Animal models play a vital role in nutrition and health research because they provide a means to study physiology and metabolism in vivo, bridging the gap between basic science and clinical studies. The successful use of animal models has contributed significantly to the development of nutritional guidelines (1), as well as improved diagnostic and therapeutic approaches in cases of disease (2). Nonhuman primates offer a unique opportunity because their genetic homology, physiology, and behavior are closely related to that of humans. In particular, rhesus macaques (Macaca mulatta) are ideal because they share 93% DNA sequence identity (3), are easy to maintain in captivity, and have a lifespan that closely mimics human biology on a shorter timescale. Like humans, rhesus monkeys are susceptible to infectious and metabolic diseases (3), and have been shown to develop adult onset obesity and related comorbidities (4), making them suitable for long-term dietary and disease-related research. Rhesus macaque reproduction is also similar to human reproduction (5), and infant rhesus monkeys experience similar programming events both in utero and during the postnatal period (6). Their nutritional needs reflect those of human infants and for this reason, rhesus infants have been used extensively in dietary studies (7,8) and are considered by most the best possible animal model for human infant nutrition research (9); however, despite the body of research that exists, few studies to date have incorporated metabolomic and microbiota sequencing technologies when comparing rhesus and human profiles.
Metabolomics is the study of small molecules or metabolites in biological samples. In recent years a number of metabolomic studies in humans have successfully identified metabolites and metabolic pathways that differentiate healthy and diseased (10), dietary habits (11), and response to intervention (12). Metabolomic techniques have also been widely applied in mechanistic studies with animals; however, the majority of these studies focus on rodent models. To our knowledge, only 1 metabolomics study involving a rhesus model of human disease has been published, which showed metabolic changes in urine resulting from type 2 diabetes mellitus (13) that were similar to changes observed in humans (14,15), in contrast to rodent models (14).
In the present study, the metabolomic and 16S rRNA gene profiles of milk from human and rhesus mothers, as well as the metabolomic profiles of urine and serum from human and rhesus infants are compared. The study validates rhesus macaques as a valuable model for human nutritional research.
The rhesus monkeys (Macaca mulatta) used for the study were provided by the California National Primate Research Center (CNPRC), 1 of 8 such centers in the United States supported by the National Institutes of Health for studying human health and disease. All of the monkeys were maintained at the CNPRC at the University of California, Davis, under the constant care of nursery and veterinary staff. Protocols for all the animal studies were approved by the University of California, Davis Institutional Animal Care and Use Committee and conducted in accordance with the Department of Agriculture Animal Welfare Act. All human samples were collected as part of ongoing research at UC Davis. The UC Davis institutional review board approved the study protocol, and the subjects consented to the study before participation. Research was conducted in accordance with the ethical standards outlined in the Helsinki Declaration.
Blood and urine samples were collected from 5 infant rhesus macaques that were fed milk from their mothers and were maintained indoors at the CNPRC. Blood samples were drawn via femoral venipuncture at 8 weeks of age. Samples were allowed to clot at room temperature for 30 minutes. Following centrifugation, serum aliquots were stored at −80°C until analysis. Urine was collected from the same rhesus infants, and 3 monkeys provided samples at 2 time points, between 3 and 12 weeks of age. In some cases, urine was collected immediately as the rhesus infants were handled. If no urine was immediately produced, the rhesus infants were placed in a metabolic unit with mesh flooring engineered to effectively separate urine into a specially designed sample collection cup while retaining feces on the grate with little opportunity for contamination. Once collected, urine samples were frozen immediately at −20°C, followed by long-term storage at −80°C. Clinical practice and infant care standards dictate that infants are fed frequently and on demand. For this reason, rhesus infants were not fasted before sample collection. Milk samples were collected from 6 rhesus mothers at weeks 5 and 14 of lactation. The milk sample collection procedure has been described in detail previously (16). Briefly, mothers were lightly sedated (5-mg ketamine hydrochloride per kilogram body mass administered by intramuscular injection) and placed in mesh jackets (ProMed-Tec, Bellingham, MA) to prevent nursing and to allow milk accumulation for 3.5 to 4 hours. Nipple areas were cleaned and mothers were administered exogenous oxytocin for myoepithelial cell contraction and milk let down (0.1 mL/kg). Milk was collected separately from each mammary gland by gentle hand stripping of the nipple. Full mammary evacuation occurred within 10 to 15 minutes for all subjects. Once collected, all samples were frozen at −80°C. Samples were subsequently stored at −80°C. Given the nature of this protocol, it can be assumed that rhesus mothers were fasted for at least 4 hours before milk sample collection.
