1 Cross-feeding between Bifidobacterium infantis and caccae on

2 lactose and human milk oligosaccharides

3

4 Loo Wee Chia1, Marko Mank2, Bernadet Blijenberg2, Roger S. Bongers2, Steven

5 Aalvink1, Kees van Limpt2, Harm Wopereis1,2, Sebastian Tims2, Bernd Stahl2, Clara

6 Belzer1*#, Jan Knol1,2*

7 * these authors contributed equally

8 1 Laboratory of Microbiology, Wageningen University and Research, Stippeneng 4,

9 6708 WE Wageningen, the Netherlands.

10 2 Nutricia Research, Uppsalalaan 12, 3584 CT Utrecht, the Netherlands.

11

12 Running Head: Microbial cross-feeding in infant gut

13

14 # Address correspondence to [email protected].

15

16 Conflict of Interest statement:

17 This project is financially supported by Nutricia Research. MM, BB, RB, HW, KvL, ST,

18 BS and JK are employed by Nutricia Research.

1

19 Abstract

20 The establishment of the immediately after birth is a dynamic process

21 that may impact lifelong health. At this important developmental stage in early life,

22 human milk oligosaccharides (HMOS) serve as specific substrates to promote the

23 growth of gut microbes, particularly the group of Actinobacteria (bifidobacteria). Later

24 in life, this shifts to the colonisation of and Bacteroidetes, which generally

25 dominate the human gut throughout adulthood. The well-orchestrated transition is

26 important for health, as an aberrant microbial composition and/or SCFA production

27 are associated with colicky symptoms and atopic diseases in infants. Here, we study

28 the trophic interactions between an HMOS-degrader, Bifidobacterium longum subsp.

29 infantis and the butyrogenic using carbohydrate substrates that

30 are relevant in this early life period, i.e. lactose and HMOS. Mono- and co-cultures of

31 these bacterial species were grown at pH 6.5 in anaerobic bioreactors supplemented

32 with lactose or total human milk carbohydrates (containing both lactose and HMOS).

33 A. cac was not able to grow on these substrates except when grown in co-culture

34 with B. inf, leading concomitant butyrate production. Cross-feeding was observed, in

35 which A. cac utilised the liberated monosaccharides as well as lactate and acetate

36 produced by B. inf. This microbial cross-feeding is indicative of the key ecological role

37 of bifidobacteria in providing substrates for other important species to colonise the

38 infant gut. The symbiotic relationship between these key species contributes to the

39 gradual production of butyrate early in life that could be important for host-microbial

40 cross-talk and gut maturation.

41

2

42 Importance

43 The establishment of a healthy infant gut microbiota is crucial for the immune,

44 metabolic and neurological development of infants. Recent evidence suggests that

45 an aberrant gut microbiota early in life could lead to discomfort and predispose

46 infants to the development of immune related diseases. This paper addresses the

47 ecosystem function of two resident microbes of the infant gut. The significance of this

48 research is the proof of cross-feeding interactions between HMOS-degrading

49 bifidobacteria and a butyrate-producing microorganism. Bifidobacteria in the infant

50 gut that support the growth and butyrogenesis of butyrate-producing , could

51 orchestrated an important event of maturation for both the gut ecosystem and

52 physiology of infant.

53

54 Keywords

55 Bifidobacteria, butyrate, , microbiome, pH

56

3

57 Introduction

58 The succession of microbial species in the infant gut microbiota is a profound

59 process in early life (1, 2), which coincides with the important development of the

60 immune, metabolic and neurological systems (3-5). At this developmental stage,

61 human milk is recognised as the best nourishment for infants (6). Human milk

62 contains a range of microbial active components and among all human milk

63 oligosaccharides (HMOS) have a vital role in the development of the infant gut

64 microbiota (7). HMOS are complex carbohydrates composed of a lactose core, which

65 may be elongated by N-acetylglucosamine (GlcNAc), galactose and/or decorated

66 with fucose and/or sialic acid residues (8). The composition of HMOS in human milk

67 is highly individual and driven by maternal genetic factors and varies with the phases

68 of lactation (9).

69 The majority of the HMOS escape digestion by the host’s enzymes in the

70 upper gastrointestinal tract (10). HMOS confer important physiological traits by acting

71 both as a decoy for the binding of pathogenic bacteria and viruses, and as a prebiotic

72 to stimulate the growth and activity of specific microbes in the infant gut (11). These

73 complex carbohydrates exert therefore a selective nutrient pressure to promote the

74 HMOS-utilising microbes, especially bifidobacteria belonging to the Actinobacteria

75 phylum (12). Bifidobacteria are specifically adapted to utilise HMOS by employing an

76 extensive range of glycosyl hydrolases and transporters, which leads to their

77 dominance in the infant gut (13). Upon weaning, the relative abundance of

78 bifidobacteria decreases with the increase of Firmicutes and Bacteroidetes phyla,

79 whilst the gut microbial diversity increases (14).

80 The early dominance of bifidobacteria could be important for the maturation of

81 the overall microbial community. In healthy children, the relative abundance of 4

82 bifidobacteria is positively associated with the butyrate-producing Firmicutes from the

83 family of Lachnospiraceae (also known as Clostridium cluster XIVa) and

84 Ruminococcaceae (also known as Clostridium cluster IV) (15). This butyrogenic

85 community often presents at a much lower relative abundance in the gut of new-

86 borns (16). The subdominant butyrogenic species could however quickly become

87 more dominant upon weaning as a result of the introduction of solid food and the

88 cessation of breast-feeding (2, 17). The colonisation by the strict anaerobic, butyrate-

89 producing bacteria could be a critical step for the gut and immune maturation (18, 19).

90 The interactions between lactate-producing bacteria (such as bifidobacteria) and

91 lactate-utilising bacteria (such as Ruminococcaceae and Lachnospiraceae) are

92 suggested to be associated with a lower risk of colicky symptoms and atopic disease

93 in infants (18-21). To date, cross-feeding between glycan-degrading bifidobacteria

94 and butyrate-producers using complex dietary carbohydrates (including starch, inulin,

95 fructo-oligosaccharides, and arabinoxylan oligosaccharides) has been demonstrated

96 in in vitro co-culturing experiments (22-26). However, limited studies have shown the

97 cross-feeding between these groups of bacteria on host-secreted glycans such as

98 HMOS (27) and mucins (28).

99 In this study, we investigated the trophic interaction between an HMOS-

100 degrader, Bifidobacterium longum subsp. infantis and a butyrogenic non-degrader of

101 human milk carbohydrates. To this end the butyrate-producer Anaerostipes caccae

102 was used as the representative species for the Lachnospiraceae family as it is

103 detected in the early life gut microbiota (2, 29) and is one of the prevalent members

104 of the gut microbiota in human adults (30). We show that B. inf supports the

105 development of the microbial ecosystem by metabolising lactose and HMOS into

106 monosaccharides and short chain fatty acid (SCFA) including lactate and acetate, to

5

107 support the growth and concomitant butyrate production by A. cac. This butyrogenic

108 cross-feeding demonstrates the importance of bifidobacteria in the establishment of a

109 healthy microbial ecosystem in early life.

110

111 Results

112 The occurrence of B. inf and A. cac across the life span

113 A published dataset (29) was mined for the occurrence of B. inf and A. cac in the

114 microbiota across life stages. The two infant-associated bacteria demonstrated

115 opposite trajectories in early life. Bifidobacterium genus showed high abundance at

116 the first year followed by a sharp decline, with a negative correlation between age

117 and relative abundance (Spearman ρ = -0.38, p < 0.05) (Fig. 1). On the contrary,

118 Anaerostipes genus (Spearman ρ = 0.56, p < 0.05) and Lachnospiraceae family

119 (Spearman ρ = 0.37, p < 0.05) were present at low abundance early in life and

120 increased in relative abundance in the first 1000 days of life (Fig. 1).

