ABSTRACT
Milk, in addition to nourishing the neonate, provides a range of complex glycans whose construction ensures a specific enrichment of key members of the gut microbiota in the nursing infant, a consortium known as the milk-oriented microbiome. Milk glycoproteins are thought to function similarly, as specific growth substrates for bifidobacteria common to the breast-fed infant gut. Recently, a cell wall-associated endo-β-N-acetylglucosaminidase (EndoBI-1) found in various infant-borne bifidobacteria was shown to remove a range of intact N-linked glycans. We hypothesized that these released oligosaccharide structures can serve as a sole source for the selective growth of bifidobacteria. We demonstrated that EndoBI-1 released N-glycans from concentrated bovine colostrum at the pilot scale. EndoBI-1-released N-glycans supported the rapid growth of Bifidobacterium longum subsp. infantis (B. infantis), a species that grows well on human milk oligosaccharides, but did not support growth of Bifidobacterium animalis subsp. lactis (B. lactis), a species which does not. Conversely, B. infantis ATCC 15697 did not grow on the deglycosylated milk protein fraction, clearly demonstrating that the glycan portion of milk glycoproteins provided the key substrate for growth. Mass spectrometry-based profiling revealed that B. infantis consumed 73% of neutral and 92% of sialylated N-glycans, while B. lactis degraded only 11% of neutral and virtually no (<1%) sialylated N-glycans. These results provide mechanistic support that N-linked glycoproteins from milk serve as selective substrates for the enrichment of infant-associated bifidobacteria capable of carrying out the initial deglycosylation. Moreover, released N-glycans were better growth substrates than the intact milk glycoproteins, suggesting that EndoBI-1 cleavage is a key initial step in consumption of glycoproteins. Finally, the variety of N-glycans released from bovine milk glycoproteins suggests that they may serve as novel prebiotic substrates with selective properties similar to those of human milk oligosaccharides.
IMPORTANCE It has been previously shown that glycoproteins serve as growth substrates for bifidobacteria. However, which part of a glycoprotein (glycans or polypeptides) is responsible for this function was not known. In this study, we used a novel enzyme to cleave conjugated N-glycans from milk glycoproteins and tested their consumption by various bifidobacteria. The results showed that the glycans selectively stimulated the growth of B. infantis, which is a key infant gut microbe. The selectivity of consumption of individual N-glycans was determined using advanced mass spectrometry (nano-liquid chromatography chip–quadrupole time of flight mass spectrometry [nano-LC-Chip-Q-TOF MS]) to reveal that B. infantis can consume the range of glycan structures released from whey protein concentrate.
INTRODUCTION
Protein glycosylation is a common modification that adds a major dimension of complexity to protein structure and has been linked to significant roles in protein functions, such as protein folding, biological recognition, and enzymatic protection (1, 2). In eukaryotes, N-glycosylation and O-glycosylation are the two major types of glycosylation, whereby N-linked glycans are linked to a polypeptide at a specific asparagine, while O-linked glycans occur at a serine/threonine residue. N-Linked glycoproteins are classified into the following three groups based on composition: high-mannose, complex, and hybrid (combination of high-mannose and complex types) glycoproteins (3). The nearly infinite combinations of linkages, compositions, and structures provided by complex glycan synthesis can create a similarly diverse array of possible glycoprotein structures (4).
While the structure of glycoproteins may be critical for their function, for a variety of purposes, certain bacteria have the ability to deglycosylate and degrade glycoproteins (5–9). This activity has been described primarily for pathogens deglycosylating various host defense glycoproteins, including human IgG (10), RNase B (7), and lactoferrin (11). It has also been shown that released glycans from N-linked glycoproteins serve as a carbon source for pathogen growth. Endo D, an endo-β-N-acetylglucosaminidase from Streptococcus pneumoniae, acts to release asparagine-linked oligosaccharides (6); however, additional degradation of the released oligosaccharides by exoglycosidases is required for subsequent growth (5, 9, 12, 13). Commensal Bacteroides species can release and consume the O-linked glycans in mucin, providing a fitness advantage in vivo (14, 15). These Bacteroides species also consume the structurally similar oligosaccharides found in milk (16). Curiously, there is little information regarding whether the reported bifidogenic effect of milk oligosaccharides is restricted to free oligosaccharides or if the structurally similar glycans found conjugated to proteins and lipids may serve a similar function (17–19).