Blood samples were collected from 5 healthy term infants at 17 weeks of age. Samples were allowed to clot at room temperature for 30 minutes, and following centrifugation serum aliquots were stored at −80°C until analysis. Milk samples were collected from mothers of the same term infants at 17 weeks postpartum. All milk samples were collected by manual expression and immediately frozen at −20°C, followed by long-term storage at −80°C. Urine samples were collected from 8 healthy term infants ages between 5 and 24 weeks. Urine was collected into 3 cotton balls held within disposable diapers. The cotton balls were removed as soon as they were wet and immediately stored at −20°C. Soiled samples were discarded. Samples were moved to −80°C for long-term storage. As for rhesus infants, human infants were not fasted before sample collection.
Female monkeys were maintained on a commercial diet and fed twice daily (Monkey Lab Diet, PMI Nutrition, Brentwood, MI). The commercial diet was supplemented semiweekly with fresh fruit and vegetables. Water was available ad libitum. Rhesus infants were exclusively fed milk from their mothers. Human mothers followed their habitual diet and all human infants were exclusively breast-fed.
1H Nuclear Magnetic Resonance Spectroscopy
All samples were analyzed independently. Urine samples were removed from −80°C storage and allowed to thaw. For human infant samples, urine was collected from defrosted cotton balls by placing the cotton ball in the barrel of a syringe, inserting the plunger, and squeezing out the absorbed urine. Samples were prepared for analysis by the addition of an internal standard containing 5 mmol/L 3-(trimethylsilyl)-1-propanesulfonic acid-d6 (DSS-d6) and 0.2% NaN3 in 99.8% D2O; 65 μL was added to 585 μL of urine. The pH of each sample was adjusted to 6.8 ± 0.1 by adding small amounts of NaOH or HCl. A 600-μL aliquot was subsequently transferred to a 5-mm Bruker nuclear magnetic resonance (NMR) tube, and stored at 4°C until NMR acquisition (within 24 hours of sample preparation).
Serum and milk samples were prepared for metabolomic analysis as follows. After removal from −80°C storage and defrosted, each sample was filtered through a 3000 MW cutoff filter (Pall, Ann Arbor, MI) to remove lipids and proteins. The filtrate volume was adjusted to 585 μL with ultrapure water (Millipore Synergy UV system, Millipore, Billerica, MI) where required. Sixty-five microliters of internal standard was added and pH adjusted as described above. A 600-μL aliquot was subsequently transferred to a 5-mm Bruker NMR tube, and stored at 4°C until NMR acquisition (within 24 hours of sample preparation).
NMR spectra were acquired as previously described (17) on a Bruker Avance 600-MHz NMR equipped with a SampleJet autosampler (Bruker Biospin Corp, Fremont, CA) using a NOESY-presaturation pulse sequence (noesypr) at 25°C. Water saturation was achieved during the prescan delay (2.5 seconds) and mixing time (100 ms). Spectra were acquired with 8 dummy scans and 32 transients over a spectral width of 12 ppm with a total acquisition time of 2.5 seconds.
NMR Spectral Processing and Metabolite Identification
Once acquired, all spectra were zero-filled to 128k data points, Fourier transformed with a 0.5-Hz line broadening applied, and manually phased and baseline corrected using NMR Suite version 6.1 Processor (Chenomx Inc, Edmonton, Canada). Metabolite quantification was achieved using the 600-MHz library from Chenomx NMR Suite version 6.1 Profiler (Chenomx Inc), which uses the concentration of a known reference signal (in this case DSS) to determine the concentration of individual compounds as previously described (18). Metabolites were quantified in micromolar (μmol/L) units and exported from Chenomx for analysis. Where appropriate, corrections for sample dilution were made.
NMR Data Analysis
Data are presented in tables as fold change calculated based on median values. Statistical analyses were performed using PASW Statistics version 18.0 for Windows (SPSS Inc, Chicago, IL). Metabolite concentrations in human and rhesus samples were compared using the Mann-Whitney U test.
