against human type A RBC. Total cellular RNA was extracted from hybridomas cells. PCR amplification and sequencing of anti-A heavy and light chain cDNAs were performed. The VH and VK sequences of antibody AC12 were shown to be very homologous to that used by other antibodies recognizing carbohydrates as well as glycoproteins, peptides or haptens constituting self antigens as well as nonself antigens. The VH sequence of antibody AC12 presented important homology with a previously reported monoclonal anti-blood group B antibody. The antibody AC12 also presented homology with the VH and VK sequences of a previously reported human anti-blood group A antibody which contributes additional evidence in favor of a restricted usage of V segments by antibodies directed against red blood antigens.

T-Independent IgA Responses to Microbial Polysaccharides

There is accumulating evidence indicating the presence in vivo of T-independent routes of IgA response in addition to the conventional T-dependent IgA response. Factors influencing these alternative pathways of IgA responses may include the structural characteristics of a stimulating antigen, the nature of responding B cells, and the microenvironment. The structural complexity of polysaccharide antigens has made it difficult to summarize a general scheme for the antibody responses they induce. Instead, one may expect that each individual polysaccharide may be able to create a unique microenvironment by activation of specific cell populations in the repertoires of non-T cell types. A specific pattern of B cell response may thus be elicited by TI stimulation. Recognition of such a unique property of a TI antigens is necessary for us to better understand the T-independent IgA response. Information obtained may have an impact on the development of vaccination strategies directed at the mucosal immunity mediated by IgA antibodies.

Binding specificity of hydrolytic catalytic antibody: lessons from the diverse immune system

The diverse immune system has the potential to provide numerous antibodies that catalyze a great variety of reactions and exhibit diversity in mechanism and substrate specificity. However, to find such an antibody is not practical, unless we understand the immune response precisely and utilize it. We applied an alternative screening methodology based on the binding specificity of antibodies for the structure of the transition-state analog p-nitrophenyl phosphonate in order to study the built-in diversity of the immune response. This study uncovered the unpredictable binding and substrate specificity of catalytic antibodies derived from the diverse immune response.

Improving the binding affinity of an antibody using molecular modeling and site-directed mutagenesis

Activated Factor X releases F1.2, a 271-amino acid peptide, from the amino terminus of prothrombin during blood coagulation. A nine-amino acid peptide, C9 (DSDRAIEGR), corresponding to the carboxyl terminus of F1.2 was synthesized and used to produce a monoclonal antibody, TA1 (K(D)) 1.22 x 10(-6) M). To model the TA1 antibody, we entered the sequence information of the cloned TA1 Fv into the antibody modeling program, ABM, which combines homology methods, conformational search procedures, and energy screening and has proved to be a reliable and reproducible antibody modeling method. Using a novel protein fusion procedure, we expressed the C9 peptide fused to the carboxyl terminus of the PENI repressor protein from Bacillus licheniformis in Escherichia coli. We constructed fusion proteins containing alanine substitutions for each amino acid in the C9 epitope. Binding studies, using the C9 alanine mutants and TA1, and spatial constraints predicted by the modeled TA1 binding cleft enabled us to establish a plausible conformation for C9 complexed with TA1. Furthermore, based on binding results of conservative amino acid substitutions in C9 and mutations in the antibody, we were able to refine the complex model and identify antibody mutations that would improve binding affinity.

Partial analysis of the flagellar antigenic determinant recognized by a monoclonal antibody to Clostridium tyrobutyricum

In order to count Clostridium tyrobutyricum spores in milk after membrane filtration, murine 21E7-B12 monoclonal antibody was produced. Elution of the monoclonal antibody from this antigen, the flagellar filament protein, by carbohydrate ligands was used to study the epitope structure. A competitive elution of an anti-dextran monoclonal antibody by carbohydrate ligands served as a control in order to validate the immunological tool applied to flagellin epitope study. The carbohydrate moiety of flagellin contained D-glucose and N-acetyl-glucosamine in a molar ration of 11:1 as determined by gas-liquid chromatography and 2 low-abundancy unidentified compounds. In ELISA, D-glucose and N-acetyl-glucosamine did not dissociate the antibody-flagellin complex contrary to maltose, maltotriose, maltotetraose and maltopentaose. The efficiency of elution increased from the dimer to the pentamer and became nil for maltohexaose and maltoheptaose. The fact that the hexamer and heptamer could not react with the 21E7-B12 monoclonal antibody could be explained by a drastic conformational change. The over-all stretched maltopentaose switch to a helical-shaped maltoheptaose which could not fit the 21E7-B12 monoclonal antibody antigen-combining site. Thus, flagellin epitope may contain alpha (1-->4) linked glucose residues plus either N-actyl-glucosamine or an unidentified compound that maintain it in an extended shape.

