Human Lymph Nodes Maintain TCF-1hi Memory T Cells with High Functional Potential and Clonal Diversity throughout Life

Abstract

Translating studies on T cell function and modulation from mouse models to humans requires extrapolating in vivo results on mouse T cell responses in lymphoid organs (spleen and lymph nodes [LN]) to human peripheral blood T cells. However, our understanding of T cell responses in human lymphoid sites and their relation to peripheral blood remains sparse. In this study, we used a unique human tissue resource to study human T cells in different anatomical compartments within individual donors and identify a subset of memory CD8 T cells in LN, which maintain a distinct differentiation and functional profile compared with memory CD8 T cells in blood, spleen, bone marrow, and lungs. Whole-transcriptome and high-dimensional cytometry by time-of-flight profiling reveals that LN memory CD8 T cells express signatures of quiescence and self-renewal compared with corresponding populations in blood, spleen, bone marrow, and lung. LN memory T cells exhibit a distinct transcriptional signature, including expression of stem cell-associated transcription factors TCF-1 and LEF-1, T follicular helper cell markers CXCR5 and CXCR4, and reduced expression of effector molecules. LN memory T cells display high homology to a subset of mouse CD8 T cells identified in chronic infection models that respond to checkpoint blockade immunotherapy. Functionally, human LN memory T cells exhibit increased proliferation to TCR-mediated stimulation and maintain higher TCR clonal diversity compared with memory T cells from blood and other sites. These findings establish human LN as reservoirs for memory T cells with high capacities for expansion and diverse recognition and important targets for immunotherapies.

Longevity and replenishment of human liver-resident memory T cells and mononuclear phagocytes

“…However the CD8 + T RM within the hepatic hilar lymph nodes lacked the expression of CXCR6 seen on those within the liver (Fig. 4 f); this suggested that either CD8 + T RM capable of migrating from the liver to the draining lymph node down-regulate CXCR6, or those within the lymph node represented an independent population that had developed residence in situ, in line with the identification of CD8 + T cells with a tissue-resident signature in many human lymphoid sites (Buggert et al, 2018;Kumar et al, 2017;Miron et al, 2018). Future studies could use TCR clonotype tracking of donor allograft T cells in draining lymph nodes to distinguish these two scenarios.…”

A single-cell reference map for human blood and tissue T cell activation reveals functional states in health and disease

“…Recent studies in human tissues have revealed that the majority of human T cells are localized in lymphoid mucosal and barrier tissues 8 and that T cell subset composition is a function of the specific tissue site 9,10 . Human TRM cells can be defined based on their phenotypic homology to mouse TRM and are distinguished from circulating T cells in blood and tissues by a core transcriptional and protein signature [10][11][12][13] . However, the role of tissue site in determining T cell functional responses, and a deeper understanding of the relationship between blood and tissue T cells beyond composition differences are key unanswered questions in human immunology.…”

A robust and interpretable, end-to-end deep learning model for cytometry data

“…In this study, we developed and tested a framework for modeling cytometry data using a deep convolutional neural network (CNN), in which multiple hidden layers are used to model the high-dimensional cytometry data. Leveraging multiple large publicly-available CyTOF datasets (472 samples from 9 studies) available in ImmPort [16][17][18][19] , we demonstrate that the deep CNN model is able to diagnose asymptomatic cytomegalovirus infection with high accuracy, even in the presence of strong heterogeneity between datasets. In addition, we developed a permutation-based method to interpret the full deep CNN model.…”

“…To test the performance of the deep CNN model, we applied it to nine CyTOF datasets to train it to diagnose asymptomatic cytomegalovirus (CMV) infection. The dataset spans nine human immunology studies and contains 596 peripheral blood mononuclear cells (PBMC) samples from 313 subjects [16][17][18][19] . We split the nine studies into training, validation, and testing datasets.…”

“…We first queried the ImmPort database to identify samples from healthy individuals with both CyTOF and CMV antibody titer data. The query identified 472 samples from nine studies, including SDY112, SDY113, SDY305, SDY311, SDY472, SDY472, SDY515, and SDY519 as of March 2019 [16][17][18][19] . We downloaded CyTOF data and transformed the raw cytometry signal using arcsinh transformation (y = arcsinh(x/5)).…”