The Role of Protein and Free Amino Acids on Intake, Metabolism, and Gut Microbiome: A Comparison Between Breast-Fed and Formula-Fed Rhesus Monkey Infants

Background: Compared to breast-fed (BF), formula-fed (FF) infants exhibit more rapid weight gain, a different fecal microbial profile, as well as elevated serum insulin, insulin growth factor 1 (IGF-1), and branched chain amino acids (BCAAs). Since infant formula contains more protein and lower free amino acids than breast milk, it is thought that protein and/or free amino acids may be key factors that explain phenotypic differences between BF and FF infants. Methods: Newborn rhesus monkeys (Macaca mulatta) were either exclusively BF or fed regular formula or reduced protein formula either supplemented or not with a mixture of amino acids. Longitudinal sampling and clinical evaluation were performed from birth to 16 weeks including anthropometric measurements, intake records, collection of blood for hematology, serum biochemistry, hormones, and metabolic profiling, collection of urine for metabolic profiling, and collection of feces for 16s rRNA fecal microbial community profiling. Results: Reducing protein in infant formula profoundly suppressed intake, lowered weight gain and improved the FF-specific metabolic phenotype in the first month of age. This time-dependent change paralleled an improvement in serum insulin. All lower protein FF groups showed reduced protein catabolism with lower levels of blood urea nitrogen (BUN), urea, ammonia, albumin, creatinine, as well as lower excretion of creatinine in urine compared to infants fed regular formula. Levels of fecal microbes (Bifidobacterium and Ruminococcus from the Ruminococcaceae family), that are known to have varying ability to utilize complex carbohydrates, also increased with protein reduction. Adding free amino acids to infant formula did not alter milk intake or fecal microbial composition, but did significantly increase urinary excretion of amino acids and nitrogen-containing metabolites. However, despite the lower protein intake, these infants still exhibited a distinct FF-specific metabolic phenotype characterized by accelerated weight gain, higher levels of insulin and C-peptide as well as elevated amino acids including BCAA, lysine, methionine, threonine and asparagine. Conclusions: Reducing protein and adding free amino acids to infant formula resulted in growth and metabolic performance of infants that were more similar to BF infants, but was insufficient to reverse the FF-specific accelerated growth and insulin-inducing high BCAA phenotype.

o SI Table 2. Differentiating OTUs between the breast-fed and the formula-fed rhesus infants.     (1) [b] Value is adapted from the true protein measurement of mature rhesus milk (2). True protein was estimated using (nitrogen in whole milknon-protein nitrogen) X 6.25. Nitrogen was estimated by micro-Kjeldahl analysis.
[c] Value is adapted from measurement of rhesus milk collected after 36 days of lactation (3). Total lipid was measured by a colorimetric method through sulfo-phospho-vanillin reaction.
[d] Value is adapted from measurement of milk collected at 1 month of lactation (4). Total lipid was measured by a micromodification of Rose-Gottleib procedure.
[e] Value is adapted from measurement of milk collected at 1 month of lactation (4). Total carbohydrate was determined by phenol-sulfuric acid method.
[f] Value is adapted from measurement of milk collected after 36 days of lactation (3). Lactose was measured using an enzymatic approach involveing lactase and glucose oxidase.
[g] The amino acid composition was converted using measurements adapted from (5) and corrected using true protein measurement adapted from (2). Amino acid concentration was measured spectrophotometrically after HCl digestion. However, tryptophan was destroyed by acid hydrolysis therefore not included in the total amino acid concentration. Due to this limitation of excluding tryptophan as a part of total amino acid, the concentration for each amino acid present in the table is over-estimated.
Data on rhesus milk are presented from literature as Mean ± SD. For unit conversion, rhesus milk density is conventionally assumed as 1.03 g/ mL. ND: not determined. Table 2. Differentiating OTUs between the breast-fed and the formula-fed rhesus infants. • High in breast-fed, ○ High in the formula-fed. Significance is evaluated using Analysis of Composition of Microbiomes (ANCOM) with p < 0.05 after FDR correction.  Figure 1. Comparison of serum urea, blood urea nitrogen (BUN) and ammonia between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Serum urea, BUN and ammonia were quantified using NMR, biochemical assay and AAA, respectively. Data are presented as mean ± SEM. Figure 2. Comparison of serum creatinine between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data were quantified using NMR (left) and biochemical assay (right). Data are presented as mean ± SEM.