The Escherichia coli-derived Fab fragment of the IgM/kappa antibody IN-1 recognizes and neutralizes myelin-associated inhibitors of neurite growth

A recombinant Fab fragment was prepared from the monoclonal IgM/kappa antibody IN-1, which neutralizes central nervous system myelin-associated neurite growth inhibitors both in vitro and in vivo. The variable domain gene sequences were amplified and cloned after cDNA synthesis from the hybridoma RNA. After insertion into the tet promoter vector pASK85, which provided the constant domains of class IgG1/kappa, equipped with a His6 tag, large amounts of the Fab fragment were produced in Escherichia coli by medium cell density fermentation. The Fab fragment was purified to homogeneity by immobilized metal-affinity chromatography and its biochemical activity was compared with the original IN-1 antibody. In an assay for neurite outgrowth and fibroblast spreading, the Fab fragment showed a similar neutralizing effect on inhibitory substrate properties of central nervous system myelin as the unpurified IgM, although an approximately tenfold higher concentration was necessary. Immunoprecipitation experiments revealed a more selective antigen-binding behaviour for the Fab fragment. The Fab fragment was also successfully applied for antigen detection in immunohistochemical analyses. Therefore, the recombinant Fab fragment of IN-1 shows full functionality in vitro and appears to be well suited for replacing the monoclonal IgM in investigations on fiber tract regeneration in vivo.

Structural diversity of monoclonal CD4 antibodies and their capacity to block the HIV gp120/CD4 interaction

A number of monoclonal antibodies have been raised against CD4, the receptor on T cells for the HIV envelope glycoprotein gp120. In the present paper we describe biological activities and sequence analysis of seven CD4 MAb. Five of these MAb preparations compete with HIV/gp120 for CD4 binding. The sequences of the variable regions for these MAb were determined in order to ascertain any correlation with selective V gene usage. A relationship was found between the expressed variable region genes and the CD4 recognition pattern. The VH genes that are used can be subdivided into two major groups expressing either a VH gene belonging to the J558 family or to the VGam family. The usage of the VL genes varies, indicating that the epitope specificity is predominantly determined by the rearranged VH genes. The distinct cross-reactivity pattern of these MAb also correlates with their capacity to block binding of recombinant gp120 to CD4 in vitro. Although five of these MAb were able to block gp120 binding none of the CDR sequences shows a relevant homology to the gp120 sequence. This indicates a steric hinderence mechanism for blocking gp120 binding and not a direct interaction with the receptor binding site on CD4. The data also confirm the failure of these MAb as a potential target for receptor mimicry.

Temperature and pH dependence of fluorescein binding within the monoclonal antibody 9-40 active site as monitored by hydrostatic pressure

In a comparative study, the thermodynamic parameter, DeltaV, was obtained using hydrostatic pressure-induced dissociation of fluorescein (Fl) from the active site of monoclonal antibody (mAb) 9-40 and its mutant and native derivatives equilibrated at six pH values (8.0, 7.5, 7.0, 6.5, 6.0, and 5.5) and four temperatures (35, 25, 15, and 5 degrees C). mAb 9-40 and its Fab and single-chain Fv (scFv) derivatives at pH 8.0 were found to have identical Fl dissociation behavior under pressure as a function of temperature. The pressure dissociation at 25 degrees C as a function of pH showed a sigmoidal dependence of DeltaV with a midpoint value at pH 7.4 for mAb 9-40. Comparison of experimental results for scFv 9-40/212 with its mutant scFv 9-40/212Arg-34L indicated that the pH dependence of mAb 9-40 was due to the titration of His-34L in the active site. Iodide quenching of bound Fl showed that the hapten in this active site was solvent accessible. Imperfect packing, which leads to increased conformational dynamics, was determined as a possible cause of the low affinity for mAb 9-40.

