Why patients with THBD c.1611C>A (p.Cys537X) nonsense mutation have high levels of soluble thrombomodulin?

Background Recently our group has described a new autosomal dominant bleeding disorder characterized by very high plasma levels of soluble thrombomodulin (TM). The THBD c.1611C>A (p.Cys537X) mutation in heterozygous state was found in the propositus. This mutation leads to the synthesis of a truncated TM which has lost the last three amino-acids of the transmembrane domain and the cytoplasmic tail. Objective We investigated the mechanism responsible for TM shedding in endothelial cells with THBD c.1611C>A mutation. Methods Complementary DNA of TM wild type (TM-WT) was incorporated into a pcDNA3.1 vector for transient transfection in COS-1 cells. Mutagenesis was performed to create the c.1611C<A (TM1-536) mutant and 4 other TM mutants (TM1-515, TM1-525, TM1-533 and TM1-537) with a transmembrane domain having different lengths. The effect of shear stress, metalloprotease inhibitor, certain proteases and reducing agents were tested on TM shedding. Results Western blot and immunofluorescent analysis showed that TM1-536 was produced and a certain amount of TM1-536 was anchored on the cell membrane. A significantly higher levels of soluble TM was observed in the TM1-536 cell medium in comparison with TM-WT (56.3 +/- 5.2 vs 8.8 +/- 1.6 ng/mL, respectively, p = 0.001). The shedding of TM1-536 was 75% decreased in cells cultured in the presence of a metalloprotease inhibitor. No difference was observed between TM1-536 and TM-WT shedding after cell exposure to cathepsin G, elastase, several reducing agents and high shear stress (5000 s-1). Significantly higher levels of soluble TM were observed in the cell media of TM1-533, TM1-525, TM1-515 in comparison with TM-WT (p < 0.05). Conclusion The mechanism responsible for TM shedding is complex and is not completely understood: higher sensitivity of the TM1-536 to the proteolysis by metalloproteases and a defect of synthesis due to the decreased size of the transmembrane domain might explain the high levels of soluble TM in plasma of the carriers.

Introduction Thrombomodulin (TM, CD 1441) is a 557 amino acid type-1 transmembrane glycoprotein expressed at high levels on vascular endothelial cells [1]. This protein is encoded by an intronless gene (THBD) located on the chromosome 20p12-cen [2]. The mature protein have a molecular weight ranging from 70 to 100 kDa [3] and has an amino terminal lectin-like domain, a chain of 6 epidermal growth factor-like domains, a serine threonine rich region containing chondroitin sulfate glycosaminoglycan, a transmembrane domain and a short cytoplasmic tail [1,4]. The anticoagulant function of TM is mediated by its interaction with thrombin and protein C (PC). Thrombin-TM complex activates PC and thrombin-activatable fibrinolysis inhibitor (TAFI) [4,5]. Activated PC formation prevents the amplification of thrombin generation via proteolysis of activated coagulation factors V and VIII and TAFI down-regulates fibrinolysis. In addition to anticoagulant properties, TM exhibits other functions such as anti-inflammatory properties and role in embryogenesis [1].
TM also exits in a soluble form in plasma and urine, but in healthy human subjects, the circulating levels of soluble TM (sTM) are low (<10 ng/mL) [6]. Soluble TM is generated by enzymatic or chemical cleavage of the membrane bound-protein [7][8][9]. Several pathological conditions are associated with moderate increase of TM levels in plasma which has been suggested to be a marker of endothelial damage [10,11].
Recently, our group has described a family with an autosomal dominant disorder characterized by trauma-and surgery-related severe bleeding despite normal preoperative hemostasis tests. This bleeding disorder is characterized by very high levels of sTM in patient's plasma e.g. 180 times higher than normal controls [12,13]. Using whole exome sequencing, the underlying genetic defect has been identified as the THBD c.1611C>A mutation (p.Cys537Stop), which is responsible for the synthesis of a truncated TM which has lost the last three amino-acids of the transmembrane domain and the cytoplasmic tail. The aim of the present study was to investigate the mechanisms responsible for excessive TM shedding from endothelial cells of patients with THBD c.1611C>A nonsense mutation.

