Diverse Substrate Recognition Mechanisms for Rhomboids Thrombomodulin Is Cleaved by Mammalian Rhomboids

Abstract

The rhomboids are a recently discovered family of intramembrane proteases that are conserved across evolution. Drosophila was the first organism in which they were characterized, where at least Rhomboids 1–3 activate EGF receptor signaling by releasing the active forms of EGF-like growth factors [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar, 2.Urban S. Lee J.R. Freeman M. A family of Rhomboid intramembrane proteases activates all Drosophila membrane-tethered EGF ligands.EMBO J. 2002; 21: 4277-4286Crossref PubMed Scopus (201) Google Scholar]. Subsequent work has begun to shed light on the role of these proteases in bacteria and yeast [3.Urban S. Schlieper D. Freeman M. Conservation of intramembrane proteolytic activity and substrate specificity in prokaryotic and eukaryotic rhomboids.Curr. Biol. 2002; 12: 1507-1512Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar, 4.McQuibban G.A. Saurya S. Freeman M. Mitochondrial membrane remodelling regulated by a conserved rhomboid protease.Nature. 2003; 423: 537-541Crossref PubMed Scopus (314) Google Scholar, 5.Gallio M. Sturgill G. Rather P. Kylsten P. A conserved mechanism for extracellular signaling in eukaryotes and prokaryotes.Proc. Natl. Acad. Sci. USA. 2002; 99: 12208-12213Crossref PubMed Scopus (99) Google Scholar, 6.Esser K. Tursun B. Ingenhoven M. Michaelis G. Pratje E. A novel two-step mechanism for removal of a mitochondrial signal sequence involves the mAAA complex and the putative rhomboid protease Pcp1.J. Mol. Biol. 2002; 323: 835-843Crossref PubMed Scopus (140) Google Scholar, 7.Sesaki H. Southard S.M. Hobbs A.E. Jensen R.E. Cells lacking Pcp1p/Ugo2p, a rhomboid-like protease required for Mgm1p processing, lose mtDNA and mitochondrial structure in a Dnm1p-dependent manner, but remain competent for mitochondrial fusion.Biochem. Biophys. Res. Commun. 2003; 308: 276-283Crossref PubMed Scopus (111) Google Scholar, 8.Herlan M. Vogel F. Bornhovd C. Neupert W. Reichert A.S. Processing of Mgm1 by the rhomboid-type protease Pcp1 is required for maintenance of mitochondrial morphology and of mitochondrial DNA.J. Biol. Chem. 2003; 278: 27781-27788Crossref PubMed Scopus (297) Google Scholar], but nothing is known about the function of rhomboids in vertebrates beyond evidence that the subclass of mitochondrial rhomboids is conserved [4.McQuibban G.A. Saurya S. Freeman M. Mitochondrial membrane remodelling regulated by a conserved rhomboid protease.Nature. 2003; 423: 537-541Crossref PubMed Scopus (314) Google Scholar]. Here, we report that the anticoagulant cell-surface protein thrombomodulin [9.Weiler H. Isermann B.H. Thrombomodulin.J. Thromb. Haemost. 2003; 1: 1515-1524Crossref PubMed Scopus (214) Google Scholar] is the first mammalian protein to be a rhomboid substrate in a cell culture assay. The thrombomodulin transmembrane domain (TMD) is cleaved only by vertebrate RHBDL2-like rhomboids. Thrombomodulin TMD cleavage is directed not by sequences within the TMD, as is the case with Spitz but by its cytoplasmic domain, which, at least in some contexts, is necessary and sufficient to determine cleavage by RHBDL2. These data suggest that thrombomodulin could be a physiological substrate for rhomboid. Moreover, the discovery of a second mode of substrate recognition by rhomboids implies mechanistic diversity in this family of intramembrane proteases. The rhomboids are a recently discovered family of intramembrane proteases that are conserved across evolution. Drosophila was the first organism in which they were characterized, where at least Rhomboids 1–3 activate EGF receptor signaling by releasing the active forms of EGF-like growth factors [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar, 2.Urban S. Lee J.R. Freeman M. A family of Rhomboid intramembrane proteases activates all Drosophila membrane-tethered EGF ligands.EMBO J. 2002; 21: 4277-4286Crossref PubMed Scopus (201) Google Scholar]. Subsequent work has begun to shed light on the role of these proteases in bacteria and yeast [3.Urban S. Schlieper D. Freeman M. Conservation of intramembrane proteolytic activity and substrate specificity in prokaryotic and eukaryotic rhomboids.Curr. Biol. 2002; 12: 1507-1512Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar, 4.McQuibban G.A. Saurya S. Freeman M. Mitochondrial membrane remodelling regulated by a conserved rhomboid protease.Nature. 