Maternal Milk Contains Antimicrobial Factors That Protect Young Rabbits from Enteropathogenic Escherichia coli Infection

Experimental infections.This experiment was carried out with PS Hyplus 19 × PS Hyplus 39 rabbits (Grimaud group) from a breeding unit free of E. coli contamination. Litters, normalized to nine kits, were weaned either at 21 days (W21 group, five litters) or 35 days of age (W35 group, five litters). The environmental temperature was maintained at 19 ± 3°C. The lighting schedule was 7:00 to 23:00 h with the lights on. From 14 days of age, young rabbits had access to an experimental feed (on an as-fed basis, crude protein, 17.3%; nondigested fibers, 34.3%; starch, 9.1%), whereas does were fed with a commercial feed. None of the diets was supplemented with antibiotics. Water and feed were provided ad libitum. At this time, the does were separated from their pups, except for suckling. The does were moved to new cages, whereas young rabbits remained in their birth cage until the end of the experiment (63 days of age), unless they were moved to new cages to respect density standards. Indeed, if more than five rabbits from the same litter reached 35 days of age, half of the litter was moved to another cage by alternating the weight ranks. Does were not inseminated during the suckling period. Animals were handled according to the care of animals in experimentation, in agreement with French national legislation (decree 2001-486, 06/06/2001), and the study received approval from the regional ethical committee of Midi-Pyrenees (France).

Bacterial strains.An EPEC strain belonging to the O103:H2:K− serotype (strain E22) was used. This strain, described previously (4, 8, 30), does not metabolize rhamnose and is AF/R2 fimbria positive. REPEC strain E22 was cultured overnight at 37°C with shaking in Luria-Bertani (LB) broth. All young rabbits were inoculated at 28 days of age by the oral route with 1 to 2 ml of a phosphate-buffered saline (PBS) solution containing 10⁴ CFU per ml, which corresponds to 1 50% lethal dose of this strain (4). The rabbits in the W35 group were infected 3 h after suckling.

For the inhibition of adhesion assay, we used a genetically modified pEGFP-E22 strain (5). pEGFP-E22 was cultured in Penassay broth (antibiotic medium 3; Difco) supplemented with 25 μg/ml of carbenicillin, as described previously (5).

Clinical follow-up.From birth until the time of infection, the mortality of the rabbits was controlled every day and each rabbit underwent a clinical examination twice a week. From 28 days onwards, mortality checking was done at least twice a day, and a clinical examination was performed each morning. Prostration, loss of appetite, diarrhea, and dehydration were systematically checked for. When required, to shorten the length of suffering, the rabbits were euthanized after anesthesia with an intramuscular injection of ketamine (Imalgène 1000), followed by an intracardiac injection of a solution containing embutramide, mebezonium, and tetracaine (T 61). An autopsy was carried out for each dead rabbit to detect macroscopic lesions. The cecal content was sampled for bacterial analysis (see below). The enumeration of E. coli E22 in the cecal contents, combined with a postmortem examination of the lesions, permitted the identification of rabbits that had died from colibacillosis.

Bacterial analysis.Twice a week, a sample of feces (about 10 g) from the different cages was collected to carry out bacterial analysis. At 63 days of age, all the young rabbits were killed, and enumeration of the E. coli cells in the cecal contents of individual rabbits was done. Cecal and fecal content enumeration of E. coli cells was performed by plating 10-fold dilutions of samples on eosin-methylene blue agar medium. After 24 h of incubation at 37°C, followed by a selective enumeration (dark blue colonies), screening for the inoculated strain was done with a set of randomly selected isolates (10 per sample) by using rhamnose as a marker. The growth of E. coli was evaluated on a phenol red agar containing 1% rhamnose (24 h, 37°C). In the case of doubtful biochemical characteristics of the colonies, a slide agglutination test was done with specific antibodies directed toward the O103 lipopolysaccharide (LPS) (7).

Preparation of milk and lactoserum.Seven suckling does kept in an area free of the O103:H2 REPEC strain were used. The does were intravenously injected with 3 IU of ocytocin at the time of suckling. After a brief contact with their pups, the does were milked with a vacuum pump connected to a milker tube. For each doe, between 2 and 10 ml of milk was obtained and kept at 4°C before testing. Blood was also sampled from these does, and the sera were kept at −20°C. Sampling was done at 17, 24, and 30 days after birth.

