Evolution of the Contact Factors
Genomic DNA sequences for multiple organisms are now available, facilitating studies of the natural histories of complex processes such as blood coagulation. This section covers evolution of the contact factors. For the purpose of discussion, homologs are proteins descended from a common ancestor, of which there are two types. Orthologs are direct descendants usually with the same function, while paralogs result from gene duplications and may have similar or different functions. A schematic diagram of vertebrate phylogeny is shown in Fig. 175-4. Orthologs for the proteases factors II, VII, IX, and X, and cofactors factors V and VIII are present in the genomes of teleosts (boney fish) such as the puffer fish (Fugu rubripes)191 and zebra fish (Danio rerio).192 Orthologs for factor XI, factor XII and PK are not found in fish.192 This implies that the original mechanism for factor IX activation was independent of factor XIa, and probably involved factor VIIa/TF.
The Contact Factors and Vertebrate Evolution. Shown is a cladogram depicting relationships between major classes of vertebrates, with line junctions indicating relative points of divergence of major groups (time running from left to right). The presence (+) of genes for the contact factors and their precursors in vertebrate genomes are shown at the top of the diagram. Abbreviations: HGFA - Hepatocyte growth factor activator, XII - factor XII, HGF - hepatocyte growth factor, XI - factor XI, PK - prekallikrein, and PK/XI Predecessor - the ancestor of PK and FXI. Numbers in the chart indicate the following: (1) reptilian ancestors of birds had the factor XII gene, but it has been lost along the lineage leading to birds, (2) the PK/FXI predecessor gene has undergone duplication, creating genes for factor XI and PK, and (3) in Cetaceans, such as whales, factor XII is absent from plasma because the factor XII gene has been converted to a pseudogene by a point mutation. The question mark for HGFA in the zebra fish column indicates that an ortholog for this protein was not identified during a recent search, although it is assumed to be present.
Factor XI and PK are paralogs; the products of a duplication involving a common ancestral gene. Both proteins have four N-terminal disulfide bond-constrained repeats called apple domains (discussed in the section on Factor XI ).193–195 Apple domains are members of the PAN (Plasminogen, Apple, Nematode) domain family, with homology to the N-terminal domains of plasminogen and hepatocyte growth factor (HGF) (Fig. 175-5).196 HGF is found in the genomes of all vertebrates, and it is likely that the factor XI and PK apple domains are descended from an HGF-like protein. A gene coding for a trypsin-like protease with four PAN domains is found in the frog (Xenopus), green anole lizard (Anolis carolinensis), and chicken ( Gallus gallus).192,197 The chicken gene is flanked by a cytochrome P-450 and a melatonin receptor gene, and the same genes flank the factor XI and PK loci in humans. Thus, the human genes are derived from an ortholog of the frog, lizard, and chicken genes, which we will refer to as the PK/FXI predecessor. A PK/FXI predecessor gene also is found in the egg-laying mammal the platypus (Ornithorrhynchus anatinus), while distinct genes for factor XI and PK are clearly identifiable in the genome of the opposum (Monodelphis domestica), a marsupial, as well as in placental (eutherian) mammals.192 The duplication of the gene for the PK/FXI predecessor that gave rise to the factor XI and PK genes, then, likely occurred during mammalian evolution, probably after monotremes diverged from the lineage leading to marsupials and eutherian mammals.
Domain Structure of the Contact Factors. Shown are schematic diagrams depicting the relative positions of the major domains in the contact factors and their evolutionary predecessors. The domain abbreviations used are: PAN - plasminogen, apple, nematode; K - Kringle, EGF - epidermal growth factor; F1 - fibronectin type I; and F2 - fibronectin type II. The apple (PAN) domains of prekallikrein and factor XI are probably descendents from the N-terminal PAN domain of hepatocyte growth factor (HGF) or plasminogen. The factor XII gene was created by a duplication event involving the gene for hepatocyte growth factor activator (HGFA).
