Multimodal machine learning for modeling infant head circumference, mothers' milk composition, and their shared environment.

Links between human milk (HM) and infant development are poorly understood and often focus on individual HM components. Here we apply multi-modal predictive machine learning to study HM and head circumference (a proxy for brain development) among 1022 mother-infant dyads of the CHILD Cohort. We integrated HM data (19 oligosaccharides, 28 fatty acids, 3 hormones, 28 chemokines) with maternal and infant demographic, health, dietary and home environment data. Head circumference was significantly predictable at 3 and 12 months. Two of the most associated features were HM n3-polyunsaturated fatty acid C22:6n3 (docosahexaenoic acid, DHA; p = 9.6e-05) and maternal intake of fish (p = 4.1e-03), a key dietary source of DHA with established relationships to brain function. Thus, using a systems biology approach, we identified meaningful relationships between HM and brain development, which validates our statistical approach, gives credence to the novel associations we observed, and sets the foundation for further research with additional cohorts and HM analytes.

Variation in Human Milk Composition Is Related to Differences in Milk and Infant Fecal Microbial Communities.

Previously published data from our group and others demonstrate that human milk oligosaccharide (HMOs), as well as milk and infant fecal microbial profiles, vary by geography. However, little is known about the geographical variation of other milk-borne factors, such as lactose and protein, as well as the associations among these factors and microbial community structures in milk and infant feces. Here, we characterized and contrasted concentrations of milk-borne lactose, protein, and HMOs, and examined their associations with milk and infant fecal microbiomes in samples collected in 11 geographically diverse sites. Although geographical site was strongly associated with milk and infant fecal microbiomes, both sample types assorted into a smaller number of community state types based on shared microbial profiles. Similar to HMOs, concentrations of lactose and protein also varied by geography. Concentrations of HMOs, lactose, and protein were associated with differences in the microbial community structures of milk and infant feces and in the abundance of specific taxa. Taken together, these data suggest that the composition of human milk, even when produced by relatively healthy women, differs based on geographical boundaries and that concentrations of HMOs, lactose, and protein in milk are related to variation in milk and infant fecal microbial communities.

Human milk oligosaccharide profiles and allergic disease up to 18 years.

BACKGROUND: Human milk oligosaccharides (HMO) are a diverse range of sugars secreted in breast milk that have direct and indirect effects on immunity. The profiles of HMOs produced differ between mothers. OBJECTIVE: We sought to determine the relationship between maternal HMO profiles and offspring allergic diseases up to age 18 years. METHODS: Colostrum and early lactation milk samples were collected from 285 mothers enrolled in a high-allergy-risk birth cohort, the Melbourne Atopy Cohort Study. Nineteen HMOs were measured. Profiles/patterns of maternal HMOs were determined using LCA. Details of allergic disease outcomes including sensitization, wheeze, asthma, and eczema were collected at multiple follow-ups up to age 18 years. Adjusted logistic regression analyses and generalized estimating equations were used to determine the relationship between HMO profiles and allergy. RESULTS: The levels of several HMOs were highly correlated with each other. LCA determined 7 distinct maternal milk profiles with memberships of 10% and 20%. Compared with offspring exposed to the neutral Lewis HMO profile, exposure to acidic Lewis HMOs was associated with a higher risk of allergic disease and asthma over childhood (odds ratio asthma at 18 years, 5.82; 95% CI, 1.59-21.23), whereas exposure to the acidic-predominant profile was associated with a reduced risk of food sensitization (OR at 12 years, 0.08; 95% CI, 0.01-0.67). CONCLUSIONS: In this high-allergy-risk birth cohort, some profiles of HMOs were associated with increased and some with decreased allergic disease risks over childhood. Further studies are needed to confirm these findings and realize the potential for intervention.

Breastmilk Feeding Practices Are Associated with the Co-Occurrence of Bacteria in Mothers' Milk and the Infant Gut: the CHILD Cohort Study.

