Tissue factor links inflammation, thrombosis, and senescence in COVID-19

Senescence is a common biological response to COVID-19 infection

VIS has been proposed to trigger the hyperactive inflammatory response in a subset of severe COVID-19 patients¹². To confirm the above hypothesis in a broad range of COVID-19 cases, we analyzed the single-cell profiles obtained from multiple independent datasets. First, we analyzed the single-cell RNA-seq data of the bronchoalveolar lavage fluid (BALF) cells derived from 12 individuals; 3 mild and 6 severe COVID-positive, and 3 healthy control individuals (GSE145926)¹⁷. Three out of 9 COVID-positive patients had chronic diseases and two patients died in the hospital. All infected patients were treated with antiviral and anti-inflammatory interventions¹⁷.

To determine what cell types were present in the patient samples, we used representative cell type-specific markers to profile the BALF cell populations (Fig. 1a). Most BALF cells from healthy control individuals were macrophages (92%). In contrast, COVID-19-infected patients had more diverse cell types within their BALF cell populations, with a notable increase in epithelial cells (about 6%) and Natural killer (NK)/T cells (about 17%) while macrophages are about 60 to 70%. First, we surveyed the status of cellular pathways that were significantly enriched in the BALF cells from COVID-positive patients in comparison to those from healthy controls by performing Gene Set Enrichment Analysis (GSEA)^18,19. The top upregulated pathways in the BALF from COVID-19-infected patients include SARS-CoV-2 infection, cytokine response (GOBP Response to cytokine, KEGG Cytokine-cytokine receptor interaction), thrombosis (HP arterial thrombosis, GOBP Positive regulation of coagulation, GOBP coagulation), and senescence (Fridman senescence UP, Reactome cellular senescence) (Fig. 1b, Table 1). In addition, chemokine signaling and NF-kappa B signaling pathways, two pathways directly linked to senescence, are among the highly enriched pathways in COVID patient-derived BALF cell populations¹⁴ (Table 1).

Senescence is a common response to COVID-19 infection. (A) UMAP projection of the major cell types in bronchoalveolar lavage fluid (BALF) cell population from healthy controls and COVID-19 patients (Control, n = 3; Mild, n = 3; Severe, n = 6). Each point represents a single cell. Major cell types were marked with different colors and quantitated per individual samples. (B) Representative GSEA plots of Reactome Cellular Senescence and Fridman Senescence UP gene sets. Positive normalized enrichment score (NES) indicates enrichment compared to the controls. (C) UMAP representation of BALF cell population from each COVID-19 patient. Each gray point represents a single cell. Age and sex were listed next to sample name. SASP factor expression levels using the SASP Schleigh gene list³⁹ are marked with a blue color gradient. Prominent cell types are outlined and labeled. E = Epithelial cells and M = Macrophage. (D) Heat map plots of the SASP gene sets in BALF cells populations (Severe COVID patients: 41,323, Mild COVID patients: 8613, healthy control: 23,486 cells). (E) Heat map plots of the SASP gene sets in BALF epithelial cells (Severe: 2705, Mild: 551, healthy control: 72 cells). Note that F3 is not on the SASP SCHL list but added for comparison (boxed in green).

To investigate which cell types were responsible for the upregulation of the senescence pathway, we created UMAP single-cell profiles for each COVID-19-infected patient and clustered the single cells based on their cell types. Then, we inferred the level of SASP factor expression in each cell by color gradient (Fig. 1c). Moderate levels of SASP factor expression were noted in macrophage populations from both control and COVID patients. Notably, BALF epithelial cells of severe patients show a particularly strong SASP upregulation in comparison to healthy controls. These findings are consistent with the fact that SARS-CoV-2 virus infects lung epithelial cells via ACE2 receptor³.

We then aimed to quantitate which SASP factors were upregulated in the BALF cell population from COVID patients. Only a subset of genes can be reliably quantitated through single-cell RNA sequencing data. To account for this gene dropout, we converted single-cell profiles to pseudo-bulk profiles in which gene expression profiles of the clustered single-cell subpopulations are pooled. Using this method, we created three pseudo-bulk RNA profiles of total BALF cells derived from healthy control, mild COVID patients, and severe COVID patients. Heatmap plot analysis indicates that about 80% of representative SASP factor genes were upregulated in response to COVID-19 infection, with further upregulation in severe COVID patients (Fig. 1d). Next, we focused on the epithelial cells from the total BALF population (Fig. 1e). Our heatmap analysis showed that an even larger proportion of SASP signature genes were strongly upregulated in BALF epithelial cells from severe COVID patients (Fig. 1e). These findings demonstrate a strong positive correlation between the senescence phenotype and COVID-19 infection.

