Target discovery for IBD has traditionally relied on genetic associations, which lack the cellular resolution needed to identify novel, actionable, cell type-specific disease pathways. Here, we describe an integrated analytical and experimental framework that leverages harmonized single-cell data to systematically discover novel therapeutic strategies for IBD. We used AMICA DBTM, Immunai's harmonized database of single-cell RNA datasets to construct a harmonized 1 million single-cell atlas of the human intestine. We applied a machine learning framework (Immune Patient Representation, IPR) to identify disease-associated transcriptional programs and cell type-specific gene targets. Candidate targets were prioritized using atlas-derived metrics, refined using custom criteria emphasizing translational actionability, and validated across independent clinical cohorts. Select candidates were evaluated in human primary-cell models reflecting the target's cell-type context. The IPR framework identified 85 disease-associated transcriptional programs and ranked 400 cell type-specific target genes across immune and stromal lineages. Disease-associated programs were interpreted using a structured AI-assisted reasoning framework for structured biological reasoning, linking them to IBD-relevant pathways and guiding the identification of novel, promising gene targets. Functional validation of two cell-type-specific candidates, PTGIR in myeloid cells and IL6ST in fibroblasts, confirmed the reduction of inflammatory and fibrotic pathways linked to IBD pathology. Multi-omic profiling and projection of in vitro phenotypes to patient datasets demonstrated the reversal of disease-associated programs via mechanisms distinct from those of existing biologics. Our single-cell anchored, machine-learning framework integrates in silico discovery with experimental validation, revealing new cell type-specific therapeutic opportunities and supporting a scalable approach for precision target discovery in IBD and other immune-mediated diseases. ### Competing Interest Statement All authors declare potential competing interests. Authors affiliated with Immunai and AstraZeneca are employees of their respective organizations and may hold equity or stock options as part of standard compensation. The authors' institutions have filed provisional patent applications related to aspects of the work described in this manuscript. No other financial, professional, or personal conflicts relevant to this manuscript are reported.

Crohn's disease (CD) is a chronic inflammatory bowel disease exhibiting substantial heterogeneity in clinical presentation and response to therapy. To explore its molecular basis, we developed IBDverse, a large single-cell RNA sequencing (scRNA-seq) dataset of terminal ileal biopsies, profiling over 1.1 million cells from 111 patients with CD and 232 healthy controls. This resource integrates discovery and replication cohorts for the robust identification of CD-associated cell types, genes and pathways. We uncovered epithelial changes marked by interferon-driven upregulation of major histocompatibility complex class I molecules that persisted in progenitor cells after macroscopic inflammation resolution. ITGA4+ macrophages were identified as key inflammatory drivers, showing enriched JAK-STAT signaling and cytokine expression (interleukin-6 (IL-6), IL-12 and IL-23). Heritability analysis linked inflammatory monocytes and macrophages to CD susceptibility, implicating resident and recruited immune cells in pathogenesis. These findings establish a comprehensive cellular and molecular framework for CD, offering insights into disease mechanisms and therapeutic opportunities.