Gut microbiota play a critical role in infant health. It is now accepted that breastmilk contains live bacteria from endogenous and exogenous sources, but it remains unclear whether these bacteria transfer to the infant gut and whether this process is influenced by breastmilk feeding practices. Here, we show that certain bacteria, including Streptococcus spp. and Veillonella dispar, co-occur in mothers' milk and their infants' stool, and co-occurrence is reduced when infants receive pumped breastmilk. The relative abundances of commonly shared species are positively correlated between breastmilk and stool. Overall, gut microbiota composition is strongly associated with breastfeeding exclusivity and duration but not breastmilk feeding mode (nursing versus pumping). Moreover, breastmilk bacteria contributed to overall gut microbiota variation to a similar extent as other modifiers of the infant microbiome, such as birth mode. These results provide evidence that breastmilk may transfer bacteria to the infant gut and influence microbiota development.

Exercise-induced 3'-sialyllactose in breast milk is a critical mediator to improve metabolic health and cardiac function in mouse offspring.

Poor maternal environments, such as under- or overnutrition, can increase the risk for the development of obesity, type 2 diabetes and cardiovascular disease in offspring1-9. Recent studies in animal models have shown that maternal exercise before and during pregnancy abolishes the age-related development of impaired glucose metabolism10-15, decreased cardiovascular function16 and increased adiposity11,15; however, the underlying mechanisms for maternal exercise to improve offspring's health have not been identified. In the present study, we identify an exercise-induced increase in the oligosaccharide 3'-sialyllactose (3'-SL) in milk in humans and mice, and show that the beneficial effects of maternal exercise on mouse offspring's metabolic health and cardiac function are mediated by 3'-SL. In global 3'-SL knockout mice (3'-SL-/-), maternal exercise training failed to improve offspring metabolic health or cardiac function in mice. There was no beneficial effect of maternal exercise on wild-type offspring who consumed milk from exercise-trained 3'-SL-/- dams, whereas supplementing 3'-SL during lactation to wild-type mice improved metabolic health and cardiac function in offspring during adulthood. Importantly, supplementation of 3'-SL negated the detrimental effects of a high-fat diet on body composition and metabolism. The present study reveals a critical role for the oligosaccharide 3'-SL in milk to mediate the effects of maternal exercise on offspring's health. 3'-SL supplementation is a potential therapeutic approach to combat the development of obesity, type 2 diabetes and cardiovascular disease.

Human milk fungi: environmental determinants and inter-kingdom associations with milk bacteria in the CHILD Cohort Study.

BACKGROUND: Fungi constitute an important yet frequently neglected component of the human microbiota with a possible role in health and disease. Fungi and bacteria colonise the infant gastrointestinal tract in parallel, yet most infant microbiome studies have ignored fungi. Milk is a source of diverse and viable bacteria, but few studies have assessed the diversity of fungi in human milk. RESULTS: Here we profiled mycobiota in milk from 271 mothers in the CHILD birth cohort and detected fungi in 58 (21.4%). Samples containing detectable fungi were dominated by Candida, Alternaria, and Rhodotorula, and had lower concentrations of two human milk oligosaccharides (disialyllacto-N-tetraose and lacto-N-hexaose). The presence of milk fungi was associated with multiple outdoor environmental features (city, population density, and season), maternal atopy, and early-life antibiotic exposure. In addition, despite a strong positive correlation between bacterial and fungal richness, there was a co-exclusion pattern between the most abundant fungus (Candida) and most of the core bacterial genera. CONCLUSION: We profiled human milk mycobiota in a well-characterised cohort of mother-infant dyads and provide evidence of possible host-environment interactions in fungal inoculation. Further research is required to establish the role of breastfeeding in delivering fungi to the developing infant, and to assess the health impact of the milk microbiota in its entirety, including both bacterial and fungal components.

Development of a biochemical marker to detect current breast milk intake.