Factor XII is a paralog of hepatocyte growth factor activator (HGFA) (Fig. 175-5).192 Like its substrate HGF, HGFA is present in all vertebrates. An ortholog for the factor XII gene first appears in amphibians, and is found in mammals, but is missing from the chicken genome.192 An analysis of the region surrounding the HGFA and factor XII genes sheds light on this situation. In frogs and mammals the genes flanking those for HGFA are paralogs of those flanking factor XII, demonstrating that the factor XII gene is the result of duplication of the chromosomal segment containing the HGFA gene, followed by translocation to a different chromosome. The duplicated region is similar in the chicken, except that the factor XII gene is missing from between the flanking genes found in amphibians and mammals. Thus, the factor XII gene was lost during evolution along the lineage leading to birds. This is consistent with the observation that plasmas from birds lack factor XII activity.198–200 Factor XII gene expression has also been lost in Cetaceans (whales and porpoises) due to mutations that converted the gene into a pseudogene in a common ancestor of this order.201,202
Zhou et al. recently presented an analysis of the complex phylogency of vertebrate kininogen genes.203 Kininogens are the precursors of BK, and are ubiquitous in vertebrates. Kininogen genes in amphibians appear to be related to an original kininogen gene in early vertebrates, which has been lost in other modern vertebrate groups. Kininogens in fish, birds, and mammals are related to a paralog of this original gene, which has become a pseudogene in amphibians. Three types of kininogens have been identified in mammals. HK and low molecular weight kininogen (LK) are products of alternatively spliced messages from the K kininogen gene, while T kininogen is the product of a gene found in rats that is the result of a duplication event that occurred after the divergence of rats and mice. Of relevance to the topic of contact activation is the observation that a histidine rich binding region necessary for binding to surfaces (discussed below) is found in some amphibian kininogen genes.
In summary, factor XII and the PK/FXI predecessor first appeared in tetrapods, and may represent an adaptation to a terrestrial environment. While formation of the genes for these proteins likely involved separate events, it is possible they were linked. Indeed, the PK/FXI predecessor is related to HGF to about the same extent as factor XII is to HGFA. The observation that the histidine-rich domain of kininogens also appeared in amphibians is consistent with development of a system involving these proteins in amphibians. While factor XIIa may have been an activator of the PK/FXI predecessor, and subsequently PK and factor XI, through much of vertebrate history, the absence of factor XII in birds and Cetaceans suggests alternative mechanisms for activation of these proteins co-existed with factor XII. While we lack experimental evidence for a role for the PK/FXI predecessor in hemostasis, a comparison with mammalian factor XI (discussed under Factor XI ) suggests that it lacks features required to promote factor IX activation. If the PK/FXI predecessor was not a coagulation protease (i.e. an activator of factor IX), then factor XII may not have served a role in coagulation until the appearance of factor XI in a common ancestor of marsupial and eutherian mammals.
Factor XI (Human Protein Reference Database 07524) is the zymogen of the trypsin-like protease factor XIa.204,205 It circulates in blood plasma at a concentration of 20 to 45 nM (3 to 7 μg/ml)204,206,207 in a noncovalent complex with HK.208 Liver is the major source of plasma factor XI. Factor XI deficiency has been acquired,209,210 and been corrected,211,212 by liver transplantation. Factor XI protein and mRNA have also been identified in human pancreatic islets and kidney tubules.213,214 Factor XI mRNA is present in human platelets, but it is not known if factor XI is synthesized in megakaryoctyes or platelets.215,216 Human factor XI is 58 percent identical in amino acid sequence to its paralog PK.193–195 The human factor XI217 and PK218 genes are in a head to tail orientation on the distal end of the long arm of chromosome 4 (4q34-35). The 3′ end of exon 15 of PK is ~8 kb from exon 1 of factor XI,219 consistent with a duplication event involving the PK/FXI predecessor gene described in the previous section.192 Exon 1 of the factor XI gene is not translated. Exon 2 encodes a signal peptide, exons 3-10 the apple domains and exons 11-15 the protease domain.217 Hepatocyte specific expression of human factor XI requires the transcription factor hepatocyte nuclear factor 4α (HNF4α),219 which is also expressed in pancreatic islets and kidney, explaining the presence of factor XI in these tissues. Mice with a liver-specific knockout of HNF4α have reduced factor XI expression.220
Factor XI Subunit Structure.
