Identification and characterization of two SERPINC1 mutations causing congenital antithrombin deficiency

The human anticoagulant system works with the fibrinolytic system to maintain the balance between bleeding and coagulation. Antithrombin (AT), a 432-amino-acid single-chain glycoprotein synthesized and secreted in the liver, is the most important physiological anticoagulant in plasma [1]. AT is a member of the anti-serine protease inhibitor superfamily and binds to serine residues in the active centers of thrombin and coagulation factors IXa, Xa, XIa, and XIIa (FIXa, FXa, FXIa, and FXIIa), thereby "sealing" and inactivating their active centers with a pronounced anticoagulant effect [2]. In addition, AT inhibits activated factor VIIa (FVIIa) by accelerating the dissociation of the FVIIa/tissue factor (FVIIa/TF) complex and preventing its recombination, thereby preventing coagulation initiation [3]. Physiologically, the anticoagulant activity of AT is low, but when combined with high-affinity heparin, it can be rapidly amplified 1000-fold [4]. Owing to the unique anticoagulant mechanism and extensive anticoagulant activity of AT, a mild deficiency increases the risk of thrombosis, whereas a deficiency increases the risk of venous thrombosis 5–50-fold [5]. This is a major risk factor for thrombophilia, conferring a much higher risk than protein C or protein S deficiency.

Currently, there is no consensus on the minimum AT level defining a deficiency, but AT levels of at least 70% are necessary to effectively inhibit the coagulation cascade [6]. AT deficiency can be categorized as hereditary or secondary. Hereditary AT deficiency is a rare autosomal incompletely dominant disease characterized by clinical manifestations of recurrent thrombosis, of which lower extremity deep venous thrombosis and pulmonary embolism are the most common. There is a population incidence of approximately 1/5000 to 1/500 [7, 8], but the incidence may be about 1/200 to 1/20 in patients with venous thrombosis [9]. This genetic defect is mainly caused by mutations in SERPINC1, which encodes AT. SERPINC1 is located on chromosome 1 q23.1–25, is 13.5 kb in length, and consists of seven exons and six introns [10]. Furthermore, SERPINC1 is highly susceptible to alterations, and even small changes in its nucleotide sequence can cause severe structural and functional changes promoting thrombosis [11]. Most clinical cases are heterozygous because it is difficult for homozygotes to survive, and most die during embryonic development. Homozygous deletion of SERPINC1 in mouse models has been demonstrated to result in severe thrombosis and hemorrhage, culminating in embryonic death [12, 13].

The SERPINC1 mutation profile is highly heterogeneous, with 409 SERPINC1 pathogenic mutations recorded in the HGMD database as of April 2021 (http://​www.​hgmd.​cf.​ac.​uk/​ac/​gene.​php gene = SERPINC1). Most of these mutations alter a single protein structural unit in AT, thereby disrupting its ability to inhibit coagulation [14]. Martinez-Martinez et al. found that genetic mutations may affect the mobile domains of conformational instability induced by AT, leading to protein polymerization associated with severe clinical phenotypes [15]. The variant types include point mutations, splicing, and small fragment insertion/deletion mutations [16, 17]. More than 90% of these mutations are point mutations [18]. The HGMD database showed that missense/nonsense mutations accounted for 55.0% of all mutations, frame-transfer mutations caused by small fragment deletions or insertions accounted for 26.5%, splicing mutations accounted for 7.6%, and other types of mutations accounted for 10.9%. In this study, two pedigrees with AT deficiency and recurrent venous thrombotic episodes were screened for causative genes to analyze the relationship between clinical phenotypes, gene variants, and cell function.

Congenital AT deficiency was first reported by Egeberg in 1965 [20]. AT deficiency can be divided into two types based on AT activity and plasma antigen levels. Type I: Classical deficiency, characterized by impaired AT synthesis and a parallel decrease in plasma AT antigen and activity in patients. Its clinical manifestations are severe and usually associated with severe mutations in SERPINC1, which lead to a significant decrease in the level of AT in plasma owing to mutations that destabilize mRNA, misfolded proteins, and intracellular retention, or lead to abnormal degradation [21‐23]. Type II mainly affects the domain function bound to thrombin or heparin, with normal plasma AT antigen content, but its activity is weakened [24], its clinical manifestations are mild, and its incidence is relatively high [25]. Moreover, the mutant forms are mostly missense mutations, which only affect the functional domain of AT [26‐28]. Type II deficiency can be divided into three subtypes based on the location of the mutation: type II with reactive site defects (type IIRS), type II with heparin-binding defects (type IIHBS), and type II with pleiotropic defects (type IIPE). In type IIPE deficiency, mutations are commonly found in the highly conserved C-terminal hinge region (strands 1C–5 B) of AT [29‐31]. Patients with type IIHBS have the lowest risk of VET [32]. Some patients with SERPINC1 mutations may have normal AT activity. Carriers of the close hotspot mutations p.Val295Met, p.Arg294Leu, and p.Arg294Cys in SERPINC1 have normal AT levels and anticoagulant activity in the Chinese population, as demonstrated in recombination models. However, mutation carriers have significantly increased endogenous thrombin potential. Patients with these variants have an approximately 20-fold increased risk of thrombosis [33].

