Thrombomodulin: A multifunctional receptor modulating the endothelial quiescence

Summary

Thrombomodulin (TM) is a type 1 receptor best known for its function as an anticoagulant cofactor for thrombin activation of protein C on the surface of vascular endothelial cells. In addition to its anticoagulant cofactor function, TM also regulates fibrinolysis, complement and inflammatory pathways. TM is a multidomain receptor protein with a lectin-like domain at its N-terminus that has been shown to exhibit direct anti-inflammatory functions. This domain is followed by six epidermal growth factor-like domains that support the interaction of TM with thrombin. The interaction inhibits the procoagulant function of thrombin and enables the protease to regulate the anticoagulant and fibrinolytic pathways by activating protein C and thrombin activatable fibrinolysis inhibitor. TM has a Thr/Ser rich region immediately above the membrane surface that harbors chondroitin sulfate glycosaminoglycans, and this region is followed by a single spanning transmembrane and a C-terminal cytoplasmic domain. The structure and physiological function of the extracellular domains of TM have been extensively studied and numerous excellent review articles have been published. However, the physiological function of the cytoplasmic domain of TM has remained poorly understood. Recent data from our laboratory suggests that intracellular signaling by the cytoplasmic domain of TM plays key roles in maintaining quiescence by modulating phosphatase and tensin homolog (PTEN) signaling in endothelial cells. This article will briefly review the structure and function of extracellular domains of TM and then focus on the mechanism and possible physiological importance of the cytoplasmic domain of TM in modulating PTEN signaling in endothelial cells.

1. Introduction

Thrombomodulin (TM, CD141) is a multidomain type-1 integral membrane protein which was initially discovered by Esmon and Owen in 1981 for its ability to function as a cofactor receptor for thrombin in promoting the activation of protein C to activated protein C (APC) on endothelial cells [1]. Since its discovery, it has been demonstrated that TM is expressed in most cells, though its expression level on vascular endothelial cells has been found to be the most abundant [2–8]. TM plays critical roles in regulating several physiological processes including hemostasis, coagulation, fibrinolysis, inflammation, and angiogenesis [9]. The role of TM in blood coagulation and fibrinolysis pathways has been well studied [10–12]. The interaction of the epidermal growth factor (EGF)-like domains of TM with basic residues of exosite-1 of thrombin switches the specificity of thrombin from a procoagulant to an anticoagulant and anti-fibrinolytic protease. Upon interaction, TM inhibits the proteolytic activity of thrombin toward fibrinogen and promotes its activity toward protein C and thrombin-activatable fibrinolysis inhibitor (TAFI) [10–13] (Fig, 1). In addition to regulation of coagulation and fibrinolysis, TM expressed on the surface of endothelial cells plays critical protective roles in changing the cleavage specificity of protease-activated receptor 1 (PAR1), inhibiting endothelial cell permeability and inflammation [9,10,14]. Various pathological inflammatory conditions can cause endothelial dysfunction, leading to increased vessel adhesiveness to leukocytes, increased endothelial cell migration, proliferation, apoptosis, and loss of barrier function. TM plays cytoprotective roles in preventing endothelial cell dysfunction under different pathophysiological conditions [15]. The anticoagulant and anti-inflammatory properties of TM have led to the approval of the soluble form of TM lacking its cytosolic and transmembrane domains as a drug for the treatment of disseminated intravascular coagulation and sepsis in Japan [15,16]. Several excellent reviews have been published on the structure and physiological function of TM [9,10,17–21]. This article will briefly review the literature with respect to the structure and function of various domains of TM and then focus on the novel data related to the role of the cytoplasmic domain of TM in intracellular signaling function of the receptor.

