bifidobacterium longum subsp. infantis uses two - glycobiology

Bifidobacterium longum subsp. infantis uses two differentβ-galactosidases for selectively degrading type-1and type-2 human milk oligosaccharides

Erina Yoshida2, Haruko Sakurama2, Masashi Kiyohara2,Masahiro Nakajima3, Motomitsu Kitaoka3,Hisashi Ashida4, Junko Hirose5, Takane Katayama1,2,Kenji Yamamoto2, and Hidehiko Kumagai2

2Research Institute for Bioresources and Biotechnology, Ishikawa PrefecturalUniversity, Nonoichi, Ishikawa 921-8836, Japan; 3National Food ResearchInstitute, National Agriculture and Food Research Organization, Tsukuba,Ibaraki 305-8642, Japan; 4Graduate School of Biostudies, Kyoto University,Sakyo-ku, Kyoto 606-8502, Japan; and 5Department of Food Science andNutrition, The University of Shiga Prefecture, Hikone, Shiga 522-8533, Japan

Received on July 1, 2011; revised on August 11, 2011; accepted onAugust 13, 2011

The breast-fed infant intestine is often colonized by par-ticular bifidobacteria, and human milk oligosaccharides(HMOs) are considered to be bifidogenic. Recent studiesshowed that Bifidobacterium longum subsp. infantis cangrow on HMOs as the sole carbon source. This ability hasbeen ascribed to the presence of a gene cluster (HMOcluster-1) contained in its genome. However, the metab-olism of HMOs by the organism remains unresolvedbecause no enzymatic studies have been completed. In thepresent study, we characterized β-galactosidases of thissubspecies to understand how the organism degrades type-1 (Galβ1-3GlcNAc) and type-2 (Galβ1-4GlcNAc) isomersof HMOs. The results revealed that the locus tagBlon_2016 gene, which is distantly located from the HMOcluster-1, encodes a novel β-galactosidase (Bga42A) with asignificantly higher specificity for lacto-N-tetraose (LNT;Galβ1-3GlcNAcβ1-3Galβ1-4Glc) than for lacto-N-biose I(Galβ1-3GlcNAc), lactose (Lac) and type-2 HMOs. Theproposed name of Bga42A is LNT β-1,3-galactosidase.The Blon_2334 gene (Bga2A) located within the HMOcluster-1 encodes a β-galactosidase specific for Lac andtype-2 HMOs. Real-time quantitative reverse transcrip-tion–polymerase chain reaction analysis revealed thephysiological significance of Bga42A and Bga2A in HMOmetabolism. The organism therefore uses two differentβ-galactosidases to selectively degrade type-1 and type-2HMOs. Despite the quite rare occurrence in nature ofβ-galactosidases acting on type-1 chains, the close

homologs of Bga42A were present in the genomes ofinfant-gut associated bifidobacteria that are known toconsume LNT. The predominance of type-1 chains inHMOs and the conservation of Bga42A homologs suggestthe coevolution of these bifidobacteria with humans.

Keywords: human milk oligosaccharides / infant-gutassociated bifidobacteria / lacto-N-tetraose / β-galactosidase

Introduction

Human milk contains oligosaccharides (the degree ofpolymerization between 3 and 14) as the third most abundantsolid component (�20 g/L) after lactose (Lac) and lipids(Newburg and Neubauer 1995; Bode 2006; Urashima et al.2011). These oligosaccharides (human milk oligosaccharides,HMOs) have diverse structures and more than 100 molecularspecies have been identified (Kunz et al. 2000; Kobata 2010;Urashima et al. 2011). HMOs are resistant to host gastrointes-tinal digestion, and thus the majority of the HMOs reach thecolon (Gnoth et al. 2000). The structures of HMOs mimic thesugar chains of the host epithelial cell surfaces, and conse-quently, HMOs act as decoys to inhibit the binding of patho-gens and toxins to the host intestinal cells (Morrow et al.2004). In addition, HMOs are believed to promote the growthof particular bifidobacteria in the infant gut, and the resultingbifidobacteria-rich gut ecosystem is thought to be importantfor infant health (Penders et al. 2006).In the genus Bifidobacteria, Bifidobacterium bifidum, B.

breve, B. longum subsp. infantis and B. longum subsp.longum are frequently isolated from infant feces (Benno et al.1984). Recently, Marcobal et al. (2010) showed that B.longum subsp. infantis is an avid consumer of HMOs and canuse HMOs as the sole carbon source, whereas most of theother intestinal bacteria examined were unable to assimilateHMOs. The genome of B. longum subsp. infantisATCC15697 contains a large gene cluster that comprisesseveral glycosidases and ATP-binding cassette (ABC) trans-porters predicted to be involved in the metabolism of HMOs(Sela et al. 2008). A comparative genomic survey showed thatthe occurrence of the cluster is correlated with the survivalof this subspecies on HMOs (LoCascio et al. 2010).Consequently, LoCascio et al. (2010) suggested that theHMO-consuming phenotype (HMO+) of the subspecies can

1To whom correspondence should be addressed: Tel: +81-76-227-7513;Fax: +81-76-227-7557; e-mail: [email protected]

Glycobiology vol. 22 no. 3 pp. 361–368, 2012doi:10.1093/glycob/cwr116Advance Access publication on September 16, 2011

© The Author 2011. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected] 361

