Comment on 'SARS-CoV-2 suppresses anticoagulant and fibrinolytic gene expression in the lung'

Abstract

Mast et al. analyzed transcriptome data derived from RNA-sequencing (RNA-seq) of COVID-19 patient bronchoalveolar lavage fluid (BALF) samples, as compared to BALF RNA-seq samples from a study investigating microbiome and inflammatory interactions in obese and asthmatic adults (Mast et al., 2021). Based on their analysis of these data, Mast et al. concluded that mRNA expression of key regulators of the extrinsic coagulation cascade and fibrinolysis were significantly reduced in COVID-19 patients. Notably, they reported that the expression of the extrinsic coagulation cascade master regulator Tissue Factor (F3) remained unchanged, while there was an 8-fold upregulation of its cognate inhibitor Tissue Factor Pathway Inhibitor (TFPI). From this they conclude that "pulmonary fibrin deposition does not stem from enhanced local [tissue factor] production and that counterintuitively, COVID-19 may dampen [tissue factor]-dependent mechanisms in the lungs". They also reported decreased Activated Protein C (aPC) mediated anticoagulant activity and major increases in fibrinogen expression and other key regulators of clot formation. Many of these results are contradictory to findings in most of the field, particularly the findings regarding extrinsic coagulation cascade mediated coagulopathies. Here, we present a complete re-analysis of the data sets analyzed by Mast et al. This re-analysis demonstrates that the two data sets utilized were not comparable between one another, and that the COVID-19 sample set was not suitable for the transcriptomic analysis Mast et al. performed. We also identified other significant flaws in the design of their retrospective analysis, such as poor-quality control and filtering standards. Given the issues with the datasets and analysis, their conclusions are not supported.

Keywords: COVID-19; SARS-CoV-2; coagulation; human; infectious disease; microbiology.

Figure 1.. Graphical representation of health meta-data from control group samples as reported in Garvin et al.

(Supplementary file 1 – “Patient Meta-data”).

Figure 2.. Box and Whisker plot of the percent of rRNA reads.

Control group and SARS-CoV-2+ fastq files were accessed from public repositories and aligned with CLC genomics workbench as described in Garvin et al. Count data for known RefSeq rRNA features were sorted from CLC generated count tables and summed per sample. rRNA count sums were then divided by the total counts per sample to generate percentages.

Figure 3.. Histograms of the total number of reads per sample after adapter and quality trimming was performed with TrimGalore.

This graphic was generated by analyzing trimmed fastq files using the package FastQC, and then processing the FastQC output into the package MultiQC. MultiQC was used to generate the histogram.