Figure 1: Schematic illustration of TM domains and their key functions:

TM is a multidomain receptor protein composed of: (1) an N-terminal lectin-like domain, which maintains an anti-inflammatory signaling by inhibiting expression of cell adhesion molecules (CAMs), sequestering, and inhibiting pro-inflammatory signaling by damage/pathogen-associated molecular patterns (DAMP) receptors. (2) Six epidermal growth factor (EGF)-like domains with EGF domains 4, 5, and 6 (E4–6) binding thrombin and initiating the anticoagulant pathway by activating protein C to activated protein C (APC), which proteolytically degrades procoagulants factors FVa and FVIIIa. Through interaction with E3–6 of TM, thrombin also activates the thrombin activatable fibrinolysis inhibitor TAFI to TAFIa, thereby regulating fibrinolysis. Angiopoietins (Ang1/Ang2) has been reported to bind EGF-like domains and inhibit protein C and TAFI activation by thrombin. EGF1–6 domains may regulate cell proliferation by binding to fibroblast growth factor receptor 1 (FGFR1) and the G protein-coupled receptor 15 (GPR15). (3) The transmembrane domain of TM is followed by (4) a cytoplasmic domain that is involved in maintaining endothelial cell quiescence by regulating AKT activation, inhibiting permeability, constitutive VWF release, and cell migration. The image was created with BioRender.com.

2. TM structure and function

The protein structure of mature TM (aa19-575) is comprised of five domains: i) an N-terminal C-type lectin-like domain (aa31–169), ii) six EGF-like domains (aa 241–281, aa 284–324, aa 325–363, aa 365–405, aa 404–440, and aa 441–481) with each domain having six Cys residues forming three disulfide bonds, iii) an extracellular membrane proximal Ser/Thr rich domain harboring chondroitin sulfate (CS) glycosaminoglycans, iv) a single spanning transmembrane domain (aa 516–539), and v) a C-terminal cytoplasmic domain (aa 540–575) [17,18] (Fig. 1).

3. The C-type lectin-like domain (lectin domain)

The lectin domain of TM is located at the N-terminus of the receptor and its physiological significance was first studied in TM transgenic mice harboring a deleted lectin domain [22]. Mice deficient in TM lectin domain were viable and healthy, however, bacterial lipopolysaccharides (LPS) challenge in these mice led to elevated cytokine production and mortality [22]. A normal activation profile for protein C in these mice suggested that the lectin domain of TM has a potent anti-inflammatory function independent of its cofactor function for thrombin during APC generation [22]. Numerous in vitro and in vivo studies have demonstrated that the lectin domain of TM inhibits NF-kB activation, expression of cell adhesion molecules (CAMs), and rolling of neutrophils on vascular endothelial cells in response to pro-inflammatory cytokines [20,23]. The lectin domain of TM is thought to bind a vascular cell surface receptor to elicit anti-inflammatory signaling effects; however, no such receptor has been identified for the lectin domain. The known mechanisms explaining the anti-inflammatory functions of the lectin domain of TM appears to involve a direct interaction of this domain with specific carbohydrate moieties present on inflammatory ligands such as high mobility group box 1 (HMGB1), histones and Lewis Y antigen on LPS (Fig. 1), thereby TM sequestering and inhibiting binding of these pro-inflammatory molecules to damage/pathogen-associated molecular patterns (DAMP) receptors [19,24–26]. It has been also reported that TM can function as a cofactor for thrombin in enhancing the cleavage of HMGB1 and inhibiting its pro-inflammatory signaling by DAMP receptors [27,28]. In human leukemia monocytic THP-1 and murine macrophage RAW 264.7 cells, the lectin domain of TM has been shown to decrease LPS-induced production of tumor necrosis factor α (TNFα), NF-κB and associated MAPK signaling pathways [25]. The anti-inflammatory effect of the lectin domain of TM has been studied in vivo and shown that intravenous injection of the isolated lectin domain attenuates the serum levels of TNFα in mice challenged with LPS or Klebsiella pneumoniae [25]. In keratocytes, a 31 amino acids peptide derived from the lectin domain of TM inhibited LPS-induced expression of intercellular adhesion molecule 1 (ICAM-1) and nuclear factor (NF)-kB signaling [29] by attenuating the formation of a complex between toll-like receptor 4 (TLR4) and its ligands; CD14 and LPS [30]. In addition to direct binding to LPS, the lectin domain of TM induces agglutination of Klebsiella pneumoniae and E coli DH5α, thereby promoting efficient phagocytosis of bacterial pathogens by macrophages [29]. The lectin domain of TM has also been shown to neutralize activation of the complement cascade, thus transgenic mice lacking this domain of TM develop inflammatory arthritis with increased complement C3 deposition in the joints [31] and kidneys [32]. Infusion of isolated lectin domain of TM inhibited activation of complement pathways in these mice [31]. This function of TM may be of clinical significance in patients with hemolytic uremic syndrome where several point mutations have been identified in the lectin domain (A43T, D53G, V81I) which are associated with complement dysregulation [33].

