Main

Oral tolerance is defined as the systemic unresponsiveness to a previously ingested Ag when encountered on the parenteral route(1–3). Small quantities of orally administered proteins escape enzymatic digestion in the intestine and can be detected in the circulation antigenically intact(2, 4). It is thus an important physiologic phenomenon in preventing immune reactions to dietary protein Ag. Oral tolerance is known to suppress IgE Ab responses involved in the development of food-hypersensitivity reactions(5–7). Young children are more frequently subject to food hypersensitivity than adults, particularly during the first year of life, although the reasons are still not perfectly understood(7, 8). Adults and young children differ substantially in the composition and stability of the gut microflora, found mainly in the colon(9, 10). In the adult, the composition is remarkably stable. In the newborn, bacterial colonization follows a sequence, and also depends largely on type of nutrition-breast or formula feeding(11). At weaning, the switch of diet results in wide-ranging changes to the gut flora equilibrium(10). This could alter the colonization resistance and predispose the child to enteric infections(12), especially by enterotoxigenic bacteria. At weaning, countless children throughout the world suffer from diarrheal diseases due to enterotoxigenicEscherichia coli(13–15). The transient presence of enterotoxins in young children may thus interfere with the physiologic process of oral tolerance to new dietary proteins and result in food hypersensitivities.

Studies on the mouse have shown that bacterial factors such as enterotoxins can affect the induction of oral tolerance. In conventional mice orally co-administered CT + Ag or LT + Ag, the serum IgG Ab responses were almost identical to those of the saline-administered controls, demonstrating that oral tolerance had been abrogated(16, 17). More recently, Snider et al.(18) showed that oral co-administration of CT and a protein Ag primes mice for systemic IgE Ab response to the Ag.

On the other hand, the indigenous gut flora also influences oral tolerance(19). The extended absence of serum-specific IgG Ab after systemic immunization shows oral tolerance to be a long-lasting process(20). In our laboratory, oral tolerance was also found in conventional mice 3 mo after tolerogenic feeding, whereas in germ-free mice it lasted no more than 20 d(19) (M. C. Moreau, manuscript in preparation). The above results show the gut flora to be an important environmental factor in the persistence of Ab unresponsiveness after Ag feeding, and suggest that it could also play a regulatory role in the oral tolerance process.

The present study examines the short- and long-term effects of the gut flora on CT- or LT-mediated abrogation of oral tolerance to a dietary protein, OVA. Both conventional and germ-free mice were used, and serum anti-OVA IgG and IgE Ab were measured to monitor the toxin-mediated effects. Whereas abrogation persisted in germ-free mice, in conventional mice it was transient, indicating a critical role of gut flora.

METHODS

Animals. Conventional and germ-free C3H/He female mice (5-10 per group) were used. The animals were bred in our Institute, and were between 6 and 8 wk old at the start of the experiment. The germ-free mice were bred in a Trexler plastic isolator fitted with a rapid transfer system (La Cahlène, Velizy, France). The conventional mice were bred under normal animal-housing conditions. All mice were fed ad libitum on a commercial diet (R03-40, UAR, Villemoisson-sur-Orge, France) sterilized byγ-irradiation (40 kGy).

Antigen and toxins. OVA (grade V) and CT were obtained from Sigma Chemical Co. (St. Quentin Fallavier, France). LT was obtained from ICN Biochemicals (Orsay, France).

Feedings and immunizations. The experimental procedure was the same for conventional and germ-free mice (Fig. 1). Mice were force-fed with Ag and/or toxins in 0.5 mL of 0.2 M Bic using blunt-tipped feeding needles. Oral inoculations consisted of a dose of 10 μg of CT + 20 mg of OVA [(CT + OVA)-fed groups] administered either once or twice, the second dose being 1 wk later. Three control groups were fed respectively with single doses of 20 mg of OVA (OVA-fed group), 10 μg of CT (CT-fed group), or 0.5 mL of Bic alone (Bic-fed group). All solutions were sterilized by filtration through a 0.45-μm membrane (Analypore, Osi, Paris, France). Similar experiments were performed with an equimolar amount of LT (10μg).

