Galacto- and Fructo-oligosaccharides Utilized for Growth by Cocultures of Bifidobacterial Species Characteristic of the Infant Gut

The kinds of bacteria that form the collection of microbes (the microbiota) in the gut of human infants may influence health and well-being. Knowledge of how the composition of the infant diet influences the assemblage of the bacterial collection is therefore important because dietary interventions may offer opportunities to alter the microbiota with the aim of improving health. Bifidobacterium longum subspecies infantis is a well-known bacterial species, but under modern child-rearing conditions it may be disadvantaged in the gut. Modern formula milks often contain particular oligosaccharide additives that are generally considered to support bifidobacterial growth. However, studies of the ability of various bifidobacterial species to grow together in the presence of these oligosaccharides have not been conducted. These kinds of studies are essential for developing concepts of microbial ecology related to the influence of human nutrition on the development of the gut microbiota.

ABSTRACT

Bifidobacterial species are common inhabitants of the gut of human infants during the period when milk is a major component of the diet. Bifidobacterium breve, Bifidobacterium bifidum, Bifidobacterium longum subspecies longum, and B. longum subspecies infantis have been detected frequently in infant feces, but B. longum subsp. infantis may be disadvantaged numerically in the gut of infants in westernized countries. This may be due to the different durations of breast milk feeding in different countries. Supplementation of the infant diet or replacement of breast milk using formula feeds is common in Western countries. Formula milks often contain galacto- and/or fructo-oligosaccharides (GOS and FOS, respectively) as additives to augment the concentration of oligosaccharides in ruminant milks, but the ability of B. longum subsp. infantis to utilize these potential growth substrates when they are in competition with other bifidobacterial species is unknown. We compared the growth and oligosaccharide utilization of GOS and FOS by bifidobacterial species in pure culture and coculture. Short-chain GOS and FOS (degrees of polymerization [DP] 2 and 3) were favored growth substrates for strains of B. bifidum and B. longum subsp. longum, whereas both B. breve and B. longum subsp. infantis had the ability to utilize both short- and longer-chain GOS and FOS (DP 2 to 6). B. breve was nevertheless numerically dominant over B. longum subsp. infantis in cocultures. This was probably related to the slower use of GOS of DP 3 by B. longum subsp. infantis, indicating that the kinetics of substrate utilization is an important ecological factor in the assemblage of gut communities.

IMPORTANCE The kinds of bacteria that form the collection of microbes (the microbiota) in the gut of human infants may influence health and well-being. Knowledge of how the composition of the infant diet influences the assemblage of the bacterial collection is therefore important because dietary interventions may offer opportunities to alter the microbiota with the aim of improving health. Bifidobacterium longum subspecies infantis is a well-known bacterial species, but under modern child-rearing conditions it may be disadvantaged in the gut. Modern formula milks often contain particular oligosaccharide additives that are generally considered to support bifidobacterial growth. However, studies of the ability of various bifidobacterial species to grow together in the presence of these oligosaccharides have not been conducted. These kinds of studies are essential for developing concepts of microbial ecology related to the influence of human nutrition on the development of the gut microbiota.

INTRODUCTION

Bacteria belonging to the genus Bifidobacterium are commonly detected as predominant members of the gut microbiota of infants, especially prior to the introduction of complementary (weaning) foods to the diet (1–14). Four bifidobacterial species, Bifidobacterium breve, Bifidobacterium bifidum, Bifidobacterium longum subspecies longum, and B. longum subspecies infantis are characteristic of the infant gut microbiota in that they have been detected consistently in feces (2, 8, 15). However, from the results of more recent studies, there appears to be a paucity of B. longum subsp. infantis in the microbiota of infants inhabiting westernized countries (16–18). In theory, this may be due to the short duration of breast milk feeding in Western countries relative to that of other countries where B. longum subsp. infantis dominates the microbiota (17). B. longum subsp. infantis is well-equipped biochemically to internalize, degrade, and ferment human milk oligosaccharides (HMOs) (19). In the absence of HMOs, B. longum subsp. infantis lacks an important growth substrate and thus may be disadvantaged with respect to competitive growth with other bifidobacterial species that do not have the same reliance on HMOs for growth (20). In support of this view, in Indonesia, where breast milk feeding tends to be of long duration, the predominance of B. longum subsp. infantis is driven by the niche model of microbiota assemblage (trophic adaptation results in preferential colonization). In contrast, in New Zealand, where breast milk feeding is usually of shorter duration, B. longum subsp. infantis does not dominate the microbiota, and microbiota assemblage follows the neutral model (species randomly assemble and are functionally equivalent, so competition will be a feature of community assembly) (17). B. longum subsp. infantis can be detected as a minor component of the gut microbiota in Western infants, so it is not an extinct bacterial lineage but appears to be less capable of dominating the ecosystem under these circumstances (17, 18, 21).