Western blot analysis
Thirty-six hours after transfection, the medium was removed, changed by serum-free IMDM. One night latter, the medium was recovered, aliquoted and stored at -20˚C. For cell lysis, cells were washed twice in phosphate-buffered saline (PBS, GIBCO BRL) and lysis buffer (HEPES pH 7.5 50mM, KCl 0.1M, MgCl2 2mM, Triton X-100 0.5%) containing protease inhibitors (Complete Mini, Roche, Bâle, Suisse) was added to each plates and incubated on ice 20 minutes (min). Then, the lysate was clarified with a centrifugation step and stored at -20˚C.

TM enzyme-linked immunosorbent assay (ELISA)
Media and cell lysate TM levels were quantified using the Thrombomodulin human ELISA Kit according the manufacturer's instruction (Abcam).

Enzymes and reducing agents for TM shedding experiments
Reduced glutathione (GSH), N-acetyl-L-cysteine (NAC), human leucocyte elastase and Cathepsin G from human neutrophils were purchased from Sigma-Aldrich (Saint-Louis, MI, US). GM6001 was from Calbiochem (San Diego, CA, US).

Shear stress experiments
For experiments evaluating the effect of shear stress, 1.25.10 5 COS-1 cells were cultured on 24-well plate and transiently transfected as previously described. Forty eight hours after transfection, shear stress (5000s-1) mimicking conditions in capillaries was applied to the cells during 0, 2, 5 or 10 min using cone and plate viscometer. Then, the concentration of TM was measured in supernatant.

Immunofluorescent labelling and confocal microscopy analysis
The cells were cultured on Labtek chambers (Nalge Nunc, Naperville, USA). The cells were washed three times in PBS and fixed for 15 min in PBS containing 4% paraformaldehyde at room temperature. After three PBS washes, the cells were blocked with IMDM containing 0.5% of bovine serum albumin (BSA) for 30 min. Mouse monoclonal antibodies against TM (Abcam) were diluted 1:1000 in IMDM containing 0.5% BSA and incubated for 1h at room temperature. After 3 PBS washes, cell membranes were labeled with Alexa Fluor 594 1 conjugated WGA (Wheat Germ Agglutinin conjugate, Invitrogen) diluted 5μg/mL in IMDM containing 0.5% BSA and incubated for 10 min at room temperature. Then, the chambers were washed 3 times with PBS before incubation for 30 min at room temperature with the secondary antibody Alexa Fluor 588 1 conjugated anti-mouse immunoglobulin (Invitrogen) diluted 1:800 in IMDM containing 0.5% BSA. The slides were washed 3 times in PBS before being mounted in Dako fluorescent mounting medium and images were analyzed using a Leica SP5 X microscope (Leica, Wetzlar, Germany).

Thrombin generation assay
The inhibitory effect of sTM on the coagulation capacity of normal plasma from controls was evaluated by Thrombin generation assay (TGA). Thrombin generation was measured using the Calibrated Automated Thrombography method (CAT, Stago, Asnières, France) using a low concentration of tissue factor 1pM (Innovin, Siemens, Marburg, Germany) and phospholipids vesicles 4μM (Avanti Polar Lipids, Alabaster, AL, USA) following the protocol previously described [14,15]. The main parameter of TGA used in this study was the Endogenous Thrombin Potential (ETP). ETP is the area under the thrombin generation curve. It has been previously shown that ETP is better correlated to the clinical outcome of patients with coagulation disorders compared to routine coagulation tests [16,17].

Statistical analysis
The statistical analysis was performed using GraphPad Prism Instat 3.0 software (La Jolla, CA, USA). Means (m), variances and standard error (SEM) were calculated for data sets and all results were expressed as mean +/-SEM. Comparison between mean values was performed with a Student t-test. A p value < 0.05 was considered statistically significant.

