003 1-399819112904-0396$03.00/0 PEDIATRIC RESEARCH Copyright O 199 1 International Pediatric Research Foundation, Inc.
Vol. 29, No. 4, 199 1 Printed in U.S.A.
The Antibody Response in Breast-Fed and NonBreast-Fed Infants after Artificial Colonization of the Intestine with Escherichia coli 0 8 3 RAJA LODINOVA-ZADN~KOVA,MILADA SLAV~KOVA,HELENA TLASKALOVA-HOGENOVA, INGEGERD ADLERBERTH, LARS A. HANSON, AGNES WOLD, BARBRO CARLSSON, CATHARINA SVANBORG, AND LOTTA MELLANDER
Institute for Care ofMother and Child [R.L.-Z.,M S . ] and Institute of Microbiology [If.T.-H.] Czechoslovak Academy of Science, Prague, Czechoslovalcia and Department of Clinical Immunology [I.A., L.A.H., A. W., B.C., C.S., L.M.], Universily of Goteborg, Goteborg, Sweden
ABSTRACT. The local and systemic antibody response after oral administration of a nonenteropathogenic type 1 fimbriated Escherichia coli 0 8 3 strain was followed in nine breast-fed and eight formula-fed infants during their first 15 wk of life. Five breast-fed and six formula-fed infants were followed as controls. E. coli 0 8 3 was detected in the stools of colonized infants from d 2 after colonization and persisted in the intestine for up to 26 wk. The percentage of children successfully colonized with E. coli 083 was higher among breast-fed than among formula-fed colonized infants. Also, the 083 bacteria isolated from the breastfed children had a higher capacity to attach to colonic epithelial cells of the HT-29 cell line than those isolated from bottle-fed infants. E. coli 0 8 3 IgA and IgM antibodies estimated by ELISA were significantly elevated in the saliva of colonized as compared with control infants 2-7 wk after colonization. IgA antibodies against 0 8 3 were also higher in the stool of colonized formula-fed infants than in formula-fed controls. The results suggest that the mucosal immune system of the newborn infant can be triggered early to produce specific antibodies against bacteria colonizing the intestine. (Pediatr Res 29: 396-399, 1991)
Starting at birth, the infant is exposed to a multitude of microbes, including potential pathogens. Within the first postnatal days, E. coli appear in the stool, the strains originating from the mother and/or the environment (1-3). The young infant can usually cope with these microorganisms with the support of transplacentally transferred IgG antibodies and breast milk, which contains various defense factors including secretory IgA antibodies to various E. coli antigens (4). In addition, secretory IgA antibodies against E. coli antigens have been detected as early as the first days of life in infants' saliva with a subsequent increase relating to the degree of microbial exposure (5). An E. coli 0 8 3 strain has been found to very efficiently colonize the colon of neonates and remain for months (6). This strain has been used to colonize premature newborn infants to reduce colonization with potential pathogens (7). In our study, healthy full-term babies were deliberately colonized with E. coli 083. The antibody response against the 0 8 3 antigens in serum, saliva, and stools was followed to determine if the common mucosal immune system could be triggered by such colonization. Received November 4, 1989; accepted October 3, 1990. Correspondence: R. ~odinova:~adnikovi,Institute for Care of Mother and Child. Podolski: nabrezi 157, 147 10 Prague 4, Czechoslovakia.
MATERIALS AND METHODS
Infants. All infants included in our study were kept in a unit, where the infants were admitted for social reasons such as being born to seriously ill or single mothers or to parents without accommodation (8). Some of the infants were born to mothers who had decided to give up their child for adoption. The criteria for inclusion into our study were uncomplicated term pregnancy and delivery by a healthy mother. Fourteen infants were breastfed for 13 wk or longer and their mothers lived in the unit; 14 infants were admitted without mothers and received a "humanized" formula prepared from cow's milk. The mothers of the latter group were unable to breast-feed or uninterested in breastfeeding. After 4 mo, all infants received solids (fruits, vegetables) and cow's milk formula. The clinical characteristics of the infants are summarized in Table 1. The study was approved by the hospital scientific committee and the Ministry of Health, including ethical aspects according to the Helsinki Declaration. The infants