Abstract
Background/Aim: How tumors regulate the genes of the coagulome is crucial for cancer-associated thrombosis and the occurrence of venous thromboembolic complications in patients with cancer. We have previously reported potent yet complex effects of glucocorticoids (GC) on the expression of three genes that play a key role in the regulation of thrombin/plasmin activation (F3, PLAU, and SERPINE1). This study aimed to extend the investigation of GC effects to the whole tumor coagulome and assess the resulting impact on the ability of cancer cells to activate thrombin and plasmin. Materials and Methods: Cancer RNA expression data were retrieved from various sources. Additionally, oral squamous cell carcinoma (OSCC) cells exposed to GC in vitro were analyzed using QPCR, enzymatic assays measuring thrombin and urokinase-type Plasminogen Activator (uPA) activity, and D-dimer concentrations. Results: Our findings highlight the potent and specific stimulatory effect of GC on SERPINE1 expression across different types of cancer. Consistently, GC were found to inhibit uPA proteolytic activity and reduce the concentrations of D-dimers in OSCC in vitro. Conclusion: Fibrinolysis inhibition is a key consequence of cancer cell exposure to GC, possibly leading to the stabilization of the fibrin clot in cancer.
Venous thromboembolism (VTE) is a major cause of morbidity and mortality in cancer patients (1, 2). Among the mechanisms that contribute to VTE, the establishment of a local hypercoagulable state, typically accounted for by the over-expression of tissue factor (TF) by tumor cells, is thought to be an essential step (1). Importantly, TF is a key regulator of thrombin activation and constitutes an essential trigger for the formation of the fibrin clot (1). How cancer cells regulate their expression of F3, the gene that encodes TF, has been a research question for investigators aiming to decipher the process of cancer-associated thrombosis (CAT) (1). In addition to F3, an array of genes regulates the activation of the proteases involved in coagulation and fibrinolysis. The recent application of systems biology, especially using transcriptomics, has allowed for a more precise, pan-cancer description of the human tumor coagulome, i.e., the extended set of genes with direct or indirect effects on the regulation of coagulation and fibrinolysis (3, 4). This increasingly precise description of the tumor coagulome has unveiled great differences between different tumor types and among individual tumors (4). Some primary tumor types, such as glioblastoma multiforme (GBM) and pancreatic adenocarcinoma (PAAD), are characterized by high expression of F3. Oral squamous cell carcinoma (OSCC), one of the most frequent tumors of the upper aerodigestive tract, have been shown to express the highest levels of F3 and PLAU, coding for the main regulators of coagulation and fibrinolysis TF and urokinase-type plasminogen activator (uPA), respectively (5, 6). The simultaneous high expression of the main regulators of coagulation and fibrinolysis makes OSCC a great model for the study of the coagulome, and in particular to address the biological determinants that shape its composition.
Glucocorticoids (GC) are steroid hormones produced by the adrenal gland, secreted during situations of stress. GC are involved in several biological processes, often related to their ability to interact with the receptor NR3C1, a member of the nuclear receptor superfamily (7, 8). Synthetic GC agonists are often used in the clinic for multiple purposes that include the control of deleterious inflammatory reactions, direct targeting of tumor cells, as well as against the secondary effects induced by some anticancer treatments (9). Natural GC and synthetic agonists of the GC receptor can regulate a large variety of genes (7, 10). In a previous study, we have shown that GC are transcriptional regulators of three key coagulation-related genes in OSCC (11). While F3/TF and PLAU/uPA expression decreased upon cell exposure to dexamethasone, we noted a major direct positive transcriptional effect of dexamethasone on SERPINE1/plasminogen activator inhibitor-1 (PAI-1) expression. From these observations, we concluded that the expression of coagulation-related genes likely involves a dual regulation, with both direct transcriptional regulation and indirect effects possibly explained by the anti-inflammatory action of GC (11). By analyzing RNA-seq data from human tumors in The Cancer Genome Atlas (TCGA), we have suggested the existence of a potential GR-PAI-1 axis linked to increased levels of fibroblasts and endothelial cells in OSCC and PAAD (11). While interesting, our previous study did not address the effect of GC on the coagulome, and it did not assess the functional effects of GC on coagulation and fibrinolysis. In this present work, we further examined the effects of GC by extending our analysis to a set of 85 coagulation-related genes. In addition to providing a broader characterization of the effects of GC on the tumor coagulome, we explored the functional consequences of GC regulation in cancer cells in culture using enzymatic tests measuring coagulation / fibrinolysis and D-dimer concentrations.