16S rRNA Gene Sequencing
DNA was extracted from human and rhesus monkey milk by centrifuging 1 mL of milk to pellet bacterial cells, followed by resuspension in 0.2 mL of ice-cold phosphate buffer saline (pH 7). DNA was extracted using the QIAamp DNA Stool Mini kit (Qiagen, Valencia, CA) according to the manufacturer's protocol with the following modifications: after addition of ASL buffer, samples were subjected to bead beating for 2 minutes using Mini-Beadbeater-16 (Biospec Products Inc, Bartlesville, OK) and subsequent heating for 5 minutes at 95°C. DNA extracts were stored at −20°C until further analysis.
Phylogenetic Analysis of 16S rRNA Gene Sequences
The V4 region of 16S rRNA genes was amplified using F515/806 primer pair modified according to Bokulich et al (19) with 5 pmol of each primer added for every polymerase chain reaction. Purified libraries were submitted to the UC Davis Genome Center DNA Technologies Core for a 150 bp on Illumina GAIIx platform. Paired-end Illumina sequencing of the larger V4 region has previously been successfully used to construct phylogenetic trees (20). Quantitative sequence analysis was performed using the Quantitative Insights Into Microbial Ecology suite of software tools version 1.5.0 (UCLUST, PyNAST, RDP and FastTree) (21) on the forward amplicons. Operational taxonomic unit (OTU) picking was performed using Quantitative Insights Into Microbial Ecology implemented in UCLUST (22) against the most recent version of the Greengenes (Lawrence Berkeley National Laboratory, Berkeley, CA) core database, and clustered with a threshold of 97% identity excluding reads that did not match the database. Taxonomy assignment was performed using the parameters as described in Yatsunenko et al (23). The most abundant sequence in each OTU was used as the representative sequence for comparison to Greengenes. Representative sequences for each OTU were chosen using Python Nearest Alignment Space Termination (PyNAST) (24,25) with a relaxed neighbor-joining tree built using FastTree (26). OTUs were classified taxonomically using the Ribosomal Database Project classifier with a 0.8 confidence threshold. A retrained Ribosomal Database Project classifier using taxa from the Greengenes reference at the genus level was used to obtain a better classification resolution to the genus level.
To compare the serum and urine metabolites between human infants and infant rhesus monkeys, 8 human infants (3 male, 5 female) between 5 and 24 weeks of age, and 5 infant rhesus monkeys (2 male, 3 female) between 3 and 17 weeks of age were recruited in the present study. To account for differences in the developmental age (27), samples from human infants were collected at approximately twice the rhesus age. To compare the milk metabolites, milk from 5 women at 16 to 18 weeks postpartum, and from 6 dams at 5 and 14 weeks postpartum, were collected. Inspection of the 1H NMR spectra from urine, serum, and milk samples highlighted several metabolic similarities between rhesus and human mother infant dyads.
Fifty-nine metabolites were quantified from urine 1H NMR profiles, and all metabolites present in human samples were also present in monkey samples (Fig. 1A). Approximately, 65% of the metabolites quantified had similar concentrations (Mann-Whitney U test, P > 0.05) for human and rhesus infants, including, several amino acids (glutamate, alanine, histidine, isoleucine, leucine, serine, threonine, tryptophan, tyrosine, valine, and 1-methylhistidine), sugars and derivatives (fucose, galactose, glucose, lactose, and galactitol), vitamins (ascorbate), tricarboxylic acid cycle intermediates (citrate, succinate, fumarate, 2-oxoglutarate, and cis-aconitate), microbial metabolites (acetate, formate, dimethylamine, methylamine, trimethylamine-N-oxide, methanol), ketones, and fatty acid metabolites (3-hydroxybutyrate and carnitine), as well as creatine, pyruvate, guanidoacetate, taurine, betaine, trigonelline, N,N-dimethylglycine, myo-inositol, and uracil. The remaining metabolites that were present at different concentrations are listed in Table 1. Concentration differences ranged from 2.5- (for lysine and trans-aconitate) to 17.5-fold (for 1-methylnicotinamide).