2003; 423: 537-541Crossref PubMed Scopus (314) Google Scholar, 5.Gallio M. Sturgill G. Rather P. Kylsten P. A conserved mechanism for extracellular signaling in eukaryotes and prokaryotes.Proc. Natl. Acad. Sci. USA. 2002; 99: 12208-12213Crossref PubMed Scopus (99) Google Scholar, 6.Esser K. Tursun B. Ingenhoven M. Michaelis G. Pratje E. A novel two-step mechanism for removal of a mitochondrial signal sequence involves the mAAA complex and the putative rhomboid protease Pcp1.J. Mol. Biol. 2002; 323: 835-843Crossref PubMed Scopus (140) Google Scholar, 7.Sesaki H. Southard S.M. Hobbs A.E. Jensen R.E. Cells lacking Pcp1p/Ugo2p, a rhomboid-like protease required for Mgm1p processing, lose mtDNA and mitochondrial structure in a Dnm1p-dependent manner, but remain competent for mitochondrial fusion.Biochem. Biophys. Res. Commun. 2003; 308: 276-283Crossref PubMed Scopus (111) Google Scholar, 8.Herlan M. Vogel F. Bornhovd C. Neupert W. Reichert A.S. Processing of Mgm1 by the rhomboid-type protease Pcp1 is required for maintenance of mitochondrial morphology and of mitochondrial DNA.J. Biol. Chem. 2003; 278: 27781-27788Crossref PubMed Scopus (297) Google Scholar], but nothing is known about the function of rhomboids in vertebrates beyond evidence that the subclass of mitochondrial rhomboids is conserved [4.McQuibban G.A. Saurya S. Freeman M. Mitochondrial membrane remodelling regulated by a conserved rhomboid protease.Nature. 2003; 423: 537-541Crossref PubMed Scopus (314) Google Scholar]. Here, we report that the anticoagulant cell-surface protein thrombomodulin [9.Weiler H. Isermann B.H. Thrombomodulin.J. Thromb. Haemost. 2003; 1: 1515-1524Crossref PubMed Scopus (214) Google Scholar] is the first mammalian protein to be a rhomboid substrate in a cell culture assay. The thrombomodulin transmembrane domain (TMD) is cleaved only by vertebrate RHBDL2-like rhomboids. Thrombomodulin TMD cleavage is directed not by sequences within the TMD, as is the case with Spitz but by its cytoplasmic domain, which, at least in some contexts, is necessary and sufficient to determine cleavage by RHBDL2. These data suggest that thrombomodulin could be a physiological substrate for rhomboid. Moreover, the discovery of a second mode of substrate recognition by rhomboids implies mechanistic diversity in this family of intramembrane proteases. Based on the hypothesis that rhomboids in Drosophila and mammals might have a conserved role in EGFR signaling [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar, 10.Lee J.R. Urban S. Garvey C.F. Freeman M. Regulated intracellular ligand transport and proteolysis controls EGF signal activation in Drosophila.Cell. 2001; 107: 161-171Abstract Full Text Full Text PDF PubMed Scopus (324) Google Scholar], we initially investigated whether mammalian membrane-tethered EGFR ligands were cleaved by human RHBDL1 and RHBDL2; of the seven tested, none were cleaved by either human rhomboid (see Supplemental Data). We next searched the mouse genome sequence for single transmembrane domain proteins with the widely conserved rhomboid substrate motif, previously characterized in the Drosophila ligand Spitz [4.McQuibban G.A. Saurya S. Freeman M. Mitochondrial membrane remodelling regulated by a conserved rhomboid protease.Nature. 2003; 423: 537-541Crossref PubMed Scopus (314) Google Scholar, 11.Urban S. Freeman M. Substrate specificity of rhomboid intramembrane proteases is governed by helix-breaking residues in the substrate transmembrane domain.Mol. Cell. 2003; 11: 1425-1434Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar]. This substrate motif depends on predicted protein conformation rather than primary sequence, making the search somewhat subjective and difficult to automate. A manual search through about 50% of genes in the mouse genome annotated as having a TMD and signal peptide (approximately 1200 searched) revealed that only 12 appeared to be good candidates for Spitz-like substrates (Table S1). One of these was the anticoagulant protein thrombomodulin, which comprises a large N-terminal domain with homology to lectins and six EGF repeats; a TMD; and a highly conserved, short cytoplasmic domain with no recognizable motifs [9.Weiler H. Isermann B.H. Thrombomodulin.J. Thromb. Haemost. 