N-Glycan Profile of Cell Membrane as Probe for Lipopolysaccharide-induced Microglial Neuroinflammation Uncovers the Effects of Common Fatty Acid Supplementation

2.1. Cell culture and LPS treatment

The cell lines, SVG p12 – CRL-8621 (also SVG, human brain fibroblast astroglia), DAOY – HTB-186 (human desmoplastic cerebellar medulloblastoma) and HMC3 – CRL-3304 (human microglial cell) were all obtained from the American Type Culture Collection (ATCC, VA, USA). Both DAOY and HMC3 were grown in ATCC Eagle’s Minimum Essential Medium (EMEM) while SVG p12 was grown in ATCC RPMI-1640. Media were supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin (Thermo Scientific, MA, USA) and 30 uM of bovine serum albumin (1g of lyophilized BSA, Sigma Aldrich, MO, USA) which is used as a carrier to ensure FA dissolution. Cells were cultured in T75 flasks at 37 °C under 5% CO₂, with the media changed every other day. Upon reaching 80% confluency, cells were treated with O55:B5 lipopolysaccharide (Sigma Aldrich, USA) to a final concentration of 100 ng/mL and fatty acids with the following final concentration for long/medium-chain at 100uM and for short-chain at 1000uM, accordingly. After 24 hours incubation, cells were harvested for mass spectrometry-based analysis using previously reported general protocol (M. R. S. Alvarez et al., 2022; Li et al., 2020, 2019). Experiments were conducted with three biological replicates for each treatment.

2.2. Sample preparation for mass spectrometry analyses

Glycomic, proteomic and glycoproteomic analyses of the control and LPS- and fatty acid-treated HMC3 cell lines were performed following previously developed protocol which was optimized and described in Alvarez et al. (2022) and Li et al. (2020, 2019). Samples were prepared for N-glycan, peptide, and glycopeptide extraction from the cell surface membrane fractions prior mass spectrometric analyses. The samples were kept on ice during all steps of preparation. During harvest, samples were washed twice with cold PBS buffer (Dulbecco’s phosphate-buffered saline, Gibco) and were added with 1.2 mL of homogenization buffer (HB) containing 0.25 M sucrose, 20 mM HEPES-KOH (pH 7.4), 1:100 protease inhibitor cocktail (EMD Millipore, Temecula, CA, USA). The sample was then lysed using a sonicator probe (QSonica, Newtown, CT, USA). Then, the samples were centrifuged at 2000 g for 10 minutes at 4°C. The pellet containing cell debris and nucleus was discarded, and the supernatant was subsequently centrifuged at 42,000 rpm for 45 minutes at 4°C to separate cell debris from biomolecules of interest, e.g. N-glycans, proteins and glycoproteins. The pellet obtained enriched in cell membrane proteins and glycoproteins were washed with 500 μL of 0.2 M Na₂CO₃ and centrifuged at 42,000 rpm for 45 minutes at 4°C. Lastly, the subsequent pellet was washed with deionized water for 42,000 rpm for 45 minutes (4°C) to wash out excess salts and contaminants. The membrane fractions obtained were stored at −20°C until further analysis.

2.3. N-glycan release, PGC cleanup, and chip-QToF LC-MS/MS analysis

To one aliquot of the supernatant, treatment with dithiothreitol (DTT) and glycerol-free Peptide N-glycosidase F (PNGase F from New England Biolabs, MA, USA) and subsequent centrifugation will release the N-glycans into the supernatant. Prior to mass spectrometry analysis, N-glycans were desalted using porous graphitic carbon (PGC) SPE plates (Glygen Corp), then dried in vacuo using miVac (SP Scientific, PA, USA) and reconstituted in 30 μL ultrapure water.

LC-MS/MS was performed using Chip/Q-ToF LC-MS/MS at 5uL injection. N-glycan profiles were obtained using an Agilent nanochip-QTOF (quadrupole time-of-flight)-MS mass spectrometer (M. R. Alvarez et al., 2022; M. R. S. Alvarez et al., 2022; Li et al., 2020). N-glycan samples were reconstituted in 40 μL of water, and 5 μL of the resulting solution was used for injection into the nano-LC-MS/MS system. Separation was performed using an Agilent PGC-Chip II with a 40 nL enrichment and 43 mm x 75 μm analytical column (with a particle size of 5 μm) and a binary solvent system composed of mobile phase A (3% v/v acetonitrile and 0.1% v/v formic acid in water) and mobile phase B (90% v/v acetonitrile and 1% v/v formic acid in water). The gradient sequence for separation used was: 0–2.5 min, 1% B; 2.5–20 min, 16% B; 20–35 min, 58% B; 35–40 min, 100% B 40–50 min, 100% B; 50.01–65 min, 0% B with a flow rate of 0.3 μL/min. Tandem MS spectra were acquired via collision-induced dissociation (CID), with spectra measured at 0.8 s per spectrum in positive ion mode.

