Differences in human milk peptide release along the gastrointestinal tract between preterm and term infants

Abstract

Background and aims:

Preterm infants are born with a gastrointestinal tract insufficiently developed to digesting large quantities of human milk proteins. Peptides released from the digestion of human milk proteins have been identified with bioactivities that may be beneficial to the developing infant. However, it is unknown how prematurity affects total and bioactive peptide release along the gastrointestinal tract. The aim of this study was to compare milk peptide release from milk to stomach to stool between preterm and term infants.

Methods:

Milk, gastric, and stool samples were collected from preterm infants as early collection (days 8 and 9 of life) and late collection (days 21 and 22 of life), and from term infants as early collection. Milk peptides were extracted from the samples and identified using Orbitrap mass spectrometry. Peptide abundance and count were compared across digestion and between the three infant groups at each stage of digestion.

Results:

Total milk peptide count and abundance increased from milk to stomach then decreased in stool. Total peptide release was similar among the three infant groups for milk and stool samples. In the stomach, preterm early collection had significantly higher peptide abundance and count than the other two groups. Patterns for peptide release from individual milk proteins were distinct from total peptide release both across digestion and among the infant groups. When analyzing single peptides, term early collection gastric samples had significantly higher peptide abundance than preterm early collection for a known antimicrobial peptide, QELLLNPTHQIYPVTQPLAPVHNPISV.

Conclusions:

Preterm and term infants digest milk proteins differently along their gastrointestinal tracts. While preterm infants released more total peptides in the stomach, term infants released specific bioactive peptides at higher abundance. We identified a region at the C-terminus of β-casein that is conserved from milk through stool and from which are released known and potential antimicrobial peptides.

Graphical Abstract

1. Introduction

Over 10% of all infants born in the US each year are premature, or born <37 weeks gestational age (1). Premature infants face a variety of nutritional challenges due to their restricted development time. In order to match intrauterine rates of growth, the enteral energy and protein needs of preterm infants are far greater than those of term infants, at up to 135 kcal/kg/day and 4.5 g protein/kg/day (2, 3). Furthermore, the preterm gastrointestinal system is underdeveloped at birth and is not optimized for the complete digestion of large quantities of macronutrients. In the stomach, preterm infants produce less gastric acid, have lower pepsin activity (4, 5), and have increased gastric emptying time (6, 7), all of which can impact protein digestion.

Human milk with fortification is widely accepted as the best source of nutrition for the preterm infant. Human milk proteins contain the optimal balance of amino acids to meet infant growth requirements (8). Many of the proteins, such as lactoferrin, immunoglobulins, and growth factors, can remain undigested to the benefit of the infant’s health and development (9). Upon being fed to the infant, most milk proteins are cleaved into smaller peptides by a variety of proteases either present in the milk or secreted by the infant’s gastrointestinal tract. These peptides may also benefit the infant, as hundreds of milk peptides have been identified with antimicrobial, antihypertensive, antioxidant, immunomodulatory, cell proliferative, and nutrient-delivery activity (10, 11). Peptidomics, an offshoot of proteomics, has evolved as a method to identify and quantify these smaller peptides as they are released during digestion and assess how the proteins are digested at various stages (12, 13). Peptidomics research on milk and milk products have found that bioactive peptides are released from milk proteins during in vitro digestions (14, 15), digestion in animal models (16), and in vivo gastric digestion in preterm infants (17-19). However, only a few comparisons between preterm and term infants using peptidomic methods have been performed, and only on undigested milk samples (20-22). No research has yet investigated the release of human milk peptides in the intestinal tract.

The full impact of prematurity on protein digestion in the stomach and along the gastrointestinal tract remains unknown. Impaired protein digestion may not only affect the quantity of amino acids available to the infant for protein synthesis but also the release of bioactive peptides from the milk proteins. If bioactive peptides are to have an effect on infant health, they must be released in significant quantities and survive to their site of activity. For most activities, the peptides must reach the intestinal tract and either cross into the bloodstream to have systemic effects (23) or act locally on gut bacteria and intestinal tissues or immune cells (24). Infants who are unable to digest milk proteins in a manner that releases and preserves bioactive peptides are thus unable to take advantage of the full benefits of human milk.

