Building on our previous work, here we assessed the persistence of memory Th2 cells in lymphoid and non-lymphoid organs up to 3 months (12 weeks) following the cure of a primary nematode infection ( Figure 1A ). We could show that H. polygyrus-cured mice harbor overall stable frequencies ( Figures 1B , S1A ) and absolute numbers (data not shown) of memory Th2 cells up to 3 months post-cure in the PEC and siLP at 8 and 12 weeks post-cure, and a trend for mildly elevated frequencies of memory Th2 cells in the mLN and lungs of mice by 12 weeks post-cure, possibly as a contribution of recirculating populations of memory Th2 cells at these tissue sites.

To characterize the transcriptional profiles of nematode-induced memory T cells, next we performed single cell transcriptomics on sorted CD4⁺CD45.2^- T cells from two H. polygyrus-cured mice at 8 weeks post-cure ( Figures 1C–E and S1B , S2 ; Tables S1 – 3 ). CD4⁺ T cells sorted from PEC, siLP, mLN and lung at a purity of > 90% and displayed distinct clustering primarily based on their tissue origin ( Figures 1C, D and S1B ). Expectedly, mLN-sorted cells (cluster I) clustered more distinctly from CD4⁺ T cells sorted from non-lymphoid organs (cluster II-V), while clusters comprising high proportions of cells sorted from the same organ (e.g. clusters IV and V) related more closely to each other. Interestingly, for clusters II and III comprising predominantly PEC and siLP-sorted cells, respectively, we observed a shared transcriptional profile of peritoneal-resident and gut-resident CD4⁺ T cells ( Figure 1D ). We also found upregulated expression of several genes like Ccr7 (CCR7) and Sell (CD62-L) in mLN cells, markers typically associated with recirculating T_CM cells ( Figure 1E ; Table S1 ). In contrast, Th2-associated genes like Gata3, Il4, Il5, Il13, costimulatory markers Icos, Tnfrsf4 (Ox40), Tnfrsf9 (CD137) and Tnfrsf18 (GITR), as well as genes like Cd69 and CD40l encoding tissue-homing and activation markers were differentially expressed in PEC, siLP and lung, consistent with the expected tissue-resident phenotype of memory cells in these tissues ( Figure 1E and Table S1 ).

To better evaluate the heterogeneous transcriptional profile of peritoneal T cells in nematode-cured mice, we further analyzed the three smaller sub-clusters of peritoneal sorted CD4⁺ T cells ( Figures 1C–E , S1B , S2 ). Here, marker genes of cluster 1 (PEC1) showed prominent over-representation in gene ontology (GO) terms like cell migration, cell-cell adhesion, interleukin-4 secretion and CD4-positive alpha-beta T cell differentiation, but no cellular response to interleukin-7, in line with the expected transcriptional profile of quiescent memory T cells. In contrast, over-representation of genes in cluster 3 (PEC3) can be found in GO terms like positive regulation of T cell mediated cytotoxicity, positive regulation of T-cell tolerance induction and regulation of T-cell anergy, rather indicative of a cluster of peritoneal regulatory T cells (Treg) ( Figure S2 ). Thus, a diverse pool of memory Th2 cells could be identified in the peritoneal cavity 8 weeks after cure of infection.

Overall, our RNAseq analysis in H. polygyrus-cured mice revealed that, at 8 weeks following cure of a primary nematode infection, mice harbor both a circulating pool of central memory-type CD4⁺ cells in secondary lymph organs like the mLN, and CD4⁺ T cells with a tissue-resident, antigen-experienced Th2 phenotype. Moreover, distinct sub-clustering of PEC cells further highlights the presence of both memory Th2 and Treg-like populations, highlighting the accumulation of diverse CD4⁺ populations in the peritoneal cavity of nematode-cured mice.

