Lactic acid bacteria fermentation of human milk oligosaccharide components, human milk oligosaccharides and galactooligosaccharides


  • Editor: Wolfgang Kneifel

Correspondence: Clarissa Schwab, Department of Agricultural, Food and Nutritional Sciences, University of Alberta, 4-10 Ag/For Centre, Edmonton, AB, Canada T6G 2P5. Tel.: +1 780 492 3634; fax: +1 780 492 4265; e-mail:


Human milk contains about 7% lactose and 1% human milk oligosaccharides (HMOs) consisting of lactose with linked fucose, N-acetylglucosamine and sialic acid. In infant formula, galactooligosaccharides (GOSs) are added to replace HMOs. This study investigated the ability of six strains of lactic acid bacteria (LAB), Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus fermentum, Lactobacillus reuteri, Streptococcus thermophilus and Leuconostoc mesenteroides subsp. cremoris, to digest HMO components, defined HMOs, and GOSs. All strains grew on lactose and glucose. N-acetylglucosamine utilization varied between strains and was maximal in L. plantarum; fucose utilization was low or absent in all strains. Both hetero- and homofermentative LAB utilized N-acetylglucosamine via the Embden–Meyerhof pathway. Lactobacillus acidophilus and L. plantarum were the most versatile in hydrolysing pNP analogues and the only strains releasing mono- and disaccharides from defined HMOs. Whole cells of all six LAB hydrolysed oNP-galactoside and pNP-galactoside indicating β-galactosidase activity. High β-galactosidase activity of L. reuteri, L. fermentum, S. thermophilus and L. mesenteroides subsp. cremoris whole cells correlated to lactose and GOS hydrolysis. Hydrolysis of lactose and GOSs by heterologously expressed β-galactosidases confirmed that LAB β-galactosidases are involved in GOS digestion. In summary, the strains of LAB used were not capable of utilizing complex HMOs but metabolized HMO components and GOSs.


Human milk contains about 7% lactose and 1% human milk oligosaccharides (HMOs) of complex composition. Galactose, fucose, N-acetylglucosamine (GlcNAc) and sialic acids are attached to lactose forming a wide variety of oligosaccharides with different linkages and degree of branching (Kunz et al., 2000). HMOs stimulate growth of intestinal bifidobacteria, inhibit the adhesion of infectious bacterial pathogens or bacterial toxins, and potentially have immunomodulatory properties (Kunz et al., 2000).

Bifidobacteria are the dominating population in the faeces of healthy breast milk or formula-fed infants (Euler et al., 2005; Haarman & Knol, 2006). Bifidobacterium longum subsp. infantis, whose genome possesses several clusters predicted to act on HMOs, is especially well adapted to metabolize HMOs (Sela et al., 2008). Other bifidobacteria are able to ferment HMOs and components of HMOs to various extents (Harmsen et al., 2000; Ward et al., 2006, 2007; Sela et al., 2008; Wada et al., 2008). Human milk also contains high amounts of lactose. Accordingly, the induction of lacZ of Bifidobacterium longum during growth in human milk indicated a role of β-galactosidases in lactose and/or HMO digestion, for example the release of terminal nonreducing galactose units (González et al., 2008).

Lactic acid bacteria (LAB) are present in lower numbers than bifidobacteria in faeces of neonates but are nonetheless routinely detected (Kleessen et al., 1995; Harmsen et al., 2000; Euler et al., 2005; Haarman & Knol, 2006; Ziegler et al., 2007). Compared with Bifidobacterium infantis, Lactobacillus gasseri only poorly digested HMOs (Ward et al., 2006). However, LAB are capable of utilizing the lactose in breast milk after uptake via lactose phosphoenolpyruvate-phosphotransferase system and the activity of phospho-β-galactosidases, or after internalization by lactose permeases and hydrolysis by β-galactosidases.

Efforts to produce HMOs on a commercial scale have failed so far. In contrast, galactooligosaccharides (GOS) consisting of galactose and glucose can be obtained from lactose by the use of fungal and bacterial β-galactosidases (Gosling et al., 2010). GOSs are commercially used in infant formula either alone or in combination with other nondigestible glycans and their inclusion increased numbers of bifidobacteria and lactobacilli in a dose-dependent effect (Moro et al., 2002; Ziegler et al., 2007; Nakamura et al., 2009). Individual strains of LAB were reported to digest GOSs. Lactobacillus rhamnosus preferred GOSs with a low degree of polymerization (Gopal et al., 2001; Ignatova et al., 2009); growth of Lactobacillus delbrueckii on GOSs was strain dependent. However, to date few data exist on HMO or GOS metabolism, or fermentation of HMO components N-acetylglucosamine (GlcNAc) and fucose by LAB.