Materials and Methods
In silico gene expression analysis. Our analysis was centered on the 85 genes in the Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway “Coagulation and complement cascades” (hsa04610). RNA-seq expression [log2(FPKM+1)] from the NCI-60 was retrieved through CellMiner (12). The GR-activity score was calculated as described previously (11), based on the study by West et al. (13). For each coagulation-related gene, a correlation coefficient was calculated between specific gene expression and the GR-activity score. Gene expression data (RNA SeqV2 data normalized using RNA-Seq by Expectation Maximization: RSEM) were retrieved for 19 cancer types from TCGA, using cBioportal (14). The GR-activity score was calculated using pan-cancer z score values. Data from a study exploring the impact of the glucocorticoid hydrocortisone (2.75 μM, 8 h) on global gene expression in five human non-small cell lung carcinoma (NSCLC) cell lines (A549, H2122, H460, H1975 and H1944) were retrieved from the Gene Expression Omnibus (GSE159546) (15). Fold change in gene expression with hydrocortisone treatment was determined for the 5 cell lines (log2).
Cell culture and reagents. The human OSCC cell line PE/CA-PJ41 was purchased from European Collection of Authenticated Cell Cultures (ECACC, Salisbury, England). Cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS), 2 mM glutamine and antibiotics (streptomycin and penicillin). Human recombinant tumor Necrosis Factor-α (TNFα, ref. 210-TA) was purchased from R&D Systems Biotechne (Noyal Chatillon sur Seiche, France). Dexamethasone (D4902), aldosterone (A9477), GW4064 (G5172), vitamin D3/cholecalciferol (679101), bexarotene (SML0282) and T3 thyroid hormone (T2877) were purchased from Sigma Aldrich (Saint-Quentin-Fallavier, France).
Quantitative PCR. Total RNA was extracted in RLT buffer and purified using RNeasy minikit (74104, Qiagen, Courtaboeuf, France). The samples were reverse transcribed using High-Capacity cDNA Reverse Transcription kit (4368814, ThermoFisher, Villebon-sur-Yvette, France). cDNAs were amplified using gene-specific probes for F3 (Hs01076029_m1), PLAU (Hs01547054_m1), SERPINE1 (Hs00167155_m1) and GAPDH (4333764F) (all from ThermoFisher). We used TaqMan Fast Advanced Master Mix (4444557, ThermoFisher) on a QuantStudioTM 7 Pro Real-Time PCR System (ThermoFisher).
Fibrinolysis and coagulation assays. Thrombin and uPA activities were measured using colorimetric substrates (ab234620 from Abcam (Amsterdam, the Netherlands) and AG4000 from Merck Millipore (Molsheim, France). Assays were conducted using cell culture supernatants from OSCC cells cultured in medium with 0.5% FBS, exposed to TNFα and dexamethasone, as indicated. For the thrombin coagulation assay, we added 10% human plasma (Human pooled plasma, Dominique Dutscher, Bernolsheim, France) immediately before the measurements. The enzymatic activity was assessed using a spectrophotometer at 414 nm (measured every 5 min over 2 h).
D-dimer measurements. D-dimer levels were assessed using the clinically-approved INNOVANCE® D-Dimer test (Siemens, Courbevoie, France). OSCC cells were cultured in DMEM containing 0.5% FBS, exposed to TNFα and dexamethasone as indicated. Coagulation was triggered by adding 10% human plasma 24 h before D-dimer measurements.