participated in the study with the written informed consent of their mothers. In nine breast-fed and eight formula-fed infants, artificial colonization of the intestine was started during the first 6 d of life by oral administration of a suspension prepared from a live 24-h-old culture of the serotype E. coli 083:K24:H3 1 containing 5 x 10' organisms in 1 mL. One mL of the suspension was given to each child diluted in 10 mL of tea. three times a wk for 4 successive wk. Five breast-fed and six formula-fed infants were followed as controls. There was no marked difference in the duration of breastfeeding between those colonized with E. coli 0 8 3 (16.6 5 3.4 wk) and the noncolonized breast-fed controls (15.2 + 2.8 wk). All infants were healthy during the investigation. Bacteriologic examinations. Stool smears were cultivated aerobically on Drigalski agar plates. The last three free-lying colonies were isolated and placed in deep agar tubes that were transported to Goteborg for further analysis. Collecting those three colonies gave a probability of including the dominant strain of 97% (9). The strains were subcultured once on Drigalski agar to check the purity, and then once on tryptic soy agar. E. coli strains were identified by limited biotyping. The presence of E. coli 0 8 3 was detected by agglutination with a specific rabbit antiserum, prepared by repeated i.v. injections of lo9 formalinized bacteria once a week for 4 wk. Activity of the antiserum measured by ELISA exceeded a titer of 1/ 10 000. The presence of E. coli bacterial adhesins and their specificity was tested by agglutination of guinea pig erythrocytes in the absence and presence of mannose. Type 1 fimbriated strains agglutinate guinea pig red blood cells in the absence but not in the presence of mannose. Agglutination of latex beads coated
3s)6
397
ARTIFICIAL E COLI COLONIZATION IN INFANTS
with the disaccharide Gal a 1 4 Gal 13 was used to confirm the presence of type P fimbriae (10). All isolates were tested for adhesion to the human adenocarcinoma cell line HT-29 as described earlier (1 1). A total of 40 HT-29 cells were scored and the adherence value was defined as the mean number of adhering bacteria per cell. Analysis of' antibody responses. Samples of blood, stool, unstimulated saliva from the infants and breast milk from the mothers were taken before colonization and in the 2nd to 3rd, 4th to 7th, 8th to 1lth, and 12th to 15th wk after colonization. The same schedule of sampling was followed in the controls. Serum and saliva samples were kept at -20°C until use. One g from each stool sample was diluted in 5 mL of PBS, homogenized, and centrifuged at 600 x g for 20 min. The supernatant was kept frozen until further examination. Antibodies of the IgM, IgA, and IgG isotypes against the 0 8 3 antigen were measured by ELISA as described (12). The optimal dilution of the supernatant (150) of the sonicated suspension used for coating of microplate wells was determined by previous checkerboard titration, as well as the optimum of all other components used. After adding samples diluted 1 :5, peroxidaseconjugated swine anti-human antisera against IgM, IgA, or IgG (Institute of Sera and Vaccines, Prague, Czechoslovakia) were used and the reaction was developed with phenylendiamine (Lachema, Brno, Czechoslovakia) diluted in 0.01 M PBS, pH 6, containing H202.The reaction was stopped by adding 50 pL of 1 M H2SO4.The plates were read on a Microelisa Minireader MR 590 (Dynatech Laboratories, Alexandria, VA) at 495 nm. The level of antibody activity was expressed in percentage of references prepared from samples of, respectively, saliva, serum, stool, and milk from two mothers colonized with E. coli 083. The specificity of the ELISA assay was determined by inhibition studies. Ninety-seven percent of the binding activity of the hyperimmune serum was inhibited by incubation with an E. coli 0 8 3 sonicate. If an E. coli 0 8 6 sonicate of comparable OD was used, only 14% of the antibody activity was inhibited. Also, in adults colonized with E. coli 0 8 3 the inhibition of serum IgA antibody activity was significantly different if we compared E. coli 0 8 3 and E. coli 0 8 6 sonicates. Coefficient of variation intraassay was 6.9% and interassay 13.5%. Table 1. Clinical characteristics ofstudied infants Breast-fed
Formula-fed
n Maleifemale Birth wt (g) Birth ht (cm) Gestational age (wk) Apgar score l min 5 min
* Mean + SD.
