The human large intestine is a complex ecosystem harbouring a vast range of bacteria (1). To address specific questions on the human gut microbiota, in vitro and in vivo models have been developed. In the in vitro model, an array of techniques have been used, ranging from simple batch culture and screening the fermentative capabilities of specific compounds to more complex continuous culture models (2,3). Such models maintain a microbiota that in many respects is similar to that of the human colon. Hence, the microbial composition of these models have correlated well with in vivo samples and thus allowed the studies of a variety of ecological situations, including the fermentation of both dietary and endogenous substances (4). These models have been extensively applied to various aspects on both probiotic and prebiotic research. However, to more definitively assess the complete functionality and safety of novel food ingredients, in vivo approaches are required (5).
Animal models such as human microflora–associated rats and neonatal pigs are more able to reflect the complexity of the microbial gut flora in terms of species and numbers, as well as reproducing some biotic parameters such as a mammalian mucosa, immune system and digestive functions. However, subtle differences such as different anatomy and nutrient requirements may put the accuracy of such results in doubt. A more realistic model of the human gut is the use of primates, even though there are logistical and economic problems with this approach. Nevertheless, primates such as baboons and rhesus monkeys have been successfully used in studying novel food ingredients (6–8). These primates also have the advantage of similar nutrient requirements and weaning periods as human infants, thus making them a useful animal model (9).
The final assessment of prebiotic effects, however, is to feed candidate substrates or food to human volunteers and assess concomitant microbiological changes in stools (5). Clinical intervention trials have previously been conducted using a wide variety of candidate prebiotics, although the doses, duration of the intervention and number of volunteers have varied. The main drawback with clinical trials is that faeces is the only readily available test material. Therefore, in contrast to in vitro continuous cultures, one cannot assess fermentation results in the proximal colon or metabolic events in the proximal gastrointestinal tract. However, if a candidate prebiotic is able to induce changes in the faecal microbiota or in faecal metabolic factors, then it may be assumed that more saccharolytic environments (eg, the right side of the large gut) are also affected (10). In effect, a randomised placebo-controlled intervention human trial is seen as the ultimate assessment of a candidate prebiotic/probiotic or other food ingredient because it allows researchers to record data from an individual for whom the novel ingredient potentially is intended.
Two milk components have previously been shown to modify virulence and stimulate bifidobacteria in vitro while inhibiting gastrointestinal infections in infant rhesus monkeys (6). Casein glycomacropeptide (GMP), a cleavage-product of κ-casein, has been shown to prevent haemagglutination by Streptococcus mutans, S. sanguis and Actinomyces viscosus(11). Furthermore, Strömquist et al (12) showed that when sections of formalin-fixed, paraffin-embedded stomach tissue were incubated with purified human κ-casein, adhesion of Helicobacter pylori to the tissue was inhibited. In vitro assays also showed that GMP strongly promoted the growth of Bifidobacterium breve, B. bifidum, B. infantis and Lactococcus lactis(13,14). α-Lactalbumin (α-lac), which has a tertiary structure comparable to c-type lysozymes, has similar effects (15). Pihlanto-Leppälä et al (16) previously showed that α-lac lowered the metabolic activity of Escherichia coli JM103 to 80% of normal. In addition, the testing of purified cow's milk whey proteins showed that α-lac was a potent growth promoter for B. infantis and B. breve(17). The mechanism behind the effect of α-lac is unclear, but it is speculated that whey products hamper cellular metabolism in a manner similar to that of lactoferrin-derived antibacterial peptides, that is, by affecting the cytoplasmic membrane.
In this study, we evaluated the prebiotic effect of α-lac– and GMP-supplemented infant formulae in healthy term infants. We tested whether enrichment of standard formula with bovine α-lac and GMP altered the gastrointestinal flora, making it closer to that of breast-fed infants as compared with infants fed control formula.
PATIENTS AND METHODS
This randomised, double-blinded, placebo-controlled clinical intervention was carried out in Umeå, Sweden, and was approved by the ethics committee of the Faculty of Medicine, Umeå University (Umeå, Sweden). Infant formulae with different levels of GMP and α-lac were provided by ARLA Food Ingredients (Viby, Denmark).
