Multimodal memory T cell profiling identifies a reduction in a polyfunctional Th17 state associated with tuberculosis progression

Abstract

Mycobacterium tuberculosis (M.tb) results in 10 million active tuberculosis (TB) cases and 1.5 million deaths each year 1 , making it the world's leading infectious cause of death 2 . Infection leads to either an asymptomatic latent state or TB disease. Memory T cells have been implicated in TB disease progression, but the specific cell states involved have not yet been delineated because of the limited scope of traditional profiling strategies. Furthermore, immune activation during infection confounds underlying differences in T cell state distributions that influence risk of progression.Here, we used a multimodal single-cell approach to integrate measurements of transcripts and 30 functionally relevant surface proteins to comprehensively define the memory T cell landscape at steady state (i.e., outside of active infection). We profiled 500,000 memory T cells from 259 Peruvians > 4.7 years after they had either latent M.tb infection or active disease and defined 31 distinct memory T cell states, including a CD4+CD26+CD161+CCR6+ effector memory state that was significantly reduced in patients who had developed active TB (OR = 0.80, 95% CI: 0.73-0.87, p = 1.21 x 10 -6 ).This state was also polyfunctional; in ex vivo stimulation, it was enriched for IL-17 and IL-22 production, consistent with a Th17-skewed phenotype, but also had more capacity to produce IFNγ than other CD161+CCR6+ Th17 cells. Additionally, in progressors, IL-17 and IL-22 production in this cell state was significantly lower than in non-progressors.Reduced abundance and function of this state may be an important factor in failure to control M.tb infection. Main textInterindividual immune differences may underlie host variation in response to pathogens, such as M.tb. Only 5-15% of individuals who are infected with M.tb develop TB disease during their lifetime 2 . Disease progression is influenced by host immune and genetic factors that implicate T cells, which are major contributors to defense against intracellular pathogens 3-11 . These markedly different outcomes of infection raise the question of whether steady-state differences in T cell composition underlie divergent host response to M.tb.Studies examining immunophenotypes in active TB have identified numerous memory T cell changes, including in the CD4+ 12 , activated 13 , exhausted 14-16 , Th1 17,18 , and IL-17+ compartments [19][20][21][22][23] . However, these studies typically profiled patients during ongoing infection and focused on antigen-specific T cells, rather than broad, intrinsic differences in memory T cell composition outside of acute infection or active disease.Moreover, it is challenging to acquire an adequate sample size, account for confounders influencing T cell composition 24 , and overcome limitations of surface marker or bulk RNA-seq-based technologies that only capture certain cell state changes.Here, we profiled total memory T cells at single-cell resolution from patients more than four years after TB disease in order to identify broad steady-state differences in progre...

Efficient and precise single-cell reference atlas mapping with Symphony

“…To demonstrate this, we used a CITE-seq dataset that measures the expression of whole-transcriptome mRNA and 30 surface proteins on 500,089 peripheral blood memory T cells from 271 samples 43 . We leveraged both mRNA and protein features to build a multimodal reference from 80% of samples (n=217) and map the remaining 20% of samples (n=54).…”

Multimodal profiling of lung granulomas reveals cellular correlates of tuberculosis control

“…Illustratively, single-cell transcriptomics has been used to identify fundamental alterations in cellular ecosystems associated with the severity and persistence of inflammation (Ordovas-Montanes et al, 2018;Smillie et al, 2019), the cellular bases of disease (Kazer et al, 2020;Montoro et al, 2018) and responses to it, and actionable features of the tumor-immune microenvironment (Hovestadt et al, 2019;Tirosh et al, 2016). While scRNA-seq has been applied to understand peripheral immune or in vitro responses in Mtb infection (Gierahn et al, 2017;Huang et al, 2019;Nathan et al, 2020), it has yet to be leveraged to empower global analyses of cellular responses linked to bacterial control in TB lung granulomas, potentially given challenges associated with tracking, identifying, and isolating these small heterogeneous structures from NHP in a biosafety-level 3 suite.…”

Maximizing statistical power to detect clinically associated cell states with scPOST

“…To demonstrate the utility of scPOST, we applied our framework to three diverse scRNA-seq datasets derived from synovial tissue (rheumatoid arthritis: RA) 10 , peripheral blood memory T cells (Tuberculosis: TB) 14 , or intestine (ulcerative colitis: UC) 15 ( Table S1 ). The RA dataset is smaller than the TB and UC datasets, both in number of cells and number of samples assayed: 5,265 cells (21 samples) versus 496,517 cells (259 samples) and 235,229 cells (30 samples) respectively.…”

“…scPOST currently operates using a low-dimensional PCA embedding of cells. With multimodal technologies such as CITE-seq 5 becoming more available, analyses of these new data types may include dimensionality reduction with alternative methods, such as canonical correlation analysis (CCA) 10,14,34 and nonlinear embeddings 35,36 . Simulating new types of data in the context of these alternative tools, such as simulating canonical variate coordinates instead of PC coordinates, represents a possible extension of scPOST.…”

“…For example, single-cell RNA sequencing (scRNA-seq) 1,2 enables the measurement of the entire transcriptome of individual cells to reveal previously unobserved cell state heterogeneity 3,4 . Technological advances allowing for simultaneous measurement of an increasing diversity of modalities [5][6][7][8] from increasing numbers of cells 9 are enabling disease-focused single-cell studies to identify cell states that correlate with disease in blood or tissue [10][11][12][13][14][15] . Shifts in cell state frequencies (expansions or depletions) are often associated with biologically meaningful phenotypes; expanded populations in disease states are often associated with pathogenic mechanisms, which can be further defined by differential gene expression programs 16 .…”