Cytotoxic, natural killer-like ex-tissue resident memory T-cells circulate in human chronic graft-versus-host disease at diagnosis

Abstract Emerging evidence implicates tissue resident memory T-cells (TRM) in chronic graft-versus host disease (cGVHD) immunopathology. While traditionally considered confined to tissues, recent studies indicate TRM can re-enter the circulation as “ex-TRM” in inflammatory conditions. However, the role of ex-TRM in cGVHD, and the link between peripheral blood (PB) and tissue-based immunopathology in cGVHD are not well understood. To identify and characterize ex-TRM in cGVHD, we utilized 10X Genomics 5' GEM-X technology to perform single-cell RNA sequencing (scRNA-Seq) and single-cell TCR sequencing (scTCR-Seq) on T-cell selected PBMC samples from patients with newly diagnosed, treatment-naive cGVHD (n=8) and post-allogeneic stem cell transplant (ASCT) matched controls (MC; n=5) who did not develop relapse or acute/chronic GVHD. Quality control (QC), normalization, clustering, principal component analysis, dimensionality reduction, and integration were performed with Seurat v5.2.1 in R v4.4.2. CD8+ effector memory (EM) subsets were re-clustered to enhance resolution for ex-TRM. TCR clonality was assessed in ScRepertoire, and antigen specificity was determined for alpha/beta TCR amino acid sequences using ImmuneWatch DETECT. Differences in continuous variables were assessed using the Wilcoxon rank-sum test, and Bonferroni correction was applied for differential gene expression (DGE) analysis. Significance was set at p < 0.05 All patients received matched donor transplants and tacrolimus and methotrexate for GVHD prophylaxis. There were no significant differences between cGVHD and MC for age, sex, donor type (related vs. unrelated), CMV serostatus, conditioning regimen, or sample timepoint post-ASCT. After QC, 29,107 CD8+ T-cells (17,938 cGVHD and 11,1169 MC) were analyzed. Re-clustering of CD8+ EM cells revealed a distinct cluster expressing canonical TRM markers (CXCR6, ITGA1, ITGAE, CD69) as well as TRM-associated genes CXCL13 and CRTAM, consistent with ex-TRM. To further validate the TRM-like signature, we performed cluster-based module scoring using: 1) the top 100 upregulated genes from our recent publication in TRM vs. non-TRM in explanted lung tissue from pulmonary cGVHD and 2) the top 200 upregulated genes from an external dataset comparing lung TRM to circulating EM T-cells in healthy controls. Module scores for both gene sets were highest in the ex-TRM cluster, confirming TRM-like identity. The ex-TRM cluster included all cGVHD patients and MC in similar proportions. We then compared abundance and gene expression of ex-TRM in cGVHD patients and MC. Ex-TRM as a fraction of CD8+ EM was similar between groups (0.08 vs 0.09). However, cGVHD ex-TRM showed upregulation of cytotoxicity genes (GZMB, GNLY, PRF1), NK-like (NKL) markers (KLRD1, FGFBP2, NKG7), and T-cell exhaustion (Tex)-associated genes (DUSP4, HAVCR2). The median module score for an external Tex gene signature was higher in cGVHD than MC (0.029 vs 0.016, p <0.001). DGE results were consistent after stratification by CMV serostatus. TCR analysis showed similar proportions of hyperexpanded (>100 cells/clonotype) and large (20-100 cells/clonotype) clones in ex-TRM between conditions (30% versus 28%). Normalized entropy scores for ex-TRM were identical (0.92), indicating comparable repertoire diversity. However, hyperexpanded and large clones in cGVHD exhibited even greater upregulation of cytotoxicity, NKL, and Tex genes by log2 fold change than the overall ex-TRM population. Finally, ImmuneWatch DETECT did not find viral epitope-specific clonotypes in ex-TRM from cGVHD patients, suggesting that expansion may reflect alloantigen recognition. In conclusion, we identified circulating ex-TRM during post-ASCT immune reconstitution using canonical TRM markers and module scoring. In cGVHD, ex-TRM had a distinct cytotoxic, NKL, and Tex gene signature, supporting possible antecedent tissue antigen exposure and pathogenicity. Future work will explore protein level validation and assess phenotypic and clonal overlap between ex-TRM and bona fide TRM in affected tissues. With further validation, ex-TRM may provide a surrogate for tissue-resident populations and the foundation for non-invasive biomarkers in cGVHD.