To obtain lactoserum, cold acetic acid was added to the milk to lower the pH to a value of 3. The milk was immediately centrifuged at 10,000 rpm for 20 min at 4°C. The clot was removed and the pH was adjusted to 7 by addition of 4 N NaOH. An aliquot of serum from these does was subjected to the same treatment, prior to its use in the tests for the inhibition of adhesion.

Bacterial growth in milk or lactoserum.Strain E22, which was cultured in LB broth for 5 h at 37°C with rotary shaking (250 rpm), was transferred in milk for 18 h under the same conditions. From this culture, 100 μl of a 10⁶ dilution was added to 400 μl of milk, lactoserum, or LB broth medium (approximately 10² CFU). The same amount of overnight culture was spread onto an LB agar plate to determine the exact amount of bacteria added to the medium. The bacteria were cultured for 5 h at 37°C with rotary shaking (250 rpm). For counting of the bacteria, dilutions of media were spread on LB agar plates in duplicate, the plates were cultured overnight, and the numbers of CFU were determined for each culture.

Quantification of bacterial adhesion on HeLa cells.HeLa cells (ATCC CCL-2) were seeded at 5 × 10⁴ cells per ml in 12-well plates (Nunc) and cultivated for 24 h in Eagle minimum essential medium (MEM) supplemented with 10% fetal calf serum (FCS; Gibco BRL), glutamine (2 mM) (Sigma), and gentamicin (80 μg/ml) at 37°C in a 5% CO₂-95% air atmosphere. Strain pEGFP-E22 was precultured for 24 h in Penassay broth (antibiotic medium 3; Difco) supplemented with 25 μg/ml of carbenicillin. Incubation of the bacteria with the cells was realized in 500 μl of MEM buffered with 5% FCS and 1% mannose, with a starting inoculum of 20 μl (about 2 × 10⁷ bacteria per well). To test the inhibitory capacity of adhesion of the serum, milk, or lactoserum of the does, 20 μl of strain pEGFP-E22 bacteria grown overnight (10⁹ CFU/ml) in Penassay broth was preincubated with different amounts of serum, milk, or lactoserum for 1 h at room temperature, prior to incubation for 30 min with HeLa cells, as described previously (5). After 30 min of incubation of the cells with the bacteria at 37°C, the cells were washed three times with PBS. The cells were then incubated for 15 min at 37°C in PBS containing 2 mM EDTA. The cells were harvested in PBS, washed two times in PBS, and immediately acquired with a Facscalibur instrument (Becton Dickinson). A total of 10⁴ HeLa cells were acquired per sample. Quantification of adhesion was monitored by using Cellquest software (Becton Dickinson).

The percentage of inhibition was calculated by using the following formula: (percentage of pEGFP-E22-positive HeLa cells without serum, milk, or lactoserum − percentage of pEGFP-E22-positive HeLa cells preincubated with serum, milk, or lactoserum)/percentage of pEGFP-E22-positive HeLa cells without serum, milk, or lactoserum.

Tissue sampling for immunohistochemical studies.A litter of nine suckling New Zealand rabbits and a litter of weaned rabbits (group W21) were orally inoculated with 2 × 10⁴ CFU of strain E22 at day 28 and housed as described above. The animals in each group were killed at 3 days postinfection by intravenous overdosing with sodium pentobarbital. Tissues from the distal ileum (10 cm from the ileo-cecal junction) were removed immediately after euthanasia. The tissues were fixed in 10% formaldehyde and embedded in paraffin. Transversal tissue sections (3 μm) were used for immunochemical staining. The slides were treated with hydrogen peroxidase and an ABC kit (Vector, Burlingame, CA) to remove the background before staining. The bacteria fixed on the epithelia were labeled by using serum prepared in the laboratory by intradermal injection of a sheep with formalin-fixed whole E22 (4). To detect the fixation of anti-E22 antibodies, we used a biotinylated anti-sheep immunoglobulin antibody (Vector Laboratories). Anti-sheep immunoglobulin fixation was revealed with streptavidin coupled with peroxidase (Vector Laboratories) and revealed with diamino-benzidine (Sigma). All tissue samples were scored blindly by using light microscopic observations (Leica).

