Comparative Analysis of the Properties of Bifidobacterial Isolates From Fecal Samples of Mother–Infant Pairs

Abstract

Objectives: 

The aim of the study was to evaluate the possibility of Bifidobacterium breve transfer from the mother to her infant during the perinatal period.

Materials and Methods: 

Hundred isolates of B breve and 80 isolates of Bifidobacterium longum were collected from paired fecal samples of mothers and their infants. Bacterial DNA from the samples was comparatively analyzed by random amplification of polymorphic DNA. The growth of the bacteria was examined in vitro.

Results: 

The analysis of diversity in B breve population showed that infant-derived isolates had significantly less diversity than mother-derived isolates. The analysis of the similarity between these samples revealed that the number of shared type isolates tended to be higher in infants than in their mothers. In the isolates of B longum, however, no such difference was found in the diversity between mother- and infant-derived isolates. Examination of the growth of B breve strains revealed that the shared type strains have a significantly higher growth than nonshared strains both in the presence of galactooligosaccharides and at a higher redox potential.

Conclusions: 

These results suggested that subpopulations of B breve strains in the mothers may be transferred to their infants. Such populations may become dominant in the gut of infants at an early time after birth, during which time the transmission of the bifidobacteria of environmental origin is not yet established. B breve strain possessing a higher growth advantage in these conditions may be advantageous for colonization in the infant gut.

Under normal conditions, bifidobacteria become the most populous bacteria of a neonate within 1 week following birth, and these bacteria remain predominant in the gut during infancy. Moreover, bifidobacteria play a crucial role in protection of gut mucosa against pathogenic bacteria, and contribute to the development of both the intestinal tract and the mucosal immune system in infants ^(1,2). Although the colonization of the neonatal gut by bifidobacteria depends on the mother's microbiota ⁽³⁾, the mode of delivery ⁽⁴⁾, the nutrition after the birth ^(5,6), and infant hospitalization with antibiotic use ⁽⁶⁾, it remains unclear how bifidobacteria present in the mother's gut and vagina influence the development of bifidobacteria in the gut of her infant.

In a previous study ⁽⁷⁾, the influence of maternal bifidobacteria from feces and vagina on bifidobacterial colonization of the gut in infants was examined. Fecal samples from 110 healthy pregnant mothers within 1 month before delivery and their babies at 1 month of age, and 100 vaginal swabs from the mothers were collected. The bifidobacterial diversity and quantity were analyzed by polymerase chain reaction (PCR) and quantitative PCR, respectively, and the detection of B breve in the mother's feces was found to be significantly associated with an increase in both the diversity and the quantity of Bifidobacterium species in the baby's feces. Moreover, B breve also was the most frequent species in baby's feces. It therefore raised the hypothesis that B breve is a Bifidobacterium species capable of frequent transmission from mothers to newborn babies and a high degree of proliferation during the perinatal period.

To assess this possibility, the present study was designed to comparatively analyze the diversity, matching, and growth using both mother- and infant-derived isolates of B breve in the pair fecal samples. Isolates of B longum were used as controls due to the observations in a previous study that the maternal presence of this species had less effect on the status of bifidobacteria in the Japanese infants ⁽⁷⁾.

MATERIALS AND METHODS

Study Population For Fecal Sampling

In a previous study ⁽⁷⁾, a total of 110 mother–infant paired fecal samples were collected, and we examined whether both of the mother and her infant samples in the pair contained DNA of B breve or B longum. As a result, the number of these positive paired samples was 21 for B breve and 34 for B longum. In the present study, we attempted to collect >5 B breve isolates from the fecal samples of the mother and her infant in each pair of B breve-DNA–positive paired sample. Consequently, we isolated B breve strains from 15 of 21 DNA-positive paired samples, but obtained sufficient numbers of strains from only 10 of those 15 paired samples. Therefore, we collected a total of 100 strains (50 from the mother and 50 from the infant) from those 10 paired samples and used them in the following analyses. We excluded the other 5 paired samples from this analysis because we could not obtain sufficient numbers of strains from these samples. In the examination of B longum, we isolated the strains from 8 of 34 paired samples and, in total, were able to isolate 10 strains from these 8 pairs. There was no pair sample that was used for the detection of both B breve and B longum isolates (Fig. 1). The clinical characteristics of those mother–infant pairs used as donors in the present study are shown in Table 1.

