Cancer cell-derived tissue factor positive extracellular vesicles: biomarkers of thrombosis and survival

Abstract

Purpose of review:

Tissue factor (TF) is released from cancer cells and tumors in the form of extracellular vesicles (EVs). This review summarizes our current knowledge of the mechanisms of release of TF-positive EVs (TF+EVs) from cancer cells and the effect of these TF+EVs on cultured endothelial cells. In addition, we will summarize the contribution of TF+EVs to thrombosis in mice, and the association between plasma EVTF activity and venous thrombosis as well as survival of cancer patients.

Recent findings:

The release of TF+EVs from cancer cells is regulated by multiple factors, including hypoxia, epithelial-mesenchymal transition, and various intracellular signaling pathways. Cancer cell-derived, TF+EVs confer procoagulant activity to endothelial cells and induce the expression of adhesion proteins and IL-8. In addition, they contribute to thrombosis by directly activating the coagulation system and by generating thrombin that activates platelets in mouse models. Finally, there is an association between EVTF activity and venous thrombosis in pancreatic cancer patients as well as mortality in cancer patients.

Summary:

Cancer cell-derived TF+EVs bind to and activate endothelial cells. In addition, they serve as biomarkers of survival of cancer patients and venous thrombosis in pancreatic cancer patients.

Introduction

Tissue factor (TF) is a transmembrane receptor for factor (F) VII and VIIa [1]. The TF/FVIIa complex is the primary initiator of the coagulation protease cascade [1]. This complex and downstream coagulation proteases, including FXa and thrombin, also induce cellular responses via activation of protease-activated receptors (PARs), such as PAR1 and PAR2 [2]. TF is expressed by many tumor types, especially adenocarcinomas [3]. Cellular TF has been shown to contribute to many aspects of cancer biology, such as tumor growth, invasion, and metastasis, in in vitro and in vivo models (reviewed in [4–6]).

Cancer cells also constitutively release extracellular vesicles (EVs) (also known as microparticles, microvesicles, ectosomes, and exosomes) [7,8]. If a cancer cell expresses TF, the EV derived from the cells will be procoagulant due to the presence of TF.

In this review, we summarize the current knowledge of the regulation of the release of TF-positive EVs (TF+EVs) from cancer cells, the effect of cancer cell-derived, TF+EVs on endothelial cells, the effect of endogenous and exogenous TF+EVs on venous thrombosis in mice, the association between plasma EVTF activity and venous thrombosis in pancreatic cancer patients and the survival of cancer patients.

Regulation of the release of tissue factor-positive extracellular vesicles from cancer cells

Dvorak and colleagues were the first to describe that cancer cells release membrane vesicles (now called EVs) carrying procoagulant activity [9]. Subsequent studies revealed that this procoagulant activity was due to TF [10–12]. More recently, studies have investigated conditions that increase the release of TF+EVs from cancer cells. For instance, different groups found that exposing the human ovarian cancer cell line OVSAYO and human glioblastoma multiforme cell line U87-MG to hypoxia increased the release of TF+EVs [13,14].

Epithelial-mesenchymal transition (EMT) is a tumorigenic event where epithelial cells lose cell-cell adhesion and polarization and gain mobility and display a mesenchymal phenotype [15]. EMT contributes to tumor growth, invasion, dissemination and metastasis as well as drug resistance [16]. Garnier and colleagues showed that release of TF+EVs is enhanced by stimulation of a mesenchymal phenotype in the human squamous cell carcinoma cell line A431 and the human colorectal cancer cell line DLD-1 [17]. These data indicate that environmental factors, such as hypoxia, or a change in cellular phenotype, such as EMT, increase the release of TF+EVs.

Multiple signaling pathways regulate the release of TF+EVs from cancer cells. For instance, one study found that FXa and a PAR2 agonist enhanced the release of TF+EVs from the human breast cancer line MDA-MB-231 [18], which suggested that the effect of FXa was mediated by PAR2. The same group also found that activation of PAR2 enhanced the release of TF+EVs in other human cancer cell lines, including the pancreatic cancer cell line AsPc-1 [19]. We found that reducing mitogen-activated protease kinase kinase kinase 2 (MEKK2) expression, but not extracellular signal-regulated kinase 5 or Jun N-terminal kinase expression, in MDA-MB-231 cells reduced the release of TF+EVs into the culture supernatant [20]. These data indicate that FXa-PAR2 signaling increases the release of TF+EVs whereas MEKK2 is required for constitutive release of TF+EVs.