Milk from many species contains an abundant amount of complex oligosaccharides (20–22). These can be found free, such as human milk oligosaccharides (HMOs), or as conjugated glycans, such as those found attached to proteins (glycoproteins) or lipids (glycolipids) (22). HMOs possess many diverse structures (23), the result of an array of configurations of the component monosaccharides: glucose, galactose, N-acetylglucosamine, fucose, and N-acetylneuraminic acid. HMOs are clearly a factor that influences populations of beneficial microorganisms, such as Bifidobacterium, in the infant gut (22, 24). Several Bifidobacterium species are known to grow to high cell densities during in vitro growth on HMOs as the sole carbon source (25–28), which mirrors observations of breast-fed infants consuming human milk supplemented with Bifidobacterium longum subsp. infantis (B. infantis) (29). Of these species of Bifidobacterium, only B. infantis possesses a large cassette of genes associated with HMO consumption. The possession of this cluster of genes uniquely enables this subspecies to grow to high cell densities on a broad array of HMOs as sole carbon sources, while many other bifidobacteria do not have this ability (30) or are able to consume only structures with limited diversity (31). In addition, several different strategies are employed by groups of Bifidobacterium to utilize HMOs. B. infantis ATCC 15697 is believed to import HMOs intact and subsequently to employ intracellular glycoside hydrolases, such as β-galactosidases, N-acetyl-β-d-hexosaminidases, α-fucosidases, and α-sialidases, to deconstruct HMOs (31–35). In contrast, other species of Bifidobacterium (e.g., B. bifidum) employ extracellular glycosidases to sequentially degrade these structures and to transport mono-, di-, or oligosaccharides rather than the complex, compositionally diverse, and often branched structures which comprise HMOs (31).
Similar to the diverse structures of HMOs, many different types of conjugated glycans are found in human milk. While HMOs have long been known to serve as prebiotic substrates for Bifidobacterium, it is unclear whether glycoproteins, owing to their structural similarity to HMOs, perform a similar function. Some glycoconjugates, such as lactoferrin and the κ-casein-derived glycomacropeptide, have previously been associated with a bifidogenic effect (8, 17, 18, 36, 37), and strains of Bifidobacterium have even been shown to grow on a structurally distant yeast mannoprotein as the sole carbon source (8). In the case of glycoprotein enrichment of bifidobacteria, it remained unclear to what extent the individual glycan and protein components are responsible for the enrichment (24). The ability to deglycosylate or degrade glycoproteins has been demonstrated for select species of Bifidobacterium (8, 38, 39). We recently showed that B. infantis ATCC 15697 contains an endoglycosidase, EndoBI-1 (glycosyl hydrolase family 18), that has activity on major types of N-linked glycans found in glycoproteins and can release glycans from human lactoferrin and immunoglobulins (8).
Bovine milk glycans have similarities to human milk glycans (40–43). We recently determined the optimal conditions and kinetic parameters of N-glycan release from concentrated bovine whey for large-scale N-glycan production using EndoBI-1 (13, 44, 45). In this work, EndoBI-1 treatment of milk glycoproteins demonstrated that freed glycans, as opposed to deglycosylated milk protein, are a good substrate for rapid and selective growth, similar to HMOs. Moreover, the selectivity of consumption of individual N-glycans was determined using advanced mass spectrometry (nano-liquid chromatography chip–quadrupole time of flight mass spectrometry [nano-LC-Chip-Q-TOF MS]) to reveal that B. infantis can consume the range of glycan structures released from whey protein concentrate.
MATERIALS AND METHODS
Bacteria and media.Bifidobacterium longum subsp. infantis (B. infantis) ATCC 15697 and Bifidobacterium animalis subsp. lactis (B. lactis) UCD316 cultures were routinely propagated in deMan-Rogosa-Sharpe (MRS) broth (Becton Dickinson, Franklin Lakes, NJ) supplemented with 0.05% (wt/vol) l-cysteine at 37°C under anaerobic conditions. Escherichia coli strains Top10 (GeneTarget Inc., San Diego, CA) and BL21* (Invitrogen, Carlsbad, CA), used for gene cloning and expression, were propagated in Luria broth (LB) under selective conditions (100 μg/ml carbenicillin).