FIGURE 1: Flowchart of the selection of paired samples for collection of bifidobacterial isolates.

TABLE 1: Clinical characteristics of the mother and infant pairs

Bifidobacterial Isolation and Identification

To isolate bifidobacterial strains, paired fecal samples were suspended in an anaerobic broth and serial dilutions were prepared. An aliquot from each dilution was spread on cephalothin-phosphomycin-lithium chloride-xylose (CPLX) agar plates to isolate the bifidobacterial colonies. The same amount of aliquots were also spread on bifidobacterium-liver agar (Nissui, Tokyo) plates to confirm that the aliquots for the CPLX agar plate contain a total of >1000 colonies of bacteria for screening. The plates were incubated for 2 days at 37°C in a jar equipped with AnaeroPack (Mitsubishi Gas Chemical, Tokyo), a disposable O₂-absorbing and CO₂-generating agent. Approximately, 10 independent colonies on the plates were randomly picked to obtain 5 B breve or B longum isolates from subject feces. Specific PCR primers targeting 16S rRNA genes of B breve and B longum were used to identify fecal bifidobacterial isolates ⁽⁸⁾. CPLX agar for the selection of bifidobacteria consists of a basal solution, a subsolution 1, and a subsolution 2. The basal solution consists of 10 g mannitol for the selection of B breve or fructooligosaccharides for selection of B longum, 10 g Trypticase (Japan Becton Dickinson [BD], Tokyo), 10 g yeast extract (Japan BD), 3 g KH₂PO₄, 4.8 g K₂HPO₄, 3 g (NH₄)₂SO₄, 0.5 g MgSO₄·7H₂O, 0.5 g L-cysteine·HCl·H₂O, and 15 g agar in 900 mL of water. Subsolution 1 consisted of 20 g D-xylose, 3 g LiCl, and 6 g sodium propionate in 100 mL of water. Subsolution 2 for B breve consisted of 0.78 mg cephalothin and 25 mg fosfomycin. Fradiomycin (0.1 mg) and 0.05 mg puromycin were further added to the subsolution 2 for B breve.

DNA Extraction

Bacterial DNA was extracted from feces using an UltraClean Soil DNA Isolation Kit (Mo Bio Laboratories, Carlsbad, CA) as described previously ⁽⁹⁾. Bacterial DNA from cells was isolated with a Wizard Genomic DNA Purification Kit (Promega, Madison, WI) according to the manufacturer's instruction. Briefly, the bacterial cell pellet from 1 mL of overnight culture was suspended in 480 μL of 50 mmol/L ethylenediaminetetraacetic acid. After 120 μL of 10 mg/mL lysozyme was added to the tube, cells were incubated at 37°C for 60 minutes and centrifuged for 2 minutes at 13,000g. After the addition of a nuclear lytic solution and RNase, the DNA-containing supernatant was transferred to a clean tube. Isopropanol was added to the supernatant, and the thread-like DNA was removed from the tube. The resultant DNA was suspended in rehydration solution and kept at 4°C until use.

