Biology of factor XI

The biology of coagulation factor XI (FXI) can be appreciated at multiple length scales, spanning from the nucleotide level to complex multicellular organisms. Changes at the nanoscale are often linked to changes in functionality and pathogenicity at the microscale and macroscale, emphasizing that studying physiological and pathological processes at multiple scales is crucial for a complete biological understanding of FXI.

The prominent nanoscale building block, DNA, can play a significant role in the pathology of FXI. Mutations in the F11 gene can lead to FXI deficiency, also known as hemophilia C. The F11 gene spans 23 kilobases of the long arm of chromosome 4 and comprises 15 exons.¹ FXI deficiency was first discovered in 1953 in patients who experienced excessive bleeding after tooth extraction.² Hemophilia C is rare, with 1 in 100 000 individuals being affected; however, the proportions are increased up to 8% in the Ashkenazi Jewish population. Although FXI deficiency is usually diagnosed after symptoms of excessive bleeding after traumatic injuries, surgery, or childbirth, FXI deficiency does not always correlate with a specific bleeding phenotype.³ Hemophilia C can be caused by reduced levels of FXI protein in the blood or reduced catalytic activity of the active form of FXI, FXIa.

Scaling up, the mysteries of FXI protein structure have been uncovered and proven to be homologous to the protein structure of plasma prekallikrein (PK). The evolutionary origins of FXI began with a duplication of the KLKB1 gene that encodes for PK. PK is a zymogen within the contact system, or KKS, and is activated by activated FXII (FXIIa) to form kallikrein. Kallikrein subsequently cleaves high molecular weight kininogen (HK) to liberate the vasoactive peptide bradykinin.⁴ Bradykinin mediates inflammation through inducing vascular permeability and vasodilation. Due to structural homology, almost all FXI and 75% to 90% of PK circulate in complex with HK.⁵ After the gene duplication event, FXI acquired a unique role in coagulation and thrombin generation, whereas PK maintained a primary role in kinin generation. Because of this evolutionary origin, FXI is structurally unique compared with other coagulation factors.

Although many coagulation factors are vitamin K–dependent and contain a γ-carboxyglutamate domain, FXI breaks this mold and is uniquely composed of 2 identical subunits, each including 4 apple domains (A1-A4) and 1 serine protease domain.⁶ The apple domains are tandem repeats of 90 or 91 amino acids, reinforced by cysteine disulfide bonds.⁷ The structure of zymogen FXI was crystalized by Papagrigoriou et al, revealing a unique “cup and saucer” configuration.⁸ The apple domains collectively form the “saucer,” and each consists of a single α helix that is supported by seven β sheets that curve antiparallel to one another. The α helix is secured by 2 disulfide bonds between the β-4 and β-5 strands. A1 and A2 run antiparallel to A3 and A4, separated by a 180° turn in the polypeptide chain. Altogether, the apple domains are packed into a 60 × 60 × 20 Å planar structure.⁹ The catalytic domain sits within the apple domain “saucer” and forms the “cup.” The FXI subunits are connected by a disulfide bridge at Cys321 on the A4 domains. The interface between A4 domains connecting the 2 dimers is 886 Å. Stability between subunits is supported by salt bridges between Lys331 on 1 subunit and Glu287 on the other subunit.⁸ Spatially, the A4 domains are closest, and the A2 domains of the FXI dimer are the farthest apart. Sitting between A4 and A2, the A1 and A3 domains of 1 subunit are adjacent to A1 and A3 of the opposite subunit. The FXIa structure is less defined, although low resolution electron microscopy and small angle X-ray scattering have shown large conformational changes accompanying FXI activation into FXIa.¹⁰ 

This review walks through the different domains of FXI (N- to C-terminus), highlighting FXI biology at various scales, including common DNA mutations, molecular structure, and protein–protein interactions (Figure 1), and data from animal models that have provided rationale for pharmacological targeting of FXI (Figure 2) to treat the disease. Throughout this review, the FXI amino acid numbering system will assign glutamic acid at the start of the A1 domain as residue 1. For other serine proteases mentioned in the paper, chymotrypsin numbering has been specified when used. The referenced publications were collected using the PubMed database to filter for recent work involving FXI biology largely published in 2018 or later. For FXI biology that has been well established, no date parameters were used.

