One of the major differences between breast-fed and formula-fed infants is the development of the intestinal flora (1–11). Initial colonization of the aseptic intestine of the newborn under normal circumstances happens during delivery when it comes in contact with the vaginal flora of the mother and the normal flora of the parents. This leads to an inoculation with a diverse flora of bifidobacteria, enterobacteria, Bacteroides, clostridia, and Gram-positive cocci (1,12). After this first inoculation, the flora changes rapidly, presumably under the influence of diet. In the infants fed with solely human breast milk, within a few weeks a flora is established that is dominated by bifidobacteria, possibly caused by selective agents (bifidobacterial factors) that are present in human milk (11). Only after weaning does the flora become more diverse and begin to resemble that of adults (1). In contrast, formula-fed infants develop a more diverse flora, consisting, in addition to bifidobacteria, also of Bacteroides, enterobacteria, enterococci, and clostridia (1,7,12). In addition to the obvious influence of early diet, other factors also may influence the flora development such as method of delivery and hygiene (2,4,13).
Several investigators have speculated that the difference in intestinal flora between breast-and formula-fed infants contributes to the functional benefits that breast-feeding has over formula-feeding—that is, protection against gastrointestinal infections (14) and induction of oral tolerance to dietary allergens (15). Therefore, it is obvious that one of the goals for improvement of current infant formulas is to achieve an intestinal flora with formula feeding that is identical with that in breast-fed infants.
The difficulty in accurately assessing the gut flora from fecal samples is one of the major handicaps in the studies on the effects of early diet on the development of the intestinal flora. Most of the enumeration of the different groups of bacteria in these studies is accomplished by culturing on specific media. However, it is clear that culture techniques are hampered because certain bacterial species cannot be cultured (16) and because most media used for quantification of the groups are nonspecific (17). Alternative methods for reliable identification and detection of microorganisms are based on the molecular detection of 16S rRNA or its encoding gene (18). Polymerase chain reaction (PCR) amplification of the 16S rRNA gene directly from colonies found on agar plates, followed by sequence analysis, makes it possible to identify these colonies without further culture steps (19). Furthermore, specific 16S rRNA-based oligonucleotide probes have been developed that detect different groups of bacteria directly in fecal samples by means of fluorescent in situ hybridization (FISH), without further cultivation (20–22).
In this study, we investigated the development of the fecal flora in six breast-fed and six formula-fed infants during the first 20 days after birth, by using these molecular techniques. Although our results confirm the major conclusions from the previous studies in which conventional plating techniques were used, the new molecular identification and detection methods provide more accurate and additional data on the diversity, dynamics, and succession of bacterial strains after the initial colonization of the neonatal gut.
MATERIALS AND METHODS
Study Group and Sample Collection
Mothers were approached for participation in the study by U-Gene (Utrecht, The Netherlands). Those intending to breast feed their infants during the first 20 days of life were eligible for the breast-feeding group, and mothers intending to bottle feed their infants immediately after birth were eligible for the formula-feeding group. To prevent possible variability in the intestinal flora of the formula-fed group caused by composition-related differences among the formulas, only infants were included whose mothers decided to use the whey-predominant infant formula Nutrilon Premium (Nutricia, Zoetermeer, The Netherlands). The study included 12 infants: 6 breast-fed and 6 formula-fed. The children were mostly born at home or polyclinically, which means they were in the hospital for only 1 or 2 days. The mothers were asked to collect fecal material on day 3 or 4, day 4 or 5, day 5 or 6, day 6 or 7, day 8 or 9, day 12 or 13, and day 20 or 21 after the birth of the infant. These samples were collected and reduced in 20 ml 0.5% cysteine-HCl (pH 7.0). The samples were stored at 4°C and transported to NIZO Dairy Research. The samples were processed within 24 hours after collection.
Culturing on Selective Media
The weight of the fecal samples added to 20 ml cysteine-HCl was determined, and dilution series were made of the samples under anaerobic conditions using an anaerobic cabinet (Coy, Laboratory Products, Ann Arbor, MI, U.S.A.). The samples were analyzed using the selective media described in Table 1. All plates were incubated for 5 days at 37°C under anaerobic conditions, after which the colonies were counted. The colonies found on the different selective media were identified using random amplified polymorphic DNA (RAPD) fingerprinting and 16S rRNA sequence analysis (19). This molecular-based identification was chosen instead of biochemical characterization for two reasons: First, the identification of bacteria based on biochemical characteristics has been shown not to be fully reliable (23). Second, biochemical characterization of more than 2,500 isolates is too laborious to perform.