Gene cloning and protein expression and purification.Gene amplification and protein expression in E. coli BL21* were performed as described by Sela et al. (31). A pEco-T7-cHis Eco cloning kit (GenTarget Inc., San Diego, CA) was used for gene cloning. Blon_2468 (EndoBI-1) was PCR amplified from B. infantis ATCC 15697 genomic DNA by use of a high-fidelity polymerase with the following primers: 5′-TTTGTACAAAAAAGCAGGCACCATGAATGCGGACGCCGTTTCTCCGAC-3′ and 5′-TTTGTACAAGAAAGCTGGGTTGCCGGTCGCACTCAGTTGCTTCGG-3′. Transmembrane domains and the signal peptide were not amplified to facilitate protein expression and purification, and a C-terminal polyhistidine tag was added to facilitate purification. E. coli BL21* containing the vector was grown for 3 h at 37°C to reach an optical density at 600 nm (OD600) of ∼0.6. Protein expression was induced by the addition of IPTG (isopropyl-β-d-thiogalactopyranoside) to a final concentration of 0.5 mM, and cells were grown at 37°C for an additional 6 h. Bacteria were collected by centrifugation at 4,000 rpm (model A-4-81 rotor; Eppendorf) for 20 min at 4°C. All subsequent steps for bacterial lysis were achieved at 4°C. All solutions were supplemented with a protease inhibitor cocktail (Roche, San Francisco, CA). Cell pellets were incubated in 100 ml of BugBuster (Novagen, Billerica, MA) for 10 min at 24°C. DNase I (200 μl) (Roche, San Francisco, CA) and 100 μl of lysozyme (100 mg/ml) were added to the sample, and the mixture was placed on ice for 30 min. Cell lysates were centrifuged at 13,000 rpm (model F45-24-11 rotor; Eppendorf) for 30 min to remove cell debris.
Protein was purified by affinity chromatography using 5-ml prepacked Ni-charged columns (Bio-Rad, Hercules, CA). All chromatographic steps were performed using a model EP-1 Econo pump (Bio-Rad) and a model 2110 fraction collector (Bio-Rad) at a flow rate of 5 ml/min. The column was equilibrated with 25 ml of a solution containing 300 mM KCl–50 mM KH2PO4, and 5 mM imidazole buffer (pH 8). Fifty milliliters of sample was loaded into the column. The flowthrough was collected, and the column was washed successively with 30 ml of 300 mM KCl–50 mM KH2PO4–5 mM imidazole buffer (pH 8) and then 20 ml of 300 mM KCl–50 mM KH2PO4–10 mM imidazole buffer (pH 8). The bound protein was eluted with a stepwise gradient, using imidazole concentrations ranging from 100 to 300 mM. The purity of the EndoBI-1 fractions was evaluated by SDS-PAGE. Purified protein was concentrated using a 15-ml, 30-kDa-cutoff centrifugal filter device (Amicon, Millipore, Billerica, MA), and buffer was exchanged for 1× saline sodium citrate (SSC; 1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate), using Bio-Gel P-30 in SSC buffer columns (Bio-Rad). The protein concentration was determined by using a Qubit protein assay kit (Life Technologies, Grand Island, NY), and the purified enzyme was stored at −80°C.
Gene expression analysis.Biological triplicate cultures of B. infantis ATCC 15697 were grown to an OD600 of 0.6 (early mid-log phase) in MRS medium containing lactose (2%) as the sole carbon source, harvested by centrifugation, and washed in prewarmed (37°C) MRS containing no carbohydrates. Cells were pelleted and resuspended in the original volume of prewarmed medium containing lactose (2% [wt/vol]; positive control), treated whey protein (10% [wt/vol]), or untreated whey protein (10% [wt/vol]) as the sole carbon source. After 2 h of anaerobic incubation at 37°C, 2-ml aliquots of culture were harvested by centrifugation and immediately resuspended in an equal volume of RNALater (Ambion/Life Technologies, Grand Island, NY) and frozen at −80°C for later use. RNA was extracted using an RNAqueous kit according to the manufacturer's instructions, except that cell lysis was preceded by a 15-min incubation with lysozyme (50 mg/ml) and mutanolysin (120 U) at 37°C and a 5-min incubation at 42°C with 10 μl of proteinase K (Qiagen, Valencia, CA). RNA was subjected to a DNase treatment using TURBOfree DNase (Ambion/Life Technologies, Grand Island, NY) before reverse transcription-PCR (RT-PCR).