Random Amplification of Polymorphic DNA

Each PCR mixture (50 μL) was composed of 5 μL of 10x ExTaq buffer (TaKaRa Co, Kyoto, Japan), 2 mmol/L MgCl₂, 200 μmol/L deoxyribonucleoside triphosphate mixture, 2 μL of template DNA (25 ng/μL), 2 μL of primers (10 pmol/μL), and 1.25 U of Taq polymerase (TaKaRa ExTaq, TaKaRa). The following random primers reported by Vincent et al ⁽¹⁰⁾ as follows were used: OPA-02 (5′-TGCCGAGCTG-3′), OPA-13 (5′-CAGCACCCAC-3′), and OPL-07 (5′-AGGCGGGAAC-3′). PCR was carried out with a Mastercycler ep gradient S (Eppendorf, Hamburg, Germany). The following amplification program was used: 1 cycle at 94°C for 3 minutes, followed by 45 cycles of 35°C for 1 minute, 72°C for 2 minutes, 94°C for 1 minute, 1 cycle at 35°C for 1 minute, and 1 final cycle at 72°C for 10 minutes. After synthesis, the PCR samples were rapidly cooled at 4°C. Next, 2 μL of 5x loading buffer (TaKaRa) and 3 μL of water were added to 5 μL of the amplified random amplification of polymorphic DNA (RAPD) product. Ten microliters was resolved on a 1.4% agarose gel in 1x Tris-acetate-EDTA buffer at 100 V/25 cm on ice for 12 hours, and visualized by ethidium bromide staining and UV transillumination. We used the test strains for B breve (T104, JCM1192, and JCM1273) and B longum (JCM1217, JCM7050, and JCM7052) ⁽¹¹⁾. As a result, the RAPD profile was clearly and reproducibly distinguishable among the test strains of B breve and B longum (data not shown).

Bifidobacterial Cultures

Shared and nonshared isolates of B breve obtained from mothers and infants were examined for growth in vitro. For the culture analyses, 1 strain of the mother-derived or infant-derived isolates was randomly taken in each pair sample. These strains were suspended in 4 mL of Gifu anaerobic medium broth (Nissui Pharmaceutical, Tokyo) and incubated in an anaerobic jar overnight. Then, 0.1 mL of the suspension was used to inoculate a culture containing 4 mL of fresh Gifu anaerobic medium broth and was incubated overnight. Ten microliters of the second bacterial suspension, adjusted to an optical density (OD) value of 0.1, was added to a tube containing 1 mL of Eggerth Gagnon (EG) broth (Nissui Ltd) for the examination of growth in the presence of oligosaccharides. One percent (wt/vol) of galactooligosaccharide (Oligomate 55 NP, Yakult Pharmaceutical Industry, Tokyo, Japan), 1-kestose (β Food Science Co, Ltd, Tokyo, Japan), or glucose was added to the culture tube before the start of incubation. Next, the culture was anaerobically incubated in a jar at 37°C for 12 hours. At 6, 9, and 12 hours after incubation, an aliquot was removed to determine the bacterial density by spectrophotometric absorbance at 550 nm. For the examination of growth at different redox potentials, EG broth in a flask was aseptically aerated using an air tube with a 0.20-μm syringe filter (Iwaki Glass, Tokyo, Japan). The oxidation-reduction potentials of the broth before and after aeration were approximately –190 mV and –40 mV, respectively, when measured by a pH meter (Navi D-52S, Horiba, Kyoto, Japan). Ten microliters of the second bacterial suspension (at OD 0.1) was added to a tube containing 1 mL of EG broth at –190 mV or –40 mV and then incubated in a jar at 37°C.

Statistical Analysis

Differences in the clinical characteristics and the OD values were examined by Student t test. The Wilcoxon signed-rank test was used to compare the differences in the diversity of RAPD patterns and the number of the isolates showing shared or nonshared RAPD patterns. The matching of the RAPD patterns between the mother- and infant-derived isolates was analyzed by Fisher exact probability test. Values of P < 0.05 were considered significant. The SSPS (SSPS, Chicago, IL) software program version 16.0 for Windows was used for the data analysis.

Ethical Considerations

The present study protocol was approved by the Committee on Ethical Practice of Tokai University School of Medicine. Written informed consent was obtained from all of the mothers.