Infant formula is a major nutritive source for infants when breast feeding is not possible, as well as during the complementary (weaning) feeding period (22). Modern formulas often contain prebiotics that increase the oligosaccharide content of the products, which are usually based on ruminant milk. Galacto- and fructo-oligosaccharides (GOS and FOS, respectively) are common prebiotics used for this purpose (23–37). Commercially available GOS contain oligosaccharides composed of galactose residues that are variously linked β-d-1,2, β-d-1,3, β-d-1,4, and β-d-1,6 and range in degrees of polymerization (DP) from 2 to 5 (designated herein according to the DP as G2 to G5) (38). Each oligosaccharide is terminated with a glucose residue. FOS contain two series of β-d-2,1-linked fructo-oligosaccharides, one of which contains a terminal glucose (GF series, with DPs of 2 to 6 and designated as GF2 to GF6) and one which does not (F series, with DPs of 2 to 6 and designated as F2 to F6). As such, GOS and FOS do not chemically resemble HMOs, which are complex, diverse molecules derived of glucose, galactose, N-acetylglucosamine, fucose, or sialic acid (39). GOS and FOS are nevertheless utilized for growth by at least some species of bifidobacteria (23–37). However, the relative abilities of bifidobacterial species to utilize GOS and FOS in coculture have not been tested. We wondered whether B. longum subsp. infantis could compete with other bifidobacteria when GOS and FOS were provided as growth substrates for cocultures. If they could not, formula feeding might be inimical to the growth of B. longum subsp. infantis in the gut. The purpose of our work, therefore, was to compare the in vitro growth profiles of Bifidobacterium bifidum, Bifidobacterium breve, B. longum subspecies infantis, and Bifidobacterium longum subspecies longum in batch and continuous cocultures in relation to the usage of GOS and FOS.

RESULTS

Growth of bifidobacterial species in culture medium containing GOS or FOS.

The variability of bifidobacterial strains to use GOS or FOS as growth substrates was tested using pure bacterial cultures. In general, after 24 h of incubation, the B. longum subspecies cultures showed similar, interstrain growth patterns in GOS and FOS cultures. A variety of growth patterns in FOS medium were evident among B. bifidum strains, whereas there was generally more growth in GOS cultures than in FOS cultures in the case of B. breve strains (Fig. 1). Four strains that utilized both GOS and FOS for growth were chosen for further experiments in coculture: B. bifidum G22, B. breve G9, B. longum subsp. infantis ATCC 15702, and B. longum subsp. longum G15.

FIG 1.

Growth of pure cultures of bifidobacterial species in basal medium or in basal medium containing GOS (0.2%, wt/vol) or FOS (0.2% wt/vol) incubated anaerobically for 24 h at 37°C, as indicated. Means and standard errors of optical density measurements of triplicate cultures are shown.

Growth of bifidobacterial species and utilization of oligosaccharides in batch culture.

The growth of the selected strains in pure culture in GOS/FOS medium was followed during an 8-h incubation in order to determine the oligosaccharide preferences of the bacteria. Under these conditions, B. breve and B. longum subsp. infantis attained the highest numbers of CFU/milliliter and optical densities at 8 h. B. bifidum tended to lag behind the other strains during the growth period (Fig. 2). The slower growth of the B. bifidum strain was associated with low usage of GOS; only GOS G2 was used to any extent (Fig. 3A). The other strains used GOS prioritized in order of DP, but the B. longum subsp. longum strain was the only one to utilize GOS G4 to an appreciable extent (Fig. 3D). The activity of this strain in relation to usage of GOS G4 was evident in cocultures (Fig. 3E and F). None of the individual strains or combinations of strains utilized FOS in 8-h batch culture, showing that GOS were the preferred growth substrates under these culture conditions. Since all four strains had been shown to grow in FOS medium with incubation for 24 h (Fig. 1), we determined the extent of GOS and FOS utilization in these cultures. B. bifidum mainly used GOS G2 and G3 for growth, with very little use of the FOS series (Table 1). B. longum subsp. longum used GOS G2, G3, and G4 as well as the shorter-chain-length FOS GF and F series. B. longum subsp. infantis utilized all of the GOS and FOS. Similarly, B. breve consumed all of the GOS and FOS although GOS G4 use was limited compared to that of the B. longum subsp. longum (Table 1).