To better characterize the early memory Th2 recall responses to a secondary nematode infection, we infected and cured C57BL/6 mice of H. polygyrus (cured). 8 weeks post-cure, 8-12 of the cured mice were challenged with a secondary H. polygyrus infection and were analyzed 3 days post-challenge (reinfected, Figure 2A ). One major novel finding here was the significant expansion of Gata3⁺CD44⁺CD62-L^-Foxp3^- memory Th2 cells observed in the peritoneal compartment only, but not in the siLP, mLN or lungs of reinfected mice ( Figure 2B ). Furthermore, the expanding population of peritoneal memory Th2 cells was strongly cytokine competent, evidenced by the significantly elevated numbers of IL-4, IL-5 and IL-13-producing memory Th2 cells detected in the PEC ( Figures 2C, D ). In contrast, a significant expansion of cytokine⁺ memory Th2 cells was not observed in the siLP, mLN or lung of reinfected mice 3 days post-challenge infection ( Figures 2B, D and S3B–D ). Interestingly, the marked increase in peritoneal IL-5⁺ Th2 cell numbers in the H. polygyrus-reinfected group was further paralleled by a significant influx of eosinophils into the peritoneal cavity, but not in other tissue sites, indicative of memory Th2-driven peritoneal eosinophilia ( Figure S4 ). This finding was further emphasized by a lack of notable eosinophil infiltration into the peritoneal cavity of mice at 3 days post-primary infection (data not shown) and therefore confirms peritoneal eosinophilia as a particular feature of host recall responses to a secondary nematode infection. Thus, our results reveal the presence of a prominent Th2 recall response, localized specifically in the peritoneal cavity early following secondary H. polygyrus infection.

Among the several upregulated costimulatory markers highlighted by the scRNA-seq analysis of non-lymphoid organs, including the PEC, was Ox40 (CD134) ( Figure 1E and Table S1 ). Ox40 is an early inducible costimulatory molecule, normally promoting the survival and function of memory Th2 cells. Mice deficient for its only known ligand Ox40L (Ox40L^-/- mice) show impaired adult worm expulsion, lower IL-4 production and parasite-specific IgE responses following a secondary H. polygyrus infection, highlighting Ox40-Ox40L interactions as key for host immunity to a challenge nematode infection (17). Here, we profiled the ex vivo Ox40 expression on memory Th2 cells, comparing our cured and reinfected groups as a measure of the early induction and reactivation of host T cell responses. We found that H. polygyrus induces a prominent expansion of Ox40⁺ memory Th2 cell numbers specifically in the PEC, but not in other organs, as early as 3 days post-challenge infection ( Figure 3A ). Furthermore, peritoneal Ox40⁺ memory Th2 cells expressed significantly higher levels of the activation marker CD69 compared with Ox40^- memory Th2 cells, in line with the early inducible nature of Ox40 upon T-cell engagement with antigen-presenting cells (APCs) ( Figure 3B ).

Next, we asked whether this early memory Th2 cell reactivation in the peritoneal cavity constitutes a parasite-specific recall response. For this, we restimulated whole PEC and mLN cells from naïve and cured mice in vitro with anti-CD3/CD28 antibodies or with H. polygyrus excretory-secretory products (HES, 10µg/mL) for 72h (3 days), matching the ex vivo end-point of secondary infection in the reinfected group ( Figures 3C–F and S5 ). While HES failed to induce a detectable parasite-specific Ox40 expression in naïve mice, we observed a significant increase in Ox40 expression on PEC, but not mLN cells from cured mice ( Figures 3E and S5A ). This was paralleled by increased levels of HES-specific IL-5 release in restimulated samples from the PEC of cured mice ( Figure 3F ). In contrast, no significant parasite-specific IL-5 was detected in restimulated mLN cells ( Figure S5B ). These findings therefore clearly indicate that the murine peritoneal compartment specifically conditions an early parasite-specific memory Th2 recall response against an intestinal parasitic nematode.

To establish whether a corresponding increase in Ox40L expression on peritoneal APCs occurs during early recall responses to H. polygyrus, next we assessed the expression of Ox40L on CD11c⁺MHC-II⁺ dendritic cells (DCs) in the PEC, siLP, mLN and lungs of naïve, cured and reinfected mice ( Figures 3G, H ). DCs are a known source of Ox40L and provide key signaling for the priming and induction of both primary and memory Th2 cell responses (17, 18). Here, analysis of total CD11c⁺MHC-II⁺ DC numbers revealed a significant influx of DCs in the peritoneal compartment and to a lesser extent into the mLN of H. polygyrus-reinfected mice ( Figure 3G ). More importantly, however, a significant increase in Ox40L expression was only detectable on peritoneal DCs following secondary nematode infection ( Figure 3H ). A correlation analysis also revealed that the upregulated Ox40L expression on peritoneal DCs strongly correlates with the increased Ox40 expression on peritoneal memory Th2 cells ( Figure 3I ). Overall, these results therefore suggest that the murine peritoneal cavity is an early site of Ox40-Ox40L interactions between DCs and resident memory Th2 cells as early as 3 days post-challenge infection with H. polygyrus.