Dow

nloaded from https://academ

ic.oup.com/glycob/article/22/3/361/1988231 by guest on 22 January 2022

be attributed to the presence of this cluster (designated as theHMO cluster-1). However, the characteristics of the encodedenzymes have not been determined.The core structures of HMOs consist of Lac at the reducing

ends, that is elongated by β-(1→ 3)-linked lacto-N-biose I(LNB, type 1; Galβ1-3GlcNAc) and/or β-(1→ 3/6)-linkedN-acetyllactosamine (LacNAc, type 2; Galβ1-4GlcNAc) (Kunzet al. 2000; Kobata 2010). A prominent feature of HMOs is thepredominance of the type-1 chain over the type-2 chain, aslacto-N-tetraose (LNT; Galβ1-3GlcNAcβ1-3Galβ1-4Glc) rep-resents the most abundant core structure. Such type-1-richcompositions have not been observed in the milk of othermammals (Urashima et al. 2011).We have recently elucidated that B. bifidum, which is

another consumer of HMOs (Turroni et al. 2010), has a dedi-cated pathway for degrading type-1 HMOs (Kitaoka et al.2005; Wada et al. 2008). B. bifidum uses a secretorylacto-N-biosidase to hydrolyze LNT into LNB and Lac. Theliberated LNB is then incorporated into the cells by an ABCtransporter specific for LNB and galacto-N-biose (GNB;Galβ1-3GalNAc) (the GNB/LNB transporter; Suzuki et al.2008). The disaccharide is subsequently phosphorolyzed by aGNB/LNB phosphorylase to produce galactose-1-phosphateand N-acetylglucosamine (GlcNAc) (Kitaoka et al. 2005) andfurther metabolized (Nishimoto and Kitaoka 2007a). Thispathway (the GNB/LNB pathway), which involves LNB as akey intermediate, occurs in B. bifidum and some strains of B.longum subsp. longum (Wada et al. 2008), but apparentlydoes not work in B. longum subsp. infantis because all of thegenomes determined so far do not contain lacto-N-biosidasehomologs (LoCascio et al. 2010). B. longum subsp. infantispossesses a GNB/LNB phosphorylase, but this enzyme exclu-sively acts on the disaccharide and never on LNT (Kitaokaet al. 2005; Hidaka et al. 2009). Therefore, how the organismdegrades the type-1 HMOs and what enzymes are involvedremain unresolved.In this study, we show that B. longum subsp. infantis

directly incorporates LNT and hydrolyzes it inside the cellby a specific β-galactosidase (suggested name: LNTβ-1,3-galactosidase) encoded by the locus tag Blon_2016gene. β-Galactosidase (Blon_2334) encoded in the HMOcluster-1 was specific for Lac and type-2 chains. The resultsindicate that the organism uses two different β-galactosidasesto selectively degrade the type-1 and type-2 HMOs and alsoindicated that the HMO+ phenotype of this subspecies shouldnot be attributed solely to the presence of the HMO cluster-1.Interestingly, LNT β-1,3-galactosidase homologs were alsopresent in other infant-related bifidobacteria.

ResultsDifference in the LNT degradation pathway betweenB. longum subsp. infantis ATCC15697 and B. bifidumJCM1254Sela et al. (2008) suggested, through genomic analysis, thatB. longum subsp. infantis ATCC15697 imports HMOs in theirintact forms. Garrido et al. (2011) recently determined thesubstrate specificities of 20 solute-binding proteins of ABCtransporters of the strain and found that the gene product of

Blon_2177 binds LNT, the most abundant HMO core struc-ture. When LNT was anaerobically incubated with the cellsand the culture supernatant was analyzed by thin-layer chrom-atography (TLC), the spot corresponding to LNT graduallydisappeared (20–90 min, Figure 1A). LNT remained uncon-sumed when verapamil, an inhibitor of ABC transporters(Li et al. 1996), was included in the reaction mixture. In bothcases, no sugar spot other than LNT was detected during thecultivation period. The inhibition of the LNT uptake was notobserved when the proton uncoupler carbonyl cyanidem-chlorophenyl hydrazone (CCCP) was added to the reactionmixture (data not shown). On the other hand, when B.bifidum JCM1254 were incubated with LNT (Figure 1B), thespots corresponding to Lac and Gal appeared with a concomi-tant disappearance of LNT. In the presence of verapamil,LNB and Glc were accumulated in the culture supernatant.The results indicated that, while B. bifidum JCM1254 uses theGNB/LNB pathway to degrade LNT, B. longum subsp. infan-tis ATCC15697 directly incorporates LNT using an ABCtransporter.

Degradation of LNT by the cell-free extract of B. longumsubsp. infantisWe then incubated LNT with the cell-free extract of B.longum subsp. infantis and analyzed the reaction products(Figure 1C). The spots corresponding to Gal and a trisacchar-ide appeared immediately (5 min), and the trisaccharide wassubsequently hydrolyzed into its constituent monosaccharidesgalactose (Gal), glucose (Glc) and GlcNAc (30–180 min). Afaint spot corresponding to Lac was also observed. No spotcorresponding to LNB, which is formed if the organism pos-sesses lacto-N-biosidase, or a spot corresponding togalactose-1-phosphate, which is formed if the GNB/LNBphosphorylase acts on LNT, was detected. No degradationoccurred when the heat-inactivated cell-free extract wasincluded in the reaction mixture (data not shown). The resultsshowed that LNT is sequentially hydrolyzed by β-galactosi-dase and β-N-acetylhexosaminidase inside the cells.

Characterization of β-galactosidase (Blon_2334) encodedin the HMO cluster-1The HMO+ phenotype of B. longum subsp. infantis has beenattributed to the HMO cluster-1 (LoCascio et al. 2010).Initially, the gene product of Blon_2334 (located within thecluster) was analyzed as a candidate β-galactosidase acting onHMOs. The gene encodes a protein of 1023 amino acid (aa)residues and does not code for a signal peptide or a cellwall-anchoring motif. A Pfam search (Finn et al. 2010)revealed that the protein belongs to the glycoside hydrolase(GH) family 2 (Cantarel et al. 2009). The purified protein(designated as Bga2A) that was expressed in an Escherichiacoli ΔlacZ background migrated as a single band in sodiumdodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE; Supplementary data, Figure S1A). The apparent mol-ecular mass (120 kDa) agreed with the calculated mass of theprotein (115 kDa). Size-exclusion chromatography indicatedthat the native molecular mass of the protein is 248 kDa, indi-cating that Bga2A forms a dimer in solution. The enzymerequired 1 mM Mg2+ for maximum activity (Goulas et al.

E Yoshida et al.

