The role of early life nutrition in the establishment of gastrointestinal microbial composition and function

ABSTRACT

The development of the human infant intestinal microbiota is a sequential process that begins in utero and continues during the first 2 to 3 years of life. Microbial composition and diversity are shaped by host genetics and multiple environmental factors, of which diet is a principal contributor. An understanding of this process is of clinical importance as the microbiota acquired in early life influence gastrointestinal, immune and neural development, and reduced microbial diversity or dysbiosis during infancy is associated with disorders in infancy and later childhood. The goal of this article was to review the published literature that used culture-independent methods to describe the development of the gastrointestinal microbiota in breast- and formula-fed human infants as well as the impact of prebiotic and probiotic addition to infant formula, and the addition of solid foods.

Introduction

Intestinal microbiota development is a complex process that begins in utero and continues for the first 2–3 years of life.^1-3 Host genetics and environmental factors such as gestational age, delivery mode, diet, pre- and probiotics, antibiotics, maternal weight and stress influence the process.⁴ The microbiota acquired in early life have long-term implications for host metabolism and gastrointestinal (GI), immune and neurological function.^5,6 Reduced diversity or dysbiosis are linked to childhood and later life disorders, including necrotizing enterocolitis,⁷ eczema,⁸ asthma,⁹ inflammatory bowel diseases,¹⁰ irritable bowel syndrome,¹¹ obesity,¹² diabetes¹³ and autism.¹⁴

Diet is one of the major determinants of GI microbial diversity. Bacterial composition and alpha-diversity differ between breastfed (BF) and formula-fed (FF) infants,^15-17 and solid food introduction has been associated with rapid and sustained alterations in the fecal microbiota.^1,18 The goal of this review is to summarize our current knowledge on how nutrition may influence GI microbial composition and function in early infancy. The impact of pre- and probiotic supplementation on the colonization of the infant microbiota will also be addressed.

Breast- and formula-feeding

The fecal microbial composition of BF and FF infants has been investigated for more than 40 years. Prior to the early part of the 21st century, microbes were studied primarily using culture-dependent methods, which limited our understanding of microbial diversity and often resulted in conflicting reports.¹⁹ Findings from 16S ribosomal ribonucleic acid (16S rRNA)-based approaches (For a review, see ref. 20) have shown that the infant microbiota is more diverse than previously appreciated. However, the findings are still somewhat varied, likely due to the small number of infants included in some studies (Table 1), geographical differences,²¹ differences in analytical methods and sampling time points,¹⁷ changes in infant formula composition over time, and differences in oligosaccharide composition of human milk (HM) and infant formula.^17,20 This review focuses on studies that used 16S rRNA-based methods to investigate the fecal microbial composition in infants over the first yr of life (Table 1).

Table 1.

Differences in fecal microbial composition among infants exposed to different feeding patterns.

