1 Cross-feeding between Bifidobacterium infantis and Anaerostipes 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 gut microbiota 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 Firmicutes 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 Anaerostipes caccae 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 bacteria, could
51 orchestrated an important event of maturation for both the gut ecosystem and
52 physiology of infant.
53
54 Keywords
55 Bifidobacteria, butyrate, Lachnospiraceae, 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 (Roseburia 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 taxonomy 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).