Mouse antibodies to the insulin receptor developing spontaneously as anti-idiotypes. I. Characterization of the antibodies

Mice immunized to ungulate insulins were found to develop antibodies of two specificities: insulin antibodies that were mostly IgG1 and IgG2 antibodies that acted both as anti-idiotypes to specific mouse insulin antibodies and as antibodies to the insulin receptor. There was a negative association between the presence of anti-idiotypic receptor antibodies and insulin antibodies bearing the specific idiotype; the specific idiotypic antibodies were confined to the early phase of the primary response while the anti-idiotypic receptor antibodies were detected only after the idiotypic antibodies had disappeared. To map the insulin epitope that triggered the specific idiotypic response, we chemically altered the insulin molecule so as to inhibit its interaction with the insulin receptor. The altered insulins triggered high titers of antibodies binding to antigenic determinants on native insulin, but no anti-idiotypic receptor antibodies. Thus, the epitope responsible for the specific idiotypic-anti-idiotypic network was probably the part of the insulin molecule whose conformation is recognized by the insulin receptor.     

Autoantibodies against human insulin.

Sera from 680 non-diabetic subjects with suspected autoimmune disease were screened for 13 different antibodies. Of the 582 sera found to contain these antibodies, nine bound insulin in an IgG specific enzyme linked immunosorbent assay (micro ELISA). Four of the sera bound human, porcine, and bovine insulins and five bound exclusively human insulin. "Cold" human, porcine, and bovine insulins each displaced, in a dose dependent manner, the four sera which bound all three insulins, but only human insulin displaced the remaining five, porcine and bovine insulins having little or no effect in concentrations up to 1000 U/1. These observations point to the existence of autoantibodies specifically against human insulin in some subjects with established autoimmunity.     

Influence of insulin antibodies on pharmacokinetics and bioavailability of recombinant human and highly purified beef insulins in insulin dependent diabetics.

Sixteen insulin dependent diabetics of long standing, with undetectable fasting plasma C peptide concentrations, and eight non-diabetic controls were each infused intravenously with biosynthetic human and highly purified beef insulin (1 mU/kg/min) while euglycaemia was maintained by a Biostator. No difference was observed between the two insulins in respect of insulin pharmacokinetics or biological action. The diabetics showed appreciable insulin resistance, manifested by a 40% reduction in the rate of insulin mediated glucose disposal, which was unrelated to the presence of insulin antibodies. Insulin binding antibodies, however, increased insulin''s clearance rate and distribution space and prolonged its pharmacological and biological half lives. The rate at which insulin action was lost, after an intravenous infusion, was more rapid in diabetics without insulin antibody binding than in controls. In respect of their influence on insulin pharmacokinetics, moderate concentrations of insulin antibodies may be of positive advantage to all diabetics without endogenous insulin secretion and are not responsible for the insulin resistance of type 1 diabetes.     

Human insulin

The two human insulins of clinical importance are (a) semisynthetic human insulin prepared from pork pancreas by enzymatically substituting threonine for alanine-the last amino acid in the beta chain-thereby transforming pork insulin in vitro to human insulin; and (b) biosynthetic human insulin synthesized biotechnologically in Escherichia coli-K12. Using this latter technique, it is possible to produce mass quantities of highly purified insulin for the treatment of insulin-dependent diabetics, avoiding the problems inherent in supplies of insulin produced from animal pancreas. It has been suggested that to avoid confusion the two human insulins should be called semisynthetic human insulin of pork origin and biosynthetic human insulin of E. coli origin, respectively. These insulins have four advantages over highly purified animal insulins: (a) they induce lower titers of circulating insulin antibodies; (b) their subcutaneous injection is associated with fewer skin reactions; (c) they are absorbed more rapidly from the injection site; and (d) less degradation occurs at the site of injection. These data indicate that newly diagnosed insulin-dependent diabetes, particularly in children, should be treated with either of the two human insulins. The warranty against inadequate supplies of insulin offered by biosynthetic human insulin makes the use of pork insulins unnecessary and beef insulins totally useless.     