The aim of this research was to test the hypothesis that there are differences in human milk peptide release between preterm and term infants across digestion. Peptidomics analysis was used to identify peptides from the milk, gastric fluid, and stool samples of preterm and term infants. Total and individual peptides were compared between the infants at each phase of digestion. Milk protein cleavage and peptide release were mapped across digestion to gain a deeper understanding of where bioactive peptides first appear in the infant.

2. Methods

2.1. Materials

Ammonium bicarbonate and HPLC-grade acetonitrile were obtained from Thermo Fisher Scientific (Waltham, MA), trifluoroacetic acid and HPLC-grade formic acid were obtained from EMD Millipore (Billerica, MA), HPLC-grade ethanol, iodoacetamide, and trichloroacetic acid were obtained from Sigma-Aldrich (St. Louis, MO), and dithiothreitol was obtained from Promega (Madison, WI).

2.2. Participants and enrollment

Ethical approval for this study was granted by the Institutional Review Boards at Legacy Health Systems and Oregon State University. Milk, gastric, and stool samples were collected from term and preterm infant subjects enrolled at Randall Children’s Hospital. Informed consent was obtained for all subjects. Stool samples from this infant population were previously analyzed for peptidomics in our previous study (19).

Infant enrollment criteria and sample collection procedures were identical to those described in our previous publications (19, 25). Briefly, collections of milk and gastric samples from preterm and term infants were attempted at 8, 9, 21, and 22 days of life (DOL). However, we were unable to complete the full four days of sample collection for many of the infants due to insufficient gastric volume or removal of infants from the NICU. Mother’s or donor milk was prepared by the nurses at Randall Children’s Hospital using aseptic techniques. Milk was prepared without human milk fortifier only for the sample feeding. A sample of the milk was collected upon thawing and prior to feeding and immediately frozen at −80°C. Gastric residual was removed and the volume recorded, and the milk was administered to the infants through naso/orogastric feeding tubes over 30–60 min. Thirty minutes after completion of the feed, a gastric sample was collected from the infant and immediately frozen at −80°C. The volume of gastric fluid collected was recorded and replaced with fresh feed. Stool samples were collected from each infant over each two-day period (8/9 and 21/22 DOL). Nurses attempted to collect all stool produced by the infant after the sample milk feeding. Each stool sample was immediately frozen at −80°C. All samples were transported from Randall Children’s Hospital to Oregon State University on dry ice for sample analysis. The number of infants and samples collected at each DOL are listed on Supplementary Table 1.

2.3. Sample preparation

Milk and gastric samples were prepared as previously described (26) with the following modifications. A 20-μL aliquot was taken from each sample and mixed with 80 μL of nanopure water. Lipids were skimmed by centrifugation at 4,000 × g for 10 min at 4°C after mixing. To reduce disulfide bonds, 100 μL of 200 mM ammonium bicarbonate was mixed with each sample, and dithiothreitol was added to a final concentration of 40 mM. The samples were then incubated at 56°C for 45 min. Iodoacetamide was added to a final concentration of 100 mM, and the samples were incubated at room temperature in the dark for 1 h. Protein precipitation with 200 μL of 24% TCA and peptide enrichment via C18 reverse-phase extraction were performed as previously described (27). The data from the stool samples were taken from a previous study (19).

2.4. Liquid chromatography mass spectrometry

Peptides were analyzed with mass spectrometry as previously described (19). The order of sample loading on the mass spectrometer was randomized to minimize bias caused by instrumental drift over time. UPLC and mass spectrometry settings were as previously described (19).

Peptides were identified from the raw files with Thermo Proteome Discoverer 2.2.0.388 using a Sequest HT search on an in-house human milk protein database. Samples were grouped for analysis based on their sample type (milk, gastric fluid, or stool), DOL (8/9 or 21/22), and infant maturity (preterm or term). Allowed peptide modifications were serine and threonine phosphorylation, methionine oxidation, and cysteine carbamidomethylation. Peptide identification was validated using a decoy database search strategy with a false-discovery rate of P < 0.01. Only peptides with high confidence were included in the final data analyses.