N-Glycomic data was analyzed using Agilent MassHunter Qualitative Analysis (B.08.00), with database matching to an in-house glycan database at 10 ppm mass error. Quantification was also performed to determine percent relative abundances of each glycan types – high-mannose, fucosylated, sialylated, and sialofucosylated. Discovery of significantly different glycans was analysed using multiple t-tests at a false discovery rate (FDR) of 5%. Potential biosynthetic pathway and structure was annotated by database matching and literature analysis.

2.4. Protein digestion, peptide cleanup and glycopeptide enrichment, nanoLC-Orbitrap LC-MS/MS analysis

After cell lysis and ultracentrifugation, samples were treated with 8 M urea and subsequently sonicated to dissolve the proteins and glycoproteins. Afterwards dithiothreitol was added and the samples were incubated for 50 minutes at 55 °C. An alkylating agent, iodoacetamide (IAA), was also added and subsequently incubated for 20 minutes in the dark to alkylate the cysteine residues. To quench the reaction, 50 mM ammonium bicarbonate was added to each reaction. Finally, 10uL of trypsin (with additional 10uL of Glu-C in glycoproteomics) was added to each reaction and subsequently incubated at 37 °C for 18 hours to cleave the proteins into peptide fragments. After the incubation period, samples for proteomic analysis were cleaned-up using C18 solid phase extraction (SPE) cartridges (Supelco DSC-18) while samples for glycoproteomic analysis were cleaned up using Hydrophilic Interaction Liquid Chromatography (HILIC) using iSPE HILIC cartridges (HILICON, 1mL, 100mg,). Prior to mass spectrometric analysis using a nanoLC paired with OrbiTrap, samples were dried and then dissolved in ultrapure water to make 0.5 μg/μL concentration.

Digested peptides and glycopeptides in one (1) μL of injected sample were separated using an Acclaim^™ PepMap^™ 100C18 LC Column (75 μm x 150 mm, particle size 2 μm; ThermoFisher Scientific) at a flow rate of 300 nL/min, and then analyzed in an UltiMate^™ WPS-3000RS nanoLC system coupled with an Orbitrap Fusion Lumos MS system (ThermoFisher Scientific). Solvents A and B were prepared using water containing 0.08% formic acid and 80% acetonitrile containing 0.1% formic acid, respectively. MS spectra with a mass range of m/z 700–2000 for MS1 and m/z ≥ 120 for MS2 were collected at a rate of 1.5 s per spectrum in positive ionization mode. The filtered precursor ions in each MS spectrum were subjected to fragmentation through 30% higher-energy C-trap dissociation (HCD) using nitrogen gas as a carrier.

Glycoprotein and protein expression data was calculated from total ion chromatogram (TIC)-normalized peptide intensities. Glycoproteomic and proteomic mass spectrometry data was analyzed using ProteinMetrics’s Byonic (v.3.5.0) for database matching and quantified using Byologic software (v.3.5.0). Specifically, glycoproteins were identified through filtering based on Score >100, Delta Mod score >10, and categorized based on the N-glycan type- high mannose, undecorated, fucosylated, sialylated and sialofucosylated. Differentially expressed proteins and glycoproteins were then analyzed using multiple t-tests at an FDR of 5%. Functional analysis was performed using the UniProt Knowledgebase, also UniProtKB (uniprot.org), and plotted using GraphPad prism (version 9.3.1) (M. R. S. Alvarez et al., 2022).

2.4. In silico molecular docking and protein-protein modeling

3D crystal structure of the human TLR4-human MD-2 complex, 3FXI (PDB DOI: https://doi.org/10.2210/pdb3fxi/pdb) (Park et al., 2009) from Protein Data Bank was selected and loaded onto PyRx (Dallakyan and Olson, 2015) and docked against dietary fatty acid and gut-microbe derived fatty acids using AutoDock VINA docking protocol (Trott and Olson, 2010). Binding against propionic acid (short chain fatty acid) and palmitic acid (long chain fatty acid) were visualized using Discovery Studio^™ (version 21.1.0.20298, Dassault Systemes, BIOVIA Corp, San Diego, CA, USA).