Healthy term infants ages 6 ± 2 weeks were recruited from well baby clinics in Umeå, Sweden. All infants were exclusively breast-fed from birth and were weaned directly to either one of the study formulas or a standard infant formula for a short time before inclusion in the study. Infants whose parents gave informed written consent were randomised to either (1) a control, commercially available formula (Standard) (11% α-lac, 14% GMP) (n = 17); (2) a formula enriched in α-lac (Alpha-10) (25% α-lac, 15% GMP) (n = 21); or (3) a formula enriched in α-lac but with reduced GMP (PSNU) (25% α-lac, 10% CGMP) (n = 16) (Table 1). Breast-fed infants served as reference (n = 31). The formulae were in powdered form and dry-blended minimize the denaturation of the proteins. All subjects and their parents were in regular contact with the research nurse and had access to 24-h medical advice through the Department of Pediatrics. The intake of either formula or breast milk was ad libitum, and no other food except 1 to 2 tablespoons of fruit purees per day from 4 months of age was allowed during the study. All subjects were free to leave the trial at any time for any reason.
Faecal samples were taken at the start of the study (baseline), at 2 months and at 6 months of age. On the day of sampling, approximately 1 g of fresh faeces was scraped off the draper, placed into an Eppendorf tube and immediately frozen. Samples were stored at −80°C and subsequently transported, blind coded, to the University of Reading for further analysis. All faecal material was analysed for predominant populations of the gut microbiota using fluorescence in situ hybridisation (FISH) (6).
Chemical reagents were obtained from Sigma (Poole, UK), unless otherwise stated. Each sample was defrosted, placed in preweighed 50-mL centrifuge tubes and weighed. Phosphate-buffered saline (PBS; 0.1 mol/L), pH 7.4, containing 0.00001% (wt/vol) cetyltrimethylammonium bromide was added to make a 1:10 (wt/vol) solution. Approximately 6 glass beads (4.5 mm Ø, BDH, Poole, UK) were then added and the samples vigorously vortexed. Three hundred seventy-five microliters of homogenate was then added to 1125 μL of filter-sterile (0.2 μm) 4% (wt/vol) paraformaldehyde in PBS (pH 7.2) and fixed for at least 4 h at 4°C (triplicates). The fixed sample was washed twice in 1 mL of filtered PBS (pH 7.4) and pelleted at 13,000g, before resuspending in 150 μL filtered PBS and 150 μL 96% (vol/vol) ethanol. The sample was mixed and stored at −20°C for at least 1 h before analysis by FISH (as previously described) using probes for Lactobacillus, Bifidobacterium, Clostridium, Bacteroides and E. coli and 4′,6-diamidino-2-phenylindol for total counts (2) (Table 2).
Statistical analysis was performed using one-way ANOVA for determining significance at P < 0.05. Variation of the intestinal microflora at the start of the study and after 4 and 6 months was compared for each diet in addition to deviation between diets (Breast Milk—Standard formula, Breast Milk—PSNU, Breast Milk—Alpha-10, PSNU—Alpha-10, PSNU—Standard, Alpha-10—Standard). Statistical design and analysis was decided on in agreement with the Department of Applied Statistics, The University of Reading, UK.
There was no difference in gastrointestinal or other symptoms between the groups (to be reported elsewhere). Parents did not express any concern regarding the formulas used and were generally satisfied.
Lactobacillus, Bifidobacterium, Clostridium, Bacteroides and E. coli were quantified using FISH. Breast-fed infants demonstrated a significant increase (P < 0.05) between baseline and 4 months in total counts. Alpha-10 showed a significant increase in the total bacteria quantified between baseline to 4 months (P < 0.012) and 4 to 6 months of age (P < 0.05). In the PSNU and Standard groups, large but statistically insignificant variations between baseline to 6 months were seen. Average levels of total bacteria decreased slightly by 6 months of age in infants fed PSNU and the breast-fed group. However, this was not statistically significant due to large individual variations within the groups. Bifidobacterium was the predominant population in all infants, and no significant changes in bifidobacteria populations were observed throughout the study in any group, with the exception of breast-fed infants. Breast-fed infants showed a peak of increased bifidobacteria levels at 4 months of age when compared with the other groups. However, levels were again indistinguishable at 6 months. A bifidogenic effect was seen in the percentage difference from baseline to the sixth month (16.8%–50.2%) in the infants fed standard formula with decreased α-lac and increased GMP (Standard formula). However, this did not reach statistical significance, probably due to large individual variations within the group. Lactobacilli levels remained unchanged in all groups throughout the study. Standard formula displayed a dramatic rise in average Clostridium populations (P < 0.039) between baseline and 6 months of intervention. In the percentage-quantified profile, E. coli showed an increase during these 2 time points (4–6 mo), for Alpha-10, but again, due to the large standard deviation, no statistical significance was found. On the whole, Alpha-10 showed a varied, more “adult-like” bacterial population (Figs. 1–4).