Two microliters of DNase-treated RNA was used with 20 μl of SYBR master mix (Applied Biosystems/Life Technologies, Grand Island, NY) in a reaction mixture containing 0.1 μl of RNase inhibitor (Applied Biosystems/Life Technologies, Grand Island, NY), 10 pmol of each primer, and 0.1 μl of MultiScribe reverse transcriptase (Applied Biosystems/Life Technologies, Grand Island, NY). Blon_2468 expression was measured via primers Blon_2468F (5′-ACCGGCAAGATCTACACAGC-3′) and Blon_2468R (5′-GCACTCAGTTGCTTCGGTTG-3′) and compared to expression of a tRNA synthase gene by use of primers Bl_0301F (5′-CAACCGCCGCGATCTTC-3′) and Bl_0301R (5′-CCAGCTGTGAAAGCAACGTGTT-3′), previously described for B. longum and used in studies of B. infantis expression (46) but modified to account for a single-base mismatch to B. infantis ATCC 15697. RT-PCR was preceded by a synthesis step (48°C for 30 min and 95°C for 10 min), followed by 40 cycles of 95°C for 15 s and 60°C for 60 s. A subsequent melting curve step was conducted to ensure amplification product specificity. Normalized expression was calculated as previously described (31), with outliers removed by Grubb's test, and the results were compared to those for cells incubated on lactose by a ratio-paired t test, using GraphPad Prism 6.
Pilot-scale concentration of bovine milk glycoproteins.Protein concentration from bovine colostrum whey was carried out using a pilot-scale cross-flow membrane system (model L; GEA Filtration, Hudson, WI). The system was composed of a 2.5-in.-diameter spiral membrane housing (1 to 2 m2), a 95-liter jacketed stainless steel feed tank, a Proline Promass 80 E flowmeter (Endress+Hauser, Reinach, Switzerland), a heat exchanger, and a 7.0-horsepower feed pump (model D10EKSGSNECF; Hydra-Cell, Minneapolis, MN). After upstream lactose hydrolysis (0.1% Aspergillus oryzae beta-galactosidase [EC 3.2.1.23], 30 min, 40 to 43°C), 74 liters of bovine colostrum whey was ultrafiltered in a single batch with a 10-kDa-cutoff polyethersulfone spiral-wound membrane (effective area of 1.86 m2) to a concentration factor of 5.4 (concentration factor = volume of feed/volume of retentate). Whey protein concentration was performed at a constant temperature of 40 to 43°C, a transmembrane pressure of 300 kPa, and a recirculation flow rate of 10 liters/min. After a concentration factor of 5.4 was achieved, the protein-rich retentate was diluted back to its original volume with water. Simple sugars (e.g., residual lactose and the free monosaccharides derived from its hydrolysis) lack selective prebiotic activity and can be a confounding factor in all functional studies, so they were removed by diafiltration. Additionally, in order to test only the prebiotic activity of the N-glycans newly released from glycoproteins, we performed two discontinuous diafiltrations (by volume reduction) to increase the removal of free oligosaccharides from the ultrafiltration retentate.
N-Glycan release from milk proteins.One hundred milliliters of concentrated bovine colostrum whey was incubated for 18 h at 37°C with 20 mg of EndoBI-1 in 20 mM Na2HPO4, pH 5. Seven hundred milliliters of pure cold ethanol was added to the mixture to precipitate proteins. The mixture was incubated for 2 h at −20°C. The mixture was centrifuged at 4,000 rpm (model A-4-81 rotor; Eppendorf) for 10 min at 4°C. The resulting supernatant, containing the soluble released N-glycans, was dried in a rotary evaporator (Heidolph 36000130 Hei-Vap value collegiate rotary; Fisher Scientific, Waltham, MA). Samples were rehydrated in 100 μl of water, vortexed, and sonicated for glycan quantification and purification.
Glycan quantification.A microplate colorimetric carbohydrate assay (Biovision, Milpitas, CA) was used to quantify the purified glycans. A commercial mannose standard (0, 2, 4, 6, 8, and 10 μl of a 2-mg/ml solution) was used to create a standard curve. The volume of each sample was adjusted to 30 μl per well with water. A sample and 150 μl of concentrated sulfuric acid (98%) were added to each well. Samples were mixed on a shaker for ∼1 min and then incubated at 85°C for 15 min. After incubation, 30 μl of developer (provided by the manufacturer) was added to each well. Samples were again mixed on the shaker for 5 min. The OD490 of each sample was measured. The OD490 was applied to the mannose standard curve linear function to calculate the quantity of carbohydrate in the sample.