362

Dow



http://clustalw.ddbj.nig.ac.jp/top-j.html

2009), and the optimal pH was 6.0 (Supplementary data,Figure S1B). The enzyme was stable in the pH range of5.5–9.0 and below 45°C for 30 min (Supplementary data,Figure S2). Bga2A hydrolyzed 4-nitrophenyl-β-D-galactopyra-noside (pNP-β-Gal) efficiently (kcat/Km = 1800 mM−1 s−1) andalso hydrolyzed pNP-β-D-fucopyranoside (pNP-β-Fuc) at alower catalytic rate (kcat/Km = 0.4 mM−1 s−1), but did nothydrolyze other pNP-monosaccharides (Table I). Among theHMO-related oligosaccharides examined, Bga2A most effi-ciently hydrolyzed Lac with Km and kcat values of 1.4 mM and

600 s−1, respectively. LacNAc and lacto-N-neotetraose (LNnT;Galβ1-4GlcNAcβ1-3Galβ1-4Glc) (type-2 chains) were alsoeffectively hydrolyzed (kcat/Km = 86 and 20 mM−1 s−1, respect-ively); however, the enzyme did not act on type-1 structures.The catalytic efficiency for LNT (0.0045 mM−1 s−1) was19,000- and 4000-fold lower than those for Lac and LNnT,respectively. Despite the significantly lower kcat values forLNB and LNT than those for Lac, LacNAc and LNnT, the Km

values were comparable among these substrates (1.4–8.7 mM).The results revealed that Bga2A is responsible for degrading

Fig. 1. Difference in LNT-consuming behavior between B. longum subsp. infantis and B. bifidum. The cells of B. longum subsp. infantis ATCC15697 (A) andB. bifidum JCM1254 (B) were anaerobically incubated with 1 mM LNT in the presence and the absence of 5 mM verapamil. The culture supernatants wereanalyzed by TLC. The cell-free extract of B. longum subsp. infantis ATCC15697 was incubated with 5 mM LNT (C). The samples were taken at the indicatedtimes and analyzed by TLC. The standard sugars (Std) are Gal, Glc, GlcNAc, Lac, LNB and LNT.

Lacto-N-tetraose β-1,3-galactosidase of GH42

363

Dow



http://glycob.oxfordjournals.org/lookup/suppl/doi:10.1093/glycob/cwr116/-/DC1




Lac and type-2 HMOs and essentially cannot be involved inthe degradation of type-1 HMOs.

Locus tag Blon_2016 gene encodes LNTβ-1,3-galactosidaseDue to the above results, we scrutinized the genome of theorganism with the aid of the Carbohydrate-Active enZYmes(CAZy) database (Cantarel et al. 2009) and found the gene pro-ducts of Blon_0268 (606 aa, GH2), Blon_2016 (691 aa,GH42), Blon_2123 (720 aa, GH42), Blon_2411 (326 aa,GH43) and Blon_2416 (706 aa, GH42) as possible enzymeshydrolyzing β-galactosides. All of these genes were predictedto encode intracellular enzymes. These proteins were similarlyexpressed in an E. coli ΔlacZ background and purified. Theproducts of Blon_0268 and Blon_2411 did not exhibit anydetectable activity toward the tested substrates, and these pro-teins were excluded from further investigation. The products ofBlon_2016 (designated as Bga42A), Blon_2123 (Bga42B) andBlon_2416 (Bga42C) gave single bands on SDS–PAGE withapparent molecular masses of 75.5, 79.5 and 73.5 kDa(Supplementary data, Figure S1A), which were consistent withtheir calculated masses (77.4, 80.2 and 77.4 kDa), respectively.The native molecular masses of Bga42A, Bga42B and Bga42Cwere estimated to be 229, 233 and 312 kDa, respectively; there-fore, Bga42A and Bga42B should form trimers, whereasBga42C may form a tetramer in solution. The enzymes hydro-lyzed pNP-β-Gal with catalytic efficiencies of 1600 (Bga42A),120 (Bga42B) and 14 mM−1 s−1 (Bga42C). The activities of

Bga42A and Bga42B toward pNP-β-Fuc were �1% of thosetoward pNP-β-Gal, whereas the activity of Bga42C towardpNP-β-Fuc was similar to that observed for pNP-β-Gal. Theother pNP-sugars were not digested by these enzymes.Optimal pHs of the three enzymes were between 5.5 and

6.0 (Supplementary data, Figure S1B). The enzymes werestable between pH 5.0 and pH 8.0. Bga42A retained 80%activity after 30 min incubation at 55°C, whereas Bga42B andBga42C lost their activities under the same conditions(Supplementary data, Figure S2). All enzymes were stable inthe presence of 0.05% polyoxyethylene sorbitan monolaurate(Tween-20) for at least 3 weeks, but rapidly lost their activitiesin the absence of the detergent.Bga42A released Gal from LNT with a high catalytic effi-

ciency (120 mM−1 s−1), but interestingly, the activity drasticallydecreased when LNB was used as a substrate (3 mM−1 s−1).The enzyme also acted on Lac, LacNAc and LNnT; however,the kcat/Km values for these substrates were 20–300-fold lowerthan that for LNT. No saturation curve for LNnT wasobserved up to 20 mM. Bga42A appears to have a high affi-nity for LNT compared with the other HMO-related substrates(Table I). Bga42B showed low activity on Lac and was essen-tially inactive on both type-1 and type-2 HMOs. Bga42Chydrolyzed Galβ1-4Glc/GlcNAc linkages with very low cata-lytic efficiencies and did not hydrolyze type-1 chains. Theresults revealed that Bga42A is a β-galactosidase highlyspecific for LNT and is essentially the sole β-galactosidaseacting on type-1 HMOs.