It was the aim of this study to investigate the ability of LAB to ferment defined HMOs, HMOs components and GOSs. Emphasis was placed on the role of β-galactosidases in oligosaccharide digestion. Four of the six LAB strains chosen are affiliated to intestinal microbial communities (Lactobacillus acidophilus, Lactobacillus plantarum, Lactobacillus fermentum and Lactobacillus reuteri); two strains are associated with fermented foods (Leuconostoc mesenteroides subsp. cremoris and Streptococcus thermophilus).

Materials and methods


Lactobacillus plantarum FUA3112, L. mesenteroides FUA3143, L. reuteri FUA3148, L. fermentum FUA3177, L. acidophilus FUA3191, and S. thermophilus FUA3194 were derived from the Food Microbiology culture collection of University of Alberta (FUA). For preparation of whole cell assays, LAB were grown in 10 mL modified MRS (10 g L−1 tryptone, 10 g L−1 beef extract, 5 g L−1 yeast extract, 2 g L−1 tri-ammonium citrate, 3 g L−1 sodium acetate, 0.1 g L−1 magnesium sulphate heptahydrate, 0.038 g L−1 manganese sulphate monohydrate, 2 g L−1 dipotassium phosphate, 1 mL−1 Tween 80, pH 6.2) in the presence of 20 g L−1 lactose as sole carbohydrate source at 37 °C for 16 h, washed once in 50 mM phosphate buffer (PB) pH 6.5 and resuspended in 100 μL PB containing 1 mM MgCl2 and 10% glycerol.

Heterologous expression of β-galactosidases

Lactococcus lactis MG1363 was used for heterologous expression of the glycosyl hydrolase family (GH)2 β-galactosidases LacLM L. plantarum FUA3112 (FN424350, FN424351, Schwab et al., 2010), LacLM L. mesenteroides subsp. cremoris (Israelsen et al., 1995), and LacZ S. thermophilus FUA3194 (FN424354, Schwab et al., 2010) using a p170-derived expression vector which is induced by pH below 6 and transition to stationary growth phase of glucose grown cultures (Israelsen et al., 1995; Madsen et al., 1999). β-Galactosidases were obtained as described previously (Schwab et al., 2010). Briefly, L. lactis harbouring the respective plasmids were plated on M17 agar plates, single colonies were picked from plates, inoculated in 10 mL M17 and subcultured at 1% in 500 mL M17 with 0.5% glucose. Cells were incubated at 30 °C for 24 h and harvested by centrifugation. The cell suspension was washed once in PB pH 6.5, resuspended in PB with 10% glycerol and 1 mM MgCl2 and disrupted using a bead beater. Protein content in the L. lactis crude cell extract (CCE) was adjusted to 0.3 mg protein mL−1.

Preparation of GOS

GOS were prepared using the LacZ-type β-galactosidase of S. thermophilus FUA3194. LacZ was expressed in L. lactis MG1363 as described above. Lactococcus lactis CCE (50 μL) containing LacZ S. thermophilus was incubated in the presence of 0.78 M lactose (950 μL) at 56 °C for 16 h. GOS crude extracts were enriched in di- and oligosaccharides by fractionation using gel permeation chromatography with a Superdex200 column (GE Healthcare, Baie d'Urfe, Canada) using water as eluent at a flow rate of 0.4 mL min−1. Fractions containing di- and higher oligosaccharides were freeze-dried and resuspended in PB, pH 6.5. To verify removal of monosaccharides in the GOS preparation, the enriched GOS preparations were analysed on a Dionex ICS-300 system equipped with a CarbopacPA20 column (Dionex, Oakville, Canada). Water (A) and 200 mM NaOH (B) were used as solvents at a flow rate of 0.25 mL min−1 with the following gradient: 0 min 6% B, 20 min 100% B followed by washing and regeneration.

