Functional Study of the Vitamin K Cycle Enzymes in Live Cells.

Vitamin K-dependent carboxylation, an essential posttranslational modification catalyzed by gamma-glutamyl carboxylase, is required for the biological functions of proteins that control blood coagulation, vascular calcification, bone metabolism, and other important physiological processes. Concomitant with carboxylation, reduced vitamin K (KH2) is oxidized to vitamin K epoxide (KO). KO must be recycled back to KH2 by the enzymes vitamin K epoxide reductase and vitamin K reductase in a pathway known as the vitamin K cycle. Our current knowledge about the enzymes of the vitamin K cycle is mainly based on in vitro studies of each individual enzymes under artificial conditions, which are of limited usefulness in understanding how the complex carboxylation process is carried out in the physiological environment. In this chapter, we review the current in vitro activity assays for vitamin K cycle enzymes. We describe the rationale, establishment, and application of cell-based assays for the functional study of these enzymes in the native cellular milieu. In these cell-based assays, different vitamin K-dependent proteins were designed and stably expressed in mammalian cells as reporter proteins to accommodate the readily used enzyme-linked immunosorbent assay for carboxylation efficiency evaluation. Additionally, recently emerged genome-editing techniques TALENs and CRISPR-Cas9 were used to knock out the endogenous enzymes in the reporter cell lines to eliminate the background. These cell-based assays are easy to scale up for high-throughput screening of inhibitors of vitamin K cycle enzymes and have been successfully used to clarify the genotypes and their clinical phenotypes of enzymes of the vitamin K cycle.

Structural and functional insights into enzymes of the vitamin K cycle.

Vitamin K-dependent proteins require carboxylation of certain glutamates for their biological functions. The enzymes involved in the vitamin K-dependent carboxylation include: gamma-glutamyl carboxylase (GGCX), vitamin K epoxide reductase (VKOR) and an as-yet-unidentified vitamin K reductase (VKR). Due to the hydrophobicity of vitamin K, these enzymes are likely to be integral membrane proteins that reside in the endoplasmic reticulum. Therefore, structure-function studies on these enzymes have been challenging, and some of the results are notably controversial. Patients with naturally occurring mutations in these enzymes, who mainly exhibit bleeding disorders or are resistant to oral anticoagulant treatment, provide valuable information for the functional study of the vitamin K cycle enzymes. In this review, we discuss: (i) the discovery of the enzymatic activities and gene identifications of the vitamin K cycle enzymes; (ii) the identification of their functionally important regions and their active site residues; (iii) the membrane topology studies of GGCX and VKOR; and (iv) the controversial issues regarding the structure and function studies of these enzymes, particularly, the membrane topology, the role of the conserved cysteines and the mechanism of active site regeneration of VKOR. We also discuss the possibility that a paralogous protein of VKOR, VKOR-like 1 (VKORL1), is involved in the vitamin K cycle, and the importance of and possible approaches for identifying the unknown VKR. Overall, we describe the accomplishments and the remaining questions in regard to the structure and function studies of the enzymes in the vitamin K cycle.

Evaluation of warfarin resistance using transcription activator-like effector nucleases-mediated vitamin K epoxide reductase knockout HEK293 cells.

BACKGROUND: Single nucleotide polymorphisms in the vitamin K epoxide reductase (VKOR) gene have been successfully used for warfarin dosage prediction. However, warfarin resistance studies of naturally occurring VKOR mutants do not correlate with their clinical phenotype. This discrepancy presumably arises because the in vitro VKOR activity assay is performed under artificial conditions using the non-physiological reductant dithiothreitol. OBJECTIVES: The aim of this study is to establish an in vivo VKOR activity assay in mammalian cells (HEK293) where VKOR functions in its native milieu without interference from endogenous enzymes. METHODS: Endogenous VKOR activity in HEK293 cells was knocked out by transcription activator-like effector nucleases (TALENs)-mediated genome editing. RESULTS AND CONCLUSIONS: Knockout of VKOR in HEK293 cells significantly decreased vitamin K-dependent carboxylation with vitamin K epoxide (KO) as substrate. However, the paralog of VKOR, VKORC1L1, also exhibits substantial ability to convert KO to vitamin K for carboxylation. Using both VKOR and VKORC1L1 knockout cells, we examined the enzymatic activity and warfarin resistance of 10 naturally occurring VKOR mutants that were reported previously to have no activity in an in vitro assay. All 10 mutants are fully active; five have increased warfarin resistance, with the order being W59R>L128R≈W59L>N77S≈S52L. Except for the L128R mutant, this order is consistent with the clinical anticoagulant dosages. The other five VKOR mutants do not change VKOR's warfarin sensitivity, suggesting that factors other than VKOR play important roles. In addition, we confirmed that the conserved loop cysteines in VKOR are not required for active site regeneration after each cycle of oxidation.