Table I. Substrate specificities of one GH2 and three GH42 β-galactosidases from B. longum subsp. infantis ATCC15697

Enzyme Substrate Structure Km (mM) kcat (s–1) kcat/Km (mM–1 s–1)

Bga2A(Blon_2334) pNP-β-Gal 1.7 (±0.1) × 10−1 3.0 (±0.0) × 102 1.8 (±0.1) × 103

pNP-β-Fuc 4.0 (±0.2) × 10−1

Lac Galβ1-4Glc 1.4 (±0.1) × 100 6.0 (±0.2) × 102 4.2 (±0.2) × 102

LacNAc Galβ1-4GlcNAc 5.3 (±0.2) × 100 4.6 (±0.1) × 102 8.6 (±0.3) × 101

LNnT Galβ1-4GlcNAcβ1-3Galβ1-4Glc 6.0 (±0.7) × 100 1.2 (±0.1) × 102 2.0 (±0.1) × 101

LNB Galβ1-3GlcNAc 8.7 (±0.9) × 100 3.0 (±0.1) × 10−2 3.4 (±0.2) × 10−3

LNT Galβ1-3GlcNAcβ1-3Galβ1-4Glc 4.3 (±0.1) × 100 1.9 (±0.0) × 10–2 4.5 (±0.1) × 10−3

Bga42A(Blon_2016) pNP-β-Gal 6.7 (±0.6) × 10–1 1.1 (±0.0) × 103 1.6 (±0.1) × 103

pNP-β-Fuc 1.6 (±0.4) × 101

Lac Galβ1-4Glc 1.6 (±0.2) × 101 9.7 (±0.3) × 101 6.1 (±0.5) × 100

LacNAc Galβ1-4GlcNAc 4.4 (±0.3) × 101 1.6 (±0.0) × 101 3.8 (±0.2) × 10−1

LNnT Galβ1-4GlcNAcβ1-3Galβ1-4Glc 1.1 (±0.1) × 100

LNB Galβ1-3GlcNAc 2.8 (±0.2) × 101 8.6 (±0.3) × 101 3.0 (±0.2) × 100

LNT Galβ1-3GlcNAcβ1-3Galβ1-4Glc 2.2 (±0.3) × 100 2.7 (±0.1) × 102 1.2 (±0.1) × 102

Bga42B(Blon_2123) pNP-β-Gal 1.7 (±0.5) × 100 2.1 (±0.2) × 102 1.2 (±0.3) × 102

pNP-β-Fuc 4.7 (±0.5) × 100 6.4 (±0.2) × 100 1.4 (±0.1) × 100

Lac Galβ1-4Glc 4.4 (±0.1) × 101 8.6 (±0.1) × 100 1.9 (±0.0) × 10−1

LacNAc Galβ1-4GlcNAc 4.2 (±0.4) × 101 1.8 (±0.1) × 10–2 4.3 (±0.2) × 10−4

LNnT Galβ1-4GlcNAcβ1-3Galβ1-4Glc n.d.LNB Galβ1-3GlcNAc 1.7 (±0.1) × 10−3

LNT Galβ1-3GlcNAcβ1-3Galβ1-4Glc 3.9 (±0.3) × 10−4

Bga42C(Blon_2416) pNP-β-Gal 1.4 (±0.1) × 101

pNP-β-Fuc 8.5 (±0.3) × 100

Lac Galβ1-4Glc 7.6 (±0.2) × 10−2

LacNAc Galβ1-4GlcNAc 1.8 (±0.1) × 10−2

LNnT Galβ1-4GlcNAcβ1-3Galβ1-4Glc 1.2 (±0.1) × 10−2

LNB Galβ1-3GlcNAc n.d.LNT Galβ1-3GlcNAcβ1-3Galβ1-4Glc n.d.

Assays were performed in 50 mM sodium phosphate buffer (pH 6.0) containing 0.05% Tween-20. These enzymes did not act on pNP-β-GalNAc,pNP-β-GlcNAc, pNP-α-Ara, pNP-β-Glc, pNP-β-GlcA and pNP-β-Xyl. n.d., not detected.

E Yoshida et al.

364

Dow






Comparison of the expression levels of the fourβ-galactosidase genesTo examine the physiological relevance of theseβ-galactosidases, the transcription levels were quantitated byreal-time reverse transcription (RT)–polymerase chain reaction(PCR) using the cells grown in the HMO or Glc medium.The absence of contaminating DNA was confirmed by theRT-omitted reactions (data not shown). The amounts ofthe transcripts were shown as percentages relative to those ofthe dnaA transcript (Figure 2). The bga2A gene was mostabundantly expressed, and its mRNA level was 3-fold higherthan that of the housekeeping dnaA gene. The amounts ofbga42A transcripts were comparable with those of the dnaAgene. On the other hand, the level of the bga42B expressionwas considerably lower, and in the case of bga42C, thetranscription was negligible. The expression of bga2A wasmarginally increased when the cells were grown in thepresence of HMOs, whereas the transcription of bga42A wasnot influenced by the type of carbon source.

Discussion

The HMO+ phenotype is a unique feature of particular infant-gut associated bifidobacteria with a few exceptions (Marcobalet al. 2010). Nevertheless, two representative HMO+ species,B. bifidum and B. longum subsp. infantis, appear tohave evolved different pathways to degrade HMOs. B. bifidumhydrolyzes LNT, the most abundant core structure, outside thecell by lacto-N-biosidase into LNB and Lac and then importsLNB by a specific ABC transporter (Suzuki et al. 2008). Lacis hydrolyzed by a secretory β-galactosidase (Miwa et al.2010), and the liberated Glc is preferentially consumed. Incontrast, B. longum subsp. infantis directly incorporated LNTand hydrolyzed it into monosaccharides inside the cells.GHs acting on β-galactosides occur in GH1, GH2, GH35,