1H NMR spectra of serum samples obtained from human and rhesus infants are presented in Figure 1B. In total, 41 metabolites could be identified and quantified in 1H NMR spectra. Similar to urinary profiles, there were no metabolites in serum from one species that were not present in the other. Of the 41 metabolites that were quantified, 16 metabolites (∼40%) had similar concentrations (Mann-Whitney U test, P > 0.05) in human and rhesus samples including, amino acids (alanine, arginine, aspartate, histidine, isoleucine, leucine, lysine, threonine, valine), creatine, taurine, carnitine, cis-aconitate, dimethylamine, propylene glycol, and ethanol. Metabolite differences in human and rhesus infant serum are highlighted in Table 2. Fold differences between monkey and human serum ranged from 1.2 (for tyrosine) to 7.4 (for acetone) (Table 2).
The profile of metabolites present in human and rhesus milk samples was also comparable (Fig. 1C). All samples contained lactose as the major component at a concentration of approximately 0.15 to 0.20 mol/L, with low concentrations (micromolar to low millimolar range) of a variety of free amino acids and other nutrients and components, including oligosaccharides. Of the 47 metabolites that were quantified, 14 metabolites (∼30%) were present at similar concentrations (Mann-Whitney U test, P > 0.05) in human and rhesus samples. These included some amino acids (aspartate, glutamine, and histidine), essential nutrients (choline, and phosphocholine), 2-aminobutyrate, acetylcarnitine, creatinine, lactate, gut microbiota metabolites (dimethylamine, and methanol), and oligosaccharides (3′-sialyllactose, 6′-sialyllactose, and lacto-N-tetraose. The metabolites that had different concentrations in human and monkey milk are presented in Table 3. All of the amino acids that were different between human and monkey milk were higher in human milk. Interestingly, the branched-chain amino acids valine and leucine were both ∼4-fold higher in human milk, whereas isoleucine was ∼2-fold higher. Fold differences for 4 oligosaccharides are presented in Table 2, all of which are higher in human samples. Three additional oligosaccharides were identified in human samples (2′-fucosyllactose, lacto-N-fucopentaose I, and lactodifucotetraose), all of which were low or below the limit of detection in milk from rhesus mothers.
Milk Microbial Composition
Analysis of microbial DNA from milk samples was accomplished by sequencing the V4 regions of the 16S rRNA bacterial genes. Sequence data were quality filtered, taxonomically assigned, and organized by phylogeny. The aggregate microbiota obtained in these 17 samples, based on a total of 281,925 quality-filtered reads (about 17,000 reads per sample), demonstrated that Firmicutes (human 57%; monkey 56%) and Proteobacteria (human 37%; monkey 40%) are the predominant phyla in milk. Actinobacteria were of low abundance in both human and rhesus milk, with the genus Bifidobacterium detected at 0.1% in humans and 0.06% in rhesus (Fig. 2). Class analysis indicated that the bacterial class of greatest abundance in both rhesus and human milk was Lactobacillales followed by Sphingomonadales and Rhodocyclales (Supplementary Figure 1A and B, http://links.lww.com/MPG/A184). Interestingly, the only major difference observed between rhesus and human milk was a higher concentration of bacteria from the Clostridiales class in rhesus milk (P = 0.005). In support of this finding, rarefaction curves (data not shown) indicate that bacterial diversity in rhesus milk is higher than in human milk, reflecting the high proportion of Clostridiales in the rhesus milk as compared with human milk.
Significant progress in medicine and related fields can be attributed to rhesus monkeys (2). In addition, macaques have provided data for nutritional sciences (1), behavioral investigations (7), and various aspects of age-related research (28). The clear advantage of this model is that rhesus infants have similar nutritional requirements as well as neurological, metabolic, and behavioral development. Unlike rodents, human and rhesus mothers rear individual infants over an extended lactation, facilitating the complex socioemotional and psychobiological organization that is shaped by the quality of maternal care (29) and physiological investment, for example, milk (6). Moreover, humans and rhesus macaques are at a more mature stage of brain development when they begin nursing than rats and mice (7), and their gastrointestinal development is different because of the different genetic background. Accordingly, the comprehensive effects of different nutritional practices can be determined with rhesus monkeys (7,8) that provide information on similar processes in humans, unlike other animal models.