121

122 Model for B. inf and A. cac co-occurrence

123 Bacterial strains were cultured in anaerobic bioreactors controlled at pH 6.5 and

124 supplemented with either lactose or total human milk (HM) carbohydrates. B. inf

125 monoculture reached maximal cell density around 12 h (ODmax = 1.40 ± 0.38 in

126 lactose and ODmax = 1.37 ± 0.25 in total HM carbohydrates) (Fig. 2). For A. cac

127 monoculture, no growth or substrate degradation was detected in identical media

128 (ODmax = 0.02 ± 0.01 in lactose and ODmax = 0.03 ± 0.02 in total HM carbohydrates)

129 (Table S1). The co-culture of B. inf with A. cac grew rapidly reaching maximal optical

6

130 density at 11 h in lactose (ODmax = 3.63 ± 0.61) and at 9 h in total HM carbohydrates

131 (ODmax = 3.54 ± 0.60). The community dynamic in the co-cultures was monitored

132 over time by qPCR. An equal amount of B. inf and A. cac (around 106 copy

133 number/ml) was inoculated at the start of the fermentation. During the first 7 h, B. inf

134 and A. cac increased 100-fold based on the increase of 16S rRNA gene copy

135 number, after which growth slowed down. FISH analysis of samples harvested at 11

136 h showed B. inf to A. cac ratio of 1:6. This was observed for both conditions either in

137 lactose or total HM carbohydrates supplemented cultures.

138

139 B. inf supported the growth and metabolism of A. cac in lactose and HMOS

140 The substrate consumption and SCFA production were monitored over time (Fig. 3).

141 A similar profile was observed between the fermentation of lactose and total HM

142 carbohydrates, probably because total HM carbohydrates consists of approximately

143 10% HMOS and 90% lactose (Fig. S1). On both substrates, the monoculture of B. inf

144 degraded the lactose into glucose and galactose resulting in the accumulation of

145 monomeric sugars in the supernatant (Fig. 3). Lactose was completely degraded at 9

146 h. At the same time point, 17.49 ± 1.83 mM of glucose and 15.24 ± 2.06 mM of

147 galactose were detected in the media supplemented with lactose, whereas 14.77 ±

148 1.59 mM of glucose and 10.91 ± 1.77 mM of galactose were detected in the media

149 supplemented with total HM carbohydrates. The monomeric sugars were fully

150 consumed after 31 h. B. inf produced acetate (56.96 ± 4.48 mM in lactose and 50.76

151 ± 3.23 mM in total HM carbohydrates), lactate (22.73 ± 3.02 mM in lactose and 17.69

152 ± 1.21 mM in total HM carbohydrates) and formate (6.56 ± 0.09 mM in lactose and

153 8.04 ± 0.21 mM in total HM carbohydrates) as the major end metabolites. The final

7

154 acetate to lactate ratio for B. inf in lactose was 2.4:1 and 2.6:1 in total HM

155 carbohydrates.

156 The co-culture of B. inf with A. cac also degraded lactose completely within 9 h.

157 However, the co-cultures depleted glucose and galactose faster compared to the

158 monocultures of B. inf. The concentration of monomeric sugars peaked around 7 h in

159 media supplemented with lactose, with 4.62 ± 1.21 mM glucose and 7.10 ± 0.97 mM

160 galactose. In media supplemented with the total HM carbohydrates, the maximum

161 concentration for glucose (4.20 ± 2.10 mM) and galactose (7.39 ± 4.45 mM) was

162 detected after 5 h. Only traces of monomeric sugars were still detectable after 9 h.

163 The major end products of fermentation in the co-cultures were butyrate, a signature

164 metabolic end product of A. cac (31.39 ± 2.15 mM in lactose and 25.80 ± 2.45 mM in

165 total HM carbohydrates), acetate (5.44 ± 0.30 mM in lactose and 9.05 ± 0.71 mM in

166 total HM carbohydrates) and formate (2.53 ± 0.16 mM in lactose and 4.78 ± 1.16 mM

167 in total HM carbohydrates). In contrast to the B. inf monocultures, no lactate was

168 detected after 11 h in the co-cultures.

169 The low molecular weight HMOS structures in the total HM carbohydrates

170 were determined by ESI-LC-MS for 0 h and 24 h cultures in order to monitor the

171 specific glycan utilisation by these bacteria (Fig. 4). The monoculture of B. inf

172 completely degraded the full range of neutral trioses (including 2’-fucosyllactose [2’-

173 FL] and 3-fucosyllactose [3-FL]), tetraoses (including difucosyllactose [DFL], lacto-N-

174 tetraose [LNT], lacto-N-neotetraose [LNnT]), pentaoses (lacto-N-fucopentaose I

175 [LNFP I], lacto-N-fucopentaose II [LNFP II], lacto-N-fucopentaose III [LNFP III], lacto-

176 N-fucopentaose V [LNFP V]), and acidic trioses (including 3’-sialyllactose [3’-SL] and

177 6’-sialyllactose [6’-SL]). No degradation of HMOS was observed in the A. cac

178 monoculture. On the other hand, the glycan utilisation pattern in the co-culture was

8

179 nearly identical to the profile of B. inf monoculture, indicative of the primary degrader

180 role of B. inf in the co-cultures. Specific HMOS-derived sugars such as GlcNAc and

181 fucose were not detected likely because these were below the detection limit of 0.5

182 mM.

183

184 Microbial cross-feeding results in a shift of SCFA pool

185 The cultures were maintained at pH 6.5 with the addition of 2 M NaOH. B. inf

186 monocultures required a higher amount of base addition compared to the co-culture

187 with A. cac (Fig. 5a). The acidification of the cultures was reflected in the composition

188 of SCFAs. The total amount of SCFAs at 31 h was higher in the monocultures (86.76

189 ± 7.78 mM in lactose and 76.75 ± 3.86 mM in total HM carbohydrates) in comparison

190 to the co-cultures (39.36 ± 1.68 mM in lactose and 39.88 ± 3.97 mM in total HM

191 carbohydrates). Furthermore, as a result of microbial cross-feeding in the co-cultures,

192 lactate (pKa = 3.86) produced by B. inf monocultures was converted to butyrate (pKa

193 = 4.82) by A. cac. The pKa value indicates the quantitative measurement of the

194 strength of an acid in the solution with lower values for stronger acid. As the pKa

195 values are expressed in log scale, the decrease by one numerical value in lactate

196 compared to butyrate may result in a 10-fold increase of soluble protons. To

197 investigate the dynamics of pH in early life, the data from Wopereis et al. (19) was

198 employed. We observed that the faecal pH for infants (n=138) increased from pH 5.7

199 at 4 weeks to pH 6.0 at 6 months of life (Fig. 5b).

200

201 Discussion

9

202 The infant gut ecosystem is highly dynamic and marked by the succession of

203 bacterial species (2). In this important window of growth and development, breast-

204 feeding has a central role and generally leads to the efficient colonisation of the

205 gastrointestinal tract with bifidobacteria (2). Bifidobacteria could prime the

206 development of gut barrier function and immune maturation (31), as well as play an

207 important ecological role in the development of the gut microbiota. In this study, we

208 showed that B. inf supports the metabolism and growth of another important species

209 in early life, the butyrate-producing A. cac via cross-feeding. This microbial cross-

210 feeding resulted in the shift of the SCFA pool and an increase butyrate production.

211 Physiologically, butyrate is associated with the enhancement of colonic barrier

212 function and it could regulate host immune and metabolic state by signalling through

213 G-protein-coupled receptors (GPR) and by inhibiting histone deacetylase (HDAC)

214 (32-35). Although the mechanistic evidences for butyrate are mostly generated from

215 animal and adult studies, a gradual shift in the ecosystem with slow induction of

216 butyrate could be important for the maturation of the infant gut.

217 The dominance of bifidobacteria is often observed in the infant gut microbiota

218 (36). Bifidobacteria have evolved to be competitive in utilising human milk as

219 substrate by employing a large arsenal of enzymes to metabolise HMOS (37). We

220 showed that B. inf effectively degraded the full range of the low molecular weight

221 HMOS structures including neutral trioses, tetraoses, and pentaoses as well as acidic

222 trioses. This is consistent with the unique HMOS utilisation capability of B. inf by

223 encoding a 43kb gene cluster that carries the genes for different oligosaccharides

224 transport proteins and glycosyl hydrolases (38). No signal peptide or transmembrane

225 domain was predicted for B. inf enzymes involved in the cleavage of the monitored

226 HMOS structures (Table S2), indicating intracellular degradation of these substrates.

10

227 Furthermore, the distinct “bifid shunt pathway” centred around the enzyme fructose-

228 6-phosphate phosphoketolase (F6PPK) could also account for the competitiveness of

229 bifidobacteria (37). The fermentation of sugars via F6PPK-dependent bifid shunt

230 pathway yields more energy compared to the usual glycolysis or Emden-Meyerhof

231 Parnas (EMP) pathway which could give bifidobacteria an additional advantage

232 compared to other gut bacteria (39).