cDNA and amino acid sequences for factor XI are available for multiple species, facilitating structure-function analyses. Human factor XI is a 160 kDa glycoprotein consisting of identical 80 kDa (607 amino acid) (Fig. 175-6) subunits193,194,204 connected by a disulfide bond.221,222 The dimeric configuration is unusual for a trypsin-like enzyme, and unique among coagulation proteases. Each subunit contains four 90 to 91 amino acid apple domains (A1 to A4) at the N-terminus, and a C-terminal protease domain.193,194 Our understanding of factor XI has been greatly facilitated by availability of a 2.9 Å crystal structure for the zymogen,132 and structures for the isolated protease domain in complex with inhibitors.223–225 Each apple domain consists of seven β-strands folded into a curved antiparallel sheet cradling an α-helix (Fig. 175-7A). This core topology is equivalent to PAN domains from HGF (Fig. 175-7B),196 leech anti-platelet protein (Fig. 175-7C),226 micronemal proteins from apicomplexan protozoans such as EtMIC5 in Eimeria Tenella (Fig. 175-7D),227 and AMA1 in Plasmodium falcipurum.228 The PAN domains shown in Fig. 175-7 have two disulfide bonds attaching the helix to the central β4 and β5 strands, and all except HGF have a disulphide bond linking the N and C termini. The β4-β5 loop and the β5-β6 crossover loop form a small pocket on the opposite side of the β-sheet from the α-helix. The apple domains of each factor XI subunit are packed into a flat 60 x 60 x 20 Å disk-like structure, with A1 and A2, and A3 and A4, arranged in tandem (Fig. 175-8A). The polypeptide takes a 180° turn between A2 and A3 so that A3 and A4 run anti-parallel to A1 and A2. The protease domain, which has a typical serine-protease chymotrypsin-like topology, rests on the apple domain disk in a "cup and saucer" arrangement (Fig. 175-8B). Within the disk, two types of interfaces are formed. Side interfaces (A1-A2 and A3-A4) bury substantial surface areas of 441 Å2 and 444 Å2, respectively. End interfaces (A1-A4 and A2-A3) bury 380 Å2 and 284 Å2, respectively, and are structurally less similar to each other than the side interfaces. In the factor XI dimer the planes of the apple domain disks of the two subunits are inclined at a 70° angle relative to each other (Fig. 175-9 center). One result of this arrangement is that the two A3 domains are on opposite sides of a major longitudinal plane of the molecule (Fig. 175-9 top and bottom), a feature that has implications for binding interactions (discussed under Factor XI binding to platelets ). The factor XI apple domain disk contains binding sites for macromolecules involved in protease function and regulation (Fig. 175-9) that are discussed in the following sections.
Human Factor XI - Amino Acid Sequence and Disulfide Bond Locations. Schematic diagram of the primary structure of factor XI as reported by Fujikawa et al.193 The apple domains are designated A1 to A4. Cysteine residues involved in disulfide bonds within the factor XI subunit are shown in yellow.194 The unpaired Cysteine at position 321 in the A4 domain is involved in dimer formation. Amino acids of the protease domain catalytic triad are shown in red. Conversion of factor XI to factor XIa involves cleavage after Arg369. Amino acids highlighted by black circles are the sites of mutations that are relatively common causes of CRM- factor XI deficiency (Cys38, Gln88, Glu117, Cys128, and Phe283) or have been associated with CRM+ deficiency (Gly155, Arg184, Ser248, Val371, Pro520, Glu555, and Thr575). The results of the mutations (amino acid substitution or stop codon) are shown adjacent to the involved residues in green. Locations that may be N-linked glycosylation sites are indicated by the orange diamonds. Printed with Permission.
PAN Domains. The apple domains of factor XI and PK are members of the PAN (Plasminogen, Apple, Nematode) domain family. Shown are topology diagrams of (A) two views of the factor XI A1 domain from the zygomen structure reported by Papagrigoriou et al.132 (B) the N-terminal domain of hepatocyte growth factor, (C) the PAN domain from leach anti-platelet protein, and (D) the PAN domain from the microneme protein EtMIC5. The strands of the PAN domain β-sheet (numbered) are shown in yellow, and the α-helix is red. The disulfide bond linking the N and C-termini of the PAN domains (Cys2-Cys85 for factor XI A1) are shown in orange. Note that this bond is missing in hepatocyte growth factor.