In silico modeling between SSRA, SSRB and STT3B was also done to predict involvement of sialofucosylation in TRAP-OST protein complexes interactions. The crystal structures of selected proteins were downloaded from PDB. Select glycans from glycopeptides identified from glycoproteomics analysis were attached to SSRA, SSRB and STT3B using CHARMM-GUI Glycan modeler using CHARMM36 force field for both proteins and carbohydrates (Huang et al., 2017). Using VMD, protein-protein interactions were visualized, and the number of interacting hydrogen bonds between the glycans and proteins was analyzed (Humphrey et al., 1996).

3.1. LC-MS/MS characterization of glycosylation in microglia

More than 70% of proteins in the brain are glycosylated which makes protein glycosylation necessary for proper brain development and maintenance (Barboza et al., 2021; Fang et al., 2020; Kizuka and Taniguchi, 2018; Sytnyk et al., 2021; Tena et al., 2022a, 2022b) wherein alterations in glycosylation may cause detrimental changes in the brain’s function (Akasaka-Manya et al., 2010; Gaunitz et al., 2021). N-Glycosylation in human and mouse brains are reportedly to commonly have high-mannose, fucosylated and sialofucosylated (Barboza et al., 2021; Tena et al., 2022a). The sialylation of the microglia is reported to attenuate the development of neurodegenerative disorders like AD by inhibiting neuroinflammation and phagocytosis (Puigdellívol et al., 2020). The neurodegenerative phenotype can be mimicked using induction of LPS in cells and since has been used as an in vitro model for Alzheimer’s Disease (Puigdellívol et al., 2020; Zhan et al., 2018).

To investigate the effect of LPS on glial activation, we first started with a panel of three cell lines – SVG p12 (astroglia), HMC3 (microglia) and DAOY (medulloblastoma) – on which we conducted N-glycomic and glycoproteomic analyses (Supplementary Figure 1). The N-glycomics results showed significant changes with high mannose and sialofucosylated structures upon LPS treatment. We observed the greatest changes, especially in the glycoproteome of HMC3, hence we chose the microglia cell model for further analyses.

Subsequently, we employed in vitro microglia cell models of LPS-induced neuroinflammation (Batista et al., 2019; Sheppard et al., 2019; Skrzypczak-Wiercioch and Sałat, 2022; Zhan et al., 2018; Zhao et al., 2019) to explore the effects of dietary and gut microbiota-derived fatty acids. We co-treated microglial cells with LPS and a physiological dose of dietary fatty acid (oleic, palmitic, or lauric) or gut microbiota-derived fatty acid (valeric, isobutyric, butyric, propionic) The LC-MS/MS workflow is depicted in Figure 1. These fatty acids were selected due to their high abundances in diet and in gut microbial fermentation. After treatments, cells were harvested and prepared for analysis using nanoLC-QtoF (for N-glycomics) and nLC-Orbitrap (for site-specific glycoproteomics). With the N-glycomic method, we quantified more than 300 unique N-glycan structures while the site-specific glycoproteomic method quantified more than 2,000 glycopeptides. Proteomics was also carried out to track the N-glycosylation pathways affected by LPS and the fatty acid treatments. We detected 19 proteins involved in N-glycan precursor synthesis and attachment of N-glycan precursor using label-free proteomics method (Supplementary Figure 2).