233 Lactose and HMOS fermentation by bifidobacteria results in acetate and

234 lactate as major end products. In addition to bifidobacteria, other primary colonisers

235 like Lactobacillus, Streptococcus, Staphylococcus, and Enterococcus spp. also

236 contribute to lactate production in the infant gut (20). In the gut of breast-fed infants,

237 the overall digestion and fermentation leads to a relatively high concentration of

238 acetate and lactate with slightly acidic pH (40, 41). The pH of the luminal content has

239 a significant impact on the microbiota composition (42). Various bacterial groups

240 have been shown to be inhibited by a low pH, such as opportunistic pathogens

241 including Salmonella Typhimurium, Staphylococcus aureus, Escherichia coli,

242 Enterococcus faecalis, Pseudomonas aeruginosa, and Klebsiella pneumoniae (43)

243 as well as Bacteroides spp. (42, 44). In contrast, a low pH may promote butyrate

244 production and the butyrogenic community (44, 45). Given the above, the

245 circumstances in the infant gut may favour the colonisation of butyrate-producers.

246 In the first months of life butyrate levels in the faeces are generally low (40, 41)

247 and the major adult-type butyrate-producing bacteria ( and

248 Faecalibacterium spp.) remain undetectable up to 30 days postnatal (16). Data

249 mining of a published dataset showed an increase of relative abundance for

250 Lachnospiraceae family and Anaerostipes genus in the first year of life (29). The

251 majority of butyrate-producing bacteria from the family Lachnospiraceae and

11

252 Ruminococcaceae are not capable of utilising HMOS (46). For A. cac, no growth or

253 metabolism was detected in the media containing lactose and HMOS. These

254 subdominant butyrogenic bacteria in the infant gut could depend on cross-feeding

255 with species like bifidobacteria. Our results indicated that A. cac could utilise the

256 monomeric sugars and end products like acetate and lactate derived from B. inf for

257 metabolic activity and growth. A. cac is known to convert 1 mol of acetate and 2 mol

258 of lactate to yield 1.5 mol of butyrate (47). This metabolic interaction could also

259 benefit the microbial community by reducing the metabolic burden (48), shown by the

260 formation of a relatively weaker acid pool. The infant faecal pH showed an increasing

261 trend in the first 6 months of life (19). Acetate and lactate as well as a small amount

262 of propionate and butyrate can be detected in the faeces of infants (19, 41). However,

263 the typical SCFA ratio in adult faeces is around 3:1:1 for acetate, propionate and

264 butyrate respectively (49, 50). The shift of the SCFA pool goes hand in hand with the

265 transition of the gut microbiota, likely induced by dietary changes. Upon weaning, the

266 diversification of indigestible fibres due to the introduction of solid foods results in

267 conditions leading to the decrease of the relative numbers of bifidobacteria and the

268 relative increase of Lachnospiraceae, Ruminococcaceae, and Bacteroides spp. (14).

269 Although the contributing factors to the progression from a bifidobacteria-

270 dominant community to Firmicutes and Bacteroides dominant community remain

271 elusive, the well-orchestrated transition is important for health. An aberrant microbial

272 composition and/or SCFA production are associated with colicky symptoms and

273 atopic diseases in infants (18-21, 51). We demonstrate the possible role of B. inf in

274 driving the butyrogenic trophic chain by metabolising human milk carbohydrates. This

275 microbial cross-feeding is indicative of the key ecological role of infant-type

276 bifidobacteria as substrate provider for subdominant butyrate-producing bacteria. The

12

277 compromised health outcomes as a result of the aberrant transition from

278 bifidobacteria-dominant to butyrogenic microbial community highlight the importance

279 of proper developmental stages in the infant gut.

280

281 Materials and Methods

282 16S rRNA gene amplicon libraries screen. 16S rRNA gene amplicon sequencing

283 datasets published by Yatsunenko et al. (29) were downloaded from European

284 Nucleotide Archive (PRJEB3079). The sequencing data of 529 faecal samples with

285 known age of the sample donors was analysed using the Quantitative Insights Into

286 Microbial Ecology (QIIME) release version 1.9.0 package (52). Sequences with

287 mismatched primers, a mean sequence quality score <15 (five nucleotides window)

288 or ambiguous bases were discarded. In total 1,036,929,139 sequences were retained

289 with an average of 1,960,168.5 sequences per sample. The retained sequences

290 were grouped into Operational Taxonomic Units with the USEARCH algorithm (53)

291 set at 97% sequence identity and subsequently, the Ribosomal Database Project

292 Classifier (RDP) (54) was applied to assign to the representative

293 sequences by alignment to the SILVA ribosomal RNA database (release version

294 1.1.9) (55).

295 Bacterial strains and growth conditions. Bacterial pre-cultures were grown in

296 anaerobic serum bottles filled with gas phase of N2/CO2 (80/20 ratio) at 1.5 atm. Pre-

297 cultures were prepared by overnight 37ºC incubation in basal minimal medium (56)

298 containing 0.5% (w/v) tryptone (Oxoid, Basingstoke, UK), supplemented with 30mM

299 lactose (Oxoid, Basingstoke, UK) for Bifidobacterium longum subsp. infantis

300 ATCC15697; and 30 mM glucose (Sigma-Aldrich, St. Louis, USA) for Anaerostipes

13

301 caccae L1-92 (DSM 14662) (57). Growth was measured by a spectrophotometer at

302 an optical density of 600 nm (OD600) (OD600 DiluPhotometerTM, IMPLEN,

303 Germany).

304 Carbohydrate substrates. Lactose (Oxoid, Basingstoke, UK) and total human milk

305 (HM) carbohydrates were tested as the carbohydrate substrates for bacterial growth.

306 For preparation of total HM carbohydrates, a total carbohydrate mineral fraction was

307 derived from pooled human milk after protein depletion by ethanol precipitation and

308 removal of lipids by centrifugation as described by Stahl et al. (58). Deviant from this

309 workflow, no anion exchange chromatography (AEC) was used to further separate

310 neutral from acidic oligosaccharides present in the resulting total carbohydrate

311 mineral fraction. The total HM carbohydrates contained approximately 90% of lactose,

312 10% of both acidic and neutral HMOS as well as traces of monosaccharides, as

313 estimated by gel permeation chromatography (GPC) described below (Fig. S1).

314 Anaerobic bioreactor. Fermentations were conducted in eight parallel minispinner

315 bioreactors (DASGIP, Germany) with 100 ml filling volume at 37°C and a stirring rate

316 of 150 rpm. Culturing experiments were performed in autoclaved basal minimal

317 media (56) containing 0.5% (w/v) tryptone (Oxoid, Basingstoke, UK), supplemented

318 with 0.2 µM filter-sterilized lactose or total HM carbohydrates. Anaerobic condition

319 was achieved by overnight purging of anaerobic gas mixture containing 5% CO2, 5%

320 H2, and 90% N2. Overnight pre-cultures were inoculated at starting OD600 of 0.05 for

321 each bacterial strain. Online signals of pH values and oxygen levels were monitored

322 by the DASGIP control software (DASGIP, Germany). Cultures were maintained at

323 pH 6.5 by the addition of 2 M NaOH.

324 Gel permeation chromatography (GPC). Total HM carbohydrates were analysed

325 using GPC. Glycans were separated by the GPC stationary phase and eluted 14

326 according to size and charge. Neutral mono-, di-, and oligosaccharides, and acidic

327 oligosaccharides with different degree of polymerisation (DP) could be detected. HM

328 carbohydrate solution was prepared by dissolving 0.2 g/ml of total HM carbohydrates

329 in ultrapure water (Sartorius Arium Pro) containing 2% (v/v) 2-propanol at 37⁰C. 5 ml

330 of 0.2 µM filter-sterilized HM carbohydrate solution was injected for each GPC run.