Fermentation of glucose, lactose, N-acetylglucosamine (GlcNAc) and fucose

LAB were grown in modified MRS supplied with the HMO components lactose, GlcNAc, fucose or glucose (approximately 20 g L−1) as sole carbohydrate sources at 37 °C for 24 h. OD595 nm was monitored in 4-h intervals using a Varioskan microplate reader (Thermo Scientific, Canada). Organic acids, alcohols and sugars concentrations after 24 h of fermentation were determined by HPLC with an Aminex HPX-87 column (300 mm × 7.8 mm, Bio-Rad) at a temperature of 70 °C and a flow rate of 0.4 mL min−1 with 5 mM H2SO4 as the eluent. Refractive index detector was used for detection. For sample preparation, 7.5% perchloric acid was added to the supernatants, which were incubated at 4 °C overnight. Precipitated protein was removed by centrifugation. The concentrations of lactose, glucose, galactose, N-acetylglucosamine, fucose, lactate, acetate and ethanol were determined using external standards. Acetate present in MRS was subtracted from the amount synthesized by the strains. Data were obtained from three independent experiments.

Hydrolysis activity of whole cells of LAB

Whole cell hydrolysis activity was tested at 37 °C using oNP-galactoside (oNPG), pNP-galactoside (pNPG) and pNP analogues pNP-mannoside (pNPM), pNP-glucoside (pNPGl), pNP-fucoside (pNPF), pNP-N-acetylglucosamide (pNPGlcNAc) and pNP-arabinoside (pNPara) as substrates (all obtained from Sigma, Oakville, Canada). Whole cells (5 μL) were mixed with 95 μL 2 mM oNPG or pNP analogues resuspended in PB. Hydrolysis kinetics were recorded in a Varioskan microplate reader at 420 nm. Specific activity (enzyme activity relative to amount of whole cells) was determined as: units hydrolysis activity=(ΔA420 nm) × (min−1 μL−1 whole cells).

Whole cell and β-galactosidases hydrolysis of GOSs and HMOs

HMOs (2′-fucosyl-lactose, 3′-fucosyl-lactose, lacto-N-fucopentaose I, lacto-N-tetraose, 3′-sialyl-lactose, 6′-sialyl-lactose, 3′-sialyl-N-acetyl-lactosamine) were obtained from V-LABS (Covington) and resuspended at 2 mM in PB. GOS preparations and HMO (95 μL) were used as substrates for LAB whole cells (5 μL) and heterologously expressed β-galactosidases LacLM L. plantarum, LacLM L. mesenteroides subsp. cremoris and LacZ S. thermophilus (5 μL). GOS, lactose and HMO degradation after incubation at 37 °C for 1 h was monitored by HPAEC-PAD (Dionex ICS-300 system, CarbopacPA20 column). Water (A), 200 mM NaOH (B) and 1 M Na-acetate (C) were used as solvents at a flow rate of 0.25 mL min−1 with the following gradient: 0 min 15% B, 0.5% C, 20 min 15% B, 0.5% C, 30 min 50% B, 0.5% C, 40 min 50% B, 0.5% C, 45 min 50% B, 20% C, 47 min 50% B, 20% C, followed by washing and regeneration. GOS and lactose degradation was indicated by the release of glucose and galactose; HMO degradation was indicated by the release of mono- or disaccharides; N-acetylglucosamine, galactose, glucose and lactose were used as external standards.


GOS preparation and enrichment

Enriched GOS preparations synthesized using LacZ of S. thermophilus contained 2 mM lactose, respectively, and did not contain glucose or galactose (Fig. 1). GOSs of different linkage type are separated at different retention times by the HPAEC-PAD system used (Splechtna et al., 2006). Accordingly, the GOS preparation contained at least eight structurally different GOSs.

Figure 1.

 Enriched GOS preparation synthesized by LacZ Streptococcus thermophilus FUA3194 (FN424354). GOSs were enriched in di- and oligosaccharides using gel permeation chromatography and freeze drying. GOSs were analysed by HPAEC-PAD. lac, lactose.

Fermentation of lactose, glucose, N-acetylglucosamine and fucose

LAB strains reached highest OD during growth on lactose and glucose (Fig. 2). All strains except L. plantarum and L. acidophilus grew on N-acetylglucosamine with an extended lag phase; essentially no growth was observed with fucose as substrate.

Figure 2.