F3 expression in COVID-19

Thrombosis and hyper-coagulation are prominent responses to SARS-CoV-2 infection^8,9. GSEA analysis of BALF cell population derived from COVID-19-infected patients indicates that coagulation pathways are highly enriched in COVID-positive patients (Fig. 2a, Table 1). Tissue factor (F3) is a critical initiator of extrinsic coagulation pathway and is associated with hypercoagulation in various diseases²⁰. We tested whether F3 is upregulated upon COVID-19 infection, using the single-cell RNA profiles of BALF samples. High levels of F3 mRNA expression were found only in COVID-19 patients (Fig. 2b). Furthermore, BALF epithelial cells derived from severe COVID-19 patients showed particularly strong upregulation of F3 (Fig. 2b). To determine the F3 mRNA expression levels in different cell types within the BALF cell populations, we plotted single-cell expression of F3 against their cell types. We found that F3 is highly upregulated in epithelial cells, in addition to macrophages to a lesser degree (Fig. 2c). This is consistent with previous reports that F3 is primarily expressed in epithelial cells, platelets, fibroblasts, and endothelial cells^21,22.

F3 is highly expressed in BALF epithelial cells derived from COVID-19 patients. (A) GSEA plots of GOBP coagulation pathways in BALF cells from COVID patients compared to those from healthy control¹⁸. (B) UMAP representation of F3 expression in control, mild, and severe COVID patients. Each gray dot represents a single cell. F3 expression levels are marked with blue color. Epithelial cells are outlined. (C) Dot plot of F3 levels across the major cell types within BALF cell populations. (D) Plot of the percentage of F3-positive (F3⁺) cells and average F3 expression level (arbitrary unit) in BALF cell populations. Each point represents one patient. (E) % of F3⁺ cells and average F3 expression in the BALF epithelial cell populations. Each point represents one patient. * p value < 0.05.

To determine if this increase in F3 expression is prominent on a patient level, we calculated both the expression levels of F3 and the percentage of F3-positive (F3⁺) cells per patient. The percentages of F3⁺ cells and F3 levels are significantly higher in severe patients, compared to those in healthy control individuals (Fig. 2d). Additional analysis of epithelial BALF cells showed that severe COVID patients had 40 to 70% of F3⁺ BALF epithelial cells, which is significantly higher than the healthy controls or mild COVID patients (Fig. 2e).

To determine if F3 upregulation is a common event upon SARS-CoV-2 infection, we analyzed additional datasets. First, we determined the F3 expression in SARS-CoV-2 infected Calu3 lung epithelial cells (GSE148729)²³. In vitro infected samples had a significantly higher F3 expression compared to the control, as early as 1-day after infection (data not shown). We also analyzed the data from SARS-CoV-2 infected golden hamster studies. Three lung samples were taken 1, 2, 4, 6, 8, and 14 days post-infection. In the lung tissue of SARS-CoV-2-infected golden hamsters (GSE161200)²⁴, infected lungs showed the highest expression of F3 at the beginning of the infection (data not shown). Taken together, these data support the notion that SARS-CoV-2 infection induces F3 upregulation.

F3 levels correlate with the senescence and thrombosis signatures in COVID-19

The above data implicate a strong correlation between F3 level and the senescence phenotype. Comparison of the SASP and F3 expression levels in single-cell plots showed a positive correlation between the two, especially in the BALF epithelial cells derived from severe COVID patients (Figs. 1 and 2). To further investigate the relationship between F3 levels and the senescence signature in COVID infection, we grouped the cells into F3-positive (F3⁺) and F3-negative (F3^–) subpopulations based on F3 expression and performed GSEA analysis. While the highest-ranked pathway enriched in the F3⁺ BALF epithelial cells from severe COVID patients is the epidermal differentiation signature, other highly enriched pathways include the senescence pathway, inflammatory SASP factors, and coagulation (Fig. 3a). Additional analysis with GSEA and KEGG database showed that F3⁺ BALF cell populations from severe COVID patients have enriched gene signatures for thrombosis and coagulation, as well as tight junction and extra-cellular matrix hallmarks (Fig. 3a, Table 2). Heatmap analysis to determine the levels of representative gene sets including inflammatory cytokines and pro-thrombosis genes further indicates a strong correlation between F3 expression and the upregulation of the senescence and coagulation pathways (Fig. 3b and c).