3.2. LPS targets high-mannose and sialofucosylated N-glycans in microglia

Upon treatment with LPS, we observed a drastic and significant drop in high-mannose N-glycans with a corresponding increase in the abundance of sialofucosylated N-glycans (Figure 2a, b). Analysis of each individual N-glycan structure revealed that LPS treatment of microglial cells decreased the abundance of all high-mannose type structures: Hex₉HexNAc₂, Hex₈HexNAc₂, Hex₇HexNAc₂, Hex₆HexNAc₂, Hex₅HexNAc₂ (Figure 2c). On the other hand, LPS treatment increased the abundance of specific sialofucosylated structures with varying degree of branching (bi-, tri-, and tetra-antennary) and varying degree of sialylation (mono- or bi-sialylated) (Figure 2c). A tandem MS spectra of select differentially expressed N-glycans were mapped to confirm structures based on fragments, i.e. high mannose N-glycan structure, Hex₉HexNAc₂ (Supplementary Figure 3a), and sialofucosylated tetra-antennary glycan, Hex₇HexNAc₆Fuc₁NeuAc₂ which is composed of 2 isomers with either core fucose or terminal fucose residues (Supplementary Figure 3b, 3c). Therefore, LPS significantly alters the microglial glycocalyx, increasing complex type of N-glycans. In the glycosylation pathway, high mannose structures represent early-stage N-glycans, typically added to nascent glycoproteins. Thus, decreases in high mannose is linked to dysregulated activities of mannosidases, enzymes responsible for trimming mannose residues during glycan maturation. On the other hand, increased sialofucosylation points towards enhanced addition of sialic and fucose residues to glycan structures. Sialofucosylated glycans are often found in mature glycoproteins on the cell surface. Therefore, the increase may be associated with hyperactive fucosyltransferases and sialytransferases, which are enzymes involved in adding fucose and sialic acid, respectively. These results coincide with the N-glycan biosynthesis pathway, such that with increased activity of mannosidase, pathway leads to creation of more complex-type N-glycans wherein can be acted upon by several fucosyltransferase and sialyltransferase enzymes. The increase in fucosyltransferase activity following LPS treatment aligns with previous findings linking FUT8 (fucosyltransferase catalyzing core-fucosylation) activity to the regulation of AβO-induced microglial activation(Jin et al., 2023). This finding is consistent with noticeable increase in core fucose levels observed during LPS treatment, a phenomenon detectable using our MS/MS fragmentation-based method and relative retention time analysis. As an example of probing the fucose position (either core or antennary) in N-glycans, we present herein sialofucosylated tetra-antennary N-glycan, Hex7HexNAc6Fuc1Sia1, which has an apparent increase in core fucosylation during neuroinflammation (Supplementary Figure 4).

We next determined the specificity of these changes in microglial glycosylation by LPS using site-specific glycoproteomics. Upon LPS treatment, we observed 28 glycoproteins altered glycosylation in over 30 unique glycosites (Figure 2d). We observed underexpression of SSRB and LAMP1 high mannose glycoforms, and overexpression of CD166, SSRA, and CD59 sialofucosylated glycoforms. ITB1, CD166, and CD59 glycoproteins are known to be important cell surface signaling molecules, while SSRA, SSRB, and LAMP1 are proteins involved in protein processing. The observed decrease in sialylation is consistent with the report that neuroinflammation can promote desialylation (Puigdellívol et al., 2020).

3.3. Dietary and gut microbiota-derived fatty acids ameliorate the effect of LPS in microglial glycosylation

We further investigated whether our diet and gut microbiota can alleviate the changes of LPS on microglia glycosylation. Diet-derived (oleic, palmitic, lauric) and gut microbiota-derived (valeric, butyric, isobutyric, propionic) fatty acids were co-treated with LPS into microglial cells followed by N-glycomic and site-specific glycoproteomic analyses.

Comparing the effects of diet-derived fatty acids (oleic, palmitic, lauric), we observed a distinctive reversal of the LPS-induced effects by oleic acid both in terms of high-mannose and sialofucosylated N-glycans (Figure 3a). We analyzed each individual high-mannose structure – Hex₉HexNAc₂, Hex₈HexNAc₂, Hex₇HexNAc₂, Hex₆HexNAc₂, Hex₅HexNAc₂ – and found that oleic acid increased high-mannose abundances to levels equivalent to that of the controls (Figure 3b). The equivalent effect was observed in sialofucosylated N-glycans, wherein oleic acid co-treatment reduced the effect of LPS in bi-, tri-, and tetra-antennary sialofucosylated N-glycans to levels comparable to the control (Figure 3c). Thus, among the diet-derived fatty acids studied, oleic acid was demonstrated the most pronounced ability to counteract the effects of LPS.

