Production Of IgA Antibodies Research Articles

Immune response to bacterial dextrans. II. T cell control of antibody isotypes.

The isotype distribution of Dextran B 512 (Dex)-specific plaque-forming cells (PFC) and serum antibodies was studied after in vivo immunization in C57BL/6 mice. Although IgG2b and IgG3 could also be detected in most individuals, the majority of non-IgM PFC were of the IgA isotype. All classes other than IgM were T cell-dependent, as shown by their complete absence in athymic "nude" mice. This unusual isotype pattern was further investigated by studying the antibody responses to the same Dex epitope coupled to a protein carrier, and to a different hapten coupled to the carrier Dex or to a protein. The results show that IgA responses are epitope-related and selectively associated with anti-Dex antibodies: no IgA PFC are detected against a hapten coupled to Dex or proteins, while the enhanced levels of helper cell reactivity provided by protein carrier to Dex result in the appearance of IgG1 antibodies in addition to IgA. These results indicate that T cells that modulate isotype patterns in these responses can discriminate between Dex- and DNP-specific B cells in the response to the same carrier. Since the same idiotype is detected on a large fraction of the IgM and IgA anti-Dex response and antiidiotypic helper cells have previously been detected in normal C57BL/6 mice, we suggest that idiotype-specific T cells control the production of IgA antibodies upon immunization with Dex.

Polyclonal antibody production induced by parotid protein and its active glycopeptide in mouse and human lymphocytes

Parotid protein (extracted from bovine parotid gland), one parotid subunit, and the active fragment (Fr AA-1) induced polyclonal IgM, IgG, and IgA antibody production in murine spleen cells in vivo and in vitro. The parotid subunit also elicited polyclonal IgM antibody responses in immune defective CBA/N and LPS nonresponder C3H/HeJ mouse. When one carbohydrate residue in the subunit was removed by treatment with mixed galactosidase, the ability to generate nonspecific responses was markedly reduced. In human peripheral lymphocytes, the parotid subunit and Fr AA-1 induced considerable polyclonal IgM antibody production. The subunit was not toxic to human lymphocytes. Therefore, it seems possible that parotid protein could be applied to immunopharmacological therapy for conditions such as immunodeficiency.

T cell hybridomas coexpressing Fc receptors (FcR) for different isotypes. II. IgA-induced formation of suppressive IgA binding factor(s) by a murine T hybridoma bearing Fc gamma R and Fc alpha R.

T2D4 murine T hybridoma cells have previously been shown to express Fc receptors (FcR) for IgG (Fc gamma R) and for IgA (Fc alpha R) and to produce an IgG binding factor (IgGBF) that suppresses IgG and IgM responses. In the present work we report on the behavior of IgA bound to T2D4 cells and on the production of IgA binding factor (IgABF) and its ability to suppress IgA antibody production. A dose-dependent binding of MOPC315 IgA with anti-TNP activity by T2D4 cells was demonstrated by rosette formation with trinitrophenylated ox red blood cells (TNP-ORBC) and fixation of iodinated DNP-BSA. IgA bound to the cells disappeared after a short-term culture of 3 hr at 37 degrees C, but not at 4 degrees C. Because this phenomenon was inhibited by 0.1% sodium azide and 100 microM dansylcadaverine, a transglutaminase inhibitor, Fc alpha R-IgA complexes seemed to be released by an active process involving receptor movement. In the culture supernatant of IgA-treated T2D4 cells, we detected a factor(s) that binds to IgA-Sepharose and competitively inhibits the binding of IgA to T2D4 cells. The factor (IgABF) failed to inhibit the rosette formation of Fc gamma R(+) cells with IgG-sensitized ORBC (EAox gamma), indicating that it binds specifically to IgA. IgABF was undetectable in the culture supernatants of untreated T2D4 cells of Fc alpha R(-) BW5147 T lymphoma cells used as parent cells for the establishment of the hybridoma. To study the effect of IgABF on antibody formation, culture filtrates of IgA-treated or untreated T2D4 cells were fractionated on IgA-Sepharose beads and were added to BALB/c spleen cells cultured with pokeweed mitogen. By use of a reverse plaque assay, it was shown that the IgA plaque-forming cell (PFC) response was suppressed by the acid eluate but not by the effluent of IgA-Sepharose beads incubated with the filtrates of IgA-treated T2D4 cell cultures. The suppression was IgA specific, because neither IgG nor IgM responses were suppressed by the eluate. As expected, there was no significant IgA suppressive activity in the acid eluates of the beads incubated with the culture filtrate of untreated T2D4 cells or IgA-treated BW5147 cells. IgA-specific suppressive activity proved to be due to IgA binding factor(s), because suppressive activity in the eluate was completely adsorbed by IgA-Sepharose but not by IgG- nor BSA-Sepharose.