4. EGF-like domains

TM possesses six EGF-like domains (EGF1–6). The cofactor function of TM in promoting the activation of protein C by thrombin is mediated via interaction of EGF4–6 of TM with exosite-1 of thrombin. Structural and mutagenesis data have indicated that EGF56 binds to basic residues of exosite-1 [13], and this binding not only induces conformational changes in the structure of thrombin to accommodate the activation peptide of protein C in the catalytic pocket [34], but also positions EGF4 at a location that functions as a platform for interacting with a basic exosite of the zymogen, thereby facilitating formation of a productive ternary TM-thrombin-protein C activation complex [35]. In the case of thrombin activation of TAFI, EGF3–6 domain of TM is required for its cofactor function [36]. Despite overlaps in the EGF domains requirement for activation of the two proteins by thrombin, it has been found that protein C and TAFI are both concurrently activated in a TM-dependent manner on cultured endothelial cells without competing for the thrombin-TM complex, raising the possibility that they interact with distinct activation complexes [37]. However, in a kinetic study using purified protein components, TAFI activation by the thrombin-TM complex has been shown to be competitively inhibited by protein C [38]. A recent study reported that angiopoietins 1 and 2 (Ang1/Ang2) can also bind to EGF-like domains of TM (Fig. 1), thereby competitively inhibiting activation of both protein C and TAFI by thrombin [39]. Whether the interaction of Ang1/Ang2 influences the intracellular signaling of TM has not been investigated.

It has been reported that the binding of EGF5 domain of TM to the G-protein coupled receptor 15 (GPR15) mediates a cytoprotective function by upregulating the expression of myeloid cell leukemia sequence 1, thereby attenuating cyclosporine A-induced apoptosis in human umbilical vein endothelial cells [40]. A pro-angiogenic function for this domain of TM through interaction with GPR15 has also been reported in a murine model [40]. It has been also reported that EGF5 domain of TM alleviates LPS-induced sepsis through the same mechanism [41]. Moreover, recombinant soluble TM (extracellular domains of TM including lectin and all EGF-like domains) has been shown to downregulate transforming growth factor β1 (TGFβ1)-induced epithelial-mesenchymal transition (EMT) in human podocyte primary cells and kidney fibrosis in glomerulus-specific human TGFβ1 transgenic mouse [42]. Interestingly, the siRNA knockdown of GPR15 in human podocyte primary cells eliminated TGFβ1 inhibitory function of soluble TM [42], raising the possibility that this effect of soluble TM is mediated through EGF5. A role for the pro-angiogenic effect for EGF1–6 domains of TM has been reported through interaction with type I fibroblast growth factor receptor (FGFR1) [43]. It is worth mentioning that these nontraditional functions for TM have all been derived from studies employing synthetic or recombinant isolated domains of the receptor, thus additional investigation may be required to fully understand the true physiological significance of these domains in the context of the intact cell-bound receptor.