Figure 1
figure 1

Protocol of feedings and immunizations. Conventional and germ-free mice were fed CT and LT as follows: a and b= 10 μg of toxin + 20 mg of OVA in 0.5 mL of Bic [(toxin + OVA)-fed groups]; c = 20 mg of OVA in Bic (OVA-fed group); d = 10μg of toxin in Bic (toxin-fed group); e = Bic alone (Bic-fed group). All groups were immunized i.p. as follows: OVA i.p. = 10 μg of OVA in alum; OVA i.p.* = 50 μg OVA in alum. Bl = blood samples taken for ELISA.

Full size image

To study the kinetics of the IgG and IgE Ab responses, the mice were immunized by intraperitoneal (i.p.) injection as follows. On d 0, 1 wk after the last oral dose: 10 μg of OVA absorbed by 1 mg of alum; on d 14:50 μg of OVA in 4 mg of alum; and on d 34: 10 μg of OVA in 1 mg of alum. Conventional mice were immunized with 10 μg of OVA in 1 mg of alum once again between d 60 and 70.

Blood samples were taken from the retro-orbital plexus after anesthetization by intramuscular injections of 0.1 mL of Imalgene 500(Rhône Mérieux, Lyon, France) diluted in the same volume of saline. The blood was centrifuged for 10 min at 3000 rpm, and sera were stored at -20°C until assayed.

Antibody titrations. Anti-OVA IgG Ab were assayed by ELISA as described earlier(19). Briefly, OVA (Pierce, Montluçon, France, 10 μg/mL of 0.1 M Bic buffer) was coated onto flat-bottomed microtiter plates (Immulon II, Dynatech, Guyancourt, France). The wells were then incubated, first with duplicated serial dilutions of the samples, then with sheep anti-mouse IgG coupled to peroxidase (Diagnostics Pasteur, Marne la Coquette, France). Finally, o-phenylenediamine(Sigma Chemical Co., 0.5 mg/mL of 0.05 M citrate buffer) was added, and the enzymatic reaction was stopped by 4 N H2SO4. Plates were read at 490 nm in a Titertek Multiskan ELISA Reader (Flow Laboratories, McLean, VA). The IgG concentration in each sample was calculated with reference to purified polyclonal mouse anti-OVA IgG Ab prepared as described by Saklayen et al.(21).

Anti-OVA IgE Ab were assayed by ELISA. The wells of the Nunc plate were coated with 100 μL of OVA (Pierce; 50 μg/mL of 0.1 M Bic buffer). After saturation with BSA (Sigma Chemical Co.; 0.2% in PBS) for 1 h at 37°C, the wells were incubated with 100 μL of serial twofold dilutions of samples for 2 h at 37°C. Serum from mice immunized with the first two i.p. doses of OVA and alum was used as a standard on each plate. Bound IgE Ab were revealed with peroxidase-conjugated rat MAb (LO-ME-2, Biosys, Compiègne, France). Finally, 100 μL of TMB Microwell Peroxidase Substrate System (KPL, Gaithersburg, MD) were added to each well, and the enzymatic reaction was stopped by 50 μL of 4 N H2SO4. Absorbencies were read at 450 nm. Anti-OVA IgE titers were expressed as (OD ratio of sample to standard at 1/80th dilution) × 100.

Statistical analysis. Data are expressed as means ± SD of groups of 5 to 10 mice. Variances were homogenized by logarithmic transformation, and statistical differences analyzed by two-tailed t test. Given that four independent experiments were carried out, i.e. CT or LT with conventional or germ-free mice, data were compared only within each experiment.

RESULTS

Effect of feeding (CT + OVA) on systemic anti-OVA Ab responses. Doses of 10 μg of CT plus 20 mg of OVA were orally administered to conventional and germ-free mice, and anti-OVA IgG and IgE Ab levels were quantified on d 23, i.e. 10 d after the second i.p. immunization with OVA. As shown in Tables 1 and 2, respectively, significantly higher anti-OVA IgG and IgE Ab responses were detected in both conventional and germ-free mice fed (CT + OVA) than in those fed OVA alone(p < 0.025). The Ab levels were not significantly different from those of the Bic-fed groups. The Ab responses in the conventional mice fed (CT+ OVA) twice at a 1-wk interval (data not shown) did not differ significantly from those obtained from mice fed once.