The greatest similarities were seen in the urinary metabolome in which an identical set of metabolites were found for rhesus and human infants, and several amino acids, ketones, and other metabolites related to energy metabolism, sugar metabolism, and gut microbiota showed similar concentrations. This finding is even more striking given the known metabolic diversity inherent in urinary profiles (10,30), and the fact that hydration may play a role in the concentration of certain metabolites. There are numerous genetic and metabolic explanations that could account for the observed differences. For example, urinary hydroxyproline is considered a marker of growth rate in infants (31); higher concentrations in rhesus urine could reflect the faster growth rate of rhesus infants. Other metabolites can be linked to obvious lifestyle differences. For example, increased concentrations of propylene glycol in human infant urine likely reflect routine hygiene practices because propylene glycol is one of the components of several brands of baby wipes. The fact that urine is the most closely matched biofluid is particularly relevant for infant nutrition research given the noninvasive nature of sample collection and its recognized value in metabolomics investigations of human diet (11,32) and health (10).
Serum profiles were also closely linked. Like urine, of the metabolites that were different between humans and rhesus, the majority were present at higher concentrations in rhesus. One noteworthy metabolite is glutamate, which was present at higher concentrations in human serum. It is the most abundant amino acid in human milk and present at significantly higher concentrations when compared with rhesus milk also. Another interesting observation relates to choline, which is an essential nutrient required for neurotransmitter synthesis, cell membrane signaling, lipid transport, and methyl-group metabolism (33). Its importance for human infants is supported by its presence in milk (34), and the fact that the capacity of the brain to extract choline from blood is greatest during the neonatal period (35). Although higher serum choline was found in human infants, higher concentrations of metabolites related to choline metabolism were observed in rhesus samples, namely betaine and dimethylglycine, which could reflect divergent developmental priorities in rhesus and human infant metabolism.
Comparison of human and rhesus milk metabolomes also showed some similarities; however, unlike other biofluids, there was no unifying trend when comparing metabolite concentrations between species. The most abundant amino acids in human and rhesus milk were glutamate, glutamine, and taurine, although the actual concentrations were higher in human milk. Indeed, free amino acid concentrations were generally higher in human milk as compared with rhesus milk, which supports previous findings (36). The abundance of glutamate in milk has led to speculation that it is essential in infant nutrition for the protection of intestinal growth and integrity, in addition to providing a supply of functional substrates to the nervous tissues (37). Metabolites with notably higher concentrations in rhesus milk samples included glycerophosphocholine, myo-inositol, and trimethylamine-N-oxide. Although each of these metabolites has several functions, they have 1 common role as osmolytes in renal medullary cells (38,39). Higher concentrations of osmolytes in animal kidneys compared with humans have been reported previously (40); HOWEVER, to our knowledge, these metabolite patterns have not been identified in milk samples until now.
The major difference between human and rhesus milk metabolomes was the detection of certain oligosaccharides in human milk that were either not present or at low concentration in rhesus milk, namely 2′-fucosyllactose, lacto-N-fucopentaose I, and lactodifucotetraose. These oligosaccharides are synthesized by the enzyme α(1,2) fucosyltransferase present in mammary tissue. The low concentrations found in rhesus milk could potentially reflect limited functionality of the α(1,2) fucosyltransferase enzyme in mammary tissue. We know that this condition exists for humans who are known as nonsecretors (41). Alternatively, it is possible that these particular oligosaccharides are more important for rhesus infants earlier in life and may be present at higher concentrations at an earlier stage of lactation. A detailed study of oligosaccharides at multiple time points in rhesus lactation will be necessary in future studies. Nevertheless, in support of previous findings, the present study revealed higher concentrations for most oligosaccharides in human milk compared with rhesus milk (42,43).
Given the similarity between the overall composition of human and rhesus milk, it is not surprising that the milk microbiome is also similar. The major difference between human and rhesus milk was a higher proportion of bacteria from the Clostridiales order. Interestingly, gut microbe–related metabolites were also present at different concentrations in rhesus and human infant urine samples, which supports findings reported previously describing a direct relation between mother's milk and the establishment and maintenance of the infant's microbiota (44). Although the symbiotic effect of oligosaccharides and bacteria present in mammalian milk has been described previously, the present study shows species-based differences in the composition of oligosaccharides and the microbial components of milk samples. Based on these findings, it could be speculated that declining levels of milk oligosaccharides may increase the bacterial diversity as is seen in rhesus milk samples.