Sample Timepoint	Feeding Mode (n)	Method	Results	Reference (Country)
4 mo	BF (63), MF (22), FF (9)	Mixed Sanger sequencing	• Solid foods did not affect microbial composition	22 (Norway)
3−4 mo	BF (19), MF (5), FF (9)	Illumina sequencing	• No difference in proportion of Bifidobacterium	15 (Canada)
			• Lower prevalence of C. difficile in BF
			• Higher proportions of Verrucomicrobiaceae (Akkermansia) and Peptostreptococcaceae in FF
			• BF exclusivity was negatively correlated with proportion of Peptostreptococcaceae and prevalence C. difficile
3 mo, 12 mo	BF (102), MF (60), FF (36)	Illumina sequencing	• BF had higher proportions of Actinobacteria and Proteobacteria	23 (Canada)
			• At 3 mo, BF had higher proportions of Bifidobacterium and lower proportions of Bacteroides and Clostridiales
			• At 3 mo, BF exclusivity was inversely associated with diversity and proportions of Bacteroidetes and Clostridiales
			• Higher proportion of Clostridiales at 1 y associated with BF exclusivity at 3 mo
			• BF exclusivity and duration associated with higher diversity at 1 y.
			• BF exposed to intrapartum antibiotics (IAP) had higher proportions of Clostridium
			• At 12 mo, FF & MF had lower proportions of Bacteroidaceae and higher proportions of Clostridiales if exposed to IAP during emergency CS (compared with FF & MF infants who were not exposed)
1 wk, 4 mo, 12 mo	1 wk: BF (73), MF (24), FF (1)Four mo: BF (67), MF (19), FF (11)	Illumina sequencing	• No differences in MF and BF infants at 1 wk	24 (Sweden)
			• At 4 mo, BF had higher proportions of L. johnsonii/L.gasseri, L. paracasei/L.casei and B. longum
			• At 4 mo, FF had higher proportions of C. difficile, Granulicatella adjacens, Citrobacter spp, Enterobacter cloacae, Bilophila
			wadsworthia and B. adolescentis
			• Cessation of breastfeeding followed by increased proportions of Bacteroides, Bilophila, Roseburia, Clostridium, Anaerostipes.
			• Bifidobacterium, Lactobacillus, Collinsella, Megasphaera, and Veillonella dominated infants receiving breast milk at 12 mo.
birth−24 mo	BF-dominant (31),FF-dominant (12)	Illumina sequencing	• α and β-diversity affected by delivery mode, antibiotic use, and diet.	25 (US)
			• Solid foods associated with a ↑ in Clostridiales
			• BF- dominant (> 50% of feedings were breast milk from 0–3 mo) had higher abundance of Lactobacillus, Staphylococcus, Megasphaera, and Actinobacteria
			• FF-dominant (> 50% of feedings were formula from 0–3 mo) had greater abundances of Clostridiales and Proteobacteria
			• Diversity decreased from 12–24 mo in FF-dominant
3−6 mo	BF (6), FF (6)	Illumina sequencing	• BF had greater proportions of Actinobacteria and Firmicutes, while FF had higher proportions of Proteobacteria	26 (China)
			• BF dominated by Streptococcus, Bifidobacterium, Lactobacillus, Clostridium.
			• FF dominated by Klebsiella, Streptococcus, Lactobacillus, Enterococcus, Bifidobacterium.
			• BF had greater proportions of Lactobacillus and were 2x greater abundance of Bifidobacterium
			• Klebsiella was 30x more abundant in FF
			• Staphylococcaceae and Pasteurellaceae were only present in BF infants.
			• Bacillaceae, Veillonellaceae, Ruminococcaceae and Comamonadaceae were only found in FF infants.
d4−6, 9−14, 25−30	BF (7)	Sanger sequencing, pyrosequencing, qPCR	• Actinobacteria was most abundant phyla	27 (Switzerland)
			• B. breve was most dominant strain of Bifidobacterium
			• Lactobacillus levels were variable and unstable
			• Parabacteroides and Bacteroides were most abundant genera from Bacteroidetes
			• Escherichia and Klebsiella were most abundant genera from Proteobacteria
			• Bacteroides levels ↑ over sampling period; timing of appearance differed among infants.
			• Levels of Bacteroides and Bifidobacterium were negatively correlated
			• B. catenulatum, B. adolescentis, Roseburia, Eubacterium, Faecalibacterium, or Ruminococcus not detected
birth−2.5 y	BF until d134; MF until 9 mo; Solids started at 4 mo.; FF until 1 y (1)	Pyrosequencing	• Firmicutes dominated until d92; Actinobacteria and Proteobacteria then began to ↑	1 (US)
			• Introduction of peas and formula led to a ↑ in Bacteroidetes.
			• Solid foods caused a shift in phyla abundances, ↑ bacterial load, and ↑ SCFA.
			• Abundance of Firmicutes and of actinobacterial and proteobacterial genes were negatively correlated.
			• Composition was consistent after weaning.
4 wk	BF (10), FF (10)	Pyrosequencing	• All infants dominated by Actinobacteria, Firmicutes, and Proteobacteria	28 (S. Korea)
			• BF had higher proportion of Actinobacteria and lower proportions of Firmicutes and Proteobacteria
			• Actinobacteria and Firmicutes were almost equally abundant in FF
			• Proportions of Bifidobacteriaceae, Micrococcaceae, Bifidobacterium, L. gasseri, B. longum greater in BF
			• FF had higher proportions of Escherichia, Veillonella, Enterococcus, and Enterobacteria
			• Low abundance of Lactobacillus in FF.
			• B. breve, B. bifidum, L. brevis, and L. reuteri only detected in FF
6 wk	BF (70), MF (25), FF (6)	Illumina sequencing	• Interaction between delivery mode and diet had no impact on composition.	29 (US)
			• Greater phylogenetic differences between BF and FF infants than between BF and MF infants
			• Compositional differences only seen between BF and MF or FF
			• FF had a greater abundance of Lactococcus compared with BF
			• MF had largest intra-group compositional differences
3 mo	BF (6), FF (6)	Pyrosequencing	• BF had greater α-diversity and were less homogenous at phylum-level.	30 (US)
			• Five of the BF infants dominated by 1 phylum (Actinobacteria, Proteobacteria, or Bacteroidetes).
			• Actinobacteria and Firmicutes were equally dominant in FF.
d1, 6 wk, 12 wk	BF (13), FF (13)	Pyrosequencing, qPCR	• In both groups, bifidobacterial counts ↑ from birth to 6 wks	31(France, Poland)
			• qPCR showed B. bifidum and B. breve as the dominant species of bifidobacteria in all infants
			• Pyrosequencing identified B. breve and B. longum as the dominant bifidobacterial species at different times in all infants
			• Bifidobacterium-dominated 2/3 of BF at 6 wks
			• B. pseudocatenulatum became dominant over time in BF
			• FF had higher levels of Bacteroidetes and a greater relative abundance of Peptostreptococcaceae
			• Some FF dominated by Ruminococcus gnavus at 6 and 12 wks.
			• Larger ↓ in Escherichia counts in BF over the first 6 wks
2 mo (on average)	BF (30), FF (60- 30 CM, 30 GM)	Pyrosequencing	• BF had greater proportions of Bifidobacteriaceae, Bacteroidaceae, and B. breve and lower proportions of Lachnospiraceae	32 (Australia)
			• BF and goat's milk (GM) formula-fed infants dominated by Ruminococcus gnavus.
			• GM-fed had a Lachnospiraceae composition more similar to BF.
			• Greater α-diversity in cow's milk (CM) formula group.
			• Both groups of FF had higher proportions of Erysipelotrichaceaes (Clostridium ramosum, Clostridium innocuum - most common)
d13−14 mo	Diet was variable (9)	Pyrosequencing	• BF had a core microbiome of Bifidobacterium and Coprobacillus	33 (US)
			• MF & FF had a core microbiome of Veillonella, Clostridiales, and Bacteroides
			• BF had higher abundances of Actinobacteria and Bifidobacterium
			• Greater proportions of Bacteroidetes, Eggerthella, Clostridiales, Lachnospiraceae, Blautia, and Faecalibacterium in MF & FF
			• Introduction of solids followed by a ↑ in proportions of Bacteroidetes, Eggerthella, Blautia, Neisseria, Peptostreptococcaeceae and ↓ in Staphylococcus and Roseateles in BF.
			• Introduction of solids in MF & FF followed by a ↑ in Bifidobacterium, Ruminococcaeceae, and Blautia.
			• BF had higher proportions of Lactobacillus and Ruminococcaeceae and MF & FF had greater proportions of Clostridiales, Blautia, Faecalibacterium, Anaerotruncus, and Eubacterium after introduction to solids
			• EBF were more compositionally similar to BF + solids than non-EBF infants.
1.5−3 mo	BF (7), MF (1), FF (3)	Pyrosequencing	• Bifidobacteriales, Lactobacillales, and Clostridiales were most abundant classes respectively.	34 (Italy, Spain, Ireland)
			• 10/11 infants dominated by Bifidobacteriaceae
			• Large variations in proportions of Bifidobacterium species, with B. longum being the dominant bifidobacterial species
1 wk, 1 mo, 3 mo,7 mo, 12 mo	Diet was variable; solids introduced between 3–7mo (13)	Pyrosequencing	• Introduction of solid foods followed by large permanent shift toward an adult-like composition.	35 (Spain)
			• ↓ in richness (recovered by 1 yr) and ↑ in diversity throughout introduction of solids.
			• Escherichia ↓ throughout introduction of solid foods.
			• At 7 mo., Bifidobacterium and Bacteroides still dominated and Ruminococcus and Akkermansia began to colonize
			• Similar composition early in life did not predict similar microbial development throughout the first year.
2−5 mo	BF (8), FF (10)	Pyrosequencing, shotgun sequencing	• 48 OTUs discriminated BF and FF	36 (Venezuela, Malawi, US)
			• BF had higher proportion of Bifidobacterium, Actinomyces, Erwinia, and Haemophilus
			• FF had greater levels of Firmicutes and Bacteroidetes
d1−12mo.	BF d1–17 (1), BF d1–130 (1), Diets varied after EBF	PCR-DGGE	• Bifidobacteria were present from d3-d4.	37 (The Netherlands)
			• Ruminococcus, Clostridium, and Enterobacter appeared at different times in infants.
			• At 3 mo, BF composition was more diverse and more dominated by bifidobacteria
			• Ruminococcus levels ↑ upon supplementation and weaning
			• Streptococci and enterococci ↓ upon weaning.
birth−d160	Diet was variable (5)	PCR-DGGE	• Bifidobacteria appeared earlier in BF infants than those who received formula for 1 d	38 (not stated)
			• Diversity of FF infants ↑ within first few days
			• Enterobacteriaceae and enterococci present at higher counts and for a longer duration in FF infants
8 mo (on average); samples collected for 8 wk	BF (2), FF (3) Some solid foods consumed after 6 mo	PCR-DGGE	• B. longum subsp infantis (major species), B. breve, B. bifidum, and B. longum identified in all infants.	39 (Netherlands)
			• BF had a more stable bacterial diversity and greater total numbers of bifidobacteria
			• FF had a greater diversity among bifidobacterial species
			• Low proportion of B. adolescentis in FF; none detected in BF
			• Bifidobacteria counts and species in FF began to resemble BF after starting on prebiotic formula
5, 13, 21 wk (on average)	Infants sampled while EBF, weaning (MF), post-weaning (11)	PCR-TGGE	• EBF period dominated by Bifidobacterium	40 (Algeria)
			• Prevalence of B. breve, B. infantis/longum, and Ruminococcus ↑ across feeding periods.
			• Prevalence of E. coli and Bifidobacterium remained consistent throughout dietary changes.
			• Proportion of bifidobacteria not affected by diet changes
			• Diversity ↑ upon weaning but ↓ post-weaning
			• Greater relative abundance of E. coli and Ruminococcus post-weaning
∼d31	BF (27)	PCR	• B. breve and B. infantis were most frequently detected bifidobacterial species	41 (Japan)
			• B. longum and B. bifidum detected in at least 20% of samples.
			• Low prevalence of B. catenulatum, B. dentium, B. adolescentis, and B. angulatum
1−18 mo	BF (7), FF (7)	DGGE, RISA	• Diversity of bifidobacterial species increased upon weaning	42 (UK)
			• B. bifidum and B. breve were observed in both groups across timepoints.
			• BF had a more diverse bifidobacterial population and a higher prevalence of B. longum and B. breve
			• FF had a more stable Bifidobacterium profile across dietary phases
∼1, 7 mo.	BF (7), FF (6); Solids introduced between 12–21 wks	DGGE, PCR, Dot blot analysis	• Shifts in bifidobacterial populations pre- and post-weaning differed among infants; no significant difference in bifidobacterial	43 (Scotland, The Netherlands,)
			profiles between groups at either timepoint
			• B. infantis was the most prevalent species across both feeding groups.
			• B. breve not detected in FF infants
			• Lactobacillus was prevalent in <50% of infants and shifts pre- and post-weaning were variable among infants
			• L. acidophilus was the most common species of Lactobacillus
			• Composition and diversity did not significantly differ between groups pre- or post-weaning
d30	BF (31), MF (26), FF (11)	PCR, T-RFLP	• No differences in Bifidobacterium species between feeding groups.	44 (Japan)
			• The 3 most common strains were B. longum longum, B. longum infantis, and B. breve respectively.
1, 2, 4 wk; Two, 6, 12 mo	BF until 4 mo; FF + introduction of solids (1). MF until 6 mo; introduced to solids at 4 mo. (1)	T-RFLP	• BF infant dominated by Bacteroides spp, E. coli, Citrobacter, Veillonella spp, and members of clostridia cluster at 1 wk• MF infant dominated by E. coli and Bacteroides at 1 wk	8 (Sweden)
			• Bacteroides spp was dominant at all sample timepoints.
			• Bacteroides spp was dominant at all sample timepoints.
			• Bifidobacterium not detected during BF period but was detected during MF period.
			• Composition during the weaning period was different than during BF or MF.
			• Weaning followed by a ↑ in clostridia, ↓ in Enterobacteriaceae, and ↑ in overall diversity.
9 mo	Partially BF - 168 Weaned - 141	qPCR	• BF infants had lower levels of butyrate-producing bacteria	45 (Denmark)
			• BF duration was positively correlated with Lactobacillus and Bifidobacterium and negatively correlated with C. leptum,
			E. hallii, Roseburia spp, Bacteriodes/Prevotella groups, Bacteroides fragilis, Bacteroides vulgatus, Desulfovibrio spp, and A.
			muciniphila.
			• BF was positively associated with greater abundance of Lactobacillus spp, Bifidobacterium spp, and B. longum and lower
			levels of Delsulfovibrio spp, Akkermansia muciniphila, and Bacteriodetes.
			• Intake of formula was negatively associated with abundance of Lactobacillus spp
			• Introduction of solids followed by ↓ in Lactobacillus spp and Enterobacteriaceae and a ↑ in Bacteroidetes.
			• Composition heavily influenced by the cessation of breastfeeding.
birth−12 mo (26 samples)	Diet was variable across infants and sample timepoints (14)	qPCR	• Each infant had a distinct microbial profile.• Infants were dominated by Flexibacter, Ctyophaga, and Bacteroides.• Infants were dominated by Flexibacter, Ctyophaga, and Bacteroides.	2 (US)
			• Bifidobacteria were not present for several months and only in minor amounts
			• Appearance of Bacteroides was largely variable among infants
			• Introduction to solid foods was followed by a more adult-like microbiota.
			• Composition at 1 y resembled an adult-like microbiota and was largely different than earlier timepoints.
4 wks	BF (50), FF (50)	qPCR	• No significant difference in Bifidobacterium spp counts between groups	46 (The Netherlands)
			• Counts and prevalence of E. coli and C. difficile were higher in FF infants.
1 mo	BF (200), MF (98), FF (232)	qPCR	• BF born by VD at home had highest counts of Bifidobacterium and lowest counts of C. difficile and E. coli.	16 (The Netherlands)
			• FF & MF infants had higher counts and prevalence of E. coli, C. difficile, and B. fragilis
			• Lactobacilli counts were greater in FF than BF infants.
Meconium, d2, d7, d30, d90, d180, 1 wk post weaning	Diet was variable across infants and timepoints (108)	qPCR	• BF delivered by VD were most often colonized by B. bifidum and L. gasseri.	47 (Belgium)
			• FF & MF more often colonized by L. reuteri compared with BF, CS-delivered infants
			• BF, CS-delivered infants reached B. bifidum levels more similar to BF, VD infants sooner than FF or MF, CS-delivered infants.
			• Enterococcus, C. coccoides, Atopobium cluster, B. vulgatus, C. leptum group and B. longum subsp. longum were more prevalent
			in MF & FF
			• Counts of Staphylococcus were higher in BF over the first 3 mo
			• BF had lower counts of Bifidobacterium at earlier timepoints. (May have been affected by some FF infants who received
			probiotic-supplemented formula)
			• BF for the first 3 mo. was associated with greater counts of Bifidobacterium and greater prevalence of L. gasseri, L. casei, and B.
			adolescentis at 6 mo
			• C. perfringens and L. casei were less prevalent in BF, but prevalence was more likely to increase over the first 3 mo. than in FF.
			• Prevalence of C. leptum group increased more over first 3 mos. in MF & FF
			• Introduction of solids was associated with a higher prevalence of Atopobium cluster, C. coccoides, and B. longum subsp longum
			• Enterobacteriaceae and Staphylococcus ↓ after introduction of solids but prevalence of infant-type Bifidobacterium did not change
d21−30	BF (71), MF (77), FF (80), FF + Prebiotic (77)	qPCR	• BF infants had higher levels of lactobacilli and S. aureus.	48 (US)
			• S. aureus colonization rate was higher in infants BF and VD
			• Colonization of S. aureus ↑ at 60 d in FF + prebiotic group.
0−6 mo (for BF/FF)	BF (7), FF (15), Weaned (18)	qPCR; Northern blot	• BF had higher proportions of Bifidobacterium; however, results were only significant if not age-matched.	49 (Scotland)
			• qPCR showed greater proportions of Bacteroides and Enterococcus faecalis in BF.
d1−21 (7 samples)	BF (6), FF (6)	FISH; RAPD-PCR	• Bifidobacterium and Escherichia coli dominated in both groups.	50 (The Netherlands)
			• BF characterized by streptococci and lactobacilli, while FF characterized by staphylococci and clostridia
			• Levels of lactic acid bacteria were higher in BF
			• FF had higher counts of clostridia and Bacteroides
			• FISH showed bifidobacteria and Bacteroides numbers to be approximately equal in FF infants. PCR showed much lower numbers.
d11−22 (BF)	BF (6), FF (6)	FISH	• Both groups were dominated by Bifidobacterium, Bacteroides, and Atopobium.• E. coli colonization was similar across both feeding groups.	51 (Greece)
d14−36 (FF)
			• FF had higher levels of Bacteroides and Atopobium
			• BF had 2x the number of Bifidobacterium
			• Streptococcus and Staphylococcus were present at low levels (< 1%) in BF
			• Bifidobacterial populations were more stable in BF
6 wk	BF (312), MF (112), FF (182)	FISH	• BF had lower proportions C coccoides and Lactobacillus than FF	52 (Italy, Spain, Scotland, Sweden, Germany)
			• Proportions of Bacteroides were lower in BF compared with both FF and MF
			• FF had lower levels of C perfringens compared with BF infants.
			• BF and MF infants had higher levels of bifidobacteria
			• No differences in levels of C. difficile
6 wk, 4 wk post intro to solids	BF (358), MF (85), FF (162)	FISH-FC	• Pre-weaning feeding method impacted composition after introduction of solid foods.	18 (Italy, Spain, Scotland, Sweden, Germany)
			• Infants BF (pre-weaning) had greater post-weaning proportions of bifidobacteria and lower levels of C. coccoides and Bacteroides
			than FF (pre-weaning).
			• Post-weaning proportions of C. coccoides were greater in infants that were MF compared with BF pre-weaning.
			• Changes in C. leptum, C. difficile, and C. perfringens group, before and after solids foods, were slower in BF infants compared with
			FF and MF.
			• Introduction of solid foods correlated with ↑ in proportions of C. leptum, C. coccoides and ↓ in proportions of bifidobacteria,
			C. difficile + C. perfringens, and enterobacteria.
			• Bifidobacteria were more diverse post-weaning but remained dominant
			• Proportions of Bacteroides remained stable before and after weaning.
			• Greater post-weaning increase in C. leptum in infants who were previously FF.
			• The decrease in C. difficile + C. perfringens post-weaning was greater in those who had been FF compared with MF.
12 wk	BF (31), FF (27)	FISH-FC	• BF had lower proportions of Atopobium cluster, Lactobacillus, C leptum, and Streptococcus groups	53 (Spain)
			• High levels of Bifidobacterium were associated with lower levels of Enterobacteriaceae in both groups and lower levels of
			Streptococcus in BF and C. leptum in FF infants
			• Proportions of C. coccoides were positively associated with proportions of Atopobium cluster in BF and Streptococcus and C.
			perfringens + C. difficile in FF.
			• In BF, lower levels of Bacteroides were associated with higher levels of Lactobacillus.
			• Higher levels of Atopobium cluster were associated with higher levels of Bacteroides in FF but lower levels of in BF
6, 12 mo	BF (46), FF (50)	FISH	• At 6 mo, Bifidobacterium counts were greater in infants BF for the first 3 mo compared with those BF for less than 3 mos.	54 (Finland)
			• Infants still receiving breast milk at 6 mo had higher counts of Bifidobacterium and Lactobacillus/Enterococcus.
			• No differences in the counts of Bifidobacterium, Lactobacillus, and Enterococcus between groups at 12 mo.
1−18 mo	BF (7), FF (7)	FISH, DGGE	• Similarity within and between the feeding groups was higher 10 weeks post-weaning	55 (UK)
			• Bifidobacterium levels were higher in BF before and during the weaning period.
			• BF exhibited greater intraindividual changes, especially among Clostridium, throughout the study.
			• FF had higher levels of bacteria from Clostridium clusters throughout sampling.