Considering the notable peritoneal localization of early host recall responses to secondary H. polygyrus infection shown here, next we aimed to decipher what mechanisms potentially drive this influx of immune cells into the peritoneal cavity of nematode-reinfected mice. The peritoneal membrane is typically lined by a layer of squamous peritoneal mesothelial (PeM) cells, expressing various cell adhesion and migration markers (19–21). Returning to our scRNAseq analysis, we found that compared with sorted gut CD4⁺ T cells, peritoneal T cells show elevated gene expression levels of cell adhesion-associated markers like Itgb2 (encodes lymphocyte function associated antigen-1 (LFA-1), which binds induced cell adhesion marker 1 (ICAM-1)), Ccr2 (encodes chemokine receptor 2 (CCR2), which binds monocyte chemoattractant proteins 1 and 3 (MCP-1/3)), Itga4 and Itgal (encode the integrin α4β1 (CD49d/CD29), which binds vascular cell adhesion marker 1 (VCAM-1)) ( Figure 4A ).

Assessing the gene expression of Vcam1 and Icam1 on PeM cells via immunofluorescence microscopy and qPCR, we found that while in naive and cured mice there are stable gene expression levels of both Vcam1 and Icam1, reinfected mice display significantly lower expression of both genes ( Figures 4B, C ). In contrast, a similar downregulation of Vcam1 or Icam1 gene expression was not observed in mice killed at 3 days post-primary infection (data not shown). Next, to see whether these changes in Vcam1 and Icam1 expression potentially correlate with later shifts in host Th2 recall responses towards the site of infection in the gut, we also assessed the numbers of Th2 cells in both the PEC and the gut-draining mLN of mice at 14 days post-challenge infection ( Figure 4D ). Here, by day 14 post-challenge we found that while total numbers of Gata3⁺ Th2 cells in the PEC were no longer increasing, there was a contrasting significant increase in Th2 cell numbers in the mLN ( Figure 4D ). In summary, these findings indicate that while the host peritoneum maintains stable levels of cell adhesion marker expression of VCAM-1 and ICAM-1 under homeostasis and following cure of a primary infection, a H. polygyrus challenge induces a significant downregulation of gene expression levels of both markers, likely enabling immune cell influx out of the peritoneal compartment. In addition, the peritoneal cavity appears to prime and support only early host recall responses to a strictly intestinal pathogen, as by day 14 post-challenge we start observing stronger host Th2 responses in the gut-draining lymph nodes instead.

Considering the high gene expression levels of cell adhesion and chemokine receptors LFA-1, α4β1 and CCR2 found on peritoneal compared with intestinal CD4⁺ T cells ( Figure 4A ), we also quantified the amounts of secreted cytokines and chemokines in peritoneal exudates of naïve, cured and reinfected mice using a magnetic bead multiplex assay ( Figure 4E ). We detected expectedly elevated levels of Th2 cytokines like IL-4, IL-5, IL-6 and IL-13, in line with the marked reactivation and expansion of peritoneal memory Th2 cells shown earlier. One novel finding here was the markedly elevated levels of the APC-secreted immunomodulatory cytokine IL-27 in the peritoneal cavity of reinfected mice, fitting well with the observed influx of peritoneal DCs following H. polygyrus challenge ( Figure 4E ). Considering the known immunomodulatory functions of both IL-27 and nematode-derived molecules, we also checked whether HES/IL-27-mediated signaling potentially contributes to the observed downregulation of Vcam1 and Icam1 in the peritoneum of reinfected mice. For this, we performed an in vitro restimulation of whole peritoneal membrane snips with either the pro-inflammatory cytokine IFNγ, known to increase ICAM-1 and VCAM-1 expression on human pleural mesothelial cells (22), or with HES either in the presence or absence of IL-27. Here, while IFNγ restimulation induced an increase in gene expression of Icam1, but not Vcam1, HES and IL-27 did not significantly modulate neither Vcam1 nor Icam1 expression on murine PeM cells in vitro, indicating that alternative signaling mechanisms in the peritoneum regulate cell adhesion marker expression in response to a nematode infection ( Figure S6 ). Finally, we found that reinfected mice displayed significantly elevated levels of several chemokines. including MCP-1/3 and the eosinophil-attracting chemokine Eotaxin (CCL11) ( Figure 4F ), fitting with the differential gene expression of CCR2 on peritoneal T cells and the marked peritoneal eosinophilia, respectively ( Figure 4F ).