ANOVA analysis showed significant differences in the following bacterial populations: Bacteroides, E. coli, Clostridium and total counts. Infants fed Alpha-10 outperformed all other feeding groups in the fourth month with the lowest bacteroides counts. Significant differences were found between Breast-fed and Alpha-10 (P < 0.03) and Standard and Alpha-10 (P < 0.02) bacteroides populations. PSNU-fed infants displayed similar but not significantly different results (P < 0.07) when compared with infants fed Standard formula and Breast-fed infants. By 6 months of age, average E. coli counts differed significantly in infants that were exclusively breast-fed. Although varying carriage of E. coli amongst the time points and groups was observed, average counts remained most stable only in the breast-fed group (Breast-fed and PSNU, P < 0.032; Breast-fed and Alpha-10, P < 0.03; Breast-fed and Standard, P < 0.035). Between the fourth and sixth months, all formula-fed infants showed an increase in Clostridium spp. Breast-fed infants displayed a higher average number of clostridia compared with all of the formula-fed groups at 4 months, with a significant difference in the total number of cells between the breast-fed group and infants fed Alpha-10 (P < 0.047).
In the present double-blinded, randomised, placebo-controlled intervention, we compared the effects of feeding 2 experimental formulas enriched with α-lac, but with 2 different levels of GMP, with feeding a standard formula or breast milk on the gut microflora of healthy infants.
No major variation in the Bifidobacterium and Lactobacillus populations of either group of formula-fed infants compared with the breast-fed group was observed. The microflora of the breast-fed infants was stable throughout the study and clearly dominated by bifidobacteria. Furthermore, at 6 months of age, E. coli counts in the breast-fed infants seemed to be stabilised. The breast-fed group also maintained a lower number of total bacteria per gram of faeces compared with the formula-fed groups, which showed increasing average counts of bacteria during the study. In the formula-fed infants, bifidobacteria was also the predominant group and did not vary greatly during the study, except for infants fed Standard formula. However, in Standard formula–fed infants, Bacteroides spp. and Clostridium spp. also varied more. Because all infants had been breast-fed before being recruited into the study, it is possible that a threshold level of bifidobacteria was already established, which prevented further increase in the population. Thus, a prebiotic effect of α-lac and GMP, as observed previously in in vitro cultures (2), may only be seen with low starting populations of beneficial microbiota, for example, in formula-fed infants. This was noted previously by Roberfroid et al. (19), who suggested that a correlation may exist between initial Bifidobacterium spp. population levels and the magnitude of the increase in bifidobacterial numbers in individuals fed prebiotics. Consequently, a dose of α-lac and/or GMP could result in the production of similar high numbers of bifidobacteria irrespective of the initial starting population levels. However, only if the starting population of bifidobacteria is low will a significant increase in cell counts result.
A similar explanation could be plausible for lactobacilli. Infants fed Alpha-10 and PSNU formulae displayed higher Lactobacillus counts at baseline as did breast-fed infants. Thus, no significant changes could be observed because counts were already higher than those of breast-fed infants. It is also interesting to note that PSNU-fed and breast-fed infants had comparable Lactobacillus counts at 6 months of age, even though Alpha-10–fed infants had the highest count of all groups.
Another study determined that in formula-fed infants, Bacteroides were found in significantly higher numbers compared with breast-fed infants (20). However, the present study did not show this; all groups displayed variable average Bacteroides counts with both the Standard formula and Breast milk infants having the highest populations at 4 and 6 months.
In summary, this study has shown that a formula enriched in α-lac may promote the growth of certain bacterial groups in a manner similar to breast milk. Bifidobacterium levels did not show any substantial difference among groups, but this may reflect the fact that all infants were initially breast-fed. A formula enriched in α-lac and CGMP may promote an intestinal microflora composition closer to that of breast-fed infants. However, due to a high variation of carriage between individuals, it is difficult to draw a firm conclusion on the degree of bifidogenesis induced by either α-lac or GMP because large population fluctuations between single individuals within groups existed. Due to the fact that in previous batch culture and continuous culture models, levels of bifidobacteria were substantial, it may be speculated that enrichment of formula with the 2 milk factors may confer an increase of the indigenous beneficial microflora when levels are below a certain point (ie, below 10% of the faecal flora) (2). Furthermore, it becomes evident that well-controlled human volunteer trials are essential to complete the assessment of new food ingredients because it is virtually impossible to model all aspects of the human colon.
The study was financially supported by Arla Foods, Viby J, Denmark, and the Swedish Research Council, Medicine. Gitte Graverholt is employed by the main sponsor (Arla Foods). She was involved in discussions regarding study design and contributed to writing the article but did not take part in the analysis or interpretation of data, nor in the decision to write the report and to submit for publication. Wolfram Brück and Michele Redgrave had PhD scholarships paid by the main sponsor. W. Brück did most of the analyses and was the main author of the article. M. Redgrave and K. Tuohy conducted part of the analyses and contributed to data analyses and to writing the article. None of the other researchers had any financial or personal interest in the organizations and companies sponsoring the study. Thus, none of the other authors has any conflict of interest to declare. B. Lönnerdal, O. Hernell and G. Gibson took active part in designing the study, interpreting the data and writing the article.