Identification of the Colonies
From all samples 20 to 30 colonies were randomly selected from all selective plates (three to five isolates per plate) and analyzed by RAPD fingerprinting. In this way, all isolates were classified into groups with similar RAPD patterns. RAPD patterns were generated using the following primer: 5´-GTCGTTATGCGGTA-3´. The RAPD patterns were compared per infant using Phoretix version 3.0 (Nonlinian Dynamics, Newcastle-upon-Tyne, UK).
Isolates with the same RAPD patterns were grouped, and one or two representatives of each of these groups were identified by 16S rRNA sequence analysis. This identification of species was achieved by amplification and sequencing the first part of the 16S rRNA (19). The sequences were compared to those present in the database of the Ribosomal Database Project (RDP) (24). Most RAPD groups were identified, with more than 99% identity with sequences of species present in the RDP database. Some of them showed less identity (between 95% and 99%) with known sequences. In those cases, only the genus name of the closest relative was used, followed by the suffix sp.
Fluorescent In Situ Hybridization Analysis
The fecal samples of six breast-fed and six formula-fed infants were fixed, using a 0.5-ml sample in cysteine-HCl storage buffer and 4.5 ml 4% paraformaldehyde in phosphate-buffered saline (PBS). These samples were stored at −80°C. Before analysis, the samples were transferred and concentrated into 1200 μl 50% ethanol in PBS. Depending on the amount of cells in the sample, part of the ethanol/PBS stock (usually 20 μl) was hybridized with group-specific fluorescent 16S rRNA–targeted oligonucleotide probes or stained with 4´,6-diamidino-2-phenylindole (DAPI) for total cell counts, as described before (21). The probes used in the study are listed in Table 2. For hybridizations with the Lab158 probe the samples were pretreated with lysozyme and lipase to permeabilize the Gram-positive cell wall, as described elsewhere (22). Subsequently, samples were washed and filtered on a 0.22-μm pore size filter, mounted in Vectashield (Vector Laboratories, Burlingame, CA, U.S.A.) on microscope slides, and enumerated using epifluorescent microscopy. Ten to 25 fields with a total of approximately 300 positive cells were counted for each sample.
Culturing on Selective Media
The bacterial population of the collected fecal samples was enumerated on nine different selective media. Generally, the log numbers from the total counts enumerated on the Colombia blood agar medium increased during the first days after birth from 9.5 to 10.1, and then to 10.4 to 11.3 after 2 weeks. On all of the specific media high counts were reached after 8 or 9 days. Table 1 shows the number of colony-forming units from the samples of breast-fed infant 4 cultivated on the different selective media. These data show a trend that was representative for the plate counts of the floras of the other infants. To analyze the biodiversity of the culturable fraction, 20 to 30 colonies representing the dominant flora were randomly selected from the different selective media and analyzed by RAPD–PCR fingerprinting. This showed that the main flora of the formula-and breast-fed children consisted of 10 to 20 different RAPD groups. Subsequent 16S rRNA sequence analysis of representatives of the RAPD groups resulted in a species identification. Table 3 shows that the media used were selective; however, unwanted species also grew on the different media. Furthermore, species such as Bifidobacterium spp, which were dominant in the samples, grew on most of the selective media. The species isolated from the fecal microflora from all 12 children are listed in Table 4. This table shows that bifidobacteria and Escherichia coli belong to the dominant normal flora isolated from the samples of either group of infants. Also enterococci, staphylococci, Bacteroides, and Veillonella could be isolated from both groups. From breast-fed infants more lactic acid bacteria were isolated, whereas from formula-fed infant feces, more clostridia and Bacteroides were isolated.
Fluorescent In Situ Hybridization Analysis
From each infant, six or seven fecal samples were analyzed by FISH, with three exceptions: the samples of the second and the last breast-fed infant and of the first formula-fed infant. Some of the samples of these infants did not contain enough biomass to be measured accurately. Figure 1 shows the flora development in the breast-fed infants measured by FISH, using the specific oligonucleotides. The relative numbers of the bacterial groups found in the samples are relative to the total number of cells counted after DAPI staining of the cells. Diverse flora was detected in the fecal samples obtained during the first days. Average bifidobacterial numbers were below 40%. Bacteroides counts ranged from 0% to 80% and E. coli numbers ranged from 0% to 30%. After this beginning, all bacterial floras became dominated by bifidobacteria, which make up from 60% to 91% of the flora between days 12 and 20. The flora of the first infant was dominated by bifidobacteria from the first day on. The first sample from infant 2 is missing, and the sample of day 5 shows flora apparently already dominated by bifidobacteria. This infant had a peak of streptococci of 16% on day 7 followed by an increase of E. coli with a maximum of 32% on day 12. Infant 4 had an increase of Bacteroides up to 11% beginning approximately at day 12, which coincides with a small peak of streptococci (9%) and a decrease in the bifidobacteria population.