331 Two connected Kronlab ECO50 columns (5×110 cm) packed with Toyopearl HW 40

332 (TOSOH BIOSCIENCE) were used. Milli-Q water was maintained at 50°C using

333 heating bath (Lauda, RE 206) for columns equilibration. Milli-Q water containing 2%

334 (v/v) of 2-propanol was used as the eluent. The flow rate of the eluent was set at 1.65

335 ml/min. Eluting glycans were monitored by refractive index detection (Shodex, RI-

336 101). The resulting chromatograms were analysed by using the Chromeleon®

337 software (ThermoScientific 6.80).

338 High-performance liquid chromatography (HPLC). For metabolites analysis, 1 ml

339 of bacterial culture was centrifuged and the supernatant was stored at -20°C until

340 HPLC analysis. Crotonate was used as the internal standard, and external standards

341 tested included lactose, glucose, galactose, N-acetylglucosamine (GlcNAc), N-

342 acetylgalactosamine (GalNAc), fucose, malate, fumarate, succinate, citrate, formate,

343 acetate, butyrate, isobutyrate, lactate, 1,2-propanediol, and propionate. Substrate

344 conversion and product formation were measured with a Spectrasystem HPLC

345 (Thermo Scientific, Breda, the Netherlands) equipped with a Hi-Plex-H column

346 (Agilent, Amstelveen, the Netherlands) for the separation of carbohydrates and

347 organic acids. A Hi-Plex-H column performs separation with diluted sulphuric acid on

348 the basis of ion-exchange ligand-exchange chromatography. Measurements were

349 conducted at a column temperature of 45°C with an eluent flow of 0.8 ml/min flow of

15

350 0.01 N sulphuric acid. Metabolites were detected by refractive index (Spectrasystem

351 RI 150, Thermo, Breda, the Netherlands).

352 HMOS extraction. HMOS were recovered from 1 ml aliquots of bacterial cultures.

353 Internal standard 1,5-α-L-arabinopentaose (Megazyme) was added, at the volume of

354 10 µl per sample to minimize pipetting error, to reach a final concentration of 0.01

355 mmol/l. The solution was diluted 1:1 with ultrapure water and centrifuged at 4,000 g

356 for 15 min at 4°C. The supernatant was filtered through 0.2 μM syringe filter followed

357 by subsequent centrifugation with a pre-washed ultra-filter (Amicon Ultra 0.5 Ultracel

358 Membrane 3 kDa device, Merck Milipore) at 14,000 g for 1 h at room temperature.

359 Finally, the filtrate was vortexed and stored at -20°C until further electrospray

360 ionisation liquid chromatography mass spectrometry (ESI-LC-MS) analysis.

361 Electrospray ionisation liquid chromatography mass spectrometry (ESI-LC-MS)

362 analysis. The identification and relative quantitation of HMOS were determined with

363 ESI-LC-MS. This method allowed the study of distinct HMOS structures differed in

364 monosaccharide sequence, glycosidic linkage or the molecular conformation.

365 Thereby even the HMOS isobaric isomers such as Lacto-N-fucopentaose (LNFP) I, II,

366 III and V could be distinguished. Micro ESI-LC-MS analysis was performed on a 1200

367 series HPLC stack (Agilent, Waldbronn, Germany) consisting of solvent tray,

368 degasser, binary pump, autosampler and DAD detector coupled to a 3200 Qtrap

369 mass spectrometer (ABSciex, USA). After HMOS extraction (see above) 5 µl of

370 HMOS extract was injected into the LC-MS system. Oligosaccharides were

371 separated by means of a 2.1x30 mm Hypercarb porous graphitized carbon (PGC)

372 column with 2.1x10 mm PGC pre-column (Thermo Scientific, USA) using water-

373 ethanol gradient for 19 min protocol. The gradient started with a ratio of 98% (v/v)

374 water and 2% (v/v) ethanol in 5 mM ammonium acetate at 0 min and ended with a

16

375 ratio of 20% (v/v) water and 80% (v/v) ethanol in 5 mM ammonium acetate at 13 min.

376 Re-equilibration was established between 13 and 19 min with 98% (v/v) water and 2%

377 (v/v) ethanol in 5 mM ammonium acetate. Eluent flow was 400 µl/min and the

378 columns were kept at 45⁰C. The LC-effluent was infused online into the mass

379 spectrometer and individual HMOS structures were analysed qualitatively and

380 quantitatively by multiple reaction monitoring (MRM) in negative ion mode. Specific

381 MRM transitions for neutral HMOS up to pentaoses and acidic HMOS up to trioses

382 were included. The spray voltage was -4500 V, declustering potential was at -44 V,

383 and collision energy was set to -29 eV. Each MRM-transition was performed for 50

384 ms. The instrument was calibrated with polypropylene glycol according the

385 instructions of the manufacturer. Unit resolution setting was used for precursor

386 selection whereas low resolution setting was used to monitor fragment ions of the

387 MRM transitions.

388 Quantitative real-time PCR (q-PCR). The abundance of B. inf and A. cac in mono-

389 and co-culture were determined by quantitative real-time PCR. Bacterial cultures

390 were harvested at 16,100 g for 10 min. DNA extractions were performed using

391 MasterPure™ Gram Positive DNA Purification Kit. The DNA concentrations were

392 determined fluorometrically (Qubit dsDNA HS assay; Invitrogen) and adjusted to 1

393 ng/μl prior to use as the template in qPCR. Primers targeting the 16S rRNA gene of

394 Bifidobacterium spp. (F-bifido 5'-CGCGTCYGGTGTGAAAG-3'; R-bifido 5'-

395 CCCCACATCCAGCATCCA-3'; 244 bp product (59)) and A. cac (OFF2555 5'-

396 GCGTAGGTGGCATGGTAAGT-3'; OFF2556 5'-CTGCACTCCAGCATGACAGT-3';

397 83 bp product (60)) were used for quantification. Standard template DNA was

398 prepared by amplifying genomic DNA of each bacterium using primer pairs of 35F

399 (5'-CCTGGCTCAGGATGAACG-3' (61)) and 1492R (5'-GGTTACCTTGTTACGACTT-

17

400 3') for B. inf; and 27F (5'-AGAGTTTGATCCTGGCTCAG-3') and 1492R for A. cac.

401 Standard curves were prepared with nine standard concentrations of 100 to 108 gene

402 copies/μl. PCRs were performed in triplicate with iQ SYBR Green Supermix (Bio-Rad)

403 in a total volume of 10 μl with primers at 500 nM in 384-well plates sealed with optical

404 sealing tape. Amplification was performed with an iCycler (Bio-Rad) with the following

405 protocol: 95°C for 10 min; 40 cycles of 95°C for 15 s, 55°C for 20 s, and 72°C for 30 s;

406 95°C for 1 min and 60°C for 1 min followed by a stepwise temperature increase from

407 60 to 95°C (at 0.5°C per 5 s) to obtain the melt curve data. Data was analysed using

408 the Bio-Rad CFX Manager 3.0.