Abstract Background Crohn’s Disease (CD) is a chronic condition that can impact any part of the gastrointestinal tract. The sustained inflammation is driven by the continuous influx of inflammatory leukocytes into the gut mucosa, which is regulated through chemokine gradients and adhesion molecules, including integrins. In the gut, the migration of inflammatory cells to specific segments is thought to be mediated by selected integrins and/or chemokine axes. The CCL25/CCR9 axis has been identified as a key pathway for immune cell migration into small bowel segments, such as the ileum. In this study, CITE-seq and spatial transcriptomic analysis was used to integrate proteomic and transcriptomic profiles to characterise CCR9+ immune cells in the ileum compared to the colon. Methods Biopsies were obtained via ileocolonoscopy from macroscopically inflamed and normal mucosa with appropriate consent from healthy volunteers and IBD patients visiting the Department of Gastroenterology at the Sahlgrenska Hospital. Biopsies were frozen in Cryostor CS10 or formalin fixed and paraffin embedded. Cells from frozen biopsies were sorted, CD45+ cells stained with CCR9 and Totalseq-C human universal antibody cocktail, followed by amplification, separation of antibody-derived DNA from mRNA derived cDNA, library construction and sequencing of both libraries. For spatial transcriptomics, tissue sections were deparaffinised, stained with H&E and decrosslinked before human whole transcriptome probe hybridisation, ligation, extension, library construction and sequencing. Cells were annotated using a transfer learning approach combining several reference single cell datasets including detailed CD4+ and CD8+ T cell datasets and utilising both antibody data and mRNA transcriptomic data. Results CCR9 positive cells appeared scattered throughout the lamina propria and expression at mRNA level was enriched in the ileum compared to the colon with about 10% of CD4 and 17% of CD8 T cells expressing CCR9. Interestingly at protein level, expression was more restricted with the highest expression on cells that exhibited proteomic and transcriptomic features associated with resident CD4 memory T cells (Trm). This population almost exclusively (>90%) consisted of CCR9 positive cells and their proteomic and transcriptomic profile consistent with the recently described effector and innate-like phenotype of Trm cells that is thought to contribute to Crohn’s disease pathology. Conclusion These data may point to CCR9 positive cells, and in particular Trm cells, in the ileum as important pathological cells in Crohn’s disease and provide further support for depleting CCR9 cells as a potential therapeutic target for the treatment of ileal Crohn’s disease.

Abstract Background Single-cell technologies enable the fine mapping of disease and treatment mechanisms in inflammatory bowel disease (IBD). We sought to leverage these insights to discover novel drug targets with precise therapeutic hypotheses. To this end, we previously constructed one of the largest integrated single-cell atlas of IBD patient-derived tissue samples1. We then used Immunai’s ImmunoDynamics EngineTM, a proprietary machine learning (ML) framework, to identify transcriptional signatures of inflammation and non-response to anti-TNF treatment in specific cell populations, and extracted a ranked gene list that was highly enriched in targets of currently marketed IBD therapies. We selected top-ranked genes for in vitro functional genomic validation and leveraged the IBD single-cell atlas to support clinical relevance of in vitro observations. Here, we present a target case study demonstrating this approach. Methods The top 400 genes were triaged to choose 20 candidate targets based on novelty, druggability, genetic association with IBD, and experimental feasibility. Each of these 20 targets was individually deleted under optimized culture conditions in the primary human cell type(s) in which it showed a significant disease association in the IBD single-cell atlas, including monocyte-derived macrophages (MDM), CD8, CD4 helper and/or regulatory T cells, and intestinal fibroblasts. Knockout (KO) cells were characterized by flow cytometry, secreted proteome profiling, and bulk transcriptomics, with suitable positive and negative controls per cell type. KO-induced transcriptomic changes were interpreted via gene-set enrichment of canonical pathways and of IBD-associated transcriptional signatures derived from orthogonal clinical cohorts. KO-induced transcriptomic signatures were also computed in each sample in the IBD single-cell atlas. Results Deletion of one selected target in MDMs led to a reduction in TNFα and IFNγ signaling and in the expression of IBD inflammation-associated signatures from orthogonal patient cohorts. Genes downregulated by the deletion of this target in MDMs were enriched in macrophages from inflamed tissue in the IBD single-cell atlas. Conclusion The functional genomic data indicate that deletion of the selected target shifts MDMs away from a disease-associated inflammatory state, suggesting that its pharmacologic inhibition in macrophages may have a therapeutic benefit. These findings highlight the relevance of cell type-specific methods to understand disease mechanisms and characterize putative novel targets. Applying these experimental and clinical contextualization methods to additional candidate genes will further expand the basis for novel therapeutic strategies in IBD. References 1. De Baets G, d’Rozario J, Gehrmann U, et al. Inflammatory Bowel Disease single cell atlas construction to enable cell type-specific target identification. J Crohns Colitis 2024;18(Supplement_1), i143. doi: 10.1093/ecco-jcc/jjad212.0080

