Sokzini, Peacha

Single-cell studies have defined immune states in inflammatory bowel disease (IBD), but how adaptive receptor histories organize circulating immunity remains unclear. We generated a single-cell transcriptomic atlas of peripheral blood from 249 participants with Crohn's disease, ulcerative colitis, or non-IBD control status, including 182 with productive TCR and BCR recovery. Expanded TCR clonotypes marked inflammatory-memory and cytotoxic states, while distinct but similar paired TCRs shared inflammatory programs across participants. BCR lineage maturation linked IgA-associated mucosal and plasma B cell programs to somatic mutation and class switching, distinguishing maturation-associated biology from clonal expansion. Helper, regulatory, and cytotoxic T-cell programs covaried with B-cell states, and inferred interactions nominated reciprocal antigen-presentation and helper pathways. Repertoire-based machine learning distinguished diagnosis, inflammation, and contemporaneous six-month treatment-response status. Together, this atlas connects receptor architecture to coordinated systemic immune remodeling, establishes a foundation for repertoire-informed patient stratification, and prioritizes candidate mechanisms of IBD pathogenesis.

Abstract Background Inflammatory bowel disease (IBD), including ulcerative colitis (UC) and Crohn’s disease (CD), is characterized by immune dysregulation1,2. Blood-based biomarkers in diagnosing patients with IBD are lacking. We tested the hypothesis that circulating immune cell types and immune repertoire could distinguish patients with IBD from controls. Methods We performed deep immunophenotyping with scRNA-seq with T-cell receptor (TCR) and B-cell receptor (BCR) sequencing on peripheral blood mononuclear cells (PBMCs) from patients with IBD and healthy controls (87 UC, 127 CD, 35 controls). We applied machine learning to quantify cell abundance (miloR3), construct diagnostic classifiers (immuneML4), and infer cell-cell ligand-receptor interactions (MultiNicheNet5) between patients with IBD and controls. Results A scRNA-seq atlas of over 1.1 million PBMCs consisting of 53 distinct cell types was generated (Fig1A). In classifying UC vs controls, support vector machine model had a balanced accuracy of 0.90, consisting of increased IgA and IgG plasma B cells, switched memory B cells, TRegs, and decreased CD16 monocytes, cDC1, non-switched B cells, and MAIT (Fig.1B). In classifying CD vs controls, K-nearest neighbors model had a balanced accuracy of 0.87, consisting of increased IgA plasma B cells, non-switched B cells, CD4 Th22, TReg cytotoxic, CD4 Th1/Th17, and pDC and decreased CD16 monocytes, gdT, CD4 Th17, CD4 Th2, and MAIT cells.Fig.1C). Inflammation was associated with increased IgG plasma B in patients with both UC and CD. Using 4-mer amino acid motif of the BCR complementary-determining region 3 (CDR3) logistic regression model, immuneML had a balanced accuracy of 0.82 (0.75-0.92) in classifying CD vs controls (Fig.1D with models showing top 25 BCR predictive features). Using 4-mer amino acid motif of the BCR CDR3 logistic regression model, immuneML had a balanced accuracy of 0.90 (0.70-1.00) in classifying UC vs controls (Fig.1E with models showing top 25 BCR predictive features). Using 3-mer amino acid motif of the TCR CDR3 logistic regression model, immuneML had a balanced accuracy of 0.76 (0.72-081) in classifying CD vs UC. (Fig.1F with models showing top 25 TCR predictive features). In UC vs controls, MultiNicheNet demonstrated