The WHO recommends exclusive breastfeeding for 6 months, but despite interventions, breastfeeding rates remain stubbornly low. Financial voucher incentives have shown promise but require a biomarker for validation of intake. This study aimed to develop a simple biochemical assay of infant urine that would tell if an infant was receiving any breast milk to validate maternal report. Urine samples were collected and snap frozen from 34 infants attending with minor illness or feeding problems, of whom 12 infants were exclusively breastfed, nine exclusively formula fed, and 11 mixed breast/formula fed. High-performance anion exchange chromatography was used to identify discriminating patterns of monosaccharide composition of unconjugated glycans in a sequence of three experiments. The absolute concentration of all human milk oligosaccharides measured blind could detect "any breastfeeding" only with a sensitivity of 48% and specificity of 78%. Unblinded examination of N-acetylglucosamine (GlcNAc) measured as GlcNH2 after hydrolysis of GlcNAc improved sensitivity to 75% at the expense of a specificity of 28%. Estimation of the relative abundance of GlcNH2 (GlcNH2[%]) or the ratio of GlcNH2 to endogenous mannose (Man) improved accuracy. In a further blind experiment, the GlcNH2/Man ratio with a cut-off of 1.5 correctly identified all those receiving "any breast milk," while excluding exclusively formula fed infants. The GlcNH2/Man ratio in infant urine is a promising test to provide biochemical confirmation of any breastfeeding for trials of breastfeeding promotion.

Integrated Analysis of Human Milk Microbiota With Oligosaccharides and Fatty Acids in the CHILD Cohort.

Background: Human milk contains many bioactive components that are typically studied in isolation, including bacteria. We performed an integrated analysis of human milk oligosaccharides and fatty acids to explore their associations with milk microbiota. Methods: We studied a sub-sample of 393 mothers in the CHILD birth cohort. Milk was collected at 3-4 months postpartum. Microbiota was analyzed by 16S rRNA gene V4 sequencing. Oligosaccharides and fatty acids were analyzed by rapid high-throughput high performance and gas liquid chromatography, respectively. Dimension reduction was performed with principal component analysis for oligosaccharides and fatty acids. Center log-ratio transformation was applied to all three components. Associations between components were assessed using Spearman rank correlation, network visualization, multivariable linear regression, redundancy analysis, and structural equation modeling. P-values were adjusted for multiple comparisons. Key covariates were considered, including fucosyltransferase-2 (FUT2) secretor status of mother and infant, method of feeding (direct breastfeeding or pumped breast milk), and maternal fish oil supplement use. Results: Overall, correlations were strongest between milk components of the same type. For example, FUT2-dependent HMOs were positively correlated with each other, and Staphylococcus was negatively correlated with other core taxa. Some associations were also observed between components of different types. Using redundancy analysis and structural equation modeling, the overall milk fatty acid profile was significantly associated with milk microbiota composition. In addition, some individual fatty acids [22:6n3 (docosahexaenoic acid), 22:5n3, 20:5n3, 17:0, 18:0] and oligosaccharides (fucosyl-lacto-N-hexaose, lacto-N-hexaose, lacto-N-fucopentaose I) were associated with microbiota α diversity, while others (C18:0, 3'-sialyllactose, disialyl-lacto-N-tetraose) were associated with overall microbiota composition. Only a few significant associations between individual HMOs and microbiota were observed; notably, among mothers using breast pumps, Bifidobacterium prevalence was associated with lower abundances of disialyl-lacto-N-hexaose. Additionally, among non-secretor mothers, Staphylococcus was positively correlated with sialylated HMOs. Conclusion: Using multiple approaches to integrate and analyse milk microbiota, oligosaccharides, and fatty acids, we observed several associations between different milk components and microbiota, some of which were modified by secretor status and/or breastfeeding practices. Additional research is needed to further validate and mechanistically characterize these associations and determine their relevance to infant gut and respiratory microbiota development and health.

Composition and Variation of the Human Milk Microbiota Are Influenced by Maternal and Early-Life Factors.

Breastmilk contains a complex community of bacteria that may help seed the infant gut microbiota. The composition and determinants of milk microbiota are poorly understood. Among 393 mother-infant dyads from the CHILD cohort, we found that milk microbiota at 3-4 months postpartum was dominated by inversely correlated Proteobacteria and Firmicutes, and exhibited discrete compositional patterns. Milk microbiota composition and diversity were associated with maternal factors (BMI, parity, and mode of delivery), breastfeeding practices, and other milk components in a sex-specific manner. Causal modeling identified mode of breastfeeding as a key determinant of milk microbiota composition. Specifically, providing pumped breastmilk was consistently associated with multiple microbiota parameters including enrichment of potential pathogens and depletion of bifidobacteria. Further, these data support the retrograde inoculation hypothesis, whereby the infant oral cavity impacts the milk microbiota. Collectively, these results identify features and determinants of human milk microbiota composition, with potential implications for infant health and development.