The actin-binding protein filamin A binds to the cytoplasmic domain of TF [21]. One study reported that reducing levels of filamin A in the human ovarian cancer cell line OVISE reduced both the release of EVs and TF incorporation into EVs [22]. Another study found that reducing levels of filamin A in MDA-MB-231 cells, human coronary artery endothelial cells (HCAECs), or human dermal blood endothelial cells (HDBECs) did not affect TF expression or the release of EVs but reduced incorporation of TF into EVs [18]. These studies suggest that filamin A regulates the release of EVs in cells and also the incorporation of TF into EVs.

TF itself also regulates the release of EVs from cancer cells. Rondon and colleagues showed that reducing levels of TF in MDA-MB-231 cells significantly reduced the number of EVs released from these cells in vitro [23]. The authors also found that expression of Rho A, which mediates EVs release, is reduced in MDA-MB-231 cells with reduced levels of TF [23].

Effect of cancer cell-derived tissue factor-positive extracellular vesicles on endothelial cells

Recent studies have demonstrated that EVs are cell-cell mediators that transfer material between cells [24–29]. Several studies found that cancer cell-derived, TF+EVs can be transferred to other cells. For instance, one study found that TF+EVs from high TF-expressing MDA-MB-231 cells could be transferred to a low TF-expressing human breast cancer cell line MCF-7 [24]. TF+EVs were incorporated into the plasma membrane and increased the procoagulant activity of MCF-7 cells. Collier and colleagues also found that TF+EVs from MDA-MB-231 cells could be transferred to HDBECs resulting in an increase in TF activity of HDBECs [25]. Interestingly, inhibition of dynamin, which is required for endocytosis, significantly reduced the internalization of TF+EVs and recycling of TF to the cellular surface in HDBEC. More recently, Adesanya and colleagues showed that TF+EVs from a human head and neck squamous cell carcinoma line UMSCC81B could be incorporated into human umbilical vein endothelial cells (HUVECs) [29]. Again, the TF was functional because the procoagulant activity of TF+EV HUVECs was significantly increased compared with untreated HUVECs.

Cancer cell-derived, TF+EVs also activate ECs (Figure 1). Svensson and colleagues found that U87-MG-derived, TF+EVs were incorporated into hypoxic HUVECs [14]. U87-MG-derived, TF+EVs induced phosphorylation of ERK1/2 and expression of heparin-binding EGF-like growth factor via PAR2 signaling. Similarly, Che and colleagues showed that TF+EVs from MDA-MB-231 cells and the human pancreatic cancer cell lines BxPC-3 and Capan-1, together with FVIIa and FX, induced E-selectin and IL-8 expression in HUVECs [26]. This induction of E-selectin and IL-8 in HUVECs was PAR1-dependent but PAR2-independent. These data demonstrate that cancer cells-derived, TF+EVs induce a pro-coagulant, pro-adhesive and pro-inflammatory phenotype of endothelial cells. We speculated that these changes may enhance tumor growth by stimulating angiogenesis.

Figure 1. Activation of endothelial cells by cancer cell-derived, tissue factor-positive extracellular vesicles.

(A) Cancer cell-derived, tissue factor (TF)-positive extracellular vesicles (EVs) binds to endothelial cells and induce phosphorylation of ERK1/2 and expression of heparin binding EGF-like growth factor (HB-EGF) via protease-activated receptor (PAR) 2 signaling. (B) Cancer cell-derived, TF+EVs induce E-selectin and IL-8 expression in the presence of factor (F)VIIa and FX(a). Cells and proteins were modified from Servier Medical Art, licensed under Creative Common Attribution 3.0 Unported License. (http://www.servier.fr/servier-medical-art)

Cancer cell-derived extracellular vesicle tissue factor activity activate coagulation and enhance thrombosis in mouse models

The role of exogenous and endogenous cancer cell-derived TF+EVs in thrombosis have been investigated using mouse models. Exogenous TF+ EVs are useful to study thrombosis but they do not recapitulate the more complex situation of enhanced thrombosis in tumor-bearing mice or cancer patients that involves interactions between multiple prothrombotic pathways.