Human and mouse monoclonal antibodies to blood group A substance, which are nearly identical immunochemically, use radically different primary sequences

A human monoclonal antibody (HuA) specific for blood group A substance with two fucose groups was found to be immunochemically almost identical with that of a previously characterized mouse monoclonal anti-A, AC-1001. The VH and VL chain cDNAs of HuA were sequenced and compared with those of AC-1001. The human and mouse antibodies used VH and Vk genes that came from different families and shared minimal nucleotide and amino acid sequence identity. Thus, two antibodies from two different species can use evolutionarily unrelated sequences to bind the same carbohydrate epitope. The cloned HuA VH and VL genes were then transfected into a mouse myeloma cell line and re-expressed, together, and each separately with an irrelevant VH or VL. Only the original HuA VH and Vk had anti-A activity, demonstrating that both the heavy and light chains contributed to specificity.

Sequences of cDNAs encoding immunoglobulin heavy- and light-chain variable regions from two anti-dioxin monoclonal antibodies

We report the cDNA cloning of the gamma- and kappa-chain-encoding genes for two mouse monoclonal antibodies (mAb) which recognize dioxins. The nucleotide sequences encoding the variable regions of these mAb were also determined. Although the mAb have similar dioxin-binding characteristics, their deduced variable region amino-acid sequences are very different.

Type 2 T-cell-independent murine immune response to the human AB0 blood group antigens

The human AB0 blood group determinants are simple carbohydrate structures which are widely distributed in nature. Much work has been done on the structure of the A and B antigens but little is known on the regulation of anti-A and anti-B immune responses. To develop a model system, we have characterized the AB0 immunity of normal Balb/c mice and found a significant level of serum natural anti-A but almost no anti-B. This finding and the known IgM predominance among immune anti-A produced in the Balb/c mouse indicate that the AB0 immunity of this mouse strain is comparable to the one of human blood group B individuals. Following immunization with human red blood cells, similar levels of anti-A were produced in normal and athymic Balb/c mice showing that the anti-A response is T cell independent. Furthermore, no anti-A or anti-B antibodies were produced in CBA/xid mice indicating a type 2 T-cell-independent immune response. These results may contribute to a better understanding of human AB0 responses and establish the mouse as a suitable model to study the immunobiology of AB0 antigens.

Antibody design: beyond the natural limits

Dissection of antibody-antigen interactions requires a knowledge of antibody structure, the ability to model accurately the conformation of antibody-combining sites, and an understanding of the energetic factors governing the interactions. When this understanding has reached the point where the molecular shape and chemical character of a combining site necessary to define a particular specificity and binding requirement can be designed, the antibody repertoire will have been extended 'beyond the natural limits'.

Reaction of germinal centers in the T-cell-independent response to the bacterial polysaccharide alpha(1-->6)dextran

Primary immunization of BALB/c mice with alpha(1-->6)dextran (DEX), a native bacterial polysaccharide, induces an unexpected pattern of splenic B-cell responses. After a peak of antibody-secreting B-cell response at day 4, deposition of dextran-anti-dextran immune complexes, as revealed by staining with both dextran and antibodies to dextran, occurs and persists in splenic follicles until at least the fourth week after immunization. Antigen-specific B cells appear and proliferate in such follicles, leading by day 11 to development of DEX-specific germinal centers as characterized by the presence of distinct regions of DEX+ peanut agglutinin-positive (PNA+) cells. At this time, fluorescence-activated cell sorter analysis also reveals the appearance of a distinct population of DEX+ PNA+ splenic B cells. In contrast, DEX+ PNA- cells, characterized by intense cytoplasmic staining, are present outside of splenic follicles, peak at day 4 to day 5, and persist until at least day 28. The frequency of these cells correlates with DEX-specific antibody-secreting cells, as detected by the ELISA-spot assay. Thus, in addition to the expected plasma cellular response, the typical T-cell-independent type II antigen, DEX, surprisingly also elicits the formation of antigen-specific germinal centers. These observations raise fundamental questions about the roles of germinal centers in T-cell-independent immune responses.