Note. Abbreviations: BF, breastfed; FF, formula-fed; MF, mixed-fed (receiving both human milk and infant formula); EBF, exclusively breastfed; d, day; wk, week; mo, month; yr, year; VD, vaginal delivery; CS, cesarean section delivery; qPCR, real-time polymerase chain reaction; PCR, polymerase chain reaction; GM, goat's milk; CM, cow's milk; DGGE, denaturing gradient gel electrophoresis; TGGE, temperature gradient gel electrophoresis; RISA, ribosomal intergenic spacer analysis; T-RFLP, terminal restriction fragment length polymorphism; FISH, fluorescence in situ hybridization; RAPD, random amplified polymorphic DNA; FC, flow cytometry

Initially after birth, the newborn fecal microbiota is composed primarily of facultative anaerobes, such as Staphylococcus, Streptococcus, Lactobacillus, and Enterobacteriaceae,^1,27,56,57which are thought to consume oxygen and prime the GI tract for colonization with obligate anaerobes such as Bifidobacterium, Bacteroides, and Clostridium. From the first wk of life, different colonization patterns are present in BF and FF infants.^26,49 The fecal microbiota of BF infants is more stable over time,^26,39 and characterized by a lower alpha-diversity compared to FF infants.^15,23,31,33 Exclusive breastfeeding is inversely related to alpha-diversity in 3 mo-old infants.²³ Conversely, exclusive breastfeeding and breastfeeding duration,²³ as well as predominant breastfeeding for the first 3 mos,²⁵ are positively related to alpha-diversity at 12 mos.

Many studies have reported that BF infants are colonized with higher proportions of microbes from the phylum Actinobacteria and less Firmicutes than FF infants.^{23,25,26,28,33} Other studies report that BF infants have similar levels of Actinobacteria and Firmicutes^28,30 but lower proportions of Bacteroidetes^31,36 and Proteobacteria^26,28 compared to FF infants.

At the genus level, Bifidobacterium has long been described to predominate in BF versus FF infants¹⁹; however, results are varied. Several studies found bifidobacteria to be significantly greater in BF infants^{18,23,26,28,33,36,37,39,47,51}; Fan and colleagues²⁶ reported roughly double the relative abundance of bifidobacteria in BF than FF infants. Bifidobacteria have been shown to account for 70% of the sequences of exclusively breastfed (EBF) infants⁴⁰ and appear earlier in the feces of EBF than in FF infants³⁸ In addition, infants who were exclusively breastfed in early life maintained greater colonization with bifidobacteria later in infancy.⁵⁴ Although this evidence is substantial, other findings suggest that Bifidobacterium exists in similar proportions^18,34 and numbers^44,46 in BF and FF infants and dominate the microbiota across both groups.^34,50,51 Conversely, Palmer and colleagues² reported very low levels of bifidobacteria in all infants with varied dietary patterns; however, these findings are likely due to the 8F universal primer that was used in the study having a three base pair mismatch against B. longum, and that Bifidobacterium genus in general does not have 100% sequence identity to the 8F primer sequence.⁵⁸ Infants harboring similar relative abundances of Bifidobacterium still exhibit some variance in diversity,^34,39,42 and stability⁵¹ of populations at the species level. For example, B. longum and B. breve are the dominant species in BF infants,^27,41-43 whereas FF infants also harbor adult-associated bifidobacterial species, such as B. adolescentis.^24,39,42