In summary, the application of 1H NMR spectroscopy to compare rhesus macaques and humans has revealed substantial metabolic similarities. Metabolites that vary in concentration between species, and in particular in milk samples, could reflect the divergent developmental priorities of human and rhesus infants. Application of 16S rRNA gene sequencing reveals substantial microbial similarities in milk. These similarities provide unique biological information that highlights the importance of this model for infant nutrition and developmental research.
The authors are indebted to Dr Jennifer Smilowitz for providing human samples, and Dr Darya Mishchuk for isolating bacterial DNA from milk. We are particularly grateful to Dr David Mills for graciously providing us with DNA extraction and 16S rDNA amplification protocols. We also thank Dr Karen Kalanetra and Nicholas Bokulich for their skillful technical assistance, and the UC Davis CA&ES Computing Cluster Center (Farm) for providing access to computing resources. Macaque milk samples were made available through the ARMMS program (Archive of Rhesus Macaque Milk Samples).
1. Baker DH. Animal models in nutrition research. J Nutr 2008; 138:391–396.
2. Gardner MB, Luciw PA. Macaque models of human infectious disease. ILAR J 2008; 49:220–255.
3. Gibbs RA, Rogers J, et al. Evolutionary and biomedical insights from the rhesus macaque genome. Science 2007; 316:222–234.
4. Bodkin NL, Hannah JS, et al. Central obesity in rhesus monkeys: Association with hyperinsulinemia, insulin resistance and hypertriglyceridemia? Int J Obes Relat Metab Disord 1993; 17:53–61.
5. Adams Waldorf KM, Rubens CE, et al. Use of nonhuman primate models to investigate mechanisms of infection-associated preterm birth. BJOG: Int J Ob Gynaecol 2011; 118:136–144.
6. Hinde K, Capitanio JP. Lactational programming? Mother's milk energy predicts infant behavior and temperament in rhesus macaques (macaca mulatta). Am J Primatol 2010; 72:522–529.
7. Golub MS, Hogrefe CE, et al. Neurobehavioral evaluation of rhesus monkey infants fed cow's milk formula, soy formula, or soy formula with added manganese. Neurotoxicol Teratol 2005; 27:615–627.
8. Kelleher SL, Casas I, et al. Supplementation of infant formula with the probiotic lactobacillus reuteri and zinc: impact on enteric infection and nutrition in infant rhesus monkeys. J Pediatr Gastroenterol Nutr 2002; 35:162–168.
9. Lonnerdal B. Preclinical assessment of infant formula. Annals Nutr Metabol 2012; 60:196–199.
10. Slupsky CM, Steed H, et al. Urine metabolite analysis offers potential early diagnosis of ovarian and breast cancers. Clin Cancer Res 2010; 16:5835–5841.
11. O'Sullivan A, Gibney MJ, et al. Dietary intake patterns are reflected in metabolomic profiles: potential role in dietary assessment studies. Am J Clin Nutr 2011; 93:314–321.
12. O'Sullivan A, Gibney MJ, et al. Biochemical and metabolomic phenotyping in the identification of a vitamin D responsive metabotype for markers of the metabolic syndrome. Mol Nutr Food Res 2011; 55:679–690.
13. Patterson AD, Bonzo JA, et al. Metabolomics reveals attenuation of the slc6a20 kidney transporter in nonhuman primate and mouse models of type 2 diabetes mellitus. J Biol Chem 2011; 286:19511–19522.
14. Salek RM, Maguire ML, et al. A metabolomic comparison of urinary changes in type 2 diabetes in mouse, rat, and human. Physiol Genomics 2007; 29:99–108.
15. van Doorn M, Vogels J, et al. Evaluation of metabolite profiles as biomarkers for the pharmacological effects of thiazolidinediones in type 2 diabetes mellitus patients and healthy volunteers. BrJ Clin Pharmacol 2007; 63:562–574.
16. Hinde K, Power ML, et al. Rhesus macaque milk: magnitude, sources, and consequences of individual variation over lactation. Am J Phys Anthropol 2009; 138:148–157.
17. Slupsky CM, Rankin KN, et al. Investigations of the effects of gender, diurnal variation, and age in human urinary metabolomic profiles. Anal Chem 2007; 79:6995–7004.
18. Weljie AM, Newton J, et al. Targeted profiling: quantitative analysis of 1 h nmr metabolomics data. Anal Chem 2006; 78:4430–4442.
