Lung Spatial Profiling Reveals a T Cell Signature in COPD Patients with Fatal SARS-CoV-2 Infection

Abstract

People with pre-existing lung diseases such as chronic obstructive pulmonary disease (COPD) are more likely to get very sick from SARS-CoV-2 disease 2019 (COVID-19). Still, an interrogation of the immune response to COVID-19 infection, spatially throughout the lung structure, is lacking in patients with COPD. For this study, we characterized the immune microenvironment of the lung parenchyma, airways, and vessels of never- and ever-smokers with or without COPD, all of whom died of COVID-19, using spatial transcriptomic and proteomic profiling. The parenchyma, airways, and vessels of COPD patients, compared to control lungs had (1) significant enrichment for lung-resident CD45RO⁺ memory CD4⁺ T cells; (2) downregulation of genes associated with T cell antigen priming and memory T cell differentiation; and (3) higher expression of proteins associated with SARS-CoV-2 entry and primary receptor ubiquitously across the ROIs and in particular the lung parenchyma, despite similar SARS-CoV-2 structural gene expression levels. In conclusion, the lung parenchyma, airways, and vessels of COPD patients have increased T-lymphocytes with a blunted memory CD4 T cell response and a more invasive SARS-CoV-2 infection pattern and may underlie the higher death toll observed with COVID-19.

Keywords: COVID-19; T cells; chronic obstructive pulmonary disease (COPD).

Figure 1

COVID-19 patient characteristics: (A) Survival curve for the days of the clinical course of never-smokers (NS), ever-smokers (ES), and COPD patients infected with severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). The survival analysis was performed using day one of COVID-19 symptoms and day of death in hospital, using a log-rank test. (B,C) The gene expression levels of the envelope (E) (B) and nucleocapsid (N) (C) SARS-CoV-2 genes were quantified in nasal swabs. (D) Thoracic computed tomography (CT) coronal and sagittal views in a representative never smoker (left panel), ever-smoker (middle panel), and patient with COPD (right panel) within four days before death due to SARS-CoV-2 infection.

Figure 2

NanoString GeoMX spatial transcriptomic profiling: (A) Image of a lung tissue section from a COPD subject stained with pan-cytokeratin (green) to identify epithelial cells, CD45 (red) to identify lymphocytes, and SYTO13 (blue) to identify nuclei. Sixteen randomly selected regions of interest (ROIs) were sampled for spatial protein and RNA analysis and are highlighted in white. The higher-magnification inserts show representative airways (green), vessels (red), and parenchyma (blue) structures selected as ROIs. (B–G) Volcano plots showing the genes expressed in the lung parenchyma (B,C), airways (D,E), and vessels (F,G) of patients with COPD who died of COVID-19 versus never-smoker controls (NS, panels B,D,F) and ever-smoker controls (ES, panels C,E,G). The data were assessed using generalized estimating equations (GEEs) accounting for multiple ROIs per case and were adjusted for age. The red dots indicate upregulated genes and the blue dots downregulated genes in COPD patients versus never- and ever-smokers for the specific tissue structures. The horizontal dotted line indicates the significance threshold of FDR < 0.05, and the vertical dashed line indicates the effect size of 0. The gray dots indicate genes that were not significant at FDR < 0.05. The top genes ranked by p-value are labeled with their gene symbols. (H,I) Venn diagram comparing the transcriptomic signatures in the parenchyma, airways, and vessels structures in COPD vs. ever-smoker controls (H) and vs. never-smoker controls (I). The up/down arrows indicate genes that are up/downregulated in the specific structure, and the double-sided arrows indicate markers with the inconsistent direction of change between tissue structures.

Figure 3

ACE2 spatial proteomic expression: (A) ACE2 protein expression levels were quantified in the parenchyma, airway, and vessel ROIs in formalin-fixed paraffin embedded (FFPE) lung tissue sections from patients with COPD versus ever- (ES) and never-smoker (NS) controls by NanoString GeoMX digital spatial profiling. The data were assessed using generalized estimating equations (GEEs) accounting for multiple ROIs per case and were adjusted for age. The box plots represent the median and interquartile range, the error bars are the 5th and 95th percentile, and each dot represents a single ROI. (B–D) ACE2⁺ cells were quantified in the parenchyma (B), airways (C), and vessels (D) in patients with COPD versus ever- (ES) and never-smoker (NS) controls by double immunofluorescence staining for ACE2 (green) and epithelial (pan-cytokeratin, red) or endothelial cells (von Willebrand factor, red). The non-normally distributed data were assessed using Kruskal–Wallis and Mann–Whitney tests, the box plots represent the median and interquartile range, the error bars indicate the 5th and 95th percentile, and each dot represents a single subject. * Indicates p < 0.05, ** indicates p < 0.01, ***: p < 0.001.

Figure 4

Deconvolution of spatial transcriptomics and T cell levels in the blood. (A) The cellular proportions of the immune cell types between tissue structures (parenchyma, airways, and vessels) and disease conditions (NS, ES, and COPD) were inferred using the R package “SpatialDecon”, a deconvolution method developed for the NanoString GeoMX RNA assay. (B) The predicted number of T cells between tissue structures (parenchyma, airways, and vessels) in never- and ever-smokers with and without COPD. (C) Inverse correlation between predicted cell proportion in the lung obtained by cell deconvolution and absolute T cell count in blood on admission. * Indicates p < 0.05, ** indicates p < 0.01.

Figure 5

CD45RO spatial protein expression. (A) CD45RO protein expression levels were quantified in the parenchyma, airway, and vessel ROIs in formalin-fixed paraffin-embedded (FFPE) lung tissue sections from patients with COPD versus ever- (ES) and never-smoker (NS) controls by NanoString GeoMX digital spatial profiling. The data were assessed using generalized estimating equations (GEEs) accounting for multiple ROIs per case and were adjusted for age. The box plots represent the median and interquartile range, the error bars indicate the 5th and 95th percentile, and each dot represents a single ROI. (B–D) CD45RO⁺ cells were quantified in the parenchyma (B), airways (C), and vessels (D) in patients with COPD vs. ever- (ES) and never-smoker (NS) controls by double immunofluorescence staining for CD45RO (green) and epithelial (pan-cytokeratin, red) or endothelial cells (von Willebrand factor, red). The non-normally distributed data were assessed using Kruskal–Wallis and Mann–Whitney tests, the box plots represent the median and interquartile range, the error bars indicate the 5th and 95th percentile, and each dot represents a single subject. *: p < 0.05; **: p < 0.01; ***: p < 0.001.