Human insulin autoantibody fine specificity and H and L chain use

Fine specificity and H and L chain isotypes of insulin autoantibodies in sera from 11 subjects were examined. None of these 11 subjects was treated with exogenous insulin. Two patterns of fine specificity were found. In one, the autoantibodies were specific for human insulin, with a requirement for threonine at B30. The conservative substitution in pork insulin (threonine to alanine) abrogated IgG binding by these sera. Insulin autoantibodies in other sera cross-reacted with beef, pork, and human insulin; not requiring threonine at B30. Reciprocal competitive inhibition experiments showed that epitopes recognized by the human specific insulin autoantibodies were exclusively on the B chain, whereas the cross-reactive sera contain autoantibodies that recognize both the B chain and combinatorial (A and B chain) epitopes. The fine specificity of cross-reactive insulin autoantibodies are thus similar to insulin antibodies from insulin-treated subjects. When IgG subclasses and L chains of insulin autoantibodies were examined, however, restricted C region usage was found. The hierarchy was IgG3 greater than G1 greater than G2 greater than G4; with one subclass dominant in each serum, although others were used. L chain use was similarly restricted. There was no correlation between isotype and fine specificity or between H and L chain type. It is concluded that heterogeneity of insulin autoantibodies is restricted. The response is probably more oligo- or pauciclonal than insulin antibody from insulin-treated subjects.     

Insulin-specific human T cells. Epitope specificity, major histocompatibility complex restriction, and alloreactivity to a diabetes-associated haplotype

T cells from an insulin-treated diabetic (ML, HLA DR1, w6) were stimulated in vitro with insulin, cloned at limiting dilution, and examined for their fine specificity and genetic restriction. T cell lines (TCL) derived from beef insulin stimulation were highly specific for epitopes on beef insulin, whereas pork insulin stimulation generated T cells that recognized determinants shared with beef insulin. Included among TCL reactive with pork insulin is one line (P2/9) that is autoreactive with human insulin. Antigen-presenting cells of known HLA type and monoclonal antibodies directed at class II major histocompatibility complex antigens were used to confirm the role of HLA-DR in restricting the response of insulin immune T cells. No preference or determinant selection within the donor's haplotypes was identified because either DR1 or DRw6 antigen-presenting cells could present the A loop of beef insulin. A TCL that recognized the A loop of beef insulin in association with DR1 was also alloreactive to HLA DR3, or a molecule closely linked to it, in the absence of insulin. A second T cell clone with insulin specificity and alloreactivity was also derived by allo stimulation of the donor's cells with DR3+ cells. When tested with a series of DR3+ stimulator cells, the alloreactivity was directed at diabetes-associated haplotypes. These data show that the T cell repertoire for insulin of a single diabetic donor encompasses that of multiple inbred animal strains and includes fine specificity for one to two amino acids, recognition of autologous insulin, and cross-reactivity with an allogeneic major histocompatibility complex antigen.     

Correlation between HbA1c, purified insulins, diabetic control, and insulin antibodies in diabetic children

Insulin antibodies were determined in sera from 38 children diagnosed as having juvenile diabetes for a duration of 0.7-15.2 years (median = 4.9 years). 8 children were treated with purified porcine insulins from the beginning of their disease, 16 children with bovine insulin NPH alone, and 14 children with non-purified, of whom 9 were later transferred to purified insulins. Serum insulin antibodies were measured by non-specific and specific methods using beef (B) and pork (P) antigens as described by Welborne and Sebriakova, respectively. 12/38 children had insulin binding levels similar to those of normal children, irrespective of the type of insulin used. The concentration of antibodies using radiolabelled B or P insulins as antigens were strongly correlated, by both the non-specific (p less than 0.01) and the specific (p less than 0.01) methods. Children with better score for diabetic control had significantly lower levels of insulin antibodies against B (p less than 0.05) and P (p less than 0.05) than those with poor diabetic control. There was also a significant positive correlation between mean HbA1c concentration and both B and P mean insulin antibody concentration (p less than 0.01). Finally, patients treated with purified porcine insulin had significantly lower levels of antibodies than patients with non-purified bovine insulin (p less than 0.05).     

