Despite their crucial role in health and disease, our knowledge of immune cells within human tissues remains limited. We surveyed the immune compartment of 16 tissues from 12 adult donors by single-cell RNA sequencing and VDJ sequencing generating a dataset of ~360,000 cells. To systematically resolve immune cell heterogeneity across tissues, we developed CellTypist, a machine learning tool for rapid and precise cell type annotation. Using this approach, combined with detailed curation, we determined the tissue distribution of finely phenotyped immune cell types, revealing hitherto unappreciated tissue-specific features and clonal architecture of T and B cells. Our multitissue approach lays the foundation for identifying highly resolved immune cell types by leveraging a common reference dataset, tissue-integrated expression analysis, and antigen receptor sequencing. Description Immune cell diversity in the human body The human immune system is composed of many different cell types spread across the entire body, but little is currently known about the fine-grained variations in these cell types across organs. Using single-cell genomics, Domínguez Conde et al. examined the gene expression profile of more than 300,000 individual immune cells extracted from 16 different tissues in 12 deceased adult organ donors (see the Perspective by Liu and Zhang). Cell identity was assigned using CellTypist, an automated cell classification tool designed by the authors. In-depth data analysis revealed insights into how the immune system adapts to function effectively in different organ contexts. —LZ and DJ An immune cell atlas of human innate and adaptive immune cells across lymphoid, mucosal, and exocrine sites reveals tissue-specific compositions and features. INTRODUCTION Immune cells that seed peripheral tissues play a vital role in health and disease, yet most studies of human immunity focus on blood-derived cells. Immune cells adapt to local microenvironments, acquiring distinct features and functional specialization. Dissecting these molecular adaptations through the systematic assessment of cells across the human body promises to transform our understanding of the immune system at the organismal level. RATIONALE To comprehensively assess immune cell types, we collected donor-matched tissues from 12 deceased organ donors. We isolated immune cells and performed single-cell RNA sequencing and paired VDJ sequencing for T cell and B cell receptors, resulting in high-quality data for ~330,000 immune cells. To resolve cell identities, we developed CellTypist, a logistic regression–based framework using stochastic gradient descent learning. This cross-tissue annotation enabled interrogation of shared and tissue-specific expression modules and cell states within myeloid and lymphoid cell lineages. RESULTS We developed CellTypist by curating and harmonizing public datasets to assemble a comprehensive immune cell type reference database (https://www.celltypist.org). CellTypist was then applied to our data, generated across multiple tissues and individuals (https://www.tissueimmunecellatlas.org/). Altogether, we detected 101 immune populations and performed extensive cross-tissue comparisons for each subset. Although macrophages displayed prominent tissue-restricted features, some convergent features were also detected. For example, macrophages expressing erythrophagocytosis-related genes were widely found in spleen, liver, bone marrow, and lymph nodes. Heterogeneity within defined subpopulations was also observed, such as migratory dendritic cell adaptations. Within adaptive immune lineages, we identified tissue-specific distributions of memory populations. Plasma cells showed a restricted tissue distribution, whereas memory B cells were more widely distributed. Similarly, tissue-resident memory T (TRM) cells were more restricted in distribution compared with central and effector memory T cells. Notably, TRM cells harbored significant diversity, including αβ and γδ lineages, ascertained by VDJ sequencing. Assessment of clonal dynamics revealed the highest clonal expansions in TRM cells and the most frequent clonal sharing between resident and effector memory populations. CONCLUSION Here we present an immune cell atlas of myeloid and lymphoid lineages across adult human tissues. We developed CellTypist for automated immune cell annotation and performed an in-depth dissection of cell populations, identifying 101 cell types or states from more than one million cells, including previously underappreciated cell states. We also uncovered convergent phenotypes across tissues within given lineages and described tissue adaptation signatures for a number of cell types, including macrophages and resident memory T cells. Together, we have extended our understanding of how human immunity functions as an integrated, cross-tissue network, and we provide the scientific community with several key new resources. Cross-tissue atlas of human immune cells. Donor-matched tissues from 12 deceased organ donors were profiled using single-cell RNA sequencing [including single-cell VDJ sequencing (scVDJ-seq)]. Cell type annotation was achieved with CellTypist, an automated cell type annotation pipeline, followed by manual curation. In-depth analysis of transcriptional features revealed convergent and divergent gene expression programs for each cell lineage across lymphoid and nonlymphoid tissues. scVDJ-seq enabled repertoire analysis across cell subsets and tissues. scRNA-seq, single-cell RNA sequencing; TCR, T cell receptor; BCR, B cell receptor.