Of all the patients with AT deficiency, 80% had defects in SERPINC1. Moreover, up to 20–30% of cases do not have any SERPINC1 mutations, which may be associated with genetic variants, such as proteins encoding AT transcription, chaperones involved in AT folding, and post-translational modifications [34], up to 1/3 of these cases may be associated with N-glycosylation disorders of SERPINC1 [35, 36]. N-glycosylation is an important post-translational modification associated with protein folding, function, and secretion [37]. SERPINC1 encodes 32 amino acid signal peptides and 432 amino acid maturation proteins [38, 39]. The primary function of this signal peptide is to direct nascent polypeptide chains into the endoplasmic reticulum, followed by post-translational processing, including disulfide bond formation, N-glycosylation, signal peptide cleavage, and correct folding of the protein into a metastable conformation, which is required for the inhibitory activity of this serine protease inhibitor [22]. A portion of exons 2 and 3 encodes the heparin-binding site region [40]. A portion of exon 6 encodes the C-terminal region [41]. AT has a structure similar to serine protease inhibitors: three β-folds (A–C), nine α-helices (A–I), and a reactive center loop (RCL) [42]. It has two important functional structural regions: the protease inhibitory active region at the C-terminus and heparin-binding region at the N-terminus [43, 44]. Protease-binding sites are mainly distributed between A-β folds and C-β folds, and RCL can be formed by stretching outward on the molecular surface in this region. The spatial structure of RCL is complementary to that of the substrate-binding groove in the active site of serine proteases, thus exerting its role in protease inhibition. Heparin-binding sites are mainly distributed on the basic D helix, and they can bind pentameric polysaccharides in heparin polysaccharide chains and achieve tight binding with heparin by changing their conformation [45‐48]. Lys 11, Arg 13, Arg46, Arg47, Lys 114, Lys125, and Arg129 have high affinity for heparin [49]. AT also has four glycosylation sites (Asn128, Asn167, Asn187, and Asn224) and three disulfide bonds (Cys40-Cys160, Cys53-Cys127, and Cys279-Cys462). Variations in these key residues can affect the structure and function of thrombin proteins [50]. During coagulation inhibition, RCL first identifies the substrate-binding site of the target protease, breaks its own Arg425-Ser426 peptide bond (P1-P1′), inserts the hinge of the broken RCL into the notch region of the A- β fold, and then covalently binds to Ser150 in the thrombin active center to form a stable AT-thrombin complex such that thrombin activity is lost or weakened. When the pentasaccharide sequence in the heparin molecule binds to AT, the hinge of the RCL is removed from the A-β fold, leaving the RCL completely exposed, which binds more thrombin and rapidly amplifies the anticoagulant activity of AT [51‐54].