The N-terminal region of FXI begins with the A1 domain (Cys¹-Cys⁸⁵).⁷ The A1 domain mediates interactions with thrombin.^11-14 A broad range of crucial residues, Ala45-Arg54 and Val59-Arg70, within the A1 domain, were the first identified as required for thrombin-mediated activation of FXI (Table 1).¹⁴ More recently, the residues Glu66, Lys83, and Gln84 in the A1 domain were found to mediate FXI-thrombin binding.^5,8,13 In concert with computationally predicted binding sites for thrombin located within the A3 and A4 domains of FXI,⁷ the interaction of FXI with thrombin opens a biochemical avenue to amplify thrombin generation via feedback activation of FXI. This pathway is unique to FXI, because thrombin is incapable of activating the FXI homolog, PK.^34,35 The relevance of the role of FXI in propagating thrombin generation is evident in the setting of low concentrations of initiators such as tissue factor.³⁶ The fact that the pathway of feedback activation of FXI may play only a minor role at high concentrations of tissue factor, such as at sites of breaches in the vasculature, may explain, in part, why FXI is only a minor component of hemostasis although representing a druggable target to limit runaway thrombin generation in disease settings.

There have been several notable mutations within the A1 domain of FXI. Using the interactive FXI variant database, the study by Harris et al highlights the identified variants in each domain of FXI. The database highlights 33 mutations in the A1, making up 11% of the total mutations identified.^37,38 Of particular interest, mutation Cys38Arg, described as the founder mutation of the French Basque population, has a high prevalence in the French and Spanish populations (Table 2). None of the 16 patients with severe or mild FXI deficiency exhibited a bleeding phenotype.³⁹ A population cohort study screened 60 000 individuals and identified that 109 individuals from 24 unrelated families carried the Cys38Arg mutation, concluding that FXI deficiency may be more common than previously thought.⁴⁰ It is important to note that the Cys38Arg mutation interferes with a disulfide bond between Cys38 and Cys32, which forms a small loop in A1. A second interesting mutation within the A1 domain, Cys28Phe, impedes proper formation of the cysteine disulfide bridge, affecting protein folding. All reported patients with this mutation exhibited a bleeding phenotype.^41,42 

The A2 domain of FXI includes residues Cys92-Cys175 and is connected to the A1 domain by a 6-residue linker peptide.^5,7 The initial function discovered for the A2 domain of FXI was in service as the binding partner for HK.¹² HK is a nonenzymatic glycoprotein that circulates in complex with FXI and PK.⁵¹ The earliest investigations of the HK binding site on FXI identified residues Phe56-Ser86 and Val59-Lys83 in the A1 domain.^15,16 However, more recent studies have followed-up and identified that HK binding is mainly mediated by A2, with A1, A3, and A4 having supporting roles.^5,24 

Domain 6 of HK, residues Asn583-Pro-Ile-Ser-Asp-Phe-Pro-Asp590, binds to FXI through a pocket on the A2 domain β-sheet.⁵ The tripeptide sequence motif, Asp-Phe-Pro (DFP), serves as the binding region of HK for the A2 domain of FXI. Within the HK DFP motif, the aspartate residue forms a salt bridge with FXI Lys103 and hydrogen bonds to Asn106, the proline residue interacts with FXI His143 in the β4–β5 loop, and the phenylalanine side chain of the peptide forms contacts with the side chains of the A2 β-sheet, specifically FXI residues Leu148, Ile146, Thr132, and Ala134.¹⁷ Mapping studies of the binding of a peptide sequence of HK spanning Ser565-Lys595 confirmed that this region of HK contained the binding domains for the A2 region of FXI, with the residues Leu579-Asn583 binding to a unique pocket formed between the FXI A2 and 3 domain.¹⁸ Several structural similarities and unique differences were observed for the binding of HK to the member of the contact system homologous to FXI, PK. The same region within HK, Phe582-Thr591, was found to mediate HK binding to the A2 domain of PK, whereas, rather than binding to the A3 domain of PK, binding of HK relied on a second exosite within the A1 domain of PK.