409 Fluorescent in situ hybridization (FISH). FISH was performed as described

410 previously (62). Bacterial cultures were fixated by adding 1.5 ml of 4%

411 paraformaldehyde (PFA) to 0.5 ml of cultures followed by storage at -20°C. Working

412 stocks were prepared by harvesting bacterial cells by 5 min of 4°C centrifugation at

413 8,000 g, followed by re-suspension in ice-cold phosphate buffered saline (PBS) and

414 96% ethanol at a 1:1 (v/v) ratio. 3 μl of the PBS-ethanol working stocks were spotted

415 on 18 wells (round, 6 mm diameter) gelatine-coated microscope slides. Hybridization

416 was performed using rRNA-targeted oligonucleotide probes specific for

417 Bifidobacterium genus (Bif164m 5'-CATCCGGYATTACCACCC -3' [5']Cy3) (63). 10

418 µl of hybridization mixture containing 1 volume of 10 μM probe and 9 volumes of

419 hybridization buffer (20 mM Tris–HCl, 0.9 M NaCl, 0.1% SDS, pH 7.2 – pH 7.4) was

420 applied on each well. The slides were hybridized for at least 3 h in a moist chamber

421 at 50°C; followed by 30 min incubation in washing buffer (20 mM Tris–HCl, 0.9 M

422 NaCl, pH 7.2 – pH 7.4) at 50°C for washing. The slides were rinsed briefly with Milli-

423 Q water and air-dried. Slides were stained with 4,6-diamine-2-phenylindole

424 dihydrochloride (DAPI) mix containing 200 μl of PBS and 1 μl of DAPI-dye at 100

18

425 ng/μl, for 5 min in the dark at room temperature followed by Milli-Q rinsing and air-

426 drying. The slides were then covered with Citifluor AF1 and a coverslip. The slides

427 were enumerated using an Olympus MT ARC/HG epifluorescence microscope. A

428 total of 25 positions per well were automatically captured in two colour channels (Cy3

429 and DAPI) using a quadruple band filter. Images were analysed using Olympus

430 ScanR Analysis software.

431 Carbohydrate-active enzymes (CAZymes) prediction. CAZymes were predicted

432 with dbCAN version 3.0 (64), transmembrane domains with TMHMM version 2.0c (65)

433 and signal peptides with signalP 4.1 (66).

434

435 Author contributions

436 LWC, BS, JK, and CB contributed in conception. LWC, MM, RB, KvL, JK and CB

437 contributed in experimental design. LWC, BB and SA performed experiments and

438 analysed data. LWC and MM wrote the manuscript. LWC, MM, BB, RB, SA, KvL, HW,

439 ST, BS, JK and CB interpreted data and revised manuscript.

440

441 Acknowledgements

442 We thank Heleen de Weerd for the bioinformatic analysis of 16S rRNA amplicon

443 sequencing data.

444

445

19

446 Reference

447 1. Koenig JE, Spor A, Scalfone N, Fricker AD, Stombaugh J, Knight R, Angenent

448 LT, Ley RE. 2011. Succession of microbial consortia in the developing infant

449 gut microbiome. Proc Natl Acad Sci U S A 108 Suppl 1:4578-85.

450 2. Backhed F, Roswall J, Peng Y, Feng Q, Jia H, Kovatcheva-Datchary P, Li Y,

451 Xia Y, Xie H, Zhong H, Khan MT, Zhang J, Li J, Xiao L, Al-Aama J, Zhang D,

452 Lee YS, Kotowska D, Colding C, Tremaroli V, Yin Y, Bergman S, Xu X,

453 Madsen L, Kristiansen K, Dahlgren J, Jun W. 2015. Dynamics and

454 Stabilization of the Human Gut Microbiome during the First Year of Life. Cell

455 Host Microbe 17:690-703.

456 3. Arrieta MC, Stiemsma LT, Amenyogbe N, Brown EM, Finlay B. 2014. The

457 intestinal microbiome in early life: health and disease. Front Immunol 5:427.

458 4. Sherman MP, Zaghouani H, Niklas V. 2015. Gut microbiota, the immune

459 system, and diet influence the neonatal gut-brain axis. Pediatr Res 77:127-35.

460 5. Thompson AL. 2012. Developmental origins of obesity: early feeding

461 environments, infant growth, and the intestinal microbiome. Am J Hum Biol

462 24:350-60.

463 6. Neville MC, Anderson SM, McManaman JL, Badger TM, Bunik M, Contractor N,

464 Crume T, Dabelea D, Donovan SM, Forman N, Frank DN, Friedman JE,

465 German JB, Goldman A, Hadsell D, Hambidge M, Hinde K, Horseman ND,

466 Hovey RC, Janoff E, Krebs NF, Lebrilla CB, Lemay DG, MacLean PS, Meier P,

467 Morrow AL, Neu J, Nommsen-Rivers LA, Raiten DJ, Rijnkels M, Seewaldt V,

468 Shur BD, VanHouten J, Williamson P. 2012. Lactation and neonatal nutrition:

20

469 defining and refining the critical questions. J Mammary Gland Biol Neoplasia

470 17:167-88.

471 7. Zivkovic AM, German JB, Lebrilla CB, Mills DA. 2011. Human milk glycobiome

472 and its impact on the infant gastrointestinal microbiota. Proc Natl Acad Sci U S

473 A 108 Suppl 1:4653-8.

474 8. Smilowitz JT, Lebrilla CB, Mills DA, German JB, Freeman SL. 2014. Breast

475 Milk Oligosaccharides: Structure-Function Relationships in the Neonate. Annu

476 Rev Nutr doi:10.1146/annurev-nutr-071813-105721.

477 9. Thurl S, Munzert M, Boehm G, Matthews C, Stahl B. 2017. Systematic review

478 of the concentrations of oligosaccharides in human milk. Nutr Rev

479 doi:10.1093/nutrit/nux044.

480 10. Engfer MB, Stahl B, Finke B, Sawatzki G, Daniel H. 2000. Human milk

481 oligosaccharides are resistant to enzymatic hydrolysis in the upper

482 gastrointestinal tract. Am J Clin Nutr 71:1589-96.

483 11. Bode L. 2012. Human milk oligosaccharides: every baby needs a sugar mama.

484 Glycobiology 22:1147-62.

485 12. Marcobal A, Barboza M, Froehlich JW, Block DE, German JB, Lebrilla CB,

486 Mills DA. 2010. Consumption of human milk oligosaccharides by gut-related

487 microbes. J Agric Food Chem 58:5334-40.

488 13. Sela DA, Mills DA. 2010. Nursing our microbiota: molecular linkages between

489 bifidobacteria and milk oligosaccharides. TRENDS in Microbiology 18:298-307.

490 14. Laursen MF, Bahl MI, Michaelsen KF, Licht TR. 2017. First Foods and Gut

491 Microbes. Front Microbiol 8:356.

21

492 15. Cheng J, Ringel-Kulka T, Heikamp-de Jong I, Ringel Y, Carroll I, de Vos WM,

493 Salojarvi J, Satokari R. 2015. Discordant temporal development of bacterial

494 phyla and the emergence of core in the fecal microbiota of young children.

495 ISME J doi:10.1038/ismej.2015.177.

496 16. Jost T, Lacroix C, Braegger CP, Chassard C. 2012. New insights in gut

497 microbiota establishment in healthy breast fed neonates. PLoS One 7:e44595.

498 17. Laursen MF, Andersen LBB, Michaelsen KF, Molgaard C, Trolle E, Bahl MI,

499 Licht TR. 2016. Infant Gut Microbiota Development Is Driven by Transition to

500 Family Foods Independent of Maternal Obesity. Msphere 1.

501 18. Arrieta MC, Stiemsma LT, Dimitriu PA, Thorson L, Russell S, Yurist-Doutsch S,

502 Kuzeljevic B, Gold MJ, Britton HM, Lefebvre DL, Subbarao P, Mandhane P,

503 Becker A, McNagny KM, Sears MR, Kollmann T, Investigators CS, Mohn WW,

504 Turvey SE, Brett Finlay B. 2015. Early infancy microbial and metabolic

505 alterations affect risk of childhood asthma. Sci Transl Med 7:307ra152.

506 19. Wopereis H, Sim K, Shaw A, Warner JO, Knol J, Kroll S. 2017. Intestinal

507 Microbiota in Infants at High-risk for Allergy: Effects of Prebiotics and Role in

508 Eczema Development. Journal of Allergy and Clinical Immunology

509 doi:10.1016/j.jaci.2017.05.054.

510 20. Pham VT, Lacroix C, Braegger CP, Chassard C. 2017. Lactate-utilizing

511 community is associated with gut microbiota dysbiosis in colicky infants.

512 Scientific Reports 7.

513 21. de Weerth C, Fuentes S, Puylaert P, de Vos WM. 2013. Intestinal microbiota of

514 infants with colic: development and specific signatures. Pediatrics 131:e550-8.

22

515 22. Belenguer A, Duncan SH, Calder AG, Holtrop G, Louis P, Lobley GE, Flint HJ.

516 2006. Two routes of metabolic cross-feeding between Bifidobacterium

517 adolescentis and butyrate-producing anaerobes from the human gut. Applied

518 and Environmental Microbiology 72:3593-3599.