RESULTS

Analysis of Diversity by RAPD Profiles

Using optimized RAPD condition, B breve isolates derived from mother–infant paired samples were compared. Figure 2 shows a representative of the banding pattern of 10 B breve isolates from a pair of samples (pair no. 18). Primer OPA13 exhibited 3 different banding patterns (a, B, C) in mother-derived isolates and 3 different patterns (C, d, B) in infant-derived isolates in the paired samples (Fig. 2). Therefore, the numbers of isolates in mother- and infant-derived isolates by this primer were 3 and 3, respectively. The same RAPD profile analysis was also performed in the other paired samples using this primer. As a result, the mean numbers of RAPD banding patterns were 2.2 and 1.4 in mother- and infant-derived isolates, respectively (Table 2). According to this methodology, the analyses were conducted using 3 different primers in both B breve and B longum isolates (Table 2). Consequently, in B breve, infant-derived isolates showed significantly less diversity than mother-derived isolates (P = 0.01, 0.05, and 0.02 in OPA13, OPL07, and OPA02, respectively). In B longum no significant difference in the diversity between mothers and infants was found when examined using the 3 primers.

FIGURE 2: RAPD profiles of bifidobacteria using primers OPA13. The banding patterns of mother-derived (isolates numbered 1–5) and infant-derived (isolates numbered 6–10) B breve isolates are shown. All of the isolates were picked up from paired sample number 18. In the “Pattern” at the bottom, each letter represents a particular banding pattern. Large and small letters indicate the isolate with the shared and that with nonshared isolates, respectively, between the mother-derived and infant-derived isolates. RAPD = random amplification of polymorphic DNA.

TABLE 2: Less diversity of RAPD patterns in infant-derived B breve isolates than in mother-derived ones in the paired samples

Analysis of Similarity by RAPD Profiles

For the analysis of similarity, RAPD banding patterns were compared between the mother- and infant-derived isolates in each pair using the OPA13, OPL07, and OPA02 primers together. In 1 pair of samples (pair no. 18), primer OPA13 showed that the patterns of 3 isolates from the mother (no. 3, 4, 5) were shared with the patterns of isolates from the infant, whereas the patterns of 4 isolates from the infant (no. 6, 8, 9, 10) were shared with those from the mother (Fig. 2). Primers OPL07 and OPA02 showed the same pattern in all of the isolates of this paired samples except for isolate no. 7 (data not shown). Therefore, there were 3 isolates of shared type (large letters) and 2 isolates of nonshared type (small letters) in the mother, and 4 of shared type and 1 of nonshared type in the infant in the paired samples. According to this methodology, we also examined the other paired samples for B breve and those for B longum (Table 3). Consequently, the mean number of B breve isolates bearing a shared type tended to be higher in infants than in mothers (1.9 ± 2.5 vs 1.0 ± 1.3; P = 0.07) in the paired samples. In the isolates of B longum, no such difference in the number of shared type was found at all between the isolates from infants and those from the mothers (0.25 ± 0.71 vs 0.13 ± 0.35; P = 0.32).

TABLE 3: Isolates showing shared and nonshared type

Comparison of Matching Between B breve and B longum

In the comparison between 2 Bifidobacterium species, the number of shared type isolates was far higher in B breve than in B longum (mean numbers 1.0–1.9 vs 0.13–0.25; Table 3). Because the shared type implies being matched in the pair, a matching analysis of the RAPD pattern between the mother- and infant-derived isolates was then conducted regarding B breve or B longum. In mother-derived isolates, the total number of isolates whose RAPD patterns matched with those of the infant-derived isolates was 10 and 1 of 50 B breve and 40 B longum isolates, respectively (Table 4). In addition, these differences in the rate of matching between the B breve and B longum isolates were found to be statistically significant (P = 0.02) (Table 4). This difference was also significant in the analyses using the infant-derived isolates (P = 0.0002).