F3⁺ cell populations from COVID patients harbor the upregulated gene signatures for the senescence and coagulation pathways. (A) representative GSEA plots for reactome epidermal differentiation, senescence, and coagulation pathways in the F3⁺ populations from COVID patients¹⁸. (B) Heat map plot of the SASP factor expression in the F3⁺ and F3⁻ populations from BALF cells (Severe F3⁺: 2100, Severe F3⁻: 39,223, Mild F3⁺: 201, Mild F3⁻: 8411, Control F3⁺: 180, Control F3⁻: 23,306 cells). (C) Heat map plot of the SASP factor expression in the F3⁺ and F3⁻ populations of BALF epithelial cells. (Severe F3⁺: 1454, Severe F3⁻: 1251, Mild F3⁺: 137, Mild F3⁻: 414, Control F3⁺: 10, Control F3⁻: 62 cells).

To confirm the above association further, we conducted a similar analysis using the data from nasopharyngeal swabs (NS) studies²⁵. This study consists of NS samples from 24 individuals (5 control, 8 mild, and 11 severe COVID-19 patients). Patients ranged from 21 to 75 years old, of which 11 were admitted to the Intense care unit (ICU), and 2 died from COVID-19²⁵. We profiled about 130,000 single cells harvested from the NS swabs and clustered them by cell types (Fig. 4a,b). While most of the control NS samples were primarily epithelial cells, NS cell populations from COVID patients contained more diverse cell types, including neutrophils, macrophages, and natural killer/T cells (Fig. 4b). Similar to the BALF data, we found a strong upregulation of the SASP factor expression and F3 expression in the NS samples from COVID patients (Fig. 4c and d), further indicating a possible link between F3, the senescence, and hyper-coagulation phenotype in COVID infection.

F3 and SASP factor genes are highly expressed in nasopharyngeal (NS) cell population from the COVID-19 patients. (A) UMAP projection of the major cell types in NS of COVID-19 patients (control, n = 5; mild, n = 8; severe, n = 11). Cell type composition in NS was quantitated in (D). (B, C) UMAP plot of NS cells derived from healthy control and COVID-19 patients. SASP factor expression level (B) and F3 level (C) of each cell are marked with blue color.

Potential roles of F3 and SASP in other respiratory viral infections

Respiratory syncytial virus (RSV) and Influenza A virus (IAV) bear similarities to SARS-CoV-2. For example, RSV and IAV enter the body through airway epithelial cells and they can induce severe inflammation of the upper and lower respiratory systems^26,27,28. RSV is one of the most common causes of bronchiolitis and pneumonia in infants, but it can cause cold-like symptoms regardless of age²⁶. IAV is a major cause of seasonal flu and it is the only influenza virus known to cause flu pandemics²⁸. To investigate whether the senescence phenotype and F3 upregulation are involved in these viral infection models, we analyzed RSV-infected A549 cell studies (GSE147507)²⁹ and IAV-infected A549 cell studies (GSE118773)^30,31. Upon IAV infection, both the SASP factors and F3 expression were significantly upregulated (Fig. 5a and b). We also found a concurrent induction of F3 and the SASP factors upon RSV infection(Fig. 5c and d), suggesting the possibility that the induction of senescence and F3 can be one of common anti-viral responses in humans.

SASP factors and F3 are upregulated upon RSV and IAV infection. (A) The levels of representative SASP factor expression in IAV infected A549 cells. (B) F3 expression in A549 cells infected with IAV (mock, n = 1; IAV n = 3). *p value < 0.05. **p value < 0.005. (C) Heat map plot of the SASP factor expression in the Mock and RSV-infected A549 cells (n = 3 per group). (D) F3 expression in RSV-infected A549 cell lines. **p value < 0.005.