Human milk oligosaccharides, milk microbiome and infant gut microbiome modulate neonatal rotavirus infection.

Neonatal rotavirus infections are predominantly asymptomatic. While an association with gastrointestinal symptoms has been described in some settings, factors influencing differences in clinical presentation are not well understood. Using multidisciplinary approaches, we show that a complex interplay between human milk oligosaccharides (HMOs), milk microbiome, and infant gut microbiome impacts neonatal rotavirus infections. Validating in vitro studies where HMOs are not decoy receptors for neonatal strain G10P[11], population studies show significantly higher levels of Lacto-N-tetraose (LNT), 2'-fucosyllactose (2'FL), and 6'-siallylactose (6'SL) in milk from mothers of rotavirus-positive neonates with gastrointestinal symptoms. Further, these HMOs correlate with abundance of Enterobacter/Klebsiella in maternal milk and infant stool. Specific HMOs also improve the infectivity of a neonatal strain-derived rotavirus vaccine. This study provides molecular and translational insight into host factors influencing neonatal rotavirus infections and identifies maternal components that could promote the performance of live, attenuated rotavirus vaccines.

Infants Are Exposed to Human Milk Oligosaccharides Already in utero.

Human milk oligosaccharides (HMOs) are complex carbohydrates that are highly abundant in and, in their complexity, unique to human milk. Accumulating evidence indicates that exposure to HMOs in the postnatal period affects immediate as well as long-term infant health and development. However, studies reported in the 1970s showed that HMOs already appear in maternal urine and blood during pregnancy and as early as the first trimester. In this pilot study we aimed to determine whether or not HMOs also appear in amniotic fluid. We enrolled women during pregnancy and collected their urine and amniotic fluid at birth as well as their milk 4 days postpartum. We analyzed the samples by high-performance liquid chromatography (HPLC) and mass spectrometry and identified several HMOs including 2'-fucosyllactose, 3-fucosyllactose, difucosyllactose, and 6'-sialyllactose to be present in different relative abundancies in all three tissues. This is the first report that HMOs appear in amniotic fluid and that the fetus is already exposed to HMOs in utero, warranting future research to investigate the immediate and long-term implications on fetal and infant health and development.

Human Milk Oligosaccharide Concentrations Are Associated with Multiple Fixed and Modifiable Maternal Characteristics, Environmental Factors, and Feeding Practices.

Background: Human milk oligosaccharides (HMOs) shape the developing gut microbiome and influence immune function. Aside from genetic Secretor status, the factors influencing HMO synthesis and secretion are largely unknown. Objective: We aimed to identify modifiable and nonmodifiable factors associated with HMO concentrations. Methods: This prospective observational study included a representative subset of 427 mothers participating in the CHILD birth cohort (mean age: 33 y, 73% Caucasian). Breast milk was collected at 3-4 mo postpartum. Concentrations of 19 predominant HMOs were measured by rapid high-throughput HPLC. Secretor status was defined by the presence of 2'-fucosylactose. Associations with maternal, infant, and environmental factors were explored using multivariable regression. Breastfeeding duration was explored as a secondary outcome. Results: Overall, 72% of mothers were Secretors and the mean ± SD duration of any breastfeeding was 12.8 ± 5.7 mo. HMO profiles were highly variable; total HMO concentrations varied 3.7-fold and individual HMOs varied 20- to >100-fold. Secretor mothers had higher total HMO concentrations than did non-Secretors (mean: 15.91 ± 2.80 compared with 8.94 ± 1.51 µmol/mL, P < 0.001) and all individual HMOs differed by Secretor status, except for disialyllacto-N-tetraose (DSLNT). Most HMO concentrations were lower in milk collected later in lactation, although some were higher including DSLNT and 3'-sialyllactose. Independent of Secretor status and lactation stage, seasonal and geographic variation was observed for several HMOs. Parity, ethnicity, and breastfeeding exclusivity also emerged as independent factors associated with some HMOs, whereas diet quality and mode of delivery did not. Together, these factors explained between 14% (for 6'-sialyllactose) and 92% (for 2'-fucosyllactose) of the observed variation in HMO concentrations. Lower concentrations of lacto-N-hexaose or fucodisialyllacto-N-hexaose were associated with earlier breastfeeding cessation. Conclusions: HMO concentrations vary widely between mothers and are associated with multiple characteristics beyond genetic Secretor status, as well as feeding practices and environmental factors. Further research is warranted to determine how these associations affect infant health. This study was registered at clinicaltrials.gov as NCT03225534.