1. Borriello SP. Microbial flora of the gastrointestinal tract. In: Hill MJ, ed. Microbial Metabolism in the Digestive Tract
. Boca Raton, FL: CRC Press Inc, 1986:2–16.
2. Brück WM, Graverholt G, Gibson GR. Use of batch culture and a two-stage continuous culture system to study the effect of supplemental alpha-lactalbumin
and glycomacropeptide on mixed populations of human gut bacteria. FEMS Microbiol Ecol 2002; 41:231–237.
3. Brück WM, Graverholt G, Gibson GR. A two-stage continuous culture system to study the effect of supplemental alpha-lactalbumin
and glycomacropeptide on mixed cultures of human gut bacteria challenged with enteropathogenic Escherichia coli
and Salmonella typhimurium
. J Appl Microbiol 2003; 95:33–53.
4. Steer T, Carpenter H, Tuohy K, et al
. Perspectives on the role of the human gut microbiota and its modulation by pro- and prebiotics. Nutr Res Rev 2000; 13:229–254.
5. Gibson GR, Roberfroid MB. Dietary modulation of the human colonic microbiota: introducing the concept of prebiotics. J Nutr 1995; 125:1401–1412.
6. Brück WM, Kelleher SL, Gibson GR, et al
. rRNA probes used to quantify the effects of glycomacropeptide and alpha-lactalbumin
supplementation of the predominant groups of intestinal bacteria of infant Rhesus monkeys challenged with enterophathogenic Escherichia coli
. J Pediatr Gastroenterol Nutr 2003; 37:273–280.
7. Brinkley AW, Mott GE. Anaerobic fecal bacteria of the baboon. Appl Environ Microbiol 1978; 36:530–532.
8. Lindberg T, Engberg S, Jakobsson I, et al
. Digestion of proteins in human milk, human milk fortifier, and preterm formula in infant rhesus monkeys. J Pediatr Gastroenterol Nutr 1997; 24:537–543.
9. Kunz C, Lönnerdal B. Protein composition of rhesus monkey milk: comparison to human milk. Comp Biochem Physiol 1993; 104A:793–797.
10. Tuohy KM, Kolida S, Lustenberger AM, et al
. The prebiotic effects of biscuits containing partially hydrolysed guar gum and fructo-oligosaccharides—a human volunteer study. Br J Nutr 2001; 86:341–348.
11. Neeser JR, Grafström RC, Woltz A, et al
. A 23 kDa membrane glycomprotein bearing NeuNAcα2-3Galβ1-3GalNAc O-linked carbohydrate chains acts as a receptor for Streptococcus sanguis
OMZ 9 on human buccal epithelial cells. Glycobiology 1995; 5:97–104.
12. Strömqvist M, Falk P, Bergström S, et al
. Human milk κ-casein and inhibition of Helicobacter pylori
adhesion to human gastric mucosa. J Pediatr Gastroenterol Nutr 1995; 21:288–296.
13. Bouhallab S, Favrot C, Maubois JL. Growth-promoting activity of tryptic digest of casein glycomacropeptide
for Lactococcus lactis
. Lait 1993; 73:73–77.
14. Idota T, Kawakami H, Nakajima I. Growth-promoting effects of N
-acetylneuraminic acid-containing substances on bifidobacteria. Biosci Biotechnol Biochem 1994; 58:1720–1722.
15. Xue Y, Lui JN, Sun Z, et al
. α-Lactalbumin mutant acting as lysozyme. Proteins 2001; 42:17–22.
16. Pihlanto-Leppälä A, Marnila P, Hubert L, et al
. The effect of α-lactalbumin and β-lactoglobulin hydrolysates on the metabolic activity of Escherichia coli
JM103. J Appl Microbiol 1999; 87:545.
17. Petschow BW, Talbott RD. Response of Bifidobacterium
species to growth promoters in human and cow milk. Pediatr Res 1991; 29:208–213.
18. DHSS. Artificial feed for the young infant. In: Report on Health and Social Subjects
. London, UK: HMSO, 1980:18.
19. Roberfroid MB, Van Loo J, Gibson GR. The bifidogenic nature of chicory inulin and its hydrolysis products. J Nutr 1998; 128:11–18.
20. Farano S, Chierici R, Guerrini P, et al
. Intestinal microflora in early infancy: composition and development. Acta Paediatr Suppl 2003; 91:48–55.