Factor XI Subunit Structure. Shown are topology images as reported by Papagrigoriou et al.132 (A) Factor XI apple domain disk (heavy chain). Shown is the disk structure formed by the apple domains (A1, A2, A3, and A4) of human factor XI, which are shown in light blue, gray, orange, and yellow, respectively. Locations of amino acids implicated in ligand binding are shown as space filling images in red for thrombin, green for HK, black for glycoprotein Ib (GpIb), blue for heparin and orange for factor IX. Cys321 in A4 forms the interchain disulfide bond between the subunits of the dimer. (B) Factor XI zymogen subunit. The protease domain (white) of factor XI rests on the apple domain disk like a cup on a saucer. The color coding for the apple domains is the same as in Panel A. Note that Arg184 in the A3 domain is buried in the zymogen factor XI structure by the protease domain. (B inset) Arg184. The side chain of Arg184 forms non-covalent interactions with the side chains of Ser268 (dark gray) from the A3 domain, and Asp488 (green) and Asn566 (white) from the protease domain in zymogen factor XI. These interactions must be broken during conversion of factor XI to factor XIa, freeing Arg184 to interact with factor IX. Reprinted with permission
Factor XI Structure. (A-C) Structures of zymogen factor XI. Shown are space filling models of dimeric zymogen factor XI as reported by Papagrigoriou et al.,132 viewed from three perspectives rotated 90 degrees about the longitudinal access. The A1, A2, A3, and A4 domains are shown in light blue, purple, orange, and yellow, respectively, and the protease domains are in white. Sites of residues implicated in ligand binding are colored red for thrombin (top and middle images), light green for HK (bottom), black for glycoprotein Ib (top and middle), blue for heparin (top, middle and bottom) and dark green for factor IX (top and middle). Positions for the activation cleavage site Arg369-Ile370 (top and middle) and the protease active site serine (bottom) are in magenta. Reprinted with permission.
Factor XI Dimer Formation.
The factor XI dimer interface involves the A4 domains of each subunit, burying a surface area of 886 Å2, with the β-sheets packed against each other (Fig. 175-10A).132 Cys321 is located in a finger like loop in A4. It projects away from the body of the domain, and forms the interchain disulfide bond. Cys321 is conserved in all mammals studied to date with the exception of the rabbit, where it is replaced by histidine (Fig. 175-10B).229 However, rabbit factor XI and human factor XI lacking Cys321 are noncovalently associated dimers,221,222,229 indicating other residues are involved in dimerization. Leu284, Ile290, and Tyr329 form the hydrophobic core of the dimer interface,132,230,231 and are conserved in factor XI across mammal species, but not in PK (a monomer)130 or the PK/FXI predecessor192 (Fig. 175-10B). Salt bridges also form between Lys331 from one subunit and Glu287 from the other subunit. Cys321 in PK, and probably in the PK/FXI predecessor, forms an intra-subunit disulfide bond with Cys326,195 a residue not present in factor XI. A number of mutations in the A4 domain have been shown to interfere, or are predicted to interfere with dimer formation. The best characterized is Phe283Leu,221 which causes a partial dimerization defect. Riley et al. showed that the Phe283Leu substitution causes increased dimer dissociation, and stabilizes the monomer through altered side-chain packing that is unfavorable for dimer formation.232
Factor XI Dimer Interface. (A) Shown are two perspectives of the A4 domain dimer interface. In the image on the left, one A4 domain is presented as a charge-surface representation (blue positive, red negative), and the other A4 domain is shown as a ribbon diagram showing the side chains of residues that interact with the opposite A4 domain. On the right is a ribbon drawing showing the interface with hydrogen bonds and electrostatic interactions in magenta. The Cys321 interchain disulfide bond is shown at the top of the right hand image in orange. Residues Leu284, Ile290 and Tyr329 form the core of the hydrophobic interface. Salt bridges are also formed between Lys331 from one subunit and Glu287 from the other subunit. From Papagrigoriou et al.,132 with permission. (B) A4 domain amino acid sequences. Leu284, Ile290, and Tyr329 are conserved in factor XI (FXI), but not in prekallikrein (PK) or the PK/FXI predecessor (PX). PK and PX have a Cys residue at position 326 that forms an interchain disulfide bond with Cys321. Note that rabbit factor XI, while lacking Cys321, still forms non-covalently associated homodimers.
