Spatial transcriptomic profiling of coronary endothelial cells in SARS-CoV-2 myocarditis
Description
Objectives: Our objective was to examine coronary endothelial and myocardial programming in patients with severe COVID-19 utilizing digital spatial transcriptomics. Background: Severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) has well-established links to thrombotic and cardiovascular events. Endothelial cell infection was initially proposed to initiate vascular events; however, this paradigm has sparked growing controversy. The significance of myocardial infection also remains unclear. Methods: Autopsy-derived cardiac tissue from control (n = 4) and COVID-19 (n = 8) patients underwent spatial transcriptomic profiling to assess differential expression patterns in myocardial and coronary vascular tissue. Our approach enabled transcriptional profiling in situ with preserved anatomy and unaltered local SARS-CoV-2 expression. In so doing, we examined the paracrine effect of SARS-CoV-2 infection in cardiac tissue. Results: We observed heterogeneous myocardial infection that tended to colocalize with CD31 positive cells within coronary capillaries. Despite these differences, COVID-19 patients displayed a uniform and unique myocardial transcriptional profile independent of local viral burden. Segmentation of tissues directly infected with SARS-CoV-2 showed unique, pro-inflammatory expression profiles including upregulated mediators of viral antigen presentation and immune regulation. Infected cell types appeared to primarily be capillary endothelial cells as differentially expressed genes included endothelial cell markers. However, there was limited differential expression within the endothelium of larger coronary vessels. Conclusions: Our results highlight altered myocardial programming during severe COVID-19 that may in part be associated with capillary endothelial cells. However, similar patterns were not observed in larger vessels, diminishing endotheliitis and endothelial activation as key drivers of cardiovascular events during COVID-19. Dataset Description: Tab 1: RNA Sequencing results from all myocardial samples organized by patient COVID19 status. These results are independent of spatial location. Tab 2: RNA Sequencing results from all coronary endothelial samples organized by patient COVID19 status. These results are spatially-resolved to include only the endothelium. Tab 3: RNA Sequencing results from myocardial samples organized by SARS-CoV-2 nucleocapsid expression (Positive vs. Negative immunostain). These results are spatially-resolved and segmented to highlight sequencing from SARS-CoV-2 infected cells.
Files
Steps to reproduce
Paraffin embedded tissues were processed and analyzed locally using a combination of fluorescently labeled anti-SARS-CoV-2 nucleocapsid (GeneTex, GTX135361; 1:500) and anti-CD31 (Abcam, ab9498, 1:200) antibodies. Fluorescent antibodies were combined with the GeoMX Cancer Transcriptome (Nanostring) and COVID-19 Immune Response Atlas gene sets with custom probes specific for SARS-CoV-2 lung infection and tissue responses (see Table S3 for SARS-CoV-2 related gene list), totaling 1860 genes. Selection of regions of interests (ROIs) was performed based on 1) immunofluorescent viral staining, 2) cellular immunofluorescent profile and anatomic features consistent with coronary vascular tissue, and 3) cellular anatomic features consistent with myocardial tissue observed in the hematoxylin and eosin (H&E) stained sections. To ensure even and representative selection of ROIs, general myocardial regions were selected evenly across the entire tissue section. Endothelial ROIs were sampled whenever possible due to the heterogeneity of coronary arterial and venous vessels. Quantification. Libraries of the oligo tags collected with the Nanostring GeoMx platform were prepared using the GeoMx Seq code plates and reagents (Nanostring Technologies) with unique I7 and I5 indices. Paired sequencing was performed on the NovaSeq 6000 (Flow cell S100, Medgenome) of the oligo tags and not on the transcript itself, providing a more accurate transcript count with less sequencing bias. FASTQ files were converted into DCC files using the online Nanostring pipeline available on Basespace (Illumina). RNA probe counts used in the analyses were selected following a sequencing QC according to Nanostring protocols. Specifically, counts from each identified region of interest are analyzed, and under-sequenced samples are dropped (field of view percentage of 75% and Binding density from 0.1 to 2.25). A QC probe was also used where mRNAs are targeted by multiple probes and outlier probes are dropped from downstream data analysis (positive spike-in normalization factory between 0.3 and 3) 1. Then RNA counts were normalized using a signal-based normalization, in which individual counts are normalized against the 75th percentile of signal from their own region of interest. The final list of detectable genes was then obtained by dropping genes in each specific group (i.e. endothelial, COVID status, SARS-CoV-2 nucleocapsid expression) by using a limit of quantification (LOQ) of 20% coverage within replicates. The LOQ was calculated using the geometric mean and geometric standard deviation of negative probes in the dataset. Counts were normalized to log2, and statistical comparisons were performed using a mixed linear model with Benjamini-Hochberg correction to account for false discovery rate1, 2.