FIG 2.

Growth of pure cultures of bifidobacterial strains. (A) Population levels of pure cultures of bifidobacterial strains representing four species in basal medium containing GOS and FOS (both at 0.2%, wt/vol) incubated anaerobically for 8 h at 37°C. Values are means and standard errors of triplicate cultures. (B) Optical densities of pure cultures as described for panel A. Means and standard errors of A₆₀₀ measurements of triplicate cultures are shown.

FIG 3.

(A to D) Utilization of GOS by pure cultures of bifidobacterial strains representing four species in basal medium containing GOS and FOS (both at 0.2%, wt/vol) incubated anaerobically for 8 h at 37°C. (E and F) Utilization of GOS by cocultures. Values are means and standard errors of triplicate cultures.

TABLE 1.

Utilization of GOS and FOS by bifidobacterial species in pure cultures incubated for 24 h

Oligosaccharide	Oligosaccharide utilization (%) by speciesa
	B. bifidum G22	B. breve G9	B. longum subsp. infantis ATCC 15702	B. longum subsp. longum G15
G2	100.0	100.0	100.0	100.0
G3	96.0	100.0	100.0	98.0
G4	5.7	65.6	100.0	100.0
GF2	15.1	100.0	100.0	90.1
GF3	2.2	99.5	100.0	10.0
GF4	2.7	99.4	100.0	15.2
GF5	3.3	100.0	100.0	16.8
GF6	3.3	100.0	100.0	19.2
F2	4.2	100.0	100.0	90.3
F3	1.1	100.0	100.0	98.6
F4	0.6	100.0	100.0	55.4
F5	1.5	100.0	100.0	16.6
F6	0.8	99.7	100.0	15.5

^aValues are means of duplicate assays.

Growth of bifidobacterial species and utilization of oligosaccharides in continuous coculture.

The growth and utilization of GOS and FOS under steady-state conditions (continuous culture) were investigated because the use of FOS did not occur during the relatively short log-phase growth period in batch culture. In continuous culture, all bacterial cells are maintained in the same physiological state, under substrate limiting conditions, during the entire experiment (40). Continuous cultures were inoculated with mixtures of either B. breve G9, B. longum subsp. longum G15, and B. bifidum G22 or of B. breve, B. longum subsp. longum, and B. longum subsp. infantis ATCC 15702. It was not possible to test the four strains in coculture because differential plate counts could not be used satisfactorily. Nevertheless, the combination of most interest was B. breve and the B. longum subspecies because these strains used GOS and FOS to a major extent for growth (Table 1) and would thus be potentially in competition. In general, apart from discrepancies in GOS G4 and FOS GF2 values in cocultures of B. breve, B. longum subsp. longum, and B. bifidum, there was good agreement between continuous culture runs with regard to usage of GOS and FOS by the cocultures (Fig. 4). FOS utilization was greater in cocultures containing B. longum subsp. infantis. Cocultures of the two combinations of strains produced similar population sizes (Fig. 5), including the numerical dominance of B. breve over the B. longum subspecies. That B. longum subsp. longum had less ability than B. breve to use longer-chain FOS was the most likely explanation for the differences in population sizes. However, B. longum subsp. infantis had an ability equal to that of B. breve to use all GOS and FOS for growth, yet B. breve was numerically dominant. This is probably explained by the more rapid utilization of GOS G3 by B. breve than by B. longum subsp. infantis (Fig. 3B and C) and the associated more rapid doubling time of the respective strains in GOS/FOS medium. B. breve G9 doubling time was 85 min, whereas that of B. longum subsp. infantis ATCC 15702 was almost twice as long at 156 min. Thus, despite the similar abilities of these strains to use oligosaccharides, temporal kinetics of utilization and doubling times under the same growth conditions were different and influenced population outcomes.

FIG 4.

Utilization of GOS and FOS by continuous cocultures of combinations of bifidobacterial strains in basal medium containing GOS and FOS (both at 0.2%, wt/vol) incubated anaerobically at 37°C. Means and standard errors of triplicate samples collected from two chemostat runs (1 and 2) per combination of strains are shown.