Secretory IgA specific for herpes simplex virus in lacrimal fluid from patients with herpes keratitis--a possible diagnostic parameter.

In the present study a solid-phase radioimmune assay was used for the demonstration of herpes simplex virus-specific IgG and secretory IgA antibodies in the lacrimal fluid from patients with active recurrent herpes keratitis. The method was quantitative and made it possible to test specifically for the production of secretory IgA antibodies produced during an active herpes simplex virus infection. The production of secretory IgA was followed in 2 patients with fresh recurrent lesions. The HSV-specific secretory IgA could be demonstrated during the first 10 days of infection, where the maximal concentration was reached 3-5 days after the first symptoms occurred. The secretory antibodies were locally produced, and it is shown for the first time that herpes virus-specific secretory antibodies were of diagnostic value.

Sera from breast cancer patients contain an IgA antibody to a breast cyst fluid component

We have previously demonstrated that aspirated breast cyst fluids from women with benign breast disease contained a particulate component (BCF-PC) that reacted with antiserum to the murine mammary tumor virus (MuMTV). Since women with breast cancer contain antibodies and antigens reactive with MuMTV it was of interest to measure the immune response in breast cancer patients to the BCF-PC. Sera were obtained from breast cancer patients prior to surgery and 5 days postmastectomy, from women with benign breast disease and from healthy women. Analysis for antibody to the BCF-PC was by an enzyme-linked immunosorbent assay. IgA from premastectomy breast cancer patients was significantly more reactive with the BCF-PC than was IgA from women in the other categories ( P < 0.001). In contrast, IgG binding to the BCF-PC did not differ among the various groups. It was further demonstrated that the reactivity of IgA from any particular serum for the BCF-PC remained constant for different BCF-PC isolates. Other fractions of breast cyst fluids were unreactive with IgA from breast cancer patients. Determination of serum IgA levels in these women revealed that the mean IgA level was significantly higher in healthy women than in the other categories ( P < 0.001). Individual IgA levels and levels of IgA antibody to the BCF-PC were found to be not correlated in breast cancer patients. These results indicate that the production of a specific IgA antibody to the BCF-PC increases in sera of patients with breast cancer. Lastly, it was shown that preincubation of patients' sera with disrupted MuMTV, but not intact MuMTV, severely reduced subsequent IgA binding to the BCF-PC. The analysis of sera for IgA antibody to the BCF-PC may be of value for the detection of breast cancer or its recurrence.

Preferential Production of IgA Antibodies by Spleen Fragments Immunized and Maintained in Vitro

Primary immune responses were induced with a hapten conjugate of Ficol (NIP38-Ficoll) in cultures of mouse spleen fragments. The production of anti-NIP antibodies and mouse immunoglobulins IgA and IgG was monitored by solid-phase radioimmunoassays. Cultures produced IgA and IgG without any stimulation. IgA at a rate of c.60ng/mg of wet tissue per day throughout the 10-day culture period. Background production of IgG gradually declined from almost 1000 ng/mg/day at the beginning to 100 ng/mg/day at the end. Immunization modestly reduced the decline of the IgG production. It increased the production of IgA fourfold between days 6 and 8. It induced an antibody production that had its peak between days 6 and 8. Up to 50 ng of anti-NIP antibody was produced during this period per mg of wet tissue per day. Of this antibody 75-90% was IgA. When the same antigen was injected into mice, more than 75% of anti-NIP antibodies detected in their sera on day 8 were IgG. One cause of the high ratio of IgG/IgA antibodies detected in their sera on day 8 were IgG. One cause of the high rate of IgG/IgA antibodies in vivo is the efficient elimination of oligomeric IgA by the liver. Another possible cause is that the production of IgA is less sensitive than the production of IgG to the adverse conditions of the culture.