5. Chondroitin sulfate (CS) attached to the Ser/Thr rich domain

The physiological function of the CS moiety of TM is not fully understood. It is known that by binding to basic residues of exosite-2 of thrombin, CS attached to TM not only significantly increases the affinity of the receptor for thrombin [44], but also alters the calcium-dependence of protein C activation by the thrombin-TM complex [45]. It has also been demonstrated that the CS moiety of TM binds to a second molecule of thrombin which may have significance for its anticoagulant function [46]. It has been reported that recombinant EGF domains plus the Ser/Thr-rich domain of TM have a high stimulatory effect on endothelial cell proliferation, migration, and angiogenesis in comparison to EGF domains alone [43,47]. In this context, it has been found that the TM fragment containing EGF and Ser/Thr-rich domains directly interacts with proangiogenic fibroblast growth factor receptor 1 (FGFR1)-syndecan-4 complex to potentiate FGFR1 signaling in endothelial cells [43]. Further studies are required to investigate the physiological significance of this observation and determine whether cell bound TM can exert FGFR1-syndecan-4-dependent downstream signaling in endothelial cells. Finally, in another study it was demonstrated that CS containing TM is more potent inhibitor of histone-induced platelet aggregation and neutrophil extracellular trap (NET) release in comparison to TM lacking its CS moiety [48]. In this case, stronger binding of CS containing TM to histones H3 and H4 appears to be an important factor in neutralizing the inflammatory effects of these nuclear proteins [48].

6. Cytoplasmic domain

The function of the 39 amino acids long cytoplasmic domain of TM remains unknown. In the mouse, global deletion of the cytoplasmic domain of TM did not affect embryonic development [49]. The most prominent phenotypic change that was observed in mice expressing TM lacking its cytoplasmic domain was a significantly lower plasmin generation [49]. This study speculated a modulatory effect for the TM cytoplasmic domain in fibrinolytic activity, however, the study did not rule out other roles for the cytoplasmic domain of TM including in tumor/cell proliferation [49]. It has been reported that the cytoplasmic domain of TM interacts directly with the actin cytoskeleton in epithelial cells through membrane-anchoring proteins called ERM (ezrin, radixin and moesin) proteins [50], which are involved in controlling cell shape by facilitating the attachment of cortical actin microfilaments to the cell membrane [51,52]. A direct interaction between TM cytoplasmic domain and ezrin, one of the three closely related proteins of ERM, has been reported in epithelial cells that is mediated through three basic cytoplasmic juxtamembrane Arg-Lys-Lys residues of TM [50]. Whether the cytoplasmic domain of TM similarly interacts with ERM proteins in endothelial cells has not been investigated. However, it has been shown that under inflammatory conditions, TNFα-induced phosphorylation of ezrin results in its dissociation from ERM proteins, leading to subsequent increase in the barrier-permeability of endothelial cells [53]. Moreover, in the monocytic THP-1 cell line, the protein intersectin-1 has been shown to connect the cytoplasmic domain of the cell surface TM to actin filaments, thereby regulating cell movements [54,55]. Deletion of the cytoplasmic domain of TM has been shown to result in increased THP-1 chemotaxis [55].

We have demonstrated that TM is co-localized with protease-activated receptor 1 (PAR1) and endothelial protein C receptor (EPCR) in lipid-rafts of endothelial cells where cellular signaling molecules are enriched [56]. Thus, the activation of the EPCR-bound protein C by the TM-bound thrombin is mediated in lipid-rafts of endothelial cells. The consensus view is that unlike APC, which initiates a cytoprotective response through cleavage of Arg46 of PAR1, thrombin elicits an inflammatory response through cleavage of Arg41 of PAR1 in endothelial cells. In a recent study we discovered that TM switches the signaling and PAR1 cleavage specificity of thrombin so that the protease functions only in the cytoprotective pathways [57]. This is facilitated by thrombin binding to TM, thereby recruiting both β-arrestin-1 and −2 to the plasma membrane by a mechanism that requires the cytoplasmic domain of TM. Thus, upon interaction with TM, thrombin initiates protective responses by β-arrestin-biased PAR1 signaling through cleavage of either Arg41 (mediated by β-arrestin-1) or Arg46 (mediated by β-arrestin-2) [57]. Interestingly, by analyzing functional data and cleavage rates we discovered that relative to APC thrombin cleaves Arg46 of PAR1 with >10-fold higher efficiency. We believe that this finding may have important implications for the thrombin regulation of immunothrombosis in microcirculation since the expression of TM is markedly high in microvascular endothelial cells [58], but the expression of EPCR while abundant on large vessels is insignificant in micro vessels such as lung capillaries [59]. Based on our findings, we have hypothesized that the pathological effects of immunothrombosis which are mediated by activation of inflammation and coagulation in microcirculation is primarily regulated by thrombin through activation of PAR1 and APC though inhibition of coagulation [57].

