Proteomic and N-glycomic comparison of synthetic and bovine whey proteins and their effect on human gut microbiomes in vitro

Advances in food production systems and customer acceptance have led to the commercial launch of dietary proteins produced via modern biotechnological approaches as alternatives to traditional agricultural sources. At the same time, a deeper understanding of how dietary components interact with the gut microbiome has highlighted the importance of understanding the nuances underpinning diet-microbiome interactions. Novel food proteins with distinct post-translational modifications resulting from their respective production systems have not been characterized, nor how they may differ from their traditionally produced counterparts. To address this, we have characterized the protein composition and N-glycome of a yeast-synthesized whey protein ingredient isolated from commercially available  ice cream and compared this novel ingredient to whey protein powder isolate derived from bovine milk. We found that despite strong similarities in protein composition, the N-glycome significantly differs between these protein sources, reflecting the biosynthetic machinery of the production systems. Further, the composition profile and diversity of proteins found in the synthetic whey protein were lower relative to bovine whey protein, despite both being predominantly composed of β-lactoglobulin. Finally, to understand whether these differences in N-glycome profiles affected the human gut microbiome, we tested these proteins in an in vitro fecal fermentation model. We found that the distinct whey protein sources generated significant differences in microbial compositions, which we hypothesize is a product of differences in N-glycan composition and degradation by these representative microbial communities. This work highlights the need to understand how differences in novel biotechnological systems affect the bioactivity of these proteins, and how these differences impact the human gut microbiome.

Methods
Enzymatic deglycosylation of denatured protein samples was performed with PNGase F (1 Unit/μL) obtained from Promega (Madison, WI, USA). First, dried purified protein samples (1 mg) were dissolved in 50 μL of 2% SDS and denatured by incubations at 60°C. Denatured protein samples were then mixed with 2% NP-40 solution and 5X PBS and 1 U PNGase F was added and incubated at 37°C overnight. Finally, the samples were centrifuged, and supernatants were collected for further analysis. After enzymatic deglycosylation of protein samples, released N-glycans from each sample were labeled with a 2-AA tag. 50 μL of 2-AA tag (48 mg/mL−1 in DMSO/glacial acetic acid, 7/3, v/v) and 50 μL of 2-sodium cyanoborohydride (60 mg/mL in DMSO/glacial acetic acid, 7:3 w/v) were added to the released glycan samples (50 μL). Subsequently, the mixtures were incubated at 65°C for two hours. Purification of N-glycans was achieved by solid-phase extraction cartridges containing cellulose and porous graphitized carbon materials, as previously described by Kayili et al (2021). MALDI-TOF(/TOF)-MS analysis of 2-AA labeled N-glycans from bovine and synthetic whey protein samples was carried out on a Bruker rapifleX™ MALDI Tissuetyper™ (Bruker Daltonik GmbH, Bremen, Germany) equipped with a SmartBeam 3D laser system. On the AnchorChip MALDI-target plate, the purified N-glycans (1 μL) were spotted and allowed to dry. Then, 1 μL of DHB matrix (5 mg/mL−1 in ACN/H2O, 1/1, v/v comprising 0.1% ortho-phosphoric acid) was added. The analysis included a 20 kV acceleration voltage, a 160 ns extraction delay, and the summation of 8000 shots at 2000 Hz for each spectrum. The mass range of 1000-5000 Da was used to produce all spectra using a random walk pattern in negative ion and reflectron mode. Data obtained by MALDI-TOF(/TOF)-MS analysis were processed using Flex Analysis v.4.0 software (Bruker Daltonik Gmbh). Peaks of 2-AA labeled N-glycans were inserted into ProteinScape software including the GlycoQuest algorithm (Bruker Daltonik GmbH, Bremen, Germany) for glycan identification. Total area normalization was used to determine the relative abundance of individual N-glycans (mass-intensity based). All experiments were performed with three technical replicates.