The interplay between HK and the A2 domain of PK and FXI has recently been unraveled in part through the development and use of the novel anti-HK antibody 3E8, which binds to domain 6 of HK and blocks PK and FXI activation by FXIIa.⁵² In mouse models, blocking the binding of HK to PK and FXI with 3E8 resulted in a reduction in HK cleavage and bradykinin generation in addition to PK and FXI activation. Interestingly, blocking HK binding to PK with 3E8 reduced the activation of FXII by way of blocking the kallikrein positive feedback loop. Along these lines, 3E8 was found to block kallikrein cleavage of HK.⁵³ Additionally, HK binding to FXI promotes binding to platelets by allosterically modifying FXI to expose A3 residues for glycoprotein Ibα (GPIbα) platelet binding,^21,22 and to endothelial cells. HK has also been shown to function as a negative regulator of nucleic acid–mediated FXI activation.⁵⁴ 

Although the region by which FXIIa interacts with FXI remains in the shadows, the interplay between FXII and the A2 domain of FXI was likewise deciphered through the use of the anti-mouse FXI antibody, 14E11. This antibody targets a highly conserved region of the A2 domain and specifically blocks FXIIa-mediated activation of FXI as well as, conversely, FXIa-mediated activation of FXII, in vitro.^13,19 Pharmacological targeting of the A2 domain of FXI with 14E11, or in some studies, the humanized version of 14E11, gruticibart, has been shown to provide benefit in mouse and nonhuman primate models of sepsis,⁵⁵ ischemic stroke,⁵⁶ myocardial infarction,⁵⁷ atherosclerosis,⁵⁸ and, recently, neuroinflammation associated with multiple sclerosis.⁵⁹ A common theme in these animal studies was that targeting the A2 domain of FXI caused a reduction in markers of inflammation, suggestive of cross talk between FXI and activation or regulation of the inflammatory pathways. Beyond acting as an effective anticoagulant when trialed in healthy donors, administration of gruticibart to patients with end-stage renal disease on chronic hemodialysis lowered inflammatory markers, including C-reactive protein in addition to reducing occlusive events requiring hemodialysis circuit exchange.^55,60 

Several mutations have been identified within the A2 domain of FXI. The type 2 mutation Glu117∗ is documented as the most common mutation in Jewish patients with FXI deficiency.^43,44 This mutation leads to the formation of a stop codon and can contribute to a bleeding phenotype. The FXI variant database reports 61 patients with this mutation, with 28 experiencing a bleeding phenotype.³⁷ Approximately half of the individuals with this mutation exhibiting a bleeding phenotype emphasized the heterogeneity in FXI deficiency symptoms. A second common mutation within the A2 domain is His127Arg, which is adjacent to the Cys122–Cys128 disulfide bond. Although the mutation is relatively common among East Asian individuals, most individuals are asymptomatic, likely because of the mutated residue arginine matching the size, hydrophobicity, and charge of the wild-type residue histidine.^37,44 

The FXI A3 domain encompasses residues Cys182-Cys265 and mediates binding with the coagulation factors thrombin,⁷ FIX,^20,61,62 and FV²⁵ as well as HK. Moreover, the A3 domain mediates FXI binding to the platelet receptor GPIbα,²² whereas platelet-derived polyphosphate binds to a cluster of basic amino acids, known as anion binding exosite 1, within the A3 domain.^63,64 This region has been shown to generally bind polyanions including heparin^26,27,32 and the nucleic acids DNA and RNA.^5,54 Together, the A3 domain serves to orchestrate FXI activation and the propagation of thrombin generation through assembling substrates with agents of catalysis, including the platelet surface, polyanions, and, more recently, skeletal muscle myosin.²⁸ 

Initial insights into the functional role of the FXI A3 domain in thrombus formation were made using the anti-FXI monoclonal antibody 1A6. This antibody binds the A3 domain and inhibits FXI activation by FXIIa, and FIX and FV activation by FXIa.^13,25,29,65 In a nonhuman primate model of thrombus formation, 1A6 dramatically limited platelet and fibrin deposition and prevented occlusion of collagen-coated vascular grafts. Moreover, 1A6 was found to abrogate downstream thrombin–antithrombin complex formation and the distal release of the platelet chemokine β-thromboglobulin, implicating a role of the FXI A3 domain in propagating thrombin generation as well as platelet activation.⁶⁶ The potential mechanisms of action by which the A3 domain promotes the activation of FXI and the procoagulant activity of FXIa were recently expanded after the creation and mapping of the unique nanobody, 2G3. This led to the discovery of a novel binding site for HK within the A3 domain of FXI. Mapping studies found that 2G3 binds Ala193-Ala205 and Leu239-Phe260 in the second, fourth, and fifth β strand of the FXI A3 domain, respectively, inhibiting FXI-mediated thrombin generation.²⁴ 