FIG 5.

Population levels of bifidobacterial species in continuous coculture in GOS/FOS medium (both substrates at 0.2%, wt/vol). (A) Chemostat inoculated with a combination of strains G9, G15, and G22. (B) Chemostat inoculated with a combination of strains G9, G15, and ATCC 15702. Values are means and standard errors of duplicate chemostat runs.

DISCUSSION

The bacterial community resident in the large bowel of humans is daily presented with a mélange of potential growth substrates of various structural and chemical compositions. Even in the case of infants, where the diet before weaning is exclusively milk, a variety of carbohydrates are present as potential growth substrates for bacterial species. These substrates include lactose (some escapes digestion in the small bowel), carbohydrates linked to milk proteins (glycoproteins), and HMOs that pass to the large bowel undigested by the infant (39, 41). Lactose is a growth substrate for all bifidobacterial species characteristic of the gut (42); some B. longum subsp. longum, B. longum subsp. infantis, and B. breve strains have endoglycosidases that may be important in harvesting fermentable glycans from glycoproteins (43), while HMOs are most efficiently consumed by B. longum subsp. infantis (19, 39, 44). Other bifidobacterial species detected in the gut of infants are generally less efficient in the use of HMOs for growth although B. bifidum has an important hydrolytic capacity to provide sialic acid and fucose from HMOs and mucins as growth substrates for other bacteria (39, 40, 45–47). The addition of GOS/FOS to infant formulas provides yet further fermentable substrates to the bifidobacterial environment, and in vitro and in vivo testing has shown that at least some strains of the common species can use them to support growth (23–37). Oligosaccharides of different chain lengths are contained within the general categories GOS and FOS, yet little information about the effect of the degree of polymerization on the usage of these potential growth substrates has been reported (48). Knowledge of the prioritization of individual carbohydrates by gut bacteria is increasingly appreciated as fundamental to understanding how the microbiota degrades and ferments the complex mixtures of indigestible components of human food (49, 50). Use of carbohydrates in a preferential order by bacteria can consequently affect the emergent properties of the microbiota (51). Considering the widespread use of GOS and FOS in formulas and other foods, the lack of information about the preferential utilization of these substrates by bifidobacteria is surprising, especially in the potentially competitive setting of the infant gut where at least three bifidobacterial species commonly coexist (2, 5, 8, 15, 17).

The results of our study show that B. breve has the greatest trophic potential with respect to GOS/FOS to outcompete B. longum subsp. infantis. B. breve is a common member of the infant gut microbiota. However, the in vivo effect of GOS/FOS dietary supplementation on the proportions of the different species within bifidobacterial populations has not been a feature of the trials in which the effect of GOS/FOS-supplemented formulas has been compared to that of nonsupplemented formulas (31–37). This may be due to past difficulties of differentiating between bifidobacterial species using DNA-based methodologies (20, 21). Nevertheless, the species and subspecies can be easily differentiated using biochemical tests of cultured isolates (20). Clearly, there is a need to measure the abundances of bifidobacterial species in the feces of infants in association with dietary interventions so that new, detailed, ecological perspectives can be obtained. B. breve and GOS/FOS are used as a synbiotic combination, and this is certainly appropriate according to the results of our study (52–54). Synbiotics containing B. longum subsp. infantis might best combine the bacteria with synthetic HMOs that are now available (55).

Differences in abundances of bifidobacterial species in the gut of infants of different countries are unlikely to be fully explained by variation in breast milk composition (56), general feeding practices, and GOS/FOS supplementation of formulas. Shared environments and similar diets under conditions of communal living are important in the assembly and maintenance of the microbiota of the human body because they result in greater ease of dispersal (horizontal transmission) of gut commensals (57–60). Western methods of sanitation, water treatment, and hygiene might impair the dispersal of B. longum subsp. infantis (17). The genetic impact of the human host on the colonization of the gut by bacterial species might also be important. Heritable taxa, notably members of the family Christensenellaceae, have been recognized in studies of Canadians of European descent and of Koreans (61) but, to date, genome-wide association studies (GWASs) that aimed to associate human genetic variants (single nucleotide polymorphisms [SNPs]) with microbiota compositions have led to the conclusion that, overall, the genotype of the host probably has little impact on the composition of the adult microbiota (61, 62). The situation in infants, however, is unknown. Thus, currently, niche definitions based on trophic requirements of species appear to be the pertinent means of testing mechanisms that enable cohabitation of potentially competitive strains in microbial communities (63–65). Knowledge of substrate preferences and of the kinetics of their utilization by gut bacteria, such as bifidobacteria, is therefore important and continues to require investigation.