https://www.science.org/doi/pdf/10.1126/science.abl5197?download=true

Cross-tissue immune cell analysis reveals tissue-specific features in humans

While it is now widely accepted that host inflammatory responses contribute to lung injury, the pathways that drive severity and distinguish coronavirus disease 2019 (COVID-19) from other viral lung diseases remain poorly characterized. We analyzed plasma samples from 471 hospitalized patients recruited through the prospective multicenter ISARIC4C study and 39 outpatients with mild disease, enabling extensive characterization of responses across a full spectrum of COVID-19 severity. Progressive elevation of levels of numerous inflammatory cytokines and chemokines (including IL-6, CXCL10, and GM-CSF) were associated with severity and accompanied by elevated markers of endothelial injury and thrombosis. Principal component and network analyses demonstrated central roles for IL-6 and GM-CSF in COVID-19 pathogenesis. Comparing these profiles to archived samples from patients with fatal influenza, IL-6 was equally elevated in both conditions whereas GM-CSF was prominent only in COVID-19. These findings further identify the key inflammatory, thrombotic, and vascular factors that characterize and distinguish severe and fatal COVID-19.

Inflammatory profiles across the spectrum of disease reveal a distinct role for GM-CSF in severe COVID-19.

Post-acute sequelae of COVID-19 (PASC) is emerging as global problem with unknown molecular drivers. Using a digital epidemiology approach, we recruited 8,077 individuals to the cohort study for digital health research in Germany (DigiHero) to respond to a basic questionnaire followed by a PASC-focused survey and blood sampling. We report the first 318 participants, the majority thereof after mild infections. Of those, 67.8% report PASC, predominantly consisting of fatigue, dyspnea, and concentration deficit, which persists in 60% over the mean 8-month follow-up period and resolves independently of post-infection vaccination. PASC is not associated with autoantibodies, but with elevated IL-1β, IL-6, and TNF plasma levels, which we confirm in a validation cohort with 333 additional participants and a longer time from infection of 10 months. Blood profiling and single-cell data from early infection suggest the induction of these cytokines in COVID-19 lung pro-inflammatory macrophages creating a self-sustaining feedback loop. 

http://www.cell.com/article/S2666379122001951/pdf

The IL-1β, IL-6, and TNF cytokine triad is associated with post-acute sequelae of COVID-19

Although critical for host defense, innate immune cells are also pathologic drivers of acute respiratory distress syndrome (ARDS). Innate immune dynamics during Coronavirus Disease 2019 (COVID-19) ARDS, compared to ARDS from other respiratory pathogens, is unclear. Moreover, mechanisms underlying the beneficial effects of dexamethasone during severe COVID-19 remain elusive. Using single-cell RNA sequencing and plasma proteomics, we discovered that, compared to bacterial ARDS, COVID-19 was associated with expansion of distinct neutrophil states characterized by interferon (IFN) and prostaglandin signaling. Dexamethasone during severe COVID-19 affected circulating neutrophils, altered IFNactive neutrophils, downregulated interferon-stimulated genes and activated IL-1R2+ neutrophils. Dexamethasone also expanded immunosuppressive immature neutrophils and remodeled cellular interactions by changing neutrophils from information receivers into information providers. Male patients had higher proportions of IFNactive neutrophils and preferential steroid-induced immature neutrophil expansion, potentially affecting outcomes. Our single-cell atlas (see 'Data availability' section) defines COVID-19-enriched neutrophil states and molecular mechanisms of dexamethasone action to develop targeted immunotherapies for severe COVID-19.