 OD of (a) Lactobacillus acidophilus, (b) Lactobacillus plantarum, (c) Lactobacillus reuteri, (d) Lactobacillus fermentum, (e) Streptococcus thermophilus and (f) Leuconostoc mesenteroides subsp. cremoris during growth in MRS containing (•) lactose, (○) glucose, (▾) N-acetylglucosamine or (Δ) fucose (all approximately 20 g L−1) as sole carbohydrate source. Data represent average values from three independent experiments.

Lactose and glucose were completely or partially utilized by all strains (Table 1). Accumulation of galactose from lactose was detected in culture supernatants from L. acidophilus (approximately 3.5 mM) and L. mesenteroides subsp. cremoris (approximately 13 mM). Between 35% (L. reuteri) and 85% (L. plantarum) of N-acetylglucosamine were metabolized during growth (Table 1); only L. plantarum and L. acidophilus utilized more than 5% of the available fucose (9 and 4 mM, respectively). The homofermentative or facultative heterofermentative species L. acidophilus, L. plantarum and S. thermophilus produced lactate as major metabolite from lactose or glucose, the obligate heterofermentative species L. fermentum, L. reuteri and L. mesenteroides subsp. cremoris produced lactate and the alternative end products acetate or ethanol. All strains formed lactate and acetate in a ratio of 2 : 1 when grown with N-acetylglucosamine as sole carbon source.

Table 1.   Utilization of lactose (60 mM), glucose (76 mM), N-acetylglucosamine (49.8 mM) and fucose (66.2 mM), consumption of substrate (S) and formation of lactate (L), acetate (A) and ethanol (E) during growth in MRS at 37°C for 24 h in (mM)
SubstrateLactose (mM)Glucose (mM)N-acetylglucosamine (mM)Fucose (mM)
L. acidophilus−2151 ± 13 ± 010 ± 10−2956 ± 16 ± 14 ± 4−2956 ± 725 ± 66 ± 5−46 ± 02 ± 03 ± 3
L. plantarum−53118 ± 45 ± 01 ± 2−63109 ± 34 ± 11 ± 2−4266 ± 130 ± 03 ± 2−911 ± 05 ± 14 ± 1
L. reuteri−6071 ± 01 ± 189 ± 5−7663 ± 91 ± 172 ± 10−1838 ± 819 ± 312 ± 2−23 ± 12 ± 19 ± 1
L. fermentum−6076 ± 39 ± 171 ± 3−3752 ± 613 ± 131 ± 4−2756 ± 229 ± 23 ± 0−34 ± 13 ± 10
S. thermophilus−4182 ± 87 ± 11 ± 1−4477 ± 47 ± 24 ± 2−1836 ± 314 ± 30−36 ± 02 ± 00
L. mesenteroides subsp. cremoris−3452 ± 113 ± 021 ± 10−6267 ± 010 ± 052 ± 3−2141 ± 216 ± 15 ± 0−25 ± 13 ± 17 ± 2

Activity of HMO- and GOS-converting hydrolytic enzymes in LAB whole cells

Hydrolytic activity of whole cells of LAB was tested using oNPG, pNPG and pNP analogues as substrates (Table 2). Activity was calculated relative to oNPG. All six LAB hydrolysed oNPG and pNPG. Cells of L. fermentum, L. mesenteroides and S. thermophilus were between 5 and 13 times more active on oNPG than L. acidophilus, L. plantarum and L. reuteri. Relative to the activity on oNPG, L. acidophilus and L. reuteri showed highest hydrolysing capacity on pNPM; L. plantarum and L. acidophilus most effectively hydrolysed pNPF and pNPara. Lactobacillus acidophilus, L. plantarum and L. reuteri exhibited the highest relative activity with pNPGlcNAc as substrate.

Table 2.   Activity of LAB whole cells grown in MRS with 20 g L−1 lactose for 16 h at 37°C on β-nitrophenyl analogues oNPG, pNPG, pNPM, pNPGl, pNPF, pNPGlcNAc and pNPara (all 2 mM) during incubation at 37°C
SubstrateL. acidophilusL. plantarumL. reuteriL. fermentumL. mesenteroides subsp. cremorisS. thermophilus
  1. Absorbance was measured at 420 nm. Shown are mean values of specific activity [units hydrolysis activity=(ΔA420 nm) × (min−1 μL−1 whole cells)] obtained from three different experiments; in parentheses, % activity relative to oNPG. LAB whole cells grown on MRS were also active on oNPG and pNPG; however, hydrolytic activity was lower compared with growth in MRS with lactose (data not shown).