TABLE 4: Matching of the RAPD banding pattern between mother- and infant-derived isolates of Bifidobacterium

Growth Differences Between Shared and Nonshared Type B breve Strains

Because some B breve isolates were shared between the mother and her infant, they were suggested to be directly transferred. Conversely, the other isolates were nonshared, suggesting a nontransfer mechanism. To identify the properties of the strain of B breve and how these properties affect its transfer between mothers and infants, we examined the growth under different culture conditions for both the shared and nonshared strains of B breve. First, the growth was examined in the presence of galactooligosaccharides, which are a major component of a mother's milk (Fig. 3A). As a result, in mother-derived strains, the shared strains showed a significantly higher growth than nonshared strains. Moreover, in the presence of the short-chain fructooligosaccharide 1-kestose, shared strains tended toward higher growth as well. However, in infant-derived strains, no such significant differences in the growth were observed when comparing the shared and nonshared strains in response to oligosaccharides. Next, we examined the growth of these strains under different redox potentials (Fig. 3B) because the redox potential of the gut at birth, when the bacteria from the mother first establish residence on the aseptic gut of the infant, is higher than the later redox potential. At a higher redox potential (−40 mV), the growth of shared strains was significantly higher than that of nonshared strains in mother-derived strains. At a lower redox potential (−190 mV), no significant differences were found between these 2 groups of strains. In infant-derived strains, no such difference was found between the shared and nonshared strains at a higher or a lower redox potential.

FIGURE 3: Growth of bifidobacteria in vitro. The vertical axis represents the optical density value of the suspension medium of the culture after incubation for 9 hours, and peak growth was observed at this time point. Bars and asterisks represent the mean values and the statistically significant differences (P < 0.05), respectively. Growth of B breve strains were examined in vitro in the presence of 1% (wt/vol) glucose (upper panel), galactooligosaccharides (middle panel), and 1-kestose (lower panel) using the transferred (n = 9)/nontransferred (n = 13) strains derived from the mothers and the transferred (n = 14)/nontransferred (n = 8) strains derived from infants. Growth of B breve strains were examined in vitro both at a higher (−40 mV; upper panels) and at a lower (−190 mV; lower panels) redox potential using the shared (n = 8)/nonshared strains (n = 9) strains derived from the mothers and the shared (n = 12)/nonshared strains (n = 6) strains derived from the infants.

DISCUSSION

The detection of B breve in the feces of mothers was significantly associated with increases in both the count and diversity of bifidobacteria in the feces of infants in our previous study ⁽⁷⁾. There were, however, a few limitations to that study in the elucidation of how B breve from the mother affects the status of bifidobacteria in the baby. One limitation was that the analysis regarding the strain of bifidobacteria was not previously performed. Therefore, in the present study using isolates from the feces of the mother and infant pairs, the RAPD analyses were performed to examine the possibility of the transmission of B breve from a mother to her infant.

Genomic DNA fingerprinting by a RAPD analysis is useful to differentiate closely related bacteria ⁽¹²⁾. This technique is sensitive and specific because the entire genome of a bacterium is the basis for generating a DNA profile. Genomic variation among bacteria is identifiable as differences in the sizes and numbers of DNA fragments. Vincent et al ⁽¹⁰⁾ reported that the RAPD method using 5 kinds of primers could clearly distinguish among 10 to 20 strains of a particular Bifidobacterium species. In the present study, we chose 3 kinds of primers from those used in their study for examination of 10 isolates from B breve or B longum species in each pair. Although the RAPD method cannot definitively reveal the identity among strains, it can clearly illustrate differences among them.

