Role of the afferent lymph as an immunological conduit to analyze tissue antigenic and inflammatory load

SUMMARY

The lymphatic fluid is the conduit by which part of the tissue “omics” is transported to the draining lymph node for immunosurveillance. Following cannulation of the pre-nodal cervical and mesenteric afferent lymphatics, herein we investigate the lymph proteomic composition, uncovering that its composition varies according to the tissue of origin. Tissue specificity is also reflected in the dendritic cell-major histocompatibility complex class II-eluted immunopeptidome harvested from the cervical and mesenteric nodes. Following inflammatory disruption of the gut barrier, the lymph antigenic and inflammatory loads are analyzed in both mice and subjects with inflammatory bowel diseases. Gastrointestinal tissue damage reflects the lymph inflammatory and damage-associated molecular pattern signatures, microbiome-derived by-products, and immunomodulatory molecules, including metabolites of the gut-brain axis, mapped in the afferent mesenteric lymph. Our data point to the relevance of the lymphatic fluid to probe the tissue-specific antigenic and inflammatory load transported to the draining lymph node for immunosurveillance.

In brief

Nanaware et al. report the tissue-specific composition of the cervical and mesenteric lymph in physiological and inflammatory conditions in subjects with inflammatory bowel diseases and a mouse model. The tissue specificity is mirrored in the nodal dendritic cell-MHC-II immunopeptidome, including microbiome-related peptides. The analysis showcases the lymph’s crucial role in immunosurveillance.

Graphical Abstract

INTRODUCTION

The lymphatic system is often referred to as the third circulatory system. It comprises lymphatic capillaries, draining into a network of progressively larger lymphatic vessels, which will eventually flow into the left subclavian vein and then the thoracic duct or the right subclavian vein and the right lymphatic duct. Thus, eventually, through the venous system, the lymphatic fluid will drain into the heart, making the lymphatic system an open circulation with a unidirectional flow.^1–5

Central to the lymphatic system functionality is the draining of the lymphatic fluid into one of the approximately 600–800 lymph nodes distributed across the human body, present as a single lymph node or, more often, distributed in nodal chains. As exceptions to the rule, lymphatic vessels are not found in the bone marrow and the brain parenchyma but are present in the meninges,^6,7 and the thymus and spleen do not have afferent but only efferent lymphatic vessels.

Immunosurveillance refers to the constant monitoring by the immune system for pathogens and “aberrant/transformed” cells, as well as maintenance of immune tolerance to “self.”^8–10 This function is mostly performed by the secondary lymphoid organs, which act as critical checkpoints, anatomically organized to capture the tissue-draining self-proteome, as well as invading pathogens and/or infected and malignantly transformed cells, and they orchestrate a tailored immune response.^11,12 Continuous influx of afferent lymph, derived from the interstitial fluid bathing all parenchymal organs, is critical for their function.^13–18 Acting as a conduit, the lymphatic fluid collects and transports pathogens from injured mucosal tissues, such as those in the lung, gut, and genitourinary system and damaged subcutaneous tissue. It also transports apoptotic, necrotic, and tumor cells, in addition to organ- or tumor-derived extracellular vesicles and exosomes, showcasing the lymphatic system’s comprehensive role in bodily defense and surveillance.^14,18

In addition to the cellular components, the lymphatic system also transports a soluble proteome, originating from the interstitial fluid from various anatomical locations to specific lymph nodes.^13,15,19,20 Typically, each lymph node receives lymph from five to eight afferent lymphatic vessels, transporting the incoming cellular and soluble “ome” to be phagocytosed and processed by the macrophages and dendritic cells (DCs) present in the subcapsular and medullary sinuses. The generated major histocompatibility complex class I (MHC-I) and MHC-II immunopeptidomes are then presented to T cells for maintenance of immune tolerance to “self” or generation of immunity.^21–23

Similarly, lymph-carried damage-associated molecular patterns (DAMPs) and pathogen-associated molecular patterns will initiate innate nodal immune responses.

To accurately measure the variations in lymph composition and the MHC-II antigenic load on DCs from different anatomical areas, both under normal and inflammatory conditions, we collected afferent lymph draining to the cervical and mesenteric lymph nodes. This collection was performed in healthy mice and in mice with disrupted gut barriers, caused by dextran sulfate sodium (DSS)-induced tissue damage, as well as individuals with inflammatory bowel disease (IBD). Our analysis distinctly showcases the lymphatic fluid’s crucial role as a conduit for transporting the soluble proteome to the lymph nodes designated for immunosurveillance.

RESULTS

A tissue-specific proteome is mapped in both pre-nodal cervical and mesenteric lymph

To determine the composition of the lymphatic fluid draining from the brain as well as the intestine, pre-nodal/post-nodal afferent lymph was collected from afferent lymphatics draining to the deep cervical and mesenteric nodes respectively (Figures 1A–1D, Videos S1 and S2.²⁴ Advanced label-free quantitative (LFQ) analysis (n = 4 biological replicates) determined that, albeit most of the proteome (68% of the total proteome, 5,148 proteins) was shared between cervical and mesenteric lymph (Figures 1E and 1F), quantitative and qualitative differences in the composition of the lymph harvested from the two anatomical districts could be observed (Figures 1G–1N, S1, and S2; Table S1).

Figure 1. Differential proteomic profile between mesenteric and cervical lymph.

(A) Pre-nodal mesenteric lymphatic vessel draining to the mesenteric node and inset of the same pre-nodal lymphatic vessel following peri-lymphatic fat removal.

(B) Pipette tip (2–4 μm) showing the collected lymph fluid from the pre-nodal lymphatic vessel.

(C) Pre-nodal cervical afferent lymphatic vessel draining to the deep cervical node.

(D) Detail of the cannulation of a pre-nodal lymphatic vessel.

(E) Separation of the mesenteric and cervical lymph proteome (3 μg of protein) on a silver-stained 4%–20% gradient acrylamide SDS-PAGE.

(F) Deep Venn area proportional diagram displaying the degree of overlap and differential expression profiling of proteins identified in the mesenteric and cervical lymph using a combination of LFQ proteomics platforms (see STAR Methods protocol).

(G) Volcano plot depicting the significant differential expression (n = 4, p < 0.05 by t test) of 2,752 proteins. Highlighted in blue and red are the 681 proteins showing at least 2-fold downregulation and 799 proteins showing at least 2-fold upregulation in the cervical vs. mesenteric lymph proteomes.

(H) Principal component analysis (PCA), generated by MetaboAnalyst, based on the DIA intensities of the proteins identified in n = 4 biological replicates of afferent mesenteric and cervical lymph.

(I–K) LFQ comparative analysis of selected proteins displaying significant differences in their relative abundance (extracted from DIA intensities) as determined in (G). Blue and orange dots depict the mesenteric and cervical lymph proteins respectively. Statistical significance was determined using the Holm-Sidak method, with alpha = 0.05: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

(L) Euclidian Ward’s dual heatmap of the top 2,000 proteins identified in all four biological replicates of mouse pre-nodal mesenteric and cervical lymph. The map highlights the significant difference in the proteomic signature of the two biological fluids harvested from the distinct anatomical regions. Only proteins that passed a selected significance statistical threshold (ANOVA/t test applied in PEAKS XPro, p < 0.05) are displayed in the heatmap.

(M and N) The subsequently generated METASCAPE analysis of protein networks and GO annotations defining the different molecular functions and metabolic pathways of proteins from mesenteric (M) vs. cervical (N) lymph. All proteins and details of LFQ analysis and GO annotations related to the mesenteric and cervical proteomic analysis are presented in Table S1.

Bioinformatic analysis using complementary METASCAPE, ShinyGO, and ingenuity pathway analysis (IPA) platform led to the identification of major protein networks differentiating the cervical from the mesenteric lymph proteomes (Figures 1L–1N). Most of the proteome present in the mesenteric afferent lymph showed a signature of proteins involved in different metabolic pathways associated with lipoprotein transport and lipid metabolism, such as adipocyte-type fatty acid-binding protein, apolipoproteins A, B, C, and E, and phospholipid transfer proteins, consistent with the known role of the mesenteric lymph in chylomicron transport²⁵ (Figures 1G–1N, S1, and S2; Table S1). Proteins involved in pancreas beta-cell function and glucose homeostasis (adiponectin, adipsin, insulin-binding protein, and lithostathine) or growth factors (hepatocyte growth factor-like protein, insulin-like growth factor, and macrophage colony-stimulating factor 1) were also mapped in the afferent mesenteric lymph (Figures 1G–1N, S1, and S2; Table S1). Similarly, vitamin-binding proteins, including afamin, vitamin K and D binding proteins, and biotinidase, as well as chaperones such as MUP-1, were quantitatively more abundant in the mesenteric lymph (Figures 1G–1N, S1, and S2; Table S1), consistent with the notion that the pancreas and the gut both drain to the mesenteric lymph node.²⁶ Network analysis on the mesenteric afferent lymph unique/enriched proteome highlighted pathways associated with lipase and hydrolase activity, lipoprotein remodeling, fat digestion and absorption, triglyceride catabolism, and gut-associated immune cell and cytokine responses (Figures 1L and 1M; Table S1).

A brain-specific or highly enriched proteome, including glia maturation factor, nerve growth factor, mesencephalic astrocyte-derived neurotrophic factor, alpha-crystallin, brain-specific isoform of glycogen phosphorylase, and proteins associated with voltage-dependent channels, was uniquely observed in the lymph harvested from the afferent lymphatics entering the deep cervical nodes. (Figures 1G–1N, S1, and S2; Table S1). Network analysis on the afferent cervical unique/enriched proteome highlighted pathways associated with neurotransmitter release cycle, synaptic transmission, neuronal development, mitochondrial activity, and an overall central nervous system (CNS) proteome (Figures 1L and 1N; Table S1).

At the metabolomic level, amino acids and their derivatives were also represented in the mesenteric afferent lymph, followed by monosaccharides and dicarboxylic acids, such as adipic, malonic, glutaric, and succinic (Figures S3 and S4; Table S3). Primary and secondary bile acids were also observed (Figure S3; Table S3) indicating that, though the majority of bile acids travel through the enterohepatic circulation,²⁷ a percentage is also present in the mesenteric lymph. As expected, lipids, such as phosphatidylcholine, phosphatidylserine, and phosphatidylethanolamine as well as triglycerides and fatty acids were present in the afferent lymph, considering that chylomicrons are mostly absorbed by the lacteals²⁵ (Figures S3 and S4).

These results suggest quantitative and qualitative differences in the lymph composition of mesenteric and cervical anatomical districts.

Protein tissue specificity is highlighted in the MHC-II immunopeptidome

Next, to understand the contribution of the lymph proteome to the MHC-II immunopeptidome, we eluted I-A^b complexes from DCs harvested from the deep cervical and mesenteric nodes. Heatmap representation and cluster analysis indicated differences among the two immunopeptidomes (Figures 2A–2C).

Figure 2. Differential I-A^b elution profile from dendritic cells harvested from the mesenteric or deep cervical nodes.

(A) Representative clustered heatmap contrasting the top 600 I-A^b eluted peptides from nodal dendritic cells (DCs), harvested from cervical and mesenteric lymph nodes. Peptides exhibit a significant differential expression (n = 3, p < 0.05 by t test). The colors indicate the peptides’ average relative abundance calculated from the log₂ (DIA MS exclusive intensities) indexed in TableS4: red = increase (log₂ DIA MS intensities ≥0); blue = decrease (log2 DIA MS intensities ≤0). The map was generated in MetaboAnalyst using the Pearson distance and Ward clustering method, after filtering, normalization, and autoscaling of Log₂ (MS DIA intensities) for all peptides identified with FDR <5% using both PEAKS and Scaffold DIA proteome software.

(B) Principal component analysis (PCA), generated by MetaboAnalyst, using the DIA MS intensities of the I-A^b cervical and mesenteric eluted peptidomes (n = 3) shown in the clustered map in (A).

