Thrombomodulin (p.Cys537Stop) is released from cells by an unusual membrane insertion/leakage mechanism

Abstract

Expression of the thrombomodulin (TM) variant c.1611C>A (p.Cys537Stop) leads to the synthesis of a protein with no cytoplasmic tail and a transmembrane domain shortened by 3 amino acids (TM536). However, little is known regarding the release mechanism and properties of TM536. Using umbilical vein endothelial cells and peripheral blood-derived endothelial colony-forming cells from a heterozygous carrier of the TM536 variant as well as overexpression cell models, we demonstrated that TM536 is released from cells by an unusual mechanism. First, TM536 is inserted into the endoplasmic reticulum (ER) membrane, then, because of the low hydrophobicity of its intramembrane domain, it escapes from it and follows the conventional secretory pathway to be released into the extracellular compartment without the involvement of proteolysis. This particular secretion mechanism yields a soluble TM536, which is poorly modified by chondroitin sulfate glycosaminoglycan compared with conventionally secreted soluble forms of TM, and therefore has a suboptimal capacity to mediate thrombin-dependent activation of protein C (PC). We also showed that TM536 cellular trafficking was altered, with retention in the early secretory pathway and increased sensitivity to ER-associated degradation. As expected, activation of ER-associated degradation increased TM536 degradation and reduced its release. The expression of TM536 at the cell surface was low, and its distribution in lipid raft-like membrane microdomains was altered, resulting in low thrombin-dependent PC activation on the cell surface.

Graphical abstract

Figure 1.

Patients’ description. (A) Pedigrees of the affected family. Squares denote males; circles denote females; slash denote deceased family member; solid red and black symbols represent family members with a history of bleeding and heterozygous carriers of the THBD variant (c.1611C>A giving TM₅₃₆), respectively. The arrow indicates the proband. Plasma sTM concentrations were measured by enzyme-linked immunosorbent assay (ELISA; normal value <10 ng/mL). rFIIa was measured in the serum derived from the coagulation of plasma (normal value <30%). (B) FIIa generation assays were performed in PPP and PRP plasma from a control participant and 4 family members affected by the TM₅₃₆ mutation. PPP, platelet-poor plasma; PRP, platelet-rich plasma; rFIIa, residual FIIa.

Figure 2.

Expression and release of TM₅₃₆. TM_WT and TM₅₃₆ were overexpressed in HEK293 cells then (A) TM was measured by ELISA in cell CM after 17 hours of accumulation and in cell lysates (CM/lysate ratio reflects TM release capacity) and (B) TM cell-surface expression was measured by flow cytometry. (C) Cells were homogenized for DRM preparation, and TM and Cav1 (DRM marker) were detected by immunoblot in each fraction of the sucrose density gradient. Densitometric analysis of the immunoblots (right panels) shows TM and Cav1 distribution in the gradient as a percentage of their total amount. (D) Immunoblot detection of TM and endogenous IR in lysates of cells overexpressing TM_WT and TM₅₃₆ and on their surfaces after biotinylation and enrichment of cell-surface–exposed proteins (d- indicates the position of TM homodimers). The enrichment of cell-surface–exposed proteins is attested by the detection of IR. The proIR is intracellular and only detected in cell lysates, whereas its mature form (IR) is exposed at the cell surface. (E) Lysates from cells overexpressing TM_WT or TM₅₃₆ were left untreated (−) or deglycosylated with EndoH (+), and TM was detected using immunoblot. Values are mean ± standard deviation (SD) expressed as fold change relative to TM_WT. Statistical analyses were performed using unpaired t test in panels A,B. ∗∗∗P < .001. CaV1, Caveolin 1; IR, insulin receptor; proIR, proform of IR.

Figure 3.

Expression and release of TM from HUVEC. (A) TM was measured by ELISA in HUVEC and HUVEC₅₃₆ CM and lysates (CM/lysate ratio reflects TM release capacity). (B) TM contained in the CM of HUVEC and HUVEC₅₃₆ was concentrated by Con A or FIIa affinity purification and run on SDS-PAGE alongside sTM produced by HEK293 cells overexpressing TM₅₃₆, then TM was detected by immunoblot. (C) Expression of TM on the surface of HUVEC and HUVEC₅₃₆ was measured using flow cytometry. (D) Immunoblot detection of TM and GAPDH (loading control) in lysates of HUVEC and HUVEC₅₃₆. (E) TM was detected in the lysates of HUVEC and HUVEC₅₃₆ submitted (+) or not (−) to EndoH deglycosylation. Value are mean ± SD expressed as fold change relative to TM_WT or HUVEC. Statistical analyses were performed using an unpaired t test for panel A or Mann-Whitney test for panel C. ∗P < .05; ∗∗∗P < .001. Con A, Concanavalin A; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; OE, overexposed blot.

Figure 4.