Abnormal hemostasis in a knock-in mouse carrying a variant of factor IX with impaired binding to collagen type IV.

BACKGROUND: Factor IX binds to collagen type IV, but this binding has no known consequence. OBJECTIVES: To determine the effect of reduced binding of FIX to collagen IV. METHODS: We constructed and characterized 'knock-in' mice containing the mutation lysine 5 to alanine (K5A) in the Gla domain of their FIX. The K5A mutation dramatically reduced the affinity of FIX for collagen type IV, but had no measurable effect on platelet binding, phospholipid binding, or in vitro clotting activity. However, K5AFIX mice had a mild bleeding tendency, despite their in vitro clotting activity being normal. Hemostatic protection from delayed rebleeding was intermediate between wild-type and hemophilia B mice (which had no detectable clotting activity); moreover, survival of K5A FIX mice after nascent clot removal was dramatically improved as compared with hemophilia B mice. Importantly, there was no detectable difference between K5AFIX and wild-type mice in either a laser-induced thrombosis model or the chromogenic FIX activity assay. In contrast, after ferric chloride injury, which exposes collagen IV as well as other basement membrane proteins, intravital microscopy revealed that vessel occlusion was significantly slower in K5AFIX mice than in wild-type mice. CONCLUSIONS: Our results indicate that the FIX molecule with decreased affinity for collagen IV has altered hemostatic properties in vivo and that the binding of FIX to collagen IV probably plays a significant functional role in hemostasis.

Transgene expression levels and kinetics determine risk of humoral immune response modeled in factor IX knockout and missense mutant mice.

Immune responses leading to antibody-mediated elimination of the transgenic protein are a concern in gene replacement for congenital protein deficiencies, for which hemophilia is an important model. Although most hemophilia B patients have circulating non-functional but immunologically crossreactive factor IX (FIX) protein (CRM+ phenotype), inciting factors for FIX neutralizing antibody (inhibitor) development have been studied in crossreactive material-negative (CRM-) animal models. For this study, determinants of FIX inhibitor development were compared in hemophilia B mice, in which circulating FIX protein is absent (CRM- factor IX knockout (FIXKO) model) or present (CRM+ missense R333Q-hFIX model) modeling multiple potential therapies. The investigations compare for the first time different serotypes of adeno-associated virus (AAV) vectors (AAV2 and AAV1), each at multiple doses, in the setting of two different FIX mutations. The comparisons demonstrate in the FIXKO background (CRM- phenotype) that neither vector serotype nor vector particle number independently determine the inhibitor trigger, which is influenced primarily by the level and kinetics of transgene expression. In the CRM+ missense background, inhibitor development was never stimulated by AAV gene therapy or protein therapy, despite the persistence of lymphocytes capable of responding to FIX with non-inhibitory antibodies. This genotype/phenotype is strongly protective against antibody formation in response to FIX therapy.

Purified vitamin K epoxide reductase alone is sufficient for conversion of vitamin K epoxide to vitamin K and vitamin K to vitamin KH2.

More than 21 million prescriptions for warfarin are written yearly in the U.S. Despite its importance, warfarin's target, vitamin K epoxide reductase (VKOR), has resisted purification since its identification in 1972. Here, we report its purification and reconstitution. HPC4, a calcium-specific antibody that recognizes a 12-aa tag, was used to purify and identify VKOR. Partial reconstitution is achieved on the column by washing with 0.4% dioleoylphosphatidylcholine/0.4% deoxycholate. Activity is completely recovered by dialysis against a buffer containing a reducing agent but lacking dioleoylphosphatidylcholine/deoxycholate. Removal of detergent from the eluted proteins apparently facilitates liposome formation. Purified recombinant VKOR with tag is approximately 21 kDa, as expected; fully active; and > 93% pure. The concentration of warfarin for 50% inhibition is the same for purified protein and microsomes. It has been reported that VKOR is a multisubunit enzyme. Our results, however, suggest that a single peptide can accomplish both the conversion of vitamin K epoxide to vitamin K and vitamin K to reduced vitamin K. This purification will allow further characterization of VKOR in relation to other components of the vitamin K cycle and should facilitate its structural determination.