Thymic Control of Secretory Antibody Responses in the Rat

The effect of neonatal thymectomy on salivary and serum antibody responses was studied in rats. Local immunization of thymectomized rats with a T-dependent antigen (DNPBGG) elicited negligible amounts of IgA anti-DNP antibody in saliva. In contrast, both normal and sham-thymectomized animals demonstrated substantial levels of salivary IgA antibody. All thymectomized rats locally injected with a T-independent antigen (DNP-Lys-Ficoll) exhibited salivary IgA antibody production. Salivary IgG antibodies were somewhat decreased in thymectomized rats injected with either antigen; however, the final effect of T cell deprivation on IgG synthesis was not as pronounced as on IgA synthesis. Serum IgA antibody was induced in control rats injected with DNPBGG, whereas this Ig class of antibody was absent in thymectomized rats. The results suggest that thymus-derived cells exert a regulatory influence on both serum and secretory IgA responses to antigens.

Local Production of IgG, IgA, and IgE Antibodies in Grass Pollen Hay Fever

Abstract An antigen-binding radioimmunoassay was used to measure binding activity for the group I protein of Rye grass pollen (BA). Most patients with grass pollen hay fever were found to have IgG, IgA, and IgE BA in both their serum and nasal secretions. When these results were related to the quantities of total immunoglobulin of each class, it was clear that the IgG and IgA in nasal secretions had a higher proportion of pollen-specific ab than the IgG and IgA in serum. This calculation showed that most of the IgG BA and IgA BA in nasal secretions must have been produced locally. In contrast, IgG ab to diphtheria toxin was found to represent a lower proportion of nasal IgG than serum IgG. These findings for IgG anti-toxin supported the view that the pollen-specific IgG BA must have been produced locally. Specific antipollen IgE ab in nasal secretions was compared with the other classes. The results suggested that the IgE ab in nasal secretions was also locally produced. Furthermore, saliva samples from the patients that contained similar total quantities of IgG and IgA to nasal secretions were found to have little or no pollen-specific IgG, IgA, or IgE BA. After injections of grass pollen extract, the major antibody response was an increase in serum IgG BA, which was not matched in nasal secretions. Calculations in relation to total IgG showed that the IgG BA response to injections was unlikely to have been produced in the same place as the IgG BA produced in response to natural exposure. The findings show that the ab response to pollen antigens is truly local. The possibility that serum IgG BA and IgE BA was produced predominantly in local lymph nodes whereas the BA found in nasal secretions was derived from plasma cells in the nasal mucosa is discussed.

Inhibition of Antibody Production in Plasmacytoma Cells

Abstract Previous studies have demonstrated that highly multivalent DNP conjugates of γ-globulins, but not of KLH or BSA, inhibit the production of anti-DNP IgA antibody by the mouse plasmacytoma line, MOPC 315, in vitro. The present experiments were designed to compare the fate of various DNP antigens, trace-labeled with 125I, in MOPC 315 cells. After continuous exposure of MOPC 315 cells to various antigens for up to 20 hr, suppressive DNP-globulin conjugates are extensively internalized by the cells whereas nonsuppressive DNP-KLH and DNP-BSA remain largely bound to the surface. These differences are not attributable to cooperative binding of DNP-globulins to Fc receptors, and, at least with DNP-KLH, are not due to major differences in the molecular charge of the antigens. Moreover, all internalized antigens persist within cells and are not extensively hydrolyzed for up to 18 hr of antigen-free culture. Electron microscopic studies show that internalized antigens are not enclosed within pinocytotic vesicles, and are often found close to the endoplasmic reticulum or Golgi complex. It is proposed that prolonged intracellular persistence of antigens may be a central mechanism in the inactivation of antibody-secreting cells as well as in B lymphocyte tolerance.