(C) Person correlation plot among the cervical and mesenteric replicates indicates differences between the I-A^b immunopeptidomes eluted from cervical or mesenteric DCs.

(D) The MHCMotifDecon 1.1 algorithm for motif deconvolution of multi-allele immunopetidomics data was used for motif analysis of the I-A^b eluted peptides. Peptides within a 9–35 amino acid length and percent rank of ≤30% were selected for the motif analysis. 505 I-A^b-eluted peptides from the cervical nodes and 456 I-A^b-eluted peptides from the mesenteric nodes passed the criteria for I-A^b binders.

(E) Predicted rank of the I-A^b bound peptidome did not show statistical difference between the cervical and mesenteric node ligandome. The significance was calculated using Mann-Whitney test.

(F) Length distribution for the filtered I-A^b binders. The median length distribution is 15 aa for both cervical and mesenteric lymph nodes.

(G) Deep Venn area proportional diagram displaying the degree of overlap between the cervical and mesenteric I-A^b-eluted peptidome.

(H) Deep Venn area proportional diagram displaying the percentage of protein source overlap between the cervical or mesenteric lymph proteome within the I-A^b-eluted peptidome from cervical and mesenteric lymph nodes.

(I) Fragmentation profiles of some of the identified mouse peptides matched against a generated mouse I-A^b DDA and DIA spectral library or against the FASTA mouse SwissProt database. The predicted affinity and the percent rank, predicted using NetMHCIIpan 4.1 algorithm, are indicated. The predicted binding core is underlined.

(J) The MHCMotifDecon 1.1 algorithm for motif deconvolution of multi-allele immunopetidomics data was used for motif analysis of the mesenteric and cervical peptidomes shown in Table S4N. Peptides within a 9–35 amino acid length and percent rank of ≤30% were selected for the motif analysis. The motif prediction showed that about 7,157 endogenous processed peptides identified in the cervical and mesenteric lymph by DIA timsTOF-PASEF method could be potential I-A^b MHC-II binders (Table S4N).

(K) Example of peptide epitopes derived from albumin (ALBU) and apolipoprotein E (Apo-E) found in both the mesenteric lymph and the I-A^b immunopeptidome eluted from the mesenteric nodes and identified by de novo sequencing algorithm built into PEAKS.

(L) MS/MS spectra of the peptides in (K).

Overall, all analyzed peptides displayed the expected I-A^b binding motives, binding affinity, and range of peptide length (Figures 2D–2F). A combination of data-dependent (DDA) and data-independent (DIA) analysis indicated that around 36% of the eluted peptides were shared by both the cervical and mesenteric lymph nodes (Figure 2G; Table S4). The remaining peptides eluted from each lymph node were unique, mirroring the distinct proteome characteristic of the two anatomical districts (Figures 2G–2I; Table S4).

Specifically, peptides derived from proteins functionally annotated to be involved in a variety of CNS functions, such as NREM sleep, regulation of oligodendrocytes survival, and regulation of sensory perception of sound (F107B), were highly enriched in the protein source of cervical I-A^b eluted peptidome. In addition, the neuronal membrane protein brain acidic protein 1 (BASP1), apolipoproteins B100 and E, both involved in CNS-related pathology, and several brain-specific isoforms of enzymes of the glycolytic/trichloroacetic acid (TCA) pathways were uniquely present in the I-A^b immunopeptidome eluted from cervical lymph nodes (Figures 2G–2I; Table S4). On the other hand, the I-A^b immunopeptidome eluted from mesenteric lymph nodes was enriched by peptides derived from metabolic enzymes such as enolases A and B, pyruvate kinase, lactate dehydrogenases A and B, phosphoglycerate mutase 2, cytoplasmic malate dehydrogenase, and phosphoglycerate kinase 1, among others (Table S4). Several transport proteins highly expressed in the intestinal Peyer’s patches and proteins involved in glucose transport (SLC2A4/GLUT4) were also highly represented in the I-A^b immunopeptidome eluted from mesenteric lymph nodes (Table S4).

Finally, in the immunopeptidome eluted from cervical or mesenteric nodes, we also mapped 10 and 12 peptides found in the cervical or mesenteric lymph, respectively (Figures 2J–2L; Table S4). Among these, peptides from apolipoprotein E and albumin were found in both the mesenteric lymph and the I-A^b immunopeptidome from mesenteric nodes (Figures 2J–2L). Similarly, peptides from the Sec61 protein and thioredoxin were present in both the cervical lymph and the I-A^b immunopeptidome from cervical nodes. (Table S4N).

The findings indicate that specific tissue antigens contribute to observable differences in the I-A^b immunopeptidome extracted from nodal DCs originating from various anatomical districts. Furthermore, some of these differences can be attributed to the distinct proteomic profiles present in the lymph from those regions.

DSS-induced inflammation is reflected in the proteomic composition of mouse mesenteric lymph

Next, we aimed to map possible changes in the lymph composition following inflammatory damage. For that, healthy mice were fed DSS, a sulfated polysaccharide that causes ulcerative colitis-like pathologies as well as inflammation in the intestinal mucosa²⁸ (Figure S5). Mesenteric lymph was collected in healthy and DSS-fed mice as previously described^14,24 (Figures 1A and 1B). As expected, following 1 week of DSS treatment, mice lost weight and blood was detected in the stools (Figures S5A and S5B). Macroscopically, there was a reduction in the intestinal length, as previously reported in this model (Figures S5C and S5D), with a disruption of normal mucosal integrity associated with extensive inflammation and immune cell infiltration (Figures S5E–S5H). The histological features phenocopied the epithelial disruption and inflammation observed in intestinal specimens from IBD patients (Figures S5I and S5J). Furthermore, as expected, we observed enhanced permeability of the gut epithelium due to cell damage and inflammation. This was determined by evaluating the lymph fluorescence following the oral administration of 4 kDa fluorescein isothiocyanate-dextran. (Figure S5I).

Upon DSS-induced damage, the profile of the afferent lymph drastically changed compared to the control (Figure 3A, Tables S1 and S2). The “omic” signature was characterized by several proteins present in the afferent DSS lymph but not in the control (Figure 3A, Tables S1 and S2), and principal component analysis (PCA) analyses revealed differential clustering between control and DSS pre-nodal lymph (Figure 3B). The most upregulated proteins were associated with inflammation, gastrointestinal tissue injury, oxidative stress, and nutritional diseases (Figures 3C–3E, Tables S1 and S2). Several pro-inflammatory cytokines, including IFNγ, IL-1β, IL-12, IL-13, and IL-17, were significantly upregulated in the DSS pre-nodal lymph compared to control (Figure 3F). DAMPs, including heat shock proteins, calreticulin, high-mobility group box 1 protein, and protein S100, were also strongly represented in the DSS lymph compared to control (Figure 3G, Tables S1 and S2), and as expected, a signature of organ injury, apoptosis and intestinal damage were detected (Figure 3H, Tables S1 and S2). Additionally, proteolytic enzymes such as trypsinogen, chymotrypsin, and kallikrein, among others, which are normally restricted to the intestine and absent in the control lymph, were also observed in the pre-nodal DSS lymph, confirming the compromised nature of the DSS-treated gut epithelial barrier (Figure 3I, Tables S1 and S2).

Figure 3. Differential proteomic profiles between afferent mesenteric lymph harvested from control and mice with DSS-induced colitis.

(A) Heatmap of representative biological quadruplicates (n = 4) of mouse mesenteric lymph contrasting all fold changes between the 726 nodal proteins identified in the afferent lymph (FDR < 3%) from the Ctr and DSS colitis mice and quantified by label-free (LFQ) proteomic analysis of first level of mass spectrometry (MS1) area (Table S2). Only proteins that passed a selected statistical significance threshold are displayed in the heatmap. The hierarchical clustering was generated using a neighbor-joining algorithm and Euclidean distance similarity measurement of the log₂ ratios of the abundance of each sample relative to the average abundance, built into PEAKS X + Q module (ANOVA/t test applied in PEAKS XPro, p < 0.05). In the heatmap, the positive values reflect fold increases (red color), and negative values reflect fold decreases (blue color). The clustering is shown in Table S2H.

(B) Principal component analysis (PCA) generated by Scaffold Quant LFQ, based on exclusive spectra counts of four biological replicates of afferent lymph from control mice and mice with DSS colitis, highlighting distinct proteomics signatures.

(C) IPA-enabled enrichment analysis of afferent mesenteric lymph from mice with DSS colitis depicting the increased number of proteins overrepresenting the inflammation pathways, GI tract injuries, redox stress, and immunological responses to metabolic, hepatic, and nutritional distress induced by the gut inflammation.

(D and E) IPA-predicted colitis- and enteritis-associated pathways in the afferent mesenteric lymph from DSS vs. Ctr mice. In all IPA predictions, the quantitative analysis of the log₂ fold change, corresponding to the proteomic changes in the ratio (efferent DSS/efferent Ctr) are displayed as colored-coded networks. IPA-calculated activation Z score, in orange, predicts z > 2.0 pathway activation; red and green color molecules, in each pathway, are experimentally determined to be up- and downregulated, respectively.

(F) Enzyme linked immunosorbent assay (ELISA) quantitation of IFN-g and pro-inflammatory cytokines in the afferent lymph from DSS vs. control mice (two-way ANOVA and multiple t test comparison analysis. Statistical significance was determined using the Holm-Sidak method, with alpha = 0.05: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001).

(G) IPA analysis indicates increased damage-associated molecular patterns (DAMPs) in the afferent mesenteric lymph from DSS vs. control mice.

(H) IPA-generated heatmap using the (afferent DSS/afferent Ctr) protein ratios (Table S1) and colored by Z score (orange = activation and Z score >2.0; blue = inactivation and Z score ≤2.0).

(I) IPA analysis indicates increased presence of proteolytic enzymes such as trypsin, chymotrypsin, carboxypeptidases, elastases, and cathepsin B (CTSB) in the proteome of the afferent lymph harvested from DSS vs. control mice. The quantitative analysis of the log₂ fold change corresponding to the proteomic changes in the ratio (afferent DSS/afferent Ctr) is displayed as networks colored coded by the IPA-calculated activation Z score (orange predicts Z >2.0 activation/increase). The GO annotation and pathway analysis is presented in Table S2.

The inflammatory signature was also transposed in the DSS efferent lymph (Figure S6, Tables S1 and S2), which displayed a molecular signature associated with inflammation and adaptive and innate immune responses including immunoglobulin production, complement activation, cytokine signaling, immune cell trafficking, and phagocytosis alongside increased accumulation and activation of neutrophils, as well as elevated release of prostaglandins (Figure S6, Tables S1 and S2).

These findings collectively highlight the function of draining lymph in gathering the by-products of tissue damage and transporting them to the lymph nodes.