Expression of TM 1-536 (p.Cys536X) in COS-1 cells
The mutation c.1611C>A (p.Cys537X) in the TM gene is responsible for a truncated TM molecule that has lost the last three amino-acids of transmembrane domain and the cytoplasmic tail. TM-WT and TM with c.1611C>A mutation (TM 1-536 ) were transiently expressed in COS-1 cells. Forty eight hours after transfection, western blotting of cell lysates and supernatants were performed using a monoclonal anti-TM antibody. A band of 100 kDa corresponding to TM was detected in the cell lysate for both TM-WT and TM 1-536 cells ( Fig 1A). The TM antigen levels in the cell lysate measured by ELISA showed similar levels for each TM constructs (211+/-7 vs 220 +/-5 ng/mL for TM-WT and TM 1-536 respectively, p = 0.37) (Fig 1B). However, a significantly higher levels of sTM was observed in the TM 1-536 cell media in comparison with TM-WT (56.3+/-5.2 vs 8.8+/-1.6 ng/mL, respectively, p = 0.001) (Fig 1B). These results were confirmed by western blot analysis of the TM present in concentrated cell media (Fig 1C). A single band between 75-100kDa corresponding to TM 1-536 was observed. No band was observed for TM-WT. These results demonstrated that TM 1-536 was correctly synthesized by transfected COS-1 cells and partly secreted in the cell supernatant.
Cell surface expression of TM-WT and TM 1-536 was studied using immunofluorescent labeling and confocal microscopy analysis. Representative several immunofluorescence images with labelling for TM (Alexa Fluor 594 1 ) and cell membrane (Alexa Fluor 588 1 ) are presented in Fig 2. For both TM-WT and TM 1-536 cells, superimposition of Alexa Fluor 594 labelled cell membranes and Alexa Fluor 588 labelled TM revealed yellow area demonstrating the expression of TM on the cell surface. This result confirmed that some TM 1-536 was anchored on the cell membrane whilst a certain amount of the protein was secreted in the supernatant.

Secreted TM 1-536 conserved anticoagulant activity
Secreted TM activity was evaluated by measuring the decrease of thrombin generation of a normal control plasma (ETP = 1291 nM.min) by adding the same volume of supernatants from TM-WT and TM 1-536 cell cultures (Fig 1D). No significant decrease of thrombin generation was observed when 10 μL of TM-WT or non-transfected cell media were added to 80μL of normal PPP (ETP = 1269 and 1204 nM.min, respectively). A decrease of 54% was observed when TM 1-536 cell supernatant was added to normal PPP (ETP = 595 nM.min). Thus, we showed that TM 1-536 released into the cell supernatant was enzymatically active. TM 1-536 does not exhibit a higher sensitivity to the proteases cleavage and reducing agents A low concentration of sTM is physiologically present in plasma. It's generated by enzymatic cleavage and by the action of reducing agents. We hypothesized that the TM 1-536 having a shorter transmembrane domain, might have a higher sensitivity to enzymatic or chemical cleavage compared to TM-WT. Forty-four hours after transfection, the medium was removed and replaced by FCS free medium containing different concentrations of proteases (cathepsin G and elastase at 1, 2.5 and 5 μg/mL) or reducing agents (NAC 10mM, H 2 O 2 2mM and GSH 5 mM). TM release in cell supernatant was measured at 30, 60 and 90 minutes (min) after incubation at 37˚C. TM-WT and TM 1-536 were very rapidly shedded in the supernatant after incubation with elastase and Cathepsin G. A significant release of membrane TM was detected 30 min after exposure to the enzymes (Fig 3A, 3B, 3C and 3D). The majority of TM was retrieved in the first hour of incubation. The concentration of TM 1-536 in the supernatants was lower than the TM-WT concentration because a part of TM 1-536 was spontaneously released into the media before the addition of the enzymes and eliminated when the medium has been replaced by serum-free medium. Exposure of COS-1 cells to the reducing agents such as NAC, GSH and H 2 O 2 , during 90 min, increased the shedding of both TM-WT and TM   (Fig 3E and  3F). Very low concentrations of sTM were found in the control culture medium related to physiological cell turnover.