Characterization of released sTM₅₃₆. (A) HEK293 cells overexpressing TM_WT and TM₅₃₆ were treated with TMI-5 (+; 5 μM; 17 hours), and TM was measured by ELISA in CM and cell lysates. (B) TM_WT and TM₅₃₆ were overexpressed together with inactive (−) or active (+) RHBDL2, and TM was measured by ELISA, as in panel A (inset shows RHBDL2 expression in cell lysates; Ø, no RHBDL2). (C) CS-TM₅₃₆ was overexpressed together with inactive (−) or active (+) RHBDL2, and sTM₅₃₆ in CM was detected by immunoblot. (D) Immunoblot detection of TM₅₃₆ and Flot1 (used as a cell lysate marker) contained in CM and cell lysate. Values are mean ± SD, expressed as fold change relative to TM_WT-overexpressing cells. Statistical analyses were performed using an unpaired t test for panel A or Mann-Whitney test for panel B. ∗∗P < .01; ∗∗∗P < .001. CS-TM₅₃₆, CS-devoid TM₅₃₆; Flot1, Flotilin 1.

Figure 5.

Mechanism of TM₅₃₆ release from cells. (A) CM from TM₅₃₆-overexpressing cells were left untreated (−) or deglycosylated with EndoH or PNGaseF (+), then TM was detected by immunoblot. Enzyme’s ability to deglycosylate TM₅₃₆ was confirmed using TM₅₃₆ from the cell lysate. (B) The indicated forms of TM₅₃₆ were overexpressed in HEK293 cells, then TM was measured by ELISA in the CM and cell lysates (CM/lysate ratio reflects TM release capacity). The inset shows the hydropathy scores (ΔG) of IMD. Values are mean ± SD, expressed as fold change compared with TM₅₃₆. Statistical analyses were performed using analysis of variance (ANOVA) followed by Dunnett multiple comparison tests (vs TM₅₃₆): ∗P < .05; ∗∗∗P < .001. (C) Immunoblot detection of soluble forms of TM_536C527S contained in the CM of HEK293 cells overexpressing the inactive (−) or active form (+) of RHBDL2. (D) GPMV was prepared from HEK293 cells overexpressing TM_WT and TM₅₃₆ and incubated for 5 hours. TM in CM (sTM) and GPMV lysates (mTM) was measured by ELISA. The release of TM from GPMVs is shown by the ratios of sTM/mTM from 3 independent experiments (Exp 1-3). (E) Intracellular vesicles prepared from HEK293 and HELA cells overexpressing TM_WT and TM₅₃₆ were disrupted by incubation in Na₂CO₃ or saponin, and soluble TM (sTM) and mTM levels were measured by ELISA. The release of TM in the lumen of intracellular vesicles is shown by the ratios of sTM/mTM from 5 independent experiments (Exp 1-5). Exp, experiment; mTM, membrane TM.

Figure 6.

Degradation of TM₅₃₆ by ERAD. (A) His-tagged CS-TM₅₃₆ was overexpressed with inactive (−) or active (+) RHBDL2, soluble TM was purified by immobilized metal affinity chromatography from CM, and TM and HSPA5 were detected by immunoblot. (B) The TM-HSPA5 complex was detected by ELISA in the plasma from TM₅₃₆ patients (patients) and healthy participants (controls). (C) Cells overexpressing TM_WT or TM₅₃₆ were treated or not with MG (5 μM; 17 hours) then TM in cell lystates was detected by immunoblot. (D) Cells overexpressing TM₅₃₆ were treated with MG, cell lysates were deglycosylated with EndoH, and TM was detected by immunoblot. (E) TM detected by immunoblot in lysates of HUVEC and HUVEC₅₃₆ treated with MG (5 μM; 17 hours; GAPDH, loading control). (F) Cells overexpressing TM_WT or TM₅₃₆ were treated with CL (10 μM; 17 hours) and MG, and TM was detected by immunoblot of cell lysates. (G) Cells overexpressing TM_WT or TM₅₃₆ were treated with CL, and then TM in CM and cell lysates were measured by ELISA. Values are mean ± SD expressed as fold change compared with TM_WT without CL. Statistical analyses were performed using the Mann-Whitney test. ∗∗P < .01. Arrowheads on overexposed blots (OE) indicate bands that appear with MG132 treatment. MG, MG132.

Figure 7.

Cofactor activity of TM. (A) PC activation measured on the surfaces of HUVEC and HUVEC₅₃₆. (B) HEK293 cells expressing EPCR and TM_WT or TM₅₃₉ were treated (+) or not (−) with 10 mM MβCD for 20 minutes before starting the PC activation protocol, and PC activity was measured. (C) Immunoblot detection of the indicated sTM and their CS-devoid counterparts accumulated in the CM. TM_CS+ indicates the position of the CS-modified TM, and Ø indicates the absence of the TM. (D) FIIa-dependent PC activation measured in vitro in the presence of the same amount of the indicated sTM. (E) Immunoblot detection of soluble CS-TM₅₃₆ and CS-TM₅₁₄ and their efficacy as cofactors in PC activation. Relative PC activities (mean ± SD) obtained after 30 to 50 minutes of PC substrate conversion are shown. Statistical analyses were performed using unpaired t test. ∗P < .05; ∗∗P < .01; ∗∗∗P < .001. CS-TM, CS-devoid TM; MβCD, methyl-β-cyclodextrin.