Factors affecting serum IgA antibody to Epstein-Barr viral capsid antigens in nasopharyngeal carcinoma

Irrespective of the ethnic origin of the patient, nasopharyngeal carcinoma (NPC), appears to stimulate the production of IgA antibodies against VCA. These antibodies are detected at high frequency and titres in sera from NPC patients but only rarely from control subjects. A majority of relapse-free survivors tested 1-12 years after radiotherapy (RT) sustain a detectable level of IgA anti-VCA. Serum titres of IgA anti-VCA remain relatively unchaged in individual NPC patients after RT, regardless of the disease evolution. These antibodies were detected in serum from one individual 9 months before NPC and the titre rose concomitantly with its clinical onset. Titres of IgA anti-VCA in multiple serum specimens from individual NPC patients, and in sera from different NPC patients, do not correlate with titres of IgG anti-VCA or with Serum IgA. It thus seems possible that the IgA anti-VCA in the sera of NPC patients might be largely derived near NPC. Apparently healthy individuals showing detectable IgA anti-VCA tend to aggregate in families of NPC patients. The distribution of siblings of these families who have the IgA anti-VCA reaction shows the binomial distribution expected for an autosomal recessive trait, implying the involvement of an autosomal recessive gene in the IgA anti-VCA response.

Local Production of IgG, IgA and IgE Antibody to Pollen Antigens in Patients with Hay Fever

Fifteen untreated patients with a presumptive diagnosis of hypertension were studied for periods up to 48 h with the Oxford continuous intra-arterial pressure recorder. In each case a denned protocol of rest and exercise (using an Elema bicycle ergometer) was followed before the patients left the hospital. Blood pressure was characterized in each patient by a 6-hourly daytime mean systolic and diastolic measurement, and by peak responses to increasing levels of exercise. The 6-hourly mean pressures were compared with the indirect clinic recordings, and a weak correlation was found. A stronger correlation was found on comparing the 6-hourly means with the response to maximal exercise. The four patients who showed electrocardiographic evidence of severe left ventricular hypertrophy responded to exercise in an identical fashion to the main group. It is suggested that blood pressure response to exercise may be a better indicator of raised blood pressure than a casual resting clinic recording in borderline subjects.

Defense Mechanisms Against Primary Influenza Virus Infection in Mice

Abstract To investigate the defensive roles and production of interferon and antibodies, C3H/He mice were subjected to various immunosuppressive treatments and infected with influenza virus. In infected normal control mice the pattern of pulmonary viral growth can be divided into three phases. The first phase is characterized by an exponential increase of virus titer, the second by a rapid decrease, and the third by a moderate decrease. At the time of transition from the first phase to the second in pulmonary virus growth, interferon could be detected in the tracheobronchial washings of infected mice, but neutralizing antibodies could not. In infected B cell-deprived mice and infected anti-µ-treated mice, the transition from the first phase to the second occurred without any detectable antibody production, and interferon could be induced in the early stage of infection. However, the pulmonary virus in these mice increased again exponentially until the death of the mice. In infected T cell-deprived mice which could not induce interferon, but produced IgM-neutralizing antibodies, the second phase was not observed after the first phase, but a transient plateau phase could be demonstrated, and then the pulmonary virus increased again exponentially until the death of the mice. In anti-γ-treated infected mice, pulmonary virus growth and production of interferon and neutralizing antibody were almost similar to those of infected normal control mice except for the absence of IgG neutralizing antibody production. Although anti-α-treated infected mice produced interferon and no IgA antibody, the transition from the first exponential increase of pulmonary virus to the second rapid decrease was seen, but then the virus increased exponentially again until the death of the mice. These results suggest that interferon plays an important role in the transition from the first phase to the second, and that T cells are required for interferon induction in mice infected with influenza virus. These data also suggest that IgA antibodies play an important role in the inhibition of virus propagation in the lungs after the disappearance of interferon. Moreover, infected T cell-deprived mice could produce only IgM neutralizing antibodies, but not IgG and IgA antibodies. Therefore, T cells are required for the production of IgG and IgA antibodies and eventually for defense functions in mice infected primarily with influenza virus.