Catalytic antibodies: a critical assessment

In the past few years, antibodies that catalyze a variety of reactions with enzyme-like properties have been produced. The present review is of a critical nature, rather than a survey or an introduction to the field of catalytic antibodies. Here, we examine the performance of catalytic antibodies in light of the features that define an enzyme: substrate specificity, rate enhancement, and turnover. We also refer to some limitations of the technologies currently used for their generation. In the future, antibodies may provide a new repertoire of tailor-made, enzyme-like, catalysts with possible applications in biology, medicine, and biotechnology. In the following sections, we emphasize that these applications will require far more efficient catalysts than are presently available, and we point to several trends for future research that may offer more efficient catalytic antibodies.

Cloning and characterisation of TPO autoantibodies using combinatorial phage display libraries

Thyroid lymphocyte RNA from a Hashimoto patient with high serum levels of autoantibodies to thyroid peroxidase (TPO) was used to construct a phage display antibody library in the phagemid vector pComb3. The library (100,000cfu) encoded IgG1 heavy chains together with kappa light chains. Selection of the phages displaying TPO antibody on TPO-coated ELISA plates yielded a phage population enriched for surface expression of TPO antibody Fabs. 3 different Fabs specific for TPO were subsequently isolated with affinities in the region of 10(9) molar-1. 2 of the Fabs recognised the same, or closely related, epitopes on TPO whereas the third Fab recognised a different epitope. These 2 epitopes were recognised by TPO autoantibodies in the serum of the lymphocyte donor and a series of 10 patient sera. Available sequence data showed that several non-self antibodies and non-thyroid autoantibodies use the same V kappa and VH germline genes as TPO autoantibodies. There appeared to be no clear relationship between gene sequence or gene family usage by TPO autoantibodies of the same or similar epitope specificity.

Liquid-liquid partition chromatography as a method to examine surface properties of antibodies and antigen-antibody complexes

We demonstrate liquid-liquid partition chromatography in aqueous two-phase systems (LLPC) as a simple method for examining the surface properties of immunoglobulins and antigen-antibody complexes in solution. LLPC separates molecules with respect to the properties of the exposed surfaces. As an example, the method may be used to detect changes in the conformation of IgG following chemical modification like acylation or iodination. We have studied the partitioning of antibodies and antigen-antibody complexes, modelled by rabbit antibodies against three human serum proteins, in aqueous polyethylene glycol/dextran two-phase systems at pH 7. Analysis of both polyclonal and monoclonal antibodies against various antigens suggested that the partition properties of immunoglobulins are related mainly to their antigen specificity and not to subclass-specific structures. Furthermore, experiments indicated that changes in the surface properties of antigen and/or antibody following complexation may be detected. Thus, LLPC may prove to be a new way of studying the relation between antibody structure and function in solution.

Variable region cDNA sequences of three mouse monoclonal anti-idiotypic antibodies specific for anti-alpha(1----6)dextrans with groove- or cavity-type combining sites

The variables regions of three syngeneic anti-idiotypic antibodies (Ab2s) were cloned and sequenced. They are encoded by different VL genes, two are from different members of V kappa-Ox1 superfamily. The H chains are encoded by VH genes belonging to three different VH families, J558, Q52 and 7183. Together with a previous report from this laboratory, the nucleotide sequences of four Ab2s to anti-alpha(1----6)dextrans have been presented. They are derived from a number of unrelated germline genes, and differ from similar studies in anti-NP, anti-GAT and anti-Ars systems. Three of four Ab2s in the anti-alpha(1----6)dextran system appear to have D-D fusions, which has also been reported in several other Ab2s.

Sequences of variable regions of a monoclonal antibody specific to the thyroid hormone, triiodo-L-thyronine

We have determined the nucleotide sequences of the variable regions of H and L chains of a monoclonal antibody 98QQ that interacts with the thyroid hormone triiodo-L- thyronine with high affinity. Analysis of the nucleotide sequence of the light chain V region of 98QQ revealed that the VL sequence is 99% identical to Balb/c germline Vk 21-E sequence. That is an interesting finding with this high affinity anti-T3 antibody, since occurence of predominantly germline variable region sequences is observed in some autoantibodies to self antigens but not usually in high affinity IgG antibodies. The sequence analysis also revealed that the heavy chain variable region sequence of 98QQ is similar to a V region of an anti-DNA antibody (MRL DNA 22). Thus the sequence analysis of our anti-T3 mAb 98QQ has revealed some features of autoantibodies to self antigens.