Study of insulin-receptor interactions in plasma membrane of rat liver using antibodies against insulin

Preincubation of rat liver cell plasma membranes with antiinsulin serum (AIS) results in a 3-fold increase of specific binding of [125I]insulin added to AIS-free membranes. The degree of such increase depends on preincubation time and AIS concentration; at a rise of both parameters the dependence curve shown a plateau. The Scatchard plots suggest that preincubation with AIS increases the high and low affinity of binding sites; their number remains thereby unchanged. Using several types of immune and non-immune sera for preincubation the specificity of this effect was established. It was assumed that the antibodies initiate the immunological reaction of the membrane. Study of AIS effect on dissociation of membrane-bound labelled insulins differing in their affinities for the given AIS revealed that the antibodies can interact with the newly formed hormone-receptor complex but cannot affect the receptor within insulin. It is concluded that the intact hepatocyte membrane contains a receptor-bound insulin which is a target of AIS action. The antigen-antibody interaction changes the functional state of occupied and free receptors which bind [125I]insulin. The latter becomes repeatedly involved in this process as a result of destabilization of site-site interactions provoked by antibodies against the hormones.     

Antineutrophil cytoplasmic autoantibodies in Wegener's granulomatosis recognize conformational epitope(s) on proteinase 3.

Although proteinase 3 (PR3) has been identified as a major autoantigen in Wegener's granulomatosis, the precise antibody specificity(ies) and requirements for epitope recognition have not been characterized. We analyzed 11 sera containing antineutrophil cytoplasmic antibodies (cANCA) for binding to azurophilic granule proteins extracted from neutrophils under various conditions and for binding to native or rPR3. Ten of 11 (91%) of the cANCA sera bound to PR3 extracted by nonionic detergents when tested by immunoprecipitation or by IEF followed by capillary immunoblotting. Antibody binding to PR3 was retained when IEF was performed under dissociating conditions (8 M urea) indicating that PR3 is the major autoantigen in azurophilic granules and that association with other proteins is not required for antigenicity. In contrast, antigenicity was totally destroyed by exposure of PR3 to reducing agents or to low pH (less than 3.0) and was either lost or considerably diminished after boiling in SDS. cANCA sera also showed little or no binding to rPR3 expressed as a fusion protein in Escherichia coli or synthesized by wheat germ ribosomes in vitro. Inasmuch as PR3 enzymatic activity was partially retained after acid extraction, these findings indicate that cANCA bind to a limited number of conformational epitopes on PR3. In addition, IEF followed by capillary immunoblotting appears to be a sensitive and specific method to detect anti-PR3 antibodies in Wegener's granulomatosis.     

A radioreceptor assay for insulin: direct measurement of dog pancreatic vein serum insulin.

A simple radioreceptor assay for insulin rat liver membranes as receptor sites, with sufficient specificity precision, and sensitivity to detect 10 ng or 276 muU/ml of serum insulin, has been developed. In the presence of standard porcine insulin at the concentration of 1.0 ng/tube, approximately 8% of 125I-porcine insulin was bound to the plasma membranes and ninety-five per cent of this binding was inhibited by 1.0 microgram of standard insulin per tube. Four animal insulins inhibited the binding of 125I-insulin while ACTH, glucagon, human growth hormone, and oxytocin were inert. Insulin values in dog pancreatic vein sera obtained during and after glucose loading and measured by the present radioreceptor assay agreed well with immunoreactive insulin. The ratio of IRI to the measurement by radioreceptor assay was 1.09 +/- 0.18 for the same sera.     

Thymus-dependent antiidiotype and anti-antiidiotype responses to a dinitrophenyl-specific monoclonal antibody

BALB/c mice were inoculated i.p. with graded doses of a DNP-specific, IgM mAb (designated 57.1). Injection with unmodified 57.1 in the absence of adjuvants resulted in the generation of an anti-Id response (Ab2) and an anti-anti-Id response (Ab3). The generation of serum anti-Id antibodies was found to be thymus dependent. Nude mice immunized with 57.1 were unable to produce a serum Ab2 response above nonimmunized controls whereas euthymic mice receiving identical doses of 57.1 produced strong Ab2 responses. To examine the specificity of serum anti-Id, sera from mice receiving 57.1 were screened against a panel of mAb representing at least five distinct VH gene families. Serum titers were significantly higher against 57.1 than against any of the other antibodies in the panel. Three of the antibodies in this panel bind FD5-1, a monoclonal anti-Id (Ab2) that also binds 57.1. However, sera from mice receiving 57.1 bound 57.1 only. Thus, the serum Ab2 response appears to be highly specific for idiotopes on 57.1. The predominant isotype of these anti-Id antibodies was IgG1. The number of isotypes detected increased in a dose dependent manner with all IgG subclasses having anti-Id specificity in sera from animals receiving the higher doses of 57.1. Further analysis of the serum demonstrated that approximately 8% of the Ab2 response was paratope-specific (inhibitable by the monovalent hapten DNP-lysine). The same sera were analyzed for the presence of Ab3 by binding to the monoclonal anti-Id antibody FD5-1. Lower serum titers of Ab3 were generated in comparison to serum titers of Ab2. Analysis of the binding specificity of the Ab3 response revealed that DNP-BSA was able to partially inhibit the binding of serum IgM and IgG Ab3 to FD5-1. A subset of the Ab3 response. Ab1' that is specific for DNP was observed in a direct binding assay where detectable amounts of DNP binding IgM, IgG1, and IgG3 isotypes were present. We have thus described a complete circuit (Ab1----Ab2----Ab3) of antibodies within the Id network by immunizing animals with an unmodified mAb in the absence of Ag or adjuvants.     