6.1. Cytoplasmic domain of TM and endothelial cell barrier function

We recently investigated the role of TM and its cytoplasmic domain in endothelial cells by employing CRISPR/Cas9 technology and deleting TM from the EA.hy926 endothelial cell line [14]. The TM-deficient (TM^−/−) cells exhibited increased basal permeability and hyper-permeability when stimulated by pro-inflammatory cytokines. The loss of barrier-permeability function was associated with reduced polymerization of F-actin filaments and phosphorylation-dependent loss of VE-cadherin at cell junctions in TM^−/− cells under both basal and stimulated conditions [14]. The basal NF-kB signaling, and expression of cell adhesion molecules (CAMs) were upregulated in TM^−/− cells, leading to enhanced adhesion of leukocytes to TM^−/− cells in flow chamber experiments. Furthermore, rather than storage in Weibel-Palade bodies (WPBs), TM deficiency caused a marked increase in expression and constitutive release of von Willebrand factor (VWF) from endothelial cells, thus supporting adhesion to isolated platelets and forming VWF-platelet strings on endothelial cells in a flow chamber assay [14] (Fig. 2). Unlike an increased expression and constitutive release of VWF, a significant decrease in the storage and release of Ang2 was observed in TM^−/− cells [60]. Increased VWF levels and inflammatory foci were also observed in the lungs of tamoxifen treated ERcre-TM^f/f (TM-KO) mice [14]. Re-expression of a wildtype TM construct, but not a TM construct which lacked the cytoplasmic domain (TM-des-Cyto) in TM^−/− cells normalized the cellular phenotypes [14,60], suggesting that the cytoplasmic domain of TM endows a quiescent phenotype by tightly regulating expression of procoagulant, pro-inflammatory and angiogenic molecules in vascular endothelial cells (Fig. 2).

Figure 2: Schematic illustration of the cytoplasmic domain of TM regulating endothelial cell permeability, inflammation, proliferation, and migration:

TM maintains a basal endothelial cell survival signaling function by inhibiting the activation of NF-κB, nuclear localization of FOXO1, maintenance of barrier-permeability function and fine-tuning of the AKT signaling through PTEN localization to the membrane/nucleus. The cytoplasmic domain of TM is linked to the actin cytoskeleton, which is important for the maintenance of the barrier-permeability function of endothelial cells. The deletion of the cytoplasmic domain of TM (right panel) results in loss of the barrier function (stress fiber formation), constitutive release of VWF (rather and formation of VWF-platelet strings on the surface of the endothelium. The deletion of the cytoplasmic domain also results in phosphorylation/inactivation of PTEN, leading to hyperactivation of the PI3K/AKT/mTORC1 signaling axis, upregulation of pro-inflammatory, procoagulant, proliferative, and migratory phenotypes in endothelial cells. Hyperactivated AKT induces phosphorylation-dependent nuclear exclusion and degradation of FOXO1, thereby inhibiting the transcription of FOXO1-target genes that are involved in regulation of proliferation and quiescence. TM, thrombomodulin; CS, chondroitin sulfate; VWF, von Willebrand factor; CAMs, cell adhesion molecules; WPBs, Weibel-Palade bodies. The image was created with BioRender.com.