Polymorphisms in the VKORC1 gene are strongly associated with warfarin dosage requirements in patients receiving anticoagulation.

BACKGROUND: Warfarin is a mainstay of therapy for conditions associated with an increased risk of thromboembolic events. However, the use of this common agent is fraught with complications and little is known regarding inter-individual variation in warfarin response. OBJECTIVE: We tested for association between single nucleotide polymorphisms (SNPs) in VKORC1 and CYP2C9 and average weekly warfarin dose required to maintain patients at their desired anticoagulation target. METHODS: The sample consisted of 93 European-American patients from anticoagulation clinics at the University of North Carolina at Chapel Hill. Data on mean weekly warfarin dose were collected over a mean treatment period of 20.6 months. ANCOVA models were used and haplotype analysis was performed. RESULTS: Three of six VKORC1 SNPs were found to be very strongly associated with the average warfarin dose required to achieve the target international normalised ratio (INR; p<0.0001). The mean weekly dose by genotype ranged from approximately 27 to 47 mg. There was no evidence for an association between either of the two CYP2C9 polymorphisms studied, CYP2C9*2 and CYP2C9*3. CYP2C9*3 was significantly (p = 0.05) associated with average warfarin dosage after adjustment for VKORC1*1173. CONCLUSIONS: These results are of considerable clinical interest and confirm recently published results regarding the role of these two genes in modifying warfarin metabolism and maintenance dosage. The consistent findings regarding the role of VKORC1 and CYP2C9 in warfarin metabolism and maintenance dosage represent a clinically useful proof of principal for the use of pharmacogenomic information in medicine and may lead to improved understanding of warfarin's actions.

The vitamin K cycle.

Post-translational modification of glutamate to gamma carboxyl glutamate is required for the activity of vitamin K-dependent proteins. Carboxylation is accomplished by the enzyme gamma glutamyl carboxylase (GGCX) which requires the propeptide-containing substrate and three co-substrates: reduced vitamin K, CO2, and O2. Most propeptides bind tightly to GGCX and all of the Glu residues that will be modified are modified during one binding event. Complete carboxylation is thus dependent upon the rate of carboxylation and the dissociation rate constant of the substrate from the GGCX enzyme. If the propeptide is released before carboxylation is complete, partially carboxylated vitamin K-dependent proteins are produced. The rate of carboxylation is mainly controlled by the level of reduced vitamin K available for the reactions while the dissociation rate constant is dependent upon both the propeptide and the Gla domain of the substrate. In addition, there are allosteric effects that increase the rate of dissociation of the fully carboxylated substrates. Carboxylation requires the abstraction of a proton from the 4-carbon of glutamate by reduced vitamin K and results in the conversion of vitamin K to vitamin K epoxide. The vitamin K epoxide must be recycled to vitamin K before it can be reused, a reaction catalyzed by the enzyme vitamin K epoxide reductase (VKOR). The gene for VKOR has recently been identified but the enzyme itself has not been purified to homogeneity. It appears, however, that most of the variability observed in patients response to warfarin may be attributed to variability in the VKOR gene.

Characteristics of recombinant W501S mutated human gamma-glutamyl carboxylase.

A mutation (W501S) in the vitamin K-dependent gamma-glutamyl carboxylase (VKC) that leads to a congenital bleeding disorder was recently discovered in two patients. To characterize the enzyme defect, recombinant VKC-W501S was expressed in and purified from insect cells. The major effect of the mutation appears to be to decrease the affinity of the carboxylase for the propeptide of its substrates. This observation agrees with recent data that place part of the propeptide binding site within residues 495-513 of VKC. Additionally, we demonstrate that the affinity between descarboxy osteocalcin (d-OC) and VKC remains unaffected by the W501S mutation. This confirms earlier data that the high-affinity site for d-OC is not located on the propeptide binding domain of VKC. Two properties of the enzyme suggest an explanation for the observation that vitamin K supplementation ameliorates the effects of the mutation: (i) since full carboxylation requires the propeptide to remain bound to the enzyme sufficiently long for full carboxylation, a reduced affinity can cause its premature release before carboxylation is complete; (ii) propeptide binding results in a decrease of the KM for vitamin K hydroquinone in wild-type, but not in mutant carboxylase, resulting in increased vitamin K requirement of affected subjects.