Statistical analysis. Student’s t-test was used to compare the different conditions using GraphPad Prism (5.00.288 version, Dotmatics, Boston, MA, USA). Correlation analyses were done using R version 4.3.2 (16). p<0.05 was considered as threshold for significance.
Results
Glucocorticoid receptor activity correlates with the expression of coagulome genes in cancer cells and human tumors. To perform a systems analysis encompassing the key genes of the coagulome, we retrieved RNA-seq data for genes from the KEGG pathway “Coagulation and complement cascades” (hsa04610; 85 coagulation-related genes) from the NCI-60 database, which contains data from 54 solid cancer cell lines (eight types of primary tumors). We calculated the GR-activity score for each cell line and performed a Pearson’s correlation analysis between the GR-activity scores and expression of the 85 coagulation-related genes (Figure 1A). We observed a significant correlation for eight genes (R≥|0.25| and p<0.05): SERPINE1 (r=0.40), SERPING1 (r=0.37), TFPI (r=0.35), CFD (r=0.30), F12 (r=0.30), SERPINE2 (r=−0.38), PROS1 (r=−0.43), A2M (r=−0.56). SERPINE1 was the gene with the highest positive correlation score (r=0.40). We next retrieved data from GSE159546 (15) to evaluate the expression of the 85 coagulation related-genes in NSCLC cells exposed to hydrocortisone (2.75 μM, 8 h). We calculated the fold-change in expression of the 85 coagulation-related genes in the five cell lines treated with hydrocortisone (Figure 1B). Consistent variations in expression were noted for 21 genes (median log2 change >|0.5|) (13 down-regulated and 8 up-regulated). SERPINE1 was among the genes that were up-regulated, confirming our previous observation in NCI-60. We then performed a pan-cancer analysis to correlate the GR-activity score and SERPINE1 expression across different types of human tumors. We used bulk RNA-seq data available for 19 types of tumors in TCGA. A modest positive correlation was observed between GR-activity score and SERPINE1 expression in the pan-cancer analysis (Pearson r=0.24) (Figure 1C). The correlation was however much stronger in tumor types characterized by an overall high expression of SERPINE1, such as pancreatic adenocarcinoma (PAAD) (Pearson r=0.44) (Figure 1D).
Glucocorticoid regulation of coagulome-related genes in human tumors. (A) Pearson correlation between the GR-activity score for 54 solid cancer cell lines from the NCI-60 and their expression of the 85 coagulation-related genes from KEGG. (B) RNAseq data from the Gene Expression Omnibus (GSE159546) were used to evaluate the changes in expression of coagulation-related genes from KEGG (n=85). The log2 change in expression was calculated for five NSCLC cell lines treated with hydrocortisone (2.75 μM, 8 h). The top 21 genes either up-regulated or down-regulated are shown. (C, D) Pan-cancer correlation analysis of the GR-activity score and SERPINE1 expression, using gene expression data from 19 common human cancers from TCGA (C) and PAAD-TCGA (D).
Positive transcriptional regulation of SERPINE1 by GC in OSCC in vitro. In order to experimentally address the regulation of the coagulome by GC and examine its specificity, we tested the effect of five agonists active on other members of the nuclear receptor superfamily. The human OSCC cell line PE/CA-PJ41 was exposed in vitro to dexamethasone, aldosterone (mineralocorticoid receptor agonist), GW4064 (FXR agonist), Vitamin D3 (vitamin D receptor agonist), bexarotene (RXR agonist) and T3 hormone (thyroid hormone receptor agonist) (17) (Figure 2). PE/CA-PJ41 cells were treated with these drugs for 24 h at active concentrations, either alone or in combination with the inflammatory cytokine TNFα (25 ng/ml). Using QPCR, we observed that dexamethasone and other members of the nuclear receptor superfamily had little effect on the expression of F3 (Figure 2). We noted that dexamethasone, and to a lower extent aldosterone, decreased the expression of PLAU mRNA, especially in cells exposed to TNFα. Conversely, dexamethasone sharply increased SERPINE1 expression (an almost 3-fold induction, i.e., greater than all other agonists used). In these experimental conditions, no biologically significant effect on cancer cell viability was detected (data not shown).