While many studies have focused on bifidobacterial colonization, breastfeeding supports other differences in microbiota composition. Within the phyla Bacteroidetes and Firmicutes, BF infants contain lower levels of Bacteroides,^33,50-53 Atopobium,^51,53 Clostridiales, Lachnospiraceae, and Faecalibacterium than FF.³³ Fan and colleagues²⁶ found that Bifidobacterium, Streptococcus, and Lactobacillus were dominant members in both feeding groups sampled between 3 and 6 mos; other dominant genera included Clostridium in BF infants and Klebsiella and Enterococcus in FF infants. Bergstrom and coworkers⁴⁵ reported a positive association between breastfeeding duration and levels of Lactobacillus spp.; however, effects of feeding on Lactobacillus abundance are inconsistent across other studies. Increased proportions of Lactobacillus,^25,26,48,54 as well as greater abundances of L. gasseri^28,47 in BF compared to FF infants have been reported, whereas other studies found opposite results^16,52,53 or reported intergroup variability and instability over time.^27,43

A greater prevalence¹⁵ and higher proportion²⁴ and total counts^16,46 of C. difficile in FF compared to BF infants, as well as, significantly more Peptostreptococcaceae,³¹ Akkermansia,¹⁵ Veillonella, and Enterococcus²⁸ have been described. Escherichia abundance was also higher in FF than BF infants²⁸ and decreased at a slower rate over time.³¹ Furthermore, other results have indicated E. coli to be dominant across both groups⁵⁰ or present in similar total numbers,⁵¹ while Penders et al.¹⁶ found this species significantly increased in FF infants at 4 wks of age. Bacteroides abundance has been shown to increase over time in BF infants; however, the timing of initial colonization is variable among infants.²⁷ Jost et al.²⁷ also found that EBF infants who had higher proportions of and were colonized earlier with Bacteroides also harbored higher proportions of pathogenic bacteria, such as Klebsiella and Clostridium.

Few studies have investigated the fecal microbial composition of mixed-fed (MF) infants – those receiving both HM and infant formula.^29,52 The dietary patterns of MF infants are highly variable within and across studies, and are often not well described, which presents an obstacle when trying to draw conclusions relative to the effect of the proportion of HM and formula consumed on microbial composition. At 6 wks of age, MF infants exhibited the largest within group compositional differences and harbored a microbiota significantly different than BF but not FF infants.²⁹ As further shown in Table 1, MF infants tend to group with FF infants and differ from BF infants.^16,18,47 Conversely, Fallani and associates⁵² detected, in a much larger sample of infants, significantly higher proportions of bifidobacteria in MF compared to FF infants of the same age. Azad and colleagues²³ also found breastfeeding exclusivity, at 3 mos, to be associated with fecal microbiota composition at 3 mos and 1 yr of age, with MF infants exhibiting increased abundances of Bacteroidetes, Lachnospiraceae, Ruminococcaceae and Verrucomicrobia and decreased abundances of Enterobacteriaceae and Bifidobacteriaceae at 3 mos compared to EBF infants. Differences in Bifidobacteriaceae and Bacteroidetes abundance still remained at 1 yr of age; Veillonellaceae abundance was also increased in EBF infants. Mixed feeding is a common dietary pattern in early life, and these wide-ranging results emphasize the need for further exploration of the fecal microbiota of infants receiving both HM and formula. Additional studies investigating the long-term impacts of early feeding regimes on GI profiles beyond infancy are also necessary, as existing evidence has demonstrated that infants who were fed with HM for at least 50% of feedings during the first 3 mos of life maintained a different fecal microbial composition between 12 and 24 mos than those who had received less than 50% of feedings from HM.⁵⁴ In future studies, it will be important to carefully document the relative proportions of HM and formula the MF infant is consuming in addition to when the formula was introduced to the infant. For example, the microbiota of an infant who was mixed-fed from birth would likely differ from an infant who was exclusively breastfed from birth until 3 mos of age at which time formula was introduced.

Diet also interacts with other environmental factors and life events during infancy to affect GI colonization. Breastfeeding may compensate for factors shown to negatively impact the infant's GI microbiota. For example, infants delivered by Cesarean section (CS) had different fecal microbiota depending on if they had been breast- or formula-fed.⁴⁷ The BF infants achieved numbers of B. bifidum comparable to those of vaginally-delivered (VD) infants earlier than FF or MF infants. Findings have also shown that infant feeding likely moderates the effects of intrapartum antibiotic prophylaxis (IAP) exposure on microbial colonization (Table 1).²³ Regardless of feeding mode, IAP-exposed infants, delivered by emergency CS, presented with dybioses at 3 mos. Compared to infants who were not exposed to IAP, these infants had a lower abundance of Bacteroidetes and increased abundance of Firmicutes. These differences, between IAP-exposed and non-exposed, remained at 1 yr, but only in infants who were not exclusively breastfed at 3 mos. Results, at 1 yr, were significant when compared with VD infants not exposed to IAP and of either feeding type at 3 mos. These findings further support the notion that breastfeeding may help to reconcile imbalances in an infant's microbial composition resulting from adverse life events.

To date, few have described functional changes accompanying microbial variations between BF and FF infants. Some investigators have utilized metagenomics, an analysis of the microbial genetic material, in order to identify possible roles in various metabolic pathways and therefore functions of the microbes in a community.²⁰ Bäckhed and coworkers²⁴ used Kyoto Encyclopedia of Genes and Genomes (KEGG) orthology (KO) modules to identify functional differences, according to feeding pattern, in the stool microbiome of 4 mo-old infants. The FF infant microbiome was enriched in functions characteristic of a more mature microbiota (e.g. more similar to that of an adult) such as bile acid synthesis, methanogenesis, and the phosphotransferase system, while the fecal microbiome of BF infants was enriched in KO modules associated with synthesis of B vitamins and oxidative phosphorylation. In addition, virulence factors of the microbiota of BF and FF infants affect the expression of immunity and defense genes in the host GI epithelium,³⁰ potentially contributing to programming of the intestinal immunological tone. More recently, Bokulich and colleagues²⁵ showed maturational differences, measured by age-dependent stages characterized by specific taxa that serve as markers for the rate of microbiota development, in BF and FF infant microbiota. These assessments are of interest to researchers as delayed development is often linked to physiological perturbations. Over the first 6 mos of life, FF infants matured at a faster rate, compared to BF infants, as indicated by a lower abundance of several gene pathways involved in metabolism; however, between the ages of 12–24 mos, these FF infants exhibited decreased microbiota maturation. These findings illustrate the potential impact of microbial composition to developmentally program host function and should direct the efforts of future research studies.

While it is clear that HM plays a substantial role in establishing the fecal microbiota, details regarding the interactions between HM components and infant microbes that contribute to compositional and functional differences among infants remain to be determined. Complex oligosaccharides and bacteria present in HM may contribute, but do not explain all compositional differences.

Human milk microbiota

The microbes in HM are hypothesized to be one of the contributors to the observed differences in the fecal microbiota of BF vs. FF infants. In the last two decades, HM, which was once thought to be sterile, has been shown to be a source of potentially probiotic bacteria for the infant.⁵⁹ Human milk contains bacterial genera also present in the infant fecal microbiota, including Staphylococcus, Streptococcus, Lactobacillus, and Bifidobacterium.^60-63 Hunt and coworkers⁶⁴ described a “core” HM microbiota comprised of nine OTUs that were present in all three samples collected from each mother and accounted for roughly half of the relative abundance of total bacteria in all HM samples. The remaining 50% consisted of several different genera, highlighting that each mother has a unique HM microbiota. These findings were further supported by a study showing that the HM microbiota composition was associated with mode of delivery, stage of lactation, and maternal health status.⁶⁵ Both Hunt et al.⁶⁴ and Cabrera-Rubio et al.⁶⁵ only detected few sequences of Bifidobacterium upon analysis of milk samples. However, other studies have identified this genus in over 90% of samples taken from mothers⁶¹ and have found associations between levels of bifidobacteria in maternal milk and infant feces,⁶⁶ which may partly explain the increased levels of Bifidobacterium found in the feces of BF infants. Although taxonomic differences could be attributed to primers used for analysis, as mentioned previously, incomplete cell lysis during DNA isolation, due to the thick bacterial cell wall, has been recognized to influence the ability to detect bifidobacteria in samples. This may be another factor which led to varying results seen among milk, as well as fecal microbiota, and has prompted many researchers to include a bead-beating step to adequately lyse cells.⁵⁸ Further studies have also detected identical strains of Lactobacillus, Bifidobacterium, and Staphylococcus in both HM and infant feces, reaffirming HM as a potential source of probiotic and commensal bacteria involved in colonization of the immature infant GI tract.^61,67

Culture-independent methods have detected more than 700 different bacterial species in milk, including obligate anaerobes and butyrate-producers.^59,65 However, the origin of the HM microbiota remains unclear. Previously, the bacteria present in HM were thought to be a result of contamination from the infant's oral cavity, the nipple and surrounding breast skin, as a result of backflow of milk into the mammary ducts during feeding.⁶⁴ While it is still hypothesized that breast skin may contribute a portion of the bacteria found in HM, notable differences found in bacterial phylotypes between HM and the breast skin of the same mother indicate additional sources for milk bacteria.⁶⁴ The presence of bacterial DNA in HM samples before breastfeeding initiation and the subsequent lack of typical neonate oral bacteria in HM after infants began to feed from the breast support a similar conclusion.⁶² Identification of identical strains of bacteria in maternal feces, HM, and infant feces from the same mother-infant pair have led researchers to the concept of an “enteromammary pathway” through which bacteria translocate from the maternal GI tract to the mammary gland via dendritic cells.^59,68

Jost et al.⁵⁹ proposed several possible functions of the HM microbes. Certain strains of bacteria previously isolated from HM produce bacteriocins and others prevent growth of various GI-associated pathogens, which could contribute to the decreased prevalence of illness and other health problems observed in BF vs. FF infants.⁵⁹ Bacterial genera identified in milk using molecular, but not culture-dependent, methods suggest that bacterial DNA in HM may correspond to dead bacteria. While there could be other reasons explaining this lack of identification with culture-dependent methods, including sample collection or culture techniques, this may be a result of the interaction of these bacteria with antimicrobial components in milk or of an out competition by other bacterial species.⁵⁹