DSS-induced inflammation is reflected in the metabolomic composition of mouse mesenteric lymph

We further investigated whether we could detect differences in gut-derived metabolites in the pre-nodal lymph samples from DSS vs. control mesenteric lymph using both targeted and untargeted metabolomic analyses (Figure 4; Table S3). Consistent with previous findings²⁹ in plasma of patients with acute inflammatory diseases and dysbiosis, we noted reductions in amino acids and fatty acids of short, medium, and long chain lengths (Figures 4A and 4B). Conversely, an increase in some inflammatory and metabolic by-products of amino acid catabolism, such as putrescine, citrulline, spermidine, and oxoproline, was observed (Figure 4C). Also, secondary to the inflammation/dysbiosis, a decrease in secondary bile acid (i.e., chenodeoxycholic acid, taurocholic acid, and tauroursodeoxycholic acid) could be detected (Figure 4D). A partial least squares discriminant analysis in MetaboAnalyst 5.0 to the mapped metabolites showed distinct clustering of the DSS vs. control signatures (Figures 4E and S7; Table S3), supporting the notion that the two biological fluids have distinct metabolic signatures due to the inflammatory intestinal damage. Overall, we identified 25 metabolites showing at least 2-fold increase and 18 metabolites with at least 2-fold decrease in the DSS-fed group compared to the control group (Figures 4F and S7; Table S3). Among the metabolites showing a significant increase (p < 0.05 by t test) in the DSS compared to control lymph, we identified unsaturated derivatives of butyric acid, 2-deoxy-D-galactose, products from the Trp-pathway, such as indole and 3-OH-kynurenamine, and intermediates from the tricarboxylic acid metabolism, such as isocitric acid.³⁰ Furthermore, we also observed an increase in the DSS lymph samples of several sugar isomers crucial for bacteria cell wall synthesis, including L-rhamnulose and D-arabitol, as well as bacterial lipids such as mannosyl-1beta-phosphomycoketide C32. These substances, which are known to be presented by CD1c, play a role in activating T cell responses (Figures 4F and S7; Table S3). DSS-induced inflammation also increased the concentration of oxidized amino acids such as O-tyrosine, which has been associated with cellular oxidative damage,³¹ and docosahexaenoyl-serotonin, generated by serotonin conjugation to omega 3 polyunsaturated fatty acid, known to be immunosuppressive.^32,33 Inosine, an extracellular degradation product of purines, was also upregulated in the DSS afferent lymph (Table S3).

Figure 4. Metabolic changes in the afferent lymph harvested from mice with DSS-induced colitis.

(A–D) The relative abundance of (A) amino acids, (B) short, medium, and long chain fatty acids, (C) metabolic by-products of amino acid catabolism, and (D) bile acids, as mapped in control and DSS mesenteric afferent lymph. The median values of MS1 intensities from n = 5 control and n = 6 DSS replicates are shown.

(E) Partial least squares discriminant analysis (PLS-DA) in METABOANALYST 5.0 of identified metabolites, listed in Table S3, shows different clusters for the mesenteric control and DSS afferent lymph.

(F) Untargeted metabolomic analysis identifies differentially expressed metabolites in control versus DSS afferent mouse mesenteric lymph (red: >1.5× upregulated, blue: <1.5× downregulated). Identified metabolites include molecules derived from bacterial metabolism of dietary substrates, modification of host biomolecules, such as bile acids (i.e., taurine), or de novo synthesized molecules by the gut microbiome. Metabolites IDs, quantitation, and associated metabolic pathways are presented in Table S3. The statistical significance was calculated in MetaboAnalyst 5.0 using the built-in t test/ANOVA and displayed as (*) (for p ≤ 0.05).

Finally, metabolites showing a significant decrease (p < 0.05 by t test) in the DSS compared to control afferent lymph included derivatives of the spermidine pathway, such as N1-acetylspermidine, alcohol derivatives of histidine, such as L-histidinol, and derivatives of imidazole such as 1,2-dimethylimidazole (Figure 4F; Table S3).

Our findings collectively underscore the importance of pre-nodal lymphatic fluid in transporting metabolic by-products, including bacterial metabolites, to the draining mesenteric node under inflammatory conditions, highlighting its significance in immunosurveillance.

DSS-induced inflammation increases microbiome-derived proteins in the afferent lymph

To determine whether the inflammatory disruption of the gut barrier would result in the presence of the microbiome in the lymph, we search the mesenteric proteome for bacterial proteins translated from the mouse gut microbiota 16S rRNA.³⁴ Several lymph-carried proteins mapped to phyla known to be components of the mouse gut microbiota such as Saccharibacteria, Cyanobacteriota, and Mycoplasmatota, among others (Table S4O and Figure 5A). Importantly, a statistically significant increase of microbiome-derived proteins was observed in the DSS lymph compared to control, following disruption of the gut barrier (Figures 5B and 5C). Comparative analysis of the microbiome proteins found in the DSS lymph assigned over 60% of the bacterial proteins to DNA metabolism, translational and ribosomal machinery, and ATP synthesis in the latter (Figure 5D; Table S4O). Finally, a specific increase in the number and relative abundance of proteins derived from pathogenic bacteria like Klebsiella pneumoniae was found in the DSS mesenteric lymph but not in the control (Figure 5E, Table S4O).

Figure 5. Increased microbiome-derived proteins in the afferent mesenteric lymph harvested from mice with DSS-induced colitis.

(A) Pie chart representing the frequency of microbiota phyla identified from 262 proteins mapped to the afferent mesenteric lymph harvested from DSS mice from n = 5 control and n = 6 DSS replicates is shown. The identified phyla are all known components of the mouse gut microbiota. The proteomics analysis was performed using a fused database containing FASTA sequences of proteins translated from the bacterial 16S rRNA sequenced genomes identified in the mouse gut microbiota and the UniProt Mus musculus FASTA database (17,155 entries, March 2024).

(B) MS1-based LFQ analysis showing a significant increase in the relative abundance of gut bacteria-derived proteins in the lymph harvested from DSS mice compared to control (p < 0.01 by unpaired t test).

(C) Bar graph representing the LFQ quantitation of the iron-regulated protein A (from Synechococcus elongatus, Cyanobacteriota phylum) displaying a significant increase in the afferent lymph collected from DSS vs. control mice (p < 0.01 by unpaired t test).

(D) GO annotations and functional analysis of the bacterial proteins found in the afferent mesenteric lymph of DSS mice using Pantherdb databases.

(E) MS1 LFQ analysis showing specific increase in the number and relative abundance of proteins derived from Klebsiella pneumoniae in the afferent mesenteric lymph harvested from DSS mice vs. Ctr (Table S4O).

In summary, our analysis of the gut microbiota proteomes carried by the afferent mesenteric lymph indicates the relevance of the lymphatic fluid for immunosurveillance, particularly during conditions that disrupt the gut barrier.

Composition of IBD patient pre-nodal lymph

To assess whether the proteomic inflammatory signature of the DSS lymph could be recapitulated in human IBD, pre-nodal mesenteric lymph was harvested from 5 individuals diagnosed with IBD (Figure 6A; Table S5G), who underwent intestinal ileocolic resection procedures. The harvested lymphatic fluid exhibited a prominent inflammatory signature and molecular pathways associated with tissue damage, as previously seen in the DSS mice model. (Figures 6B and 6C; Table S5). Activation of several pathways, including the complement and coagulation cascades, interleukin signaling, inflammation, gut tissue injury, and oxidative responses, and signatures of innate and adaptive immune responses were also mapped in the pre-nodal lymph (Figures 6B and 6C; Table S5). The proteome of the IBD lymph was compared with the proteome of human mesenteric lymph of subjects undergoing different surgeries due to splenic trauma.^35,36 Remarkably, only 24% of the total proteomes overlapped, while 41% of proteins were uniquely mapped to IBD and 35% to the “trauma” lymph (Figure 6D; Table S5F). Gene Ontology (GO) annotations and pathways analysis of the two datasets further indicated that the shared proteins mapped to intestine, liver, and adipocytes proteomes (Figure 6D). The lymph collected from abdominal trauma subjects reflected upregulation of extrinsic and intrinsic coagulation pathways (Figure 6E; Table S5F), whereas the unique IBD proteome indicated an inflammatory response with tissue damage and innate and adaptive immune responses (Figures 6F–6H; Table S5F). These findings support the view that the lymph proteome/peptidome signature strongly reflects the physiological state of the organs and tissues of origin.

Figure 6. Inflammatory signatures in human mesenteric lymph harvested from subjects with IBD.

(A) Cannulations of pre-nodal mesenteric afferent lymphatic vessels from a human specimen obtained from surgery of IBD subjects following ileocolic resection procedure (n = 5).

(B) IPA analysis of the top-scoring biochemical and pathophysiological pathways, derived from the 799 protein IDs identified in pre-nodal mesenteric human lymph with a statistically significant identification score (p < 0.05 by Fisher’s exact test with Benjamini-Hochberg correction). IPA identified significant cellular and molecular functions associated with inflammation, cell death, immunological response, and metabolic changes overrepresented in the IBD subjects.

(C) IPA analysis depicting the activation networks found in the IBD proteome.

(D) Venn diagram contrasting the qualitative differences between 799 proteins identified in the IBD lymph (this study) and 733 proteins retrieved from the studies of human mesenteric lymph from patients undergoing abdominal surgery following traumatic injuries. GO annotation and pathway analysis of the two-proteomics dataset confirms that both proteomes are generated from intestine, liver, and adipose tissue.

(E) IPA pathway enrichment analysis of the unique proteome characterizing the lymph harvested from an individual with traumatic abdominal injuries.

(F) IPA pathway enrichment analysis of the unique proteome characterizing the IBD lymph.

(G and H) Top protein networks describing the immunologically driven inflammatory response in the mesenteric lymph of IBD subjects (Table S5F).

Another notable observation in the lymphatic fluid collected from individuals with IBD was the detection of a unique peptidome/degradome, reflecting the ongoing tissue damage (Figure 7A). The identified peptides were traced back to the sub-cellular proteome present in the gut, including proteins from intestinal stem, epithelial, and basal cells as well as muscle and endothelial cells. Additionally, proteolytic by-products were predicted to be derived from pancreatic acinar cells. The peptides mapped to a wide range of intracellular and extracellular pathways, encompassing cellular stress, apoptosis, and extracellular matrix degradation, among others (Figures 7A and S8A; Table S6).

Figure 7. Gut inflammation in IBD subjects increases the microbiome-derived degradome in the afferent pre-nodal lymph.

(A) METSCAPE cellular and molecular pathway enrichment analysis highlights the intestine and pancreatic tissues as the primary anatomical regions of the lymph peptidome/degradome mapped in the IBD subjects.

(B) The heterogeneity of microbial proteins from acid elutions of human IBD mesenteric afferent lymph (n = 1).

(C) Length distribution of the acid-eluted peptides of the human samples showing higher fraction of smaller length peptides with a mean distribution of 8 aa.

(D–F) The NetMHCpan 4.1 predictions for frequently observed HLA alleles (D) HLA-A, (E) HLA-B, and (F) HLA-C for the acid-eluted microbial peptides present in the IBD lymph. The binders below rank percentile of 2 are indicated.

(G) Spatial distribution of the microbes identified in trypsin-digested proteome across different human IBD lymph samples (n = 6).

Additionally, by referencing the lymph peptidome using an 16sRNA gut microbiome database, generated on the microbial composition of IBD subjects,^37,38 we mapped several lymph peptides to the IBD gut microbiome including 46.1% of the firmicutes, 32.7% of actinobacteria, 7.7% bacteroidetes, 7.7% proteobacteria, and 5.8% fusobacteria (Figure 7B; Table S6). As the peptides were majorly of shorter length (Figure 7C), we used the NetMHCpan 4.2 algorithm for predicting the putative human leukocyte antigen (HLA)-A, B, and C binders (Figures 7D–7F). The prediction was done against the allele frequently expressed in the population.³⁹ With the cutoff rank percentile of 2%, we identified one peptide “ESSAPPDAK” derived from an uncharacterized protein of Collinsella sp. as a potential HLA-A binder (Figure 7D). The other two peptides, “LAKKVVIVPI” derived from uracil phosphoribosyl transferase of Eubacterium bioforme and “AEEDGGGKIIML” derived from a hypothetical protein of Erysipelotrichaceae, were both predicted to bind HLA-B proteins (Figure 7E). Similarly, we identified the spatial gut-derived microbiome in protease-digested human lymph samples (Figure 7G) comprising 56.6% of firmicutes, 25.4% of actinobacteria, 8.6% bacteroidetes, 5% proteobacteria, and 4.4% fusobacteria (Figure 7G; Table S6).

Altogether the data indicate that in both subjects with Crohn’s disease and mice post DSS treatment, microbiome-derived peptides are present in the afferent lymph consistent with disruption of the gut barrier following the inflammatory process.