Gasdermins are a family of structurally related proteins originally described for their role in pyroptosis. Gasdermin B (GSDMB) is currently the least studied, and while its association with genetic susceptibility to chronic mucosal inflammatory disorders is well established, little is known about its functional relevance during active disease states. Herein, we report increased GSDMB in inflammatory bowel disease, with single-cell analysis identifying epithelial specificity to inflamed colonocytes/crypt top colonocytes. Surprisingly, mechanistic experiments and transcriptome profiling reveal lack of inherent GSDMB-dependent pyroptosis in activated epithelial cells and organoids but instead point to increased proliferation and migration during in vitro wound closure, which arrests in GSDMB-deficient cells that display hyper-adhesiveness and enhanced formation of vinculin-based focal adhesions dependent on PDGF-A-mediated FAK phosphorylation. Importantly, carriage of disease-associated GSDMB SNPs confers functional defects, disrupting epithelial restitution/repair, which, altogether, establishes GSDMB as a critical factor for restoration of epithelial barrier function and the resolution of inflammation.

Abstract IL-33 is a pleiotropic cytokine known to possess dichotomous roles during gut health and disease. We previously described that IL-33 promotes epithelial restitution/repair in otherwise healthy (C57BL/6) mice that have been challenged with dextran sodium sulphate (DSS) to induce acute colitis, with the end result of efficient resolution of inflammation. However, in SAMP1/YitFc (SAMP) mice that spontaneously develop Crohn’s disease (CD)-like ileitis, elevated IL-33 levels persist as disease progresses, perpetuating chronic gut inflammation and severe fibrosis. While IL-33 has been implicated in the development of inflammation-associated fibrosis, the precise mechanism(s) by which this occurs remains unclear. The aim of this study was to determine IL-33-dependent events leading to intestinal fibrosis in ileitis-prone SAMP mice. Our results show IHC co-localization of IL-33 with fibrotic lesions, specifically in cells morphologically-consistent with both macrophages and subepithelial myofibroblasts (SEMFs). Bulk RNA-Seq analysis of the human SEMF cell line, CCD-18Co, stimulated +/- IL-33 identifies one of the highest-expressing transcripts as SERPINE1, encoding for plasminogen activator inhibitor-1 (PAI-1), which has previously been reported as highly-enriched in IBD patients with active disease that do not respond to anti-TNF therapy. Analysis of publicly-available scRNA-Seq data confirms the increased expression of SERPINE1 in IBD patients that localizes to mesenchymal cell populations. Of note, robust upregulation of SERPINE1 is detected in activated fibroblasts from involved vs. non-involved areas of CD patients. Furthermore, spatial transcriptomics of SAMP ilea reveal a progressive and concomitant increase in both Il33 and Serpine1, with strong clusterization compared to healthy AKR controls, particularly during later time points when fibrosis is evident. Currently, results are pending on the treatment of SAMP mice with the PAI-1 inhibitor, MDI-2268, to determine its direct effect(s) on the development of inflammation-associated intestinal fibrosis. Taken together, these findings suggest IL-33-dependent regulation of Serpine1/PAI-1 that promotes intestinal fibrosis, commonly observed in CD patients, and may provide a novel target to treat IBD patients with fibrostenoic disease.