IgA Antibody to House Dust Mite in Nasal Fluid

The biological significance of IgA antibody to inhalant allergens in nasal fluid was studied. The concentration of IgA antibodies to house dust mite in the nasal fluids which were collected by suction from patients with mite allergy, was measured by a double antibody technique. As a result, allergen-binding activity of nasal fluid antibodies to mite was found to suffice the participation of the IgA antibody in the local immunity to mite which is inhaled and deposited on the surface of the nasal mucous membrane. However the severity of reaction from nasal provocation with mite extract was found to be independent of the concentration of antibodies in nasal fluid. The binding activity of nasal fluid IgA antibody did not increase after immunotherapy with mite and found to be independent of the structure of serum or nasal fluid IgA. Moreover, on the basis that the nasal fluid-serum ratio of IgA antibody was much greater than that of IgG or IgE antibody to mite, the local production of IgA antibody to mite was strongly suggested. The biological significance of IgA antibody to inhalant allergens in nasal fluid was studied. The concentration of IgA antibodies to house dust mite in the nasal fluids which were collected by suction from patients with mite allergy, was measured by a double antibody technique. As a result, allergen-binding activity of nasal fluid antibodies to mite was found to suffice the participation of the IgA antibody in the local immunity to mite which is inhaled and deposited on the surface of the nasal mucous membrane. However the severity of reaction from nasal provocation with mite extract was found to be independent of the concentration of antibodies in nasal fluid. The binding activity of nasal fluid IgA antibody did not increase after immunotherapy with mite and found to be independent of the structure of serum or nasal fluid IgA. Moreover, on the basis that the nasal fluid-serum ratio of IgA antibody was much greater than that of IgG or IgE antibody to mite, the local production of IgA antibody to mite was strongly suggested.

Complement in the Induction of IgA and IgE Antibody Production

Abstract The IgA-antibody forming cell response to sheep erythrocytes in BALB/c and CBA mice depleted of complement by cobra factor (CoF) was suppressed to 10 to 40% of the level in controls. The IgE response to alum-precipitated ovalbumin was measured in NIH mice by passive cutaneous anaphylaxis in rat skin. Reaginic antibody was scarcely detectable in CoF-treated animals whereas controls had titers of up to 1:2000. Fourteen days after boosting with ovalbumin, control mice had peripheral eosinophilia whereas animals which had received CoF before priming 6 weeks earlier had levels only minimally raised above normal. Studies of the generation of T cell blasts, “education” of thymus cells, and priming of splenic T cells indicated that activation of T cells by antigen proceeded normally under conditions in which thymus-dependent antibody production was suppressed by in vivo complement depletion. Pure sheep IgG anti-mouse C3 and pure F(ab′)2 anti-mouse C3 antibody both inhibited the in vitro T-dependent antihapten response of mouse spleen cells to TNP-KLH.

Studies of Escherichia coli O antigen specific antibodies in human milk, maternal serum and cord blood.

Studies of Escherichia coli O antigen specific antibodies in human milk, maternal serum and cord blood. Acta Paediatr Scand, 65:216, 1976.--The quantity and class specificity of E. coli O antibodies in human milk, maternal serum and cord blood was determined by the enzyme-linked immunosorbent assay. The predominant Ig-class of these antibodies in milk was IgA. The initially high levels of antibodies decreased 10-fold during the first days, but there seemed to be a fairly constant daily production of IgA antibodies during the first two months of the nursing period. There was no obvious decline in antibody content from morning to night or during a meal. The ratio milk antibody/serum antibody was very high for IgA, suggesting a local production. The ratios for the IgG were all less than 1, suggesting a restricted transfer from the serum. Ratios around 1 for IgM did not exclude some local production. In umbilical cord serum the amount of IgG E. coli O antibodies was higher than that in maternal serum. Small amounts of IgM and IgA antibodies was also demonstrated in some cases. In milk as well as in serum there were antibodies against a wide variety of E. coli O groups, not only to the actual E. coli strain dominating the mother's gut flora.

Antibody-forming cells in human colostrum after oral immunisation.

IT has been suggested that feeding infants with human milk reduces the incidence of gastrointestinal infections1. Although macrophages, lymphocytes and antibodies in the milk are probably important in this protection2,3, the mechanism for the development of specific immunity to enteric pathogens is not known. Human colostral cells synthesise IgA (ref. 4), and we have used the haemolysis-in-gel technique to demonstrate that human colostrum contains numerous cells which produce IgA antibodies against the O antigens of commonly encountered Escherichia coli bacteria5. This suggested that the sensitised lymphocytes had either undergone antigen-induced clonal proliferation within the mammary gland or had homed to that organ from another site after becoming sensitised. Because of reports that gastrointestinal immunisation of germ-free mice led to local6 and systemic7 production of IgA antibodies, and that cells from Peyer's patches effectively repopulate the ileum of lethally-irradiated rabbits with IgA-producing cells8, we have examined the effect of gastrointestinal immunisation on the development of antibody-producing cells in colostrum of humans. We found that oral administration of a non-pathogenic strain of E. coli led to the rapid appearance of colostral cells producing antibodies against the O antigen of the organism.