Finally, we also isolated extracellular vesicles (EVs) from the Crohn’s disease pre-nodal lymph. EVs are established active participants in inflammatory processes, found within an inflamed region as “ambassadors” for their tissue of origin.⁴⁰ To map the comparable molecular profile observed in the DSS-harvested lymph, EVs isolated from the pre-nodal lymph of Crohn’s disease patients were subjected to proteomics analysis. We first examined the presence of common EV markers by mass spectrometry and then performed a qualitative enrichment analysis (Figures S9A–S9E; Table S7). The EV proteome displayed a molecular signature associated with pro/inflammatory pathways such as eukaryotic initiation factor 2 signaling, integrin signaling, neutrophil activation, actin cytoskeleton signaling, acute phase response, and epithelial damage, reflecting EV shedding from the inflamed mucosal sites and their potential contributions to inflammation and disease pathogenesis. Moreover, the functional analysis mapped the Crohn’s pre-nodal lymph EV proteome to pathways related to inflammation, intestinal disorders, metabolic changes, epithelial damage, and immune regulation (Figures S9B–S9E), thus highlighting the role of EVs in immunomodulation and inflammation. Overall, the EV proteome was similar to that of the lymph fluid, reflecting the inflammatory and tissue-damaging processes occurring in the intestines of IBD patients.

DISCUSSION

The interstitial fluid (IF) is a complex biological fluid that comprises proteins, lipids, and carbohydrate nutrients, generated from blood ultrafiltration and by trans-endothelial transport, as well as waste solute by-products deriving from cellular metabolism.^{13,14,19,20,41–44} The IF becomes “lymph” once it enters the lymphatic capillaries, and this initial transport is dependent on the cyclic transmural pressure between the IF and the surrounding initial lymphatics.^25,45 Herein, we characterized the proteomic composition of the lymphatic fluid, harvested from the mesenteric and cervical afferent lymphatics, and we mapped several proteins uniquely represented in the two anatomical districts, confirming that the lymph composition reflects the anatomical district from which it drains. Indeed, the mesenteric lymph was enriched in apolipoproteins, phospholipid transfer proteins, fatty acid- and vitamin-binding proteins, and proteins involved in pancreas beta-cell function and glucose homeostasis, all consistent with the known role of the mesenteric lymph in chylomicron transport.²⁵ On the other hand, a brain-specific rather highly enriched proteome including glia maturation factor, nerve growth factor, alpha-crystallin, brain-specific isoform of glycogen phosphorylase, and drebrin was uniquely observed in the lymph harvested from the afferent lymphatics entering the deep cervical nodes. Importantly, we also determined that the lymph-carried proteome contributes to the MHC-II immunopeptidome eluted from mesenteric and cervical dendritic cells, pointing to a pivotal role of the lymphatic fluid in immunosurveillance.^14,17,46

The lymph role in immunosurveillance was further highlighted upon organ inflammation. In the DSS mouse model, where the integrity of the gut barrier is compromised,^47–50 we mapped changes in the lymph antigenic load. At steady state, as expected, chylomicron basic components such as triglycerides, fatty acids of all lengths, and proteins involved in lipid transport constituted the bulk of the “omics” present in the afferent lymph.^{25,47,51–56} However, following DSS-mediated inflammation in mice, the most notable proteomic differences were represented by an increase in DAMP proteins, generated from the damaged epithelia as well as the molecular signatures of colitis/enteritis, organ injury, and gastrointestinal inflammation. Importantly, enzymes normally confined to the gut, such as chymotrypsin, peptidases, phospholipases, and elastases were all detected in the afferent lymph, consistent with the lipolysis following gut barrier breakdown. The DAMPs and inflammatory signatures were also recapitulated in the “omics” analysis of pre-nodal lymphatic fluid collected from patients with IBD. In these subjects, loss of gut barrier integrity resulted in ingress of bacterial by-products, encompassing bacteria-derived peptides and metabolites, into the lymphatic flow. Indeed, proteomic analysis of the afferent lymph from mice with DSS-induced colitis or from IBD subjects revealed the presence of peptides mapped to known components of the gut microbiota, as well as pathogenic microbes such as Klebsiella.

Notably, a considerable portion of these peptides exhibited putative binding affinity for various MHC-I haplotypes. Microbiome-derived peptides as well as peptides derived from inflammation-driven proteins were also eluted from MHC-II molecules harvested from mesenteric nodal DCs in DSS mice,^57–60 indicating the crosstalk between the damaged tissue and the presented MHC immunopeptidome, further underling the relevance of tissue specificity in antigen presentation.

At the biochemical level, we observed decreases in amino acid concentrations, such as glutamine, leucine, isoleucine, and valine, which are known to influence T cell activation and lineage fate, leading to the development of Th1, Th2, Th17, and regulatory T cells as well as antibody generation,²⁹ cysteine, glutamine, and glycine, which control the redox metabolism,^61,62 and arginine, pivotal for the polyamine pathway, which controls macrophage polarization and generates spermidine, which drives translational elongation through factor 5a hypusination.^62,63 These data overlap with previous observations in the blood during inflammatory conditions,⁶⁴ where amino acids, used for inflammatory protein synthesis, purine, pyrimidine, and nucleotide synthesis decrease, and an inverse correlation between inflammatory markers and amino acid plasma concentration is observed.⁶⁵ The decrease in amino acid levels following DSS-mediated damage was associated with an enrichment in proteomic pathways associated with amino acid degradation, as well as an increase in amino acid intermediate catabolites/metabolites such as putrescine, spermidine, citrulline, and 5-oxoproline, among others.

The majority of bile acids are normally reabsorbed into the enterohepatic circulation.²⁷ However, as observed in our study, a small component of mostly primary bile acids and lithocholic acid as a secondary bile acid is found in the lymph. Bile acid receptors TGR5, FXR, S1PR2, VDR, and CAR are ubiquitously present on immune cells, including monocytes, T cells, B cells, DCs, natural killer cells, and granulocytes, and bile acids have been shown to modulate immune functions.^66–68 Herein we found that DSS-mediated damage mostly compromises the conjugation and generation of secondary bile acids, such as taurocholic, taurolithocholic, and tauroursodeoxycholic acid and increases free taurine. Taurine was previously shown to counteract gut permeability by increasing tight junction proteins in epithelial cells.⁶⁹ On the other hand, tauroursodeoxycholic acid has been shown to downregulate the expression of the MHC antigen processing and presentation machinery,⁷⁰ while deoxycholic acid downregulates DC activation in different inflammatory conditions, as well as Th1/Th17 differentiation and Treg induction.^71,72 These changes in the ratio between primary and secondary bile acids were previously associated with dysbiosis.^73–75

Finally, several metabolic by-products from bacterial and gut-associated metabolism were present in the afferent lymph. L-3-cyano-alanine, a non-proteinogenic cyanoamino acid produced by bacteria to detoxify cyanide from endogenous and exogenous sources, was also found in the DSS lymph.⁷⁶ Several components of the bacterial wall were also found in the afferent lymph following DSS-mediated damage including mucic acid,⁷⁷ rhamnose,⁷⁸ and β-d-mannosyl phosphomycoketide, a glycolipid known to activate immune cells through CD1d binding. Also elevated was methylglutaric acid, which induces lipid peroxidation and mitochondrial disfunction⁷⁹

Short chain fatty acids (SCFAs) (2–6 carbons in length), among which propionate, butyrate, and acetate are the most abundant in mammals, are generated from dietary fibers that cannot be processed by digestive enzymes but only through microbiota-driven fermentation.⁸⁰ Receptors for SCFAs are widely distributed with a high degree of expression in the adipose tissue where they control lipolysis and energy homeostasis and in immune cells where they are generally associated with a decrease in inflammatory responses.^81,82,83,84 As a sign of dysbiosis, SCFAs were all decreased following DSS-mediated epithelial damage furthering the inflammation.

Inflammation also increased the concentration of oxidized amino acids such as O-tyrosine, which derives from the oxidation of the benzyl ring of phenylalanine. O-tyrosine has been shown to induce cellular damage and contribute to degenerative diseases such as Alzheimer’s disease and atherosclerosis.³¹ On the other hand, docosahexaenoyl-serotonin, an endogenously generated omega 3 polyunsaturated fatty acid-serotonin conjugate, is a powerful inhibitor of IL-17 and IL-23 pro-inflammatory responses.^32,33 Similarly, inosine, the extracellular degradation product of purines, which we identified as highly upregulated in the DSS afferent lymph, has strong anti-inflammatory effects, reducing innate immune responses and production of pro-inflammatory cytokines such as TNF-a, IL-1, IL-12, and IFNg.⁸⁵

Overall, our study underscores the role of lymphatic fluid in the “omics” transport to the lymph node as a crosstalk between the drained tissue and the immune system. The mapped changes indicate how the lymphatic fluid can provide a snapshot of organ physiology/pathology that ultimately shapes nodal immune responses.

Limitations of the study

A limitation of the analysis of the human lymph is that, differently from the blood, it is impossible to obtain samples from healthy donors, particularly from the cervical or mesenteric district. We circumvent the problem by harvesting lymph from subjects who underwent surgery for mechanical accidents and required abdominal surgery. A second limitation of the study is the lack of comprehensive human and mouse microbiome protein libraries to be used to analyze the bacteria peptides. We overcome this limitation by using FASTA files of translated 16S RNA-seq libraries corresponding to gut microbiota identified in healthy or DSS mice models or patients with IBD and performed the peptidomic and proteomics analysis with the PEAKS’s built-in de novo sequencing and database search algorithms. Nevertheless, the complexity and diversity of the human and mouse proteomes are not fully captured in the current available libraries, and as such, we may have missed some of the microbiome peptides present in the human lymph. At last, the paucity of the number of DCs retrieved from the cervical and mesenteric draining nodes makes it complex to perform MHC immunoaffinity purification and elution, and the number of eluted peptides is in the low thousands due to the limited number of cells.

A technical limitation of the study is the difficulty in performing lymphatic cannulation. In mice, the lymphatic afferent vessels are around 20 μm in diameter, and the pipet used for cannulation is 10 μm in diameter, as such it requires a trained and skilled operator to perform the cannulation.

STAR★METHODS

RESOURCE AVAILABILITY

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Laura Santambrogio (las4011@med.cornell.edu).

Materials availability

No new material was generated in this study.

Data and code availability

• The mass spectrometry proteomics data for the human lymph from IBD patients are deposited to the Proteome Xchange Consortium via the PRIDE partner repository with the dataset identifier PXD044801 and 10.6019/PXD044801. Reviewer account details: Username: reviewer_pxd044801@ebi.ac.uk, Password: T6SmFvo; and the human lymph peptidome dataset are deposited with the PRIDE identifier PXD051024 with DOI: 10.6019/PXD051024; the mesenteric lymph from control (healthy) and DSS-colitis mice are deposited with the identifier PXD044885 and 10.6019/PXD044885 and reviewer account: Username: reviewer_pxd044885@ebi.ac.uk and Password: UysnJKyP. The proteomics data corresponding to the cervical and mesenteric mouse lymph analyzed with Q exactive HF mass spectrometry are deposited to MassIVE server with the project IDs MSV000094586 and PXD051618. The I-A^b immunopeptidome datasets from cervical and mesenteric lymph nodes are deposited to MassIVE server with the project IDs MSV000094761 and PXD052267. The lymph metabolomics data are deposited into Metabolomics workbench and assigned the temporary DataTrack ID number 4234.

• This paper does not report original code.

• Any additional information required to reanalyze the data reported in this work paper is available from the lead contact upon request.

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

Mice

Mice (C57BL/6J) mice were obtained from Jackson Laboratories and housed at Weill Cornell Medicine animal facilities protocol number (2019–0024)

In all experiments both male and females (50% each) were used between 8 and 12 weeks old.

Human

Human Intestinal samples (Ileocolic resected) were obtained from informed patients in accordance with the Institutional Review Board (IRB) protocols for Weill Cornell Medical College (number 1501015812) and Cornell University (number 1609006586).

All subject’s information is included in Table S5.