The specificity of coagulome regulation by glucocorticoids in oral squamous cell carcinoma. QPCR analysis of the expression of F3, PLAU and SERPINE1 in PE/CA-PJ41. Cells were treated with dexamethasone (10 μM), aldosterone (1 μM), GW4064 (1 μM), Vitamin D3 (100 nM), bexarotene (10 μM) and T3 hormone (1 μM), with or without TNFα (25 ng/ml) for 24 h. *indicates p<0.05 compared to control conditions with Student’s t test.
The effect of dexamethasone on coagulation and fibrinolysis was determined by measuring the activity of thrombin and uPA in the cell supernatants using specific peptide substrates. PE/CA-PJ41 cells were exposed to dexamethasone (1 μM) and TNFα (25 ng/ml) for six days. The supernatants were collected to measure the activity of thrombin and uPA. We found no biologically-significant effect of TNFα or dexamethasone on thrombin activity in the supernatants of OSCC cells (Figure 3A). Conversely, TNFα produced a sharp increase in the enzymatic activity of uPA in the culture supernatants of OSCC cells. Interestingly, dexamethasone abolished the increase in uPA activity induced by TNFα (Figure 3B). In a parallel experiment, we measured the concentrations of D-dimers. We found that GC greatly reduce the appearance of these fibrin degradation products (Figure 3C), further suggesting that GC inhibit fibrinolysis in OSCC cells.
Functional consequences of glucocorticoid treatment of oral squamous cell carcinoma cells on thrombin and plasmin activation. (A, B) Functional assays using thrombin (A) and uPA (B) substrates and culture supernatants of PE/CA-PJ41 cells exposed to dexamethasone (1 μM) +/− TNFα (25 ng/ml) for six days. The enzymatic activity is expressed as arbitrary units. The left panel is a representative experiment. The right panel is a quantification based on three independent experiments. (C) D-dimer levels in PE/CA-PJ41 supernatants exposed to dexamethasone (1 μM) +/− TNFα (25 ng/ml) for three days. Coagulation was triggered with 10% plasma 24 h before D-dimer measurements. *p<0.05 compared to control condition with Student’s t-test.
Discussion
The tumor coagulome is an essential tumor-intrinsic determinant of VTE complications. While it is possible to study the tumor coagulome at an unprecedented resolution with systems approaches, a major challenge remains in identifying the biological events and molecules that could dynamically shape it (4). Here, we extended the conclusions of our previous study (11) by documenting the effects of glucocorticoids on the cancer coagulome beyond the three key genes that were explored previously and in a number of different cellular contexts. In agreement with our previous findings of a direct positive transcriptional regulation of SERPINE1 explained by the interaction of the active GR with the promoter of SERPINE1 (11), we noted a consistent correlation between SERPINE1 and GR activity in different transcriptional datasets (NCI60 and TCGA, covering multiple types of primary tumors). We tested the effects of GC on OSCC cells in an experimental setting allowing for the simultaneous examination of thrombin/uPA activity upon GC exposure, in control and inflammatory conditions (mimicked by TNFα). We observed a strong inhibition of the proteolytic activity of uPA and a simultaneous reduction of D-dimer concentrations, the most commonly measured product of fibrin degradation, in the OSCC culture supernatants. These findings suggest that GC block plasminogen activation and fibrinolysis and may therefore stabilize fibrin clots formed in the vicinity of cancer cells.