2.5. Data analysis

Milk, gastric, and stool samples were analyzed for peptide abundance (the summed area under the ion intensity curve from the mass spectra) and peptide count (the number of unique peptide sequences). All statistical comparisons were carried out in GraphPad Prism 8.4.1. Peptide abundances and counts were averaged for the milk and gastric samples for each infant across days 8/9 (Early Collection, EC) and days 21/22 (Late Collection, LC) of life. Stool samples were only collected once over each two-day period, so no averaging was necessary. Significant differences in peptide abundance and count between sample types and infant age and maturity status were determined by two-way ANOVA and post-hoc multiple comparisons corrected with the Tukey-Kramer method. Significance was determined by a P-value of <0.05.

R version 3.6.1 was used for peptide density and hierarchical clustering analyses. Peptide isoelectric point was determined using the calculator at http://isoelectric.org/ (28). Peptide grand average of hydropathy (GRAVY) score (29) was determined using the calculator at http://www.gravy-calculator.de/. Hierarchical clustering analysis was performed with the pheatmap package. Perseus version 1.6.14.0 was used to calculate permutation-based false discovery rate q-values for comparisons between infant groups of the individual peptides.

Milk peptides were analyzed for sequence homology with bioactive peptides in the Milk Bioactive Peptide Database (MBPDB) (11). The search type was “Sequence” with a similarity threshold of 80%. PepEx (http://mbpdb.nws.oregonstate.edu/pepex/) was used to map the identified milk peptides to their location in the parent protein sequence (30).

3. Results

3.1. Human milk peptide profiles across digestion

Across all milk, gastric, and stool samples analyzed, 19,612 peptides with unique sequences were identified. The number of peptides unique to the milk, gastric, or stool samples and the number of peptides shared among the samples are shown in Fig. 1A. Although all sample types were unique in terms of the milk peptides present in each, there was more crossover for peptide sequences from milk to gastric fluid (4,928 peptides in common) than from milk to stool (935 in common) or gastric fluid to stool (1,900 in common). There were 849 peptides that were present in all three samples; however, the present study was not designed to determine whether these peptides were first released in milk and survived through the entirety of digestion or were released multiple times by distinct proteolytic events in the milk, stomach, and colon.

Fig. 1.

Differences in human milk peptides from infant milk, gastric, and stool samples. (A) Venn diagram of the number of peptides in each sample. (B) Density plots of peptides in each sample for the physicochemical characteristics: molecular weight, (C) length, (D) peptide isoelectric point, (E) net charge at pH 7, and (F) GRAVY score. (G) Average percentage composition of each amino acid in the total milk peptidome of the milk, gastric, and stool samples. Amino acids are represented by their one-letter code.

Peptides from milk and gastric samples were more similar physicochemically than either was to stool. Milk and gastric peptides were distributed across higher molecular weights than were those in stool (Fig. 1B). Although the differences in molecular weight among milk, gastric, and stool peptides may be in part explained by differences in the amino acid makeup of the peptides, the larger contributing factor was the shorter peptide lengths in the stool samples (Fig. 1C). Milk and gastric peptides were also distributed at higher isoelectric points and thus were more basic, whereas stool peptides had more peptides distributed at lower isoelectric points and thus were more acidic (Fig. 1D). The higher acidity of stool peptides is further supported by the shift to lower net charges at neutral pH compared with those of the milk and gastric peptides (Fig. 1E). The GRAVY score predicts the hydropathy of a peptide, with a negative value indicating hydrophilicity and a positive value indicating hydrophobicity. There was a large amount of overlap between peptide GRAVY scores, but stool peptides were slightly more hydrophobic than milk and gastric peptides, which had longer tails in the negative direction (Fig. 1D).