METHOD DETAILS

Cannulation and lymph collection

The lymph was collected from a human specimen within 1 h of surgery. The human specimen was kept hydrated with 1X PBS (pH 7.4). Following identification of the pre-nodal lymph vessel, under the microscope – Nikon SMZ1270, fat was carefully separated from the region of interest. Care was taken to prevent excess mechanical stress to the vessel during the fat removal stages. Using glass pipettes (tip diameter − 40–60 μM), a puncture was made on to the vessel, allowing the lymph to flow and collect into the pipette. The pipette was then flushed into a microcentrifuge containing 1X PBS and 2X protease inhibitor (Thermofisher Scientific (A32963). The collected lymph was then spun at 1,000 × g at RT for 10 min to pellet the cellular component. The supernatant was then spun at 3,000 × g for 20 min to separate the tissue debris. The final cell pellet was stored in liquid nitrogen using freeze media, while the supernatant was stored at −80C for further analysis.

Mice (C57BL/6J) mice were obtained from Jackson Laboratories and housed at Weill Cornell Medicine animal facilities. Mice were anesthetized using ketamine (80 mg/kg) and Xylazine (10 mg/kg), injected intraperitoneally. Cannulation was performed 15–20 min post anesthesia, when the mice was no longer able to feel a toe pinch. Mice were placed face up with paws stretched and pinned. Incision was made at the abdominal area, exposing the underlying gut and intestinal tissues. Pre-nodal/post-nodal lymphatic vessels were visualized and identified. Fat tissues was carefully removed or move aside using forceps and prenodal lymphatics were cannulated halfway between the intestine and the mesenteric lymph node. Post-nodal mesenteric lymphatics were cannulated at the nodal exit. Pre-nodal/post-nodal lymphatic vessels were visualized and identified. Fat tissues was carefully removed or move aside using forceps. For collection of pre-nodal cervical lymph, an incision, superior to the clavicle was made and the sternocleidomastoid muscle was retracted to visualize the afferent lymphatic vessels draining to the deep cervical lymph nodes. In both collections, care was taken to keep the tissues hydrated using 1X PBS. Upon creating a small tear/cut on the lymphatic vessel, using a glass pipette (tip diameter of 40–60 μM) the lymph was collected. The pipette was then flushed into a microcentrifuge containing 1X PBS and 2X protease inhibitor (Thermofisher Scientific (A32963).

Isolation of I-A^b peptide complexes from dendritic cells harvested from cervical and mesenteric lymph nodes and peptide elution

Mice were injected subcutaneously with 4×10^6 B16-FLT3-L producing melanoma cells as previously described.⁵⁷ Twelve days after injection cervical and mesenteric lymph nodes were harvested and dendritic cells (DC) purified using a 37% BSA gradient (Sigma-Aldrich)⁸⁶ to enrich DC to a 70–80% purity. DC were further enriched using CD11c-antibody-conjugated magnetic beads (Miltenyi Biotech) according to the manufacturer protocol. DC pellets were resuspended in 50 mM Tris-HCl, 150 mM NaCl, pH 8.0, containing protease inhibitors and 5% β-octylglucoside, freeze-thawed for 5–6 times, homogenized, and the solubilized whole cell fraction was recovered by centrifugation at 100,000 × g for 1 h at 4°C. The supernatant was used for the isolation of the MHC-class-II-peptide complexes using an immunoaffinity column of M5/114 monoclonal antibody immobilized onto CNBr activated Sepharose. The column was equilibrated with buffer (50 mM Tris-HCl, 150 mM NaCl pH 8.0, containing protease inhibitors) for 2 h. The lysates were pre-cleared with the Sepharose beads only followed by isotype control antibody conjugated Sepharose beads for 1 h at 4°C. After pre-clearing the lysate was incubated with M5/114 conjugated beads and allowed to mix for 1 h at 4°C. The column was washed with several buffers in succession as follows: (1) 50 mM Tris-HCl, 150 mM NaCl, pH 8.0, containing protease inhibitors and 5% β-octyl-glucoside (5 times the bead volume); (2) 50 mM Tris-HCl, 150 mM NaCl, pH 8.0, containing protease inhibitors and 1% β-octylglucoside (10 times the bead volume); (3) 50 mM Tris-HCl, 150 mM NaCl, pH 8.0, containing protease inhibitors (30 times the bead volume); (4) 50 mM Tris-HCl, 300 mM NaCl, pH 8.0, containing protease inhibitors (10 times the bead volume); (5) 1X PBS (30 times the bead volume); and (6) HPLC water (100 times the bead volume). The I-A^b peptide complexes were eluted from beads only or isotype conjugated beads or M5/114 column using 2% TFA solution. Peptides were further separated using a Vydac C4 macrospin column (The Nest Group, USA) and lyophilized using a Speed-Vac.

Prediction of bacteria peptides binding affinity to the MHC-II alleles

NetMHCIIpan-4.0 algorithms⁸⁷ were used to predict MHC-II binding for all identified bacterial peptides, encompassing all available alleles. The selection of predicted MHC-II binders was performed based on their relative ranking within the output of NetMHCIIpan-4.0. Only peptides that were situated within the top 30% of the ranked list were considered as potential binders.

Predicted bacterial taxonomic classification

Bacterial taxonomic classification was conducted employing the National Institutes of Health (NIH) National Library of Medicine’s resource, specifically the protein Basic Local Alignment Search Tool (BLAST), accessible at https://blast.ncbi.nlm.nih.gov/Blast.cgi and the taxonomy database available at https://www.ncbi.nlm.nih.gov/taxonomy. This process was primarily executed using the protein BLAST feature (blastp) in conjunction with the standard databases.

The phylum designation was determined by considering two key parameters from the blastp output: the Expected value (E-value) and the percentage of query cover. The phylum assigned to each bacterial peptide was the one associated with the best score i.e., the lowest E-value and the highest percentage coverage for the query, ensuring the most accurate taxonomic identification.

Extracellular vesicle isolation from human mesenteric lymph (IBD patients)

Extracellular vesicles (EVs) were isolated as previously described.⁸⁸ Briefly, the pre-nodal lymph samples were collected from five IBD subjects and the volume was raised to 5mL using particle-free PBS. The samples were spun at 12,000 × g for 20 min at 4°C to remove any apoptotic bodies and cellular debris. The collected supernatant was subjected to two rounds of ultracentrifugation in 5 mL ultracentrifugation tubes (#326819, Beckman Coulter) at 100,000 × g for 70min. In the last cycle EVs were washed in PBS and pelleted by 100,000 × g ultracentrifugation in SW 50.1 swinging bucket rotor in a Beckman Coulter Optima XPN ultracentrifuge. The final pellet was carefully resuspended in 100uls of cold PBS and stored at −80°C for proteomic analysis.

Human mesenteric lymph EV proteome

For MS sample preparation, our optimized protocol was followed^89,90 with the exception that acetone precipitation, which was used given the small volume/amount of sample. Seven times the sample volume of ice-cold acetone was added, and the samples were precipitated overnight. Samples were then centrifuged at 10000g for 10 min at 4°. The supernatant was removed, and the samples were dried for 30 min.

Data-dependent acquisition (DDA) nanoLC-MS/MS analysis of EV samples

Enriched EV samples were re-dissolved in 30–50uL 8M Urea/50mM ammonium bicarbonate/10mm DTT. Following lysis, reduction and alkylation steps using 20 or 30mM iodoacetamide (Sigma), proteins were digested with Endopeptidase Lys C (Wako) in < 4M urea followed by trypsination (Promega) in < 2M Urea. Peptides were desalted and concentrated using Empore C18-based solid phase extraction prior to analysis by high resolution/high mass accuracy reversed phase (C18) nano-LC-MS/MS. nanoLC/MS/MS analysis: samples were run on the Bruker TIMS/TOF, using two column separation: waters nanoeasy symmetry trap column, and 90-min water/acetonitrile gradient on a 25 cm ionoptics separation column where 200 ng protein was injected.

Database search and validation of proteins identification in human EV samples

High resolution/high mass accuracy nano-LC-MS/MS data was processed using Proteome Discoverer 1.4.1.14 (Thermo-Scientific, 2012)/Mascot 2.5 (Perkins et al., 1999). Human data was queried against the UniProt’s Complete HUMAN proteome using the following parameters: Enzyme: Trypsin/P, maximum allowed missed cleavage sites: 2, monoisotopic precursor mass tolerance: 10 ppm, monoisotopic fragment mass tolerance: 0.02 Da, dynamic modifications: Oxidation (M), Acetyl (Protein N-term), static modification: Carbamidomethyl (I). Percolator was used to calculate peptide False Discovery Rates (FDR), which was calculated per file while one unique peptide criteria was implemented to call a protein. A false discovery rate (FDR) of 1% was applied to each separate LC-MS/MS file.

Gene ontology, pathway enrichment of human lymph and EV LFQ proteomics

Networks, functional analyses, and biochemical and cellular pathways were generated by employing the ingenuity pathway analysis (IPA; Ingenuity Systems, Redwood City, CA, USA) on the list of identified proteins in the DDA analysis (Table S7). For networks generation, datasets containing gene identifiers (gene symbols) were uploaded into the IPA application where these molecules were overlaid onto a global molecular network contained in the Ingenuity Knowledge Base. The networks of network-qualified molecules were then algorithmically generated based on their connectivity index using the built-in the IPAIPA algorithm. An independent GO assessment and annotation was performed using the ShinyGO 0.76 (ShinyGO 0.77 (sdstate.edu)) to map the identified proteins in the EV samples isolated from the human IBD subjects.

Proteomics analysis of mouse lymph harvested from cervical and mesenteric afferent lymphatics

Paired samples from afferent and efferent mesenteric lymph, collected by canulation from control and DSS-induced colitis mice (n =4 biological replicates for each sample set) were subjected to label-free quantitative (LFQ) proteomics analysis using optimized protocols.^14,24

Total protein concentration for pre- and post-nodal lymph samples was determined using the Bradford micro assay. Equal protein amounts (15 μg) were reduced with 15 mM TCEP.HCl (Thermo Scientific) in 50 mM ammonium bicarbonate buffer, at pH 8.5, for 40 min at room temperature. The reduced proteins were further alkylated with 75 mM iodoacetamide solution, for 50 min at room temperature. Three different enzymes were used for “in solution” digestion in 50 mM ammonium bicarbonate buffer, pH 8.5, for 18 h, at 37 °C: endoproteinase Lys-C (1:40 enzyme: protein ratio); trypsin (1:40 enzyme: protein ratio) and Glu-C (1:20 enzyme: protein ratio). The digestion was quenched with 0.5% acetonitrile and 1.5% formic acid. Processed peptides were then extracted through a 10-kDa MWCO (molecular weight cut-off) using 10kDa centrifugal filter units by spinning at 10,000×g for 15 min in a 20:1 microcentrifuge. The peptides mixture, extracted from all enzymatic digestions, were desalted on C18 Prep clean columns before high resolution liquid chromatography-tandem mass spectrometry (nanoLC-MS/MS).

Extraction of endogenous peptides (peptidome) from human mesenteric lymph

The endogenous processed peptides from the human mesenteric lymph (300 μg total protein by BCA assay) were extracted and processed for nanoLC-MS/MS analysis using optimized protocols^18,59,60 with modifications. Human lymph proteins were resuspended in 0.5 mL of sterile PBS buffer supplemented with a cocktail of protease inhibitors. Peptides were extracted using 0.2% TFA for 5 min, at 20°C. Samples were then filtrated through a 10,000-Da cutoff filter device at 20°C for 30 min, desalted using pepClean C-18 spin columns, and eluted with 70% acetonitrile containing 0.1% TFA for further nanoLC/MS/MS analysis.