Bacterial fermentation of N-glycans.B. infantis ATCC 15697 and B. lactis UCD316 colonies were cultured in MRS broth and incubated overnight at 37°C in an anaerobic growth chamber (Coy Laboratory Products). The resultant cultures were inoculated at 1% (vol/vol) into 100 μl of reconstituted MRS broth supplemented with 2% N-glycans (wt/vol), 10% deglycosylated bovine whey (wt/vol), or 10% bovine whey (MRSN-glycan) as the sole carbohydrate and then overlaid with 25 μl of sterile mineral oil in a 96-well microtiter plate to prevent evaporation. Cell growth was monitored in real time by assessing the OD600 by use of a BioTek PowerWave 340 plate reader (BioTek, Winooski, VT) every 30 min, preceded by 15 s of shaking at variable speed. Three biological replicates were performed for each sample. Once harvested, culture supernatants were centrifuged at 3,000 × g for 15 min and filtered through a 0.22-μm-pore-size membrane (Millipore, Billerica, MA) prior to storage at −80°C. An inoculated sample (control) with no carbon source was performed in parallel as a plate control.
Glycan purification for mass spectrometry.Recovered N-glycans from the bacterial supernatants and control samples were loaded on a C18 plate (Glygen, Columbia, MD). The plate was conditioned three times with 100 μl of 80% acetonitrile (ACN) containing 0.1% trifluoroacetic acid (TFA) in water, followed by three times with 100 μl of water. Samples were loaded, and N-glycans were eluted with 3 volumes of water. The N-glycan solution was loaded on a PGC SPE (porous graphitic carbon, solid-phase extraction) plate (Glygen, Columbia, MD) previously optimized for the purification of free glycans. The plate was conditioned as described previously (47). After sample loading, wells were washed six times with 200 μl of water, and N-glycans were eluted thrice with 200 μl of 40% ACN containing 0.1% TFA in water. The enriched N-glycan fractions were dried overnight by vacuum drying. Samples were rehydrated in 50 μl of water, vortexed, sonicated, and diluted 50 times prior to mass spectrometry analysis.
Nano-LC-Chip-Q-TOF MS.N-Glycans were analyzed using an Agilent 6520 accurate-mass Q-TOF LC-MS with a microfluidic nano-electrospray chip (Agilent Technologies, Santa Clara, CA). N-Glycans were separated using a high-pressure liquid chromatography (HPLC) chip with a 40-nl enrichment column and a 43-mm by 75-μm analytical column, both packed with 5-μm porous graphitized carbon (PGC). The system was composed of a capillary and a nanoflow pump, and both used binary solvents consisting of solvent A (3% [vol/vol] ACN, 0.1% [vol/vol] formic acid in water) and solvent B (90% [vol/vol] ACN, 0.1% [vol/vol] formic acid in water). Two microliters of sample was loaded with solvent A at a capillary pump flow rate of 4 μl/min. N-Glycan separation was performed with a 65-min gradient delivered by the nanopump at a flow rate of 0.3 μl/min. The 65-min gradient used the following program: 0% B (0.0 to 2.5 min), 0 to 16% B (2.5 to 20.0 min), 16 to 44% B (20.0 to 30.0 min), 44 to 100% B (30.0 to 35.0 min), and 100% B (35.0 to 45.0 min). The gradient was followed by equilibration at 0% B (45.0 to 65.0 min). Data were acquired within the mass range of 450 to 3,000 m/z for N-glycans in the positive ionization mode, with an acquisition rate of 1 spectrum/s for N-glycans. An internal calibrator ion of 922.010 m/z from the tuning mix (ESI-TOF tuning mix G1969-85000; Agilent Technologies) was used for continual mass calibration. For tandem MS analysis, N-glycans were fragmented with nitrogen as the collision gas. Spectra were acquired within the mass range of 100 to 3,000 m/z. The collision energies corresponded to voltages (Vcollision) based on the following equation: Vcollision = m/z (1.5/100 Da) V − 3.6 V; the slope and offset of the voltages were set at 1.5/100 Da and −3.6, respectively. Acquisition was controlled by MassHunter workstation data acquisition software (Agilent Technologies).