Analyses of Idiotypes on Anti-α-1, 6-Dextran Antibodies in Mice

When mice of strains C57BL/6, C3H/He, and BALB/c were immunized with native dextran B512, only a small amount of IgM antibody was produced, but a substantial amount of anti-dextran antibody of IgG class was produced after immunization with a conjugate of dextran T10 and keyhole limpet hemocyanin regardless of the mouse strain used. Isoelectric focusing (IEF) spectra revealed limited heterogeneity of anti-dextran antibody of IgG class with strict consistency in all individual sera from C57BL/6 mice, even after secondary immunization, whereas antibodies from C3H/He and BALB/c mice showed more heterogeneous IEF spectra with some individual variations. Rabbit anti-idiotypic (Id) antibodies were raised by immunization with a subfraction of anti-dextran antibody of IgG class from C57BL/6 mice, which showed major bands focused at around pH 7.7 upon IEF. It was found by using the anti-Id antibodies that virtually all anti-dextran antibody molecules of both IgG and IgM classes from C57BL/6 mice possessed common Id determinants which can be classified into two specificities, one specific for antibody from C57BL/6 mice and the other cross-reactive with antibodies from BALB/c and C3H/He mice. About 80% of the antibody molecules from BALB/c and less than 20% of those from C3H/He mice were positive for the interstrain cross-reactive Id. Both Id determinants seemed to be closely related to the antigen binding sites, or at least to reside in the vicinity of the antigen binding sites of anti-dextran antibody.     

Anti-Ro autoantibody with cross-reactive binding to the heavy chain of immunoglobulin G.

Autoantibodies directed at the intracellular Ro ribonucleoprotein complex are found in the serum of patients with systemic lupus erythematosus (SLE) and related autoimmune diseases. The antigenic stimulus for the induction of these autoantibodies is unknown, although we have previously demonstrated that the Ro protein and immunoglobulin G (IgG) share immunologic determinants bound by anti-Ro antibodies. The present study further defines the fine specificity of this cross-reactive binding. Using both patient autoanti-Ro antibodies and antigen-induced rabbit anti-Ro serum, the binding specificity for IgG was located to the heavy chains of IgG outside the Fc domain. F(ab')2 fragments of IgG were observed to inhibit specific Ro binding by either human or antigen-induced rabbit sera, while Fc fragments of IgG failed to inhibit Ro binding. Anti-Ro sera were found to bind the heavy chains of IgG in immunoblots, and the antibodies eluted from these heavy chains were capable of immunoprecipitating the Ro particle from human cell extracts. Not all patient sera with anti-Ro antibodies possessed IgG binding antibodies. Studies of cyanogen bromide digestion fragments of IgG implicate the hinge region of IgG as the region cross-reactive with the Ro protein. The nature of this cross-reactivity may be important in understanding the induction and/or perpetuation of the anti-Ro response in patients with autoimmune disease.     