Complete and sustained phenotypic correction of hemophilia B in mice following hepatic gene transfer of a high-expressing human factor IX plasmid.

Therapeutic correction of hemophilia B was achieved by rapid infusion of a large-volume solution containing a high-expressing human factor IX (hFIX) plasmid into the tail vein of hemophilia B mice. hFIX circulated at therapeutic levels (1-5 micro g mL-1) in all animals for more than 1 year as determined by both species-specific antigen assay and an activated partial thromboplastin time (APTT)-based clotting assay. There was acute, transient hepatic tissue damage by the infusion procedure and no significant inhibitory anti-hFIX antibodies developed. No bleeding episode was observed during or after treatment. Immunohistochemical studies indicated that the hFIX gene was exclusively expressed in hepatocytes, and that transduced cells had readily detectable hFIX protein at 4 h postinfusion, and stainable protein persisted for up to 1 year. Repeated infusions of hFIX plasmids boosted the hFIX expression to higher levels. These results demonstrate that hemophilia B can be treated by gene transfer of naked hFIX plasmids.

Identification of a gene encoding a typical gamma-carboxyglutamic acid domain in the tunicate Halocynthia roretzi.

We report the identification of a gene capable of encoding a novel Gla (gamma-carboxyglutamic acid) protein from the tunicate Halocynthia roretzi, a primitive member of the phylum Chordata. We call this new hypothetical protein Gla-RTK; it has a Gla domain typical of human vitamin K-dependent coagulation factors, a transmembrane domain, and a receptor tyrosine kinase domain. The receptor tyrosine kinase domain is very similar to the ARK (adhesion-related kinase) family of receptor tyrosine kinases. The ARK family includes Axl, Tyro3, and c-Mer. This gene also encodes a propeptide that binds to the human gamma-glutamyl carboxylase within a range of affinities observed for mammalian propeptides. The cDNA for this putative protein is found distributed throughout the oocyte and embryo but the cDNA is apparently not transcribed except during oogenesis. One of the most interesting aspects of this hypothetical protein is that its Gla domain is highly homologous to the Gla domain of Gas6, a ligand for Axl, while its receptor tyrosine kinase domain is highly homologous to Axl.

Treatment of hemophilia B in mice with nonautologous somatic gene therapeutics.

The implantation of nonautologous cells encapsulated in immunoprotective microcapsules provides an alternative nonviral method for gene therapy. This strategy was successful in reversing the disease phenotypes of dwarfism and a lysosomal storage disease, mucopolysaccharidosis VII, in murine models. In this article we implanted transgenic hemophilic B mice with microcapsules enclosing factor IX-secreting C2C12 myoblasts to study the clinical potential of this approach in the treatment of hemophilia. Treated mice showed increased plasma factor IX levels as high as 28 ng of human factor IX per milliliter of plasma and decreased activated thromboplastin times (reduced by 20% to 29%). However, the level of factor IX decreased to baseline levels by day 7, coinciding with emergence of anti-human factor IX antibody, the titer of which increased greater than 10-fold by day 28. Monoclonal anti-CD4 antibodies were used to deplete CD4+ T cells to suppress the immune response against the recombinant factor IX. In the treated hemophilic mice, the anti-factor IX antibody response was totally suppressed to beyond day 28 accompanied by a significant decrease in activated thromboplastin time compared with that seen in untreated hemophilic mice. When the microcapsules were recovered from the intraperitoneal cavity after 38 days of implantation, the encapsulated cells continued to secrete factor IX at preimplantation levels, but both cell viability and microcapsule mechanical stability were reduced. Hence although the polymer chemistry of the microcapsules and cell viability may need to be improved for long-term delivery, nonautologous gene therapy with microencapsulated cells has been shown to be effective, at least for the short-term, in alleviating the hemophilic hemostatic anomaly. Coadministration of an immunosuppressant is effective in inhibiting antibody development against the delivered factor IX and should be considered for recipients at risk of inhibitor development.