Figure 2 shows the flora development of the formula-fed infants. In these infants initial flora similar to that of breast-fed infants was seen. In these cases, however, the bifidobacteria did not become as dominant in the following days as was seen in breast-fed infants. Their relative numbers ranged from 28% to 75%, with an average of approximately 50%. An exceptional flora was established in the last formula-fed infant in whom the relative numbers of bifidobacteria decreased below the detection limits of 0.01% of the total flora. In most fecal samples of the formula-fed infants the Bacteroides populations decreased after the high initial relative numbers and increased again toward day 20, ranging at that time from 35% to 61%. Only in the third formula-fed infant did the Bacteroides numbers decrease to 1%, which coincides with an increase in E. coli numbers up to 24%. In the remaining formula-fed infant floras, the relative E. coli numbers slowly decreased from approximately 40% in the first sample to approximately 5% in the last.
In the fecal samples of both groups the relative numbers of streptococci and lactococci varied from 0% to 7.5%, with one peak of 16% in breast-fed infant 2, as mentioned. Clostridia belonging to the Clostridium histolyticum/C. lituseburense group, including C. perfringens and C. difficile, were measured with the His150/Lit135 probes. Cells of this group were detected in some of the samples, but the numbers did not exceed 1%, except in the sample of day 12 of the second breast-fed infant in which 2.2% was measured. This coincided with the E. coli peak on that day. Lactobacilli and enterococci were usually detected in the samples of the first days varying from 0% to 4.6%. After this beginning, the relative numbers did not exceed 0.7%.
In this study, two new approaches were used to evaluate the development of gut microflora in newborn infants. Culturing on specific media and molecular identification of the growing colonies demonstrated the succession of species and the species diversity of the microflora of the two groups of newborns. The results also revealed that culturing cannot result in reliable counts, because the plates chosen for their selectivity were actually not selective. The FISH analysis with the specific probes provided quantitative data on the relative amounts of the different bacterial groups. The results from this new method confirmed the dominance of bifidobacteria and the dynamics of mainly Bacteroides and E. coli.
Although the counts on the selective plate were not useful to compare with the FISH data, for most of the species found in a particular sample by cultivation and sequence identification, the corresponding bacterial groups were usually also found by FISH analysis of the same sample. The exception to this was Bacteroides. For example in the first formula-fed infant, Bacteroides made up approximately 50% of the population. However, it was never cultured from the feces of this child, despite the good anaerobic transport and culture techniques. Either the sample had stayed too long in the aerobic diaper, or the Bacteroides species in this population were not able to grow on the media that were used. A comparison of the diversity of species cultured from samples with that of the probes used in the FISH is only possible to a limited extent, because probes for a few groups were not yet available. A probe (Erec482) for the C. coccoides/Eubacterium rectale group has recently become available (21). This group represents 29% of the adult fecal flora and should be incorporated in the next study. Interestingly, Table 4 shows that clostridia belonging to this group were only present in the flora of formula-fed infants. Therefore, low numbers can be expected with the Erec482 probe in breast-fed infants, whereas in formula-fed infants, higher numbers up to 30% may be present. Probes for C. innocuum and relatives, Veillonella, Staphylococcus, Propionibacterium and other probes are currently being designed and validated, and all these probes will be included in future studies.
The results obtained, confirm other studies in which the colonization of the intestine of newborn breast-fed and formula-fed infants are compared (1,3,5,7,10,11). In all these studies bifidobacteria became the dominant bacteria (>60%) within 1 week after birth when the newborn was fed with breast milk. Formula-fed infants showed development of a more diverse flora. Remarkably, in this study, the FISH results show that the number of Bacteroides cells equaled the number of bifidobacteria in formula-fed infants, whereas in the culture-based studies mentioned earlier the Bacteroides numbers remained 100-to 1000-fold lower. This clearly indicates that there is a problem in culturing this group of anaerobic bacteria, which can lead to large biases. In contrast, other researchers describe no significant differences between both regimens and note a flora with Bacteroides as dominant group. However, in these cases the numbers of cultured bifidobacteria are remarkably low, suggesting a problem in culturing of bifidobacteria (2,4,9). Other groups of organisms that are difficult to culture could have been overlooked. Besides bifidobacteria and Bacteroides, in breast-fed infants mostly lactic acid bacteria, such as streptococci and lactobacilli, are found, whereas formula-fed infants possess a flora with more staphylococci and clostridia. This could be an effect of pH, because earlier studies show that the stool of breast-fed infants has a lower pH than that of formula-fed infants (5).