Data analysis. Statistical evaluation included analysis of variance, Fisher's test, and the single nonparametric one-sided Rosenbaum's test (13, 14). All data in the figures are expressed as mean rt SEM. RESULTS
Bacteriologicfindings. The strain E. coli 0 8 3 was detected in the stools of all colonized infants from the 2nd d after oral administration and persisted throughout the study period. E. coli 0 8 3 represented a significantly larger fraction of E. coli strains isolated from breast-fed than formula-fed infants (46 versus 24%, p < 0.05; Table 2). In addition, the E. coli 0 8 3 from the breastfed infants adhered more efficiently to the colonic cell line HT29 than did those from the formula-fed infants ( p < 0.05). No mothers were found to be colonized with E, coli 083. Antibody response against E. coli 083. IgA antibodies against E. coli 083, found in breast-fed babies' stools whether they had been colonized or not, presumably originated from the mothers' milk (Fig. 1). In the formula-fed infants, the IgA antibodies to E. coli 0 8 3 increased from the 2nd wk after being colonized ( p < 0.05). The IgM anti-083 in the stool did not differ between the colonized and the control infants. IgA and IgM antibodies against the E. coli 0 8 3 were detected in saliva of colonized infants, whether breast- or formula-fed, and reached the highest levels between the 4th and 7th wk in comparison with the control groups ( p < 0.01) (Fig. 2). In the serum, no increase of IgG, IgM, or IgA antibodies could be detected against the 0 8 3 antigens. DISCUSSION
Bacterial colonization of the intestine or oral administration of killed bacteria has been tested as a prevention against enteric infections and as a stimulation of the local immune system in animals (15, 16), human adults (17), and also infants (1 8, 19). The antigenic structure of strains of E. coli 0 8 3 was described in 1978 by the International Escherichia and Klebsiella Center in Copenhagen as 083:K:H31. Recently, the K antigen was identified as K24 (0rskov F, personal communication). In E. coli strains isolated from infants with urinary tract infections, meningitis, or septicemia, the K1, K2, K3, K12, and K13 antigens were mostly detected (20-22). The strain E. coli 0 8 3 isolated from an infant with meningitis and septicemia by Czirok et al. (23) had no K and H antigens identified. The strain E. coli 083:K24:H31 used in our work has never been isolated from infants with meningitis or septicemia and has caused no complications in 1500 infants colonized in our studies of newborns (LodinovCZadnikovi R, manuscript in preparation). E. coli 0 8 3 was significantly more often recovered from colonized infants who were breast-fed than from those who were formula-fed. Also, the 0 8 3 isolates recovered from breast-fed infants had a higher mean adherence to colonic cells of the HT29 cell line than isolates of the same strain recovered from
Table 2. Detection of E. coli adhesins in stools of breast-fed and forrnzila-fed infants after natural and artificial colonization of intestine*
No. of infants
No. of samples
No. of isolates
P-fimbriated strains (%)
Type 1 fimbriated strains except E. coli 0 8 3 (%)
Breast-fed colonized
9
45
72
13
I1
Formula-fed colonized Breast-fed controls Formula-fed controls
8 5
40 25 30
94 39 52
10 11 14
19 22 37
Groups
6
Type I fimbriated E. coli 0 8 3 (%)
~ ~ 2 adherence E. coli 0 8 3 (mean, SEM)
46 p < 0.05 24 0 0
18 (1.4) p < 0.05 1 1.6 (2.7)
9
* Stool samples were taken before and 2-3, 4-7, 8-1 1, and 12-15 wk after colonization. The E. coli were subcultured on tryptic soy agar and tested for the presence of type 1 fimbriae by agglutination (mannose-sensitive) of guinea pig erythrocytes and for the presence of P-fimbriae by agglutination (mannose-resistant) of human erythrocytes and Gal a1 -+ 4 Gal 0-coated latex beads. Adherence to HT-29 cells was determined as stated in Materials and Methods.
% of
10'1
ref.
IgA
COLONIZED H breast fed ( ~ a 9 ) .--+ b o t t l e fed ( N - 8 ) I
CONTROLS I breast fed (N.5)
IgM lool
'before'
%---a b o t t l e fed ( ~ : 6 )
I
I
1 - 3
'4-7
' 8 - 1 1 '11-1Sdeeks
Fig. 1. Levels o f E. coli 083 IgA a n d IgM antibodies in stools. Results o f ELISA a r e expressed a s a percentage o f reference i n m e a n s S E M . T h e IgA 0 8 3 antibodies a r e significantly higher i n breast-fed colonized a n d control groups t h a n i n formula-fed controls i n all intervals ( p < 0.05). I n formula-fed colonized infants, t h e IgA 0 8 3 antibodies a r e significantly higher t h a n i n formula-fed controls i n intervals o f 2-3 a n d 4-7 wk after colonization ( p < 0.05).