6.2. Cytoplasmic domain of TM is involved in modulating AKT/FOXO1 signaling

Deletion of TM from endothelial cells resulted in hyperactivation of AKT (protein kinase B) and nuclear exclusion of the forkhead box O1 (FOXO1) transcription factor in TM^−/− cells and TM-KO mice [60]. AKT plays a key role in modulation of proliferation and angiogenesis via mediating phosphorylation-dependent nuclear exclusion of FOXO1 in endothelial cells [61–64]. In un-phosphorylated form, FOXO1 enters the nucleus and upregulates expression of several target genes that are involved in regulation of quiescence, apoptosis, cell cycle arrest, glucose metabolism, oxidative stress, and longevity [61–64]. However, when AKT is activated, it phosphorylates FOXO1 leading to its nuclear exclusion, proteasomal degradation and inhibition of expression of its target genes, including Ang2 which is one of the target genes of FOXO1, thus explaining the basis for the lack of Ang2 expression in TM^−/− cells [14]. TM deficiency increased cell migration and proliferation under basal conditions or when TM^−/− cells were stimulated with growth factors [i.e., Ang1 and vascular endothelial growth factor (VEGF)] [14,60]. The activation of AKT and phosphorylation of FOXO1 were inhibited under both basal and stimulated conditions when TM^−/− cells were transfected with a wildtype TM construct, but not with TM-des-Cyto, suggesting that the cytoplasmic domain of TM is involved in regulating AKT/FOXO1 signaling pathways in endothelial cells [60].

6. 3. TM regulation of phosphatase and tensin homolog (PTEN) signaling

The PI3K-dependent phosphorylation and activation of AKT is regulated by PTEN, which is a tumor suppressor protein involved in antagonizing the phosphoinositide 3-kinase (PI3K) signaling pathway by dephosphorylating the membrane-bound phosphatidylinositol 3,4,5-trisphosphate (PIP3) to phosphatidylinositol 4,5-trisphosphate (PIP2), thereby inhibiting recruitment of AKT to the plasma membrane and its subsequent activation by downstream protein kinases, PKD1 and PKD2, at Thr308 and Ser473 sites, respectively [65–67]. PTEN can be localized to the cytoplasm, the membrane or the nucleus, and a phosphorylation-dependent structural change in the C-terminal tail regulates its function by trapping the protein in an inactive closed conformation [67]. Interestingly, in a recent study, we discovered that the expression of total PTEN was decreased and its C-terminal tail phosphorylation in both TM^−/− cells and liver tissues of TM-KO mice were significantly increased, suggesting that TM deficiency is associated with downregulation of expression and inactivation of PTEN [60]. Moreover, the phosphorylation of proline-rich AKT substrate of 40 kDa (PRAS40) in the liver samples of the TM-KO mice was elevated, indicating that TM deletion results in activation of mammalian target of rapamycin complex 1 (mTORC1) in the liver endothelial cells of the TM-KO mice [60]. PRAS40 is a specific component of mTORC1 and acts as an inhibitor by binding to the regulatory associated protein of mTOR (raptor), thereby inhibiting the kinase activity of mTORC1 [68,69]. When PRAS40 is phosphorylated by AKT, its inhibitory effect on raptor is lifted resulting in activation of the mTORC1 pathway [68,69]. Interestingly, the expression of raptor and phosphorylation of PRAS40 were found to be also enhanced in both TM^−/− cells and TM^−/− cells transfected with TM-des-Cyto but not with wildtype TM [60]. On the other hand, the level of rapamycin-insensitive component of mTOR (rictor, an mTORC2 component) was reduced in TM^−/− cells transfected with TM-des-Cyto but not with wildtype TM [60]. These results suggested that the cytoplasmic domain of TM may be specifically involved in PTEN-dependent regulation of the mTORC1 signaling pathway in endothelial cells (Fig. 2).