Modeling analysis of the AT protein suggested that this frameshift mutation caused breakage of the disulfide bond between Cys279-Cys462 and affected the structure and function of the protein. Mutations involving disulfide bond breakage have previously been reported to cause AT deficiency, including Cys279-Cys462 (Cys279Trp, Cys462Phe), Cys40-Cys160 (Cys160Tyr, Cys160Term, Cys160Trp), and Cys53-Cys127 (Cys53Ser, Cys53Term, Cys127Arg) [55]. In contrast, intramolecular disulfide bonds are essential for the correct folding, structural stabilization, and secretion of AT proteins [56]. For example, p.Cys127Arg causes the rupture of disulfide bonds between Cys53-Cys127, significantly slows the rate of AT secretion, and persists in the intracellular reticulum in the form of oligosaccharides, which cause AT deficiency [57]. Mutations 13,387-9delG and c.964A > T (p.Lys322*) cause breakage of the disulfide bond between Cys279 and Cys462, impairing secretion and structural stability of AT-truncated proteins degraded intracellularly [58, 59]. In addition, AT protein truncation caused by c.964A > T (p.Lys322*) led to the loss of the P1-P1′(Arg393-Ser394) bond, which does not act as an inactivated protease [58]. The Cys462Tyr mutation also breaks the disulfide bond between Cys247 and Cys430, triggering aberrant AT aggregation and leading to impaired secretion and eventual intracellular degradation [58]. Notably, p.Leu450fsX9 and p.Val458Cys462delGly variant, which are adjacent to Asn460 on exon7, may cause thrombopoiesis and AT activity of less than 50% [60, 61]. AT activity of less than 50% has been observed in many frameshift mutations in SERPINC1. When AT activity is less than 50%, the sensitivity of AT to heparin decreases rapidly, which increases observably the risk of thrombosis. Frameshift mutations often mean “loss of function”. Frameshift mutations can cause the lack of full-length proteins and damage multiple potential functional domains of AT, easily leading to severe decline in AT activity.

Bioinformatic analysis of the p.Arg229* variant suggested that this might affect protein viability because of the truncated product, while AT position 229 loses the domain interacting with FIX. Carriage of p.Arg229* has been found in patients with hereditary AT deficiency type I (Accession Number: CM920111) [19]. Approximately 12% of inherited disease-causing mutations in humans are nonsense mutations, which may lead to the production of truncated proteins that are hypoactive or inactive as well as gain-of-function or dominant-negative effects [62]. The p.Glu271* mutation is associated with recurrent DVT, cerebral artery thrombosis, and pulmonary embolism, and the p.Arg391* mutation is associated with cerebral venous sinus thrombosis [58, 63]. Modeling analysis of the AT protein revealed that the mutated AT protein lost its domain when interacting with FIXa. Heparin-binding AT action involves two distinct mechanisms: allosteric activation and bridging [64, 65]. Among these, FIXa and FXa are the main target proteases that are rapidly inhibited by heparin through the allosteric activation of AT [66, 67]. Heparin promotes the interaction of exosites in the RCL of AT with FXa and FIXa by mediating allosteric activation of AT. Furthermore, the external binding site is a key factor in determining the response of AT to FXa and FIXa [48, 68]. Several external binding sites for FIXa and FXa are found in the region consisting of amino acids 230–310 of AT, such as Asn233, Arg235, Glu237, Tyr253, and Glu255 [69]. It has also been suggested that the extracellular binding site of AT is on the 232–255 contiguous loop, which forms a folding chain of 3 and 4C-β that can bind 143–153 and Glu74 on the autolytic loop residues of FIXa [38]. Therefore, Arg229, which is located near this region, may also belong to this category, which led us to consider the impact of such positional mutations. Tyr253 and His319 are critical sites for the external binding of AT. They specifically bind to Arg150 on FXa and FIXa, and mutations in these two locations could reduce the response of AT to FXa and FIXa by 10–200-fold [70, 71]. Adjacent exosites Tyr285 and Glu287 enhance the interaction of heparin-allosterically activated AT with target proteins [33]. This implies that Arg229 may be associated with an external binding site on the RCL of AT, and mutations at this position may reduce the ability of AT to react with FIXa and enhance the activity of FIXa. FIXa is a key factor in determining the intensity and efficiency of coagulation reactions [72, 73]. The F helix in the AT protein structure belongs to a highly conserved region and is related to the stability of AT protein conformation [74]. The AT Rouen-VI (Asn187Asp) variant of the F helix affects the stability of the A–β fold and forms polymers that are retained intracellularly [75]. The Gly228 residue is located at the end of the F helix, and when replaced, the AT protein conformation is unstable, resulting in polymer formation, and thus, is not degraded and trapped in the cell [76]. Similarly, Arg229 is also on the F helix, and mutations at this position mostly affect the stability of the AT conformation, occasionally forming polymers that may be associated with slowing the rate of RCL insertion into the A-β-fold or disruption of the stability of AT-target protein complexes [77‐79]. The heparin-induced conformational change may involve the C-terminal part of the AT molecule and not N-terminal sites alone [80], and Arg229* mutant can cause lack of C-terminal part of the AT molecule just right, which indicates that Arg229* variant may also affect functional domains HBS. The other key sites in the RCL are behind position 229. The reaction site bond of AT is located at the Arg425-Ser426 bond. The presence of Arg425 at P1 position leads to the fairly broad specificity of AT, enabling many coagulation proteinases to be inhibited. Arg229* mutant does not just affect the domain of interaction with FIX of the AT molecule, but also the whole functional domains RCL. Therefore, we hypothesize that this variant may also affect two functional domains, HBS and RCL, with loss of the corresponding functions in the protein.