+
COLONIZED H breast fed (Nag) *--a b o t t l e fed (N=B)
% of ref. IgA
secretory IgA in the gut, as in the breast-fed infants, may favor mannose-binding type 1-fimbriated strains and increase their expression of fimbriae. In our previous work, oral administration of a nonenteropathogenic E. coli strain 0 8 3 evoked an increase in total stool secretory IgA as well as a specific serum antibody response in infants (26). In the present work, the significant increase of the specific IgA and IgM response shown in saliva after oral colonization of the gut agrees with the concept of the common mucosal immune system. The presence of low levels of IgA and IgM antibodies already before colonization, i.e. in the first days of life, is in agreement with the findings of Mellander et al. (5). They showed a secretory IgA response in saliva against E. coli 0 antigens at birth and a subsequent increase. It was suggested that these salivary antibodies in newborns may result from priming in utero via maternal antiidiotypic antibodies (27, 28), and/or some degree of cross-reactions. The E. coli 0 8 3 IgA antibodies detected in stool samples, higher in both the colonized and noncolonized breast-fed groups, probably originated from the maternal milk. Significant production of E. coli IgA antibodies in the intestine after colonization could, however, be shown in the formula-fed group 4 to 7 wk arter colonization. The IgM and IgA antibodies to the 0 8 3 antigen in the stool most likely have the same origin as the saliva antibodies mentioned above. Actually, Mellander et al. (27) found IgA and IgM antibodies to E. coli 0 antigens already in meconium. The best way to protect young infants from infection is presumably breastfeeding (29). Some infants at high risk, such as premature infants, are unfortunately often deprived of fresh breast milk. The protective effect of intentional colonization with E. coli 0 8 3 has already been indicated to be eficient clinically. In infants, including high-risk neonates and premature infants, the presence of the 0 8 3 strain in the intestine prevented colonization with pathogens and displaced those present without any unfavorable side effects (30). Colonization with E. coli 0 8 3 might protect formula-fed infants by an early induction of antibody production, with the possibility of reactivity beyond the E. coli 083. REFERENCES
CONTROLS breast f e d ( ~ : 5 )
IgM
O
before
1-3
X-x
4- 7
8 -11
12-15 weeks
Fig. 2. Levels o f IgA a n d IgM E. coli 0 8 3 antibodies i n saliva. Results o f ELISA a r e expressed a s a percentage o f reference i n m e a n s +. S E M . T h e IgA levels a r e significantly higher i n both colonized groups c o m p a r e d with bottle-fed controls in t h e intervals o f 2-3, 4-7, a n d 8-1 1 wk ( p < 0.0 I ) a n d i n t h e interval 4-7 wk after colonization c o m p a r e d with breastfed controls. T h e IgM levels a r e significantly higher i n colonized groups t h a n i n both control groups i n intervals o f 2-3 a n d 4-7 wk after colonization ( p < 0.01).
formula-fed infants. The adherence of E. coli 0 8 3 to HT-29 cells is mediated via a mannose-specific interaction. Mannose-containing oligosaccharides functioning as receptors for type 1fimbriated E. coli are found in secretory IgA (24, 25). This raises the interesting possibility that the presence of high levels of
1. Gothefors L, Carlsson B, Ahlstedt S, Hanson LA, Winberg J 1976 Influence of maternal gut flora and colostral and cord serum antibodies on presence of E. coli in faeces of the newborn infant. Acta Pediatr Scand 65:225-232 2. Adlerberth I, Carlsson B, de Man P, Jalil F, Larsson P, Mellander L, Svanborg C, Wold A, Hanson LA 1990 Intestinal colonization with enterobacteria in Pakistani and Swedish hospital-delivered infants. Acta Paediatr Scand (in press) 3. Brskov F, Sorensen KB 1975 Escherichia coli serogroups in breast-fed and bottle-fed infants. Acta Pathol Microbiol Scand [B) 83:25-30 4. Hanson LA, Adlerbert I, Carlsson B, Dahlgren U, Jalil F, Khan SR, Zaman S, Larsson P, Mellander L, Sheikh AK, Soderstrom T, Wold A 1987 The ontogeny of the immune response: the role of maternal factors. In: Burgio GR, Hanson LA, Ugazio AG (eds) Immunology of the Neonate. SpringerVerlag, Berlin, pp 51-58 5. Mellander L, Carlsson B, Jalil F, S6derstriim T, Hanson LA 1985 Secretory IgA antibody response against Escherichiu coli antigens in infants in relation to exposure. J Pediatr l07:433-436 6. Lodinova R, Jouja V, Lanc A 1967 Experimentelle Besiedlungdes Darmtrakts von Neugeborenen mit dem E, coli 0 8 3 und Untersuchungen iiber die Bildung van Antikorpern. Z Immunitaetsforsch 133:229-237 7. Lodinova R, Jouja V, Vinsova N, Vocel J, Melkova J 1980 New attempts and possibilities of prevention and treatment of intestinal coli-infections in infants. Czech Med 3:47-58 8. Paul K, Vondracek J 1976 Individual differences in infant's sleep. Dev Med Child Neurol 18:182-188 9. Lidin Jansson G,Kaiiser B, Lincoln K, Olling S, Wedel H 1978 The homogeneity of the faecal coliform flora of normal schoolgirls, characterized by serological and biochemical properties. Med Microbiol Immunol (Berl) 164:247-253 10. de Man P, Cedergren B, Enerback S, Larsson A-C, Lemer H, Lundell A-L, Nilsson B, Svanborg Eden C 1987 Receptor-specific agglutination tests for detection of bacteria that bind globoseries glycolipids. J Clin Microbial 2540 1-406 I I . Wold AE, Thom6n M, Hull S, Svanborg Eden C 1988 Attachment of Escherichiu coli via mannose- or gal 1-4gal R-containing receptors to human colonic epithelial cells. Infect Immun 56:2531-2537
Vol. 29, No. 4, 199 1 Printed in U.S.A.