A novel fluorescence assay to study propeptide interaction with gamma-glutamyl carboxylase.

The vitamin K-dependent gamma-glutamyl carboxylase catalyzes the posttranslational modification of select glutamate residues of its vitamin K-dependent substrates to gamma-carboxyglutamate. In this report, we describe a new fluorescence assay that is sensitive and specific for the propeptide binding site of active carboxylase. We employed the assay to make three important observations: (1) A tight binding fluorescein-labeled consensus propeptide can be used to quantify the active fraction of the enzyme. (2) The off-rate for a fluorescein-labeled factor IX propeptide was 3000-fold slower than the rate of carboxylation, a difference that may explain how carboxylase can carry out multiple carboxylations of a substrate during the same binding event. (3) We show evidence that substrate binding to the active site modifies the propeptide binding site of carboxylase. The significant (9-fold) differences in off-rates for the propeptide in the presence and absence of its co-substrates may represent a release mechanism for macromolecular substrates from the enzyme. Additionally, sedimentation velocity and equilibrium experiments indicate a monomeric association of enzyme with propeptide. Furthermore, the carboxylase preparation is monodisperse in the buffer used for our studies.

Residues Y179 and H101 of a hydrophobic patch of factor VII are involved in activation by factor Xa.

We studied factor Xa activation of human factor VII in hopes of identifying factor VII residues, not adjacent to the cleavage site, involved in this interaction. We made eight factor VIIs with single mutations (N100A, H101A, D102Q, L144A, R147A, Y179A, D186A, and F256A) and two factor VIIs with multiple mutations [MM3 (L144A/R147A/D186A) and MM4 (N100A/H101A/Y179A/F256A)]. Residues in MM3 have previously been identified as affecting factor X activation, and the residues of MM4 are located at a hydrophobic patch of factor VII on the opposite side of the catalytic domain from those in MM3. Only H101A, Y179A, and MM4 were activated significantly more slowly than the wild type. Results of our kinetic analyses showed that the catalytic efficiency of factor Xa for activation of factor VII was 176- and 234-fold higher than that for H101A andY179A, respectively. All the mutants with measurable activity had affinities for tissue factor similar to those of the wild type. The activated hydrophobic patch residues, except N100A, which is adjacent to one of the catalytic residues, had normal activities toward both a small peptide substrate and factor X. The rest of the activated mutants (except D102Q with no activity) had reduced activities toward the small substrate (except R147A) and factor X. We conclude that factor VII activation by factor Xa and factor VIIa's catalytic interaction with factor X involve different regions in the catalytic domain, and residues H101 and Y179, part of an aromatic hydrophobic patch, are specifically involved in factor Xa activation of factor VII.

Location of the platelet binding site in zymogen coagulation factor IX.

The assembly of the tenase complex on the surface of the platelet is an essential step in maintaining normal hemostasis as evidenced by the serious hemorrhagic diathesis associated with either factor IX (FIX) or factor VIII deficiencies. Understanding the regions and or residues of FIX crucial for proper binding to platelets has important clinical implications. The ability of FIX to bind activated platelets in the presence of 4 mmol/l CaCl2 was examined using electrophoretic light-scattering experiments. Wild-type FIX binds to activated platelets with dissociation constant Kd = 7.9 nmol/l. Activated FIX binds to activated platelets with Kd = 2 nmol/l. Activated factor VII does not bind activated platelets at physiological concentrations. The Gla domain of FIX is important for the binding of FIX to activated platelets since a chimera with a factor VII (FVII) template and FIX Gla [FVII(FIXGla)] has Kd = 9.6 nmol/l, and a chimera with a FVII template and FIX Gla, A and the first epidermal growth factor domain (EGF1) [FVII(FIXGla,A,EGF1)] has Kd = 9.7 nmol/l, but a chimera with a FIX template and a FVII Gla [FIX(FVIIGla)] does not bind activated platelets. Altering the fifth residue of FIX from a lysine to an alanine (Lys5<--Ala) abolishes the mutant from binding to collagen but does not affect FIX binding to the activated platelet (Kd = 9.8 nmol/l). Point mutations involved with residues 4 and 5 (Gly4<--Phe and Lys5<--no residue), residue 9 (Phe9<--Ala), residue 10 (Val10<--Lys) and residues 9-11 (Phe9<--Met, Val10<--Lys, Glu11<--Lys) do not bind to activated platelets.