Table 1 Anti-OVA IgG Ab responses after oral co-administration of OVA with either CT or LT in conventional and germ-free mice
Full size table
Table 2 Anti-OVA IgE Ab responses after oral co-administration of OVA with either CT or LT in conventional and germ-free mice
Full size table

Effect of feeding (LT + OVA) on systemic anti-OVA Ab responses. The same experiments were done with the related enterotoxin LT. In germ-free mice (Tables 1 and 2), the anti-OVA IgG Ab level after one oral dose of (LT + OVA) was about 7.5 greater than that of the Bic-fed group (p < 0.0025). The response observed in the LT-fed group was not significantly different from that of the Bic-fed, suggesting that oral LT administration did not modify the OVA-specific immune response after i.p. immunization. No hyperstimulation of the IgE Ab response was observed: the Ab response in (LT + OVA)-fed mice did not differ significantly from that of the Bic-fed.

In conventional mice (Tables 1 and 2) fed once with (LT + OVA), the Ab responses were not significantly different from those in the OVA-fed group. Two (LT + OVA) doses at a 1-wk interval were necessary to induce anti-OVA IgG and IgE Ab responses, but they did not differ significantly from those of the Bic-fed group. No hyperstimulation was observed. A difference in the LT-mediated effect on the systemic Ab responses between germ-free and conventional mice was thus observed.

Kinetics of the anti-OVA IgG Ab response. The kinetics of the specific IgG Ab responses were observed from d 0 to 55 in germ-free mice, and up to d 90 in the conventional, i.e. beyond a third i.p. immunization. The experimental data are presented in Figs. 2 and 3.

Figure 2
figure 2

Kinetics of IgG anti-OVA Ab responses in germ-free mice. Four groups of 5 to 10 mice (see Table 1) tested with CT (A) or LT (B), were fed with OVA (-•-), toxin+ OVA (-○-), toxin (-□-), or Bic (-▪-), 7 d before i.p. immunization (arrow) with OVA (d 0). For immunization and blood samples, see Fig. 1. Each point represents the mean(±SD). **p < 0.0025 compared with Bic-fed group. Data are representative of two similar experiments.

Full size image
Figure 3
figure 3

Kinetics of IgG anti-OVA Ab responses in conventional mice. Four groups of seven to nine mice (see Table 1) tested with CT (A) or LT (B), were fed with OVA(-•-), toxin + OVA (-○-), toxin (-□-), or Bic (-▪-), 7 d before i.p. immunization (arrow) with OVA (d 0). In (B), the toxin + OVA group (-○-) was fed twice, 14 and 7 d before i.p. immunization (arrow). For immunization and blood samples, see Fig. 1. Each point represents the mean (±SD). Differences between toxin + OVA and Bic are given for two representative points: *p < 0.025, **p < 0.001. Data are representative of two similar experiments.

Full size image

In germ-free mice fed once with either (CT + OVA) or (LT + OVA) (Fig. 2,A and B, respectively), the third i.p. immunization with OVA on d 34 resulted in a further increase in OVA-specific IgG Ab response, similar to that observed in the Bic-fed.

On the other hand, in conventional mice fed either once with (CT + OVA) or twice with (LT + OVA), the third i.p. immunization with OVA resulted in a lower anti-OVA IgG Ab response than that of the Bic-fed mice[ Fig. 3,A and B (CT + OVA) on d 55; (LT + OVA) on d 43, both p < 0.025]. The difference was further corroborated after a fourth i.p. immunization with OVA: the IgG Ab responses in the (CT + OVA) and(LT + OVA) groups remained at the same level as after the third i.p. injection(d 78, p < 0.001). Thus, a hyporesponsive state was observed in the conventional mice as from about 40 d after the (CT + OVA) and the second(LT + OVA) feeding.

DISCUSSION

The present comparison of conventional and germ-free mice showed that the presence of gut flora was not necessary for CT- and LT-mediated abrogation of oral tolerance (i.e. systemic, or peripheral tolerance to orally administered Ag), although it could modulate it. When given orally with OVA, both CT and LT elicited serum IgG and IgE Ab responses in conventional and germ-free mice. However, in the presence of indigenous gut flora, the CT- and LT-mediated abrogation of oral tolerance was short-lasting.

When orally co-administered with an Ag, both CT and LT abrogate specific systemic IgG Ab unresponsiveness in conventional mice(16, 17, 22). The same was found to occur in germ-free mice. The bacterial enterotoxins found in the digestive tract during infectious diarrhea are transient. We therefore examined whether a single oral co-administration of toxin and OVA could abrogate IgG Ab unresponsiveness. We found that a single feeding with CT and OVA was effective not only in conventional mice, as previously observed by Pierre et al.(22), but also in germ-free mice. This therefore suggests that the gut flora is not directly involved in the toxin-mediated abrogation of oral tolerance.

