INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Humoral immunity in human secretions is mainly associated with polymeric immunoglobulin A (IgA) bound to the secretory component as secretory IgA (S-IgA) (11, 27, 35-37). This isotype is synthesized in the subepithelial stroma and is then actively transported throughout epithelial cells with the help of the transmembrane form of the secretory component, called polymeric Ig receptor. The antibody activity of human S-IgA can differ from that of human serum IgA (32) because the secretory and systemic immune systems are largely independent (17). The secretory immune system comprises inductive sites of mucosa-associated lymphoid tissue and effector sites with dispersed Ig-producing cells. As reviewed by Brandtzaeg and Haneberg (14), it has been hypothesized that human mucosa-associated lymphoid tissue includes different compartments from which antigen-primed B cells migrate to different effector areas. In normal adults, cells from Waldeyer's ring would form nasal gland-associated lymphoid tissue and preferentially emigrate to lacrimal, nasal, salivary, and bronchial glands. Alternatively, cells from gut-associated lymphoid tissue are known to mainly migrate from Peyer's patches to the small intestinal mucosa or from the appendix and colonic-rectal follicles to the large intestinal mucosa. Cells of mammary glands might be issued from both nasal gland- and gut-associated lymphoid tissues, whereas those of the urogenital tract would originate from the appendix and colonic-rectal follicles.

In addition to S-IgA, human secretions also contain a variable amount of IgG (47), which is usually considered serum derived. Indeed, serum IgG antibodies seem to translocate throughout the epithelium of the lungs (41), nasal mucosa (48), gingival sulcus (46), and endometrium (25), and a transient paracellular translocation of serum proteins has been described after minor irritation of the mucosal surface of both airways (42) and the gut (43). Increased diffusion can occur during mucosal (15) and glandular inflammations, and a large release of IgG from the serum to the gut lumen via the biliary tract is the normal method of catabolism of serum IgG (45, 49). The serum origin of IgG in the gut lumen has been proved by injection of radiolabeled molecules (49), whereas this origin has been assumed for other secretions by detection of tetanus antitoxins (8, 25, 45), which are considered good markers of serum-derived immunity. However, these data have not ruled out the possibility of an additional immune response, with local IgG displaying a specificity independent from that of the systemic immune system, as suggested by the higher IgG/albumin ratio in secretions than in serum (8, 40).

Percentages ranging from 3% (duodenum-jejunum) to 17% (nasal glands) IgG-producing immunocytes have been observed for mucosal and glandular tissues in the absence of inflammation (12). Higher percentages, corresponding to a majority of Ig-positive immunocytes, have even been reported for the endometrium (5). Whether these cells belong to the systemic immune system or are associated with mucosal immunity is still undetermined, but it has been found that IgG-producing cells in normal mucosa contain J chains (6, 9), like polymeric IgA- and IgM-producing cells. Differences between salivary or vaginal IgG and serum IgG have been reported in terms of the proportions of the four  subclasses among total IgG molecules (25, 46) or among IgG antibodies of known specificities (18). Finally, the number of local IgG-producing cells has been found to increase markedly during inflammatory bowel diseases. Many of the locally produced antibodies are directed against fecal anaerobic bacteria (reviewed in reference 15), as demonstrated by their increasing specific activity in contrast to the unchanged serum IgG activity (33). These quantitative variations may also suggest that qualitative differences in antibody specificities can exist between serum-derived and locally produced mucosal IgG antibodies.

A global investigation of the antibody reactivity against a large number of antigens has been rendered possible by the use of a computer-assisted immunoblot analysis (3). A similar method has allowed the investigation of antibody repertoires (20, 23, 24, 30, 34, 38, 39) and has recently demonstrated a clear-cut variation in repertoires between serum and autologous salivary IgA autoantibodies (44). This finding led us to use this method to analyze the activity spectrum of IgG antibodies directed toward surface antigens of a frequent pathogen and to compare the results obtained for serum and for autologous secretions.