Abstract Background Advances in single-cell technologies enable the unbiased study of cellular heterogeneity. Recently, single-cell RNA sequencing (scRNA-seq) has been utilised on intestinal and blood samples from patients with inflammatory bowel disease (IBD) and healthy individuals, often in conjunction with cell surface proteome and TCR repertoire analyses. These individual studies revealed novel cell subpopulations of immune, mesenchymal, and epithelial cells in UC and CD, but are not sufficiently powered to consistently identify granular cell subsets or establish their association with disease status and response to treatment. Integration of these studies into a unified IBD single-cell atlas would provide a more robust data foundation for therapeutic target discovery. Here, we constructed a comprehensive, multi-modal single-cell atlas of human IBD tissue, combining a large breadth of meta-analysis with the depth of single-cell resolution. Methods Raw data from 20 public datasets were curated and reprocessed to generate a harmonised data foundation to support downstream discovery efforts. Low-quality cells and doublets were removed using thresholds for gene number, gene count, and percentage of counts originating from mitochondrial genes, and the resulting data were normalised and adjusted for batch effects. Published clinical metadata from each study were re-annotated with controlled vocabularies. Associations between clinical metadata and cellular/molecular features were discovered using computational and ML approaches. Results We generated a large-scale integrated single-cell atlas for IBD comprising >500 tissue samples from >200 IBD patients and relevant controls, with harmonised clinical metadata including treatment history and response. This tissue atlas comprises >990k high-quality cells with granular annotations of 129 cell types/states. IBD inflammation and non-response to anti-TNF treatment were associated with unique transcriptional signatures in specific mononuclear phagocyte, CD4 T cell, and fibroblast subpopulations. Further prioritisation and validation of genes comprising these signatures may yield future therapeutic targets for IBD. Conclusion This atlas integrates single-cell data across the largest available collection of IBD patient-derived tissue samples. Leveraging high-resolution cell type annotations and harmonised clinical metadata, meta-analyses of this data foundation will broaden the understanding of IBD biology to identify novel targets and pathways for drug discovery.

Abstract Background Interleukin23 (IL-23) is a cytokine that plays a crucial role in the pathogenesis of inflammatory bowel disease (IBD), making it a highly validated therapeutic target. Understanding the role of IL-23 in IBD at the histopathological level is crucial for determining effective treatment strategies, providing insights into IBD patients who fail to respond to targeted therapies, or predicting those who are likely to lose response. In this context, there is a surge in utilizing artificial intelligence (AI) for histopathological data in IBD and other disease indications. Here, we present an automated computer vision approach to predict IL-23 signalling activity directly from routinely stained Hematoxylin and Eosin (H&E) images Methods A total of 1502 samples with matched clinical data and H&E biopsy images were included from 991 Crohn’s disease (CD) and 511 ulcerative colitis (UC) samples. IL-23 signalling activity was calculated using gene set variation analysis on RNA-seq data collected from the same tissue biopsies. The data were obtained from the IBD Plexus program of the Crohn’s & Colitis Foundation. The proposed approach is based on vision transformers (ViTs) which is a type of deep learning model. ViTs divide the input image into fixed-size patches, transform it into linear embedding, and analyze it with the self-attention mechanism. This enables ViTs to incorporate relationships between different patches of the input image to identify regions predictive of IL-23. Our approach was trained in a weakly supervised manner to automatically identify tissue regions that correlate with IL-23 signaling activity. The model produces interpretable heatmaps to interrogate model predictions and allow clinicians to visualize and interpret the significance of different tissue regions predictive of IL-23 Results We performed 5-fold cross-validation on the splits obtained at the patient level, retaining the data distribution of IL-23 signaling activity, biopsy location, and diagnosis. We separately validated the performance of the proposed model on both disease categories, including CD and UC. The proposed approach achieved an area under the curve (AUC) of 0.82 ± 0.04 on unseen data from CD and an AUC of 0.80 ± 0.02 for UC. The 5-fold results for both disease categories are shown below Conclusion The presented results highlight the significance of computational pathology algorithms to identify IL-23 signalling activity from H&E images. Pathological interpretation from the heatmaps may help understand disease pathomechanism and optimize the treatment options for IBD patients by timely identification of IL-23 status. We are further validating the clinical utility of such heatmaps and expanding the use of H&E to predict other patient-centric endpoints