The differences in isoelectric point, net charge, and GRAVY score of the peptides can be explained by differences in the primary structure of the peptides. Histidine, lysine, and arginine (positively-charged amino acids) all comprised a larger percentage of milk and gastric peptides on average, whereas stool peptides had a larger percentage of aspartate (negatively-charged amino acid) (Fig. 1G). Stool peptides also had more cysteine and glycine and slightly more phenylalanine and tryptophan, which contributed to to their higher GRAVY scores.

3.2. Differences between preterm and term infants at 8/9 and 21/22 DOL

Total milk peptide count and abundance of each sample type were compared among preterm infants at 8/9 DOL (PEC), preterm infants at 21/22 DOL (PLC), and term infants at 8/9 DOL (TEC). No term infants remained enrolled in the study long enough to collect samples at 21/22 DOL. For milk and stool samples, there were no significant differences among the infant groups (Fig. 2A-B). In the gastric samples, PEC had the highest peptide abundance and count, whereas PLC and TEC were not significantly different. Peptide abundance and count were also compared across the three samples for each infant group. For all infants, both average peptide abundance and count significantly differed within the milk, gastric, and stool samples. Average peptide abundance increased from milk to gastric samples, then decreased back to levels equivalent to milk in the stool. Average peptide count similarly increased from milk to gastric samples, then decreased in stool to a level higher than in milk.

Fig. 2.

Comparison of milk peptide abundance and count across digestion and between PEC, PLC, and TEC infants. (A) Total peptide abundance and (B) total peptide counts were compared. Individual peptide abundances and counts were summed for each sample, and sample types of different infant maturity and age were compared using two-way ANVOA with post-hoc Tukey-Kramer corrected multiple comparisons. Data with different overhead letters represent significant differences between the groups. (C) Peptide abundance and (D) peptide count from individual milk proteins were similarly derived from the summed individual peptide abundances and counts for each protein and compared using two-way ANOVA and multiple comparisons. *Indicates significant differences between infant groups for the sample type and protein. PEC, preterm early collection; PLC, preterm late collection; TEC, term early collection. For milk, PEC n=16, PLC n=16, TEC n=11; for gastric, PEC n=16, PLC n=12, TEC n=11; and for stool, PEC n=12, PLC n=11, TEC n=10.

As differences in the total milk peptide abundance and count among PEC, PLC, and TEC infants were only found in the gastric samples, we next determined whether these results remained constant for peptides from individual milk proteins (Fig. 2C-D). All proteins had similar peptide abundance and count among PEC, PLC, and TEC infants in the milk samples. Due to overlap in the sequences of peptides, Ig heavy constant α (IgHA) 1 and 2 were combined for the analysis. Polymeric Ig receptor (PIgR), α-lactalbumin, serum albumin, IgHA1/2, and lysozyme had significantly different gastric abundance between PEC, PLC, and TEC, whereas β-casein and α-lactalbumin had significantly different stool abundance. Serum albumin, α-lactalbumin, and lysozyme had significantly different gastric count, and lactoferrin and IgHA1/2 had significantly different stool count. As with the total peptides, each individual protein was significantly different from milk to the infant stomach to the stool for both peptide abundance and count.

Finally, to identify differences among the PEC, PLC, and TEC infants at an individual peptide level, we performed hierarchical clustering on the peptide abundances scaled across all the samples and generated a heatmap of the peptides (Supplementary Fig. 1). As with the total abundance, there were no clear distinctions among the peptide profiles of the infant groups, even for the gastric samples. Rather than clustering under one or two branches, the PEC, PLC, and TEC samples were scattered over the entire dendrogram. However, despite heterogeneity within each of the milk, gastric, or stool samples, the peptide profiles of each phase of digestion were distinct from each other, and the samples clustered mostly within their own branches.

3.3. Bioactive milk peptides

Peptides were compared with the Milk Bioactive Peptide Database to identify bioactivities or potential bioactivities based on sequence homology. There were 271 homologous milk peptides that had ≥80% similarity of sequence with a known bioactive peptide. Sixteen of these were identical to a known bioactive milk peptide and are listed in Table 1. Of the 271 peptides, 139 peptides were homologous to peptides with antimicrobial activity, 83 with antihypertensive activity, 42 with cell-proliferative activity, 7 with antioxidant activity, 6 with DPP-IV inhibitory activity, and one peptide each with opioid, anticancer, and antithrombin activity.