Exogenous tissue factor-positive extracellular vesicles

Thomas and colleagues showed that exogenous TF+EVs derived from a murine pancreatic cancer cell line Panc02 shortened the occlusion times of mesenteric venules and arterioles exposed to ferric chloride [30]. Administration of an anti-P-selectin antibody abolished the prothrombotic effect of the EVs suggesting that binding of EVs to the endothelium was mediated by P-selectin. A second study from Thomas and colleagues showed that exogenous Panc02-derived TF+EVs enhanced thrombosis in an inferior vena cava (IVC) stenosis model in a TF- and thrombin-dependent but P-selectin and glycoprotein 1b independent manner [31]. It is unclear whether P-selectin has a different role in the 2 thrombosis models. We found that exogenous human cancer cell line HPAF-II-derived TF+EVs activated coagulation and enhanced thrombosis in the IVC stenosis model [32]. Our second study also showed that exogenous BxPc-3-derived TF+EVs enhanced thrombosis in the IVC stenosis model in a TF- dependent manner [27]. In addition, we found that exogenous BxPc-3-derived TF+EVs decreased the survival of mice in a TF-dependent manner. Finally, Stark and colleagues showed that TF+EVs from the human pancreatic cancer cell lines FG and L3.6pl and murine pancreatic cancer cells KPC enhanced thrombosis in an IVC stenosis model [33]. Pretreatment with duramycin, which binds to the negatively-charged phospholipid phosphatidylethanolamine, abolished the prothrombotic effect of the EVs. These data indicate that exogenous TF+EVs enhance thrombosis in mice.

Endogenous tissue factor-positive extracellular vesicles

Thomas and colleagues found that immunocompetent C57BL/6 mice bearing Panc02 tumors had enhanced thrombosis in the IVC stenosis model compared with controls[30]. In contrast, we found that nude mice bearing HPAF-II tumors did not have larger thrombi in the IVC stenosis model compared with controls but exhibited reduced occlusion time in a saphenous vein ferric chloride injury model [32]. In our second study, we found that nude mice bearing BxPc-3 tumors had increased thrombus area and non-significantly increased incidence in the IVC stenosis model [27]. Our third study showed that nude mice bearing BxPc-3 tumors enhanced thrombosis in the IVC stasis model and the enhanced thrombosis was abolished by administration of an anti-human TF antibody [34]. These data indicated that cancer cell-derived TF+EV mediated the enhancement of thrombosis.

Tissue factor-positive extracellular vesicles activate platelets and this contributes to increased thrombosis

Several studies have shown that cancer cell-derived, TF+EVs activate platelets. For instance, Thomas and colleagues showed that Panc02-derived EVs induce the aggregation of platelets in a TF-dependent manner [30]. We found that TF+EVs from BxPC-3 and a human pancreatic cancer cell line L3.6pl induced P-selectin expression in platelets and platelet aggregation in a TF- and thrombin-dependent manner [27]. Moreover, TF+EVs-enhanced thrombus formation in an IVC stenosis model that could be reduced in PAR4 knockout mice and in wild-type mice treated with the P2Y12 inhibitor clopidogrel. Finally, Gomes and colleagues showed that MDA-MB-231 cell-derived TF+EVs increased platelet aggregation and P-selectin expression compared with MCF-7 cell-derived TF-EVs [28]. These data suggest that cancer cell-derived, TF+EVs may contribute to thrombosis, in part, through platelet activation (Figure 2).

Figure 2. Cancer cell-derived, tissue factor positive extracellular vesicles contribute to thrombosis.

Cancer cell-derived, tissue factor (TF)-positive extracellular vesicles (EVs) contribute to thrombosis, in part, through platelet activation by thrombin.