GH42 and GH43 (Cantarel et al. 2009). B. longum subsp.infantis ATCC15697 possesses two GH2 proteins (Blon_0268and 2334), three GH42 proteins (Blon_2016, 2123 and 2416)and one GH43 protein (Blon_2411). The genes correspondingto GH1 and GH35 enzymes were not found. The Blon_2334gene is located within the HMO cluster-1, whereas the other

genes are positioned at different loci. To elucidate which gene(s) is responsible for the degradation of HMOs, we character-ized these gene products (Table I). The results revealed thatthe gene product of Blon_2016 (Bga42A) is responsible forthe degradation of type-1 HMOs, whereas the gene product ofBlon_2334 (Bga2A) is involved in the degradation of Lac andtype-2 HMOs. Given that LNT is present at the concentrationof 5 mM (Kunz et al. 2000; Urashima et al. 2011), thespecific activity (v/[E]0) of Bga2A toward LNT is calculatedto be 0.005% of that of Bga42A (Table I and Supplementarydata, Figure S3). The HMO+ phenotype of this subspeciestherefore should not be ascribed solely to the presence of theHMO cluster-1. The bga2A and bga42A genes were tran-scribed constitutively; however, this is not surprising, becausesuch “enzymatic preparedness” was previously demonstratedfor a human commensal bacterium (Sonnenburg et al. 2005).Bga42B and Bga42C should act on substrates other thanHMOs.β-Galactosidases acting on type-1 chains are rarely found in

nature. To our knowledge, only two enzymes that belong toGH35 have been isolated from Xanthomonas species (Taronet al. 1995; Yang et al. 2003). GH42 contains more than 400sequences, and several members including bifidobacterialenzymes have been characterized (Sheridan and Brenchley2000; Hung et al. 2001; Møller et al. 2001; Hinz et al. 2004;Hu et al. 2007; Di Lauro et al. 2008; Goulas et al. 2009), andone crystal structure has been determined (Hidaka et al.2002). However, most studies have focused on the synthesisof galactooligosaccharides (GOSs) using the transglycosyla-tion activity and have not examined the substrate specificitiesof the enzymes. In a few cases, high activities toward galac-tans and GOSs with β-(1→ 4)- or β-(1→ 6)-linkages weredemonstrated (Hinz et al. 2004; Goulas et al. 2009), whichled to the assumption that GH42 enzymes may be involved inplant cell wall degradation. The present study on Bga42A istherefore the first that shows the occurrence of a type-1 chain-specific enzyme in GH42, and also the first study to revealthat a GH42 enzyme is physiologically involved in the degra-dation of the human-related glycan. The high specificity ofBga42A for LNT rather than LNB indicates the presence of a(+) subsite(s) to recognize the reducing end of LNT (Gal and/or Glc residues). The proposed name for this enzyme is LNTβ-1,3-galactosidase.Based on these finding, we performed a phylogenetic

analysis using GH42 enzymes from bifidobacteria (Figure 3).The enzymes come from eight different bifidobacterialspecies/subspecies (http://www.cazy.org/GH42_bacteria.html;Cantarel et al. 2009), and the sequence identities vary from20 to 100%. Bga42A, Bga42B and Bga42C fall into distantclades that reflect their different substrate specificities.Interestingly, the close homologs of Bga42A (aa identity>95%) that constitute one clade are found to come frominfant gut-related species, i.e. B. breve, B. longum subsp.infantis and B. longum subsp. longum, all of which areknown to consume LNT (Ward et al. 2007). The conservationof the type-1-specific β-galactosidase together with the predo-minance of the type-1 chain in HMOs lead us to postulate thecoevolution of these organisms with humans. B. bifidum andsome strains of B. longum subsp. longum may have acquired

Fig. 2. Real-time quantitative RT–PCR analysis of the four β-galactosidasegenes of B. longum subsp. infantis. Total RNAs were isolated from the cellsgrown in the HMOs (white bars) and Glc (gray bars) containing media.Relative amounts of the transcripts are presented as the percentages of thoseof the dnaA gene, and the values represent the mean ± standard deviations.


365

Dow





http://www.cazy.org/GH42_bacteria.html



lacto-N-biosidases alternatively to process type-1 HMOs(Wada et al. 2008).

Materials and methodsChemicalsGal and Glc were purchased from Nacalai tesque (Kyoto,Japan). LacNAc, LNnT and LNT were obtained from DextaLaboratories (Reading, UK). pNP-glycosides, GlcNAc, Lac,verapamil hydrochloride and CCCP were from Wako PureChemical Industries (Osaka, Japan). LNB was synthesized asdescribed previously (Nishimoto and Kitaoka 2007b).

Bacterial strains and growth conditionsB. longum subsp. infantis ATCC15697 and B. bifidumJCM1254 were obtained from the American Type CultureCollection and the Japan Collection of Microorganisms,respectively. The strains were anaerobically grown at 37°C inGAM medium (Nissui, Tokyo, Japan) or in basal medium(Wada et al. 2008). The basal medium was supplementedwith a 1% carbon source (Glc or HMOs) prior to inoculation.Preparation of HMOs from human milk was performed asdescribed previously (Sumiyoshi et al. 2003) with slightmodifications (S Asakuma, T Urashima, M Kitaoka and TKatayama, in preparation).

Transport assayThe bifidobacteria were grown in the GAM media, harvestedby centrifugation and suspended in the basal medium contain-ing 1 mM LNT as a carbon source. The suspensions wereincubated at 37°C in the presence and the absence of 5 mMverapamil or 50 μM CCCP. Samples were taken at the indi-cated times and the reaction products were analyzed by TLCusing a silica gel 60 aluminum sheet (Merck, Darmstadt,Germany). The plate was developed in a solvent system of1-buthanol/acetic acid/water (2/1/1). Sugars were visualized asdescribed previously (Anderson et al. 2000).

Degradation of LNT by the cell-free extract of B. longumsubsp. infantisThe cells grown in the GAM medium were suspended in50 mM sodium phosphate buffer (pH 7.0) and disrupted bysonication. The cell-free extract was incubated with 5 mMLNT at 37°C and the reaction products were analyzed by TLCas described in the previous paragraph (Transport assay).