Further research on the HM microbiota is needed to reveal how they interact with other components found in HM and how these constituents work together to facilitate healthy maturation of the infant GI microbiota.^69,70 Continued investigation of how the HM microbiota differs between mothers as well as the effects of transient exposure to certain bacteria, without colonization, will provide further insight into the complex and long-lasting functional differences witnessed between growing infants.⁷⁰

Human milk oligosaccharides

Differences in the colonization patterns and microbial composition in BF vs. FF infants are proposed to be guided by complex sugars present in HM of each mother, known as human milk oligosaccharides (HMO). These sugars are the third largest component of HM and have been identified in over 200 distinct forms.⁷¹ Abundance of HMOs decreases throughout lactation and maternal genetic differences determine both the composition and orientations of the 5 monosaccharides known to comprise these glycans – L-fucose, D-glucose, D-galactose, N-acetylglucosamine, and N-acetylneuraminic acid.^71,72 Being resistant to enzymatic hydrolysis, the HMOs pass intact through the infant stomach and upper GI tract to the distal small intestine and colon. HMOs exert prebiotic effects and inhibit the binding of pathogenic bacteria, thereby modulating the infant immune system and shaping the composition of the resident GI microbes.^72,73

Of particular interest to researchers is the ability of Bifidobacterium to metabolize HMOs. Certain HMO show bifidogenic effects in vitro,^72,74 selectively stimulating the growth of certain commensal bacteria often identified in BF infants, such as B. infantis, B. bifidum, B. breve and B. longum.^34,42 Additionally, HMOs also serve as a carbon source for various species of Bacteroides common to the infant GI tract, but do not function as a substrate for other members of the Firmicutes phylum or Bifidobacterium species primarily present in adults, such as B. adolescentis and B. animalis; these microbes lack the enzymatic capacity to break down and consume HMOs.^59,72,75 A recent study published by Wang et al.¹⁷ revealed both positive and negative associations between multiple HMOs and the relative abundance of bacterial genera, suggesting that HMO profiles can predict infant fecal microbial composition. Certain species, namely B. infantis, exhibit the ability to grow in the presence of a wide range of HMOs, while other species of Bifidobacterium and Bacteroides metabolize fewer structural types.^17,72 Moreover, Wang et al.¹⁷ hypothesize that the inverse relationship detected between Bifidobacterium and Bacteroides, in their study as well as by other earlier investigators,^27,53 may be due to the ability of different species to thrive in the presence of particular HMOs. Therefore, differences in selectivity for certain HMOs likely contribute to variance among species of Bifidobacterium and Bacteroides isolated from feces of BF infants. Lewis et al.⁷⁶ showed that bifidobacterial colonization was delayed in infants consuming HM from non-secretor mothers that is devoid of a particular type of HMO, 2´-fucosyllactose (2´FL), further supporting this theory regarding a possible origin for variations in Bifidobacterium observed among BF infants. This further leads to the question of how these HMOs interact with commensal bacteria residing in HM, and in turn, how this complex network of HM components functions in the näive infant GI tract. Lastly, HMOs have been implicated in upregulating the expression of genes related to carbohydrate utilization,⁷⁷ effects which may continue to be important later in infancy.²⁶

Weaning, complementary feeding, and introduction of solids

The impact of the introduction of solids and weaning has been understudied in regards to their effects on infant microbiota composition and function. The transition toward a more adult-like microbiota is observed during the introduction of solids foods^1,2,35; however, more recent studies suggest that this shift toward a more mature microbiota is more likely attributed to the cessation of breastfeeding, rather than the introduction of complementary foods.^24,45 Additional findings demonstrate that pre-weaning feeding mode remains associated with the post-weaning microbial communities.^18,33,55 For example, continued breastfeeding during the introduction of solid foods seems to support consistent levels of Lactobacillus spp. and Bifidobacterium spp.^{24,40,45,47,55} (Table 1).

Thompson et al.³³ found that EBF infants and infants receiving HM and solid foods (EBF + S) clustered closer together than EBF and non-EBF infants. The addition of solid foods also appeared to have a more dramatic impact on the fecal microbiota of non-EBF infants. While there were marked differences between the EBF and EBF + S groups, the separation between clusters of non-EBF and non-EBF + S groups was more distinct. The lack of any significant microbiota changes accompanying the introduction of complementary foods reported in BF infants by some groups^22,43 further supports the impact of HM withdrawal. Others have found that infants still receiving HM at 9 mos harbored lower levels of butyrate-producing bacteria, C. leptum, C. coccoides, and Roseburia, as well as Bacteroidetes⁴⁵ which have been previously noted to increase when infants begin to consume solid foods.^1,20 Infants still partially breastfed at 9 mos, compared to those already weaned, also had an increased relative abundance of Bifidobacterium spp. and Lactobacillus spp.⁴⁵ Furthermore, in the cohort studied by Bäckhed et al.,²⁴ it was not until infants stopped breastfeeding that their microbiota showed the functional capacity to degrade polysaccharides, which is characteristic of a shift toward a more adult-like microbiota. Taken together, these findings have led to the idea that it is the withdrawal of HM that triggers the compositional maturation of the infant GI tract, marked by an increase in members of the Firmicutes and Bacteroidetes phyla, with some studies noting specific increases in Clostridium, Roseburia, Bacteroides, Bilophila, and Anaerostipes in feces.^{1,24,33,45,47} Valles et al.,³⁵ who proposed that colonization can be separated into 2 phases, before and after the introduction of solid foods, still found Bifidobacterium and Bacteroides to be dominant in samples taken before and after all infants had begun to eat solid foods, which could be due to the fact that 70% of infants were still being breastfed when the second sample was taken.

The introduction of solid foods has been associated with increases in Bacteroidetes and Firmicutes phyla. As outlined in Table 1, several changes at the genus level have been reported, with some common findings including increases in Atopobium, Clostridium, Akkermansia, Bacteroides, and Ruminococcus and decreases in Escherichia and Staphylococcus.^33,35 Thompson et al.³³ also found that all infants, despite pre-weaning feeding type, had a microbiota dominated by Actinobacteria, Firmicutes, and Bacteroidetes, respectively, after the introduction of solids. However, as previously described, whether or not the infants had been previously breastfed or mixed- or formula-fed did appear to have an impact at the family and genus levels, both during complementary feeding and weaning periods. In EBF infants, introduction of solids was associated with increases in Eggerthella, Blautia, Neisseria, Peptostreptococcaeceae and Bacteroidetes and led to higher post-weaning abundances of Lactobacillus and Ruminococcaceae.³³ Both FF and MF infants experienced significant increases in Actinobacteria, specifically Bifidobacterium, Ruminococcaceae, and Blautia at the start of complementary feeding and showed greater abundances of Faecalibacterium, Eubacterium, and Blautia post-weaning compared to infants who had been previously exclusively breastfed.³³ Fallani et al.¹⁸ found similar results, showing that feeding mode at 6 wks was associated with long-term differences in microbial composition. Those who were breastfed at 6 wks had higher proportions of fecal Bifidobacterium and lower proportions of Bacteroides and C. coccoides at 4 wks post-weaning compared to infants receiving any formula. Those previously formula-fed also had a greater increase in C. leptum post-weaning than BF infants.¹⁸ Bifidobacterium populations remain relatively consistent throughout the introduction of solids^18,35,40,42; however, some infants experience an increase in diversity at the species level throughout complementary feeding and weaning.^18,40,42

Beyond compositional shifts, perhaps the most important outcomes of these dietary modifications are the functional changes that accompany complementary feeding and weaning as infants are exposed to several metabolic substrates for the first time, which promote microbial growth that was not supported by HM and/or formula.³³ Bäckhed et al.²⁴ noted an increase in genes involved in carbohydrate and pyruvate metabolism at 12 mos, supporting new ways of obtaining energy from the increasing percentage of solid foods. Additionally, an increase in short chain fatty acid (SCFA)-producing species was observed in accordance with an elevated intake of starch and plant fibers,^1,35 and an increased expression of genes involved in amino acid metabolism and vitamin biosynthesis followed complementary feeding and weaning.^1,33 Analysis of 16s sequencing data using PICRUSt (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States) to infer metabolic effects associated with the fecal microbial community upon the introduction of solid foods, further revealed that these changes also differ among infants receiving HM or formula prior to solid foods.³³ An increase in a variety of genes associated with metabolism pathways, for example, biosynthesis of secondary metabolites and energy metabolism, was observed in the microbiota of EBF infants. In contrast, the introduction of solids coincided with enriched expression of cellular process pathways, such as genetic information processing and immune system diseases, in the microbiota of MF and FF infants. Overall, infants who were not exclusively breastfed experienced significantly more functional changes after complementary feeding began.³³ These researchers hypothesized that the consistently changing composition of HM could explain why functional differences were less dramatic in BF infants, and that some of the differentially overrepresented genes in non-EBF infants may help to explain the protective effect of breastfeeding against certain diseases.³³

Prebiotics and probiotics

As noted above, human milk contains microbes and a diverse complement of oligosaccharides. Infant formulas are sterilized, killing any live microbes. In addition, bovine milk, the base of most infant formula, contains only trace amounts of less complex oligosaccharides.⁷¹ Thus, in an attempt to narrow the compositional gap between human milk and infant formula, prebiotics and probiotics have been supplemented to formula and functional outcomes have been assessed.