Liquid chromatography and tandem mass spectrometry (nano-LC-MS/MS)

The endogenously processed peptides, from human mesenteric lymph, and the tryptic digests of mouse mesenteric lymph samples (10–15 μg each) were analyzed on a Q-Exactive HF quadrupole orbitrap mass spectrometer (Thermo Fisher Scientific, Waltham, MA, USA) coupled to an Easy nLC 1000 UHPLC (Thermo Fisher Scientific) through a nanoelectrospray ion source. The mass spectrometer was operated using an already published protocol describing data–dependent acquisition (DDA) and positive ionization mode.⁹¹

DDA and DIA PASEF analysis on a timsTOF Pro2 mass spectrometer

Peptides (n = 4 biological replicates for each cervical and mesenteric mouse lymph), were separated within 87 min at a flow rate of 400 nL/min using a reversed-phase C18 column with an integrated CaptiveSpray Emitter (25 cm × 75μm, 1.6 μm, IonOpticks). Mobile phases A and B contained 0.1% formic acid in water and 0.1% formic acid in ACN, respectively. The fraction of B was linearly increased from 2% to 23% within 70 min, followed by an increase to 35% within 10 min, and a further increase to 80% before re-equilibration. For generation of sample-based spectral library an aliquot corresponding to 20% of each mesenteric and cervical Lys/Trypsin/GluC-digested lymph sample (n = 4 biological replicates each) was pooled and subjected to mass spectrometry analysis using timsTOF Pro2 operated in DDA PASEF mode with the following settings: Mass Range 100 to 1700m/z, 1/K0 Start 0.6 Vs/cm⁻², End 1.4 Vs/cm⁻², Ramp time 100ms, Lock Duty Cycle to 100%, Capillary Voltage 1600V, Dry Gas 3 L/min, Dry Temp 200°C, PASEF settings: 10 MS/MS Frames (1.17 s duty cycle), charge range 0–5, an active exclusion for 0.4 min, Target intensity 20000, Intensity threshold 2500, CID collision energy 59eV. A polygon filter was applied to the m/z ion mobility plane to select peptide precursors rather than singly charged background ions. For further proteomics profiling, each Lys/Trypsin/GluC-digested lymph sample (n = 4 biological replicates each) was subjected to one single injection and analyzed with timsTOF Pro2 operated in diaPASEF mode, and data were acquired with variable isolation windows ranging from m/z 350 to 1,250. The collision energy was ramped linearly as a function of mobility from 59 eV at 1/K0 = 1.45 Vs. cm⁻² to 20 eV at 1/K0 = 0.6 Vs. cm^−2.92,93

Criteria for protein identification in the cervical and mesenteric mouse lymph

For LFQ evaluation of differential proteomic profiles we used the PEAKS XPRO, PEAKS ONLINE proteomic software. For samples analyzed by Q-Exactive HF quadrupole orbitrap and timsTOF Pro2 mass spectrometers operated in DDA or DIAPASEF mode we used the DIA-NN proteomic software. The raw files were filtered, de novo sequenced and assigned a protein ID using Peaks XPRO or ONLINE software (Bioinformatics Solutions, Waterloo, Canada), by searching against the mouse (Mus musculus) revised SwissProt/UniProt database (17125 entries, August 2023 and 17155 entries, March 2024). The search parameters were applied for LFQ analysis: trypsin, Lys-C and GluC restriction enzymes and two allowed missed cleavages at one or both peptide end. The parent mass tolerance was set to 15 ppm using monoisotopic mass, and fragment ion mass tolerance was set at 0.06 Da. Carbamidomethyl cysteine (+57.0215 on C) was specified in PEAKS as a fixed modification. Methionine, lysine, proline, arginine, cysteine, and asparagine oxidations (+15.99 on CKMNPR), deamidation of asparagine and glutamine (NQ-0.98) and pyro-Glu from glutamine (Q-18.01 N-term) were set as variable modifications. The acquired diaPASEF raw files were searched using the UniProt Mouse proteome (17155 entries, March 2024) in the DIA-NN³ search engine, employing the default settings of the library-free search algorithm. The false discovery rate (FDR) was set to 1% at both the peptide precursor and protein levels. Significantly changed protein abundance was determined by a t test with a threshold for significance set at p < 0.05 (permutation-based FDR correction) and 0.58 log₂FC. Results obtained from DIA-NN were further subjected to statistical and biomarker discovery analyses using MetaboAnalyst, version 5.0. Gene Ontology analysis was performed as part of the subsequent analysis process using the enrichment algorithms for cellular and molecular pathways at the METASCAPE server (https://metascape.org).

Building the spectral libraries from samples analyzed in DDA-PASEF mode on timsTOF Pro2

PEAKS 11 studio and ONLINE versions were employed to build the sample-based spectral library using the acquired DDA timsTOF PASEF raw files. Subsequently the spectral library file was exported and used to perform the spectra-library dependent search of diaPASEF raw files corresponding to each cervical and mesenteric lymph biological replicate against the UniProt Mouse proteome database (17155 entries, March 2024), also used for the DIA-NN search engine, and library-free search algorithm. The following parameters settings were established in the timsTOF module in the PEAKS software: DDA or DIA and CID were set as mode operation on timsTOF depending on the input raw files; precursor and fragment ions tolerances: 20 ppm and 0.05 Da, respectively; CCS tolerance: 0.05; peptide length 7–45. FDR at the level of PSM was set at 3% and proteins score “−10lgP ≥ 15” which determined the final attained FDR for peptides of 3.7%, corresponding to peptides scores “−10lgP ≥ 20.5”; and 3.5% for protein groups. The enzyme restriction set as “specified for each sample” ensured that the LysC/trypsin/GluC were identified as digestion modes; and one or two allowed missed cleavages at one peptide end. Proteins IDs were further assigned using at least one unique identified peptide/protein entry. Individual MS/MS scans were manually inspected for the correct sequence assignment for proteins identified with “one unique” peptide and further subjected to BLAST search against mouse database for confirmation of protein identity using the tools at NCBI.

Analysis of gut microbiota derived proteins in the mouse mesenteric lymph

To further identify putative bacterial proteins/peptides we screened the raw files from mouse lymph samples harvested from control and DSS mice against a fused database containing FASTA sequences of proteins translated from the bacterial 16S rRNA sequenced genomes identified in the mouse gut microbiota and the UniProt Mus musculus FASTA database (17155 entries, March 2024). The LFQ proteomic analysis employed the FDR filtering method built-in the PEAKS 11 software which was chosen to be <3% (at the level of PSM and peptides), and <4% at the protein level, which corresponded to a “−10lgp” PEAKS score>20 for peptides and proteins.

Analysis of endogenous processed peptides (peptidome/degradome) and identification of gut microbiota processed peptides extracted from the human lymph of IBD subjects

The analysis of endogenous processed peptides (peptidome) extracted from the human lymph was performed on the Q-Exactive HF raw files, filtered, de novo sequenced and assigned a protein ID using the Peaks XPRO software (Bioinformatics Solutions, Waterloo, Canada). The database consisted of human (Homo Sapiens) revised SwissProt/UniProt (20,364 entries, August 2022) fused with a database of bacteria species colonizing the human gut. The search parameters applied for LFQ analysis were like those used for analysis of the mouse and human mesenteric lymph proteomics except that “no enzyme restriction” was set in the search to retrieve the endogenously processed peptides.

Analysis of endogenous processed peptides (peptidome/degradome) in the afferent mouse mesenteric lymph

The analysis of endogenous processed peptides (peptidome) extracted from the afferent mouse lymph was performed on the Q-Exactive HF raw files, filtered, de novo sequenced and assigned a protein ID using the Peaks XPRO software (Bioinformatics Solutions, Waterloo, Canada). The searched database consisted of the UniProt Mus musculus FASTA database (17155 entries, March 2024), using the “no enzyme restriction” as among the searching parameters in PEAKS software.

LFQ proteomics analysis in PEAKS

Data from the label-free quantitation for both mouse and human proteomics was validated using the false discovery rate (FDR) method built in PEAKS XPRO and protein identifications were accepted if they could be identified with a confidence score (−10lgP) > 15 for peptides and (−10lgP) > 15 for proteins; a minimum of 1 or 2 unique peptide(s) per protein after filtering for less than 2.0% FDR for peptides and less than 3% FDR for proteins identifications. The label-free quantification (LFQ) was performed using the quantification algorithm supported by the PEAKS Q module (Bioinformatics Solution Inc., version X+ and higher). The data were filtered, smoothed, and aligned in retention time, followed by feature detection based on peak volume and isotopic clustering using the algorithm of PEAKS XPro. MS/MS spectra were then extracted by the same software and used to search against the target-decoy database containing all protein entries for mouse (Mus Musculus) (17125 entries, August 2023 and 17155 entries, March 2024) or all protein entries for human Homo Sapiens (20,364 entries, August 2022) from UniprotKb reviewed database using already published protocols described elsewhere.^59,60 The relative protein abundance was displayed as a heatmap of the representative proteins of each protein group after normalization of the corresponding averaged areas (abundances) with respect to the total ion current (TIC) identified in the afferent and efferent lymph samples corresponding to control and DSS-induced colitis mouse models. The representative proteins were clustered if they exhibited a similar expression trend across all replicate samples. The hierarchical clustering was generated using a neighbor-joining algorithm and Euclidean distance similarity measurement of the log₂ ratios of the abundance of each sample relative to the average abundance, built-in in PEAKS X + Q module.⁹⁴ In the heatmap the positive values reflect fold increases (red color) and negative values reflect fold decreases (blue color). Only proteins which passed a selected significance statistical threshold (ANOVA, p < 0.05 and FDR < 3% for proteins expression) are shown in the representative heat maps generated based on LFQ analysis. The volcano plot was used to display the significance −log10 (p-value) score assigned by PEAKS XPro algorithm versus fold-change of the quantified proteins in each pair of mesenteric lymph sample (e.g., DSS/Ctr corresponding to afferent and efferent lymph; Ctr efferent/Ctr afferent and DSS efferent/DSS afferent). The PEAKS built-in t test/ANOVA (significance for p < 0.05) were used to assess the statistical significance of the fold changes in the protein expression for each paired sample set described above. The correlation plots views corresponding to the intensities recorded for each sample together with the Pearson’s correlation score were used assess the reproducibility of the LFQ experiment. A Pearson correlation score of 0.95–0.99 was reported to define a high reproducibility among the sample replicates submitted to the LFQ analysis.⁹⁵

LFQ proteomics analysis in Scaffold Quant

An independent evaluation of protein fold changes across paired afferent and efferent mouse control and DSS lymph samples was performed with the Scaffold 5.0 and Scaffold Quant (Proteome software, Portland, USA). All MS/MS samples analyzed with PEAKS Studio (Bioinformatics Solutions, Waterloo, ON Canada; version 10.6 (2020-12-21)) were exported as.mzid files and further analyzed in Scaffold (version Scaffold_5.1.2, Proteome Software Inc., Portland, OR) which was used to validate MS/MS based peptide and protein identifications. Peptide identifications were accepted if they could be established at greater than 95.0% probability to achieve an FDR less than 1.0% by the Peptide Prophet algorithm (Keller, A et al. Anal. Chem. 2002; 74(20):5383–92) with Scaffold delta-mass correction. Protein identifications were accepted if they could be established at greater than 97.0% probability to achieve an FDR less than 5.0% and contained at least 1 identified peptide. Protein probabilities were assigned by the Protein Prophet algorithm. Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony. Peptide search results were further analyzed with Scaffold Quant version 5.0.3. Peptide identifications were subsequently thresholded to achieve a peptide FDR better than 2.0% on the basis of q-values computed from the score Scaffold:Peptide Probability. Protein groups with a minimum of 1 identified peptides were thresholded to achieve a protein FDR better than 3.0% on the basis of q-values computed from the score Scaffold:Protein Probability. Normalization was applied to total ion current (TIC) and Exclusive TIC. Volcano plots were generated after applying the ANOVA/t test with the BH-correction and significance level of p < 0.05.