Although the residues Glu66, Lys 83, and Gln84, in the A1 domain, were found to mediate FXI-thrombin binding, computational modeling predicted that thrombin interacts with FXI A3 and A4 domain; the nearby residues Arg413 and Arg500 of thrombin exosite 2 bind to the FXI A3 residue Asp204 to catalyze FXI activation by thrombin.⁷ Upon activation, FXIa undergoes a conformational change allowing FIX access to residues within the A3 domain.^5,61 Using recombinant FXIa mutants containing specific substitutions within the sequence of the A3 domain, Sun et al identified that residues Ile183-Val191 are required for mediating activation of FIX by FXIa, while Ser195-Ile197, and Phe260-Ser265 play a complementary role in restoring complete activation.²⁰ 

The platelet surface has been shown to act as both a source and sink for FXI activity. The FXI A3 domain residues Ser248, Arg250, Lys255, Phe260, and Gln263 mediate binding to the leucine-rich repeat on the NH₂-terminal of platelet surface receptor GPIbα.^21-23 The adhesive protein von Willebrand factor likewise binds to the leucine-rich repeat of GPIbα,⁶⁷ whereas the anionic region of GPIbα containing the 3 sulfated tyrosine residues (Tyr276, Tyr278, and Tyr279) mediates binding to residue Arg233 (chymotrypsin numbering) within thrombin exosite 2.⁶⁸ The rate of FIX activation by FXIa as well as the overall activity of FXIa has been shown to be enhanced by platelets,⁶⁹ whereas pharmacological targeting of the thrombin-driven FXI feedback loop on platelets reduced vascular pathology and hypertension in a mouse model of vascular disease.⁷⁰ Furthermore, targeting the FXI A3 domain with the antibody h1A6 reduced markers of thromboinflammation, including platelet activation and endothelial cell dysfunction in an obese nonhuman primate model of hyperlipidemia.⁷¹ 

Numerous mutations have been mapped to the A3 domain, with the FXI variant database highlighting 35 mutations of interest, comprising nearly 15% of total mutations in FXI that have been identified.³⁸ The missense mutation Gln226Arg was shown to be 1 of the most common within the A3 domain.⁴⁴ There is conflicting evidence on the pathogenic nature of this mutation, with some patients having a bleeding phenotype and others being asymptomatic. The wild-type glutamine residue and mutated arginine residue structural studies have classified this mutation as a non-disease-causing polymorphism, whereas others have identified the mutation in women with heavy menstrual bleeding.^45,46 A second noteworthy mutation is Ala181Val, present next to Cys200, which begins the A3 domain and forms the disulfide bridge with Cys283.⁴⁴ This mutation is thought to potentially interfere with the correct formation of the A3 disulfide bridge (Cys182-Cys265), thereby affecting protein folding.³⁷ 

The final apple domain of FXI is the A4 domain comprised of Cys273-Cys356.⁵ The A4 domain plays a key structural role in facilitating FXI subunit dimerization. FXI dimerization dramatically enhances the rate constant for activation of FXI as well as FXIa activity as compared with a monomeric form of FXI.³⁰ The activation of FXI as a dimer involves the initial and rapid cleavage of Arg369-Ile370 on an individual subunit, creating an intermediate form of “half-activated” FXI, which still exhibits proteolytic activity toward FIX.³¹ The enzymatic activity of FXIa toward FIX is significantly enhanced, however, by the subsequent yet slower cleavage of the mirrored Arg369-Ile370 residue on the partnered subunit of the FXI dimer. Herein, the A4 domain interchain disulfide bond at Cys321 covalently connects 2 FXI monomers in their dimeric form.⁵ The residues Leu284, Ile290, and Tyr329 then play essential roles in forming the disulfide bond between the FXI monomers, whereas salt bridges formed between residues Lys331 and Glu287 stabilize the opposite subunits of FXI.³⁰ Dimerization was unaffected by swapping the A4 domain of FXI with the homologous A4 domain of PK, suggesting that perhaps the structural and sequence homology between PK and FXI is sufficiently conserved to maintain dimerization. Alternatively, this may suggest that the FXI A4 domain is involved in stabilizing yet is not requisite for the dimerization process itself, in contrast to the loss of dimerization observed when the FXI A2 or A3 domains were swapped with the PK A2 or A3 domains, respectively.⁷² 