Table 1.

Known bioactive peptides in the milk, gastric, or stool samples.

Peptide	Protein	Activity	Milk	Gastric	Stool
TVYTKGRVMP	β-casein (107–116)	ACE-inhibitory	X	X
LTDLENLHLP	β-casein (133–142)	ACE-inhibitory		X
LENLHLPLP	β-casein (136–144)	ACE-inhibitory	X	X
ENLHLPLP	β-casein (137–144)	ACE-inhibitory		X
ENLHLPLPLL	β-casein (137–146)	ACE-inhibitory		X
NLHLPLP	β-casein (138–144)	ACE-inhibitory		X
NLHLPLPLL	β-casein (138–146)	ACE-inhibitory	X	X
QVPQPIP	β-casein (152–158)	ACE-inhibitory		X
WSVPQPK	β-casein (169–175)	ACE-inhibitory	X	X
		Antioxidant
WLAHKAL	α-lactalbumin	ACE-inhibitory		X
	(123–129)	DPP-IV inhibitory
YANPAVVRP	κ-casein (81–89)	ACE-inhibitory		X
LLNQELLLNPTHQIYPV	β-casein (197–213)	Antimicrobial	X	X
QELLLNPTHQIYPVTQPLAPVHNPISV	β-casein (200–226)	Antimicrobial	X	X
YPVTQPLAPVHNPIS	β-casein (211–225)	Antimicrobial	X	X
RETIESLSSSEESITEYK	β-casein (16–33)	Cell-proliferation	X	X	X
SPTIPFFDPQIPK	β-casein (120–132)	Cell-proliferation	X	X

Homologous peptides were disproportionately abundant in the milk and gastric samples compared with non-homologous peptides. Averaged across the milk samples, homologous peptides comprised 31.9% of the total peptide abundance despite comprising only 5.1% of identified peptides. In the gastric samples, homologous peptides comprised 28.9% of peptide abundance and 3.3% of peptide count. Only in the stool samples was the abundance and count of homologous peptides proportional, comprising 0.15% and 0.19%, respectively.

There were no significant differences in homologous peptide abundance between PEC, PLC, and TEC infants for any sample, and only between PLC and TEC for milk and gastric samples for homologous peptide count (Supplementary Fig. 2). Though the combined abundance of homologous peptides was similar among the infant groups, we wanted to compare individual peptides to see whether any homologous peptides were more abundant in the TEC or PLC infants than the PEC infants (Fig. 3). In the milk and stool samples, no homologous peptides were significantly different, although TEC infants had two non-homologous peptides significantly higher than PEC in both samples. In the gastric samples, 121 peptides were significantly higher in abundance in PEC infants than TEC, and only 24 were higher in TEC. Both TEC and PEC had four significantly higher homologous peptides, but only TEC had a known antimicrobial peptide. There were no differences in any of the samples between the PEC and PLC groups.

Fig. 3.

Volcano plots for comparisons of individual peptide abundances within milk, gastric, and stool samples. Peptides were compared between (A) PEC and TEC milk, (B) gastric, and (C) stool samples; and between (D) PEC and PLC milk, (E) gastric and (F) stool samples. Only peptides present in ≥50% of the samples for each comparison were included. Diamonds represent peptides that were both significantly different between groups (q<0.05) and with >2-fold change in abundance. Orange data represent peptides homologous with a known bioactive peptide. PEC, preterm early collection; PLC, preterm late collection; TEC, term early collection.