In summary, we could show that within the murine peritoneal cavity, the peritoneal membrane lined by PeM cells expresses stable levels of cell adhesion markers VCAM-1 and ICAM-1 under steady state and following cure of a primary infection, and significantly downregulates the gene expression of both markers in response to a challenge nematode infection. In line with this finding, within our scRNA-seq data we also observe high differential gene expression of the binding molecules for both VCAM-1 (α4β1) and ICAM-1 (LFA-1) on peritoneal CD4⁺ T cells, therefore indicating that the notable recruitment and retention of memory Th2 cells in the peritoneal cavity of mice cured of a primary H. polygyrus infection is potentially cell adhesion marker-driven. Finally, we highlight the presence of significantly elevated levels of several APC-derived and signature Th2 cytokines, as well as chemokines associated with monocyte recruitment and activation in peritoneal exudates of reinfected mice, further highlighting the murine peritoneal compartment as a site of diverse immune cell recruitment, retention and reactivation in response to an otherwise strictly enteric pathogen.

Wild-type female C57BL/6 mice (age 8-10 weeks) were purchased from Janvier Labs (Saint-Berthevin, France). All animals were maintained under specific pathogen-free (SPF) conditions and were fed standard chow ad libidum. H. polygyrus was maintained by serial passage in C57BL/6 mice (H0099/13). Mice were infected with 225 third-stage infective (L3) H. polygyrus larvae via oral gavage in 200µL drinking water. For cure of infection, mice were treated on two consecutive days during the acute stage of infection (day 14-15) with 2mg pyrantel pamoate (Sigma, St. Louis, MO, USA) in 200µL drinking water via oral gavage. On indicated days, mice were sedated via isofluorane inhalation, followed by cervical dislocation. The efficacy of curative treatment with pyrantel pamoate was confirmed at dissection by the lack of adult worms in the small intestine of cured mice. All animal experiments were performed in accordance with the National Animal Protection Guidelines and approved by the German Animal Ethics Committee for the Protection of Animals (LAGeSo, G0176/20).

The isolation of PEC, siLP and mLN was performed as previously described (4). For the isolation of lung cells, the chest cavity was opened via incision and the whole lung was perfused with 20mL ice-cold PBS via puncture to the heart using a 27G needle and syringe. Each lung was then removed and manually minced into small pieces. The minced tissue was then transferred to a 50mL falcon tube containing 10mL digestion medium (150µg/mL collagenase D and DNase I in PBS). The samples were incubated for 1 hour at 37°C under continuous shaking at 250rpm. Following the digestion step, the lung tissue was passed through a 70µm cell strainer into a fresh 50mL falcon tube. After erythrocyte lysis, the lung samples were washed and resuspended in cRPMI. All cell suspensions were counted using a CASY automated cell counter (Roche-Innovatis, Reutlingen, Germany).

Cell suspensions from PEC, siLP, mLN and lung were obtained and counted as described above from two H. polygyrus-cured C57BL/6 mice at 8 weeks post-cure. Immediately prior to dissection, each mouse was injected intravenously with anti-mouse CD45.2-A700 (clone 104) to allow for the exclusion of any blood-derived cells in circulation. Each sample was stained with anti-mouse CD4-PerCP (clone RM4-5) and was sorted for live single CD4⁺CD45.2^- T cells on a FACS Aria cell sorter (BD Biosciences, Heidelberg, Germany).

Capture and processing of single CD4⁺ T cells was performed using the Fluidigm C1 autoprep system. Sorted CD4⁺ T cells (50 cells/µl) were mixed with (ratio 6/4) C1 Cell Suspension Reagent (Fluidigm) and subsequently loaded onto a 5-10-μm-diameter C1 Integrated Fluidic Circuit (IFC; Fluidigm From the nine chips used, we captured 517 single cells (59.8%). ERCC (External RNA Controls Consortium) spike-in RNAs (Ambion, Life Technologies) were added to the lysis mix. Reverse transcription and cDNA preamplification were performed using the SMARTer Ultra Low RNA kit (Clontech). Sequencing libraries were prepared using Nextera XT DNA Sample Preparation kit with 96 indices (Illumina), according to the protocol supplied by Fluidigm. Size distribution in final libraries was analyzed by Bioanalyzer High Sensitivity DNA assay (Agilent) and library concentrations were quantified by Qubit High Sensitivity DNA assay (ThermoFisher). A total of 384 libraries were pooled equimolar and sequenced on an Illumina HiSeq 1500 system in rapid mode v1 by a paired-end dual-indexing run with 2x 125 cycle reads. Subsequently, the generation and trimming of fastq files was performed followed by alignment of the reads to mm10 mouse genome version.