Enzyme preparationThe locus tag Blon_0268, 2016, 2123, 2334, 2411 and 2416genes were amplified by PCR involving PrimeSTAR Maxpolymerase (Takara Bio, Shiga, Japan) using the genomicDNA of B. longum subsp. infantis as a template and theprimer pairs listed in Supplementary data, Table S1. Theamplified fragments were inserted into the pET23b vector to

Fig. 3. Phylogenetic analysis of GH42 enzymes from bifidobacteria. The tree was created using Clustal W (http://clustalw.ddbj.nig.ac.jp/top-j.html) and MEGA 4(Tamura et al. 2007). The GenBank ID and species/subspecies names (ad, B. adolescentis; bi, B. bifidum; br, B. breve; de, B. dentium; in, B. longum subsp.infantis; la, B. animalis subsp. lactis; lo, B. longum subsp. longum; ps. B. pseudocatenulatum) are indicated. Bga42A, Bga42B and Bga42C are indicated bybold letters. A branch including Bga42A is enclosed by a dashed oval.

E Yoshida et al.

366

Dow








generate C-terminally histidine-tagged proteins. After the con-firmation of the nucleotide sequences, the plasmids wereintroduced into a ΔlacZ derivative of E. coli BL21 (DE3)(Miwa et al. 2010). Expression and purification of the proteinsusing Ni-nitrilotriacetic-acid affinity chromatography wereperformed according to the manufacturer’s instructions(Qiagen, MD). The proteins were further purified using aSuperdex 200 10/300 GL column (GE HealthcareBio-Sciences, Uppsala, Sweden). The elution was performedusing 20 mM Tris–HCl buffer (pH 8.0) containing 150 mMNaCl and 0.05% Tween-20. The molecular masses of the pro-teins were determined using the same column. Thyroglobulin(669 kDa), ferritin (443 kDa), aldolase (158 kDa), conalbumin(75 kDa) and ovalbumin (43 kDa) were used as the standards.The purified proteins were dialyzed against 20 mM sodiumphosphate buffer (pH 7.0) containing 0.05% Tween-20. Theprotein concentrations were determined using a BCA proteinassay kit (Thermo Fisher Scientific, MA) with bovine serumalbumin as the standard.

Enzyme assayAssays were performed in 50 mM sodium phosphate buffer(pH 6.0) containing 0.05% Tween-20 in a total volume of300 μL. The substrates used were pNP-N-acetyl-β-D-galactosa-minide (pNP-β-GalNAc), pNP-N-acetyl-β-D-glucosaminide(pNP-β-GlcNAc), pNP-α-L-arabinofuranoside (pNP-α-Ara),pNP-β-Fuc, pNP-β-Gal, pNP-β-D-glucopyranoside (pNP-β-Glc), pNP-β-D-glucuronide (pNP-β-GlcA) and pNP-β-D-xylo-pyranoside (pNP-β-Xyl). The reaction was initiated by addingthe enzyme, and the mixtures were incubated for 5–120 min,depending on the substrates used. The reaction was stopped byadding 300 μL of 1 M Na2CO3. The amounts of releasedp-nitrophenol were determined by measuring the absorbance at405 nm. When Lac, LacNAc, LNB, LNnT and LNT were usedas the substrates, the reactions were carried out in a totalvolume of 100 μL and terminated by heating (90°C, 5 min),under conditions that no chemical decomposition of LNB wasobserved (Chiku et al. 2010). The amounts of Gal releasedwere determined using a galactose dehydrogenase-neocuproinemethod (Cohenford et al. 1989). Initial velocity of hydrolysiswas determined in the range where the linearity of the reactionrate was observed. The kinetic parameters were calculated bycurve fitting the experimental data with the Michaelis–Mentenequation, using KaleidaGraph 4.0 (Synergy Software). In thecase that no saturation was observed, only the kcat/Km valuewas determined at the low substrate concentration.The temperature and pH stabilities were determined by

measuring the residual pNP-Gal-hydrolyzing activities afterincubating the enzymes (1 μM) in 50 mM sodium phosphatebuffer (pH 6.0) for 30 min at various temperatures and afterdialyzing the enzyme overnight against various 10 mMbuffers at 4°C, respectively. The buffers used were citratephosphate (pH 3.5–6.5), sodium phosphate (pH 5.5–8.0) andTris–HCl (pH 7.5–9.0). The optimal pHs for hydrolyzingpNP-Gal (0.5 mM) were determined using the same buffers(10 mM). Note that 1 mM MgCl2 was added to the reactionmixture for assaying the activity of Bga2A throughoutthe study.

Real-time quantitative RT–PCRReal-time quantitative RT–PCR was carried out for the fourβ-galactosidase genes (Blon_2016, 2123, 2334 and 2416).The cells were grown in the basal medium supplemented with1% Glc or HMOs to the mid-exponential phase and the totalRNAs were isolated using a RiboPure-Bacteria kit (AppliedBiosystems, CA). Purified RNA (120 ng) was reverse-transcribed using the High Capacity cDNA ReverseTranscription kit (Applied Biosystems), and the reaction pro-ducts were directly used for the TaqMan-based assay. Theserial dilutions of the genomic DNA were used to constructthe standard curves for the respective genes. Assays were per-formed in duplicate using RNAs isolated from two separatecultures and the relative amounts of the transcripts were pre-sented as mean ± standard deviations. The dnaA gene wasused as the reference gene. The primers and probes weredesigned using the Primer Express software (AppliedBiosystems; Supplementary data, Table S2).

Ethical considerationThis study was approved by the Ethics Committee of theUniversity of Shiga Prefecture and followed the Declarationof Helsinki. The informed consents were obtained from allmilk donors.

Supplementary data

Supplementary data for this article is available online athttp://glycob.oxfordjournals.org.

Funding

This work was supported in part by a Grant-in-aid forScientific Research by Young Scientists (B) 22780072 fromthe Ministry of Education, Culture, Sports, Science andTechnology, Japan, and a Grant-in-aid from the UrakamiFoundation.

Conflict of interest

None declared.