2003; 1: 1515-1524Crossref PubMed Scopus (214) Google Scholar, 12.Esmon C.T. New mechanisms for vascular control of inflammation mediated by natural anticoagulant proteins.J. Exp. Med. 2002; 196: 561-564Crossref PubMed Scopus (107) Google Scholar]. The thrombomodulin TMD resembles the Spitz TMD in a number of significant respects (Figure 1A), so we tested whether human thrombomodulin could indeed be cleaved by human RHBDL1 or RHBDL2 [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar, 13.Pascall J.C. Brown K.D. Characterization of a mammalian cDNA encoding a protein with high sequence similarity to the Drosophila regulatory protein Rhomboid.FEBS Lett. 1998; 429: 337-340Abstract Full Text Full Text PDF PubMed Scopus (36) Google Scholar] when the proteins were coexpressed in mammalian cells using a previously described assay [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar]. First, C-terminally tagged thrombomodulin was assayed, and a cleaved band of expected size was detected in lysates from cells coexpressing RHBDL2, but not RHBDL1 (Figure 1B). The experiment was repeated with N-terminally tagged thrombomodulin and the extracellular domain accumulated in the medium, again in response to RHBDL2 only (Figure 1C). Finally, an antibody against the extracellular domain of thrombomodulin was used to detect the accumulation of untagged extracellular domains in the medium triggered by specifically RHBDL2 (Figure 1D). As with other rhomboid substrates [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar], thrombomodulin cleavage was insensitive to the broad-spectrum metalloprotease inhibitor batimastat and was abolished when the putative catalytic serine of rhomboid was mutated to alanine (Figures 1B–1D). Thrombomodulin was cleaved at significantly lower levels than cleavage of the Drosophila substrate Spitz (Figure 1E). To address whether this might indicate that thrombomodulin proteolysis was a nonspecific artifact caused by overexpression of RHBDL2, we reduced the levels of RHBDL2 expression over a range of 103-fold by reducing the amount of specific DNA in each transfection. 100-fold reduction of input rhomboid DNA reduced the expression of RHBDL2 to undetectable levels, but thrombomodulin cleavage was barely affected; and even when input DNA was reduced 1000-fold, cleavage was still detectable (Figures 1E and 1F). This substoichiometric requirement resembled the cleavage of Spitz by Drosophila Rhomboid-1 [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar]. These results confirm that at least in our cell culture assay, human thrombomodulin is cleaved by RHBDL2, but not RHBDL1. We tested whether RHBDL2 was specific for thrombomodulin by assaying whether another single TMD protein in the same functional clotting complex as thrombomodulin, the endothelial protein C receptor (EPCR) [14.Stearns-Kurosawa D.J. Kurosawa S. Mollica J.S. Ferrell G.L. Esmon C.T. The endothelial cell protein C receptor augments protein C activation by the thrombin-thrombomodulin complex.Proc. Natl. Acad. Sci. USA. 1996; 93: 10212-10216Crossref PubMed Scopus (456) Google Scholar], was cleaved. Under the same conditions, no EPCR cleavage was detected (Figure 1G). We also assayed RHBDL2 cleavage of Pref-1 (also known as Dlk1), a mouse protein with structural similarity to thrombomodulin [15.Smas C.M. Sul H.S. Pref-1, a protein containing EGF-like repeats, inhibits adipocyte differentiation.Cell. 1993; 73: 725-734Abstract Full Text PDF PubMed Scopus (562) Google Scholar], which was in our list of 12 candidate substrates; again, no cleavage was detected (not shown). Moreover, the TMDs of the other 10 proteins that we identified as potential rhomboid substrates were also uncleaved by mouse RHBDL2 when expressed as part of chimeric molecules analagous to the “TM” chimera in Figure 4 (not shown). This suggests that our ability to predict substrate TMDs from sequence alone is limited. More positively, taken with the other TMD proteins that are uncleaved by rhomboids [3.Urban S. Schlieper D. Freeman M. Conservation of intramembrane proteolytic activity and substrate specificity in prokaryotic and eukaryotic rhomboids.Curr. Biol. 2002; 12: 