In this study, we demonstrate the presence of local IgG antibodies in various secretions from healthy subjects. The specificities of these antibodies were found to differ from those of their serum counterparts, as shown by comparative analysis of the patterns of antibodies to surface proteins of Streptococcus pyogenes in serum and in various autologous secretions. These results indicate that local IgG is normally produced by cells largely independent of the systemic immune system.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Specimens. Fourteen serum, 11 saliva, 3 colostrum, and 9 genital fluid specimens were obtained from 14 healthy volunteers. Specimens from the same subject were collected simultaneously. Whole saliva, containing both salivary gland secretions and crevicular fluid, was obtained by simple spitting for 10 min several hours after meals. Vaginal fluid was collected by washing with 3 ml of saline, corresponding to an ~1:10 dilution of the neat fluid (2). Whole semen was allowed to coagulate for 1 h at 37°C. All specimens were centrifuged at 10,000 × g for 5 min. Secretions containing erythrocytes in the pellet (determined visually) or in the supernatant (determined with Hemastix [Bayer Diagnostix, Leverkussen, Germany) were eliminated. Supernatants were kept at 30°C until use and were submitted to an additional centrifugation after thawing.

IgG purification. IgG purification was carried out by incubation of the specimens for 1 h at 37°C with protein G-Sepharose (Pharmacia, Uppsala, Sweden). This Fc- and Fab-specific sorbent reacts with the sole IgG isotype irrespective of the  subclass (19). After several washes, the bound IgG was eluted with pH 2.9 0.05 M glycine HCl, neutralized with 4 M Tris-HCl, and then diluted with phosphate-buffered saline (PBS). The lack of IgG degradation during incubation with protein G was investigated by an additional 1 h of incubation at 37°C of immobilized IgG with PBS or with an undiluted vaginal fluid (from subject 14) showing no significant antistreptococcal activity. Indeed, serum contains a large amount of ₂-macroglobulin, a major multispecific protease inhibitor, whereas vaginal fluid and, to a lesser degree, saliva (which exhibits a much higher flow rate) may contain some proteolytic enzymes from resident bacteria or from lysed cells.

Microbial extracts. The growth of S. pyogenes D471 (group A, type 6 M protein) in Todd-Hewitt broth (100 ml) was monitored by measuring the absorption at 650 nm. Bacterial cells were harvested during the exponential phase of growth and collected by centrifugation at 10,000 × g for 15 min. Surface molecules (2-ml final volume) were extracted by incubation of the cells with purified group C streptococcal phage-associated lysin (21) as described previously (22).

Electrophoresis and Western blotting. Polyacrylamide gel electrophoresis was carried out in the presence of sodium lauryl sulfate as described previously (3). The gels contained 10% acrylamide, and the buffer system was that of Laemmli (30a). The extract was diluted twofold in sample buffer containing 2-mercaptoethanol. Human serum served as a molecular mass marker. After migration, proteins from the streptococcal extract or from the control lysin extract were transferred to nitrocellulose membranes (pore size, 0.45 µm; Schleicher & Schüll, Dassel, Germany) by a semidry isotachophoresis procedure. The sheets were dried at room temperature and kept dry until use. They were then cut into vertical strips, which were individually placed into incubation tray wells. The following steps were carried out at room temperature under constant shaking. Saturation took place by incubation with 0.3% (vol/vol) Nonidet P-40 in PBS for 30 min and then with 0.03% (wt/vol) gelatin in PBS for 1 h. After a wash with 0.1% (vol/vol) Tween 20 in PBS, the strips were incubated overnight with antibodies diluted in the gelatin solution. After being washed with PBS-Tween, the strips were incubated for 1 h at room temperature with 0.1 mg of purified IgG per ml and then were washed again with the same buffer. The bound antibodies were detected with peroxidase-labeled sheep antibodies against the human Fc fragment diluted in PBS containing 0.1% (vol/vol) Tween 20 and 0.03% (wt/vol) gelatin. After a wash with PBS-Tween 20, peroxidase activity was revealed with 0.03% (wt/vol) diaminobenzidine HCl (Sigma Fast) enhanced with nickel chloride. A control blot of serum proteins migrating in a separate well was not saturated but was stained with India ink. Another control strip was incubated with a monoclonal antibody to the M protein (28), and antibodies were revealed with a peroxidase-labeled antibody to mouse Ig.