N-Glycan identification.All compounds in the chromatograms were identified with MassHunter qualitative analysis software (version B.06.00 SP2; Agilent Technologies). Compounds were extracted using the molecular feature extractor algorithm. The software generated extracted compound chromatograms in the range of 400 to 3,000 m/z, with an ion count cutoff of 1,000, allowed charge states of +1 to +3, retention times of 5 to 40 min, and a typical isotopic distribution of small biological molecules. The resulting compounds were matched to a bovine milk N-glycan library (41), using a mass error tolerance of 20 ppm. The N-glycans from the library were composed of hexose (Hex), HexNAc, fucose, NeuAc, and N-glycolylneuraminic acid (NeuGc). The assignment of N-glycans was confirmed by tandem mass spectrometry.
Calculation of the relative abundances of N-glycans was performed by MassHunter Profinder software, using the batch targeted feature extraction algorithm. A database was built to contain the molecular formulas, masses, and retention times of identified N-glycans from the control sample. This library was used in combination with the batch targeted feature extraction algorithm. A minimum abundance of 1,000 counts was used to filter out low-abundance compounds. Compounds were extracted by using allowed charge states of +1 to +3, a mass error tolerance of 20 ppm, and a retention time tolerance of 1 min.
N-Glycan consumption was calculated with respect to that of the uninoculated control by normalizing the summed abundance of neutral and sialylated N-glycans in the ion count for the bacterial supernatant to that for the control by using the following equation:
where API is absolute peak intensity and n is the number of identified N-glycans. The relative amount was expressed as a percentage of the total consumption.
RESULTS
Release and purification of N-glycans from concentrated whey proteins.Release and purification of N-glycans were accomplished using EndoBI-1 as described in Fig. 1. Briefly, EndoBI-1 was incubated with concentrated bovine colostrum whey under conditions previously determined to be optimal for enzyme activity (pH 5, overnight, 37°C) (44). Cleavage was performed under nondenaturing conditions because the goal of this study was to produce the N-glycans in a native state and to preserve biological function. Using this method, a total of 80 mg of N-glycans was released from 40 ml of concentrated bovine colostrum whey. Mass spectrometry-based characterization of the released glycans identified 18 species, including 6 neutral and 12 sialylated moieties (Table 1). Interestingly, four of the compositions contained fucosylated oligosaccharides, and their detection extends the previous identification of fucosylated oligosaccharides within the bovine milk glycome to N-linked glycoproteins (48).
Methods involved in release and purification of N-glycans by use of purified EndoBI-1. EndoBI-1 was cloned and expressed in E. coli. Purified EndoBI-1 was incubated with bovine colostrum whey to release N-glycans.
Details of released N-glycansa
Growth of bifidobacteria on whey, deglycosylated whey proteins, and N-glycans released from whey.Previous work showed a bifidogenic effect upon ingestion of milk glycoproteins (17, 18, 49). Since milk glycoprotein-derived peptides have been implicated in bifidogenic responses (50), it was unclear if this effect is a result of the catabolism of the protein itself or the associated glycan (or both) of these glycoconjugates. To test this, we examined the growth of B. infantis ATCC 15697 and B. lactis UCD316 on 10% whey glycoprotein concentrate, in which the total glycan concentration matches that used in a previous study (i.e., 2%) (13). B. infantis ATCC 15697 grows well on HMOs (25, 26), while B. lactis UCD316 does not (29). B. infantis readily grew to a high cell density on released N-glycans, while no growth occurred on 10% deglycosylated whey protein concentrate (Fig. 2). Intact whey protein concentrate supported moderate growth of B. infantis; however, none of the substrates tested supported vigorous growth of B. lactis UCD316.
Representative growth curves for B. infantis (A) and B. lactis (used as a negative control) (B) on released N-glycans (
), whey (
), or deglycosylated whey (
). The growth responses of each sample were measured by determining the OD600. Growth curves were performed in biological triplicates.