Dexamethasone modulates immature neutrophils and interferon programming in severe COVID-19.

Distinct immunological signatures discriminate severe COVID-19 from non-SARS-CoV-2-driven critical pneumonia.

Hyperinflammation contributes to lung injury and subsequent acute respiratory distress syndrome (ARDS) with high mortality in patients with severe coronavirus disease 2019 (COVID-19). To understand the underlying mechanisms involved in lung pathology, we investigated the role of the lung-specific immune response. We profiled immune cells in bronchoalveolar lavage fluid and blood collected from COVID-19 patients with severe disease and bacterial pneumonia patients not associated with viral infection. By tracking T cell clones across tissues, we identified clonally expanded tissue-resident memory-like Th17 cells (Trm17 cells) in the lungs even after viral clearance. These Trm17 cells were characterized by a a potentially pathogenic cytokine expression profile of IL17A and CSF2 (GM-CSF). Interactome analysis suggests that Trm17 cells can interact with lung macrophages and cytotoxic CD8+ T cells, which have been associated with disease severity and lung damage. High IL-17A and GM-CSF protein levels in the serum of COVID-19 patients were associated with a more severe clinical course. Collectively, our study suggests that pulmonary Trm17 cells are one potential orchestrator of the hyperinflammation in severe COVID-19.

Clonal expansion and activation of tissue-resident memory-like Th17 cells expressing GM-CSF in the lungs of severe COVID-19 patients.

Background Repetitive elements contribute a large part of eukaryotic genomes. For example, about 40 to 50% of human, mouse and rat genomes are repetitive. So identifying and classifying repeats is an important step in genome annotation. This annotation step is traditionally performed using alignment based methods, either in a de novo approach or by aligning the genome sequence to a species specific set of repetitive sequences. Recently, Li (Bioinformatics 35:4408-4410, 2019) developed a novel software tool dna-brnn to annotate repetitive sequences using a recurrent neural network trained on sample annotations of repetitive elements. Results We have developed the methods of dna-brnn further and engineered a new software tool DeepGRP. This combines the basic concepts of Li (Bioinformatics 35:4408-4410, 2019) with current techniques developed for neural machine translation, the attention mechanism, for the task of nucleotide-level annotation of repetitive elements. An evaluation on the human genome shows a 20% improvement of the Matthews correlation coefficient for the predictions delivered by DeepGRP, when compared to dna-brnn. DeepGRP predicts two additional classes of repeats (compared to dna-brnn) and is able to transfer repeat annotations, using RepeatMasker-based training data to a different species (mouse). Additionally, we could show that DeepGRP predicts repeats annotated in the Dfam database, but not annotated by RepeatMasker. DeepGRP is highly scalable due to its implementation in the TensorFlow framework. For example, the GPU-accelerated version of DeepGRP is approx. 1.8 times faster than dna-brnn, approx. 8.6 times faster than RepeatMasker and over 100 times faster than HMMER searching for models of the Dfam database. Conclusions By incorporating methods from neural machine translation, DeepGRP achieves a consistent improvement of the quality of the predictions compared to dna-brnn. Improved running times are obtained by employing TensorFlow as implementation framework and the use of GPUs. By incorporating two additional classes of repeats, DeepGRP provides more complete annotations, which were evaluated against three state-of-the-art tools for repeat annotation.

/pdf/deepgrp-engineering-a-software-tool-for-predicting-genomic-1dt69fay0x.pdf

Fabian Hausmann

Papers

Clonal expansion and activation of tissue-resident memory-like Th17 cells expressing GM-CSF in the lungs of severe COVID-19 patients.

DeepGRP: engineering a software tool for predicting genomic repetitive elements using Recurrent Neural Networks with attention.