519 23. Falony G, Vlachou A, Verbrugghe K, De Vuyst L. 2006. Cross-feeding between

520 Bifidobacterium longum BB536 and acetate-converting, butyrate-producing

521 colon bacteria during growth on oligofructose. Applied and Environmental

522 Microbiology 72:7835-7841.

523 24. De Vuyst L, Leroy F. 2011. Cross-feeding between bifidobacteria and butyrate-

524 producing colon bacteria explains bifdobacterial competitiveness, butyrate

525 production, and gas production. Int J Food Microbiol 149:73-80.

526 25. Rios-Covian D, Gueimonde M, Duncan SH, Flint HJ, de los Reyes-Gavilan CG.

527 2015. Enhanced butyrate formation by cross-feeding between

528 Faecalibacterium prausnitzii and Bifidobacterium adolescentis. Fems

529 Microbiology Letters 362.

530 26. Riviere A, Gagnon M, Weckx S, Roy D, De Vuyst L. 2015. Mutual Cross-

531 Feeding Interactions between Bifidobacterium longum subsp longum NCC2705

532 and Eubacterium rectale ATCC 33656 Explain the Bifidogenic and Butyrogenic

533 Effects of Arabinoxylan Oligosaccharides. Applied and Environmental

534 Microbiology 81:7767-7781.

535 27. Schwab C, Ruscheweyh HJ, Bunesova V, Pham VT, Beerenwinkel N, Lacroix

536 C. 2017. Trophic Interactions of Infant Bifidobacteria and Eubacterium hallii

537 during L-Fucose and Fucosyllactose Degradation. Front Microbiol 8:95.

23

538 28. Bunesova V, Lacroix C, Schwab C. 2017. Mucin Cross-Feeding of Infant

539 Bifidobacteria and Eubacterium hallii. Microb Ecol doi:10.1007/s00248-017-

540 1037-4.

541 29. Yatsunenko T, Rey FE, Manary MJ, Trehan I, Dominguez-Bello MG, Contreras

542 M, Magris M, Hidalgo G, Baldassano RN, Anokhin AP, Heath AC, Warner B,

543 Reeder J, Kuczynski J, Caporaso JG, Lozupone CA, Lauber C, Clemente JC,

544 Knights D, Knight R, Gordon JI. 2012. Human gut microbiome viewed across

545 age and geography. Nature 486:222-7.

546 30. Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR,

547 Fernandes GR, Tap J, Bruls T, Batto JM, Bertalan M, Borruel N, Casellas F,

548 Fernandez L, Gautier L, Hansen T, Hattori M, Hayashi T, Kleerebezem M,

549 Kurokawa K, Leclerc M, Levenez F, Manichanh C, Nielsen HB, Nielsen T,

550 Pons N, Poulain J, Qin J, Sicheritz-Ponten T, Tims S, Torrents D, Ugarte E,

551 Zoetendal EG, Wang J, Guarner F, Pedersen O, de Vos WM, Brunak S, Dore J,

552 Meta HITC, Antolin M, Artiguenave F, Blottiere HM, Almeida M, Brechot C,

553 Cara C, Chervaux C, Cultrone A, Delorme C, Denariaz G, et al. 2011.

554 Enterotypes of the human gut microbiome. Nature 473:174-80.

555 31. Ruiz L, Delgado S, Ruas-Madiedo P, Sanchez B, Margolles A. 2017.

556 Bifidobacteria and Their Molecular Communication with the Immune System.

557 Frontiers in Microbiology 8.

558 32. Fellows R, Denizot J, Stellato C, Cuomo A, Jain P, Stoyanova E, Balázsi S,

559 Hajnády Z, Liebert A, Kazakevych J, Blackburn H, Corrêa RO, Fachi JL, Sato

560 FT, Ribeiro WR, Ferreira CM, Perée H, Spagnuolo M, Mattiuz R, Matolcsi C,

561 Guedes J, Clark J, Veldhoen M, Bonaldi T, Vinolo MAR, Varga-Weisz P. 2018.

24

562 Microbiota derived short chain fatty acids promote histone crotonylation in the

563 colon through histone deacetylases. Nature Communications 9:105.

564 33. Bolognini D, Tobin AB, Milligan G, Moss CE. 2016. The Pharmacology and

565 Function of Receptors for Short-Chain Fatty Acids. Mol Pharmacol 89:388-98.

566 34. Koh A, De Vadder F, Kovatcheva-Datchary P, Backhed F. 2016. From Dietary

567 Fiber to Host Physiology: Short-Chain Fatty Acids as Key Bacterial Metabolites.

568 Cell 165:1332-45.

569 35. Geirnaert A, Calatayud M, Grootaert C, Laukens D, Devriese S, Smagghe G,

570 De Vos M, Boon N, Van de Wiele T. 2017. Butyrate-producing bacteria

571 supplemented in vitro to Crohn's disease patient microbiota increased butyrate

572 production and enhanced intestinal epithelial barrier integrity. Sci Rep 7:11450.

573 36. Tannock GW, Lee PS, Wong KH, Lawley B. 2016. Why Don't All Infants Have

574 Bifidobacteria in Their Stool? Frontiers in Microbiology 7.

575 37. O'Callaghan A, van Sinderen D. 2016. Bifidobacteria and Their Role as

576 Members of the Human Gut Microbiota. Frontiers in Microbiology 7.

577 38. Underwood MA, German JB, Lebrilla CB, Mills DA. 2014. Bifidobacterium

578 longum subspecies infantis: champion colonizer of the infant gut. Pediatr Res

579 doi:10.1038/pr.2014.156.

580 39. Palframan RJ, Gibson GR, Rastall RA. 2003. Carbohydrate preferences of

581 Bifidobacterium species isolated from the human gut. Curr Issues Intest

582 Microbiol 4:71-5.

583 40. Oozeer R, van Limpt K, Ludwig T, Ben Amor K, Martin R, Wind RD, Boehm G,

584 Knol J. 2013. Intestinal microbiology in early life: specific prebiotics can have

25

585 similar functionalities as human-milk oligosaccharides. Am J Clin Nutr

586 98:561S-71S.

587 41. Pham VT, Lacroix C, Braegger CP, Chassard C. 2016. Early colonization of

588 functional groups of microbes in the infant gut. Environ Microbiol

589 doi:10.1111/1462-2920.13316.

590 42. Duncan SH, Louis P, Thomson JM, Flint HJ. 2009. The role of pH in

591 determining the species composition of the human colonic microbiota. Environ

592 Microbiol 11:2112-22.

593 43. van Limpt C, Crienen A, Vriesema A, Knol J. 2004. Effect of colonic short

594 chain fatty acids, lactate and pH on the growth of common gut pathogens.

595 Pediatric Research 56:487-487.

596 44. Walker AW, Duncan SH, Leitch ECM, Child MW, Flint HJ. 2005. pH and

597 peptide supply can radically alter bacterial populations and short-chain fatty

598 acid ratios within microbial communities from the human colon. Applied and

599 Environmental Microbiology 71:3692-3700.

600 45. Reichardt N, Vollmer M, Holtrop G, Farquharson FM, Wefers D, Bunzel M,

601 Duncan SH, Drew JE, Williams LM, Milligan G, Preston T, Morrison D, Flint HJ,

602 Louis P. 2017. Specific substrate-driven changes in human faecal microbiota

603 composition contrast with functional redundancy in short-chain fatty acid

604 production. ISME J doi:10.1038/ismej.2017.196.

605 46. Sheridan PO, Martin JC, Lawley TD, Browne HP, Harris HM, Bernalier-

606 Donadille A, Duncan SH, O'Toole PW, Scott KP, Flint HJ. 2016.

607 Polysaccharide utilization loci and nutritional specialization in a dominant

26

608 group of butyrate-producing human colonic Firmicutes. Microb Genom

609 2:e000043.

610 47. Duncan SH, Louis P, Flint HJ. 2004. Lactate-utilizing bacteria, isolated from

611 human feces, that produce butyrate as a major fermentation product. Appl

612 Environ Microbiol 70:5810-7.

613 48. Seth EC, Taga ME. 2014. Nutrient cross-feeding in the microbial world. Front

614 Microbiol 5:350.

615 49. Schwiertz A, Taras D, Schafer K, Beijer S, Bos NA, Donus C, Hardt PD. 2010.