Crohn’s disease (CD) is a complex inflammatory disorder of incompletely understood molecular aetiology. We generated a large single-cell RNA sequencing dataset from the terminal ileal biopsies of two independent cohorts comprising a total of 50 CD patients and 71 healthy controls. We performed transcriptomic analyses to reveal genes, cell types and mechanisms perturbed in CD, leveraging the power of the two cohorts to confirm our findings and assess replicability. In addition to mapping widespread alterations in cytokine signalling, we provide evidence of pan-epithelial upregulation of MHC class I genes and pathways in CD. Using non-negative matrix factorization we revealed intra- and inter-cellular upregulation of expression programs such as G-protein coupled receptor signalling and interferon signalling, respectively, in CD. We observed an enrichment of CD heritability among marker genes for various activated T cell types and myeloid cells, supporting a causal role for these cell-types in CD aetiology. Comparisons between our discovery and replication cohort revealed significant variation in differential gene-expression replicability across cell types. B, T and myeloid cells showed particularly poor replicability, suggesting caution should be exercised when interpreting unreplicated differential gene-expression result in these cell types. Overall, our results provide a rich resource for identifying cell-type specific biomarkers of Crohn’s disease and identifying genes, cell types and pathways that are causally and replicably associated with disease.

Inflammatory bowel disease (IBD) is an umbrella term for two conditions (Crohn’s Disease and Ulcerative Colitis) that is characterized by chronic inflammation of the gastrointestinal tract. The use of pre-clinical animal models has been invaluable for the understanding of potential disease mechanisms. However, despite promising results of numerous therapeutics in mouse colitis models, many of these therapies did not show clinical benefits in patients with IBD. Single cell RNA-sequencing (scRNA-seq) has recently revolutionized our understanding of complex interactions between the immune system, stromal cells, and epithelial cells by mapping novel cell subpopulations and their remodeling during disease. This technology has not been widely applied to pre-clinical models of IBD. ScRNA-seq profiling of murine models may provide an opportunity to increase the translatability into the clinic, and to choose the most appropriate model to test hypotheses and novel therapeutics. In this review, we have summarized some of the key findings at the single cell transcriptomic level in IBD, how specific signatures have been functionally validated in vivo, and highlighted the similarities and differences between scRNA-seq findings in human IBD and experimental mouse models. In each section of this review, we highlight the importance of utilizing this technology to find the most suitable or translational models of IBD based on the cellular therapeutic target.

The advent of single-cell technologies has revolutionized analyses of IBD-specific processes by identifying important, often novel, mucosal cells subpopulations and their associated functions. We discuss recent findings reporting transcriptomic and cellular diversity of treatment-naïve and treated patients with ileal-specific CD.

Searchable abstracts of presentations at key conferences in endocrinology ISSN 1470-3947 (print) | ISSN 1479-6848 (online)

Innate lymphoid cells (ILCs) are enriched at barrier surfaces, including the gastrointestinal tract. While most studies have focused on the balance between pathogenic group 1 ILCs (ILC1s) and protective ILC3s in maintaining gut homeostasis and during chronic intestinal inflammation, such as Crohn's disease (CD), less is known regarding ILC2s. Using an established murine model of CD-like ileitis, i.e., the SAMP1/YitFc (SAMP) mouse strain, we showed that ILC2s, compared with ILC1s and ILC3s, were increased within draining mesenteric lymph nodes and ilea of SAMP versus AKR (parental control) mice early, during the onset of disease. Gut-derived ILC2s from CD patients versus healthy controls were also increased and expanded, similarly to ILC1s, in greater proportion compared with ILC3s. Importantly, we report that the intracellular bacteria-sensing protein, nucleotide-binding oligomerization domaining-containing protein 2, encoded by Nod2, the first and strongest susceptibility gene identified for CD, promoted ILC2 expansion, which was dramatically reduced in SAMP mice lacking NOD2 and in SAMP mice raised under germ-free conditions. Furthermore, these effects occurred through a mechanism involving the IL-33/ST2 ligand-receptor pair. Collectively, our results indicate a functional link between NOD2 and ILC2s, regulated by the IL-33/ST2 axis, that mechanistically may contribute to early events leading to CD pathogenesis.