Protection against High-Dose Highly Pathogenic Mucosal SIV Challenge at Very Low Serum Neutralizing Titers of the Antibody-Like Molecule CD4-IgG2

Passive transfer studies using monoclonal or polyclonal antibodies in the macaque model have been valuable for determining conditions for antibody protection against immunodeficiency virus challenge. Most studies have employed hybrid simian/human immunodeficiency virus (SHIV) challenge in conjunction with neutralizing human monoclonal antibodies. Passive protection against SIV, particularly the pathogenic prototype virus SIVmac239, has been little studied because of the paucity of neutralizing antibodies to this virus. Here, we show that the antibody-like molecule CD4-IgG2 potently neutralizes SIVmac239 in vitro. When administered by an osmotic pump to maintain concentrations given the short half-life of CD4-IgG2 in macaques, the molecule provided sterilizing immunity/protection against high-dose mucosal viral challenge to a high proportion of animals (5/7 at a 200 mg dose CD4-IgG2 and 3/6 at a 20 mg dose) at serum concentrations below 1.5 μg/ml. The neutralizing titers of such sera were predicted to be very low and indeed sera at a 1∶4 dilution produced no neutralization in a pseudovirus assay. Macaque anti-human CD4 titers did develop weakly at later time points in some animals but were not associated with the level of protection against viral challenge. The results show that, although SIVmac239 is considered a highly pathogenic virus for which vaccine-induced T cell responses in particular have provided limited benefit against high dose challenge, the antibody-like CD4-IgG2 molecule at surprisingly low serum concentration affords sterilizing immunity/protection to a majority of animals.     

Functional, structural, and antigenic similarities of guinea pig anti-phosphorylcholine antibodies.

The humoral immune response to PC was measured in guinea pigs. PC-vaccine stimulated IgM and IgG2, but little IgG1, anti-PC -antibodies. No memory was induced and immunization in CFA produced tolerance. PC-KLH, on the other hand, stimulated IgM, IgG2, and IgG1 anti-PC antibodies with carrier-specific memory. Hapten inhibition of plaque formation showed uniform binding patterns with minor, but significant, differences between antiPC-vaccine and anti-PC-KLH antibodies. The antibodies were characterized by IEF and idiotypic analyses. Early after immunization with PC-vaccine, guinea pigs had restricted IEF patterns which in inbred, but not outbred animals were indistinguishable between individuals. These patterns remained restricted but more individualized with time after immunization. Anti-PC-KLH antibodies showed more heterogeneity and individuality. However, these structurally heterogeneous antibodies reacted equivalently with rabbit anti-idiotype antisera and therefore must share common structural features, regardless of isotype or the genetic background of the guinea pig.     

Immunocytochemical localization and immunochemical characterization of an insulin-related peptide in the pancreas of the urodele amphibian,Ambystoma mexicanum

Summary The pancreas of the axolotl, Ambystoma mexicanum, was investigated by immunocytochemical methods for the presence of immunoreactivity to a number of antisera raised against mammalian insulins. All anti-insulin antisera tested revealed substantial amounts of reaction products confined solely to the aldehyde-fuchsinophilic B cells of the endocrine pancreas. The reactive cell population was detected by use of one polyclonal antiserum against bovine insulin and eight different monoclonal antibodies against insulins from various mammalian species. Six of these antibody clones have known specificity to sub-regions of the insulin molecule. Additionally, fractions of an ethanol-HCl extract of pancreatic tissue from Ambystoma was studied in both conventional dot-blot tests by means of the same panel of antibodies and a two-site sandwich time-resolved immunofluorometric assay for human insulin involving two of the monoclonal antibodies. These experiments support the immunocytochemical observations by demonstrating the existence of an insulin-related peptide with a great deal of structural resemblance to mammalian insulins and displaying antigenic determinants in common at least with the amino acid residues A_8–10 and B_26–30. In conclusion, we interpret the findings as indicating that the immunocytochemically revealed tissue bound antigen in the Ambystoma pancreatic B-cells may be a peptide related to human insulin.Supported in part by SNF grant 11-5082 (G.N.H.). The authors are indebted to Dr. P. Rosenkilde for the gift of the Ambystoma material     

Antibodies against ganglioside GT3 in the sera of patients with type I diabetes mellitus