Prebiotics

The prebiotic concept was introduced by Glenn Gibson and Marcel Roberfroid, who defined a prebiotic as ‘a non-digestible food ingredient that beneficially affects the host by selectively stimulating the growth and/or activity of one or a limited number of bacteria in the colon and thus improves host health’.⁷⁸ Since then, the definition has been refined several times.^79-82 Most recently, Bindels and colleagues⁸³ proposed a new definition for prebiotics as ‘a nondigestible compound that, through its metabolization by microorganisms in the gut, modulates composition and/or activity of the gut microbiota, thus conferring a beneficial physiological effect on the host’. Most previous definitions required the prebiotic to be selective for stimulation of growth and/or activities of health-promoting bacteria, while the new definition removed the selectivity criterion and places more emphasis on the causal link between the microbial metabolism of the compound, the resulting modulation of the GI microbe and the beneficial physiology effects.⁸³

Several in vitro studies have shown that HMOs serve as the primary substrate for the growth of Bifidobacterium and other bacteria.^84,85 Furthermore, we¹⁷ and others^86,87 have demonstrated correlations between several bacterial genera present in BF infant feces, with the HMO in their mother's milk¹⁷ or present in the infant feces.^86,87 Until recently, HMOs have not been commercially available in large quantities; therefore, other prebiotics that share some of the functional attributes of HMOs are currently supplemented to infant formula.

The most studied prebiotic addition to infant formula is a 9:1 mixture of short-chain galactooligosaccharides (scGOS) and long-chain fructooligosaccharides (lcFOS). Investigators have identified significant clinical impacts of these prebiotics on the immune and metabolic development of infants. Infants who consume formula supplemented with this mixture have a lower fecal pH and exhibit a fecal SCFA profile and stool characteristics comparable to that of BF infants. Findings have also shown that infants exposed to scGOS and lcFOS to require fewer doses of antibiotics and have a decreased incidence of infection likely due to an increased colonization resistance to enteric pathogens.^88-90 The other prebiotics added to infant formula individually or in combination include, GOS, FOS, oligofructose and inulin, polydextrose (PDX), lactulose (LOS) and acidic oligosaccharides (AOS) (Table 2). Prebiotics are resistant to gastric acidity and enzymatic hydrolysis in the upper gastrointestinal tract and enter the colon intact, where they are metabolized by colonic microbiota.⁷⁸ Short-chain prebiotics, such as FOS and GOS, are mainly fermented in the ascending colon, while longer-chain prebiotics, like PDX and inulin, are fermented along the entire colon.¹¹⁴

Table 2.

Impact of prebiotic supplementation on the composition of infant microbiota.

Prebiotic (dose)Source	Nutrition base	Subject (feeding duration)	Study group (n)	Major effect of the prebiotic	References (Country)
GOS (2.4 g/L)	Infant formula	Healthy term infant < 28 d (3 mo)	GOS (20)	• ↑ counts of bifidobacteria and lactobacilli	91 (China)
Frisolac Advanced, Friesland			Control (18)	• ↑ acetic acid level and stool frequency
Nutrition, Netherlands			HM reference (15)	• ↓ fecal pH
			GOS + HM (29)
GOS (4 g/L) Vivinal GOS; Friesland Foods Domo, Amersfoort,	Infant formula	Healthy term infants at 15 d (up to 70 d of age)	GOS (83)	• ↓ Clostridium count	92 (Italy)
The Netherlands			Control (80)
			HM reference (199)
GOS (4.4 g/L)	Infant formula	Healthy term infants < 8 wk (until 4 mo of age), microbiota were analyzed at 4 mo of age	GOS (40)	• ↑ Bifidobacterium count	93 (Spain)
Source not reported			Control (29)	• ↑ proportion of acetate
				• ↑ stool frequency and softness
				• ↓ fecal pH, proportions of propionate and butyrate
GOS (2 g/L)	Infant formula	Healthy term infants	GOS (43)	• ↑ counts of bifidobacteria and lactobacilli	94 (Not reported)
Source not reported			Control (17)
			HM reference (20)
GOS (3 g/L)	Infant formula	Healthy term infants at 31–54 d (2 wk)	GOS (9)	• ↑ Bifidobacterium abundance	95 (Japan)
OM55N; Yakult Pharmaceutical			Control (13)	• ↓ Shannon index
Industry, Tokyo, Japan
FOS (1.5 or 3 g/L)	Infant formula	Healthy term infants at 2–6 wk (1wk)	FOS 1.5 and FOS 3 as crossover (58)	• The counts of Bifidobacterium, Lactobacillus and clostridia were similar among the 2 FOS groups	96 (US)
Raftilose®P95; Orafti, Tienen, Belgium			HM reference (14)	• FOS 3 had softer stool than FOS 1.5.
FOS (4 g/L)	Infant formula	Preterm infants at 0–14 d (14 d), microbiota were analyzed at study d 7	FOS (36)	• ↑ numbers of bifidobacteria and bacteroides	97 (Greece)
Source not reported			Maltodextrins (20)	• ↑ stool frequency
				• ↓ counts of E. coli and enterococci
scFOS (4 g/L)	Infant formula	Healthy term infants at 0–7 d (4 mo)	FOS (18)	• Greater increase in Bifidobacterium counts after 2 mo feeding	98 (France)
Acilight 950P; Beghin Meiji, Marckolsheim, France	Maltodextrins (15)
OF (3 g/L)	Infant formula	Health term infants at 5–14 d (8 wk)	OF (20)	• Greater increase in Bifidobacterium	99 (US)
OraftiR P95; BENEO-Orafti			Control (19)	• Softer stool
			HM reference (23)
FOS (2 or 3 g/ L)	Infant formula	Healthy term infants at 0–6 d (4 wk)	FOS 2 (14)	• No effects on population of bifidobacteria, lactobacilli, C. difficile, E. coli or Bacteroides.	100 (US)
Source not reported			FOS 3 (20)
			Control (14)
			HM reference (17)
FOS (4.5 g/L)	Growing –up Formula	Infants with antibiotic treatment at 1–2 y (3 wk)	FOS (57)	• ↑counts of bifidobacteria after 1 wk	101 (Chile)
(OF and inulin 70:30 ratio)			Control (56)	• The other bacteria were unaffected.
Raftilose P95 and Raftiline in 70:30 ratio, Orafti
FOS (0.75 g/25 g cereal)	Cereal	Infants at 4–11 mos (28 d)	FOS (27)	• ↑ stool frequency and softness	102 (France)
Source not reported			Maltodextrin (29)
OF (2 g/ d)	Cereal	Healthy children at 6–24 months (21 d)	OF (10)	• ↓ Clostridium levels	103 (US)
Beneo P95; Orafti			Maltodextrin (12)	• Trend for increase of Bifidobacterium
OF/Inulin (8 g/L, 1:1 ratio)	Infant formula	Healthy term newborn (until 4 months of age)	OF/inulin(63)	• ↑ Bifidobacterium counts	104 (Spain)
Synergy 1; BENEO-Orafti			Control (68)	• Softer stools and a higher deposition frequency
			HM as reference (57)
OF/inulin (4 or 8 g/L, 1:1 ratio)	Infant formula	Healthy term infants ≤ 5 d (28d)	OF/inulin 4 (21)	• Greater increase in Bifidobacterium counts with 8 g/L of OF/inulin	105 (Belgium)
Synergy 1; BENEO-Orafti			OF/inulin 8 (20)	• Softer stool
			Control (21)
			scGOS+lcFOS (19)
			HM as reference (29)
Inulin, 1.5 g/d	Infant formula	Healthy infants at 5–24 wk (3wk)	Inulin and Control as crossover (14)	• ↑ count s of Bifidobacterium and Lactobacillus	106 (Korea)
Frutafit IQ, Roosendaal, Netherlands
Inulin (0.75, 1.00 or 1.25 g/d)	Follow-up formula	Healthy infants at 5–12 mo (14d)	Inulin 0.75 (10)	• ↓ clostridia counts and fecal pH	107 (Malaysia)
Frutafit IQ			Inulin 1 (9)	• ↑ bifidobacteria numbers with 1.25 g/d inulin
			Inulin 1.25 (9)
			Control (8)
LOS (5 or 10 g/L)	Infant formula	Healthy infants at 2–10 wk (3 wk)	LOS 5, LOS 10 and Control as crossover (6)	• ↑ Bifidobacterium counts	108 (Not Reported)
Produced by alkaline				• ↓ fecal pH
isomerization of lactose
PDX/GOS (4 g/L, 1:1 ratio)	Infant formula	Healthy term infants at d14 (until 120 d of age)	PDX/GOS (58)	• Softer stool	109 (US)
PDX/GOS/LOS (8 g/L. 50:33:17 ratio)			PDX/GOS/LOS(48)
Source not reported			Control (58)
PDX/ GOS (4 g/L, 1:1 ratio)	Infant formula	Healthy term infants at 21–30 d (60 d)	GOS/PDX (78)	• ↑ Bifidobacterium counts	110 (US)
Litesse Two PDX; Danisco, Beaminster, UK) and Vivinal GOS;Friesland Foods Domo, Zwolle, The Netherlands			Control (81)	• Softer stool
			HM as reference (71)
PDX/GOS (4 g/L; 1:1 ratio)	Infant formula	Healthy term infants at 13–92 d (28 d)	PDX/GOS (23)	• No impact on the bacterial composition were observed	111 (US)
PDX/GOS/LOS (4 or 8 g/L, 3:2:1 ratio)				PDX/GOS/LOS 4 (23)
Litesse Two PDX; Danisco, Copenhagen, Denmark; and Vivinal GOS; Friesland Foods Domo, Zwolle, The Netherlands and Morinage Milk Industry, Tokyo, Japan			PDX/GOS/LOS 8 (21)
			Control (21)
			HM as reference (29)
AOS(2 g/L)	Infant formula	Healthy term newborns (6 wk)	AOS (14)	• ↓ stool pH; softer stool	112 (Italy)
AOS/GOS/FOS (2 g/L, 25:67.5:7.5 ratio)			AOS/GOS/FOS (13)	• AOS/GOS/FOS also ↑ Bifidobacterium and Lactobacillus counts and stool frequency
AOS from Richter Pharma,Wels, Austria; the source for GOS/FOS are unknown			Maltodextrin (14)	• No effect on the numbers of Clostridia, E.coli, Proteus, Klebsiella, Pseudomonas, Enterobacter, Citrobacter, Enterococci
AOS/scGOS/lcFOS/	Enteral supplementation	Preterm infants < 2d (until 30 d of age)	scGOS/lcFOS/AOS (55)	• ↑ total bacteria count at d14	113 (Netherlands)
(1.5 g/kg/d, 20:72:8 ratio)			maltodextrin (58)	• ↓ fecal pH
Danone Research, Friedrichsdorf, Germany