Comparative studies on human and rabbit exocrine IgA antibodies to an albumin

To study some mechanisms involved in mucoantibody production and to compare immunological activities of exocrine IgA antibodies and IgG antibodies, we isolated albumin from the blood serum of Alligator mississippiensis, alligator serum albumin (ASA), and used it to immunize rabbits and one human. The human subject responded to intranasal administration of ASA with the delayed production of IgG and IgA antibodies in his nasal fluids and serum. Most of the nasal IgA, and some of the nasal IgG antibodies were produced and secreted locally in the nose, as indicated by the results of measuring primary antigen-binding IgA and IgG separately. The pattern of responses to subsequent administrations of intranasal ASA moreover suggested the operation of immunological memory in the production of human exocrine IgA antibodies. The human nasal exocrine IgA and serum IgG antibodies were compared with rabbit colostral exocrine IgA and serum IgG antibodies produced in response to intramammary or parenteral immunization with ASA. Antibody concentrations were measured in terms of antigen-binding capacity by ammonium sulfate coprecipitation. Relative avidities of the antibody preparations, as reflected in the effect of antigen concentration on antigen-binding capacities, were similar for the various preparations. Nevertheless, exocrine IgA antibody was more efficient in agglutinating passively sensitized red blood cells than the IgG antibodies. Human and rabbit IgG antibodies precipitated ASA from solution, unlike either human or rabbit exocrine IgA antibodies. Low concentration of polyethylene glycol however induced precipitation of immune complexes composed of human, but not rabbit exocrine IgA antibodies. A probable explanation for this difference was found in experiments on the ability of antibodies to cross-link ASA molecules, using an insoluble immunoadsorbent containing ASA. Human exocrine IgA antibody that was combined with insoluble antigen was at least effective as IgG antibody in binding soluble ASA molecules, whereas rabbit exocrine IgA antibody was far less able to cross-link. Although these experiments do not allow a decision as to whether monogamous polyvalency, steric hindrance, or some other mechanism can explain the inability of the rabbit exocrine IgA antibody to cross-link ASA molecules, these findings suggest fundamental differences between the exocrine IgA antibodies from different species with respect to their reactions with polyvalent antigens.

The induction and characterization of secretory IgA antibodies.

In addition to humoral immunity, which is characterized by the presence of circulating antibody, a second distinct type of immunity involves the localized production of secretory antibodies (reviews 1,2,3). Numerous investigators, employing different antigens in a wide variety of animal systems, have been able to stimulate localized antibody production (4–13). The characteristic structural features of secretory IgA (review 3), which allows it to be differentiated from its serum counterpart, make IgA the ideal immunoglobulin class to be used in studies on the inductive process. Of particular relevance to our current studies was the demonstration that local immunization, by the intramammary route, with either dinitrophenylated bovine gamma globulin (6) or group A streptococci (8) could induce specific colostral IgA antibodies. Other evidence strongly suggested that gut associated lymphoid tissue was not only involved in the production of intestinal IgA antibodies (5, 9, 11, 13), but also might play a role in IgA production at remote sites (7, 9, 10, 12) including secretory sites (10, 12, 14).

Human nasal exocrine IgA antibody: formation and some activities.

Evidence was sought for anamnestic attributes of exocrine IgA antibody production. A human was immunized with a protein antigen to which prior exposure was unlikely, serum albumin from Alligator mississippiensis (ASA). One mgN of ASA, diluted in saline and delivered repeatedly as nose drops during 5 consecutive days, constituted one “dose”. Values for antigen binding by IgG, IgA and IgM were measured separately by radioimmunoprecipitation.

The Immunosuppressive Effect of Rabbit Anti-Mouse Macrophage Serum

Summary The suppressive effect of anti-macrophage serum (AMS) on the serum antibody response to trachoma agent antigens was studied in white Swiss mice. When given on days -1, 0 and +1, the antigen being given on day 0, AMS caused clumping of peritoneal macrophages but not peritoneal lymphocytes or spleen cells and interfered with the phagocytic activity of peritoneal macrophages. AMS also inhibited the primary immune response to trachoma agent antigens and specifically inhibited the production of IgM, IgA and IgG2 (7S γ2) antibodies.