DIA analysis of the I-A^b peptidomes eluted from cervical and mesenteric DCs

Peptide extracts were reconstituted in 25 μL 5% acetonitrile containing 0.1% (v/v) trifluoroacetic acid and analyzed using nanoLC MS with data-dependent acquisition (DDA) and data-independent acquisition (DIA) mode on a Orbitrap Fusion Lumos mass spectrometer (ThermoFisher Scientific).

DIA-MS samples were analyzed in Scaffold DIA (3.0.1) after the data files were converted to mzML format using ProteoWizard (3.0.11748). An independent analysis was performed using the PEAKS11 and PEAKS Xpro (Bioinformatics Solutions, Waterloo, Canada) and the built-in workflow algorithm for DIA analysis which employs both data search against sample-based in house constructed DDA/DIA spectral library and spectral library unrestricted search against FASTA database (taxonomy Mus Musculus) and UniprotKb reviewed database (17125 entries, August 2023). To generate the spectral libraries, the acquired DDA and DIA raw files corresponding to the pooled samples from the biological replicates (i.e., aliquots of 1/10 from each biological replicate for each “control” and “DSS” lymph sample set) were searched with PEAKS X+/Xpro and then filtered with Scaffold software (version 5.1.2). The enzyme restriction was set up as “no enzyme” option in PEAKS X+/XPro to fit the endogenously processed peptides. Then, the spectral library was exported as a“.blib” file using the built-in available option from the Scaffold software.

Spectral library search in Scaffold DIA

Analytic samples were aligned based on retention times and individually searched against a fused DDA/DIA sample-based spectral library “.blib” file with a peptide mass tolerance of 15.0 ppm and a fragment mass tolerance of 0.05 Da. Variable modifications considered were: Oxidation M, Deamidation Q, Pyro-glu from E, Deamidation N, Oxidation W, Pyro-glu from Q Q and Oxidation or Hydroxylation H. The digestion enzyme was assumed to be No Enzyme with a maximum of 1–2 missed cleavage site(s) allowed. Only peptides with charges in the range [1 to 7] and length in the range [9–36] were considered. Peptides identified in each sample were filtered by Percolator (3.01.nightly-13–655e4c7-dirty) to achieve a maximum FDR of 0.05. Individual search results were combined and peptide identifications were assigned posterior error probabilities and filtered to an FDR threshold of 0.05 by Percolator (3.01.nightly-13–655e4c7-dirty). Peptide quantification was performed by Encyclopedia (1.2.2). For each peptide, the 5 highest quality fragment ions were selected for quantitation. The protein intensities were calculated and normalized by summation of the peptide intensities using the Scaffold DIA’s built-in normalization algorithm.

Functional pathway enrichment and GO annotations of mouse lymph LFQ proteomics

Networks, functional analyses, and biochemical and cellular pathways were generated by the ingenuity pathway analysis (IPA; Ingenuity Systems, Redwood City, CA, USA) on the list of identified proteins extracted from LFQ analysis (Table S1) using the same protocol described for the analysis of human mesenteric lymph and EV proteomics data. For quantitative display of protein fold changes, the experimentally determined protein ratios from LFQ analysis were used to calculate the experimental fold changes by rescaling the values with a log₂ transformation, such that positive values reflected fold increases and negative values corresponded to fold decreases (as shown in heatmap representations and Table S1). The calculated fold changes for each sample corresponding to the afferent and efferent lymph from control and DSS groups were imported in IPA for generating the networks and assessing the pathways’ activation z-scores (where z < −2.0 is inhibition and z > 2.0 is activation). A right-tailed Fisher’s exact test was used to calculate p-values (significance level was set at p < 0.05). The IPA analysis identified the pathways from the IPA library of canonical pathways that were most significant to the dataset (−log (p value) > 1.3).

Pathway enrichment analysis with METASCAPE

For some proteomics dataset, including the mesenteric human lymph peptidome, we performed an independent evaluation of the cellular and molecular pathways at the METASCAPE server (https://metascape.org). For each given gene list, the pathway and process enrichment analysis have been carried out with the following ontology sources: KEGG Pathway, GO Biological Processes, Reactome Gene Sets, Canonical Pathways, CORUM, WikiPathways, and PANTHER Pathway. All genes in the human genome have been used as the enrichment background. Terms with a p-value <0.01, a minimum count of 4, and an enrichment factor >1.5 (the enrichment factor is the ratio between the observed counts and the counts expected by chance) were collected and grouped into clusters based on their membership similarities. More specifically, p-values were calculated based on the cumulative hypergeometric distribution, and q-values are calculated using the Benjamini-Hochberg procedure to account for multiple testing. To further capture the relationships between the terms, a subset of enriched terms has been selected and rendered as a network plot, where terms with a similarity >0.3 are connected by edges. We selected the terms with the best p-values from each of the top 20 clusters, with the constraint that there are no more than 15 terms per cluster and no more than 250 terms in total. The network was visualized using Cytoscape, where each node represents an enriched term and is colored first by its cluster ID and then by its p-value.

Metabolomics analysis using positive (+) and negative (−) ESI nanoLC-MS

Prior to metabolomic analyses, lymph samples (20 mL) were diluted in 180 μL of ice-cold acetonitrile:methanol:water (50:30:20 v/v/v). Suspensions were vortexed for 30 min at 4°C, and then centrifuged for 10 min, at 18,213g, and 4°C. The supernatants were further used for mass spectrometry characterization of metabolites. UHPLC-MS data acquisition and processing Analyses were performed as previously described using a Vanquish UHPLC system (Thermo Fisher Scientific, San Jose, CA, USA) coupled online to a Q Exactive mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA). The (semi)polar extracts were resolved over a Kinetex C18 column, 2.1 × 150 mm, 1.7 μm particle size (Phenomenex, Torrance, CA, USA) equipped with a guard column (SecurityGuardTM Ultracartridge – UHPLC C18 for 2.1 mm ID Columns – AJO-8782 – Phenomenex, Torrance, CA, USA) using an aqueous phase (A) of water and 0.1% formic acid and a mobile phase (B) of acetonitrile and 0.1% formic acid for positive ion polarity mode, and an aqueous phase (A) of water:acetonitrile (95:5) with 1 mM ammonium acetate and a mobile phase (B) of acetonitrile:water (95:5) with 1 mM ammonium acetate for negative ion polarity mode. The Q Exactive mass spectrometer (Thermo Fisher Scientific, San Jose, CA, USA) was operated independently in positive or negative ion mode, scanning in Full MS mode (2 μscans) from 60 to 900 m/z at 70,000 resolutions, with 4 kV spray voltage, 45 sheath gas, 15 auxiliary gas. Calibration was performed prior to analysis using the PierceTM Positive and Negative Ion Calibration Solutions (Thermo Fisher Scientific).

Raw data processing and bioinformatics analysis of lymph metabolomics

Targeted and untargeted metabolomics acquired raw files to mzXML file format using Mass Matrix (Cleveland, OH, USA). Samples were analyzed in randomized order with a technical mixture injected after every 10 samples to qualify instrument performance. Metabolite assignments were performed using MAVEN (Princeton, NJ, USA). An independent validation of untargeted metabolomics data was performed with the Elements software (version 3.0.4, Proteome Software Inc., Portland, OR).

Raw data files were converted to mz5 format using ProteoWizard version(s) pwiz: 3.0.21222. Feature finding (a/k/a peak picking) was performed using Elements (version 3.0.4, Proteome Software Inc., Portland, OR). Feature finding was conducted over a mass range of [50.0··1200.0] and the entire retention time range. A noise threshold value of 0.1% of max signal and a minimum time between scans of 0.5 s was used. MS2 spectra were detected for some features. Features were organized into isotopic clusters, and all appropriate MS2 spectra were associated to appropriate features. MS1 Peak “Groups” were formed within individual samples using a same-charge inclusion threshold of 1.0 s and a cross-charge inclusion threshold of 1.0 s. Retention time alignment was performed on all samples. Consensus MS1 Peak Groups were formed using a maximum RT difference of 5 min, and those consensus spectra identified in 75% of the samples were used for RT alignment. Following RT alignment, consensus MS1 Peak Groups were regenerated, using a post-alignment maximum RT difference of 1 min. In addition, cross-sample gap filling feature reextraction was performed. Analyte groups were formed containing all analytes with the same set of ions (peaks in the MS1 Peak Group).

Spectral library search for metabolomics analysis

Candidate analyte identifications were generated by matching experimental data to spectral library data using exact mass with a mass tolerance of 20.0 ppm. If both the experimental and library data contained MS2 spectra, MS2 peaks were matched between experimental and library spectra using a fragment mass tolerance of 10.0 ppm. The following libraries were searched to generate candidate analyte identifications: NIST_hr_msms_v20.libdb (1025741 entries); hmdb_library_elements.libdb (45905 entries) and lipidmaps_library_elements.libdb (39842 entries) for lipidomics targeted analysis. The following ion types were considered when matching features to library analytes: [M+Na]+, [M + NH4]+, [2M + H]+, [2M−H]−, [M + H]+, [M−2H]2-, [M + H−NH3]+, [M + H−H2O]+, [M−H]−, [M−H−NH3]−, [M−H−H2O]−, [M+2H]2+ and in-source fragments (at least 40% of reference MS2 spectrum max intensity).

Scoring assignments for metabolites identification

To gauge confidence in candidate analyte identifications, an Analyte ID score was calculated from individual feature - library entry matches, incorporating mass accuracy, isotopic distribution, and fragmentation pattern. Analyte identifications that were identified with more ion types received a higher score than identifications made with fewer ion types. Analyte identifications with an ID Score below 0.7 were retained as “Unknown Analytes” (instead of their original identifications).

Criteria for analyte identification

Each technical replicate group’s intensities were normalized to align the median intensities and the inner quartile widths with a bilinear mapping in log space. Identifications were accepted if they could be established with an Analyte ID Score of 0.7, based on peaks with log10 intensity levels of 0.0 or higher which are identified in 1 or more samples. Graphs, and statistical analyses (either T Test or ANOVA), multivariate analyses including Partial Least Squares-Discriminant Analysis (PLS-DA), and metabolite pathway enrichment analysis were performed using MetaboAnalyst 5.0. The MS1 intensities were normalized in MetaboAnalyst after imputation of missing values (applying the KNN model for sample or feature) and 5% feature filtering based on mean intensities. An independent validation of biochemical pathways was performed with IPA as described in the proteomics section.

SDS-PAGE and western blot validation of proteins with significant differential expression in mesenteric vs. cervical lymph

Lymph proteins (10–20 μg) were mixed with the sample buffer, heated at 95 °C for 5 min, and run on a 4–20% gradient acrylamide SDS (PAGE) gels (Novex, Tris-Glycine Mini Gels, cat# XV04205PK20, from ThermoFisher Scientific) at 160 V constant following the manufacturer protocols. Proteins were transferred to the nitrocellulose membrane (0.45 mm, ThermoFisher Scientific cat# 88014) using standard wet blot procedures. Membranes were blocked in 5% nonfat dry milk (Biorad blotting-grade blocker cat# 170–6404) in 1XPBST (0.05% Tween in 1XPBS buffer, Millipore Sigma cat# P9416 and Millipore Sigma cat# 11666789001, respectively), for 1 h at room temperature. Membranes were then incubated overnight at 4 °C with the following primary antibodies: pancreatic Amylase Polyclonal Antibody (Cat#PA5–25330) (thermofisher.com); ApoA1 Polyclonal Antibody (Cat#PA1–23059) (thermofisher.com); ApoC3 Polyclonal Antibody (Cat#PA5–78802) (thermofisher.com); FABP4 Recombinant Polyclonal Antibody (2HCLC) (Cat#710189) (thermofisher.com); VDAC1/2 Polyclonal Antibody (Cat#10866-1-AP) (thermofisher.com); Syntabulin Polyclonal Antibody (Cat#16972-1-AP) (thermofisher.com). Blots were then washed in 1X PBST and incubated at room temperature for 2 h with goat anti-rabbit IgG HRP conjugate (Southern Biotech; cat# 4055–05, dilution 1:2000) secondary Ab. The enhanced chemiluminescence assay containing Super signal West-Pico PLUS chemiluminescence substrate (ThermoScientific Pierce; cat# 34577) was used to develop the membranes. Densitometry analysis was performed with the ImageJ 1.80 112 software. The relative WB counts corresponding to relative protein abundance in each mesenteric and cervical lymph replicates were calculated after normalizing the individual mean gray values with respect to the gray values of total protein quantified for each lane. The SDS-PAGE gel replica representing the identical cervical and mesenteric lymph samples used for western blot was silver-stained with the Pierce silver stain kit (Cat#24612) (thermoscientific). Densitometry analysis was performed for the total proteins separated and displayed for each lane with the ImageJ 1.80 112 software (Figure 1E).