Enhanced gamma-carboxylation of recombinant factor X using a chimeric construct containing the prothrombin propeptide.

Factor Xa is the serine protease component of prothrombinase, the enzymatic complex responsible for thrombin generation. Production of recombinant factor X/Xa has proven to be difficult because of inefficient gamma-carboxylation, a critical post-translational modification. The affinities of the vitamin K-dependent propeptides for the gamma-carboxylase vary over 2 logs, with the propeptide of factor X having the highest affinity followed by the propeptides of factor VII, protein S, factor IX, protein C, and prothrombin [Stanley, T. B. (1999) J. Biol. Chem. 274, 16940-16944]. On the basis of this observation, it was hypothesized that exchanging the propeptide of factor X with one that binds the gamma-carboxylase with a reduced affinity would enhance gamma-carboxylation by allowing greater substrate turnover. A chimeric cDNA consisting of the human prothrombin signal sequence and propeptide followed by mature human factor X was generated and stably transfected into HEK 293 cells, and modified factor X was purified from conditioned medium. The results indicate that on average 85% of the total factor X produced with the prothrombin propeptide was fully gamma-carboxylated, representing a substantial improvement over a system that employs the native factor X propeptide, with which on average only 32% of the protein is fully gamma-carboxylated. These results indicate that the affinity of the gamma-carboxylase for the propeptide greatly influences the extent of gamma-carboxylation. It was also observed that regardless of which propeptide sequence is directing gamma-carboxylation (factor X or prothrombin), two pools of factor X are secreted; one is uncarboxylated and a second is fully gamma-carboxylated, supporting the notion that the gamma-carboxylase is a processive enzyme.

Expression and characterization of the naturally occurring mutation L394R in human gamma-glutamyl carboxylase.

Patients with mutation L394R in gamma-glutamyl carboxylase have a severe bleeding disorder because of decreased biological activities of all vitamin K-dependent coagulation proteins. Vitamin K administration partially corrects this deficiency. To characterize L394R, we purified recombinant mutant L394R and wild-type carboxylase expressed in baculovirus-infected insect cells. By kinetic studies, we analyzed the catalytic activity of mutant L394R and its binding to factor IX's propeptide and vitamin KH(2). Mutant L394R differs from its wild-type counterpart as follows: 1) 110-fold higher K(i) for Boc-mEEV, an active site-specific, competitive inhibitor of FLEEL; 2) 30-fold lower V(max)/K(m) toward the substrate FLEEL in the presence of the propeptide; 3) severely reduced activity toward FLEEL carboxylation in the absence of the propeptide; 4) 7-fold decreased affinity for the propeptide; 5) 9-fold higher K(m) for FIXproGla, a substrate containing the propeptide and the Gla domain of human factor IX; and 6) 5-fold higher K(m) for vitamin KH(2). The primary defect in mutant L394R appears to be in its glutamate-binding site. To a lesser degree, the propeptide and KH(2) binding properties are altered in the L394R mutant. Compared with its wild-type counterpart, the L394R mutant shows an augmented activation of FLEEL carboxylation by the propeptide.

A topological study of the human gamma-glutamyl carboxylase.

gamma-Glutamyl carboxylase (GC), a polytopic membrane protein found in the endoplasmic reticulum (ER), catalyzes vitamin K-dependent posttranslational modification of glutamate to gamma-carboxyl glutamate. In an attempt to delineate the structure of this important enzyme, in vitro translation and in vivo mapping were used to study its membrane topology. Using terminus-tagged full-length carboxylase, expressed in 293 cells, it was demonstrated that the amino-terminus of the GC is on the cytoplasmic side of the ER, while the carboxyl-terminus is on the lumenal side. In addition, a series of fusions were made to encode each predicted transmembrane domain (TMD) followed by a leader peptidase (Lep) reporter tag, as analyzed by the computer algorithm TOPPRED II. Following in vitro translation of each fusion in the presence of canine microsomes, the topological orientation of the Lep tag was determined by proteinase K digestion and endoglycosidase H (Endo H) cleavage. From the topological orientation of the Lep tag in each fusion, the GC spans the ER membrane at least 5 times, with its N-terminus in the cytoplasm and its C-terminus in the lumen.