SI Figure 3. Comparison of urinary creatinine level between breast-fed (BF) and formula-fed (FF) rhesus infants from the current study and from the previous work ((1), BF ref, FF ref). Data are presented as mean ± SEM.
SI Figure 4. Comparison of serum pancreatic polypeptide between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data are presented as mean ± SEM. SI Figure 5. Comparison of serum total carbon dioxide (TCO2), anion gap, albumin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALK PHOS) and chloride between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data are presented as mean ± SEM. SI Figure 6. Comparison of serum glucose, galactose and myo-inositol between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Measurement of glucose was conducted using NMR (top left) and biochemical assay (top right). Data are presented as mean ± SEM. Figure 7. Comparison of serum hemoglobin, hematocrit and mean corpuscular hemoglobin concentration (MCHC) between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data are presented as mean ± SEM. Figure 8. Comparison of serum branched chain amino acids (isoleucine, leucine, valine) between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data were quantified using NMR (left) and AAA (right). Data are presented as mean ± SEM.

SI
SI Figure 9. Comparison of serum lysine, methionine and threonine between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data were quantified using NMR (left) and AAA (right). Data are presented as mean ± SEM.
SI Figure 10. Comparison of serum alanine, asparagine and hydroxyproline between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data were quantified using NMR (left) and AAA (right). Data are presented as mean ± SEM.
SI Figure 11. Comparison of serum fumarate, lactate, malate, pyruvate and succinate between breastfed (black) and formula-fed rhesus infants (red, orange, green, blue). Data are presented as mean ± SEM.
SI Figure 12. Comparison of serum ketone bodies (acetoacetate and 3-hydroxybutyrate) between breast-fed (black) and formula-fed rhesus infants (red, orange, green, blue). Data are presented as mean ± SEM. Figure 13. Serum Phenylalanine and serine concentration that were not statistically significant between the formula-fed (FF) and breast-fed (BF) rhesus infants in the present study but significantly different in our previous work ((1), BF ref, FF ref). Statistical difference was evaluated using repeated measures ANCOVA. Data are presented as mean ± SEM.

SI
SI Figure 14. The fold change (%) of average daily protein intake from consuming the reduced protein formula. The smooth curve was fitted using loess regression and the 95% confidence interval was constructed using a t-based approximation.
SI Figure 15. Serum metabolites at 4 weeks of age that are higher in the formula-fed rhesus infants who consumed reduced protein formulas in comparison to those who consumed the regualr formulas. (A). 4hydroxyproline, (B). hydroxylysine, (C). glycine, (D). serine, (E). homocysteine, and (F). ethanolamine were significant (p<0.05) after adjustment of multiple comparison using FDR.(G). aspartate shows an increasing trend in those who consumed the reduced protein formulas (p<0.05 before FDR correction). Statistical significance was evaluated using multiple 2-way ANOVAs.
SI Figure 17. The influence of adding free amino acids to infant formula on metabolism. In urine, amino acids (aspartate, proline, threonine), amino acid derivatives (hydroxyproline, dimethylglycine, cadaverine) were significantly higher in the urine of infants fed formulas with the addition of free amino acids (p<0.05 after FDR correction, repeated measures ANOVA). Urinary asparagine, betaine, glycine isoleucine and serine showed a tendency to be higher in the urine those infants receiving formula with the addition of free amino acids (p<0.05 before FDR correction, repeated measures ANCOVA). Serum betaine showed a trend toward lower levels in infants fed formula with the addition of free amino acids (p<0.05 before FDR correction, repeated measures ANOVA). Data are presented as mean ± SEM. Figure 18. Relative abundance of bacterial genera in feces of rhesus infants consuming rhesus milk, regular formula, regular formula plus free amino acids, reduced protein formula, or reduced protein formula plus free amino acids from 2 to 16 weeks of age.

SI
SI Figure 19. Principal coordinate analysis of weighted and log transformed unifrac distances (A) over time. (B) Shifting of centroids over time. The centroids are calculated using the average of PC1 and PC2 within each cluster.