Fig 2. cell surface thrombomodulin immunofluorescent labelling and confocal microscopy analysis of transfected COS-1 cells. COS-1 cells were transiently transfected by TM-WT and TM 1-536 expression vectors and were incubated in
Labtek chambers for 48h. The immunofluorescent labeling was performed as described in the materials and methods section. Cell nuclei were labeled using DAPI (blue). Cell membranes were labelled using Alexa Fluor 594® conjugated Wheat Germ Agglutinin (red) and thrombomodulin were labelled using Alexa Fluor Fluor 588® antibodies (green). Superimposed images were obtained and revealed that both TM-WT and TM  were expressed at the cell surface of COS-1 cells. https://doi.org/10.1371/journal.pone.0188213.g002 Metalloprotease inhibitor importantly decreased the TM 1-536 shedding Several metalloproteases have been reported to be involved in ectodomain shedding of transmembrane proteins. In order to evaluate the potential role of metalloproteases in the TM 1-536 shedding, cells were cultured in medium containing GM6001 (20μM), a broad spectrum metalloprotease inhibitor (against a disintegrim and metalloproteases, ADAMs, and matrix metalloproteases, MMPs). Forty eight hours after transfection, TM release in cell supernatant was measured. A decreased of 75% of the shedding of TM 1-536 has been shown for cells cultured in presence of GM6001 ( Fig 3G). However, the concentration of sTM in cell supernatant was remained significantly higher for TM 1-536 than TM-WT. This data greatly suggests that metalloproteinase action may be one of most determinant parameter involved in TM  shedding.
Shear stress is not a major player for the shedding of TM  TM is highly expressed on the capillary vascular endothelium where exist high shear stress forces (2000 à 5000 s-1 ). We hypothesized that a short transmembrane domain might be responsible for weaker anchorage of the TM 1-536 into the cell membrane leading to increased TM shedding when exposed to high shear stress. To evaluate the potential role of shear stress on the release of TM 1-536 into the circulation, 48h after cell transfection, shear stress (5000s-1) was applied to the COS-1 cells during 0, 2, 5 or 10 min using cone and plate viscometer. Then, the concentration of sTM was determined in supernatant using ELISA method. In the cell supernatant, no significant increase of TM concentration was observed for both TM 1-536 and TM-WT before and after exposure to shear stress (p = 0.229 and 0.078, respectively) ( Fig 3H).
These data showed that the length of the transmembrane domain is a major determinant of TM anchorage into the cells. A truncated transmembrane domain leads to an abnormal and   release. Cells were treated with elastase and cathepsin G at final concentration of 1, 2.5 and 5 μg/mL for 90 minutes. (E and F) Effect of reducing agents on TM-WT and TM 1-536 release. Cells were treated with 2mM H 2 O 2 ; 10mM N-acetyl-L-cystein (NAC) and 5mM reduced glutathione (GSH) for 90 minutes. (G) Effect of GM6001, a broad-spectrum metalloproteinase inhibitor, on TM-WT and TM 1-536 release. COS-1 cells were cultured during 48h after transfection in medium containing GM6001 (20μM). Results are presented as means +/-standard error (n>3). * p<0.05 versus control without protease inhibitor. (H) Thrombomodulin shedding after shear stress exposition. Forty-eight hours after transfection by TM-WT and TM 1-536 expression vectors, COS-1 cells were exposed to shear stress (5000s -1 ) using "cone and plate" viscometer during 0, 2, 5 and 10 minutes. Thrombomodulin concentrations in the cell supernatants were measured after exposure to shear stress in order to evaluate the thrombomodulin shedding.
https://doi.org/10.1371/journal.pone.0188213.g003 weak anchorage of the protein which may spontaneously be released into the extracellular media independently of the shear stress or the effect of natural proteases.