Computer-assisted analysis of the Western blots. The bands revealed by peroxidase were integrated on the basis of optical densities with a high-resolution charge-coupled device camera system connected to a densitometer (Masterscan; Scanalytics, Billerica, Mass.) and to a personal computer. The RFLPscan program (Masterscan) was used to manipulate the camera data. Integration was carried out under visual control and was corrected by subtraction of the values for the adjacent control strip. A calibration curve was constructed by reference to the standards stained with India ink, allowing us to determine the molecular mass of each detected band. The antibody activity spectrum was then represented as a curve of optical density values (in arbitrary units) versus calibrated molecular mass (in kilodaltons).

Presence of IgG antibodies to type 6 M protein. To investigate statistically the lack of a relationship between serum-derived IgG and IgG found in secretions, the patterns of serum- and autologous secretion-derived antibodies were compared for their reactivity to the M-protein molecule. This antigen was selected because of its major pathogenic interest, and it was easily identified in the assay. In the event of a selective serum origin of IgG, all the pairs would be concordant for this band. The percentage observed was thus compared to 100% by use of the chi-square test. Because the pattern analysis was carried out at a set concentration of IgG (0.1 mg/ml), a discrepancy between two specimens from the same individual would correspond to major variations in terms of specific activity.

IgG and albumin quantitations. IgG and albumin quantitations were carried out by an enzyme-linked immunosorbent assay. The plates were coated with commercial unlabeled rabbit antibodies, and the capture molecules were revealed with the same antibodies labeled with peroxidase (Biosis, Compiègne, France). The IgG/albumin ratio in serum and secretions was compared by use of the two-tailed unpaired Student's t test.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Specificity patterns of serum IgG antibodies to streptococcal surface antigens compared with local IgG antibodies. Both qualitative and quantitative variations were observed when adjacent Western blot strips were incubated with serum IgG or IgG from different secretions from the same individual: saliva (Fig. 1), cervicovaginal secretions (Fig. 2), seminal fluid (Fig. 3), and colostrum (Fig. 4). The serum pattern varied according to the individual and was weak or negative in subjects 2 (Fig. 1), 3, 5, and 11 (Fig. 2), and 14 (data not shown). Some bands were detected very often, but their identification was uncertain, except for that of the streptococcal M protein (~65 kDa), which was identified as a single peak with the help of a specific monoclonal antibody. In control experiments, different saliva specimens were collected at 3-day intervals to investigate the reproducibility of the method. In the two subjects examined, the pattern was found to be highly reproducible. Similarly, interference as a result of IgG cleavage by proteases present in the secretions was ruled out, since the pattern was the same when IgG was incubated with or without cervicovaginal secretions of subject 14 (data not shown). This fluid was selected because of its high content of lysed cells and thus of proteases and because the specimen did not exhibit any IgG antistreptococcal activity which could interfere in the assay.

View larger version (24K): [in this window] [in a new window]   FIG. 1.   Reactivity of paired autologous saliva (broken line) and serum (solid line) IgG antibodies with surface antigens from S. pyogenes, as demonstrated by computer-assisted Western blot analysis. IgG was purified from 10 normal subjects (1 to 10). The ordinate corresponds to the intensity of bands detected by the antibodies, and the abscissa indicates the apparent molecular masses of the corresponding antigens, calculated from the electrophoretic migration reciprocal distances. Besides quantitative differences in favor of serum IgG (subject 1) or of saliva IgG (subjects 3 and 5), striking qualitative differences were observed among most subjects, many more bands being detected by saliva IgG (subject 4) or by serum IgG (subject 8). No reactivity was observed with serum IgG and saliva IgG from subject 2. The arrow indicates the position of the type 6 M protein (~65 kDa), as determined with a mouse monoclonal antibody. Reactivity with this protein was detected with serum IgG of subjects 1, 6, 7, 8, 9, and 10 and with saliva IgG of subjects 3, 4, and 5.

View larger version (27K): [in this window] [in a new window]   FIG. 2.   Reactivity of paired autologous cervicovaginal secretion (broken line) and serum (solid line) IgG antibodies with surface antigens from S. pyogenes, determined as described in the legend to Fig. 1. IgG was purified from subjects 1, 3, 4, 5, 11, 12, and 13. The reactivity of serum IgG from subjects 3, 5, and 11 was very weak or absent, whereas the corresponding cervicovaginal secretion IgG displayed well-defined patterns. In the other pairs, important variations were observed, many antigens being detected only by the cervicovaginal IgG or only by the serum IgG. Reactivity with the type 6 M protein was detected with all cervicovaginal IgG antibodies.