To determine if EndoBI-1 expression is related to substrate availability, given the growth of B. infantis on whey protein concentrate, we measured the relative expression of Blon_2468, the gene encoding the EndoBI-1 endoglycosidase (8), during growth on this substrate. Blon_2468 was significantly upregulated during incubation on untreated whey but not on whey that had been deglycosylated by pretreatment with EndoBI-1 prior to incubation, and not in the presence of lactose (Fig. 3), suggesting that the Blon_2468 induction signaling mechanism is based on the glycan portion of the glycoprotein. Given that free milk oligosaccharides do not induce Blon_2468 expression (51), this suggests that B. infantis senses intact glycoprotein and induces Blon_2468 in response. The specific mechanism for this signal remains to be determined.
Normalized gene expression levels show a significant increase in Blon_2468 (EndoBI-1) expression relative to that in lactose-grown cells (P = 0.0145) for cells incubated on whey protein concentrate but not for cells incubated on the same whey protein concentrate pretreated with EndoBI-1 to remove complex glycans from the underlying protein scaffold. Gene expression was normalized using cysteinyl-tRNA synthetase gene expression.
Glycoprofiling of N-glycan consumption by B. infantis and B. lactis.To determine the potential preferential consumption of specific N-glycan compositions, mass spectrometry-based glycoprofiling was performed on culture supernatants. Supernatants were collected after 96 h of growth, and residual N-glycans were purified by solid-phase extraction and profiled by nano-HPLC-chip-TOF MS. The separation of N-glycans was performed on a microchip packed with porous graphitized carbon, which enabled isomer-level separation with reproducible retention times. Tandem MS analysis generated specific fragment ions common to all N-glycans and allowed confirmation of N-glycan compositions, including those of both neutral and sialylated N-glycans. For this study, we specifically profiled the bifidobacterial consumption of 6 neutral and 12 sialylated N-glycans (Table 1). Figure 4 shows the extracted compound chromatograms for N-glycans obtained from the B. infantis and B. lactis growth supernatants as well as an uninoculated control. This analysis revealed the presence of 18, 16, and 6 N-glycans remaining in the uninoculated, B. lactis, and B. infantis samples, respectively. The majority of the N-glycans in the uninoculated control were also present in the B. lactis sample, with similar peak areas, suggesting a minimal consumption of N-glycans by B. lactis. Conversely, B. infantis consumed the majority of the N-glycans present.
Nano-LC-Chip-Q-TOF MS extracted compound chromatograms for free N-glycans from bovine colostrum whey concentrate remaining in the medium after bacterial fermentation (blue, uninoculated control; green, B. lactis; and red, B. infantis). Freed N-glycans from whey concentrate were produced as described in Materials and Methods. Green circles, yellow circles, blue squares, red triangles, purple diamonds, and gray diamonds represent mannose, galactose, HexNAc, fucose, NeuAc, and NeuGc residues, respectively.
The data revealed that B. infantis consumed 73% of neutral and 92% of sialylated N-glycans, whereas B. lactis degraded 11% of neutral and only 0.9% of sialylated N-glycans. The majority of released N-glycans were consumed by B. infantis, except for the neutral N-glycan 3Hex-3HexNAc (Fig. 5). Notably, three N-glycans (3Hex-5HexNAc, 4Hex-3HexNAc-1Fuc, and 5Hex-3HexNAc-2NeuGc) could not be detected after incubation with B. lactis.
Glycoprofiling of consumption of freed N-glycans extracted from the medium after fermentation by two different Bifidobacterium species. (A) Consumption of freed neutral N-glycans from whey concentrate by B. lactis or B. infantis in comparison to uninoculated medium (control). (B) Consumption of freed sialylated N-glycans from whey concentrate by B. lactis or B. infantis in comparison to uninoculated medium (control). The x axis labels depict the monosaccharidic compositions of the N-glycans (Hex-HexNAc-Fuc-NeuAc-NeuGc); for example, the 4_3_0_1_0 composition is 4Hex-3HexNAc-0Fuc-1NeuAc-0NeuGc. Inset graphs show enlarged versions of the lower-abundance glycan types depicted within the indicated portion of each graph.