Enzyme linked immunosorbent assay (ELISA) quantitation of cytokines in the mouse afferent mesenteric lymph

Cytokines concentrations were measured using ProcartaPlex multiplex immunoassays (Thermo Fisher Scientific) according to the manufacturer’s instructions. The concentration of each cytokine was calculated according to the standard curve provided with the assay. Each measurement was performed in biological and technical triplicates. Data were analyzed using a two-way ANOVA and multiple t-tests comparison. Statistical significance was determined using the Holm-Sidak method, with alpha = 0.05: *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001.

QUANTIFICATION AND STATISTICAL ANALYSIS

Statistical analysis was performed using Windows GraphPad Prism 9 (GraphPad Software, La Jolla, CA). Numerical results are reported as mean ± SE or ±SDV when appropriate. Data are derived from a minimum of three independent experiments unless stated otherwise. A comparison between more than two groups was performed using two-tailed unpaired one-way analysis of variance (ANOVA) or two-way ANOVA followed by multiple comparison tests, uncorrected Fisher’s LSD tests or Holm–Sidak method. Results were considered statistically significant if p ≤ 0.05.

KEY RESOURCES TABLE.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibiotics
Pancreatic Amylase Polyclonal Antibody	Thermo Fisher	Cat#PA5–25330; RRID: AB 2542830
ApoA1 Polyclonal Antibody	Thermo Fisher	Cat#PA1–23059; RRID: AB 2056393
ApoC3 Polyclonal Antibody	Thermo Fisher	Cat#PA5–78802; RRID: AB 2745918
FABP4 Recombinant Polyclonal Antibody (2HCLC)	Thermo Fisher	Cat#710189; RRID: AB 2532614
VDAC1/2 Polyclonal Antibody	Thermo Fisher.	Cat#10866–1-AP
Syntabulin Polyclonal Antibody	Thermo Fisher	Cat#16972–1-AP
Biological samples
Mice (C57BL/6J)	Jackson Laboratories	000664
Human IBD Lymph samples	Weill Cornell Medical College	N/A
Chemicals, peptides, and recombinant proteins
Bovine Albumin Solution	Sigma-Aldrich	Cat # A9576
Protease inhibitor cocktail	Roche	Cat# 04693116001
4–15% SDS-PAGE (Mini-PROTEAN® TGX™) Gels	BioRad, CA	Cat # 4561084
Oxyblot Protein Oxidation Detection Kit	Millipore	Cat #S715
OxiSelect™ Protein Carbonyl Immunoblot Kit	Cell Biolabs, Inc.	Cat # STA-308
Micro BCA protein assay kit	ThermoFisher Scientific	Cat # 23235
DMEM	VWR	Cat # 45000–324
Streptomycin, Penicillin and l-glutamine	Gibco	Cat # 10376–016
Antimycotic (Ambotericin B)	Gibco	Cat # 15290–026
Sodium Pyruvate	Gibco	Cat # 11360–070
Fetal Bovine Serum	Sigma-Aldrich	Cat #F0926
CFSE	Biolegend	Cat # 423801
DAPI	Millipore Sigma	Cat #D9542
eFluor 506 fixable viability dye	eBioscience	Cat # 65–0866-14
Fixation/Permeabilization Solution Kit	BD Pharmingen	Cat # 554714
Collagenase D	Roche	Cat # 11088858001
XCR1-Brilliant Violet 510	Biolegend	Cat # 148218
Endoproteinase Lys-C	Promega	Cat # VA1170
Trypsin	Promega	Cat #V5111
Percoll	Millipore Sigma	Cat #P1644-
NaCl	Millipore Sigma	Cat #S7653–1KG
EDTA	Millipore Sigma	Cat # EDS-500G
Tris-HCl	Fisher Scientific	Cat # BP153–500
ABTS substrate solution	Roche	Cat # 11–684 302 001
Pepstatin	Millipore Sigma	Cat # 10253286001
Leupeptin	Millipore Sigma	Cat #L2884–1MG
Phenylmethanesulfonyl fluoride	Millipore Sigma	Cat #P7626–1G
Methylglyoxal	Millipore Sigma	Cat #M0252–25ML
Octyl β-D-glucopyranoside	Millipore Sigma	Cat #O8001
Sodium azide	Millipore Sigma	Cat #S2002–5G
Phosphate Buffered Saline (PBS) Ph 7.4 (1X)	Gibco	Cat # 10010023
Protease Inhibitor 2X	ThermoFisher Scientific	Cat # A32963
Ketamine	Sigma	Cat # 1867–66-9
Xylazine	Sigma	Cat #X1126
NaCl 150mM pH 8.0	Millipore Sigma	Cat #S7653–1KG
β-octylglucoside	Sigma-Aldrich	Cat #O8001
CNBr activated Sepharose	GE Healthcare	Cat # 17–0981-01
HPLC-grade water	ThermoFisher Scientific	Cat # TS-51140
Vydac C4 Macrospin Column	The Nest Group, USA	Cat # NC1418587
Acetonitrile Optima™ LC/MS	Fisher Scientific	Cat # A955–4
Trifluoroacetic acid	Sigma-Aldrich	Cat #T6508
RIPA buffer	Sigma-Aldrich	Cat #R0278
Triton X-100	Fisher Scientific	Cat # BP151–500
TCEP, Hydrochloride, Reagent Grade	Millipore Sigma	Cat # 580567
Iodoacetamide	Millipore Sigma	Cat #I1149–5G
Sodium chloride	Millipore Sigma	Cat #S7653–250G
Sodium citrate	Millipore Sigma	Cat #W302600–1KG-K
Magnesium Chloride	Millipore Sigma	Cat #M9272
Disodium salt, pentahydrate	Millipore Sigma	Cat # 35675
Acetic Acid	Fisher Scientific	Cat # A38S-500
Formic Acid Optima™ LC/MS	Fisher Scientific	Cat # A11710X1-AMP
HEPES Buffer	Fisher Scientific	Cat # BP299–100
Methanol Optima™ LC/MS	Fisher Scientific	Cat # A456–500
Urea	ThermoFisher Scientific	Cat # 29700
Dithiothreitol (DTT)	ThermoFisher Scientific	Cat # 20290
EDTA Ultrapure 0.5 M Solution, pH 8.0	ThermoFisher Scientific	Cat # 15575020
Ammonium bicarbonate	ThermoFisher Scientific	Cat # BP2413500
Potassium chloride	ThermoFisher Scientific	Cat # AAA1166201
Potassium phosphate monobasic	ThermoFisher Scientific	Cat # BP362–500
Phosphoric acid	ThermoFisher Scientific	Cat # 02–003-602
Micro BCA protein assay kit	ThermoFisher Scientific	Cat # 23235
Trypsin	Promega	Cat #V5111
Endoproteinase Lys-C	Promega	Cat # VA1170
Glu-C	Promega	Cat #V1651
MilliQ water	Water Purification System, Millipore	N/A
0.2 mm pore membrane sterile filter units	Millipore Sigma	Cat # GSWP01300
Amicon Ultra 0.5 mL centrifugal filters	Millipore Sigma	Cat # UFC501024
8M Urea		N/A
Braford protein assay kit	BioRad, CA	Cat # 500006
Pierce™ TCEP-HCl	ThermoFisher Scientific	Cat # 20491
Glu-C	Promega	Cat #V1651
Formic Acid Optima™ LC/MS	Fisher Scientific	Cat # A11710X1-AMP
Ammonium acetate	Millipore Sigma	Cat # 5.43834
PierceTM Positive and Negative Ion Calibration Solutions	Thermo Fisher Scientific	Cat # 88324
Deposited data
Human lymph proteomics	PRIDE partner repository	PXD044801 and 10.6019/PXD044801
Human lymph peptidome	PRIDE partner repository	PXD051024
Mesenteric lymph from control (healthy) and DSS-colitis mice	PRIDE partner repository	PXD044885 and 10.6019/PXD044885
Cervical and mesenteric mouse lymph proteomics	MassIVE server	MSV000094586 and PXD051618
Cervical and mesenteric mouse lymph immunopeptidome	MassIVE server	MSV000094761 and PXD052267
Software and algorithms
Scaffold Q+S (version 4.6.2)	Scaffold Q+S (proteomesoftware.com)	N/A
Scaffold PTM 3.1.0 and later versions	Scaffold PTM (proteomesoftware.com)	N/A
PEAKS X+ and PEAKS 11 Studio/Online	www.bioinfor.com/peaks-software/	N/A
Scaffold DIA(v1.2.1 and higher), Scaffold 5.0 and Scaffold Quant	Proteome software, Portland, USA (proteomesoftware.com)	N/A
Gibbs cluster analysis	www.cbs.dtu.dk	N/A
Windows GraphPad Prism 7.0–9.0	www.graphpad.com/scientific-software/prism/	N/A
NetMHCIIpan-4.0	https://services.healthtech.dtu.dk/services/NetMHCIIpan-4.0/	N/A
Proteome Discoverer 1.4.1.14	Thermo-Scientific, 2012	N/A
UniProt Complete Human Proteome	UniProt	N/A
Ingenuity pathway analysis	IPA; Ingenuity Systems, Redwood City, CA, USA	N/A
ShinyGO 0.76	ShinyGO 0.77 (sdstate.edu)	N/A
SwissProt/UniProt database	17125 entries, August 2021	N/A
PEAKS Q module	Bioinformatics Solution Inc., version X+ and higher	N/A
FASTA database	Taxonomy Mus Musculus	N/A
METASCAPE 3.0	https://metascape.org	N/A
Cytoscape 3.10.1	https://cytoscape.org/release_notes_3_10_1.html	N/A
Maven	Princenton, NJ, USA	N/A
Elements software	Version 3.0.4, Proteome Software Inc., Portland, OR	N/A
ProteoWizard pwiz	version(s) pwiz: 3.0.21222.	N/A
MetaboAnalyst 5.0	https://genap.metaboanalyst.ca/MetaboAnalyst/	N/A
Other
Microscope Nikon	SMZ1270	N/A
5 mL ultracentrifugation tubes	326819, Beckman Coulter	N/A
Bruker TIMS/TOF	Bruker	N/A
Q-Exactive HF quadrupole orbitrap mass spectrometer	Thermo Fisher Scientific, Waltham, MA, USA	N/A
Easy nLC 1000 UHPLC	Thermo Fisher Scientific	N/A
Orbitrap Fusion Lumos mass spectrometer	ThermoFisher Scientific	N/A
Vanquish UHPLC system	Thermo Fisher Scientific, San Jose, CA, USA	N/A
Kinetex C18 column	Phenomenex, Torrance, CA, USA	N/A
SecurityGuardTM Ultracartridge - UHPLC C18 for 2.1 mm ID Columns - AJO-8782	Phenomenex, Torrance, CA, USA	N/A
Mass Matrix	Cleveland, OH, USA	N/A

Highlights.

• The cervical and mesenteric afferent lymph carries an “omic” tissue-specific signature

• This signature is reflected in the draining nodes’ dendritic cell-MHC-II immunopeptidome

• The lymph has a pivotal role in organ-specific immunosurveillance