PTEN shuttles between the cytoplasm and the nucleus. The regulated recruitment of PTEN from the cytoplasm to the plasma membrane is required for inhibition of the activation of AKT/mTORC1 signaling axis [70]. In the nucleus, PTEN is involved in regulation of the cell cycle arrest genes and chromosomal stability, thus playing critical roles in the maintenance of the quiescence and inhibition of aberrant cell proliferation and migration that can lead to tumorigenesis [71–72]. Interestingly, unlike wildtype cells, PTEN was not recruited to the plasma membrane and its transport to the nucleus was markedly decreased in TM^−/− cells [60]. Transfection of TM^−/− cells with wildtype TM construct, but not with TM-des-Cyto, reversed the enhanced C-terminal tail phosphorylation and altered sub-cellular localization of PTEN, suggesting a role for the TM cytoplasmic domain in facilitating the membrane recruitment of PTEN and regulation of the AKT/mTORC1 signaling axis in endothelial cells [60]. A correlation between TM and PTEN could also be gleaned from findings that over-expression of wildtype TM construct in endothelial cells resulted in significantly lower C-terminal tail phosphorylation of PTEN when compared to untransfected cells [60]. The phosphorylation of AKT, nuclear exclusion of FOXO1 and enhanced migration phenotype of cells in response to Ang1 and VEGF were also reduced in TM over-expressing cells [60]. The findings in the cellular systems may have physiological significance based on the observation that PTEN phosphorylation and AKT activation were both significantly enhanced in the liver and kidney tissues of TM-KO mice (60).

6. 4. Cytoplasmic juxtamembrane Arg-Lys-Lys (RKK) residues of TM

As indicated above, the three basic RKK residues of TM cytoplasmic domain have been found to interact with the actin cytoskeleton in epithelial cells through membrane-anchoring ERM proteins [50]. The function of these residues of TM in endothelial cells was investigated by transfecting TM^−/− cells with another TM-des-Cyto construct in which these RKK residues remained intact in the cytoplasmic domain of TM (TM-des-Cyto^RKK). Interestingly, transfection of TM^−/− cells with either wildtype TM construct or TM-des-Cyto^RKK, but not TM-des-Cyto rescued enhanced basal and thrombin-stimulated cell permeability in these cells [60], indicating that these basic residues of TM are involved in modulating the barrier-permeability function of endothelial cells (Fig. 2). Immunofluorescence analysis supported these results by showing that, like wildtype, TM-des-Cyto^RKK restores decreased polymerization of the cortical actin microfilaments and destabilization of VE-cadherin in TM^−/− cells [60]. However, unlike wildtype TM, TM-des-Cyto^RKK did not significantly affect the defective phenotype of TM^−/− cells with respect to constitutive release of VWF, inhibition of activation of AKT and membrane recruitment of PTEN, suggesting that other structural elements in the cytoplasmic domain of TM may be involved in modulation of the PTEN/AKT signaling axis. It is worth noting that the cytoplasmic domain of TM has four potential phosphorylation sites (Tyr552, Ser558, Thr568, and Thr571). Further studies will be required to determine whether phosphorylation of one or more residues of the cytoplasmic domain of TM contributes to the signaling function of TM in endothelial cells.