The formula used in this study was a whey-predominant formula with a rather low protein content (1.4 g/100 ml) and is considered to be the closest to breast milk currently available in The Netherlands. However, our results indicate that the intestinal colonization of formula-fed infants continues to follow a different pattern than that in breast-fed infants. There have been many attempts to improve infant formulas to induce colonization more similar to that in breast-fed infants (1–11,29). These attempts have included searching for an optimal casein:whey protein ratio (5,8) and the addition of lactoferrin (10) or nucleotides (29). However, so far, no such formula inducing breast milk–like colonization exists.
Our results may stimulate new ideas on formula milk development such as the addition of probiotic lactic acid-producing bacteria. The first attempts, such as the addition of Bifidobacterium LW420 (30) and Lactobacillus rhamnosus GG (31) to infant formula are promising. The indication that the latter organism is able to promote endogenous barrier mechanisms in patients with atopic dermatitis provides a link between intestinal flora and modulation of the mucosal immune system. With the novel molecular identification and detection methods, it will be possible to investigate colonization of nonmaternal species, which has been suggested to play a role in the later development of atopy (15,32).
In conclusion, the molecular 16S rRNA-based techniques used in this study provide more accurate quantitative data on gut flora development in newborns than do conventional culture techniques, although the results of both techniques show the same trends. These novel molecular methods are well suited for testing the effect of improved infant milk formulations and for analyzing the results of desired and undesired flora modulation in infants by antibiotics or prebiotics and probiotics.
1. Stark PL, Lee A. The microbial ecology of the large bowel of breast-fed and formula fed infants during the first year of life. J Med Microbiol 1982; 15:189–203.
2. Simhon A, Douglas JR, Drasar BS, Soothill JF. Effect of feeding on infants' faecal flora. Arch Dis Child 1982; 57:54–8.
3. Yoshioka H, Iseki K-I, Fujita K. Development and differences of intestinal flora
in the neonatal period in breast-fed and bottle-fed infants. Pediatrics 1983; 72:317–21.
4. Lundequist B, Nord CE, Winberg J. The composition of the faecal microflora in breast-fed and bottle-fed infants from birth to eight weeks. Acta Paediatr Scand 1985; 74:45–51.
5. Kleesen B, Bunke H, Tovar K, Noack J, Sawatzki G. Influence of two infant formulas and human milk on the development of the faecal flora of newborn infants. Acta Paediatr 1995; 84:1347–56.
6. Mevissen–Verhage EAE, Marcelis JH, Harmsen–van Amerongen WCM, De Vos NW, Berkel J, Verhoef J. Effect of iron on neonatal gut flora during the first week of life. Eur J Clin Microbiol 1985; 4:14–9.
7. Balmer SE, Wharton BA. Diet and faecal flora in the newborn: Breast milk and infant formula. Arch Dis Child 1989; 64:1672–7.
8. Balmer SE, Scott PH, Wharton BA. Diet and faecal flora in the newborn: Casein and whey proteins. Arch Dis Child 1989; 64:1678–84.
9. Hall MA, Cole CB, Smith SL, Fuller R, Rolles CJ. Factors influencing the presence of faecal lactobacilli in infancy. Arch Dis Child 1990; 65:185–8.
10. Roberts AK, Chierici R, Sawatzki G, Vigi V. Supplementation of an adapted formula with bovine lactoferrin. Acta Paediatr 1992; 81:119–124.
11. Hudault S. Microbial colonization of the intestine of the newborn. In: Bindels JG, Goedhart AC, Visser HKA Eds. Recent developments in infant nutrition
. Dordrecht, The Netherlands: Kluwer Academic Publishers, 1996; 307–17.
12. Sakata H, Yoshioka H, Fujita K. Development of the intestinal flora
in very low birth weight infants compared to normal full-term newborns. Eur J Pediatr 1985; 144:186–90.