DISCUSSION
Previous work identified EndoBI-1 as a possible mechanism by which B. infantis utilizes glycan moieties from N-linked glycoproteins in milk as growth substrates (52); however, it remained unclear if the released oligosaccharides actually serve as a better growth substrate than the glycoconjugate. To examine this, purified EndoBI-1 was used to release a range of N-glycans from whey glycoproteins at the laboratory scale. B. infantis clearly showed an ability to consume these released N-glycans as the sole carbon source, growing to a high cell density mimicking the strong growth of B. infantis on complex HMOs. In contrast, B. lactis did not readily consume released N-glycans as the sole carbon source, which is not surprising considering that the strain is unable to consume structurally analogous HMOs and lacks a homolog to EndoBI-1 (52) or other genes associated with HMO catabolism. Furthermore, released N-glycans enabled more rapid growth than that seen with the conjugated form, reflecting both a preference for free glycans (similar to HMOs) and the comparative difficulty in accessing conjugated glycans, where the requirement for B. infantis EndoBI-1 expression creates a rate-limiting step prior to consumption.
Only a select assortment of infant-associated strains of Bifidobacterium, such as B. longum subsp. longum, B. infantis, B. breve, and B. bifidum, have the ability to grow to high cell densities on HMOs as the sole carbon source (25–28). B. bifidum and B. infantis represent the two models for HMO degradation. B. bifidum, a species well known to grow on the glycoprotein mucin (32, 53), deploys extracellular glycosyl hydrolases to degrade complex HMOs, followed by importation and consumption of select components. This “external” degradation phenotype results in the potential of released sugars cross-feeding other bacterial clades (54, 55), or even pathogens, as has been witnessed for Bacteroides species (56). In contrast, B. infantis and, to a lesser extent, B. breve have an “internal” degradation phenotype. B. infantis selectively imports HMOs via an array of specific family 1 solute binding proteins that bind the oligosaccharides prior to transport by associated ABC transporters (32). Upon import, B. infantis deploys a suite of glycosyl hydrolases (α-fucosidases, α-sialidases, β-galactosidases, and N-acetyl-β-d-hexosaminidases), whose activity on HMOs has been described previously (31, 33–35). The latter, “internal” degradation phenotype is likely a competitive advantage by removing the opportunity for cross-feeding.
Compared to the role of HMOs in shaping the distal gut microbiota of infants, which has been described extensively (57), the role of glycoconjugates has been relatively unexplored. Glycoprotein degradation strategies have been examined among several bifidobacterial species (8, 58), and EndoBI-1 homologs have been identified and characterized for strains of B. longum subsp. longum, B. breve, and others (8). All are deployed on the cell wall, which facilitates the release of intact milk oligosaccharides that, like HMOs, are resistant to degradation by other gut bacteria but are readily transported and consumed by the EndoBI-1-producing strain or neighboring bacteria.
The glycoproteins in milk thus provide an additional energy source for a select group of infant-associated gut bacteria (48). In this work, we demonstrate that the oligosaccharide portions of bovine milk protein glycoconjugates, once released by an endoglycosidase, can serve as selective growth substrates for infant-associated taxa, such as B. infantis. The nature of the specific release of these larger milk oligosaccharides from glycoconjugates instead of the release of sugar monomers would cross-feed only similarly coevolved species, such as B. infantis, over less specific colonizers that are unable to internalize and degrade these released glycans (8).
Finally, the production of synthetic HMO-like structures is challenging due to the complexity of HMOs (22), and despite differences between the oligosaccharide compositions of bovine and human milks (22, 42), we demonstrate here that selective, bioactive milk oligosaccharides that mimic the specificity of HMOs can be produced from widely available and low-cost bovine dairy sources. While bovine milk is generally considered to be a poor source of complex milk oligosaccharides (42), we show here that bovine colostrum whey can be treated to release a significant amount of complex milk glycans whose biological activity and enrichment specificity are intact. Future strategies that utilize existing products with demonstrable safety and low substrate cost may be an attractive alternative to chemical synthesis of complex HMO mimics.
ACKNOWLEDGMENTS
This work was supported by the UC Davis RISE program, National Institutes of Health (NIH) awards AT008759 and AT007079, and the Peter J. Shields Endowed Chair in Dairy Food Science. S.K. was supported in part by the Ministry of Education, Turkey, and S.A.F. was supported in part by NIH grant F32AT008533.
D.A.M. and D.B. are cofounders of, and S.A.F. is an employee of, Evolve Biosystems, a company focused on diet-based manipulation of the gut microbiota.
FOOTNOTES
- Received 18 February 2016.
- Accepted 26 March 2016.
- Accepted manuscript posted online 15 April 2016.
- Copyright © 2016, American Society for Microbiology. All Rights Reserved.