Note. AOS, acidic oligosaccharides; E. coli, Escherichia coli; FOS, fructooligosaccharide, GOS, galactooligosaccharide; OF, Oligofructose; LOS, lactulose, OF; PDX, polydextrose

Clinical studies show that prebiotics modulate the composition of the infant fecal microbiota (Table 2, ^88,90). Most studies showed that supplementation of prebiotics to infant formula increased the numbers of beneficial bacteria, mainly Bifidobacterium and sometimes Lactobacillus (Table 2, ^88,90). Some studies also demonstrated a decrease in the levels of opportunistic pathogens, such as Escherichia coli, Enterococcus and clostridia with prebiotic supplementation.^92,97,103 As ‘selectivity’ was previously considered as a key characteristic of prebiotics, many studies focused on documenting changes in the abundance of specific groups/species of bacteria, such as Bifidobacterium and Lactobacillus, using traditional plate counting or by molecular methods, such as real-time quantitative PCR and fluorescence in situ hybridization.^91-94,96,99 Fewer studies have assessed changes in overall fecal microbial communities by next-generation sequencing or microarray approaches.⁹⁵ With the new prebiotic definition,⁸³ in the future, research should shift from a focus on selective microbial group/species toward ecological and functional characteristics of the GI microbiota and on defining the mechanisms whereby any health benefits of prebiotics are achieved.

Short-chain fatty acids (SCFAs), the fermentation products of anaerobic bacteria, are commonly used to reflect the metabolic activity of the intestinal microbiota. While the role of SCFAs in the development of the infant GI tract are still being investigated, they are known to be absorbed as a source of metabolizable energy while also exerting anti-inflammatory effects on the colonic epithelium and contributing to defense against gut-associated pathogens. Additionally, SCFAs are involved in enteroendocrine hormone regulation and may be involved in de novo lipid synthesis.⁹⁰ The fecal SCFA profile represents events that occur in the colon and is closely related to the composition of intestinal microbiota.⁹³ For example, a positive correlation has been reported between the abundance of Bifidobacterium and fecal acetate level; while high propionate and butyrate content in the infant feces may imply the presence of a more complex microbiota, as propionate and butyrate, among others, are produced by the members of Bacteroides and Clostridium.⁹³ Several studies have investigated the effect of prebiotics on infant fecal SCFA content (Table 2). In a randomized, double blind, placebo-controlled study, infants who were fed formula with added scGOS/lcFOS (8 g/L) for 6 wks had an increase in proportion of acetate and a decrease in proportion of propionate compared to infants fed control formula, which represented a fermentation profile similar to that of BF infants.¹¹⁵ Similarly, a higher proportion of acetate and lower proportions of propionate and butyrate were observed in infants fed formula containing GOS (4.4 g/L) for 2 mos compared with infants fed control formula.⁹³ Thus, supplementation of certain prebiotics to infant formula could modulate the fecal SCFA content to a level as human milk does.

The SCFAs are the principal luminal anions in humans and an increase in their concentrations reduces luminal pH, resulting in more acidic stools, which in turn leads to a mild laxative effect with increased stool frequency and softness.^116,117 BF infants generally have lower fecal pH and softer and more frequent stools compared to FF infants^91,92,118; therefore, several clinical studies have evaluated the influence of prebiotics on stool characteristics of infants (Table 2). Infants fed formula supplemented with GOS (2.4 or 4.4 g/L) had lower fecal pH, as seen in BF infants, in comparison with infants fed non-supplemented formula.^91,93 Similar patterns were also observed when inulin, LOS, AOS, scGOS/lcFOS or scFOS/lcFOS/AOS was administrated to FF infants.^{107,108,112,113} The effect of prebiotics on stool frequency and consistency has been evaluated in a systematic review.¹¹⁷ Four randomized controlled trials reported that the stool frequency of infants fed prebiotics was similar to that of BF infants and was higher than that of infants fed control formula.¹¹⁷ In addition, six studies assessed the stool consistency after prebiotic supplementation, and all reported that the stools were softer in infants consuming prebiotics.¹¹⁷ Taken together, the addition of prebiotics to infant formula could decrease stool pH and increase stool frequency and softness, making the stool characteristics of FF infants resemble those of BF infants.

Probiotics

Probiotics are “live microorganisms which, when consumed in adequate amounts, confer a health benefit on the host.”¹¹⁹ Most studies conducted with probiotics in pediatrics were carried out with strains of Bifidobacterium and Lactobacillus isolated from the human GI tract or dairy products (Table 3). A few studies used bacteria isolated from HM.^138,139 Several reasons have been used to support probiotic use in infant formula. Human milk, as previously described, contains a variety of potential probiotic bacteria, such as Bifidobacterium, Lactobacillus and Streptococcus, and these bacteria serve as a source of continuous inoculum to the BF infant GI tract, which may partly contribute to differences in the fecal microbial composition between BF and FF infants.^62,64 Furthermore, administration of specific probiotic bacteria has been shown to improve infant health, including shortening the duration of rotavirus diarrhea,¹⁴⁰ preventing of antibiotic-associated diarrhea,¹⁴¹ reducing the incidence of eczema in high-risk children¹⁴² and decreasing the risk of necrotizing enterocolitis in very low birth weight infants.¹⁴³ Thus, supplementation of probiotics to infant formula has been promoted as an approach to improve infant health and to mimic the probiotic property of human milk. Although the mechanisms by which the probiotic strains exert their positive impact on infant health are unclear, it could be related to their direct interaction with the host or indirect through their modulation of the GI microbiota.

Table 3.

Impact of probiotic supplementation on the composition of infant microbiota.