Quality control of the raw fastq files was done with FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/). Trimming of the raw reads was done with Trimmomatic v0.36, with the following filters and settings in this order: Nextera adaptors, ILLUMINACLIP:${ADAPTER}:2:30:10 (adapter trimming), LEADING:3 (5’ Trimming phred<3), TRAILING:3 (3’ Trimming phred<3), SLIDINGWINDOW:3:20 (mean phred score in sliding window size 3 had to be at least 20), MINLEN:36 (discard reads shorter than 36bp) (46).

Single cell RNA-Seq data were analyzed using R and the indicated R-packages. At first, counts per gene were calculated from the bam files by adding all counts mapped to the region of a corresponding gene (Rsamtools) (47). Quality control of the count data was performed with the R-package scater (48, 49). Cells with low library size, low numbers of features or a high number of mitochondrial sequences were filtered out using the quickPerCellQC function. Features (genes) without any count in all remaining cells were removed from the dataset and count data were normalized and log2-transformed after adding a pseudocount to each value. Variances in the expression profile of each gene were modelled based on a fitted mean-variance trend (scran - modelGeneVar) (50), decomposing it into technical and biological components. Hypervariable expressed genes were identified by applying a threshold of 4 for the variances of the biological component. 500 genes with the highest biological variances were selected for dimension reduction using the UMAP algorithm (scater) (48). In order to further stratify the dataset to genes driving systemic substructures and to remove noise, pairwise correlations between all genes were calculated (scran - correlatePairs) (50). P-values were adjusted for multiple testing by calculating the false discovery rates and significant genes were selected by a fdr below 0.05. These genes were used in a hierarchical clustering with euclidean distances and ward.D2 linkage (R - inbuilt dist and hclust function) (49). Five clusters were defined by the cutTreeDynamic function (dynamicTreeCut) (51). The dendrogram information (order of the columns), as well as the cluster definitions were used in the heatmap representation of pre-selected genes (heatmap3) (52).

Differential expression of genes between PEC and other tissues were determined by fitting negative binomial generalized linear models combined with likelihood ratio tests (edgeR - glmFIT and glmLRT) (53). Ranks were determined for each comparison and the whole set was ordered according to the minimal rank of each gene.

Marker genes for the three PEC clusters were determined by combined pairwise comparisons (t-Tests) using the findMarkers function of the scran package. Significant genes were selected by a combined p-value below 0.05. Functional annotation and overrepresentation analysis of these gene sets was done by DAVID using the R package “RDAVIDWebService” (54). Comparison of functional aspects between the three PEC clusters was done using the R-package “clusterProfiler” (55). Results for the biological branch of the gene ontology system were further processed. GO terms with overrepresentation in all PEC clusters were removed to focus for the differences. Each term was mapped (if possible) to the significant most specific (child) term. 15 top-ranking terms from this processed set for each PEC cluster were then compared in a dotplot.

The antibodies used for the detection of surface and intracellular markers are described in Table S4 . Dead cells were excluded using eFluor780 or eF560 fixable viability dye (Thermo Fisher, Waltham, USA). For intracellular staining of cytokines and transcription factors, cells were fixed and permeabilized using the Fixation/Permeabilization kit and Permeabilization buffer from ThermoFisher/eBioscience. Samples were analyzed on a Canto II flow cytometer and on an Aria cell sorter (BD Biosciences, Heidelberg, Germany). The data was analyzed using FlowJo software Version 10 (Tree star Inc., Ashland, OR, USA). The data for naïve, cured (8wpc and 12wpc) and reinfected mice presented in this study are pooled from two independent experiments with 3-5 mice per group in each experiment.

For the analysis of parasite-specific Ox40 expression and IL-5 secretion, 5x10⁵ whole PEC or mLN cells were plated out per well in a round-bottom 96-well cell culture plate in a final volume of 200µL RPMI medium, containing 10% FCS, 100U/mL penicillin and 100µg/mL streptomycin (all from PAA, Pasching, Austria). The cells were stimulated with either anti-CD3/CD28 antibodies (1µg/mL) or H. polygyrus excretory-secretory (HES) antigen (10µg/mL). The cells were then incubated for 72 hours at 37°C and 5% CO₂. After 72 hours, the cell pellets were collected and stained for flow cytometric analysis, as described above, while the culture supernatants were collected and used for quantifying IL-5 secretion via ELISA, as described below. The data presented for this assay are pooled from two independent experiments with 2-4 mice per group in each experiment.