We also analyzed the effects of gut microbiota-derived fatty acids (valeric, butyric, isobutyric, and propionic). These compounds are known to be products of carbohydrate fermentation by bacteria in the human gut. In the co-treatment, we found that only isobutyric acid influenced the overall high-mannose reduction by LPS (Figure 4a). However, looking into each individual high-mannose N-glycan showed that both isobutyric and propionic acids also alleviated LPS effects on specific structures (Figure 4b). Isobutyric acid significantly increased the abundances of Hex₉HexNAc₂, Hex₈HexNAc₂, Hex₇HexNAc₂, Hex₆HexNAc₂, and Hex₅HexNAc₂, while propionic acid increased the amounts of Hex₉HexNAc₂, Hex₈HexNAc₂, Hex₆HexNAc₂, and Hex₅HexNAc₂. On the other hand, comparing the effects on sialofucosylated N-glycans showed that butyric, isobutyric, and propionic acids decreased the abundances of specific sialofucosylated structures (Figure 4c). Butyric acid reverted the effects of LPS on the tetra-antennary Hex₇HexNAc₆Fuc₁Sia₁ and Hex₇HexNAc₆Fuc₁Sia₂ N-glycans. Isobutyric acid reduced the amounts of bi-antennary (Hex₅HexNAc₄Fuc₁Sia_1, Hex₅HexNAc₄Fuc₁Sia₂), tri-antennary (Hex₆HexNAc₅Fuc₁Sia₁), and tetra-antennary (Hex₇HexNAc₆Fuc₁Sia₁, Hex₇HexNAc₆Fuc₁Sia₂) N-glycans. Lastly, propionic acid reduced the effects of LPS on bi- (Hex₅HexNAc₄Fuc₁Sia₁) and tri-antennary (Hex₆HexNAc₅Fuc₁Sia₁) N-glycans. From these results, the gut microbiota-derived fatty acids isobutyric, butyric, and propionic acids alleviated the effects of LPS in an N-glycan structure-specific manner.

3.4. Effects of LPS on N-glycosylation of specific glycoproteins are alleviated by dietary and gut microbiota-derived fatty acids

Based on the N-glycomic results, we found that both dietary and gut microbiota-derived fatty acids alleviate the effects of LPS on microglia glycosylation. Subsequently, we also determined the effects of LPS on specific glycoprotein wherein it is reducing the abundance of SSRB and LAMP1 high-mannose glycoforms, while increasing the abundance of CD166, SSRA, and CD59 sialofucosylated glycoforms. We next determined whether the fatty acids alleviated the effects of LPS on these specific glycoproteins using glycoproteomic analysis.

With regard to the dietary fatty acids (oleic, palmitic, lauric), we found that both oleic and palmitic acids alleviated the effect of LPS on SSRB protein’s high-mannose N-glycan Hex₉HexNAc₂ (Figure 5a). On the other hand, oleic acid reverted the effects of LPS on sialofucosylated N-glycans in CD166 (Hex₅HexNAc₄Fuc₁Sia₂) and SSRA (Hex₆HexNAc₅Fuc₁Sia₁, Hex₅HexNAc₅Fuc₁Sia₁) (Figure 5b). Lauric acid was also able to reduce the levels of sialofucosylated N-glycan in CD166 (Hex₅HexNAc₄Fuc₁Sia₂).

Likewise, we also observed that gut microbe-derived fatty acids (valeric, butyric, isobutyric, propionic) ameliorated select LPS effects on microglia glycosylation. In terms of high mannose glycosylation, valeric acid reverted SSRB (Hex₉HexNAc₂, Hex₈HexNAc₂) glycosylation while butyric acid reverted SSRA (Hex₉HexNAc₂, Hex₈HexNAc₂) glycosylation (Figure 6a). In terms of sialofucosylated species, isobutyric acid reverted the effects of LPS on CD63 (Hex₆HexNAc₅Fuc₄Sia₂), CD166 (Hex₅HexNAc₄Fuc₁Sia₂), and SSRA (Hex₆HexNAc₅Fuc₁Sia₁, Hex₅HexNAc₅Fuc₁Sia₁) glycosylation (Figure 6b). Interestingly, all 4 gut microbe-derived fatty acids reversed LPS effect on the sialofucosylated glycosylation of CD166 (Hex₅HexNAc₄Fuc₁Sia₂).