MATERIALS AND METHODS

Bifidobacterial strains.

Strain variability with respect to utilization of GOS and FOS was tested using pure cultures of five strains each of B. bifidum (G19, G21, G22, ATCC 11863, and ATCC 15696), B. breve (G7, G9, G38, 24b, and ATCC 15700^T), B. longum subsp. infantis (ATCC 15702, DSM 20088^T, 875/4, 267/1, and DSM 20218), and B. longum subsp. longum (G12, G15, G16, 875/3, and ATCC 15707^T). Strains not originating from ATCC and DSM culture collections were isolated from infant feces and identified by 16S rRNA gene sequencing and biochemical tests, as described previously (66).

Cultures.

GOS and FOS media were composed of a previously described basal medium with the same formulation as lactobacillus deMan-Rogosa-Sharpe (MRS) medium (Difco), but with glucose omitted (67). GOS (β-d-1,4-linked oligosaccharides [68]) (Nissin Sugar Mfg., Japan) and FOS (Orafti P95; BENEO) were added to the medium at 0.2% (wt/vol). Medium was inoculated with bifidobacterial strains prepared in lactobacillus MRS medium, individually, at 1% (vol/vol) (Difco). The amounts of growth of the strains were similar in MRS medium (contains glucose as a fermentable carbohydrate), and each culture that was used as an inoculum contained ∼10⁹ viable cells/ml. Culture manipulations and incubations were carried out in an anaerobic glove box. The optical density (A₆₀₀) of cultures was determined after appropriate incubation times, and carbohydrate analysis was carried out on supernatants obtained after centrifugation of samples for 5 min at 15,000 × g, followed by filtration (0.45-μm pore size). Triplicate cultures were tested in all batch culture experiments.

Continuous cultures were prepared using Freter-type chemostats (30-ml volume), with a flow rate of 3 ml/h (giving a dilution rate [D] of 0.1 per h), and maintained in an anaerobic glove box, as described previously (40). The culture was considered to be in steady-state condition after 5 complete turnovers of reactor volume, and the culture was then sampled. Triplicate samples from duplicate runs were tested.

The doubling times in GOS/FOS medium of B. breve G9 and B. longum subsp. infantis ATCC 15702 were measured as described previously (63) in which single aliquots of triplicate cultures were removed at hourly intervals during the incubation period, and the optical density was determined. The optical density values were converted to natural logarithms and plotted against sampling times. Doubling times were calculated by log₂/slope of the linear part of the graph.

Differential CFU count determination.

As described previously (39), samples of cultures were diluted in basal medium to 1 × 10⁻⁶ in 10-fold steps, and then aliquots of each dilution were spread-plated on basal medium agar containing carbohydrate substrates at 0.5% (wt/vol) to determine the number of CFU per milliliter on lacto-N-neotetraose (Glycom) for B. bifidum or B. longum subsp. infantis (40), on salicin (Sigma) for B. breve (42), and on arabinose for B. longum subsp. longum (42). Colonies were enumerated, and the number of CFU per milliliter was calculated after 48 h of anaerobic incubation at 37°C.

Chemical analysis of culture supernatants.

As described previously (69), concentrations of GOS and FOS in cell-free culture supernatants were determined by high-performance anion-exchange chromatography (HPAEC) using a Dionex ICS 3000 system (Dionex Corp., Sunnyvale, CA). Samples were diluted 400-fold with distilled water, injected (20 μl) onto a CarboPac PA-100 (4 by 250 mm) column equilibrated in 50 mM NaOH, and eluted with simultaneous gradients of NaOH (50 to 100 mM) and sodium acetate (0 to 90 mM) from 2.5 to 10 min at a flow rate of 1 ml/min. The eluant was monitored by pulsed amperometric detection using the Dionex standard carbohydrate waveform. The eluted sugars were identified from their elution times relative to those of standard sugars. Utilization (percent) of each oligosaccharide was calculated by reference to blank (uninoculated medium) values. Duplicate samples of batch cultures and triplicate samples from continuous cultures were analyzed.