Parasite-specific IL-5 production of restimulated PEC and mLN cells was measured via sandwich ELISA using the Mouse IL-5 Uncoated ELISA kit as per the manufacturer’s instructions (Invitrogen). Absorbance was measured on a Biotek Synergy H1 Hybrid Reader at a 450nm wavelength.

Eosinophil peroxidase (EPO) levels in blood serum were quantified using the Mouse Eosinophil Peroxidase (EPX) ELISA Kit as per the manufacturer’s instructions (DLdevelop, Jiangsu, PRC). Serum samples were diluted 1:5 in Diluent Buffer and were added to the plate with serially diluted standards. Absorbance was measured on a Biotek Synergy H1 Hybrid Reader at a 450nm wavelength. The data presented for this assay are pooled from two independent experiments with 3-5 mice per group in each experiment.

Sections were cut from formalin-fixed and paraffin-embedded tissues, dewaxed and subjected to a heat-induced epitope-retrieval step. For immunofluorescence, endogenous peroxidase was blocked (Dako REAL Peroxidase-Blocking Solution, Agilent, Santa Clara, CA, U.S.) and sections were incubated with anti-VCAM-1 (clone EPR5047, Abcam, Cambridge, U.K.). For detection, the EnVision+ Single Reagent (HRP. Rabbit, Agilent) and the Opal 570 reagent (Akoya Biosciences, MA, U.S.) were used. Nuclei were stained with DAPI (Sigma-Aldrich Chemie GmbH, Munich, Germany) and slides were cover-slipped in Fluoromount G (Southern Biotech, Birmingham, AL, U.S.). Stained sections were analyzed in a blinded manner using an AxioImager Z1 microscope (Carl Zeiss Microscopy Deutschland GmbH, Oberkochen, Germany). The immunohistochemistry data shown here are representative of two independent experiments with 2-4 mice per group in each experiment.

At necropsy, 1cm² tissue snips of peritoneal membrane were excised, flash-frozen in liquid nitrogen and were stored at -80°C. RNA was isolated using the Monarch Total RNA Miniprep kit (New England BioLabs, MA, USA) according to the manufacturer’s instructions. 2μg of RNA was then reverse-transcribed to cDNA using the High-Capacity RNA-to-cDNA kit (Applied Biosystems, Foster City, CA, USA). Relative gene expression was determined via quantitative real-time PCR (qPCR) using 10 ng of cDNA and FastStart Universal SYBR Green Master Mix (Roche). Primer pairs are described in Table S5 . Efficiencies for each primer pair were determined by generating a standard curve. mRNA expression was normalized to the housekeeping gene glyceraldehyde 3-phosphate dehydrogenase (GAPDH) and was calculated by the Roche Light Cycler 480 software. The qPCR data are pooled from three independent experiments with 2-4 mice per group in each experiment.

At necropsy, 1mL of PBS buffer was carefully injected and retracted from the cavity of naïve, cured and reinfected mice using a 20G needle and syringe. These peritoneal exudate samples were then centrifuged, the supernatants were collected and stored at -20°C for performing a multiplex sandwich ELISA-based Luminex immunoassay using the Mouse Cytokine & Chemokine 26-plex ProcartaPlex kit as per the manufacturer’s instructions. The results were read using a Luminex MagPix machine. The data presented for this assay are pooled from two independent experiments with 3-4 mice per group in each experiment.

For the in vitro restimulation of murine PeM cells, snips from naïve C57BL/6 mice were carefully excised and were plated out in 200µL cRPMI medium in 96-well plates (one snip per well). Snips from each mouse were then either left unstimulated as a negative control or were incubated with IFNγ (10ng/mL), HES (10µg/mL) or HES and IL-27 (10µg/mL + 50ng/mL, respectively) for 24h at 37°C, 5% CO₂. After the 24h incubation, the snips were then snap frozen and stored at -80°C for later RNA extraction, reverse transcription and qPCR analysis of Vcam1 and Icam1 gene expression levels as described above. The data presented for this assay are pooled from two independent experiments with 2-4 mice per group in each experiment.

Statistical analysis of FACS data was performed using GraphPad Prism software version 9.0.1 (La Jolla, CA, USA). Results are displayed as mean ± SD and significance is displayed as *p<0.05, **p<0.01, ***p<0.001. Results were tested for normal distribution using the Shapiro-Wilk normality tests, followed by ANOVA or Kruskal-Wallis combined with Tukey’s or Dunn’s multiple comparison testing.