19. Bokulich NA, Joseph CML, et al. Next-generation sequencing reveals significant bacterial diversity of botrytized wine. PLoS ONE 2012; 7:e36357.
20. Werner JJ, Zhou D, et al. Comparison of illumina paired-end and single-direction sequencing for microbial 16 s rrna gene amplicon surveys. ISME J 2012; 6:1273–1276.
21. Caporaso JG, Kuczynski J, et al. Qiime allows analysis of high-throughput community sequencing data. Nature Meth 2010; 7:335–336.
22. Edgar RC. Search and clustering orders of magnitude faster than blast. Bioinformatics 2010; 26:2460–2461.
23. Yatsunenko T, Rey FE, et al. Human gut microbiome viewed across age and geography. Nature 2012; 486:222–227.
24. Caporaso JG, Bittinger K, et al. Pynast: a flexible tool for aligning sequences to a template alignment. Bioinformatics 2010; 26:266–267.
25. DeSantis TZ, Hugenholtz P, et al. Nast: a multiple sequence alignment server for comparative analysis of 16 s rrna genes. Nucleic Acids Res 2006; 34 (Web Server issue):W394–W399.
26. Price MN, Dehal PS, et al. Fasttree: computing large minimum evolution trees with profiles instead of a distance matrix. Mol Biol Evol 2009; 26:1641–1650.
27. Hinde R, Bowell T, et al. Behaviour of socially living rhesus monkeys in their first six months. Proc Zool Soc London 1964; 143:609–649.
28. Haberthur K, Engelman F, et al. Immune senescence in aged nonhuman primates. Exp Gerontol 2010; 45:655–661.
29. Nelson EE, Winslow JT. Non-human primates: model animals for developmental psychopathology. Neuropsychopharmacol 2009; 34:90–105.
30. Walsh MC, Nugent A, et al. Understanding the metabolome—challenges for metabolomics. Nutr Bull 2008; 33:316–323.
31. Younoszai MK, Andersen DW, et al. Urinary excretion of endogenous hydroxyproline by normal male infants. Pediatr Res 1967; 1:266–270.
32. Walsh MC, Brennan L, et al. Effect of acute dietary standardization on the urinary, plasma, and salivary metabolomic profiles of healthy humans. Am J Clin Nutr 2006; 84:531–539.
33. Zeisel SH, da Costa K-A. Choline: An essential nutrient for public health. Nutr Rev 2009; 67:615–623.
34. Ilcol YO, Ozbek R, et al. Choline status in newborns, infants, children, breast-feeding women, breast-fed infants and human breast milk. J Nutr Biochem 2005; 16:489–499.
35. Braun LD, Cornford EM, et al. Newborn rabbit blood-brain barrier is selectively permeable and differs substantially from the adult. J Neurochem 1980; 34:147–152.
36. Rassin DK, Sturman JA, et al. Taurine and other free amino acids in milk of man and other mammals. Early Hum Develop 1978; 2:1–13.
37. Agostoni C, Carratù B, et al. Free glutamine and glutamic acid increase in human milk through a three-month lactation period. J Pediatr Gastroenterol Nutr 2000; 31:508–512.
38. Balaban RS, Burg MB. Osmotically active organic solutes in the renal inner medulla. Kidney Int 1987; 31:562–564.
39. Beck FX, Burger-Kentischer A, et al. Cellular response to osmotic stress in the renal medulla. Eur J Physiol 1998; 436:814–827.
40. Sizeland P, Chambers S, et al. Organic osmolytes in human and other mammalian kidneys. Kidney Int 1993; 43:448–453.
41. Kunz C, Rudloff S. Potential anti-inflammatory and anti-infectious effects of human milk oligosaccharides. Adv Exp Med Biol 2008; 606:455–466.
42. Goto K, Fukuda K, et al. Chemical characterization of oligosaccharides in the milk of six species of new and old world monkeys. Glycoconjugate J 2010; 27:703–715.
43. Tao N, Wu S, et al. Evolutionary glycomics: characterization of milk oligosaccharides in primates. J Proteome Res 2011; 10:1548–1557.
44. Zivkovic AM, German JB, et al. Human milk glycobiome and its impact on the infant gastrointestinal microbiota. Proc Natl Acad Sci USA 2011; 108 (suppl 1):4653–4658.