Although the FXI A4 domain seems to take a back seat to the A2 and A3 domains in regulating dimerization, the A4 domain plays a combined and requisite role with the other apple domains to drive thrombin-mediated activation of FXI. A computational model predicted that the A4 domain coordinates with the A3 domain to form a binding partner with exosite 2 of thrombin. In particular, Asp280 and Lys301 in the A4 domain participate in ion-pair interactions with Arg409 and Asp371 of thrombin, respectively, whereas cleavage of the activation peptide connecting the A4 domain residue Arg369-Ile370 to the serine protease domain is required for activation of FXI by thrombin.⁷ Use of the A4 domain–targeted anti-human FXIa antibody, αFXI-175, verified the functional role of the A4 domain in propagating thrombin generation, in particular in the setting when low levels of tissue factor were used as the initiator.^73,74 Conversely, the activation of FXI by FXIIa was inhibited by peptides targeted to the region between Ala317 through Gly350 of the A4 domain of FXI.³¹ This is in line with the concept that FXI subunit dimerization is required for activation of FXI by FXIIa and that the activation peptide connecting the A4 and serine protease domain can also be cleaved by FXIIa as well as by FXIa itself. It appears that the ability of FXI to serve as a nexus between the contact activation and extrinsic pathways of blood coagulation is in part dependent on the function of the A4 domain.

A substantial number of F11 mutations have been mapped to changes within the A4 domain. In fact, the FXI gene database highlights 37 unique variants, making up 13.6% of identified variants in the database.^37,38 One missense mutation, Phe283Leu is particularly enriched within the Ashkenazi Jewish population. Phe283Leu and Glu117∗ within the A2 domain, alone account for 98% of cases among the Jewish population.⁴³ One study found that an individual who was severely deficient with Phe283Leu in combination with nonsense mutation Trp501∗ exhibited a bleeding phenotype.⁴⁷ The 22 patients with Phe283Leu mutation reported on the FXI variant database exhibited a range of phenotypes, with most showing mild to moderate bleeding.³⁷ Phe283Leu neighbors Leu284, which makes up part of the dimer interface, and it has been suggested that Phe283Leu affects FXI dimerization. The Phe to Leu residue change potentially changes the shape of this interface, affecting proper FXI dimer formation.⁴⁸ Mutations at the residue Glu350 have also been shown to impair FXI dimerization. Specifically, Gly350Arg, Gly350Ala, and Gly350Glu have all been identified, with most patients documented within the FXI variant database having a history of bleeding events.³⁷ The neighboring residue Tyr351 is a highly conserved residue, and it has been suggested that mutations at Tyr351 could impair dimerization in a similar way to mutations at Glu350.⁴⁹ 

Residues Ile370-Val607 comprise the serine protease or catalytic domain of FXI. The enzymatic activity of FXIa is localized within this 30 kilodalton light chain region,⁷ cleaving the various substrates chaperoned by the A1 to A4 domains of the heavy chain.⁵ FIX is the classical known, and first identified, substrate of the serine protease domain of FXIa as part of the waterfall model of coagulation. Orchestrated by binding to the A3 domain, FIX is sequentially cleaved by the FXIa protease domain at residues Arg145-Val146 and Arg180-Val181 to form the fully activated form of FIX (Table 3). The first cleavage after Arg145 leads to the formation of an FIXα intermediate, whereas the second cleavage after Arg180 forms FIXαβ. Earlier studies reported that FIX activation by FXIa proceeds without the release of the FIXα intermediate.⁷⁵ However, more recent work showed that FIXα rebinds A3 before undergoing the second cleavage to generate FIXαβ.⁷⁶ 