3.4. Changes in peptide release across the protein sequences

Peptides were differentially released from distinct regions of the milk proteins as they progressed from milk to gastric fluid to stool. The mean abundance of peptides that contained the amino acids at each position across the full sequences of nine of the most abundant milk proteins is shown in Fig. 4. Supplementary Figs. 3-5, respectively, show the milk, gastric, and stool peptides broken down by PEC, PLC, and TEC infants. Fig. 4A-D are proteins exclusively found in human milk: β-casein, α-lactalbumin, κ-casein, and α_s1-casein, respectively. From the milk to the infant stomach, peptides were released from almost identical regions of the proteins, with the only difference being a higher peptide abundance in some of the proteins in the gastric samples. In the stool, the pattern of release was distinct from milk or gastric fluid. The regions of peptide abundance in the stool overlapped regions in milk and gastric fluid for β-casein, α-lactalbumin, and α_s1-casein, indicating that some peptides released early on could have resisted complete degradation. However, the sole peptide in κ-casein was released from a region that was not digested in the milk or gastric fluid, so this peptide could only have come from additional proteolysis in the intestine.

Fig. 4.

Peptide release across the sequence of individual milk proteins during different phases of digestion. For all figures, the y-axis is the mean peptide abundance and the x-axis is the linear amino acid sequences for the proteins (A) β-casein, (B) α-lactalbumin, (C) κ-casein, (D) α_s1-casein, (E) lactoferrin, (F) PIgR, (G) serum albumin, (H) lysozyme, and (I) osteopontin. Grey shaded boxes represent regions of homology to bioactive peptides. PIgR, polymeric immunoglobulin receptor.

Shown in Fig. 4E-I are proteins that are found in human milk but can also be secreted endogenously by the infant: lactoferrin, PIgR, serum albumin, lysozyme, and osteopontin, respectively. Like the exclusively milk proteins, there was conservation in the regions of peptide abundance from milk to gastric samples, although with a greater magnitude difference. Peptides from lactoferrin, serum albumin, and PIgR were highly abundant in the stool and dispersed across the entire protein sequences, potentially due to contributions from endogenous secretions. Osteopontin had no identifiable peptides surviving in the stool, thus it was likely completely digested or absorbed within the intestinal tract.

3.5. Peptides conserved across digestion

For a milk peptide to be bioactive in the infant, it must first be released from its parent protein and resist digestion long enough to reach its site of activity. Thus, we next searched for peptides that were identified in the milk, gastric, and stool samples from the same infant on the same DOL, i.e., milk at 8 DOL, gastric fluid at 8 DOL, and stool at 8/9 DOL. From 60 complete sample sets, there were 591 total peptides that fit these criteria, with a mean of 114 ± 29 (mean ± S.D.) peptides (Fig. 5A). When considering only gastric fluid to stool conservation, which represented peptides that appeared in the stomach and survived or were re-released in the intestine, there were 1,803 total peptides with an average of 486 ± 117 peptides. For conservation from only milk to gastric fluid, which represented peptides in milk that were able to survive gastric but not intestinal digestion, there were 70 sample sets that had both samples from the same infant at the same DOL. These sample sets had 4,415 total peptides present in both milk and gastric fluid with an average of 1,525 ± 272. Finally, for conservation only in the milk and stool, which represented peptides that did not survive gastric digestion but were released by another proteolytic event at a later point, there were 734 peptides with an average of 167 ± 40.

Fig. 5.

Milk peptides conserved across infant digestion. (A) Count of peptides conserved across two or more samples in one infant at one time point. Data are shown as mean ± S.D. Density plots of peptides conserved between samples and peptides that were not conserved for (B) molecular weight, (C) isoelectric point, and (D) GRAVY score. (E) Average percentage composition of each amino acid in the conserved and unconserved milk peptides. M, milk; G, gastric; S, stool.

We next compared the chemical and structural characteristics of the peptides that were conserved across two or more samples with those that were not to determine whether there were any unique properties to the conserved peptides that could explain their apparent increased resistance to digestion. The isoelectric point, hydropathy, and molecular weight distributions of the two groups of peptides are highly similar (Fig. 5B-D). The peptide groups were also similar in terms of average amino acid composition in the primary sequence with three differences: peptides that were conserved across digestion had a higher percentage of proline and glutamine residues and a lower percentage of glycine residues (Fig. 5E).