1507-1512Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar, 11.Urban S. Freeman M. Substrate specificity of rhomboid intramembrane proteases is governed by helix-breaking residues in the substrate transmembrane domain.Mol. Cell. 2003; 11: 1425-1434Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar], these results further indicate the specificity of TMD cleavage by rhomboids, suggesting the possibility that thrombomodulin cleavage might have physiological significance. The ability of RHBDL2, but not RHBDL1, to cleave thrombomodulin led us to examine whether other rhomboid proteases could also cleave thrombomodulin, by analogy to the cleavage of Drosophila Spitz by many rhomboids. Human thrombomodulin was cleaved by human and mouse RHBDL2 and the zebrafish ortholog of RHBDL2; however, it was not cleaved by Drosophila Rhomboid-1 or the bacterial rhomboid AarA, both of which cleave Spitz [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar, 3.Urban S. Schlieper D. Freeman M. Conservation of intramembrane proteolytic activity and substrate specificity in prokaryotic and eukaryotic rhomboids.Curr. Biol. 2002; 12: 1507-1512Abstract Full Text Full Text PDF PubMed Scopus (113) Google Scholar] (Figure 2A). We also tested the cleavage of mouse thrombomodulin by all of the three nonmitochondrial mouse rhomboids identifiable in the mouse genome; only RHBDL2 showed activity (Figure 2A). In all cases where cleavage occurred, it was dependent on the presence of the catalytic serine of RHBDL-2: activity was abolished when the serine was mutated to alanine (not all shown). Note that the specificity implied by our conclusion that only RHBDL2 cleaves thrombomodulin must be qualified by the fact that no substrate has yet been found for RHBDL1 and RHBDL3 (also known as ventrhoid [16.Jaszai J. Brand M. Cloning and expression of Ventrhoid, a novel vertebrate homologue of the Drosophila EGF pathway gene rhomboid.Mech. Dev. 2002; 113: 73-77Crossref PubMed Scopus (19) Google Scholar]) (our unpublished data). Both contain all the conserved catalytic residues, so we presume they are active proteases, but it remains possible that they lack proteolytic activity. A potential explanation for the inability of RHBDL1 and RHBDL3 to cleave thrombomodulin would be that the enzymes and substrates are segregated from each other within the cell. To address this, we examined the localization of HA-tagged mouse RHBDL1, 2, and 3 using protocols previously described [10.Lee J.R. Urban S. Garvey C.F. Freeman M. Regulated intracellular ligand transport and proteolysis controls EGF signal activation in Drosophila.Cell. 2001; 107: 161-171Abstract Full Text Full Text PDF PubMed Scopus (324) Google Scholar]. All three were localized in the secretory pathway, specifically the Golgi apparatus and the plasma membrane. Although there was some variability, RHBDL1 was quite restricted to the Golgi apparatus (Figure 2B, arrow) and only weakly detected at the plasma membrane; conversely, RHBDL2 was predominantly at the plasma membrane (Figure 2C); RHBDL3 was seen weakly in the Golgi apparatus (Figure 2D, arrow) but also at the plasma membrane and in dots that appear to be endosomes. GFP-tagged thrombomodulin was also located in the secretory pathway–visible in the ER, the Golgi apparatus, and the plasma membrane (Figure 2E). These data indicate that differential compartmentalization cannot account for the specificity of RHBDL2 for thrombomodulin. Rhomboids are unique among proteases responsible for shedding eukaryotic extracellular signaling domains in that instead of cleaving in the extracellular juxtamembrane region, they are intramembrane proteases that cut within the TMD [1.Urban S. Lee J.R. Freeman M. Drosophila Rhomboid-1 defines a family of putative intramembrane serine proteases.Cell. 2001; 107: 173-182Abstract Full Text Full Text PDF PubMed Scopus (489) Google Scholar]. There was insufficient cleaved extracellular domain of thrombomodulin in medium to allow a direct biochemical determination of the RHBDL2 cleavage site. We therefore mapped it by comparing the size of the cleaved, GFP-tagged N-terminal fragment with identically tagged artificial truncations of the protein. In the first series of experiments, the cleaved fragment of full-length mouse thrombomodulin in cell extracts