Abbreviations

aa, amino acid; ABC, ATP-binding cassette; CAZy,Carbohydrate-Active enZYmes; CCCP, carbonyl cyanide m-chlorophenyl hydrazone; Gal, galactose; GH, glycoside hydro-lase; Glc, glucose; GlcNAc, N-acetylglucosamine; GNB,galacto-N-biose; GOS, galactooligosaccharide; HMO, humanmilk oligosaccharide; Lac, lactose; LacNAc, N-acetyllactosa-mine; LNB, lacto-N-biose I; LNnT, lacto-N-neotetraose; LNT,lacto-N-tetraose; PCR, polymerase chain reaction; pNP-α-Ara,pNP-α-L-arabinofuranoside; pNP-β-Fuc, pNP-β-D-fucopyrano-side; pNP-β-Gal, 4-nitrophenyl-β-D-galactopyranoside; pNP-β-GalNAc, pNP-N-acetyl-β-D-galactosaminide; pNP-β-Glc, pNP-β-D-glucopyranoside; pNP-β-GlcA, pNP-β-D-glucuronide;


367

Dow





http://glycob.oxfordjournals.org



pNP-β-GlcNAc, pNP-N-acetyl-β-D-glucosaminide; pNP-β-Xyl,pNP-β-D-xylopyranoside; RT, reverse transcription; SDS–PAGE, sodium dodecyl sulfate–polyacrylamide gel electro-phoresis; TLC, thin-layer chromatography.

References

Anderson K, Li SC, Li YT. 2000. Diphenylamine-aniline-phosphoric acidreagent, a versatile spray reagent for revealing glycoconjugates on thin-layer chromatography plates. Anal Biochem. 287:337–339.

Benno Y, Sawada K, Mitsuoka T. 1984. The intestinal microflora of infants:Composition of fecal flora in breast-fed and bottle-fed infants. MicrobiolImmunol. 28:975–986.

Bode L. 2006. Recent advances on structure, metabolism, and function ofhuman milk oligosaccharides. J Nutr. 136:2127–2130.

Cantarel BL, Coutinho PM, Rancurel C, Bernard T, Lombard V, Henrissat B.2009. The Carbohydrate-Active EnZymes database (CAZy): An expertresource for glycogenomics. Nucleic Acids Res. 37:D233–D238.

Chiku K, Nishimoto M, Kitaoka M. 2010. Thermal decomposition ofβ-D-galactopyranosyl-(1→ 3)-2-acetamido-2-deoxy-D-hexopyranoses underneutral conditions. Carbohydr Res. 345:1901–1908.

Cohenford MA, Abraham A, Abraham J, Dain JA. 1989. Colorimetric assayfor free and bound L-fucose. Anal Biochem. 177:172–177.

Di Lauro B, Strazzulli A, Perugino G, La Cara F, Bedini E, Corsaro MM,Rossi M, Moracci M. 2008. Isolation and characterization of a new family42 β-galactosidase from the thermoacidophilic bacterium Alicyclobacillusacidocaldarius: Identification of the active site residues. Biochim BiophysActa. 1784:292–301.

Finn RD, Mistry J, Tate J, Coggill P, Heger A, Pollington JE, Gavin OL,Gunasekaran P, Ceric G, Forslund K, et al. 2010. The Pfam proteinfamilies database. Nucleic Acids Res. 38:D211–D222.

Garrido D, Kim JH, German JB, Raybould HE, Mills DA. 2011.Oligosaccharide binding proteins from Bifidobacterium longum subsp.infantis reveal a preference for host glycans. PLoS One. 6:e17315.

Gnoth MJ, Kunz C, Kinne-Saffran E, Rudloff S. 2000. Human milk oligosac-charides are minimally digested in vitro. J Nutr. 130:3014–3020.

Goulas T, Goulas A, Tzortzis G, Gibson GR. 2009. Comparative analysis of fourβ-galactosidases from Bifidobacterium bifidum NCIMB41171: Purificationand biochemical characterization. Appl Microbiol Biotechnol. 82:1079–1088.

Hidaka M, Fushinobu S, Ohtsu N, Motoshima H, Matsuzawa H, Shoun H,Wakagi T. 2002. Trimeric crystal structure of the glycoside hydrolasefamily 42 β-galactosidase from Thermus thermophilus A4 and the structureof its complex with galactose. J Mol Biol. 322:79–91.

Hidaka M, Nishimoto M, Kitaoka M, Wakagi T, Shoun H, Fushinobu S. 2009.The crystal structure of galacto-N-biose/lacto-N-biose I phosphorylase: Alarge deformation of a TIM barrel scaffold. J Biol Chem. 284:7273–7283.

Hinz SW, van den Brock LA, Beldman G, Vincken JP, Voragen AG. 2004.β-Galactosidase from Bifidobacterium adolescentis DSM20083 prefersβ(1,4)-galactosides over lactose. Appl Microbiol Biotechnol. 66:276–284.

Hu JM, Li H, Cao LX, Wu PC, Zhang CT, Sang SL, Zhang XY, Chen MJ,Lu JQ, Liu YH. 2007. Molecular cloning and characterization of the geneencoding cold-active β-galactosidase from a psychrotrophic and halotoler-ant Planococcus sp. L4. J Agric Food Chem. 55:2217–2224.

Hung MN, Xia Z, Hu NT, Lee BH. 2001. Molecular and biochemical analysisof two β-galactosidases from Bifidobacterium infantis HL96. Appl EnvironMicrobiol. 67:4256–4263.

Kitaoka M, Tian J, Nishimoto M. 2005. Novel putative galactose operoninvolving lacto-N-biose phosphorylase in Bifidobacterium longum. ApplEnviron Microbiol. 71:3158–3162.

Kobata A. 2010. Structures and application of oligosaccharides in humanmilk. Proc Jpn Acad Ser B. 86:731–747.

Kunz C, Rudloff S, Baier W, Klein N, Strobel S. 2000. Oligosaccharides inhuman milk: Structural, functional, and metabolic aspects. Annu Rev Nutr.20:699–722.

Li ZS, Szczypka M, Lu YP, Thiele DJ, Rea PA. 1996. The yeast cadmiumfactor protein (YCF1) is a vacuolar glutathione S-conjugate pump. J BiolChem. 271:6509–6517.