The Antibody Response in Breast-Fed and NonBreast-Fed Infants after Artificial Colonization of the Intestine with Escherichia coli 0 8 3 RAJA LODINOVA-ZADN~KOVA,MILADA SLAV~KOVA,HELENA TLASKALOVA-HOGENOVA, INGEGERD ADLERBERTH, LARS A. HANSON, AGNES WOLD, BARBRO CARLSSON, CATHARINA SVANBORG, AND LOTTA MELLANDER
Institute for Care ofMother and Child [R.L.-Z.,M S . ] and Institute of Microbiology [If.T.-H.] Czechoslovak Academy of Science, Prague, Czechoslovalcia and Department of Clinical Immunology [I.A., L.A.H., A. W., B.C., C.S., L.M.], Universily of Goteborg, Goteborg, Sweden
ABSTRACT. The local and systemic antibody response after oral administration of a nonenteropathogenic type 1 fimbriated Escherichia coli 0 8 3 strain was followed in nine breast-fed and eight formula-fed infants during their first 15 wk of life. Five breast-fed and six formula-fed infants were followed as controls. E. coli 0 8 3 was detected in the stools of colonized infants from d 2 after colonization and persisted in the intestine for up to 26 wk. The percentage of children successfully colonized with E. coli 083 was higher among breast-fed than among formula-fed colonized infants. Also, the 083 bacteria isolated from the breastfed children had a higher capacity to attach to colonic epithelial cells of the HT-29 cell line than those isolated from bottle-fed infants. E. coli 0 8 3 IgA and IgM antibodies estimated by ELISA were significantly elevated in the saliva of colonized as compared with control infants 2-7 wk after colonization. IgA antibodies against 0 8 3 were also higher in the stool of colonized formula-fed infants than in formula-fed controls. The results suggest that the mucosal immune system of the newborn infant can be triggered early to produce specific antibodies against bacteria colonizing the intestine. (Pediatr Res 29: 396-399, 1991)
Starting at birth, the infant is exposed to a multitude of microbes, including potential pathogens. Within the first postnatal days, E. coli appear in the stool, the strains originating from the mother and/or the environment (1-3). The young infant can usually cope with these microorganisms with the support of transplacentally transferred IgG antibodies and breast milk, which contains various defense factors including secretory IgA antibodies to various E. coli antigens (4). In addition, secretory IgA antibodies against E. coli antigens have been detected as early as the first days of life in infants' saliva with a subsequent increase relating to the degree of microbial exposure (5). An E. coli 0 8 3 strain has been found to very efficiently colonize the colon of neonates and remain for months (6). This strain has been used to colonize premature newborn infants to reduce colonization with potential pathogens (7). In our study, healthy full-term babies were deliberately colonized with E. coli 083. The antibody response against the 0 8 3 antigens in serum, saliva, and stools was followed to determine if the common mucosal immune system could be triggered by such colonization. Received November 4, 1989; accepted October 3, 1990. Correspondence: R. ~odinova:~adnikovi,Institute for Care of Mother and Child. Podolski: nabrezi 157, 147 10 Prague 4, Czechoslovakia.
MATERIALS AND METHODS
Infants. All infants included in our study were kept in a unit, where the infants were admitted for social reasons such as being born to seriously ill or single mothers or to parents without accommodation (8). Some of the infants were born to mothers who had decided to give up their child for adoption. The criteria for inclusion into our study were uncomplicated term pregnancy and delivery by a healthy mother. Fourteen infants were breastfed for 13 wk or longer and their mothers lived in the unit; 14 infants were admitted without mothers and received a "humanized" formula prepared from cow's milk. The mothers of the latter group were unable to breast-feed or uninterested in breastfeeding. After 4 mo, all infants received solids (fruits, vegetables) and cow's milk formula. The clinical characteristics of the infants are summarized in Table 1. The study was approved by the hospital scientific committee and the Ministry of Health, including ethical aspects according to the Helsinki Declaration. The infants participated in the study with the written informed consent of their mothers. In nine