was compared with C-terminal truncations that contained the whole N-terminal region. The fragment was larger than truncations at residue 508 but smaller than truncations at residue 528 (Figure 3A). This located the approximate site of cleavage to between residues 510 and 525 (note that we predict the TMD to run from residues 517–539). We performed a second set of experiments in which most of the thrombomodulin N terminus had been removed, resulting in smaller protein fragments and more precise size comparisons. This series gave results consistent with the first series, and the cleaved fragment was larger than 514 but indistinguishable from the 519 truncation (Figure 3B). We conclude that the site of thrombomodulin cleavage is most likely at residues 518, 519, or 520, corresponding to the top region of the TMD (Figure 3C). Importantly, this result strongly supports the conclusion that thrombomodulin is a direct substrate of RHBDL2, as rhomboids are the only proteases known to cleave within TMDs near the luminal/extracellular side. Spitz-like rhomboid substrates depend on helix-destabilizing residues in the top part of their TMDs and require reasonably hydrophilic residues in the same region [11.Urban S. Freeman M. Substrate specificity of rhomboid intramembrane proteases is governed by helix-breaking residues in the substrate transmembrane domain.Mol. Cell. 2003; 11: 1425-1434Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar]. Based on this requirement and our previous ability to abrogate cleavage with TMD mutations, we made an extensive set of mutations in the TMD and the juxtamembrane region of thrombomodulin (Figure 3D), but none of these changes prevented or substantially reduced cleavage by RHBDL2 (not shown). This inability to block RHBDL2 cleavage suggests that it might occur by a distinct mechanism to the cleavage of Spitz. This would be consistent with our observation that bacterial and Drosophila rhomboids cannot cleave thrombomodulin, whereas they cleave Spitz efficiently. There is already evidence that Spitz-type recognition is not the only mechanism that exists: the Drosophila EGFR ligand Gurken is efficiently cleaved by rhomboids but does not appearto conform to the structural constraints identified in Spitz [2.Urban S. Lee J.R. Freeman M. A family of Rhomboid intramembrane proteases activates all Drosophila membrane-tethered EGF ligands.EMBO J. 2002; 21: 4277-4286Crossref PubMed Scopus (201) Google Scholar, 11.Urban S. Freeman M. Substrate specificity of rhomboid intramembrane proteases is governed by helix-breaking residues in the substrate transmembrane domain.Mol. Cell. 2003; 11: 1425-1434Abstract Full Text Full Text PDF PubMed Scopus (202) Google Scholar]. Since the TMD mutations did not prevent thrombomodulin cleavage, a series of domain swaps and deletions were used to map the determinants that allow it to be cleaved by RHBDL2 (Figure 4). Removal of either the N-terminal portion (JC chimera) or the whole (TM + C chimera) of the extracellular domain did not abrogate cleavage. However, a chimera comprising an extracellular tag, the thrombomodulin TMD, and a cytoplasmic domain from TGFα (TM chimera), was not cleaved. This implied that unlike Spitz, the TMD of thrombomodulin is not sufficient to confer cleavage/recognition by RHBDL2. Furthermore, it demonstrated that again, unlike Spitz, the cytoplasmic C terminus of the protein is necessary. To examine the role of the cytoplasmic domain of thrombomodulin further, we tested whether it was also sufficient for RHBDL2 cleavage. Strikingly, the cytoplasmic domain of thrombomodulin was sufficient to transform the TMD of either Drosophila Delta or human TGFα—both type 1 transmembrane proteins — into RHBDL2 substrates (Figure 4, chimeras C+D, C+T). These experiments imply that the cytoplasmic domain of thrombomodulin is both necessary and sufficient for the cleavage of the thrombomodulin TMD. It is also sufficient to direct cleavage by RHBDL2 of at least two other TMDs that are not otherwise substrates. One interpretation of these data is that the cytoplasmic domains of RHBDL2 and thrombomodulin participate in the enzyme/substrate recognition mechanism. We have begun to investigate this by deleting the N terminus of RHBDL2. This significantly reduced its activity against