Statistical analysis.For each age, a designations of 0 or 1 was attributed to each rabbit if it was dead or alive, respectively. These data, which are considered Bernoulli variables, were submitted to a variance analysis (GLM procedure; SAS, 1999); the effect of weaning age on the survival rate was tested. For E. coli excretion, the results were expressed in log₁₀ CFU·g⁻¹ of sampled feces. Among the E. coli cells excreted, the proportion of the infectious strain corresponded to the number of colonies not metabolizing the rhamnose among those tested. The effect of the weaning age on total fecal excretion of E. coli was analyzed by the GLM procedure (SAS, 1999).

Suckling temporarily protects young rabbits from colibacillosis.In the experiment performed to determine whether suckling temporarily protects young rabbits from colibacillosis, we took several precautions in the experimental setting. First, the does had to be seronegative for the strain of E. coli used here, to exclude any humoral passive protection for the young by the maternal milk. Second, the kits should not show any sign of previous contact with this strain. Thus, all does were tested for the presence of anti-LPS O103 antibodies 10 days before parturition and were found to be seronegative (data not shown). Their 24-day-old kits were also seronegative (28% of rabbits randomly tested), and no E. coli strains metabolizing the rhamnose were found in their fecal output (all cages were tested) (data not shown).

The two groups of young rabbits were inoculated 28 days after birth, and mortality was determined daily. The origin of each death was assessed by a combination of bacterial analysis of the cecal contents and necropsic examination. Colibacillosis was not responsible for the deaths of six rabbits: one rabbit from the W21 group and five rabbits from the W35 group. The data for these animals were removed from the study.

Our results indicate that after inoculation of E22, the mortality presented different patterns of evolution according to the nutritional status of the young rabbits (Fig. 1). Indeed, the first deaths were observed in the weaned group (group W21) from day 4 after experimental infection. In sharp contrast, no mortality due to colibacillosis was noticed in the suckling group (group W35) until 8 days postinfection (36 days of age), with the first death occurring only 1 day after weaning. At that time point, 50% of the group W21 rabbits were already dead. This mortality rate was reached only on day 45 in the W35 group. Moreover, the disease developed quickly in the W21 group, with the final survival rate being 20% 13 days after the first death occurred, whereas this evolution was slower for group W35 rabbits (mortality was observed over a period of 19 days). From 33 to 49 days of age, the survival rate of group W35 rabbits was higher than that of group W21 rabbits (P < 0.05). However, the difference in the final mortality rate between the two groups (36% for the W35 group versus 20% for the W21 group) was not statistically significant (P = 0.102).

• Open in new tab
• Download powerpoint

FIG. 1.

Suckling protects young rabbits from lethal colibacillosis due to an EPEC O103 strain. Rabbits in the W21 group (black squares) and the W35 group (gray squares) were inoculated by the oral route on day 28 with 2 × 10⁴ CFU of the EPEC O103 E22 strain. The mortality rate of 50% was reached at 36 days of age in the W21 group and at 45 days of age in the W35 group. Final survival rates were of 20% and 36% in the group W21 and group W35 rabbits, respectively (P = 0.102).

As for mortality, the morbidity of the rabbits was recorded daily. The main clinical signs observed in rabbits during this study were prostration, weight loss, anorexia, and abundant aqueous diarrhea sometimes accompanied by dehydration symptoms. The same time lag in the pattern for mortality between the two weaning groups was noticed for the pattern of morbidity (Fig. 2). Thus, the peak of morbidity was observed at 32 days of age in the W21 group (close to 74% of diarrheic rabbits) but was observed at 42 days of age for rabbits in the W35 group (about 52% of morbid rabbits). Morbidity reached 100% for the whole period of the study, and all rabbits presented diarrhea symptoms in at least one clinical assessment, whatever their weaning group.

• Open in new tab
• Download powerpoint

FIG. 2.

Suckling delays the evolution of diarrhea in young rabbits infected with an EPEC O103 strain. Diarrhea signs were recorded for rabbits in the W21 group (black) and the W35 group (gray) after experimental infection with the E22 strain of EPEC on day 28. This graph represents the daily evolution of the number of deaths among surviving rabbits. The number of surviving rabbits is represented by the shaded area, and the number of deaths is represented by fully colored area for each group.