Clinical and experimental data support the concept that type I diabetes mellitus results from autoimmune destruction of pancreatic beta cells. Although both proteins and glycolipids are targets of anti-islet cell antibodies, the Ag have not been purified or characterized. Previously, we observed that rat insulinoma (RIN) cell lines varied in their reactivity with both human antibodies and murine mAb A2B5, which binds to polysialo gangliosides. To determine the chemical basis of the varied immunoreactivity, we analyzed the glycosphingolipids of 5 RIN lines. Glycolipids bound by two mAb and by antibodies in the sera of type I diabetics were identified. The more immunoreactive RIN lines contained a much higher content of gangliosides and a higher proportion of complex gangliosides. The major gangliosides were GM3, GD3, and GT3. By high performance TLC immunostaining, we demonstrated that A2B5 and R2D6, an anti-beta cell murine mAb, bound most strongly to ganglioside GT3. The binding of human sera to gangliosides was analyzed by an ELISA assay. Although both normal and diabetic sera contained antibodies to various glycolipids, binding to GT3 was significantly elevated in 31 new-onset type I diabetics (p less than 0.001). The presence of the GT3 trisialosyl epitope on human islet cells was shown by immunofluorescent staining by both R2D6 and A2B5. These findings support previous suggestions that gangliosides play an important role in the immunopathology of type I diabetes, and identify for the first time a specific ganglioside Ag that is the target for autoantibodies in a subset of diabetic patients.     

Transfer from porcine insulin to human insulin in insulin-dependent diabetes mellitus: effects on insulin binding to IgG and glycemic control

Thirty type I diabetic patients who were treated for at least 2 years with a combination of regular and lente monocomponent porcine insulins were allocated in a double-blind study to either continued porcine insulin treatment or a transfer to the corresponding semi-synthetic human insulins. Insulin binding to IgG measured by an immunoelectrophoretic method, was followed at 3-month intervals for 1 year, and did not change after the transfer. The glycemic control, as assessed by hemoglobin A1 levels, tended to deteriorate in the human insulin group during the first 3 months of the trial and then return to the baseline level. It is concluded that a transfer from highly purified porcine insulin to human insulin apparently does not change the insulin binding to IgG in already sensitized patients.     

Idiotypic determinants on human anti-insulin antibodies are cyclically expressed

Insulin-treated diabetics develop a heterogeneous antibody response to the protein hormone. To understand the repertoire of this diverse immune response, we developed monoclonal antibodies (Mab) that recognize idiotypic determinants on anti-insulin antibodies. Two Mab that recognize distinct idiotopes on anti-insulin antibodies were studied in detail. These idiotopes were present on IgG, but not IgM, anti-insulin antibodies from the proband and from four of 11 insulin-treated diabetics. The determinants recognized were distinct from conventional allotype (Gm) markers and were found in both Type I and Type II diabetics. The partial inhibition of anti-idiotope binding by beef insulin suggests that idiotopes related to the binding site for beef insulin may be recognized. When the presence of idiotope-positive molecules was determined sequentially during insulin therapy, in one patient, idiotope expression was found to be cyclical. In contrast, the total amount of anti-insulin antibody was constant, whereas idiotope expression varied. These data suggest that common V region determinants are employed in the anti-insulin response, and their variable expression may reflect dynamic regulation during chronic insulin therapy.     

IgG antibodies to secretory component in normal human serum

While isolating free secretory component (FSC) by monoclonal antibody affinity chromatography, we demonstrated FSC-IgG complexes in human milk. We hypothesized that IgG antibody to secretory component (SC) might be transported into the milk from the serum. We therefore examined sera from 10 normal adults and 10 infants for IgG capable of binding to FSC in an enzyme-linked immunosorbent assay. Eight of 10 normal adult sera and nine of 10 infant sera demonstrated IgG binding to FSC with titers ranging from 1:54 to 1:4096. Quantitation of the IgG bound to FSC was hampered in adult sera by the binding of IgM and polymeric IgA to the FSC. Quantitation in five infant sera ranged from 0.5 to 6.4 micrograms/ml. A pepsin digest of an IgG fraction of serum demonstrated binding of the F(ab')2 fragments to the FSC. The specificity of the antibodies in human serum was evaluated by examining the binding to secretory IgA (sIgA) and FSC isolated from pooled human milk and polymeric IgA isolated from the ascitic fluid of a patient with an IgA myeloma. Eight of the 10 adults had antibody specific for FSC. Three of the eight, all female, also had antibody specific for sIgA. Two of the eight had antibody either to FSC and sIgA or to FSC plus an antibody that could bind to an epitope shared by sIgA and FSC. Competition experiments with monoclonal antibodies to human secretory component and sIgA were used to confirm and further define these specificities. The results of this study indicate that antibody to SC is common in normal adult and infant sera. The majority of antibodies seem to be directed against epitopes present on FSC but not on sIgA, which suggests sensitization to circulating or membrane-bound SC. The significance of these antibodies in normal human sera remains to be elucidated.