breast-fed and eight formula-fed infants, artificial colonization of the intestine was started during the first 6 d of life by oral administration of a suspension prepared from a live 24-h-old culture of the serotype E. coli 083:K24:H3 1 containing 5 x 10' organisms in 1 mL. One mL of the suspension was given to each child diluted in 10 mL of tea. three times a wk for 4 successive wk. Five breast-fed and six formula-fed infants were followed as controls. There was no marked difference in the duration of breastfeeding between those colonized with E. coli 0 8 3 (16.6 5 3.4 wk) and the noncolonized breast-fed controls (15.2 + 2.8 wk). All infants were healthy during the investigation. Bacteriologic examinations. Stool smears were cultivated aerobically on Drigalski agar plates. The last three free-lying colonies were isolated and placed in deep agar tubes that were transported to Goteborg for further analysis. Collecting those three colonies gave a probability of including the dominant strain of 97% (9). The strains were subcultured once on Drigalski agar to check the purity, and then once on tryptic soy agar. E. coli strains were identified by limited biotyping. The presence of E. coli 0 8 3 was detected by agglutination with a specific rabbit antiserum, prepared by repeated i.v. injections of lo9 formalinized bacteria once a week for 4 wk. Activity of the antiserum measured by ELISA exceeded a titer of 1/ 10 000. The presence of E. coli bacterial adhesins and their specificity was tested by agglutination of guinea pig erythrocytes in the absence and presence of mannose. Type 1 fimbriated strains agglutinate guinea pig red blood cells in the absence but not in the presence of mannose. Agglutination of latex beads coated
3s)6
397
ARTIFICIAL E COLI COLONIZATION IN INFANTS
with the disaccharide Gal a 1 4 Gal 13 was used to confirm the presence of type P fimbriae (10). All isolates were tested for adhesion to the human adenocarcinoma cell line HT-29 as described earlier (1 1). A total of 40 HT-29 cells were scored and the adherence value was defined as the mean number of adhering bacteria per cell. Analysis of' antibody responses. Samples of blood, stool, unstimulated saliva from the infants and breast milk from the mothers were taken before colonization and in the 2nd to 3rd, 4th to 7th, 8th to 1lth, and 12th to 15th wk after colonization. The same schedule of sampling was followed in the controls. Serum and saliva samples were kept at -20°C until use. One g from each stool sample was diluted in 5 mL of PBS, homogenized, and centrifuged at 600 x g for 20 min. The supernatant was kept frozen until further examination. Antibodies of the IgM, IgA, and IgG isotypes against the 0 8 3 antigen were measured by ELISA as described (12). The optimal dilution of the supernatant (150) of the sonicated suspension used for coating of microplate wells was determined by previous checkerboard titration, as well as the optimum of all other components used. After adding samples diluted 1 :5, peroxidaseconjugated swine anti-human antisera against IgM, IgA, or IgG (Institute of Sera and Vaccines, Prague, Czechoslovakia) were used and the reaction was developed with phenylendiamine (Lachema, Brno, Czechoslovakia) diluted in 0.01 M PBS, pH 6, containing H202.The reaction was stopped by adding 50 pL of 1 M H2SO4.The plates were read on a Microelisa Minireader MR 590 (Dynatech Laboratories, Alexandria, VA) at 495 nm. The level of antibody activity was expressed in percentage of references prepared from samples of, respectively, saliva, serum, stool, and milk from two mothers colonized with E. coli 083. The specificity of the ELISA assay was determined by inhibition studies. Ninety-seven percent of the binding activity of the hyperimmune serum was inhibited by incubation with an E. coli 0 8 3 sonicate. If an E. coli 0 8 6 sonicate of comparable OD was used, only 14% of the antibody activity was inhibited. Also, in adults colonized with E. coli 0 8 3 the inhibition of serum IgA antibody activity was significantly different if we compared E. coli 0 8 3 and E. coli 0 8 6 sonicates. Coefficient of variation intraassay was 6.9% and interassay 13.5%. Table 1. Clinical characteristics ofstudied infants Breast-fed
Formula-fed
n Maleifemale Birth wt (g) Birth ht (cm) Gestational age (wk) Apgar score l min 5 min
* Mean + SD.