13. Grönlund MM, Lethonen OP, Eerola E, Kero P. Fecal microflora in healthy infants born by different methods of delivery: Permanent changes in intestinal flora
after cesarean delivery. J Pediatr Gastroenterol Nutr 1999; 28:19–25.
14. Koletzko B, Agget PJ, Bindels JG, et al. Growth, development and differentiation: A functional food science approach. Br J Nutr 1998; 80:S5–S45.
15. Hanson L, Telemo E. The growing allergy problem. Acta Paediatr 1997; 86:916–18.
16. Stackebrandt E, Rainey FA. Partial and complete 16S rRNA
sequences, their use in generation of 16S rRNA
pylogenetic trees and their implications in molecular ecological studies. In: Akkermans ADL, van Elsas JD, de Bruin FJ, eds. Molecular Microbiology and Ecology Manual.
Dordrecht, The Netherlands: Kluwer Academic Publishers, 1995; Part 3.1.1.:1–17.
17. Nelson GM, George SE. Comparison of media for selection and enumeration of mouse faecal flora populations. J Microbiol Methods 1995; 22:293–300.
18. Amann RI, Ludwig W, Schleifer K–H. Phylogenetic identification and in situ
detection of individual microbial cells without cultivation. Microbiol Rev 1995; 59:143–69.
19. Klijn N, Herman L, Langeveld L, et al. Genotypical and phenotypical characterization of Bacillus sporothermodurans strains, surviving UHT sterilisation. Int Dairy J 1997; 7:421–8.
20. Langendijk PS, Schut F, Jansen GJ, et al. Quantitative fluorescence in situ
hybridization of Bifidobacterium
spp. with genus-specific 16S rRNA
-targeted probes and its application in fecal samples. Appl Environ Microbiol 1995; 61:3069–75.
21. Franks AH, Harmsen HJM, Raangs GC, Jansen GJ, Schut F, Welling GW. Variations of bacterial populations in human feces measured by fluorescent in situ hybridization
with group-specific 16S rRNA
-targeted oligonucleotide probes. Appl Environ Microbiol 1998; 64:3336–45.
22. Harmsen HJM, Elfferich P, Schut F, Welling GW. A 16S rRNA
-targeted probe for detection of lactobacilli and enterococci in faecal samples by fluorescent in situ
hybridization. Microbiol Ecol Health Dis 1999; 11:3–12.
23. Tamine AY, Marchall VME, Robinson RK. Microbiological and technological aspects of mild fermented with bifidobacteria. J Dairy Res 1995; 62:151–87.
24. Maidak BL, Olsen GL, Larsen N, Overbeek R, McCaughey MJ, Woese CR. The RDP (Ribosomal Database Project). Nucleic Acids Res 1997; 25:109–11.
25. De Vos NW, Mevissen–Verhage EAE, Harmsen–van Amerongen WCM, Marcelis JH. A new selective medium for the culture of clostridia from human faeces. Eur J Clin Microbiol 1982; 1:267–71.
26. Silvi S, Rumney CJ, Rowland IR. An assessment of three selective media for bifidobacteria in faeces. J Appl Bacteriol 1996; 81:561–4.
27. Hartemink R, Kok BJ, Weenk GH, Rombouts FM. Raffinose-bifidobacterium (RB) agar, a new selective medium for bifidobacteria. J Microbiol Methods 1996; 27:33–43.
28. Poulsen LK, Licht TR, Rang C, Krogfelt KA, Molin S. Physiological state of E. coli
BJ4 growing in the large intestines of streptomycin-treated mice. J Bacteriol 1995; 177:5840–5.
29. Balmer SE, Hanvey LS, Wharton BA. Diet and faecal flora in the newborn: Nucleotides. Arch Dis Child 1994; 70:F137–40.
30. Kok RG, de Waal A, Schut F, Welling GW, Weenk G, Hellingwerf KJ. Specific detection and analysis of a probiotic Bifidobacterium strain in infant feces. Appl Environ Microbiol 1996; 62:3668–72.
31. Majamaa H, Isolauri E. Probiotics: A novel approach in the management of food allergy. J Allergy Clin Immunol 1997; 99:179–85.
32. Soothill JF, Stokes CR, Turner MW, Norman AP, Taylor B. Predisposing factors and development of reaginic allergy in infancy. Clin Allergy 1976; 6:305–19.
33. Alm EW, Oerther DB, Larsen N, Stahl DA, Raskin L. The ogligonucleotide probe database. Appl Environ Microbiol 1996; 62: 299–306.