Our study is to the best of our knowledge the first to examine the functional consequences of GC exposure on cancer cell-associated thrombin/plasminogen activation. The radical inhibition of plasminogen activation by uPA that we observed is in accordance with a robust and specific induction of SERPINE1/PAI1, further supporting the concept of a GR/SERPINE1 axis (11). This GR/SERPINE1 axis might be of clinical relevance, considering that GC might increase the occurrence of fibrin clots and the risk of VTE in cancer patients. Importantly, GC are not only endogenous hormones, but they are also among the most commonly used therapeutics in cancer patients addressed for surgical or radiotherapeutic treatment, i.e., procedures that increase the expression of F3/TF, as was for example recently shown in GBM (18). The risk of VTE induced by GC might be especially high for tumors such as OSCC that are characterized by the coexistence of a strong pro-fibrinolytic activity counterbalancing a hypercoagulable gene expression profile (19). For these tumors, thorough prospective studies addressing the effects of GC on biomarkers of coagulation/fibrinolysis/VTE risk are warranted. Interestingly, an increased risk of thromboembolic accidents has been noted with GC-treatment in the non-oncological context. A high risk of VTE was for example reported in Cushing’s syndrome, i.e., chronic hypercortisolism (20) and in patients with COVID-19 infection receiving oral GC (21). Furthermore, treatment of healthy subjects with GC for 10 days was found to induce a procoagulant state with increased PAI-1 blood levels (22). The conclusions of these studies and ours open up the possibility that a GR-SERPINE1 axis might be of general relevance, beyond the oncological context.
Finally, we hypothesize that our observations may be important beyond the thromboembolic risk. SERPINE1/PAI-1 has consistently been reported as one the strongest biomarkers and prognostic factors for various human cancers (23, 24). The formation of a fibrin network surrounding cancer cells, recently compared to a “bird’s nest” (25), could provide a scaffold for cell migration and favorable conditions for cancer cells to resist chemotherapeutic agents (25). Besides the control of the composition and turnover of the tumor extracellular matrix, proteases of the coagulation and fibrinolysis cascades also interact with specific protease-activated receptors (PARs) present on the surface of different cell types (4). We suggest that GC might potentially modulate the active interplay that takes place between the coagulation/fibrinolytic cascades and the tumor microenvironment (4). The GR-SERPINE1 axis might for example exert potent effects on tumor-associated macrophages (26). GC are well known to modulate the activation and function of cells of the immune system (27). The existence of the GR-SERPINE1 axis, as supported here with a combination of systems and functional approaches, calls for a broad re-examination of the consequences of GC signaling in tumor physiology and anti-tumor immunity, with particular attention paid to the tumor coagulome.
Conclusion
Overall, our findings suggest that the inhibition of fibrinolysis is a key consequence of cancer cell exposure to GC, possibly leading to the stabilization of the fibrin clot in cancer. Future studies are warranted to address the role of SERPINE1 induction in this context.
Acknowledgements
The Authors would like to thank the Amiens University Hospital and the Ligue contre le Cancer, comité de l’Aisne for the financial support.
Footnotes
Authors’ Contributions
Floriane Racine: Writing – original draft, investigation, data analysis. Christophe Louandre: Methodology, Investigation; Julien Demagny, Methodology, Investigation; Corinne Godin: Methodology, Investigation; Zuzana Saidak: Writing – original draft, validation, methodology, Investigation, data analysis, Writing - review and editing; Antoine Galmiche: Writing – original draft, validation, conceptualization, funding acquisition.
Conflicts of Interest
The Authors have no conflicts of interest to declare in relation to this study.
Funding
This study was financed by the Ligue contre le Cancer, comité de l’Aisne (COAGULCAN).
- Received June 17, 2024.
- Revision received June 24, 2024.
- Accepted June 26, 2024.
- Copyright © 2024 The Author(s). Published by the International Institute of Anticancer Research.
This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY-NC-ND) 4.0 international license (https://creativecommons.org/licenses/by-nc-nd/4.0).
