SINGLE-CELL AND SPATIAL MULTI-OMICS IDENTIFY INNATE AND STROMAL MODULES TARGETED BY ANTI-INTEGRIN THERAPY IN ULCERATIVE COLITIS

BACKGROUND: Mesenteric fat is increasingly believed to exert an important influence on intestinal inflammation. A hallmark of severe Crohn's disease (CD) is the accumulation of mesenteric fat around the inflamed intestine, typically in the terminal ileum. This so-called "creeping fat" is an important indicator of regional disease activity and is often used by surgeons as a measure of disease severity. Creeping fat correlates with intestinal transmural inflammation, muscular hypertrophy, and fibrosis/stricture formation. The pathologic function of creeping fat, however, remains unclear. We investigated this matter through the lens of adipose-derived stem cells (ASCs). ASCs are mesenchymal cells with the capacity for self-renewal and multipotential differentiation. They are critical to the adipose function and, importantly, can be cultured and studied ex vivo in a controlled and robust fashion. Here we aimed to characterize the function of ASCs in CD. METHODS: ASCs were isolated from the creeping fat of patients with CD and the mesenteric fat of control patients (n =4). After purifying and testing via fluorescence-activated single-cell sorting (FACS), bulk RNA sequencing and gene set enrichment analysis (GSEA) was performed to investigate the distinctive gene pattern of ASCs. The levels of various pro-inflammatory cytokines (TNF-a, IL-6, and IL8) were measured by ELISA and the apoptosis was measured by immunostaining of cleaved- caspase-3. RESULTS: Principal component analysis (PCA) and gene heat mapping revealed a distinctive gene pattern in ASCs from creeping fat compared to those from controls. ASCs from creeping fat were characterized by less colony formation as well as reduced differentiation and migration capacity. GSEA results suggested an up-regulated pathway for inflammatory response and apoptosis in ASCs from creeping fat. ELISA assays confirmed that ASCs from the creeping fat secreted less pro-inflammatory cytokines, including TNF-α, IL-6, and IL-8. ASCs from the creeping fat showed a lower level of cleaved caspase-3, suggesting that ASCs from CD patients experience reduced levels of apoptosis. CONCLUSIONS: Our study provides an explanation of the accumulation of creeping fat in CD patients - namely, that ASCs from creeping fat undergo less apoptosis and have reduced differentiation and migration capacity, resulting in an accumulation of this distinctive mesenteric fat. Moreover, we found that ASCs from CD patients secrete less pro-inflammatory cytokines, lending support to the emerging consensus that creeping fat likely serves a protective function in CD, by serving as a barrier surrounding the inflamed intestine. As such, our study may assist in the development of novel, adipose-based CD therapies.

Background: About one third of patients with Crohn’s disease (CD) develop strictures during their disease course requiring surgical resection. The immune landscape involved in this process is poorly understood. Therefore, we aimed to characterise the fibroblast phenotype, immune cells and their mediators involved in intestinal strictures.
Methods: We included 25 CD patients with stricturing disease in the terminal ileum (TI) and 10 controls with colorectal cancer (CRC), all undergoing an ileocolonic resection. Transmural samples from the resection specimen of the TI were obtained. Macroscopically, CD tissue was divided into unaffected, fibrostenotic and inflamed regions by an experienced histopathologist. Next, mucosa was separated from deeper layers, after which single cells were isolated and fluorescently stained for flow cytometry. Protein levels were determined via the MesoScale Discovery (MSD) platform in the corresponding samples. Comparisons between CRC controls and CD patients were performed via an unpaired t-test or Mann-Whitney analysis and corrected for multiple testing.
Results: An increase in active fibroblasts and decrease in inactive fibroblasts in the fibrotic and inflamed mucosa (p=0.0002 and p<0.0001) and deeper layers (p=0.003 and p=0.02) when compared to the CRC controls was observed, confirming ongoing tissue remodelling. An enrichment of active eosinophils was only seen in the fibrotic deeper layers (p=0.02), although an increase in T helper 2 (Th2) cells was observed in both the fibrotic and inflamed deeper layers (p=0.02 and p=0.04). In contrast, T helper 1 (Th1) cells were decreased in both fibrotic and inflamed mucosa (p=0.03 and p=0.02) and deeper layers (p=0.01 for both). Regulatory T cells were significantly enriched in both fibrotic and inflamed mucosa (p<0.0001 and p=0.0005) and deeper layers (p=0.01 and p=0.006) (figure 1). Protein quantification confirmed a significant increase in transforming growth factor-β3 (TGF-β3) in the fibrotic (p=0.007) and inflamed (p=0.0002) layers, but not in the more superficial mucosa. Comparably, IL-1β was increased in the fibrotic (p=0.05) and inflamed (p=0.05) deeper layers. A similar observation was made for basic fibroblast growth factor (bFGF) (p=0.004), although only a trend could be seen in the fibrotic deeper layers (p=0.08) (figure 2).
Conclusion: The fibrotic and inflamed tissue of CD patients is characterized by increased activated eosinophils, Th2 and regulatory T cells and decreased Th1 cells, as well as many of their mediator cytokines. The current immunological characterisation can help to prioritise potential anti-fibrotic targets for stricturing CD.