Since the discovery of FIX as a substrate for FXIa, a plethora of substrates has been added to the growing list of coagulation factors and vascular proteins cleaved by the serine protease domain of FXIa. The coagulation factors FXII,⁸⁰ HK,⁸¹ FX,^25,29 FV, FVIII,⁷⁷ and even FXI³¹ itself have been shown to be directly activated by FXIa, albeit at varying rates and, thus, relevance. Moreover, the serine protease domain of FXIa serves to inactivate negative regulators of thrombin generation, including TFPI.⁶⁵ The fact that many of these reactions are catalyzed by platelet-derived short-chain polyphosphate is reflective of the link between thrombin generation, FXIa activity, and the hemostatic activity of cells within the blood microenvironment.⁷⁸ The anion binding site 2 in the protease domain acts as a scaffold for the binding of polyanion catalytic agents such as polyphosphate,^63,64 heparin,³² and nucleic acids.⁵⁴ The anion binding site 2 spans basic residues Lys529, Arg530, Arg532, Lys534, and Lys538.³² Although polyphosphate has been shown to increase the reaction rate of FXIa, the enzymatic activity of FXIa is balanced through inactivation by plasma serpins, including antithrombin, C1-inhibitor, α-1 antitrypsin, protease nexin 1, and nexin 2.^32,33,65,87 The basic amino acids Arg504, Lys505, Arg507, and Lys509 located within the autolysis loop of the serine protease domain orchestrate the inhibition of FXIa by these serpins.³³ Moreover, FXIa is inactivated on the endothelium by the plasminogen activator inhibitor-1, a process that may indirectly regulate endothelial cell barrier function by way of activation of ADAM10. The delicate balance between inhibitors and activators of FXIa activity may be essential in maintaining vessel integrity. In addition to natural inhibitors, engineered inhibitors have shown promise in the treatment of thrombotic disease, with several undergoing phase 3 clinical evaluation. One such inhibitor, monoclonal antibody abelacimab, targets the catalytic domain of FXI. Upon binding, FXI is locked into zymogen form, preventing downstream FXIa activity.⁸⁸ Inhibition of FXI by abelacimab has shown to decrease incidence of postoperative venous thromboembolism compared with standard of care while being associated with low bleeding risk.⁸⁹ More recently, the phase 2 clinical trial AZALEA-TIMI 71, in which patients with atrial fibrillation at moderate to high risk of embolic stroke were randomized to receive the FXa inhibitor rivaroxaban or the anti-FXI monoclonal antibody abelacimab, was halted based on an observed safety advantage of abelacimab over rivaroxaban. Using a different mode of action, the antibody osocimab binds adjacent to the FXI active site, leading to structural rearrangement and allosteric inhibition.⁹⁰ Similar to this mechanism, the anti-FXI antibody 10C9 has been used in animal models and binds near the FXIa active site, preventing catalytic activity.^65,91 The small-molecule FXIa inhibitor asundexian also shows promise in the reduction of thrombosis will little effect on hemostasis. Asundexian outperformed the current standard of care in the treatment of atrial fibrillation and prevention of adverse cardiovascular events after myocardial infarction.^92,93 The structure-function relationships highlighted in this manuscript make it clear that binding of function-blocking antibodies on certain domains of FXI may cause conformational change that affect how other substrates bind. This is especially crucial to consider as FXI inhibitors progress through clinical trials.