Probiotic strain (dose)	Nutrition base	Subject (feeding duration)*	Study groups (n)	Major effect of the probiotic	References (Country)
L. johnsonii La1 (108CFU/g)	Infant formula	Healthy term infants at 4 mo (13 wk)	Probiotic (48)	• ↑ fecal Lactobacillus count	120 (Chile)
			FOS (44)
			Control (61)
			BF as reference (46)
LGG (4.5 –8.5 × 107 CFU/g)	Hydrolyzed casein formula	Infant with cow's milk allergy at 1–12 mo (6 mo)	Probiotic (12)	• ↑ proportions of Roseburia and Anaerofustis	121 (Italy)
			Control (7)	• ↑ fecal butyrate level
			Healthy infants (20)
L. rhamnosus LPR + B. longum BL999	Hydrolyzed formula	Infants with familial atopic background >1 mo (4 mo), fecal microbiota were analyzed at 6 mo of age	LPR + BL999 (24)	• No effect on the proportions of Bifidobacterium, Bacteroides-Prevotella, C. histolyticum, Lactobacilluse-Enterococcus or A. muciniphila	122 (Germany)
or			BL999 (25)	• ↓ colonization rate of B. bifidum in BL999
B. longum BL999,			Control (32)
(109 CFU/strain/d)			HM as reference (8)
L. acidophilus NCFM or B. lactis Bi-07 (1010CFU/d)	In a capsule	Children with atopic dermatitis at 7–24 mo (8 wk)	NCFM (17)	• No effect on the composition and diversity of the main bacterial populations	123 (Denmark)
			Bi-07 (17)	• ↑ numbers of Lactobacillus in NCFM
			Control (16)
LGG + B. longum BB536 (2 × 107 and 107CFU/g, respectively)	Infant formula	Newborn infant at high risk of allergy (6 mo)	Probiotic (17)	• The proportions of the major bacterial groups were similar	124 (Singapore)
			Control (20)	• Bacterial colonization pattern was unaffected by probiotics
				• ↑ detection rates of B. longum and L. rhamnosus
				• ↑ lactic acid bacteria counts
L. fermentum CECT5716 (107CFU/g)	Infant formula	Healthy infants at 1 mos (5 mo), fecal samples were analyzed at 3 y of age	Probiotic (61)	• No effect on the fecal counts of lactobacilli, bifidobacteria, clostridia and Bacteroidaceae	125 (Spain)
			Control (60)
B. lactis LKM512 (6 × 109 CFU/d)	Oral feeding	Infants required surgery ≤ 3 d (until 15 d of age)	LKM+ (4)	• ↑ Streptococcaceae in LKM+	126 (Japan)
			LKM- (4)	• ↓ Bifidobacteriaceae in LKM+
			Control (no surgery, 5)
L. reuteri DSM 17938 (1.2 ×109 CFU/L)	Infant formula	Healthy VD or CD term infants ≤ 3d (6 mo)	CSLr (11)	VD infants	127 (Greece)
				• No effect on overall microbiota composition
			VDLr (9)	• ↑ Proportions of Lactobacillus at 2wk and 4 mo.
			CSCt(10)	• ↑ Abundance of Coprococcus at 4 mo
			VDCt (10)	CS infants
				• ↑ Proportion of Actinobacteria toward VD infants at 2 wk
				• ↓ unclassified Enterobacteriaceae toward VD infants at 2 wk
				• ↑ Proportions of Lactobacillus at 2 wk and 4 mo
B. longum BB536 (107 CFU/g)	Infant formula	Healthy term infants at 0–7 d (12 mo)	Probiotic (135)	• ↑ bifidobacteria level at 2 and 4 mo	128 (China)
			Control (129)
L. acidophilus LAVRI-A1 (3 × 109CFU/d)	Powder in water	Health term newborn at high risk of allergy (6 mo)	Probiotic (89)	• ↑ rate of Lactobacillus colonization	129 (Australia)
			Maltodextrin (89)
L. reuteri DSM 17938 (108 CFU/d)	Oil drop	BF colicky infants at 10–60 d (21d)	Probiotic (15)	• No effect on the overall composition of the microbiota	130 (Italy)
			Control (14)
B. breve M-16V (5 × 108 CFU/d) or	Powder in water	Low birth weight infants ≤ 7 d (6 wk)	M-16V (15)	• ↑ detection rate and numbers of Bifidobacterium	131 (Japan)
B. breve M-16V + B. infantis M-63 + B. longum BB536 (5 × 108CFU/strain/d)			M-16V+M-63+BB536 (13)	• ↓ detection rate of C. perfringens
			Control (16)	• ↓ proportion of Enterobacteriaceae in M-16V+M-63+BB536
LGG (6 × 109 CFU/d)	Preterm infant formula	Preterm infants at 0–3 d (42 d)	Probiotic (21)	• ↑ colonization rates of Lactobacillus, Enterobactericeae, Enterococcus and staphylococci	132 (Poland)
			Maltodextrins (26)
B. lactis Bb12(1.6–4.8 × 109 CFU/d)	Infant formula	Preterm infants at birth (21d)	Probiotic (37)	• ↑ numbers of Bifidobacterium	133 (Germany)
			Control (32)	• ↓ counts of Enterobacteriaceae and Clostridium spp
B. longum BB536 + LGG (4 × 108 CFU/d)	Powder in wate	Preterm infants when enteral feeding started (until discharge)	Probiotic (45)	• No alteration in the composition of intestinal microbiota, except for increase in the colonization rates of Bifidobacterium and Lactobacillus	134 (France)
			Maltodextrin (49)
LGG (1.8 × 1010 CFU/d)	In a capsule	Mothers of infants at high risk of allergy at 36 wk gestation (until delivery), infant microbiota were followed until 180 d of age	Probiotic (59)	• ↑ prevalance of B. longum group at 90 d	135 (Australia)
			Maltodextrin (57)	• Concentration of Bifidobacterium was similar at any time
B. bifidum W23 + B. lactis W52+ L. lactis W58 (109 CFU/d/strain)	Powder	Mothers of infants at high risk of allergy (6 wk before delivery); infants at birth (1y), infant microbiota were followed up until 6 y	Probiotic (15–26)	• Bacterial composition and a diversity were similar during first 6 y, except relative abundances and Shannon indices of Bacteroidetes and Proteobacteria were lower at 2 wk	136 (Netherlands)
			Maltodextrin (9–12)
L. rhamnosus LPR + B. longum BL999 or L. paracasei ST11+ B. longum BL999 (109 CFU/strain/d)	Powder in Water	Mothers of infants (2 mo before and 2 mo after delivery during breast feeding), infant microbiota were analyzed at 6 mo of age	LPR + BL999 (28)	• ↑ percentage of Lactobacillus-Enterococcus and ↓Bifidobacterium counts in LPR + BL999 at 6 mos	122 (Finland)
			ST11 + BL999 (29)	• ↓ colonization rate of B. infantis in LPR + BL999
			Control (22)	• ↓ colonization rate of B. longum in STII + BL999
LGG (1.8 × 1010 CFU/d)	Mothers: capsule; BF infants:powder in water	Mother of infants with high risk of atopy (2–4 wks before delivery); BF mothers or FF infants after delivery (6 mos), infant microbiota were analyzed at 6 and 24 mo of age	Probiotic (46)	At 6 mos · No effect on the counts of Bifidobacterium, Bacteroides, Lactobacillus/Enterococcus and total bacteria	137 (Finland)
			Control(47)	• ↑ clostridia
				At 24 mo
				• ↓ Lactobacillus/Enterococcus
				• ↓ clostridia
				• No effect on Bifidobacterium, Bacteroides and total bacteria counts

Note. AD, atopic dermatitis; BF, breastfed ; CS, cesarean section; CFU, colony-forming units; Ct, Control; FF, formula-fed; LGG, L. rhamnosus GG; L. lactis, Lactococcus lactis; VD, vaginal delivered. The information before the parentheses indicates the time point, or age of the infant, at which the probiotic treatment began. The information presented in the parentheses is the duration for which the probiotic was administered to the infant or mother. Unless otherwise stated, fecal samples analyzed, were those collected after the completion of the probiotic administration.

Clinical evidence has shown that administration of probiotics to pregnant women and nursing mothers can influence the establishment and composition of infant fecal microbiota (Table 3). Supplementation of L. rhamnosus GG to mothers of infants at high risk of allergy during late pregnancy has been shown to increase the prevalence of B. longum group in their infants at 3 mos of age.¹³⁵ A Finnish study investigated the effect of probiotic administration (L. rhamnosus LPR + B. longum BL999) to mothers for the 2 mos prior to and 2 mos after delivery. At 6 mos of age, BF infants whose mothers who received the probiotic treatment had higher percentages of Lactobacillus-Enterococcus and lower levels of Bifidobacterium compared to infants whose mothers received a placebo control.¹²²

Probiotics have also been administrated directly to infants; however, their effects on intestinal microbiota are contradictory (Table 3). For example, Mohan and coworkers¹³³ evaluated supplementation for modifying the GI microbiota of preterm infants. They found that the numbers of Bifidobacterium were higher and the counts of Enterobacteriaceae and Clostridium spp. were lower in B. lactis Bb12 group than in the placebo group after 21 days of feeding. In a randomized double-blind placebo-controlled intervention, administration of L. acidophilus NCFM or B. lactis Bi-07 to infants with atopic dermatitis for 8 wks did not affect the composition and diversity of the main bacterial populations in feces.¹²³ Similarly, no effect on the overall microbiota composition was observed when L. reuteri DSM 17938 was administrated to BF colicky infants.¹³⁰ These inconsistent results may relate to differences in probiotic strain used, daily doses administered, timing and duration of administration, and the methods applied for microbiota analysis. In addition, environmental factors that influence the development of infant microbiota might also affect the outcomes. A recent study monitored the effect of probiotic intervention on the microbiota colonization of infants with different delivery mode. In this study, infants delivered by CS or VD received (within 72h of birth) either control formula or the same formula containing L. reuteri DSM 17938 for 6 mos and the fecal microbiota was analyzed at 2 wks- and 4 mos of age. They reported administration of L. reuteri DSM 17938 did not modify the overall microbial composition of VD infants; however, it increased proportion of Actinobacteria and decreased the relative abundance of unclassified Enterobacteriaceae of CD infants, making the microbiota of the CS infants more similar to the composition of VD infants.¹²⁷ The results suggest when a study is designed to describe the impact of the probiotic intervention on the infant microbiota, not only daily doses and timing and duration of administration need to be considered; factors that affect the microbial colonization in early life, such as delivery mode, infant diet and age, should also be controlled.

Conclusions and future directions

The infant GI microbiota undergoes rapid and profound changes during the first year of life. During this process, diet plays a predominant role over other environmental factors in shaping the microbial composition. While this review has outlined significant advances in our understanding of the role of early nutrition in the development of infant GI microbiota, large gaps in our understanding of the specific factors regulating this process still exist. For example, information about the effects of mixed feeding on the composition and function of infant microbiota is scarce. Our knowledge of how feeding mode, weaning, and solid foods, as well as the timing of dietary shifts, influence the functional capacities of infant GI microbiota and host programming is also limited. As reviewed by Chu and colleagues in this supplement,¹⁴⁴ maternal nutrition during pregnancy and lactation has the potential to effect the composition and function of the microbiota of the offspring. Although compelling, much of the evidence is derived from animal models, thus, additional studies in humans and potential intervention trials are needed to assess whether maternal nutritional status and diet lead to changes in infant microbial colonization.

Looking forward, longitudinal studies are necessary to investigate how nutrition (breast-, mixed-, and formula-feeding, solid foods) interacts with other early life events (route of delivery, antibiotics, daycare, pets, etc) to impact microbial composition and function and subsequent effects on health outcomes during and beyond infancy. Given that infant microbiota composition differs based on geography, sampling from a large cohort of infants located in various sites around the globe in which detailed records of dietary intake and other environmental factors are collected is needed to provide a more complete understanding of the development of the microbiota in infants who are breast- vs. formula-fed.

Studies are also needed to further elucidate how human milk components other than HMOs, such as cytokines, macronutrients, and antimicrobial peptides, may work individually or synergistically to support specific GI microbial profiles as well as the potential for pre- and probiotics in human milk or added to formula to preferentially modulate the infant GI tract to a desired composition. Future intervention studies should aim for consistent timing, duration, and dose amounts of any supplemental ingredients (pre or probiotics), which is instrumental in isolating the effects of different feeding regimes on GI microbial colonization during early life.

Disclosure of potential conflicts of interest

The authors report no conflicts of interest

Funding

The authors report no sources of funding to support this work