Enzymatic and Chemoenzymatic Syntheses of Disialyl Glycans and Their Necrotizing Enterocolitis Preventing Effects.

Necrotizing enterocolitis (NEC) is one of the most common and devastating intestinal disorders in preterm infants. Therapies to meet the clinical needs for this special and highly vulnerable population are extremely limited. A specific human milk oligosaccharide (HMO), disialyllacto-N-tetraose (DSLNT), was shown to contribute to the beneficial effects of breastfeeding as it prevented NEC in a neonatal rat model and was associated with lower NEC risk in a human clinical cohort study. Herein, gram-scale synthesis of two DSLNT analogs previously shown to have NEC preventing effect is described. In addition, four novel disialyl glycans have been designed and synthesized by enzymatic or chemoenzymatic methods. Noticeably, two disialyl tetraoses have been produced by enzymatic sialylation of chemically synthesized thioethyl ß-disaccharides followed by removal of the thioethyl aglycon. Dose-dependent and single-dose comparison studies showed varying NEC-preventing effects of the disialyl glycans in neonatal rats. This study helps to refine the structure requirement of the NEC-preventing effect of disialyl glycans and provides important dose-dependent information for using DSLNT analogs as potential therapeutics for NEC prevention in preterm infants.

What's normal? Oligosaccharide concentrations and profiles in milk produced by healthy women vary geographically.

Background: Human milk is a complex fluid comprised of myriad substances, with one of the most abundant substances being a group of complex carbohydrates referred to as human milk oligosaccharides (HMOs). There has been some evidence that HMO profiles differ in populations, but few studies have rigorously explored this variability.Objectives: We tested the hypothesis that HMO profiles differ in diverse populations of healthy women. Next, we examined relations between HMO and maternal anthropometric and reproductive indexes and indirectly examined whether differences were likely related to genetic or environmental variations.Design: In this cross-sectional, observational study, milk was collected from a total of 410 healthy, breastfeeding women in 11 international cohorts and analyzed for HMOs by using high-performance liquid chromatography.Results: There was an effect of the cohort (P < 0.05) on concentrations of almost all HMOs. For instance, the mean 3-fucosyllactose concentration was >4 times higher in milk collected in Sweden than in milk collected in rural Gambia (mean ± SEM: 473 ± 55 compared with 103 ± 16 nmol/mL, respectively; P < 0.05), and disialyllacto-N-tetraose (DSLNT) concentrations ranged from 216 ± 14 nmol/mL (in Sweden) to 870 ± 68 nmol/mL (in rural Gambia) (P < 0.05). Maternal age, time postpartum, weight, and body mass index were all correlated with several HMOs, and multiple differences in HMOs [e.g., lacto-N-neotetrose and DSLNT] were shown between ethnically similar (and likely genetically similar) populations who were living in different locations, which suggests that the environment may play a role in regulating the synthesis of HMOs.Conclusions: The results of this study support our hypothesis that normal HMO concentrations and profiles vary geographically, even in healthy women. Targeted genomic analyses are required to determine whether these differences are due at least in part to genetic variation. A careful examination of sociocultural, behavioral, and environmental factors is needed to determine their roles in this regard. This study was registered at clinicaltrials.gov as NCT02670278.