The activity of the serine protease domain of FXIa extends beyond hemostatic functions to include activation of FXII, which itself activates PK leading to the cleavage of HK to generate the vasoactive peptide bradykinin. FXIa is capable of bypassing PK to directly cleave HK as a way to generate bradykinin and induce vascular permeability and inflammation.⁸¹ This is compounded by the ability of FXIa to cleave the metalloprotease ADAMTS13, reversing the anti-inflammatory role that ADAMTS13 plays in cleaving von Willebrand factor, resulting in an increased local platelet deposition at sites of inflammation.⁷⁹ Following this theme, the anti-inflammatory role of the complement protein factor H (CFH) is neutralized after degradation of CFH by FXIa. Our observation that, conversely, CFH inhibited FXI activation by either thrombin or FXIIa suggests a novel molecular link between FXI and the complement system.⁸² This in line with recent observations of cross talk between FXI and the complement regulatory protein properdin, which serves an opposing role to CFH by stabilizing vs promoting proteolytic degradation of the C3b complex, respectively. Properdin reportedly reduces the activity of autoactivated FXIa, whereas, in return, FXIa is able to cleave and inactivate properdin.⁸⁴ The role of FXIa in the immune and inflammatory response even extends to cleavage and activation of extracellular matrix–associated bone morphogenetic protein 7 in the heart, resulting in an inhibition of genes involved in inflammation and fibrosis.⁹⁴ Other proinflammatory substrates of FXIa that have been identified include β2 glycoprotein I (β2GPI),⁸³ prochemerin (chem163S),⁸⁵ and hepatocyte growth factor (HGF).⁸⁶ Although the mapping of the regions of FXI beyond the serine protease domain that mediate binding to these ligands remains to be described, collectively these new studies point to a link between FXIa generation, activity of the complement system, and potentiation of the immune response, points to be taken into consideration when designing active-site inhibitors of FXIa as therapeutic targets of cardiovascular disease.

Mutations within the serine protease domain have the potential to decrease protease activity through interference with active site conformation. Pro520Leu is a missense mutation documented in several large cohort studies and mutational databases.^37,40,41,44 This mutation leads to a reduction in FXIa catalytic activity. Based on crystal structure models, a key role has been identified for Pro520, which corresponds to the highly conserved chymotrypsin protease Pro161, in stabilizing the conformation of the active site.⁹⁵ Newer work using structural proteomics, molecular dynamics simulations, and binding assays suggests that Pro520 may also be responsible for mediating the first interaction of FXI with thrombin, followed by thrombin binding to residues within the A1 domain near the catalytic site, and lastly binding to Arg378, facilitating a multistaged process for thrombin activation of the individual subunits of the FXI dimer.⁷ The binding of thrombin to Pro520 is conserved for the FXI homolog PK, facilitating binding despite the inability of thrombin to activate PK. The relevance of the thrombin binding and activation of FXI can be inferred from reports of a bleeding phenotype in patients with a Pro520Leu mutation.³⁷ Mutation at Ala412 has been shown to impair the formation of the α-helix structure in the protease domain of FXI. Ala412Thr, Ala412Ser, and Ala430Val are all reported mutations, with the majority of the documented patients exhibiting a bleeding phenotype.³⁷ Ala412 is a buried residue connecting 2 antiparallel sheets, and substitution is thought to affect protein folding. Homology studies indicate that substitution at Ala412 could reduce serine protease stability and reduce FXI secretion.⁵⁰ 

It has been well documented that FXI activity or levels do not correlate with bleeding phenotype, complicating the understanding of the role that FXI plays in hemostasis. However, the growing list of FXI substrates could provide insight to bleeding risk in patients with FXI deficiency. Specifically, FXI inhibition of TFPI could contribute to bleeding risk through impairing TFPI antihemostatic functions, and could serve at a marker for bleeding risk.⁹⁶ One such study found that FXI deficiency in combination with high levels of TFPI increases bleeding risk.⁹⁷ A deeper understanding of the relationship between FXI, various substrates, and bleeding phenotypes may aid in identifying other markers that are predictive of bleeding in patients with hemophilia C.

Just as the 4 sections of the orchestra come together to play a symphony, each with their unique instruments and tempo, the 4 apple domains of FXI harmonize the coagulation, immune, complement, and inflammatory pathways, each with their unique ligands and kinetics. Select functions of FXI represent druggable targets to prevent thromboinflammation, sparing the physiological role that FXI plays in the maintenance of barrier function in select vascular beds. We look forward to witnessing the closing movement of the translation of FXI inhibitors as novel anticoagulants with limited bleeding complications.

This work was supported by grants from National Institutes of Health (NIH), National Heart, Lung, and Blood Institute (R01HL101972 and R01HL144113) and NIH, National Institute of Allergy and Infectious Diseases (R01AI157037).

Contribution: S.A.M., C.P., and O.J.T.M. wrote and edited the manuscript and reviewed, read, and approved the final manuscript.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

Correspondence: Samantha A. Moellmer, Oregon Health & Science University, 3303 S Bond Ave, Portland, OR 97239; email: moellmer@ohsu.edu.