Using this method, we have demonstrated that the B breve population in the feces of infants has a significantly lower diversity than the B breve population in the feces of their mothers. In addition, the number of shared isolates of this species was higher in infants than in their mothers. These results suggest that a limited subpopulation of B breve strains in the mothers transfers to their infants and grows up to occupy a considerable size in the gut of infants at an early time after birth when the transmission of the bifidobacteria of environmental origin has not yet been established. In B longum, no difference in the diversity between the isolates from mothers and those from their infants was found. Moreover, the number of the isolates bearing a shared type in the paired samples was about 10 times less in this species compared with B breve. These results suggested that B longum is infrequent in the transfer between mothers and infants.

B breve is the most common species of Bifidobacterium in breast-fed infants in Japan ^(8,13). In our previous study regarding the diversity of Bifidobacterium species, B breve was also the most prevalent species in the feces of infants, although this species was less prevalent in the feces of mothers ⁽⁷⁾. In contrast, B longum was most prevalent in the feces of mothers but less frequent in that of infants. A process of natural selection appears to occur among bacteria colonizing the gut of infants. The diet of the newborn primarily contributes to the selection process. The mother's breast milk contains a large amount of galactooligosaccharides, which selectively accelerate the growth of bifidobacteria ^(14,15). Such oligosaccharides in the mother's milk are a crucial factor that explains the rapid domination of bifidobacteria in the gut of breast-fed infants. Moreover, in our previous study using strains isolated from the feces ⁽⁷⁾, we found that B breve strains had a significantly higher growth than those of B longum in the presence of galactooligosaccharides in vitro. In a clinical study ⁽¹⁶⁾, a larger quantity of B breve versus B longum was also observed in the gut microbiota of subjects supplemented with galactooligosaccharides. In that study, the prevalence of positive samples for B breve and B longum was 35% and 12% after supplementation, respectively. These results indicated that the high level of galactooligosaccharides in the gut of breast-fed infants resulted in B breve becoming the most prevalent Bifidobacterium species in the subjects' guts because 95% of infants in our study had been receiving breast milk. Therefore, these results suggest that B breve is the species that can most easily colonize the gut of infants once moving to this location. Other Bifidobacterium species such as B longum do not easily colonize upon arrival at the gut of the infants because they grow at a slower rate than B breve in the presence of oligosaccharides.

In the study to explore the diversity of the bifidobacterial population in the gut of infants in Europe ⁽¹⁷⁾, B breve occupied just 2% of the population, but other species such as B longum (47%) and B pseudocatenulatum (14%) were far more predominant than B breve. It thus suggested that the predominance of B breve in infants may not be applicable to countries other than Japan. According to our data in Japan ⁽⁷⁾, the prevalence of B breve was considerably high in mothers (26%) in comparison to the prevalence in adults in that European study (∼6%). The reason for this may be the fact that the Japanese generally consume a lot of vegetables and fruits containing oligosaccharides, which thus enables B breve to demonstrate advantageous growth as shown in a previous study ⁽⁷⁾. Such a considerably high prevalence of B breve in mothers itself may contribute to the predominance of this species in the gut of infants in Japan. The country of birth of an individual is therefore thought to be one of the important variables that influence the bifidobacterial status of infants.

Among B breve strains, the properties that determine which strain can colonize and persist in the gut of infants remain unclear. To investigate these properties in the present study, we used the strains isolated from the paired samples according to their RAPD profiles. In the mother-derived strains of B breve, as a result, the shared type exhibited a significantly higher growth in the presence of galactooligosaccharides in vitro than the nonshared type. Considering that most of the donor infants were breast-fed, it is likely that the rapid growth of the shared strains in this rich milieu of oligosaccharides enabled the successful colonization in the gut of infants once they relocate from the mother. Although such a significant difference was observed in maternal strains, no such difference was observed in the infant-derived stains between the shared and nonshared ones (Fig. 3A). With regard to nonshared strains, the strains isolated from infants showed a rather higher growth than those isolated from mothers in the presence of galactooligosaccharides (Fig. 3A). This increased growth of the nonshared strains isolated from infants resulted in the disappearance of the significant difference observed between shared and nonshared strains in infant-derived strains. It is hypothesized that the galactooligosaccharides-induced growth response results in an optimal environment in which particular bacterial strains can optimally proliferate in the gut of infants. Therefore, even in nonshared strains that are considered primarily of environmental origin, the strains exhibiting higher growth ability in the presence of galactooligosaccharides prevailed in the gut of infants. In contrast, the nonshared strains exhibiting relatively low growth rates in response to galactooligosaccharides can live in the gut of the mother because fewer galactooligosaccharides are present; therefore, galactooligosaccharides-induced selection cannot occur.