Suckling temporarily limits fecal E. coli excretion.To study the development of colibacillosis and to evaluate intestinal colonization and the potential elimination of pathogenic bacteria by rabbits, we aimed at controlling the level of fecal E. coli excretion. Fecal excretion of resident E. coli was unaffected by the weaning age before challenge, namely, 24 and 28 days of age (Fig. 3). After oral inoculation, group W21 rabbits excreted more pathogenic E. coli than group W35 rabbits until 38 days of age (+3.4 log [P < 0.05], +2.4 log [P < 0.001], and +1.3 log [P = 0.053] at 31, 36, and 38 days of age, respectively). From 42 days onwards, the excretion patterns became similar between the two weaning groups and remained high until 49 days of age (>10⁶ CFU·g⁻¹ of feces). The profile of bacterial excretion clearly corresponded to the evolution of clinical signs in the two experimental groups. After day 49, the levels were maintained at 10⁴ pathogenic E. coli·g⁻¹ until 59 days of age, whatever the weaning age, but with a high variability between rabbits. In the W35 group, this variability could be explained by the elimination of the inoculated strain from some cages (three cages on day 56 and eight cages on day 59). In the group W21 rabbits, this phenomenon may result from the low number of remaining cages (only four cages on day 42 and seven cages on day 35). Individual cecal enumeration of E. coli at 63 days of age revealed that about half of the survivors had eliminated EPEC, whatever their weaning age (4/8 rabbits in the W21 group and 6/11 rabbits in the W35 group).

• Open in new tab
• Download powerpoint

FIG. 3.

Suckling limits fecal excretion of E. coli by young rabbits infected with an EPEC O103 strain. The total excretion of E. coli was recorded twice a week in all cages and is represented by the heights of the bars (in black and gray for the group W21 and W35 rabbits, respectively). Among the E. coli cells excreted, the proportion that was E22 (solid bars) was determined. Data show the arithmetic means ± standard deviations. For each time sampling, P values between groups W21 and W35 are shown on the graph for the W35 group for P values of <0.10 (**), <0.01 (*), and <0.05 (†).

Bacterial adhesion to the ileal epithelium is strongly reduced in suckling rabbits.We next wondered if the differential clinical evolution and the differential levels of fecal excretion between weaned and suckling rabbits were associated with differential ileal bacterial attachment and tissue lesions. A litter of weaned rabbits and one of the suckling rabbits were killed 3 days after oral inoculation with E22. Ileal tissues were sampled and stained with an anti-E22 serum, as described in Material and Methods.

In all weaned rabbits, we observed an important bacterial adhesion to the ileal epithelium. Bacteria almost entirely covered the epithelium, and we could detect the intimate attachment of bacteria to enterocytes, a typical lesion characteristic of EPEC (Fig. 4a). This strong bacterial adhesion was associated with a modification of the ileal architecture, characterized by the atrophy of the villi and the desquamation of the epithelium. At the time of sampling, all rabbits were prostrate and about 50% of them already presented with diarrhea.

• Open in new tab
• Download powerpoint

FIG. 4.

Suckling prevents ileal bacterial attachment and intestinal inflammation. Ileal tissue sections of weaned (a) or suckling (b and c) rabbits infected at day 28 and killed at day 31 were stained with anti-E22 sera and visualized with a Leica microscope. (a) All weaned rabbits presented an ileal epithelium covered by intimately attached E22 bacteria (arrow). Note the strong tissue destruction characterized by villous atrophy and the afflux of inflammatory cells. (b) Eight of nine suckling rabbits presented only scarce bacterial adhesion (arrow). Villi are intact, and no sign of inflammation is visible. (c) One suckling rabbit presented a more important bacterial attachment to the epithelium (arrow). This adhesion was associated with shortening and widening of the villi. Magnifications, ×400.

In contrast, bacterial adhesion was greatly reduced in suckling rabbits. Among nine rabbits tested, eight exhibited rare bacterial adhesion, with only scarce spots of bacteria detected (Fig. 4b). In addition, the epithelium was intact and no inflammatory lesions were detected. The villi remained long and thin. None of these rabbits presented with clinical signs at the time of sampling. The remaining rabbit presented a more important E22 intestinal colonization. About 30% of the ileal epithelium was covered by bacteria (Fig. 4c). The attachment of E22 was associated with an intermediate inflammatory response, characterized by a shortening and a widening of the villi; however, it was not associated with epithelial desquamation. Interestingly, this rabbit did not display any clinical signs.

These results confirm a protective effect of milk ingestion against colibacillosis. Indeed, we demonstrated a strong difference in bacterial attachment to the epithelium between suckling and weaned rabbits. In addition, this reduced adhesion was not associated with important intestinal lesions.