thrombomodulin (Figure 5A), suggesting a function for the RHBDL2 cytoplasmic domain. However, substituting the N-terminal cytoplasmic domain of Drosophila Rhomboid-1 with the equivalent domain of RHBDL2 was not sufficient to transform Rhomboid-1 into an enzyme that could cleave thrombomodulin, although the chimeric rhomboid retained activity against Spitz (Figure 5B). These results support the idea that the cytoplasmic domains of thrombomodulin and RHBDL2 are involved in the recognition of the substrate, although they also suggest that other parts of RHBDL2 participate. The role of thrombomodulin in the protein Canticoagulation pathway is well established [9.Weiler H. Isermann B.H. Thrombomodulin.J. Thromb. Haemost. 2003; 1: 1515-1524Crossref PubMed Scopus (214) Google Scholar, 12.Esmon C.T. New mechanisms for vascular control of inflammation mediated by natural anticoagulant proteins.J. Exp. Med. 2002; 196: 561-564Crossref PubMed Scopus (107) Google Scholar]. It is expressed on endothelial cells that line the blood vessels where it forms a complex with the clotting factor thrombin, inhibiting thrombin's interaction with fibrinogen. At the same time, the thrombin-thrombomodulin complex activates protein C, which proteolyses the activated coagulation factors Va and VIIIa. These two activities give thrombomodulin an important anticoagulant role. Beyond this, the biology of thrombomodulin is less well understood although it has been implicated in many processes including inflammation, adhesion, tumorigenesis, and embryonic development (reviewed in [9.Weiler H. Isermann B.H. Thrombomodulin.J. Thromb. Haemost. 2003; 1: 1515-1524Crossref PubMed Scopus (214) Google Scholar]). A circulating form of thrombomodulin, shed from the cell surface, is normally present in plasma and other fluids [17.Takano S. Kimura S. Ohdama S. Aoki N. Plasma thrombomodulin in health and diseases.Blood. 1990; 76: 2024-2029Crossref PubMed Google Scholar], implying that it is cleaved under physiological conditions. But it is not known whether soluble thrombomodulin has a function or whether it is merely a marker of endothelial cell damage. Circulating products representing a variety of cleavage sites can be found in plasma [17.Takano S. Kimura S. Ohdama S. Aoki N. Plasma thrombomodulin in health and diseases.Blood. 1990; 76: 2024-2029Crossref PubMed Google Scholar]. Most correspond to proteolysis in the region between the membrane and the EGF repeats, but some are large enough potentially to correspond to intramembrane cleavage. Little is known about the proteases responsible for thrombomodulin shedding, although neutrophil-derived enzymes including elastase, proteinase-3, and cathepsin G have been implicated [17.Takano S. Kimura S. Ohdama S. Aoki N. Plasma thrombomodulin in health and diseases.Blood. 1990; 76: 2024-2029Crossref PubMed Google Scholar, 18.Boehme M.W. Deng Y. Raeth U. Bierhaus A. Ziegler R. Stremmel W. Nawroth P.P. Release of thrombomodulin from endothelial cells by concerted action of TNF-alpha and neutrophils: in vivo and in vitro studies.Immunology. 1996; 87: 134-140PubMed Google Scholar]. Our discovery that thrombomodulin is efficiently and specifically cleaved by RHBDL2, coupled with our observations reported here and previously that most TMDs are not rhomboid substrates, suggests that this cleavage may be physiologically significant. If so, this would be the first vertebrate rhomboid substrate to be discovered and would represent a new biological function for the rhomboid family of proteases, which are conserved throughout evolution [19.Wasserman J.D. Urban S. Freeman M. A family of rhomboid-like genes: Drosophila rhomboid-1 and roughoid/rhomboid-3 cooperate to activate EGF receptor signalling.Genes Dev. 2000; 14: 1651-1663PubMed Google Scholar, 20.Guichard A. Roark M. Ronshaugen M. Bier E. brother of rhomboid, a rhomboid-related gene expressed during early Drosophila oogenesis, promotes EGF-R/MAPK signaling.Dev. Biol. 2000; 226: 255-266Crossref PubMed Scopus (50) Google Scholar, 21.Koonin E.V. Makarova K.S. Rogozin I.B. Davidovic L. Letellier M.C. Pellegrini L. The rhomboids: a nearly ubiquitous family of intramembrane serine proteases that probably evolved by multiple ancient horizontal gene transfers.Genome Biol. 2003; 4: R19Crossref PubMed Google Scholar].