Cutaneous group 2 innate lymphoid cells (ILC2) are spatially and epigenetically poised to respond to barrier compromise and associated immunological threats. ILC2, lacking rearranged antigen-specific receptors, are primarily activated by damage-associated cytokines and respond with type 2 cytokine production. To investigate ILC2 potential for direct sensing of skin pathogens and allergens, we performed RNA sequencing of ILC2 derived from in vivo challenged human skin or blood. We detected expression of NOD2 and TLR2 by skin and blood ILC2. Stimulation of ILC2 with TLR2 agonist alone not only induced interleukin-5 (IL-5) and IL-13 expression but also elicited IL-6 expression in combination with Staphylococcus aureus muramyl dipeptide (MDP). Heat-killed skin-resident bacteria provoked an IL-6 profile in ILC2 in vitro that was notably impaired in ILC2 derived from patients with nucleotide-binding oligomerization domain-containing protein 2 (NOD2) mutations. In addition, we show that NOD2 signaling can stimulate autophagy in ILC2, which was also impaired in patients with NOD2 mutations. Here, we have identified a role for ILC2 NOD2 signaling in the differential regulation of ILC2-derived IL-6 and have reported a previously unrecognized pathway of direct ILC2 bacterial sensing.

Colonic antigen-experienced lymphocytes such as tissue-resident memory CD8 + T cells can respond rapidly to repeated antigen exposure. However, their cellular phenotypes and the mechanisms by which they drive immune regulation and inflammation remain unclear. Here we compiled an unbiased atlas of human colonic CD8 + T cells in health and ulcerative colitis (UC) using single-cell transcriptomics with T-cell receptor repertoire analysis and mass cytometry. We reveal extensive heterogeneity in CD8 + T-cell composition, including expanded effector and post-effector terminally differentiated CD8 + T cells. While UC-associated CD8 + effector T cells can trigger tissue destruction and produce tumor necrosis factor (TNF)-α, post-effector cells acquire innate signatures to adopt regulatory functions that may mitigate excessive inflammation. Thus, we identify colonic CD8 + T-cell phenotypes in health and UC, define their clonal relationships and characterize terminally differentiated dysfunctional UC CD8 + T cells expressing IL-26, which attenuate acute colitis in a humanized IL-26 transgenic mouse model.

The intestinal mucosa represents a unique environment where the coordinated function of diverse epithelial, mesenchymal, and immune cells maintains a physiologically balanced environment in the presence of gut microbiota. The intestinal mucosa plays a central role in the pathogenesis of inflammatory bowel disease (IBD), yet the molecular and cellular composition of this diverse environment is poorly understood. However, the recent advent of multimodal single-cell technologies, including single-cell RNA sequencing (scRNA-seq), now provides an opportunity to accurately map the tissue architecture, characterize rare cell types that were previously overlooked, and define function at a single-cell level. In this review, we summarize key advances in single-cell technology and provide an overview of important aspects of computational analysis. We describe emerging data in the field of IBD and discuss how the characterization of novel intestinal mucosa cell populations is reshaping our understanding of this complex disease. We conclude by considering the potential clinical applications, including the definition of novel drug targets and the opportunity for personalization of care in this exciting new era of precision medicine.

Gasdermin D (GSDMD) induces pyroptosis via the pore-forming activity of its N-terminal domain, cleaved by activated caspases associated with the release of IL-1β. Here, we report a nonpyroptotic role of full-length GSDMD in guiding the release of IL-1β-containing small extracellular vesicles (sEVs) from intestinal epithelial cells (IECs). In response to caspase-8 inflammasome activation, GSDMD, chaperoned by Cdc37/Hsp90, recruits the E3 ligase, NEDD4, to catalyze polyubiquitination of pro-IL-1β, serving as a signal for cargo loading into secretory vesicles. GSDMD and IL-1β colocalize with the exosome markers CD63 and ALIX intracellularly, and GSDMD and NEDD4 are required for release of CD63+ sEVs containing IL-1β, GSDMD, NEDD4, and caspase-8. Importantly, increased expression of epithelial-derived GSDMD is observed both in patients with inflammatory bowel disease (IBD) and those with experimental colitis. While GSDMD-dependent release of IL-1β-containing sEVs is detected in cultured colonic explants from colitic mice, GSDMD deficiency substantially attenuates disease severity, implicating GSDMD-mediated release of IL-1β sEVs in the pathogenesis of intestinal inflammation, such as that observed in IBD.