pNPG9.0 (30.4)16.7 (27)31.1 (65.2)296.6 (96.7)70.4 (44.3)313.0 (74.5)
pNPM5.2 (17.6)5.6 (9.1)7.2 (15.1)5.9 (1.9)12.2 (7.7)2.3 (0.5)
pNPGl6.6 (22.3)2.8 (4.5)2.8 (5.9)7.7 (2.5)4.5 (2.8)4.1 (1.0)
pNPF7.4 (25)16.9 (27.3)2.4 (5.0)21.8 (7.1)11.5 (7.2)11.1 (2.6)
pNPGlcNAc3.4 (11.5)20.2 (32.7)5.2 (10.9)6.7 (2.2)9.1 (5.7)6.2 (1.5)
pNPara6.2 (20.9)5.0 (8.1)1.7 (3.6)9.7(3.2)1.9 (1.2)4.9 (1.2)

GOS and HMO hydrolysis by LAB whole cells

Whole cells of L. reuteri, L. fermentum, L. mesenteroides and S. thermophilus completely hydrolysed lactose and GOS during incubation at 37 °C for 1 h (Table 3). In contrast, only L. acidophilus and L. plantarum whole cells released sugar components from 2′-fucosyl-lactose, 3′-fucosyl-lactose, lacto-N-tetraose and 6′-sialyl-lactose during incubation at 37 °C for 1 h, but showed little or no activity on lactose, respectively, and no activity on GOS. Using external standards, the compounds released from 2′-fucosyl-lactose, 3′-fucosyl-lactose, lacto-N-tetraose and 6′-sialyl-lactose were tentatively identified as a monosaccharide and as lactose.

Table 3.   Hydrolysis of GOS and lactose in GOS preparations, and of defined HMO by LAB whole cells during incubation at 37°C for 1 h
 GOS*HMO (2 mM)
fucopentaose I
  • Digestion of GOSs and HMOs was determined by the release of mono- or disaccharides using HPAEC-PAD.

  • *

    Hydrolysis of glucose and galactose from GOSs or from lactose (in parentheses) present in the GOS preparation.

  • Hydrolysis of mono- or disaccharides.

  • β-Galactosidases were heterologously expressed in Lactococcus lactis using a p170-derived expression system (Israelsen et al., 1995; Madsen et al., 1999; Schwab et al., 2010).

  • +, hydrolysis; −, no activity detectable; +/−, little activity detectable.

Whole cells
 L. acidophilus−(+/−)++++/−+
 L. plantarum−(−)++++/−+
 L. reuteri+(+)
 L. fermentum+(+)
 L. mesenteroides subsp. cremoris+(+)+/−
 S. thermophilus+(+)
 LacLM L. mesenteroides subsp. cremoris+(+)
 LacLM L. plantarum+(+)
 LacZ S. thermophilus+(+)

GOS and HMO hydrolysis by LAB β-galactosidases

β-Galactosidases LacLM L. mesenteroides, LacLM L. plantarum and LacZ S. thermophilus hydrolysed the GOSs produced by LacZ S. thermophilus (Table 3). None of the heterologously expressed β-galactosidases was active on defined HMOs during incubation at 37 °C for 1 h.


Bifidobacteria are prevalent in the faeces of breast-fed infants. Species that are frequently isolated are Bifidobacterium breve, B. infantis, B. longum, Bifidobacterium bifidum, Bifidobacterium catenulatum and Bifidobacterium dentium (Sakata et al., 2005; Shadid et al., 2007). However, only B. infantis, which possesses a specialized HMO utilization cluster composed of β-galactosidase, fucosidase, sialidase and β-hexosaminidase is capable of releasing and utilizing monosaccharides from complex HMOs (Ward et al., 2006, 2007; Sela et al., 2008). In contrast, B. bifidum releases monosaccharides from HMOs but is not able to use fucose, sialic acid and N-acetylglucosamine; B. breve was able to ferment but not release monosaccharides (Ward et al., 2007). Lactobacillus species frequently isolated from neonate faeces are L. fermentum, Lactobacillus casei, Lactobacillus paracasei, L. delbrueckii, L. gasseri, L. rhamnosus and L. plantarum (Ahrnéet al., 2005; Haarman & Knol, 2006). In vitro digestion of HMOs by LAB has previously been examined for L. gasseri, L. acidophilus, S. thermophilus and L. lactis and digestion of HMOs was low in comparison with B. infantis (Ward et al., 2006; Sela et al., 2008; Marcobal et al., 2010). Accordingly, in this study, defined HMOs acted as poor substrate for the LAB tested. Only L. acidophilus and L. plantarum whole cells, which showed the widest hydrolysing activity on oNPG and pNP analogues, were capable of releasing mono- and disaccharides from defined HMOs. Hydrolysis activity was limited to tri- or tetrasaccharides; lacto-N-fucopentaose I was not metabolized, probably because higher oligosaccharides are not transported to the cytoplasm. Dedicated transport systems for oligosaccharides are generally absent in lactobacilli. To date, only two transport systems specific for fructooligosaccharides and maltodextrins have been identified in L. plantarum and L. acidophilus (Barrangou et al., 2003; Saulnier et al., 2007; Nakai et al., 2009).