Association between levels of extracellular vesicle tissue factor activity and venous thrombosis in pancreatic cancer patients

There are different rates of venous thromboembolism (VTE) in different types of cancer [35–37]. Pancreatic cancer has one of the highest incidences of VTE (5.3–42%) [35–38]. TF+EVs may be one of the mechanisms that enhances VTE in patients with pancreatic cancer. Many retrospective and prospective studies have measured plasma extracellular vesicle TF (EVTF) activity in different types of cancer (reviewed in [39,40]). Importantly, there is an association between EVTF activity and VTE in pancreatic cancer, but not in other types of cancer including brain, colorectal, gastric, lung and ovarian cancer [41–44]. Of note, one prospective study showed that there is a border line non-significant association between EVTF activity and VTE in patients with pancreatic cancer [42]. This study used a long follow-up time (2 years). In addition to follow-up term, preanalytical settings also affect the results [40]. Improved sensitivity and reduced variability of the measurement of plasma EVTF activity is needed to better determine the contribution of TF+EVs to VTE in cancer.

Association between levels of extracellular vesicle tissue factor activity and survival in cancer patients

High levels of TF expression in tumors is associated with worse prognosis in various types of cancers, including breast, prostate, colorectal and pancreatic cancer [3,45–52]. Similarly, several studies have reported an association between EVTF activity and survival of cancer patients (Table 1). All but one of these studies used either an in-house endpoint [41] or kinetic [53] EVTF activity assay developed by us and another group. One of the studies used a commercial ACTICHROME® assay to measure plasma and EVTF activity. Measurement of plasma TF activity is problematic because FVIIa and FX are added to the plasma and the amount of FXa is determined using a chromogenic substrate. One study found that the color of the plasma can affect the results [54]. Additionally, measurement of FXa generation without using an anti-TF antibody does not distinguish TF-dependent from TF-independent FXa generation.

Table 1.

Studies evaluating the association between extracellular vesicle tissue factor activity and survival in patients with cancer

Study	Method	Patients	Time of follow-up	Association between EVTF and mortality?	Cut off-value of EVTF activity
Retrospective
Tesselaar, 2007	In-house kinetic EVTF activity	23 late stage pancreatic cancer patients	n/a	Yes	273 fM Xa min−1
		27 late stage breast cancer patients		(pancreatic cancer)
Tesselaar, 2009	In-house kinetic EVTF activity	51 cancer patients with VTE	n/a	Yes	273 fM Xa min−1
		49 case-matched cancer patients without VTE
Thaler, 2013	In-house endpoint EVTF activity	73 pancreatic cancer patients (18 localized resected, 13 localized unresected, 13 metastatic recurrent, 29 metastatic non-resectable)	n/a	Yes	0.84 pg/mL
Prospective
Thaler, 2012	In-house endpoint EVTF activity	348 cancer patients (60 pancreatic, 43 stomach, 126 colorectal, 119 brain)	2 years	Yes	0.19 pg/mL
				(pancreatic cancer)
Bharthuar, 2013	In-house endpoint EVTF activity	117 pancreaticobiliary cancer patients	6 months	Yes	2.5 pg/mL
Hernandez, 2013	ACTICHROME®	252 cancer patients (121gastrointestinal, 9 pancreatic, 18 lung and 43 breast cancer, and 62 non-Hodgkin lymphoma)	10 months	Yes	detectable EVTF
Woei, 2016	In-house kinetic EVTF activity	79 pancreatic cancer patients	Not specified	Yes	155 fM Xa min−1
Hisada, 2018	In-house endpoint EVTF activity	60 cancer patients (17 different types of cancer)	3 months	Yes	0.3 pg/mL

Retrospective studies

Tesselaar and colleagues performed the first retrospective study that investigated the association between plasma TF+EVs and survival in patients with cancer in 2007 [53]. This study included 23 patients with either non-resectable locally advanced (n= 4) or metastatic (n=19) pancreatic cancer, and 27 patients with either early (n=10) or late (n=17) stage breast cancer and 37 healthy controls. They found that EVTF activity was significantly increased in the metastatic pancreatic and breast cancer compared with healthy controls. Moreover, elevated levels of EVTF activity was associated with reduced survival in pancreatic cancer patients but not breast cancer patients.

The same group performed a second study that include 51 unselected cancer patients with VTE and 49 case-matched cancer patients without VTE [55]. This study included 27 gastrointestinal carcinoma, 12 genito-urinary tract tumors, and 12 other types of tumors in VTE group. They found that the median survival in patients with elevated levels of EVTF activity (n = 32) was 3.5 months whereas the median survival of patients with low levels of EVTF activity was 13 months (n = 68). Moreover, in the group of cancer patients with VTE, patients with elevated levels of EVTF activity showed significantly reduced survival compared to patients with low levels of EVTF activity. Importantly, univariate analysis revealed that tumor-type, VTE and EVTF activity were all significantly associated with survival.