Comparing the LT-mediated effects in germ-free and conventional mice suggests that the gut flora could protect the individual from the toxin's abrogating effect. Whereas abrogation of specific IgG Ab unresponsiveness in the germ-free mouse required one oral dose of LT + OVA, in the conventional it required two. It may be conjectured that the gut flora modifies the expression of toxin-receptors and/or the proteolytic degradation of LT. However, the same results were not observed with CT. This suggests that, although CT and LT are structurally and functionally related [reviewed in Spangler(23)], their effect on oral tolerance is unlikely to be due to quantitative factors, at least under conventional conditions. The greater effectiveness of CT may result from its specific binding to GM1 ganglioside receptors, or from a variable ADP-ribosylating capacity [reviewed in Spangler(23)]. In germ-free mice, on the other hand, 10 μg of LT not only abrogated oral tolerance to OVA, but also resulted in a higher Ab response. It may therefore be proposed that a priming effect was involved. In conventional mice, Verma et al.(24) have shown that LT increases intestinal permeability to macromolecules, and Ag uptake into the circulation. Similarly, the absence of gut flora is known to modify various anatomic and functional properties of the intestine, one being the slowing down of peristalsis(9, 25). If so, increasing the time available for LT to have this effect could result in oral priming. With CT, on the other hand, neither Ag uptake into the circulation nor intestinal permeability to Ag in experimental secretory diarrhea increases(18, 26). This may explain some of the differences between CT and LT in abrogating oral tolerance. Given that one factor suspected in the development of food hypersensitivities is an increased intestinal permeability to macromolecules(24, 27), the gut flora's protective role in LT-induced intestinal permeability should therefore be examined in greater detail. This could be of clinical importance, all the more so because infectious diarrhea in children is frequently associated with enterotoxigenic E. coli(13, 14).

Snider et al.(18) reported that the presence of CT on oral immunization with a protein Ag can sensitize animals for systemic IgE Ab response. We found that oral administration of CT or LT with a tolerogenic dose of OVA could abrogate the specific IgE Ab hyporesponsiveness in both conventional and germ-free mice. IgE Ab are involved in the development of food hypersensitivities(5–7). Hence, to understand the causes of persistent, noninfectious diarrhea in the infant(15, 28), investigating the influence of enterotoxins on oral tolerance to dietary proteins and the development of food hypersensitivities could prove worthwhile.

Despite the toxin-mediated breakdown of oral tolerance, IgG Ab hyporesponsiveness did occur in conventional mice, but did not in the germ-free. This could not be due to excess OVA immunization. Toxin-mediated effects may well exist, but they were negligible compared with those of the gut flora. We therefore propose that not only is the gut flora important for the persistence of oral tolerance(19), but may also be for its recovery after a transient, enterotoxin-mediated breakdown. Although the mechanisms governing oral tolerance are not fully understood(2, 3), the present data suggest that the gut flora plays a critical role. A similar importance of gut flora is also found in the maturation and development of many lymphoid cells of the gut-associated lymphoid tissue(10). Similarly, a bacterial fermentation product, butyrate, was recently shown to influence the secretory IgA Ab response(29). Nevertheless, the relationships between gut flora, mucosal IgA immune response, and oral tolerance are still poorly understood. In young children with weaning diarrheas due to enterotoxigenic E. coli, the normal gut flora could facilitate the recovery of oral tolerance after the transient presence of enterotoxins and dietary Ag in the digestive tract. Further experiments need to be conducted on the smaller quantities of toxins synthesized by enterotoxigenic strainsin situ. Because germ-free mice are sensitive to toxin-mediated effects, gnotobiotic mice whose digestive tract can be colonized by identifiable bacterial strains, such as enterotoxigenic Vibrio cholerae or E. coli, could be useful models.

CT is used in the development of oral vaccines due to its mucosal and systemic adjuvant activities(30, 31). Our results with OVA have shown the stimulated Ab response to be short-lived, on the systemic level at least. This should be taken into consideration when using these enterotoxins in long-term oral vaccine strategies.

Based on the present results, we believe the gut flora to be an important contributor to the functioning of intestinal immunity. In treating intestinal diseases, particularly of the young, maintaining its integrity must be taken into account.