Almost all factor XI,208 and 75-90% of PK,233,234 circulates as a complex with HK. Studies with recombinant apple domains and FXI/PK chimeras indicate that the A2 domain is required for HK binding, with contributions from A1 and A4.235–239 Detailed mutagenesis studies to localize binding sites have not been performed, however, analysis of the naturally occurring factor XI mutation Gly155Glu has been useful in this regard. Gly155Glu is a rare example of a factor XI mutation associated with a variant that circulates in plasma (cross-reactive material positive [CRM+] mutation).240,241 Gly155 is located at the center of a loop connecting the β5 and β6 strands of A2 (Fig. 175-9 bottom and 175-11A). The main chain nitrogen of Gly155 forms a hydrogen bond with the carbonyl group of Thr152 in a type I β-turn (Fig. 175-11A). Gly155 points towards the center of a cavity in the apple domain disk, and lies in a positively charged channel on the A2 surface opposite the surface pointing toward the protease domain (Fig. 175-9 bottom and 175-11B).131 Recombinant factor XI-Glu155 binds HK with ~10-fold lower affinity than wild-type factor XI (D. Gailani, unpublished observation).242 The loop containing Gly155 forms hydrogen bonds with a second loop spanning residues 103-105 (Fig. 175-11A) that is the site of the factor XI mutation Gly104Asp, and borders a pocket in the A2 domain (Fig. 175-11B). A Gly104Arg mutation in the PK A2 domain is associated with a CRM+ PK deficiency phenotype due to decreased HK binding (Fig. 175-11C).243 There are differences in the charged channels of the putative HK binding sites for factor XI (Fig. 175-11B) and PK (Fig. 175-11C), suggesting HK may bind somewhat differently to the two proteins. For example, the end of the channel between A1 and A4 is open in PK (Fig. 175-11C), but is pinched off in factor XI (Fig. 175-11B) because the channel approaches the dimer interface, which is not present in monomeric PK. Consistent with an important role for the A2 domain in HK binding, conversion of α-kallikrein to β-kallikrein by cleavage in the A2 domain after Lys140 at one end of the charged channel (Fig. 175-11C) is associated with a significant reduction in HK binding.236
Binding Site for HK. (A) Gly104 and Gly155. The relative positions of Gly104 and Gly155 in the factor XI A2 domain are shown. Gly155 is involved in a tight β-turn forming a hydrogen bond with Thr152. The compact hydrogen bond network extends through Thr152 and Thr158 to form contacts to the main chain through residues Lys103 and Ile105. Changes at Gly104 or Gly155 are likely to disrupt this network. Hydrogen bonds are indicated by dashed lines. (B) Putative HK binding site on factor XI. Shown is a charged surface representation (blue - positive, red - negative) of the underside of the factor XI disk (corresponds to bottom image in figure 8), showing the positions of Gly104 and Gly155. These residues are required for proper formation of a charged channel that starts next to a pocket in the A2 domain and extends between the A1 and A4 domains. The white dotted line indicates the predicted binding site of HK. (C) Putative HK binding site on PK. Shown is an image similar to the one in Panel B for factor XI. Note that the end of the charged channel between the A4 and A1 domains is open in PK, while in factor XI it is enclosed. α-kallikrein is cleaved after Lys140 to form β-kallikrein, which binds poorly to HK. The proximity of this residue to the putative HK binding site suggests that cleavage may alter the conformation of the charged channel. Reprinted with permission.