616 Microbiota and SCFA in lean and overweight healthy subjects. Obesity (Silver

617 Spring) 18:190-5.

618 50. Scott KP, Duncan SH, Louis P, Flint HJ. 2011. Nutritional influences on the gut

619 microbiota and the consequences for gastrointestinal health. Biochem Soc

620 Trans 39:1073-8.

621 51. Stokholm J, Blaser MJ, Thorsen J, Rasmussen MA, Waage J, Vinding RK,

622 Schoos AM, Kunoe A, Fink NR, Chawes BL, Bonnelykke K, Brejnrod AD,

623 Mortensen MS, Al-Soud WA, Sorensen SJ, Bisgaard H. 2018. Maturation of

624 the gut microbiome and risk of asthma in childhood. Nat Commun 9:141.

625 52. Caporaso JG, Kuczynski J, Stombaugh J, Bittinger K, Bushman FD, Costello

626 EK, Fierer N, Pena AG, Goodrich JK, Gordon JI, Huttley GA, Kelley ST,

627 Knights D, Koenig JE, Ley RE, Lozupone CA, McDonald D, Muegge BD,

628 Pirrung M, Reeder J, Sevinsky JR, Turnbaugh PJ, Walters WA, Widmann J,

629 Yatsunenko T, Zaneveld J, Knight R. 2010. QIIME allows analysis of high-

630 throughput community sequencing data. Nat Methods 7:335-6.

27

631 53. Edgar RC. 2010. Search and clustering orders of magnitude faster than

632 BLAST. Bioinformatics 26:2460-2461.

633 54. Cole JR, Wang Q, Cardenas E, Fish J, Chai B, Farris RJ, Kulam-Syed-

634 Mohideen AS, McGarrell DM, Marsh T, Garrity GM, Tiedje JM. 2009. The

635 Ribosomal Database Project: improved alignments and new tools for rRNA

636 analysis. Nucleic Acids Res 37:D141-5.

637 55. Pruesse E, Quast C, Knittel K, Fuchs BM, Ludwig W, Peplies J, Glockner FO.

638 2007. SILVA: a comprehensive online resource for quality checked and

639 aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids

640 Res 35:7188-96.

641 56. Plugge CM. 2005. Anoxic media design, preparation, and considerations.

642 Methods Enzymol 397:3-16.

643 57. Schwiertz A, Hold GL, Duncan SH, Gruhl B, Collins MD, Lawson PA, Flint HJ,

644 Blaut M. 2002. Anaerostipes caccae gen. nov., sp. nov., a new saccharolytic,

645 acetate-utilising, butyrate-producing bacterium from human faeces. Syst Appl

646 Microbiol 25:46-51.

647 58. Stahl B, Thurl S, Zeng JR, Karas M, Hillenkamp F, Steup M, Sawatzki G. 1994.

648 OLIGOSACCHARIDES FROM HUMAN-MILK AS REVEALED BY MATRIX-

649 ASSISTED LASER-DESORPTION IONIZATION MASS-SPECTROMETRY.

650 Analytical Biochemistry 223:218-226.

651 59. Delroisse JM, Boulvin AL, Parmentier I, Dauphin RD, Vandenbol M, Portetelle

652 D. 2008. Quantification of Bifidobacterium spp. and Lactobacillus spp. in rat

653 fecal samples by real-time PCR. Microbiol Res 163:663-70.

28

654 60. Veiga P, Gallini CA, Beal C, Michaud M, Delaney ML, DuBois A, Khlebnikov A,

655 van Hylckama Vlieg JE, Punit S, Glickman JN, Onderdonk A, Glimcher LH,

656 Garrett WS. 2010. Bifidobacterium animalis subsp. lactis fermented milk

657 product reduces inflammation by altering a niche for colitogenic microbes.

658 Proc Natl Acad Sci U S A 107:18132-7.

659 61. Hayashi H, Sakamoto M, Benno Y. 2004. Evaluation of three different forward

660 primers by terminal restriction fragment length polymorphism analysis for

661 determination of fecal bifidobacterium spp. in healthy subjects. Microbiol

662 Immunol 48:1-6.

663 62. Sim K, Cox MJ, Wopereis H, Martin R, Knol J, Li MS, Cookson WO, Moffatt

664 MF, Kroll JS. 2012. Improved detection of bifidobacteria with optimised 16S

665 rRNA-gene based pyrosequencing. PLoS One 7:e32543.

666 63. Dinoto A, Suksomcheep A, Ishizuka S, Kimura H, Hanada S, Kamagata Y,

667 Asano K, Tomita F, Yokota A. 2006. Modulation of rat cecal microbiota by

668 administration of raffinose and encapsulated Bifidobacterium breve. Appl

669 Environ Microbiol 72:784-92.

670 64. Yin Y, Mao X, Yang J, Chen X, Mao F, Xu Y. 2012. dbCAN: a web resource

671 for automated carbohydrate-active enzyme annotation. Nucleic Acids Res

672 40:W445-51.

673 65. Krogh A, Larsson B, von Heijne G, Sonnhammer EL. 2001. Predicting

674 transmembrane protein topology with a hidden Markov model: application to

675 complete genomes. J Mol Biol 305:567-80.

29

676 66. Petersen TN, Brunak S, von Heijne G, Nielsen H. 2011. SignalP 4.0:

677 discriminating signal peptides from transmembrane regions. Nat Methods

678 8:785-6.

679 67. Cleveland WS, Grosse E, Shyu WM. 1992. Local regression models. In

680 Chambers JM, Hastie TJ (ed), Statistical Models in S. Wadsworth &

681 Brooks/Cole.

682 68. Varki A, Cummings RD, Aebi M, Packer NH, Seeberger PH, Esko JD, Stanley

683 P, Hart G, Darvill A, Kinoshita T, Prestegard JJ, Schnaar RL, Freeze HH,

684 Marth JD, Bertozzi CR, Etzler ME, Frank M, Vliegenthart JF, Lutteke T, Perez

685 S, Bolton E, Rudd P, Paulson J, Kanehisa M, Toukach P, Aoki-Kinoshita KF,

686 Dell A, Narimatsu H, York W, Taniguchi N, Kornfeld S. 2015. Symbol

687 Nomenclature for Graphical Representations of Glycans. Glycobiology

688 25:1323-4.

689

690

30

691 Figures and tables

692 Figure 1. The occurrence of (A) Bifidobacterium and Anaerostipes genus (B)

693 Bifidobacteriaceae and Lachnospiraceae family in the gut microbiota across

694 age. The plot was generated from a published dataset (29) using R package ggplot2

695 version 2.2.1. The trend lines represent the smoothed conditional means using local

696 polynomial regression fitting (67).

697

698 Figure 2. B. inf supported the growth of A. cac in human milk carbohydrates. (A)

699 The optical density (OD600) indicating bacterial growth and (B) qPCR results

700 showing the microbial composition in the co-cultures over time with lactose or with

701 total HM carbohydrates. Error bars represent the standard deviation for biological

702 triplicates, except for time point 31 h (n=2) and 48 h (n=1). (C) Fluorescent in situ

703 hybridisation (FISH) of co-cultures at 11h (B. inf in pink and A. cac in purple). No

704 growth or substrate utilisation was detected for A. cac monocultures in identical

705 media (Table S1).

706

707 Figure 3. B. inf supported butyrate production of A. cac. The substrate utilisation

708 and SCFA formation of co-cultures in lactose or total HM carbohydrates. Error bars

709 represent the standard deviation for biological triplicates, except for time point 31 h

710 (n=2) and 48 h (n=1).