168 peptides conserved across digestion were homologous with bioactive peptides, including 9 peptides with 100% sequence similarity. Nine of the homologous peptides were identified in the milk, gastric, and stool samples: four with antimicrobial activity, three with cell-proliferative activity, and one each with antioxidant and antihypertensive activity. Of the remaining conserved homologous peptides, 156 were conserved from milk to gastric fluid, and 3 were conserved from gastric fluid to stool. No homologous peptide was identified in the milk and stool only.

4. Discussion

To our knowledge, this is the first time peptidomics has been applied to compare in preterm and term infants the digestion of milk proteins as they pass through the gastrointestinal tract. Peptidomics has previously been used to profile human milk (27, 31-33), infant gastric and stool samples (17-19, 26), and in vitro digestions of human milk (14, 34). Compared with previous investigations, the present study identified greater numbers of milk peptides in both the milk and gastric samples. The count of peptides in undigested milk has consistently been reported to be within a range of several hundred per sample. In gastric samples, the earliest study reported a range of several hundred, but later studies reported counts in the thousands. The average peptide count of the undigested milk samples in the present study are one order of magnitude higher than previously reported. This higher count more closely aligns with the results of Dingess et al. (33), where the investigators used a combination of TCA precipitation and HCD fragmentation to identify thousands of unique milk peptides.

Importantly, the observed peptide profiles of the milk, gastric, and stool samples in the present study may differ from the true peptide profiles. Modifications to the peptide extraction procedure, liquid chromatography and mass spectrometry settings, and peptide identification software could be applied to target peptides of different physicochemical characteristics. Comparisons of peptide abundance between studies are currently impossible as label-free quantification of the spectra is relative to each experiment based on instrumentation, methods, and software (35). It is clear that the gradual refinement of the extraction protocols and mass spectrometry instrumentation over time have increased the number of peptides identified in milk and other samples, but additional methodologies are needed to unearth the “hidden” peptidome and to compare results between different studies.

Although peptidomic surveys of various human milk samples and digesta have often been performed, differences in the peptide profiles of preterm and term infants have only been analyzed a few times for human colostrum (20) and milk samples (21, 22). The three studies were aligned in identifying differences in the levels of individual milk peptides between the term and preterm infants, but conflicted in the cumulative peptide difference. Wan et al. (20) and Dingess et al. (22) found no difference in the total number of peptides, whereas Dallas et al. (21) found preterm infant milk to have higher peptide count from <14–41 DOL and higher abundance from 29–41 DOL. The results of the present study once more confirmed differences in the individual peptide levels of preterm and term milks at 8/9 DOL, but not intotal peptide count or abundance . Although preterm milk has a higher concentration of protein than term milk in the first weeks of life (36), this does not appear to correspond with a greater release of milk peptides.

In the gastric samples, preterm infants at 8/9 DOL had significantly higher total, protein-specific, and individual peptide abundance than term infants. These results seem contrary to expected results if preterm infants had reduced capacity to digest proteins than term infants, which the current literature supports. Gastric acid secretion is lower in preterm than term infants (4, 37, 38). Preterm infants have lower pepsin activity than term infants (4, 5), and lower total proteolysis and protease activity in the stomach (5). Less pepsin activity and less proteolysis would lead to less peptide content from cleaved proteins, indicating that there must be other factors to consider.

The stomach secretes water and mucus that can dilute the feed proteins and peptides, and term infant secretions are greater in volume than preterm infants in the unstimulated stomach (39). In the present study, any residual gastric fluid was removed prior to feeding, but differences in secretion volume during digestion could explain the lower peptide abundance in term infants. This study did not control for feed dilution, but future studies will address this matter to make a more accurate determination of digestive differences. An additional factor is that the settings used for mass spectra collection and peptide identification in this study were optimized for a range of 375–1,500 m/z and 6–50 amino acids. Due to more active proteases in the term infant stomach, smaller peptides may have made up a larger fraction of the term gastric samples but were unidentified. Other potential factors that could affect gastric peptide abundance include the timing of sample collection after initiation of the feed. Our previous study showed that protein digestion and peptide release in the stomach increases up to three hours after feeding in preterm infants (26). The present study only measured gastric peptides at one hour after feeding. As gastric digestion and gastric emptying is a dynamic process, there could be changes in the rate of peptide release between preterm and term infants the present study was unable to distinguish.