LoCascio RG, Desai P, Sela DA, Weimer B, Mills DA. 2010. Broad conser-vation of milk utilization genes in Bifidobacterium longum subsp. infantisas revealed by comparative genomic hybridization. Appl EnvironMicrobiol. 76:7373–7381.

Marcobal A, Barboza M, Froehlich JW, Block DE, German JB, Lebrilla CB,Mills DA. 2010. Consumption of human milk oligosaccharides bygut-related microbes. J Agric Food Chem. 58:5334–5340.

Miwa M, Horimoto T, Kiyohara M, Katayama T, Kitaoka M, Ashida H,Yamamoto K. 2010. Cooperation of β-galactosidase andβ-N-acetylhexosaminidase from bifidobacteria in assimilation of humanmilk oligosaccharides with type 2 structure. Glycobiology. 20:1402–1409.

Møller PL, Jørgensen F, Hansen OC, Madsen SM, Stougaard P. 2001. Intra-and extracellular β-galactosidases from Bifidobacterium bifidum and B.infantis: Molecular cloning, heterologous expression, and comparativecharacterization. Appl Environ Microbiol. 67:2276–2283.

Morrow AL, Ruiz-Palacios GM, Altaye M, Jiang X, Guerrero ML,Meinzen-Derr JK, Farkas T, Chaturvedi P, Pickering LK, Newburg DS.2004. Human milk oligosaccharides are associated with protection againstdiarrhea in breast-fed infants. J Pediatr. 145:297–303.

Newburg DS, Neubauer SH. 1995. Carbohydrates in milk. In: Jensen RG, editor.Handbook of Milk Composition. San Diego (USA): Academic Press. p. 273–349.

Nishimoto M, Kitaoka M. 2007a. Identification of N-acetylhexosamine1-kinase in the complete lacto-N-biose I/galacto-N-biose metabolic pathwayin Bifidobacterium longum. Appl Environ Microbiol. 73:6444–6449.

Nishimoto M, Kitaoka M. 2007b. Practical preparation of lacto-N-biose I, acandidate for the bifidus factor in human milk. Biosci Biotechnol Biochem.71:2101–2104.

Penders J, Thijs C, Vink C, Stelma FF, Snijders B, Kummeling I, van denBrandt PA, Stobberingh EE. 2006. Factors influencing the composition ofthe intestinal microbiota in early infancy. Pediatrics. 118:511–521.

Sela DA, Chapman J, Adeuya A, Kim JH, Chen F, Whitehead TR, LapidusA, Rokhsar DS, Lebrilla CB, German JB, et al. 2008. The genomesequence of Bifidobacterium longum subsp. infantis reveals adaptations formilk utilization within the infant microbiome. Proc Natl Acad Sci USA.105:18964–18969.

Sheridan PP, Brenchley JE. 2000. Characterization of a salt-tolerant family 42β-galactosidase from a psychrophilic antarctic Planococcus isolate. ApplEnviron Microbiol. 66:2438–2444.

Sonnenburg JL, Xu J, Leip DD, Chen CH, Westover BP, Weatherford J,Buhler JD, Gordon JI. 2005. Glycan foraging in vivo by anintestine-adapted bacterial symbiont. Science. 307:1955–1959.

Sumiyoshi W, Urashima T, Nakamura T, Arai I, Saito T, Tsumura N, Wang B,Brand-Miller J, Watanabe Y, Kimura K. 2003. Determination of eachneutral oligosaccharide in the milk of Japanese women during the courseof lactation. Br J Nutr. 89:61–69.

Suzuki R, Wada J, Katayama T, Fushinobu S, Wakagi T, Shoun H, SugimotoH, Tanaka A, Kumagai H, Ashida H, et al. 2008. Structural and thermo-dynamic analyses of solute-binding protein from Bifidobacterium longumspecific for core 1 disaccharide and lacto-N-biose I. J Biol Chem.283:13165–13173.

Tamura K, Dudley J, Nei M, Kumar S. 2007. MEGA4: Molecular evolution-ary genetics analysis (MEGA) software version 4.0. Mol Biol Evol.24:1596–1599.

Taron CH, Benner JS, Hornstra LJ, Guthrie EP. 1995. A novelβ-galactosidase gene isolated from the bacterium Xanthomonas maniho-tis exhibits strong homology to several eukaryotic β-galactosidases.Glycobiology. 5:603–610.

Turroni F, Bottacini F, Foroni E, Mulder I, Kim JH, Zomer A, SánchezB, Bidossi A, Ferrarini A, Giubellini V, et al. 2010. Genome analysisof Bifidobacterium bifidum PRL2010 reveals metabolic pathways forhost-derived glycan foraging. Proc Natl Acad Sci USA.107:19514–19519.

Urashima T, Kitaoka M, Terabayashi T, Fukuda K, Ohnishi M, Kobata A.2011. Milk oligosaccharides. In: Gordon NG, editor. Oligosaccharides:Sources, Properties and Applications. New York (USA): Nova SciencePublishers. p. 1–77.

Wada J, Ando T, Kiyohara M, Ashida H, Kitaoka M, Yamaguchi M, KumagaiH, Katayama T, Yamamoto K. 2008. Bifidobacterium bifidum lacto-N-biosi-dase, a critical enzyme for the degradation of human milk oligosaccharideswith a type 1 structure. Appl Environ Microbiol. 74:3996–4004.

Ward RE, Niñonuevo M, Mills DA, Lebrilla CB, German JB. 2007. In vitrofermentability of human milk oligosaccharides by several strains of bifido-bacteria. Mol Nutr Food Res. 51:1398–1405.

Yang TC, Hu RM, Hsiao YM, Weng SF, Tseng YH. 2003. Molecular geneticanalyses of potential β-galactosidase genes in Xanthomonas campestris. JMol Microbiol Biotechnol. 6:145–154.

E Yoshida et al.

368

Dow



bifidobacterium longum subsp. infantis uses two - glycobiology

Documents