Each bacterial species requires its own redox potential for optimal growth to ensure a smooth flow of energy and to promote nutritional economy ⁽¹⁸⁾. Anaerobic bacteria such as bifidobacteria grow more successfully at a low redox potential than at a high redox potential as shown in Figure 3B. Because the redox potential of the gut of newborn infants is relatively high, E coli and other enterobacteria are the first colonizers of the gut ⁽¹⁹⁾. As a result of these bacteria's metabolisms, the redox potential drops and therefore bifidobacteria attain the optimal growth conditions in the gut of infants. At a higher redox potential in the present study, the same selection was observed in mother-derived strains as was seen in the presence of galactooligosaccharides. Specifically, shared strains showed a significantly higher growth than nonshared strains, although the degree of growth itself was far lower compared with the growth at a lower redox potential. This suggests that the B breve strains with a higher growth potential at a higher redox potential can more successfully colonize the gut of newborn infants after transfer from the mother.

Taken together, there are several reasons why the residence of B breve in the gut of the mother was significantly associated with an increased bifidobacterial count in the infant gut. As previously shown ⁽⁷⁾, B breve was the Bifidobacterium species capable of growing more successfully in the presence of galactooligosaccharides compared with other species such as B longum. Therefore, it is likely that B breve rapidly proliferates in the gut of breast-fed infants if it has already been colonizing the gut of the mother and then moves to the gut of baby during delivery and shortly after birth. The B breve strain possessing a higher growth advantage both in the presence of galactooligosaccharides and at a higher redox potential is therefore considered at a greater advantage of achieving increased colonization of the gut of infants.

There is accumulating evidence that the transfer of bacteria through breast milk may be a means by which maternal microbes colonize the neonatal gut. A study in Europe ⁽²⁰⁾ indicated that Bifidobacterium was isolated in breast milk, although Streptococcus was the genus most frequently isolated. However, it was also reported that even aseptically collected breast milk is primarily composed of Lactobacillus, Streptococcus, and Enterococcus, but no Bifidobacterium^(21,22). Therefore, it remains to be elucidated whether the bifidobacteria colonizing the gut of infants in the present study were transferred from the mothers via breast milk.

Ouwehand et al ⁽²³⁾ reported that allergic infants were predominantly colonized by adult-like Bifidobacterium such as B adolescentis but, to a lesser extent, by infant-like Bifidobacterium such as B infantis and B breve. In addition, they also hypothesized that individual Bifidobacterium species rather than the entire genus can affect the manifestation of allergies. With respect to Bifidobacterium, a predominance of B breve in the gut appears to be advantageous for infants. In fact, a study by Kitajima et al ⁽²⁴⁾ using very-low-birth-weight infants found that B breve colonized the immature bowel effectively and it was associated with fewer abnormal abdominal signs such as aspirated air volume and a better weight gain than that seen in the control group. Li et al ⁽²⁵⁾ also reported that the early administration of B breve to low-birth-weight infants helped to promote Bifidobacterium colonization, thus leading to normal intestinal microbiota development. In the present study, we modified our previous conclusion, namely that the presence of B breve in the mother's gut may facilitate transfer of bifidobacteria to the infant's gut. Thus, the administration of selected B breve strains as probiotics to mothers may offer new modalities to guide the bifidobacterial microbiota in the gut of newborn infants, although future clinical studies are needed to assess its efficacy.