Background: In Crohn's disease (CD) extravasation of pro-inflammatory peripheral monocytes plays a fundamental role in innate immunity, gut homeostasis and modulating intestinal disease. However, the molecular mechanisms mediated by circulating monocytes contributing to chronic inflammation and disease progression remains largely unknown. Aim: Identify molecular pathways underlying peripheral monocytes pathophysiology in severe treatment-resistant CD patients. Methods: CD14⁺ monocytes were purified from 73 CD patients requiring surgery. Expression profiles were generated by RNAseq and pathway analysis using ENRICHR. The analysis strategy is presented in Figure 1. Results: Unsupervised clustering of CD14⁺ gene expression stratified CD-mono into 2 transcriptomic signature subtypes, one of which designated PAF-CD-mono, was clinically associated with perianal fistula (p=3.7E-04). CD14⁺ transcriptomic stratification and PAF-CD-mono association with perianal fistula was confirmed in a separate patient cohort. CD monocyte subtype signatures were not associated with gender, age, disease location/behavior or therapeutic treatment nor with WBC, platelet and monocyte count or CRP lab values at time of surgery (Table 1). Nearly all (81%) differentially expressed genes (DEG) (830, p<0.001, FC 1.5) were down-regulated in the PAF-CD-mono vs CD-mono subtype. DEG were significantly enriched in GWAS-associated IBD variant loci (47/241 loci). The top identified DEG enrichment analysis pathways were strongly associated with megakaryocyte progenitor cell markers, monocyte-platelet aggregation/activation/signaling and clotting cascade (p=E-29 to -07). These pathways provided the focus to further define perianal-fistula biomarkers utilizing DEG implicated in platelet-monocyte complex formation (SELPL, SELP), monocyte inflammatory cytokine/chemokine expression (TNF, CCL4, CXCL3,CCL3), IBD candidate thromboembolic disease (TED)-risk genes (PROS1, PROC, MTR) and monocyte activation/inflammatory markers (CD1D, CD226). Low expression of CD226 (p=0.01) and SELP (p=0.002), both flanking GWAS-identified IBD-risk loci, were associated with perianal-fistula. Initial experiments demonstrated that CD226 mRNA expression is reflected in parallel by the level of circulating CD226 in plasma at time of surgery and low levels of circulating CD226 protein are associated with perianal fistula. Conclusion: Severe CD can be stratified based on peripheral monocyte gene expression into 2 functionally diverse profiles associated with IBD GWAS loci, monocyte-platelet activation and clotting molecular pathways and clinically with perianal fistula. Diminished expression of SELP and CD226 mRNA and soluble circulating CD226 protein levels may serve as valuable diagnostic parameter for monitoring perianal disease progression and aid in the design of novel CD subtype-specific therapeutics.

Background:
Single cell RNA sequencing (scRNAseq) offers unparalleled resolution into the cellular drivers of inflammatory bowel disease (IBD). Understanding the cellular culprits of disease in context of therapy is necessary not only to actualise precision medicine and aid drug development, but also to shape clinical strategies of drug sequencing in patients.

Methods:
To assess the impact of anti-TNF therapy, we generated a longitudinal scRNA-seq therapeutic atlas of IBD. Site- and patient-matched biopsies of 38 biologic-naïve patients with Crohn’s disease (CD) or ulcerative colitis (UC) from 5 gut regions (terminal ileum, ascending colon, descending colon, sigmoid and the rectum) before and after treatment with adalimumab and biopsies from 3 healthy controls were obtained. Levels of adalimumab confirmed adequate dosing and excluded patients with anti-drug antibody-mediated loss of response.