View larger version (35K): [in this window] [in a new window]   FIG. 3.   Reactivity of paired autologous seminal fluid (broken line) and serum (top panels) or saliva (bottom panels) (solid line) IgG antibodies with surface antigens from S. pyogenes, determined as described in the legend to Fig. 1. IgG was purified from subjects 6 and 7. In addition to common peaks shared by the serum and semen patterns, some antigens were detected only by semen IgG (subject 6) or only by serum IgG (subject 7). The patterns of saliva and semen IgG seemed to be unrelated.

View larger version (30K): [in this window] [in a new window]   FIG. 4.   Reactivity of paired autologous colostrum (broken line) and serum (top panels) or saliva (bottom panels) (solid line) IgG antibodies with surface antigens from S. pyogenes, determined as described in the legend to Fig. 1. IgG was purified from subjects 8, 9, and 10. Clear differences were seen in the patterns of the pairs. No apparent reactivity with the type 6 M protein was detected with colostrum IgG.

Comparison between antibodies from different autologous secretions. Most patterns from the same subjects were significantly different, no matter what pair of samples was compared: saliva-seminal fluid (Fig. 3), saliva-colostrum (Fig. 4), or saliva-cervicovaginal secretions (Fig. 5). In the case of paired saliva-vaginal secretion samples, the patterns were similar only for subject 4 (Fig. 5). Conversely, noticeable differences were observed for the patterns for subjects 1, 3, and 5. The possibility that the variations between autologous secretions could be due to interference with different levels of serum-derived antibodies was ruled out by further analysis of the results for subjects 3 and 5. The serum IgG profile for these subjects was flat (Fig. 1), whereas the reactivities of their autologous saliva and cervicovaginal secretions were both clearly elevated and quite different. Careful comparison of the other paired samples of secretions and autologous sera resulted in the same conclusion.

View larger version (18K): [in this window] [in a new window]   FIG. 5.   Reactivity of paired autologous saliva (solid line) and cervicovaginal secretion (broken line) IgG antibodies with surface antigens from S. pyogenes, determined as described in the legend to Fig. 1. IgG was purified from subjects 1, 3, 4, and 5. A close similarity between saliva and cervicovaginal patterns was observed only with IgG from subject 4. In the other pairs, many more antigens were reactive with cervicovaginal IgG than with saliva IgG of subject 1, while samples from subjects 3 and 5 showed multiple discordant reactivities.

Quantitative comparison between serum IgG and local IgG to a single antigen. Based on the hypothesis that IgG in secretions is serum derived, all IgG antibody patterns from a single subject would be expected to be identical or at least closely related. However, most patterns from paired samples were found to be different, suggesting an alternative explanation. To quantify the observed difference, we examined the reactivity to a single defined antigen, i.e., the type 6 M protein, by using the same concentration of IgG from each source. The comparisons were made for pairs in which at least one of the two patterns displayed a clear type 6 M-protein band. Of 15 serum-secretion pairs examined, only 3 were concordant for the presence of an M-protein peak. This result is statistically different (P <2 × 10⁴) from the 15 expected reactions if IgG were derived solely from serum. Similarly, investigation of diffusion of albumin from serum to secretions showed IgG/albumin ratios of 0.68 ± 0.03 for saliva (P << 10³) and 1.70 ± 0.59 for cervicovaginal fluid compared with 0.43 ± 0.03 for serum (P << 10³).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Investigating the specificity patterns of IgG antibodies to surface proteins of a common mucosal pathogen, S. pyogenes, we observed major variations between serum and autologous secretions. The observed differences revealed that the secretions contained IgG antibodies that were absent from serum and vice versa. These data are consistent with the hypothesis that a high percentage of IgG antibodies in secretions is of local origin and that serum IgG antibodies are poorly translocated through the corresponding mucosae.