The glycoproteomics results highlight the observed effect of the fatty acids in the microglia specifically on the high-mannose and sialofucosylated glycoforms of translocon-associated proteins (TRAP) which regulate the retention of ER resident proteins. It has been previously established that N-linked glycosylation is required for endoplasmic reticulum (ER) homeostasis(Helenius and Aebi, 2004; Losfeld et al., 2014; Ng et al., 2019) wherein the TRAP complex, composed of SSRA, SSRB, SSRC and SSRD subunits, were recently discovered through model-based analysis of genome-wide CRISPR-Cas9 knockout (MAGECK) analysis, to be responsible in regulating the quality control of N-linked glycosylation during ER stress conditions in human cells (e.g. A549 and HEK-293 cells) (Phoomak et al., 2021). Our study shows that the high mannose glycoforms of these TRAP proteins especially SSRA and SSRB are significantly decreased while the sialofucosylated glycoforms are significantly increased in the presence of LPS. Interestingly, our results coincide with the previous finding that SSRA and SSRB were associated with BiP, an ER stress-induced chaperone which prevents SSRA and SSRB to function properly during excessive ER stress (Phoomak et al., 2021). Based on our results, we hypothesize that upon inducing ER stress, LPS alters SSRB and SSRA ‘s N-glycosylation. We also showed through the significant site-specific effect of oleic (Hex9HexNAc2 - SSRB glycosite Asn88), palmitic (Hex9HexNAc2 SSRB glycosite Asn88), valeric (Hex9HexNAc2 and Hex8HexNAc2 - SSRB glycosite Asn88) and butyric (Hex9HexNAc2, Hex8HexNAc2, Hex6HexNAc5Fuc1NeuAc1 and Hex5HexNAc5Fuc1NeuAc1 of SSRA glycosite Asn136) acids that select fatty acids can modulate high mannose and sialofucosylated glycoforms of SSRA and SSRB to counter the site-specific effects of LPS-induced ER stress in microglial cells. Additionally, changes in sialofucosylation of cluster of differentiation (CD) glycoproteins brought about by LPS were also observed specifically targeting CD63 and CD166 which are both considered important cell surface proteins that facilitate cell adhesion and migration during neuroinflammation (Kim et al., 2016; Lyck et al., 2017). Both CD63 and CD166 have been identified as reliable indicators in plasma and serum samples related to cognitive decline and synaptic degeneration (Chen et al., 2023; Kim et al., 2016; Lyck et al., 2017; Vaz et al., 2022).

3.5. Proteomics reveal reversal of LPS effect by dietary and gut microbiota-derived fatty acids

Label-free proteomics analyses reveal effect of LPS on the microglial glycocalyx wherein it targets proteins involved in specific N-glycosylation pathways particularly, N-glycan precursor synthesis and attachment of N-glycan precursor. The heat map in Supplementary Figure 2 show the over over- and underexpressed N-glycosylation-related proteins in different fatty acid supplementation following LPS induction. Interestingly, the fatty acids effectively counteract the effects of LPS on ALG5, a protein important in N-glycan precursor synthesis as well as on some components of oligosaccharyltransferase (OST), RPN1, RPN2 and STT3A. OST is a membrane protein complex critical in the first step in protein N-glycosylation, that initiates transfer of Glc₃Man₉GlcNAc₂ from the lipid carrier dolichol-pyrophosphate to an asparagine residue within an Asn-X-Ser/Thr consensus motif (X being any amino acid except proline) in nascent polypeptide chains in the lumen of the endoplasmic reticulum (Varki et al., 2022).

3.5. In silico docking confirms dietary fatty acids inhibit LPS

Based on our comprehensive analysis of both N-glycans and glycoproteins data, we observed a more pronounced alleviation of LPS-induced effects by dietary fatty acids compared to short-chain fatty acids derived from the gut microbiota. Looking into the mechanism of LPS, Lipid A core of LPS molecules bind to the myeloid differentiation factor 2 (MD-2) protein hydrophobic pocket in order to enact its effect (Maeshima and Fernandez, 2013; Park et al., 2009) . In silico molecular docking to the MD-2 receptor (known binding site of LPS) in immune cells showed that dietary fatty acids (long- and medium-chain) bind more than gut microbe-derived fatty acids (short-chain) as the former fatty acid lipid tail can interact with the protein’s hydrophobic core with the fatty-acyl group exposed to the solvent surface (Supplementary Figure 5).

Notably, oleic acid demonstrated the most favorable binding energy among all tested fatty acids. Thus, the modeling results show that the dietary fatty acids may enact their alleviating effect by competing with LPS and blocking the binding site. In contrast, the shorter gut microbe-derived fatty acids may enact their effect through a different pathway besides binding to the MD-2 receptor.

3.6. In silico modeling shows sialofucosylation is critical to OST- TRAP interaction

We deduced the possibility of interaction of oligosaccharyl transferase (OST) complex and translocon-associated protein (TRAP) complex from our proteomic and glycoproteomic analysis (Supplementary Figure 6a). TRAP complex is found close to the OST complex which interacts to both the translocon and the ribosome (Russo, 2020). To configure how the complexes interact, we performed an in silico analysis to show the binding between SSRA/SSRB, TRAP components and STT3B, catalytic subunit of the OST complex that catalyzes the first step in protein N-glycosylation (Supplementary Figure 6b). The modeling highlights importance of sialofucosylation in the binding between SSRA and STT3B proteins wherein the Hex5HexNAc5Fuc1NeuAc1 of glycopeptide FYIN₁₃₆FTAL from SSRA is shown to be essential for the binding to occur.