Results:
987,743 high-quality single-cell transcriptomes from 216 gut samples were obtained from 16 responders and 22 non-responders. Cellular characteristics associated with site (ileum, colon), and disease (CD, UC) were identified. Transcriptomic variance across healthy ileum and colon was restricted to epithelial cells indicating specialised functions in vitamin B12, iron, and fat-soluble vitamin absorption. Epithelial and lymphocyte compartments had disease specific features.

The impact of TNF modulation in-vivo in patients with IBD was explored. Activated CD4+ (CD: 29%, UC: 29.6%) and CD8+ T memory cells (CD: 20.2%, UC: 18.2%) represent the largest sources of TNF mRNA transcripts in the inflamed gut. The propensity of cell types to respond to TNF signalling pre-treatment negatively correlated with decreases in TNF signalling following treatment in responders; CD: R=-0.43, p=0.0031 and UC: R =-0.45, p=0.0049. Genes enriched in inflammatory monocytes (S100A9, FTH1, IL1RN) increased despite treatment in non-responders with CD. In non-responders with UC, persistent NF-KB activation amongst other pathways occurred across cell compartments including the stroma, CD4+ T cells, and epithelium.

To characterise covarying cell states, a graph-based approach identified ‘hubs’ of inflammation across CD and UC. Gene expression programs (GEPs) in hubs were associated with histological features. After projection to large scale bulk RNA sequencing data, constituent GEPs were associated with anti-TNF therapy non-response at baseline.

Conclusion:
We have generated a single cell atlas of IBD in context of anti-TNF therapy. By systematically deconstructing IBD in context of tissue site and disease, the cellular drivers of CD and UC have been distinguished. Changes induced by anti-TNF at single cell resolution have been revealed by leveraging paired sampling within patients, highlighting pathways that remain persistently active in the context of tissue architecture.

Background: Ulcerative colitis (UC) is a chronic inflammatory disorder of the intestine driven by mucosal immune and stromal subsets, culminating in intestinal epithelial injury. Vedolizumab (VDZ) is an anti-integrin monoclonal antibody that interferes with intestinal leukocyte trafficking and is effective for treating UC. VDZ is thought to primarily inhibit lymphocyte trafficking to the intestine, but its effect on other cell subsets is poorly understood.
Methods: To better understand the inflammatory cells that contribute to colitis and respond to VDZ, we performed a comprehensive single-cell transcriptomic and proteomic analysis (scRNA-seq, CITE-seq, and CyTOF) of peripheral blood and colonic biopsies in healthy controls (HC, n=4), patients with UC on aminosalicylates (n=4), and patients with UC on VDZ (n=4). On matching tissues, spatial transcriptomic and proteomic assays on formalin-fixed paraffin-embedded (FFPE) biopsies were performed using different technologies, including multiplex ion beam imaging (MIBI), co-detection by indexing (CODEX), and two platforms for multiplexed RNA-ISH with subcellular resolution (12-plex and 960-plex). Gene set enrichment analysis (GSEA) of identified gene signatures was validated using a longitudinal transcriptomic dataset (n=11 VDZ responders and n=9 VDZ non-responders). Both unsupervised and supervised methods were used to infer intestinal cell subset variation by disease and treatment status.
Results: VDZ showed small effects in peripheral blood compared to colonic tissues. We identified tissue trafficking of mononuclear phagocytes (MNPs) as a primary mechanism of action for VDZ, with comparatively modest effects on lymphocytes. Spatial proteomics and transcriptomics demonstrated increased density and proximity of inflammatory MNP and fibroblast subsets in UC compared to HC, with inhibition by VDZ. GSEA of longitudinal transcriptomic data confirmed the reduction of inflammatory MNP and stromal signatures, with epithelial healing in VDZ responders. VDZ non-responders could be differentiated from responders pre-treatment by enrichment of endothelial, activated fibroblast, and macrophage signatures.
Conclusion: This study combines multiple technologies to enable comprehensive single-cell and spatial tissue multi-omics, revealing a significant effect of VDZ on inhibiting MNP intestinal trafficking, reducing stromal cell activation, and healing the mucosa. This spatial transcriptomic and proteomic cell atlas is compatible with FFPE tissue and will aid in future studies evaluating signatures associated with disease and treatment response.