7. The potential role of TM-regulation of PTEN signaling in tumorigenesis

An inverse correlation between TM expression and tumorigenesis has been reported in numerous types of malignant tumors including colorectal, lung, prostate, bladder, melanoma, breast cancer and others [73–79]. Restoration of expression of TM within some of these tumor tissues has been found to be associated with inhibition of invasion and reduction of the transformed phenotype. In lung cancer cells, it has been reported that a decrease in the expression level of TM is associated with decreased expression of E-cadherin and increased expression of N-cadherin [80,81]. TM expression has been shown to upregulate E-cadherin and downregulate N-cadherin, thereby reversing the epithelial-mesenchymal transition (EMT) phenotype in the lung cancer cells [81]. A decreased TM expression has also been shown to result in increased cell proliferation, EMT and angiogenesis in bladder cancer in both in vitro and in vivo settings [79]. Expression of TM in the highly aggressive A375 melanoma cells has been shown to be downregulated and overexpression of TM reverses the migration properties of these cells [77]. It is not known how TM functions as a tumor suppressor. The observation that the cytoplasmic domain of TM is required for the recruitment of PTEN to the plasma membrane raises the possibility that this domain of TM possesses a binding site for interaction with PTEN, thereby facilitating its recruitment to the plasma membrane. However, co-immunoprecipitation studies have not supported a direct interaction between TM and PTEN [60]. Thus, we hypothesize that the cytoplasmic domain of TM binds cytoskeletal structures, adaptors and/or membrane-associated proteins that synergistically facilitate the recruitment of PTEN to the plasma membrane. In this context, it has been demonstrated that expression of neutral endopeptidase (NEP), another tumor suppressor receptor, is lost or downregulated in various types of malignancies [82–84]. The tumor suppressor function of NEP has been shown to require direct interaction of its cytoplasmic domain with PTEN and ERM proteins. Interestingly, like TM, NEP has three membrane proximal basic residues (RKK) that are involved in these protein-protein interactions [83,84]. Thus, it is possible that the cytoplasmic domain of TM interacts with PTEN binding partners to synergistically regulate the activity/recruitment of PTEN to the cytoplasmic membrane.

Another possibility that may account for the inverse correlation of both TM and PTEN with tumorigenesis is that common regulatory mechanisms (i.e., transcription factors, signaling networks, epigenetic mechanisms, microRNA, etc.) are involved in coordinated modulation of expression of both molecules. In this vein, it has been reported that TGFβ signaling can induce EMT and cancer metastasis through Snail/Slug family of transcription factors [81,85]. Snail is known to bind promoters of both TM and PTEN to downregulate their expression [81,85–87]. Furthermore, TGFβ has been shown to inactivate PTEN by inducing phosphorylation of the C-terminal domain of PTEN, thereby trapping the phosphatase in an inactive closed conformation [88]. Thus, activation of TGFβ pathway can downregulate expression/activity of both TM and PTEN during tumorigenesis. Krüppel-like factor 2 (KLF2), which is involved in promoting the expression of TM in endothelial cells, is another transcription factor that may regulate expression of both TM and PTEN in tumor cells [89,90]. Expression of KLF2 is downregulated in human gastric tumor [91,92]. It has been found that microRNA-32-5p binds KLF2 mRNA and downregulates expression of KLF2 and PTEN, thereby activating PI3K/AKT signaling and promoting the development of gastric cancer [91,92]. Increasing evidence suggests that microRNAs play critical roles in regulation of expression of genes involved in tumorigenesis including PTEN and TM. For instance, a tumor-associated microRNA (miR-18a-5p) that is upregulated in metastatic melanoma and endometrial cancer cells [93,94], also negatively regulates PTEN [93]. As in melanoma [76,77], TM expression has been found to be significantly downregulated in endometrial cancer cells [94]. Interestingly, the oncogenic effect of miR-18a-5p has been shown to be inhibited by overexpression of TM [94], suggesting that TM suppresses tumor phenotype by normalizing PTEN activity. There are other microRNAs known to be involved in regulation of PTEN and other tumor suppressor proteins [95–100] the discussion of which is beyond the scope of this article, but the few examples mentioned above support the hypothesis that common regulatory mechanisms may coordinately regulate PTEN and TM signaling in tumor cells.

In summary, our results indicate that the cytoplasmic domain of TM interacts with the cytoskeletal proteins to recruit PTEN and other signaling molecules to the plasma membrane, thereby playing critical roles in maintaining vascular integrity and inhibition of aberrant cellular growth and migration, biological processes that if not tightly regulated can lead to tumorigenesis. Moreover, like switching the procoagulant specificity, interaction of TM with thrombin recruits both β-arrestin-1 and β-arrestin-2 to the plasma membrane, thereby changing the PAR1 signaling specificity of thrombin. Future studies will focus on deciphering the mechanism of the ligand-dependent recruitment of signaling molecules by TM to the plasma membrane and understanding the cause-and-effect relationship between PTEN and TM in the growing number of malignancies in which both molecules are downregulated.

Addendum

HG, IB, and ARR wrote the manuscript.