The possibility of an unbalanced depletion of some antibodies by absorption by cross-reactive surface antigens found on other streptococcal species and present in normal secretions is an unlikely explanation of our findings for various reasons. Normal colostrum contains few microorganisms and can even be sterile. Alternatively, the well-documented presence of a rich microflora coated with Ig in saliva (13) does not seem to markedly impair the detection of antibodies to streptococci. For example, S-IgA to cell wall carbohydrates and to protein I/II from S. sobrinus, a caries-associated bacterial species, can be detected in the saliva of most subjects even in the presence of multiple dental caries (26). Similarly, IgG antibodies to Actinomyces actinomycetemcomitans can be detected in the crevicular fluid of patients who have periodontal disease associated with this anaerobic species (18). Moreover, nonspecific antibody depletion by indigenous bacteria could not explain why the serum IgG patterns from subjects 3 and 4 were negative or weak, whereas those of autologous saliva IgG were strongly positive. Indeed, the opposite result would have been expected if absorption were the reason for the observed differences. It is more likely that the microflora and local pathogens induced regional IgG responses instead of interfering with the detection of these antibodies. Thus, the differences observed in IgG antibodies at different sites may be a function of the local immune history of those sites with regard to a specific pathogen.

The variations found in the IgG antibody patterns to S. pyogenes in different body fluids were in agreement with the reported elevated IgG/albumin ratio in secretions (8, 40) and suggested that this IgG could not be derived solely from serum but was mainly comprised of locally produced antibodies. This idea suggests that IgG-secreting cells in the mucosa are similar to those observed in the human Waldeyer's ring (4, 10). This lymphoepithelial structure is formed by the palatine tonsils, nasopharyngeal tonsil, lingual tonsil, tubal tonsils, and lateral pharyngeal bands. It is considered to belong to the secretory immune system, but the proportion of B cells producing IgG can be as much as 65%, whereas that of IgA-producing B cells approaches only 30%. This ratio is constant despite differences in tonsillar area (16). The possibility of minor tonsil-like pathways outside the Waldeyer's ring is a reasonable hypothesis and is supported by the finding that the specific activity of IgG antibodies to human immunodeficiency virus can be higher in both saliva (31) and vaginal secretions (1) than in serum. It has already been reported that the zone adjacent to the lymphofollicles of the appendix contains numerous IgG-producing cells (7). A simple explanation is that systemic B cells selectively migrate toward areas containing the corresponding foreign antigen and therefore increase the specific activity of local IgG. However, our observation of different specificities for the same pathogen in paired secretions and serum from the same healthy subject is in favor of a more compartmentalized IgG mucosal system. Indeed, the differences between autologous patterns were found to be of the same order of magnitude as those between heterologous patterns, suggesting that systemic and local B cells are poorly related and that the rare IgG antibody peaks shared by autologous fluids were perhaps due to parallel responses to the same antigen by different immune compartments. It is likely that the local IgG response is mainly associated with antigen penetration by the mucosal route. Nevertheless, additional diffusion of some circulating antigens toward mucosae is also likely and may explain the simultaneous increases in the levels of cervicovaginal and serum antitoxins after tetanus vaccination by parenteral injection reported by our group (8). This alternative explanation leads us to reconsider the serum-derived origin of tetanus antitoxins in secretions and especially the significance of our previous results for vaginal IgG antibodies (25).

Locally derived IgG may differ from its serum counterpart by being better adapted to mucosal pathogens and therefore more efficient in microbial clearance. Thus, local IgG antibodies should be seriously considered during the analysis of immune protection induced by mucosal vaccines. Despite the fact that its concentration in secretions is lower than that of IgA, the IgG isotype can efficiently participate in antimicrobial defenses both as a first barrier against pathogens in the lumen and as a second barrier if the mucosa is breached. Most IgG molecules trigger the complement cascade, activate polymorphonuclear leukocytes, and arm cytotoxic cells. The involvement of neutralizing IgG antibodies in IgA-containing immune complexes can also protect epithelial cells from infections during the secretory component-dependent transport of pathogens toward the lumen (29). The concept of a local IgG response is thus of major interest in the context of local vaccinations. Induction of IgG-associated regional immunity may provide a permanent in situ defense of the mucosa against invasion and may complement S-IgA as an immune barrier to mucosal pathogens.