To further verify the pathogenic mechanism of the two mutations, recombinant Asn460Thrfs*20 and Arg229* recombinant expression vectors were constructed and transfected into 293 T cells. The results show that AT protein expression level was diminished in 293 T cells with the Asn460Thrfs mutant, whereas the AT truncated protein was detected in 293 T cells with the Arg229* mutant; AT protein levels in cell supernatants were also significantly decreased, suggesting that AT protein secretion was impaired. We believe that the discovery of the variant at position 460 causes rupture of the disulfide bond between Cys279 and Cys462. Consequently, the AT protein cannot be properly folded to form a highly conserved spatial structure, and instead forms an unstable structure. As AT has not yet been exported from the cell, it is rapidly degraded intracellularly, resulting in a marked lack of AT in the serum of the proband [81]. In an AT-deficient pedigree study, the presence of the Ile218 mutation was found to result in misfolding and intracellular retention of AT [82], which is similar to the results of our study. The AT truncation protein formed by the Arg229* mutation may affect the anticoagulant activity of AT by losing the functional domains HBS and RCL. The AT truncated protein is retained in the cell owing to impaired secretion and is subsequently rapidly degraded by the proteasome and lysosome, resulting in a decrease in AT content in the plasma. However, bioinformatics analysis software was only used to predict the possible effect of mutation sites on the structure of the AT protein, and experimental evidence is lacking. While cell experiments only verified impaired protein synthesis and secretion, and the specific mechanism is not clear. Further studies are needed to determine which step of the protein synthesis and secretion process is affected by the mutation site.

The choice of genetic test is crucial for improving the detection rate. Using next-generation sequencing and multiple ligand probe amplification of the seven exons and flanking regions of the coding gene, up to 80% of cases of genetic defects can be identified [83]. Whole-genome sequencing may also reveal mutations in the non-coding regions of SERPINC1, such as promoters or regulatory sequences. Glycosylation defects can also be detected in SERPINC1-negative cells [41].

The challenge in managing patients with AT deficiency is to prevent potentially life-threatening thrombosis, while minimizing the risk of bleeding associated with long-term anticoagulant therapy. Primary prophylaxis is not currently recommended for asymptomatic patients, as long-term anticoagulation increases the risk of fatal bleeding and the risk of fatal VTE is low [84]. Approximately 60% of the thrombotic events in AT-deficient individuals are caused by pregnancy. Patients remain vulnerable to VTE or even fatal thrombotic events, even with adequate anticoagulation during pregnancy and the postpartum period. In high-risk settings, the use of AT replacement therapy reduces this risk [85]. Paradoxically, according to guidelines issued in 2012 by the American College of Chest Physicians, special antithrombotic agents should not be used prophylactically in women with a history of hereditary thrombophilia and pregnancy complications (grade 2C) [86]. Low-molecular-weight heparin (LMWH) has been widely recognized in the literature as a safe and effective alternative to oral anticoagulant therapy (e.g., warfarin) in early and late pregnancy. Unlike unfractionated heparin, LMWH has stronger anti-FXa activity [87]. However, owing to the presence of heparin resistance in many patients with AT deficiency, the therapeutic efficacy of LMWH has not been unanimously recognized, and there are limited data on the use of the ideal dose of LMWH replacement therapy [88].

In conclusion, two heterozygous mutations of SERPINC1, Asn460Thrfs*20 and Arg229*, were identified as the pathogenic mutations, and Asn460Thrfs*20 variant was reported firstly. It was predicted by molecular bioinformatics software that Asn460Thrfs*20 mutation caused the breakage of the intramolecular disulfide bond, further leading to AT misfold, while Alg229* mutation caused the formation of AT truncated protein, which lack of functional domain HBS and RCL. These occur at the level of protein translation. Then it was found that Asn460Thrfs*20 and Arg229* variants could cause disorderly AT synthesis, secretion or intracellular degradation in cells due to protein translation error, and finally led to a decrease in AT levels and activity. However, the specific mechanism need further experimental verification. Our study enriched the mutation database of SERPINC1 gene, provided some new theoretical basis for gene diagnosis and genetic counseling of patients with AT deficiency, and provided experimental data for subsequent exploration of molecular pathogenesis.