increased signaling in ADAM10-NOTCH2 between TRegs and naïve B cells and IL15-IL2RA between CD14 monocytes and CD4 Th17 cells. In CD vs controls, there was increased signaling in IL15-IL2RA between CD14 monocytes and CD4 Th17 cells and ADAM10-TSPAN5 between CD14 monocytes and TRegs (Fig.1G). Conclusion Applying machine learning models to scRNA-seq and immune repertoire datasets enable blood-based deep immunophenotyping with diagnostic potential and reveals novel insights into cell signaling in IBD. References: 1.Kong L, Pokatayev V, Lefkovith A, Carter GT, Creasey EA, Krishna C, Subramanian S, Kochar B, Ashenberg O, Lau H, Ananthakrishnan AN, Graham DB, Deguine J, Xavier RJ. The landscape of immune dysregulation in Crohn’s disease revealed through single-cell transcriptomic profiling in the ileum and colon. Immunity. 2023 Feb 14;56(2):444-458.e5. doi: 10.1016/j.immuni.2023.01.002. Epub 2023 Jan 30. Erratum in: Immunity. 2023 Dec 12;56(12):2855. doi: 10.1016/j.immuni.2023.10.017. PMID: 36720220; PMCID: PMC9957882. 2.Smillie CS, Biton M, Ordovas-Montanes J, Sullivan KM, Burgin G, Graham DB, Herbst RH, Rogel N, Slyper M, Waldman J, Sud M, Andrews E, Velonias G, Haber AL, Jagadeesh K, Vickovic S, Yao J, Stevens C, Dionne D, Nguyen LT, Villani AC, Hofree M, Creasey EA, Huang H, Rozenblatt-Rosen O, Garber JJ, Khalili H, Desch AN, Daly MJ, Ananthakrishnan AN, Shalek AK, Xavier RJ, Regev A. Intra- and Inter-cellular Rewiring of the Human Colon during Ulcerative Colitis. Cell. 2019 Jul 25;178(3):714-730.e22. doi: 10.1016/j.cell.2019.06.029. PMID: 31348891; PMCID: PMC6662628. 3.Dann E, Henderson NC, Teichmann SA, Morgan MD, Marioni JC. Differential abundance testing on single-cell data using k-nearest neighbor graphs. Nat Biotechnol. 2022 Feb;40(2):245-253. doi: 10.1038/s41587-021-01033-z. Epub 2021 Sep 30. PMID: 34594043; PMCID: PMC7617075. 4.Pavlović M, Scheffer L, Motwani K, Kanduri C, Kompova R, Vazov N, Waagan K, Bernal FLM, Costa AA, Corrie B, Akbar R, Al Hajj GS, Balaban G, Brusko TM, Chernigovskaya M, Christley S, Cowell LG, Frank R, Grytten I, Gundersen S, Haff IH, Hovig E, Hsieh PH, Klambauer G, Kuijjer ML, Lund-Andersen C, Martini A, Minotto T, Pensar J, Rand K, Riccardi E, Robert PA, Rocha A, Slabodkin A, Snapkov I, Sollid LM, Titov D, Weber CR, Widrich M, Yaari G, Greiff V, Sandve GK. The immuneML ecosystem for machine learning analysis of adaptive immune receptor repertoires. Nat Mach Intell. 2021 Nov;3(11):936-944. doi: 10.1038/s42256-021-00413-z. Epub 2021 Nov 16. PMID: 37396030; PMCID: PMC10312379. 5.Browaeys, R., Gilis, J., Sang-Aram, C., De Bleser, P., Hoste, L., Tavernier, S., Lambrechts, D., Seurinck, R. and Saeys, Y., 2023. MultiNicheNet: a flexible framework for differential cell-cell communication analysis from multi-sample multi-condition single-cell transcriptomics data. BioRxiv, pp.2023-06. Conflict of interest: Dr. Gubatan, John Mark: No conflict of interest Jiayu, Ye: No conflict of interest Christin, Lund-Andersen: No conflict of interest Boye, Theresa: No conflict of interest Hoang, Jacqueline: No conflict of interest Sojwal, Raoul: No conflict of interest Fardeen, Touran: No conflict of interest Tran, Tracy: No conflict of interest Sokzini, Peacha: No conflict of interest Koh, Ally: No conflict of interest Gibson, Marielle: No conflict of interest Sinha, Sidhartha: No conflict of interest Rogalla, Stephan: No conflict of interest Rosen, Michael: No conflict of interest Nielsen, Ole Haagen: No conflict of interest Sandve, Geir Kjetil: No conflict of interest