Though the PEC infants had more peptide abundance in the gastric samples than TEC, there were several individual peptides that were more abundant in the term stomach. Only the TEC infants had a known bioactive peptide, QELLLNPTHQIYPVTQPLAPVHNPISV, in significantly higher abundance. This peptide was first identified with antimicrobial activity against a range of pathogenic bacteria(40). It derives from the C-terminus of β-casein, which encodes several peptides with antimicrobial activity (41-43) and weak antihypertensive and antioxidant activity (44-46). The present study showed that not only were more peptides released from the C-terminus of β-casein than from any region of any other protein, but that peptides from this region were consistently identified in the milk, stomach, and stool. Our previous peptidomic investigations discovered a similar phenomenon (18, 26). It is not clear what the importance of this region of β-casein is to the infant, but if it is able to survive through the intestinal tract in significant amounts, it could inhibit the colonization of pathogenic bacteria or influence the development of the early microbiome.

In the present study, milk proteins were differentially digested along the infant gastrointestinal tract. Digestion of the caseins and α-lactalbumin from milk to stomach increased at similar regions along the sequence, but between the stomach and stool, their peptide abundances were almost entirely depleted as the peptides were likely fully broken down into amino acids or di- and tripeptides and absorbed. The fate of the remaining whey protein peptides is more difficult to decipher, as the infant can also produce those proteins endogenously. The major whey proteins of human milk are lactoferrin, α-lactalbumin, secretory IgA, serum albumin, and lysozyme (47). Lactoferrin and lysozyme are produced by gastrointestinal tissues (48, 49), whereas IgA and serum albumin are major components of plasma and can be excreted in stool (50). This study also identified a large number of peptides from osteopontin and PIgR, both of which can be secreted into the gastrointestinal tract (51, 52). Distinguishing peptides from these proteins as being milk-derived versus endogenously secreted would require stable isotope labeling. While that is outside the scope of the present study design, several peptides from these proteins were found in the milk, gastric, and stool samples from the same infant on the same DOL, suggesting some of the peptides in the stool at the end of digestion could be milk-derived.

By identifying peptides in the milk, gastric fluid, and stool from the same infant at the same DOL, we were able to compare peptides present at the beginning, middle, and end of digestion . There were only a few differences in the physical properties of peptides present at more than one site versus peptides that were not. The most interesting difference was the higher percentage of proline residues in the peptides that were conserved. It has previously been noted that peptides with leucine and proline residues, particularly at the C-terminus, are more resistant to intestinal peptidase activity (53, 54). One of the major issues with using food-derived bioactive peptides as a therapeutic or food additive is their susceptibility to gastrointestinal digestion (55). Depending on where the peptide is first released, it can encounter over 20 human proteases and peptidases (56, 57) and an unknown number of microbial proteases (58). Knowing what inherent factors contribute to a peptide’s ability to resist digestion can aid in future bioactive peptide discovery by focusing on specific regions of the proteome or modifying previously discovered peptides to increase resistance.

Conclusion

This paper is the first to track the digestion of human milk proteins from milk to the infant stomach to stool and compare the peptide release between preterm and term infants. Preterm and term infants digested milk proteins differently. At the gastric phase of digestion, preterm infants had higher total peptide count and abundance than term infants of the same age, but term infants released specific bioactive peptides in significantly greater amounts. Many milk peptides were conserved across digestion, including some known bioactive peptides. We pinpointed a region encompassing the C-terminus of β-casein that was released in large amounts in the milk and stomach, survived to the stool, and hypothesized that it may play a protective role against bacterial infection in the infant intestinal tract. Future research should investigate this region of the C-terminus of β-casein to elucidate its effects on infant health, along with the factors that allow it to resist digestion better than the other milk proteins.