Data analysis. Statistical evaluation included analysis of variance, Fisher's test, and the single nonparametric one-sided Rosenbaum's test (13, 14). All data in the figures are expressed as mean rt SEM. RESULTS
Bacteriologicfindings. The strain E. coli 0 8 3 was detected in the stools of all colonized infants from the 2nd d after oral administration and persisted throughout the study period. E. coli 0 8 3 represented a significantly larger fraction of E. coli strains isolated from breast-fed than formula-fed infants (46 versus 24%, p < 0.05; Table 2). In addition, the E. coli 0 8 3 from the breastfed infants adhered more efficiently to the colonic cell line HT29 than did those from the formula-fed infants ( p < 0.05). No mothers were found to be colonized with E, coli 083. Antibody response against E. coli 083. IgA antibodies against E. coli 083, found in breast-fed babies' stools whether they had been colonized or not, presumably originated from the mothers' milk (Fig. 1). In the formula-fed infants, the IgA antibodies to E. coli 0 8 3 increased from the 2nd wk after being colonized ( p < 0.05). The IgM anti-083 in the stool did not differ between the colonized and the control infants. IgA and IgM antibodies against the E. coli 0 8 3 were detected in saliva of colonized infants, whether breast- or formula-fed, and reached the highest levels between the 4th and 7th wk in comparison with the control groups ( p < 0.01) (Fig. 2). In the serum, no increase of IgG, IgM, or IgA antibodies could be detected against the 0 8 3 antigens. DISCUSSION
Bacterial colonization of the intestine or oral administration of killed bacteria has been tested as a prevention against enteric infections and as a stimulation of the local immune system in animals (15, 16), human adults (17), and also infants (1 8, 19). The antigenic structure of strains of E. coli 0 8 3 was described in 1978 by the International Escherichia and Klebsiella Center in Copenhagen as 083:K:H31. Recently, the K antigen was identified as K24 (0rskov F, personal communication). In E. coli strains isolated from infants with urinary tract infections, meningitis, or septicemia, the K1, K2, K3, K12, and K13 antigens were mostly detected (20-22). The strain E. coli 0 8 3 isolated from an infant with meningitis and septicemia by Czirok et al. (23) had no K and H antigens identified. The strain E. coli 083:K24:H31 used in our work has never been isolated from infants with meningitis or septicemia and has caused no complications in 1500 infants colonized in our studies of newborns (LodinovCZadnikovi R, manuscript in preparation). E. coli 0 8 3 was significantly more often recovered from colonized infants who were breast-fed than from those who were formula-fed. Also, the 0 8 3 isolates recovered from breast-fed infants had a higher mean adherence to colonic cells of the HT29 cell line than isolates of the same strain recovered from
Table 2. Detection of E. coli adhesins in stools of breast-fed and forrnzila-fed infants after natural and artificial colonization of intestine*
No. of infants
No. of samples
No. of isolates
P-fimbriated strains (%)
Type 1 fimbriated strains except E. coli 0 8 3 (%)
Breast-fed colonized
9
45
72
13
I1
Formula-fed colonized Breast-fed controls Formula-fed controls
8 5
40 25 30
94 39 52
10 11 14
19 22 37
Groups
6
Type I fimbriated E. coli 0 8 3 (%)
~ ~ 2 adherence E. coli 0 8 3 (mean, SEM)
46 p < 0.05 24 0 0
18 (1.4) p < 0.05 1 1.6 (2.7)
9
* Stool samples were taken before and 2-3, 4-7, 8-1 1, and 12-15 wk after colonization. The E. coli were subcultured on tryptic soy agar and tested for the presence of type 1 fimbriae by agglutination (mannose-sensitive) of guinea pig erythrocytes and for the presence of P-fimbriae by agglutination (mannose-resistant) of human erythrocytes and Gal a1 -+ 4 Gal 0-coated latex beads. Adherence to HT-29 cells was determined as stated in Materials and Methods.
% of
10'1
ref.
IgA
COLONIZED H breast fed ( ~ a 9 ) .--+ b o t t l e fed ( N - 8 ) I
CONTROLS I breast fed (N.5)
IgM lool
'before'
%---a b o t t l e fed ( ~ : 6 )
I
I
1 - 3
'4-7
' 8 - 1 1 '11-1Sdeeks
Fig. 1. Levels o f E. coli 083 IgA a n d IgM antibodies in stools. Results o f ELISA a r e expressed a s a percentage o f reference i n m e a n s S E M . T h e IgA 0 8 3 antibodies a r e significantly higher i n breast-fed colonized a n d control groups t h a n i n formula-fed controls i n all intervals ( p < 0.05). I n formula-fed colonized infants, t h e IgA 0 8 3 antibodies a r e significantly higher t h a n i n formula-fed controls i n intervals o f 2-3 a n d 4-7 wk after colonization ( p < 0.05).