HMO hydrolysis by LAB was absent or low but extracellular hydrolysis of HMOs by other microorganisms in the intestine may liberate monosaccharides for subsequent use by LAB. It was thus investigated whether LAB could use HMO components as substrate. All LAB strains tested grew to highest OD in the presence of lactose and glucose. N-acetylglucosamine was fermented to various extents and all LAB strains formed lactate and acetate is a molar ratio of 2 : 1 from N-acetylglucosamine, in agreement with previous reports for Lactovum miscens (Matthies et al., 2004). This indicates that the glucosamine moiety was metabolized to 2 mol lactate after liberation and release of the acetyl moiety. Interestingly, both hetero- and homofermentative LAB metabolized the glucose moiety of N-acetylglucosamine via the Embden–Meyerhof pathway, whereas glucose was metabolized via the phosphoketolase pathway by all obligate heterofermentative LAB (L. reuteri, L. fermentum and L. mesenteroides subsp. cremoris). Lactobacillus reuteri harbours the enzymatic set-up to operate both metabolic pathways (Årsköld et al., 2008). Fucose was generally not metabolized and limited conversion was only observed in L. plantarum and L. acidophilus. Fucose internalization and utilization systems have been previously identified in the anaerobic human gut bacterium Roseburia inulinivorans, and in Escherichia coli (Hacking & Lin, 1977; Scott et al., 2006), but not in LAB.

Lactobacillus reuteri, L. fermentum, L. mesenteroides subsp. cremoris and S. thermophilus hydrolysed GOS but not the more complex HMOs. GOS hydrolysis generally correlated with high activity on oNPG and pNPG, which indicated the expression of β-galactosidases. Lactobacillus reuteri expresses its LacLM β-galactosidases during growth in the presence of lactose (Nguyen et al., 2006). The role of β-galactosidases in GOS degradation was further confirmed by the release of glucose and galactose by heterologously expressed GH2 β-galactosidases of the LacLM and LacZ type. In contrast to bifidobacteria, which express both intracellular and extracellular β-galactosidases (Møller et al., 2001; Goulas et al., 2007), β-galactosidases of strains of the genera Lactobacillus, Streptocococcus and Leuconostoc are located in the cytoplasm (Fortina et al., 2003; Nguyen et al., 2006, 2007). Transport enzymes for GOSs have not been identified and the lack of transport systems for GOSs explains the preference of LAB for GOSs with a low degree of polymerization (Gopal et al., 2001). GOSs synthesized by LAB β-galactosidases mainly contain di- and trisaccharides, which are dominantly β-(1–3) or β-(1–6) linked (Toba et al., 1981; Splechtna et al., 2006). Di- and trisaccharides present in GOS preparations are possibly internalized by lactose permeases of LAB.

In summary, LAB are isolated from the faeces of neonates but are not able to digest complex HMOs. Therefore LAB depend on the presence of bifidobacteria or other gut microorganisms capable of releasing monosaccharide components from HMOs. HMO components lactose, glucose, N-acetylglucosamine and fucose were fermented by strains of LAB to various extents. β-Galactosidases contribute to GOS fermentation but do not degrade HMOs. The preference of LAB for GOS might contribute to their persistence in the faeces of infants fed with a formula containing GOS preparations.


AVAC Ltd, ALIDF and Alberta Milk are acknowledged for financial support. M.G. acknowledges the Research Chairs of Canada for financial support.