Activation of a factor XI subunit involves cleavage of the Arg369-Ile370 bond (Fig. 175-6),193,194,204 resulting in a 45 kDa heavy chain (apple domains) and 35 kDa light chain (protease domain) connected by a disulfide bond between Cys326 on the A4 domain and Cys482 in the protease domain.193,194 Factor XI is activated by the proteases factor XIIa,204,205 α-thrombin,79,80 meizothrombin,82 factor XIa (autoactivation)79,80 and trypsin.84 Thrombin interacts with most of its substrates through two anion binding exosites (ABE I and II).244 Using thrombins with site-specific mutations, Yun et al. proposed that ABE I and II are required for binding and activation of factor XI in the presence of the polyanion dextran sulfate.245 ABE II binds dextran sulfate, while ABE I was thought to bind factor XI. Subsequent experiments in plasma, however, have not identified a role for ABE I.246 Thrombin binds to the A1 domain of factor XI, with residues Glu66, Lys83, and Gln84 forming part of the binding site.96,132,247 These residues are clustered near the interface with the A4 domain (Fig. 175-8A and 175-9), and are located close to the loop containing the Arg369-Ile370 cleavage site. The Factor XI residues involved in binding to factor XIIa have not been identified. The factor XIIa noncatalytic region is large, and may interact with more than one apple domain. For example, studies using peptide mimicry point to a factor XIIa binding site on the A4 domain,248 while a monoclonal antibody recognizing A2 selectively interferes with factor XI activation by factor XIIa.96
It has been proposed that the dimeric structure of factor XI is required for a trans-activation process in which thrombin binds to one subunit of the dimer while activating the other subunit.132 Wu et al. used recombinant factor XI monomers in a purified system to show the dimer is required for efficient factor XI activation by thrombin, factor XIIa and factor XIa.230 This study used dextran sulfate to facilitate factor XI activation, which may not mimic physiological conditions for activation. A second study gave contradictory results, with monomers proving more susceptible than wild-type dimer to spontaneous autoactivation.231 While nearly all studies of factor XIa have utilized a protease in which both subunits are cleaved at the Arg369-Ile370 bond, a species with only one activated subunit has been isolated and characterized (Fig. 175-12).81 This species is designated 1/2-factor XIa, and its formation has been observed in plasma during contact activation. After Arg369-Ile370 is cleaved, the new N-terminus of the catalyic domain, Ile370, transits 12Å to form a salt bridge with Asp556 (Fig. 175-13).223–225 This conformational change forms the P1 binding pocket that will be occupied by an Arg sidechain from substrates or inhibitors, and is a defining feature of an active serine protease. Changes also occur in loops on the side of the protease domain opposite the active site cleft (Fig. 175-13B). This involves unraveling of a short α-helix in the zymogen and a closing together of the loop containing Arg489 and a β-hairpin to fill a pocket left empty by removal of the A3 domain, which forms an interface with the base of the catalytic domain in the zymogen (Fig. 175-8B and 175-13B).
Factor XI Activation. Each factor XI subunit is activated by cleavage between Arg369 and Ile370. Factor XI (0 hr time point) migrates slightly faster than factor XIa (FXIa) on non-reducing SDS-PAGE. Activation of factor XI (schematic at left) by α-thrombin (shown on the gel) or factor XIIa (not shown) proceeds through an intermediate with one cleaved subunit (center), designated 1/2-FXIa. Subsequent conversion of 1/2-FXIa to fully activated factor XIa (FXIa - right) appears to be a slower process. The apple domains are shown in gray, the inactive protease domains of the zymogen in red, and active protease domains in green. The A3 domains of activated subunits are in orange. It is postulated that a conformational change after cleavage at Arg369-Ile370 results in movement of the protease domain relative to A3, exposing a factor IX binding exosite. A second exosite may also be exposed within the protease domain (cross-hatched area).
The Factor XIa Protease Domain. (A) Conformational changes in the protease domain during activation - Ile370. Shown is a ribbon diagram superimposing the protease domain structures of factor XI (yellow) and factor XIa (green; PDB entry 1ZJD). Conformational changes include a 12 Å shift in the position of Ile370 (purple) into the protease core after cleavage of the Arg369-Ile370 bond. The position of the active site serine residue (Ser557) is also shown. (B) Conformational changes in the protease domain during activation - Arg489. Shown is a ribbon diagram superimposing the protease domain structures of factor XI (white) and factor XIa (dark red). After cleavage of the Arg369-Ile370 bond there is an unraveling of an α-helix containing Arg489 to fill a pocket left empty by removal of Arg184 in the A3 domain (yellow). The P1 Arg side chain from a hypothetical factor XIa substrate is shown in blue occupying the S1 specificity pocket of the protease active site. From Papagrigoriou et al.,132 with permission.
Factor IX Activation by Factor XIa.