We also performed a retrospective study that included 73 pancreatic cancer patients and 22 healthy controls [56]. These cancers included 18 localized resected cases, 13 localized unresected cases, 13 metastatic recurrent cases, and 29 metastatic non-resectable cases. We found that all cancer patients had significantly increased EVTF activity compared with healthy controls. In addition, we found that patients with increased levels of EVTF (75^th percentile or above) exhibited an overall survival probability of 53% after 3 months and 42% after 6 months. In contrast, cancer patients without elevated EVTF activity (below 75^th percentile) had a survival probability of 93% after 3 months and 87% after 6 months.

Prospective studies

We performed the first prospective study on the association between EVTF activity and survival in 2012 [42]. This study included 60 pancreatic, 43 stomach, 126 colorectal and 119 brain cancer. We found that pancreatic cancer patients with elevated levels of EVTF activity (75^th percentile or above) had survival probabilities of 31% after 6 months and 15% after 12 months. In contrast, pancreatic cancer patients with low levels of EVTF activity (below 75^th percentile) had survival probabilities of 82% after 6 months and 59% after 12 months.

In a second study, we analyzed 117 cancer patients, including 80 pancreatic, 34 biliary and 3 with an unknown primary cancer [57]. We found that the median survival of patients with elevated levels of EVTF activity (≥ 2.5 pg/mL) was 98.5 days compared with 231 days for those with low levels of EVTF activity (< 2.5 pg/mL).

Hernandez and colleagues performed a prospective study with 252 cancer patients but used the ACTICHROME® assay to measure plasma and EVTF activity [58]. This study included 130 digestive, 9 pancreatic, 18 lung, 42 breast and 62 non-Hodgkin lymphoma. They found that patients with detectable EVTF activity showed worse prognosis compared with those without detectable EVTF activity.

Woei and Colleagues studied 79 patients with pancreatic adenocarcinoma [59]. They found that the median EVTF activity in pancreatic cancer patients was significantly higher than in healthy controls. Moreover, there was a significant association between EVTF activity and tumor stage. Importantly, overall survival was reduced in pancreatic cancer patients with elevated levels of EVTF activity than in patients with EVTF activities within the normal range observed in healthy controls (2.9 vs 8.5 months).

More recently, we performed a prospective study, including 60 cancer patients, 51 severely ill and hospitalized patients without known cancer, and 50 healthy controls [60]. This included patients with various types of cancer, including 2 pancreatic cancer patients. We found that cancer patients had significantly increased levels of EVTF activity compared with healthy controls. Interestingly, patients with adenocarcinoma showed significantly increased EVTF activity compared with patients with non-adenocarcinoma tumors. In addition, cancer patients with elevated levels of EVTF activity (≥ median) had a survival probability of 16.7% after 90 days whereas cancer patients with low levels of EVTF activity (< median) had a survival probability of 30.0% after 90 days. These studies suggest that TF+EVs can be used as a biomarker to predict survival in patients with some types of cancer, including pancreatic cancer.

Conclusion

TF+EVs are constitutively released from TF-expressing cancer cells into the circulation. They bind to the endothelium and increase the procoagulant activity and induce expression of adhesive proteins and inflammatory mediators. This suggests that these EVs contribute to tumor growth. In addition, TF+EVs directly and indirectly, via platelets, enhance thrombosis. Inhibiting the release or procoagulant activity of TF+EVs may reduce VTE. Finally, TF+EVs are associated with survival in cancer patients.

Key points.

• 1/Multiple pathways enhance release of tissue factor positive extracellular vesicles from cancer cells.
• 2/Cancer cell-derived, tissue factor positive extracellular vesicles activate endothelial cells.
• 3/Exogenous and endogenous cancer cell-derived, tissue-factor positive extracellular vesicles enhance thrombosis in mouse.
• 4/Extracellular vesicle tissue factor activity is associated with venous thrombosis in pancreatic cancer.
• 5/Extracellular vesicle tissue factor activity is associated with survival in cancer patients.