Discussion
Most of the mutations in the THBD gene were described in patients with thrombosis or atypical hemolytic-uremic syndrome [18][19][20]. Recently, two independent groups have described a THBD c.1611C>A nonsense mutation (p.Cys537Stop) in two families with an autosomal dominant bleeding disorder [12,13]. This mutation results in very high levels of soluble TM in patient's plasma. We have previously demonstrated that extremely high plasma TM levels enhance fast APC generation and down-regulate FVa and VIIIa and thrombin generation [12,13]. In this study, we investigated the underlying mechanisms responsible for excessive TM shedding from endothelial cells of patients with THBD c.1611C>A nonsense mutation.
Individuals with THBD c.1611C>A (p.Cys537Stop) mutation have a truncated TM molecule lacking the last three amino acids of the transmembrane domain and the C-terminal intracellular domain [12]. This truncated protein might be retained or degraded into the endothelial cells, or may be highly secreted because of the missing part of the transmembrane domain. We have investigated these hypotheses by performing transient transfections in COS-1 cells followed by immunoblotting and ELISA measurements in cell lysates and supernatants. Western blotting and ELISA performed in the cell lysates confirmed that both TM-WT and TM 1-536 constructs were equally expressed in COS-1 cells after transfection. Moreover, immunofluorescence analysis showed that both TM-WT and TM  were anchored in the cell membrane and expressed at the surface of the cells. However, significantly higher levels of soluble TM were observed in the TM 1-536 cell media suggesting that the TM 1-536 was partly secreted in the extracellular medium. These in vitro findings are supported by the very high levels of TM measured from the plasma of the patient carrier of this mutation and by the decrease of the membrane-bound TM observed in the index patient's ovarian tissue [12]. In the present study, ETP was 54% decreased when TM 1-536 cell supernatant was added into the normal control plasma, demonstrating that soluble TM 1-536 exhibits anticoagulant activity. This result is in accordance with the bleeding phenotype of the patients exhibiting this mutation and it is in accordance with the structure-function relationship of the molecule, as anticoagulant activity of TM is carried by extracellular EGF domains 4, 5 and 6 [21] which are normal in TM 1-536. In addition to its anticoagulant activity mediated by APC generation, the TM/thrombin complex exhibits also antifibrinolytic activity due to TAFI activation [22]. Activation of TAFI by the TM/thrombin complex requires EGF domains 3, 4, 5 and 6 which are normal in TM 1-536 [23] . Although the sTM in the serum of patients carrier the c.1611C>A mutation exhibited a complete antifibrinolytic activity [24], the very low levels of thrombin generation observed in propositus might limit the capacity of TM/thrombin complex to activate TAFI and this activity might be probably not sufficient to counteract the anticoagulant activity of sTM.
In physiological conditions, sTM released from cell surface is detectable in plasma at very low levels [6]. Ectodomain shedding is a mechanism used by cells to selectively regulate the expression of surface molecules [25]. Sheddases involved in TM release are poorly understood but in vitro data indicated that several proteinases belonging to the MMPs and ADAMs families are involved in TM shedding [9]. We hypothesized that the very high levels of sTM observed in the patients with the c.1611C>A mutation might be explained by a higher vulnerability of the molecule to enzymatic cleavage by proteases. We showed that TM 1-536 released from COS-1 cells was greatly decreased in the presence of GM6001, a broad-spectrum metalloproteinase inhibitor. This data emphasize the role of these proteases in TM 1-536 cleavage. We can hypothesize that the missing intracellular and transmembrane domains may lead to a higher vulnerability of TM  to the effect of the metalloproteinases compared to TM-WT. Recently, it has been showed that TM is a specific substrate of transmembrane serine protease known as rhomboid-like-2 (RHBDL2; [26,27]). However, the cleavage of TM 1-536 by the RHBDL2 protease was not explored because the intracellular domain of TM is mandatory for the recognition of the molecule by RHBDL2 and this particular domain is lacking in TM 1-536 [26] . Finally, several inflammatory process are associated with moderate, but statistically significant, increase of TM levels in plasma [12,13] due to the proteolysis of TM by different leukocytes-derived proteases (elastase and cathepsin) and redox-mechanisms [9,28]. Our experiments showed no significant difference of TM shedding between the TM 1-536 and TM-WT, after exposure of the cells to cathepsin G, elastase and several reducing agents.
We also explored the potential role of shear stress in shedding of this abnormal protein. TM is highly expressed on the capillary vascular endothelium, where exist high shear stress forces (2000 à 5000 s-1 ) [29]. The introduction of a negatively charged C-terminus into the hydrophobic lipid bilayer may interfere with stable anchorage of TM to the cell membrane. Our results demonstrated that TM 1-536 was spontaneously secreted into the cell medium before exposure of the cells to shear stress. After exposure, no significant difference was observed between TM 1-536 and TM-WT .
In this study, we explored the relationship between the size of the transmembrane domain and the shedding of TM. A high levels of sTM in cell media was observed for all mutants with a transmembrane domain shortened by at least three amino acids. These results showed that TM shedding was dependent on the presence or absence of an entire transmembrane domain. During its biosynthesis into the endothelial cells, the presence of a truncated transmembrane domain might lead to a certain cell localization defect, therefore a part of the TM might be spontaneously secreted, and not to be properly anchored on the cell membrane.
The main limitation of our work is the use of COS-1 cells which are not constitutively expressing TM. These preliminary data need to be confirmed using an endothelial cell line such as HUVECs.
In conclusion, in this study the pathogenesis of the gain-of-function THBD c.1611C>A (p. Cys573X) nonsense mutation was studied. Our results show that the mechanism responsible for TM  shedding is complex and involves at last two processes i.e a higher sensitivity to proteolysis induced by metalloproteases and a defect of biosynthesis due to the abnormal transmembrane domain.