711

712 Figure 4. B. inf monoculture and co-culture with A. cac utilised the full range of

713 low molecular weight HMOS. Error bars represent the error propagation for mean

31

714 of three (for A. cac) or four (for B. inf and B. inf + A. cac) biological replicates

715 measured in technical triplicates. The HMOS structures and glycosidic linkages are

716 depicted according to Varki et al. (68). Abbreviations: 2’-FL, 2’-fucosyllactose; 3-FL,

717 3-fucosyllactose; DFL, difucosyllactose; LNT, lacto-N-tetraose; LNnT, lacto-N-

718 neotetraose; LNFP I, lacto-N-fucopentaose I; LNFP II, lacto-N-fucopentaose II; LNFP

719 III, lacto-N-fucopentaose III; LNFP V, lacto-N-fucopentaose V; SL, sialyllactose.

720

721 Figure 5. The acidification of cultures and faecal pH. (A) Base (2M NaOH) added

722 to maintain the anaerobic chemostat at pH 6.5. The shaded error bars indicate

723 standard deviation for biological triplicates. (B) The faecal pH for infants (n=138).

724 Data adapted from Wopereis et al. (19).

725

32

(A) Genus level Bifidobacterium Anaerostipes First 1000 days First 1000 days 0.100 0.9 0.12

0.075 Spearman ρ = 0.56 0.75 0.6 Spearman ρ = -0.38 p < 0.05 p < 0.05 0.050 0.3 0.025 0.08 0.0 Relative abundance (%) Relative abundance (%) 0.50 0.000

0 1 2 0 1 2 Age Age

0.25 0.04

Spearman ρ = -0.45 Spearman ρ = 0.27 Relative abundance (%) Relative abundance (%) p < 0.05 p < 0.05

0.00 0.00

0 20 40 60 80 0 20 40 60 80 Age Age

(B) Family level Bifidobacteriaceae Lachnospiraceae First 1000 days First 1000 days 0.9 0.6

Spearman ρ = -0.38 Spearman ρ = 0.37 0.75 0.6 p < 0.05 p < 0.05 0.6 0.4

0.3 0.2 Relative abundance (%) 0.0 Relative abundance (%) 0.50 0.0 0.4 0 1 2 0 1 2 Age Age

0.25 Spearman ρ = 0.38 p < 0.05 Spearman ρ = -0.45 0.2 p < 0.05 Relative abundance (%) Relative abundance (%)

0.00

0.0

0 20 40 60 80 0 20 40 60 80 Age Age

Figure 1. The occurrence of (A) Bifidobacterium and Anaerostipes genus (B) Bifidobacteriaceae and Lachnospiraceae family in the gut microbiota across age. The plot was generated from a published dataset (32) using R package ggplot2 version 2.2.1. The trend lines represent the smoothed conditional means using local polynomial regression fitting (69). (A) Cell density(B) qPCR(C) FISH B. infantis + A. caccae in lactose 10 5 10 B. infantis 109 B. infantis + A. caccae 8 4 10 7 10 6 3 10 5 10 4 OD600

Lactose 2 10 3 10 B. infantis in co-culture 2 1 10 A. caccae in co-culture 1 16S rRNA copy number rRNA 16S / ml culture 10 0 0 10 0 12 24 36 48 0 12 24 36 48 Time (hours) Time (hours)

B. infantis + A. caccae in total HM carbohydrates 10 5 10 B. infantis 109 B. infantis + A. caccae 8 4 10 7 10 6 3 10 5 10 4 OD600 2 10 3 10 B. infantis in co-culture 2 A. caccae in co-culture 1 10 1 Total HM carbohydrates Total 16S rRNA copy number rRNA 16S / ml culture 10 0 0 10 0 12 24 36 48 0 12 24 36 48 Time (hours) Time (hours)

Figure 2. B. infantis supported the growth of A. caccae in human milk carbohydrates. (A) The optical density (OD600) indicating bacterial growth and (B) qPCR results showing the microbial composition in the co-cultures over time with lactose or with total HM carbohydrates. Error bars represent the standard deviation for biological triplicates, except for time point 31 h (n=2) and 48 h (n=1). (C) Fluorescent in situ hybridisation (FISH) of co-cultures at 11h (B. infantis in pink and A. caccae in purple). No growth or substrate utilisation was detected for A. caccae monocultures in identical media (Table S1). deviation forbiological triplicates,exceptfortimepoint31h(n=2) and 48h(n=1). formation ofco-cultures inlactoseortotalHMcarbohydrates.Error barsrepresentthestandard Figure 3.B.infantissupportedbutyrateproduction of

Total HM carbohydrates Lactose

Concentration (mM) Concentration (mM) Concentration (mM) Concentration (mM) 1 1 2 2 3 1 1 2 2 3 1 1 2 2 3 1 1 2 2 3 0 5 0 5 0 0 5 0 5 0 5 0 0 5 0 5 0 0 5 0 5 0 5 0 5 0 0 5 L a 0 0 0 c 0 t o s e Substrate utilization B. infantis+ 12 12 12 12 B. infantis+ T T T T G i i i i B. infantis m m B. infantis m m l u e e e e c

o ( ( ( ( 24 24 24 24 h h h h s o o o o A. caccae e u u u u A. caccae r r r r s s s s ) ) ) ) 36 36 36 36 G a l a c t o s 48 48 48 48 e A. caccae. The substrate utilisationandSCFA F o r

m Concentration (mM) Concentration (mM) Concentration (mM) Concentration (mM) 1 2 3 4 5 6 7 1 2 3 4 5 6 7 a 1 2 3 4 5 6 7 1 2 3 4 5 6 7 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 t 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 e 0 0 0 0 Product formation 12 12 12 A 12 B. infantis+ B. infantis+ c e T T T t T i a i i m m i m B. infantis t m B. infantis e e e e

e

(

( ( h 24 24 24 ( h h 24 h o o o o u u u A. caccae u r r r A. caccae s r s s s ) ) ) ) B 36 36 u 36 36 t y r a t e 48 48 48 48 L a c t a t e 150

B. infantis

100

50

Remaining HMOS after 24h (%) 0

SL 2’-FL 3-FL DFL LNT Lactose LNnT LNFP I LNFP II LNFP III LNFP V

200 A. caccae 125 125

100

75

50

25 Remaining HMOS after 24h (%) 0 SL 2’-FL 3-FL DFL LNT Lactose LNnT LNFP I LNFP II LNFP III LNFP V 150

B. infantis + A. caccae

100

50 Remaining HMOS after 24h (%) 0 SL 2’-FL 3-FL DFL LNT Lactose LNnT LNFP I LNFP II LNFP III LNFP V

6 Trioses (neutral) Tetraoses (neutral) Pentaoses (neutral) Trioses (acidic) 4 1 β 4 β 4 β 4 β 4 β 4 2 3 2 3 β 4 3 3 β 4 α 4 3 β 4 β 4 α 3 β β 3 3 β 4 β 4 3 α α α β 4 β β β 4 β β 3 3 3 3 3 α α β 3 3 2 β β α β 2 Key for linkages α

α 6 β 4

Glucose Galactose N-acetylglucosamine Fucose Sialic acid

Figure 4. B. infantis monoculture and co-culture with A. caccae utilised the full range of low molecular weight HMOS. Error bars represent the error propagation for mean of three (for A. caccae) or four (for B. infantis and B. infantis + A. caccae) biological replicates measured in technical triplicates. The HMOS structures and glycosidic linkages are depicted according to Varki et al. (70). Abbreviations: 2’-FL, 2’-fucosyllactose; 3-FL, 3-fucosyllactose; DFL, difucosyllactose; LNT, lacto-N-tetraose; LNnT, lacto-N-neotetraose; LNFP I, lacto-N-fucopentaose I; LNFP II, lacto-N-fucopentaose II; LNFP III, lacto-N-fucopentaose III; LNFP V, lacto-N-fucopentaose V; SL, sialyllactose. (A) Anaerobic chemostat (B) Faecal pH

8 8

7 - + 7 HA A + H

6 6 SCFA pKa 5 5

Lactate 3.86 pH 4 4 Acetate 4.76 3 3 Butyrate 4.82

2 2 Amount Amount of 2M NaOH added (ml) 1 1

0 0 0 7 14 21 28 4 12 24 Time (h) Age of infants (weeks)

B. infantis B. infantis + A. caccae Lactose Total HM carbohydrates Lactose Total HM carbohydrates

Figure 5. The acidification of cultures and faecal pH. (A) Base (2M NaOH) added to maintain the anaerobic chemostat at pH 6.5. The shaded error bars indicate standard deviation for biological triplicates. (B) The faecal pH for infants (n=138). Data adapted from Wopereis et al. (22).