A combined microphysiological-computational omics approach in dietary protein evaluation

Food security is under increased pressure due to the ever-growing world population. To tackle this, alternative protein sources need to be evaluated for nutritional value, which requires information on digesta peptide composition in comparison to established protein sources and coupling to biological parameters. Here, a combined experimental and computational approach is presented, which compared seventeen protein sources with cow’s whey protein concentrate (WPC) as the benchmark. In vitro digestion of proteins was followed by proteomics analysis and statistical model-based clustering. Information on digesta peptide composition resulted in 3 cluster groups, primarily driven by the peptide overlap with the benchmark protein WPC. Functional protein data was then incorporated in the computational model after evaluating the effects of eighteen protein digests on intestinal barrier integrity, viability, brush border enzyme activity, and immune parameters using a bioengineered intestine as microphysiological gut system. This resulted in 6 cluster groups. Biological clustering was driven by viability, brush border enzyme activity, and significant differences in immune parameters. Finally, a combination of proteomic and biological efficacy data resulted in 5 clusters groups, driven by a combination of digesta peptide composition and biological effects. The key finding of our holistic approach is that protein source (animal, plant or alternative derived) is not a driving force behind the delivery of bioactive peptides and their biological efficacy.


Endotoxin (EU/mL*)
Total unique peptides Number of ion-peaks NA Supplementary figure 1. Proteomic analysis of potential dietary protein in comparison to whey protein concentrate (WPC). Protein samples were first corrected for digestive enzyme ion-peaks. Thereafter, A) digestive peptide profile was made by subdividing based on molar mass; dipeptides range from 150.2 -408.4 g/mol, tripeptides is 225.3 -612.6 g/mol everything above 300.5 g/mol is considered an oligopeptide. The values were expressed relative to the total ion-peak abundance B) shows the total abundance of all ion-peaks per protein, C) shows the total amount of unique ion-peaks per protein, D) provides the percentage of WPC overlapping ion peaks (based on charge, mass/charge ratio (±0.003) and retention time (±0.5min)) in contrast to the protein its total amount of ion-peaks and E) shows the abundance of WPC overlapping ion peaks in contrast to the total abundance per protein.

Soy
Tile plot of peptide overlap and direct comparison in peptide concentration of Soy in comparison to WPC. Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

NPP
Tile plot of peptide overlap and direct comparison in peptide concentration of NPP in comparison to WPC. Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

BP
Tile plot of peptide overlap and direct comparison in peptide concentration of BP in comparison to WPC. Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

NWP Tile plot of peptide overlap and direct comparison in peptide concentration of NWP in comparison to WPC.
Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

TPP1
Tile plot of peptide overlap and direct comparison in peptide concentration of TPP1 in comparison to WPC. Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

Tile plot of peptide overlap and direct comparison in peptide concentration of LMC2 in comparison to WPC.
Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

LMC2 QM
Tile plot of peptide overlap and direct comparison in peptide concentration of QM in comparison to WPC. Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

Tile plot of peptide overlap and direct comparison in peptide concentration of TPP2 in comparison to WPC.
Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

Wheat
Tile plot of peptide overlap and direct comparison in peptide concentration of Wheat in comparison to WPC. Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

Tile plot of peptide overlap and direct comparison in peptide concentration of QrtoE in comparison to WPC.
Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

YE
Tile plot of peptide overlap and direct comparison in peptide concentration of YE in comparison to WPC. Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl. Table 1.

Tile plot of peptide overlap and direct comparison in peptide concentration of Corn in comparison to WPC.
Left shows the tile plot visualizing the amount of overlap, and non-overlapping peptides (ion peaks) sorted by ID number. The right figure visualizes quantities of WPC-matched peptides (ion peaks) sorted by ID number. ID numbers including information can be found in Suppl.  figure 3. Shows Bayesian Information Criterion (BIC) analysis of the proteomic data set. Showing EEV as the best model when using 3 different clusters.
Supplementary figure 4, all varieties of two component plots showing the protein clustering based on proteomic data. Protein clustering was achieved based on the statistical cluster model; ellipsoidal equal volume, shape and orientation (EEV). Considered data were peptide fractions, peptide abundance, total amount of unique peptides, relative overlap with whey protein concentrate (WPC) and the relative peptide abundance of the WPC overlap. Bioengineered intestinal tubules were exposed to potential dietary proteins assessing the effect on epithelial barrier, viability and alkaline phosphatase activity. Experiment ready bioengineered intestinal tubules were exposed to in vitro digested proteins (diluted 1:4 in cultivation medium) for 3 hours. As measure for epithelial barrier integrity, inulin-FITC leakage (A) and zonula occludens-1 intersects corrected for surface area and number of nuclei (B) were quantified. As measure for cell viability mitochondrial activity was assessed (C) and for brush border enzyme activity alkaline phosphatase activity was assessed (D). Values of C and D were corrected for bioengineered intestinal tubule length. Data are presented relative to positive control (experiment ready bioengineered intestinal tubule exposed to culture medium), data are presented as mean ± SEM of n=3-7 independent experiments, corrected for outliers (1 outliers of 400 data points) and tested for significance using one-way ANOVA and student T-test, ** P < 0.005. Bioengineered intestinal tubules were exposed to potential dietary proteins assessing the effect on IL-6, TGFβ and nitric oxide (NO) secretion. Experiment ready bioengineered intestinal tubules were exposed to in vitro digested proteins (diluted 1:4 in cultivation medium) for 3 hours. Thereafter, supernatant was collected and analyzed for, IL-6 (A), TGF-β (B), NO (C). Values were corrected for basal levels in culture medium of negative control (bioengineered intestinal tubule without cells). Data are presented as mean ± SEM of at least 3-7 independent experiments, corrected for outliers (12 outliers of 306 data points) and tested for significance using one-way ANOVA, * P < 0.05, *** P < 0.001, **** P < 0.0001. . Shows Bayesian Information Criterion (BIC) analysis of the in vitro biological efficacy data. Showing diagonal equal volume and shape as the best model when using 6 different clusters.