+
COLONIZED H breast fed (Nag) *--a b o t t l e fed (N=B)
% of ref. IgA
secretory IgA in the gut, as in the breast-fed infants, may favor mannose-binding type 1-fimbriated strains and increase their expression of fimbriae. In our previous work, oral administration of a nonenteropathogenic E. coli strain 0 8 3 evoked an increase in total stool secretory IgA as well as a specific serum antibody response in infants (26). In the present work, the significant increase of the specific IgA and IgM response shown in saliva after oral colonization of the gut agrees with the concept of the common mucosal immune system. The presence of low levels of IgA and IgM antibodies already before colonization, i.e. in the first days of life, is in agreement with the findings of Mellander et al. (5). They showed a secretory IgA response in saliva against E. coli 0 antigens at birth and a subsequent increase. It was suggested that these salivary antibodies in newborns may result from priming in utero via maternal antiidiotypic antibodies (27, 28), and/or some degree of cross-reactions. The E. coli 0 8 3 IgA antibodies detected in stool samples, higher in both the colonized and noncolonized breast-fed groups, probably originated from the maternal milk. Significant production of E. coli IgA antibodies in the intestine after colonization could, however, be shown in the formula-fed group 4 to 7 wk arter colonization. The IgM and IgA antibodies to the 0 8 3 antigen in the stool most likely have the same origin as the saliva antibodies mentioned above. Actually, Mellander et al. (27) found IgA and IgM antibodies to E. coli 0 antigens already in meconium. The best way to protect young infants from infection is presumably breastfeeding (29). Some infants at high risk, such as premature infants, are unfortunately often deprived of fresh breast milk. The protective effect of intentional colonization with E. coli 0 8 3 has already been indicated to be eficient clinically. In infants, including high-risk neonates and premature infants, the presence of the 0 8 3 strain in the intestine prevented colonization with pathogens and displaced those present without any unfavorable side effects (30). Colonization with E. coli 0 8 3 might protect formula-fed infants by an early induction of antibody production, with the possibility of reactivity beyond the E. coli 083. REFERENCES
CONTROLS breast f e d ( ~ : 5 )
IgM
O
before
1-3
X-x
4- 7
8 -11
12-15 weeks
Fig. 2. Levels o f IgA a n d IgM E. coli 0 8 3 antibodies i n saliva. Results o f ELISA a r e expressed a s a percentage o f reference i n m e a n s +. S E M . T h e IgA levels a r e significantly higher i n both colonized groups c o m p a r e d with bottle-fed controls in t h e intervals o f 2-3, 4-7, a n d 8-1 1 wk ( p < 0.0 I ) a n d i n t h e interval 4-7 wk after colonization c o m p a r e d with breastfed controls. T h e IgM levels a r e significantly higher i n colonized groups t h a n i n both control groups i n intervals o f 2-3 a n d 4-7 wk after colonization ( p < 0.01).
formula-fed infants. The adherence of E. coli 0 8 3 to HT-29 cells is mediated via a mannose-specific interaction. Mannose-containing oligosaccharides functioning as receptors for type 1fimbriated E. coli are found in secretory IgA (24, 25). This raises the interesting possibility that the presence of high levels of
1. Gothefors L, Carlsson B, Ahlstedt S, Hanson LA, Winberg J 1976 Influence of maternal gut flora and colostral and cord serum antibodies on presence of E. coli in faeces of the newborn infant. Acta Pediatr Scand 65:225-232 2. Adlerberth I, Carlsson B, de Man P, Jalil F, Larsson P, Mellander L, Svanborg C, Wold A, Hanson LA 1990 Intestinal colonization with enterobacteria in Pakistani and Swedish hospital-delivered infants. Acta Paediatr Scand (in press) 3. Brskov F, Sorensen KB 1975 Escherichia coli serogroups in breast-fed and bottle-fed infants. Acta Pathol Microbiol Scand [B) 83:25-30 4. Hanson LA, Adlerbert I, Carlsson B, Dahlgren U, Jalil F, Khan SR, Zaman S, Larsson P, Mellander L, Sheikh AK, Soderstrom T, Wold A 1987 The ontogeny of the immune response: the role of maternal factors. In: Burgio GR, Hanson LA, Ugazio AG (eds) Immunology of the Neonate. SpringerVerlag, Berlin, pp 51-58 5. Mellander L, Carlsson B, Jalil F, S6derstriim T, Hanson LA 1985 Secretory IgA antibody response against Escherichiu coli antigens in infants in relation to exposure. J Pediatr l07:433-436 6. Lodinova R, Jouja V, Lanc A 1967 Experimentelle Besiedlungdes Darmtrakts von Neugeborenen mit dem E, coli 0 8 3 und Untersuchungen iiber die Bildung van Antikorpern. Z Immunitaetsforsch 133:229-237 7. Lodinova R, Jouja V, Vinsova N, Vocel J, Melkova J 1980 New attempts and possibilities of prevention and treatment of intestinal coli-infections in infants. Czech Med 3:47-58 8. Paul K, Vondracek J 1976 Individual differences in infant's sleep. Dev Med Child Neurol 18:182-188 9. Lidin Jansson G,Kaiiser B, Lincoln K, Olling S, Wedel H 1978 The homogeneity of the faecal coliform flora of normal schoolgirls, characterized by serological and biochemical properties. Med Microbiol Immunol (Berl) 164:247-253 10. de Man P, Cedergren B, Enerback S, Larsson A-C, Lemer H, Lundell A-L, Nilsson B, Svanborg Eden C 1987 Receptor-specific agglutination tests for detection of bacteria that bind globoseries glycolipids. J Clin Microbial 2540 1-406 I I . Wold AE, Thom6n M, Hull S, Svanborg Eden C 1988 Attachment of Escherichiu coli via mannose- or gal 1-4gal R-containing receptors to human colonic epithelial cells. Infect Immun 56:2531-2537