Milk and lactoserum have bacteriostatic effects in vitro.The in vivo results prompted us to characterize the protective effect of doe's milk in vitro. In order to define a potential bacteriostatic or bactericidal effect of doe's milk on REPEC growth, E22 bacteria grown in doe's milk were cultured for 5 h in LB medium, milk, or lactoserum, as described in Materials and Methods. Our results showed the clear bacteriostatic effects of milk (2 log units of growth reduction compared with the growth in LB medium) and lactoserum (more than 3 log units of growth reduction compared with the growth in LB medium) (Fig. 5). Interestingly, the bacteriostatic effect of lactoserum was almost complete. Indeed, the growth rate was equal to only 1.3 after 5 h of culture. However, we could not demonstrate any bactericidal effect of milk or lactoserum in vitro.

• Open in new tab
• Download powerpoint

FIG. 5.

Milk and lactoserum have bacteriostatic effects on E22 in vitro. After 5 h of E22 culture, the mean number of CFU for each type of culture was determined by plating the samples on LB agar plates in duplicate. The results represent the mean ± standard deviation CFU obtained with the milk and lactosera of seven does.

Thus, the protective effect of suckling against REPEC for young rabbits seems to be associated with a capacity to limit bacterial growth in vitro.

Milk and lactoserum inhibit bacterial adhesion in vitro.We previously developed an in vitro test to quantify the capacity of rabbit sera to inhibit the AF/R2-dependent bacterial adhesion to HeLa cells (5). In order to assess whether the protective effect of doe's milk could be linked to an inhibiting adhesion effect in vitro, we adapted this test for milk and lactoserum samples.

Milk was tested at different dilutions, and we observed a complete inhibition of bacterial adhesion up to a dilution of 1/4 (Fig. 6A and B). The effect became variable but still important at the dilution of 1/10 but insignificant at the dilution of 1/50. A similar pattern of inhibition of adhesion was found when these experiments were performed with lactoserum (Fig. 6C). Finally, we wondered whether the inhibition of adhesion was stable over time during lactation. Milk sampling was done on days 17, 24, and 30 after birth. For practical reasons, it was difficult to obtain milk before 2 weeks of lactation. For this experiment, lactosera were tested at a dilution of 1/4. The capacity of inhibition of adhesion was observed for all the kinetic points tested, with an increase at the end of lactation (day 30), although the differences were not significant (Fig. 6D).

• Open in new tab
• Download powerpoint

FIG. 6.

Milk and lactoserum inhibit bacterial adhesion on HeLa cells in vitro. (A) Milk samples of seven does were tested for their ability to inhibit bacterial adhesion on HeLa cells. The percentage of pEGFP-E22-positive HeLa cells was determined by flow cytometry by detecting the green fluorescent protein (gfp) expression of pEGFP-E22. Samplings were done at day 30 after birth. Dilution factors are indicated under each histogram bar. Data are means ± standard deviations. (B) Fluorescence-activated cell sorter histogram visualization of the inhibition of bacterial adhesion to HeLa cells obtained with lactoserum and serum of a doe at a dilution of 1/4. Similar results were obtained for six other does. Light black line, E22 and HeLa cells (E22 adhesion, 57.3%); heavy black line, E22 preincubated with lactoserum (E22 adhesion, 4.8%); spotted line, E22 preincubated with the corresponding serum (E22 adhesion, 52.7%). (C) The lactosera of seven does sampled at day 30 after birth were tested for their ability to inhibit bacterial adhesion on HeLa cells by flow cytometry. The dilution factors are indicated under each histogram bar. Data are means ± standard deviations. (D) The lactosera of four does were tested at different time points of lactation for their ability to inhibit bacterial adhesion on HeLa cells. All lactosera were tested at a 1/4 dilution. Data are means ± standard deviations.

In order to ensure that this inhibition of adhesion was not due to the presence of anti-E22 antibodies in the females, the sera and lactosera of the does were tested for anti-O103 antibodies by an enzyme-linked immunosorbent assay. All does tested were found to be negative (data not shown). In addition, sera were tested for their capacity to inhibit adhesion (without treatment and after the same treatment that was applied to the milk samples). None of the sera inhibited the adhesion of E22 to HeLa cells (Fig. 6B and data not shown).

Taken together, our results show that the protective effect of suckling against REPEC infection for young rabbits is associated with a capacity to inhibit bacterial adhesion to HeLa cells in vitro. These results suggest that doe's milk contains one or several antimicrobial factors whose effects can be detected both in vivo and in vitro.