Factor XIa contributes to blood coagulation through proteolysis of zymogen factor IX to form the protease factor IXaβ.39–42 Factor IX activation by factor XIa requires calcium ions,41 but does not require a charged surface, phospholipid, or HK. Factor IX activation involves cleavage of the Arg145-Ala146 and Arg180-Val181 peptide bonds, releasing an 11 kDa activation peptide (Fig. 175-14).81,206,249 During initiation of coagulation, the factor VIIa/TF complex (Fig. 175-2B) activates factor IX.250 Factor VIIa/TF first cleaves factor IX after Arg145 forming the intermediate factor IXα, which accumulates, followed by cleavage at Arg180-Val181 to complete the process. In contrast, no intermediate accumulation occurs when factor IX is activated by factor XIa.205,251 It was postulated that the two active sites of a factor XIa dimer facilitate simultaneous cleavage of the two factor IX bonds.251,252 However, 1/2-factor XIa81 and recombinant factor XIa monomers230, which only have one active site, activate factor IX similarly to factor XIa. More than one mechanism is compatible with the observations. Cleavage of the factor IX bonds could be sequential, with the rate of the second cleavage greater than the first. Indeed, factor XIa, like factor VIIa/TF, cleaves factor IX initially at the Arg145-Ala146 bond.81 Alternatively, factor XIa may cleave sequentially, but not release an intermediate before the second cleavage. This would be consistent with data showing that rates of cleavage of factor IX and factor IXα by factor XIa are similar.251
Factor IX Activation by Factor XIa. Shown are schematic diagrams of factor IX, its activation intermediates factor IXα and factor IXaα, and the active protease factor IXaβ. Abbreviations for domains are: Gla - γ-carboxyglutamic acid domain, and EGF - epidermal growth factor domain. During factor IX activation by the factor VIIa/tissue factor complex or factor XIa, there is a preference for initial cleavage of the Arg145-Ala146 bond at the N-terminus of the activation peptide (green ribbon) generating factor IXα (reaction 1). This intermediate accumulates in plasma during activation by factor VIIa/tissue factor, but not during activation by factor XIa. Conversion of factor IXα to factor IXaβ involves cleavage at Arg180-Val181 (reaction 2) releasing the activation peptide. Factor IX conversion to the alternative intermediate factor IXaα (reaction 3) involves initial cleavage at Arg180-Val181. This is a minor reaction during plasma coagulation, but is the preferred mechanism of activation by a protease in Russell Viper Venom. Factor IXaα is then converted to factor IXaβ by cleavage of the Arg145-Ala146 bond to release the activation peptide (reaction 4).
The initial interaction between coagulation proteases and their substrate involves sites (exosites) on the protease distinct from the active site.253 Exosites are major determinants of substrate specificity, and generally are not available for substrate binding in the zymogen, indicating conformational changes must occur with activation. Consistent with this, factor IX binds to factor XIa but not factor XI.254 An exosite interaction is required for factor IX recognition by factor XIa,255 and work with factor XI/PK chimeras and mutants localized an exosite to the A3 domain (Fig. 175-8A, 175-9A, 175-12, and 175-15).256,257 This site is not accessible in factor XI, as it is buried in an interface between the A3 and protease domains (Fig. 175-8B). A critical residue for factor IX activation is Arg184,257 located near the N-terminus of A3 (Fig. 175-8B, 175-13B, and 175-15). An Arg184Gly mutation was reported in a CRM+ factor XI deficient patient,258 and recombinant factor XI-Ala184 is defective in factor IX activation. In the zymogen, Arg184 occupies a central location in a loop connecting A2 and A3 that is buried under the protease domain (Fig. 175-8B and 175-13B). The guanidinium group of Arg184 fills a pocket formed by the interface between the apple domain disk and protease domain, with the sidechain interacting with Asp488, Asn566 from the protease domain and Ser268 from the A3 domain (Fig. 175-8B). Arg184 is conserved in factor XI (Fig. 175-15), but is not found in PK or the PK/FXI predecessor. Arg184 must be exposed in factor XIa to facilitate factor IX binding. Factors XI and XIa have been analyzed by small angle x-ray scattering and electron microscopy, which revealed a large shape change with activation, consistent with a shift in the relative positions of the apple domain disk and protease domain.252 The opening of the A3-protease domain interface will break interactions between Arg184 and Asp488, Asn566, and Ser268, (Fig. 175-8B), exposing the Arg184 side chain to the solvent phase. Arg184 can, therefore, be considered part of a switch that holds factor XI in an inactive conformation in the zymogen, or facilitates engagement of factor IX in the protease. The Gla-domain of factor IX is required for binding to factor XIa,254 and may interact with the A3 domain.
Putative Factor IX Binding Site on the Factor XIa A3 domain. Shown are sequences between amino acids 181 and 194 in the N-terminal portion of the A3 domains of factor XI (FXI), prekallikrein (PK), and the PK/FXI predecessor (PX) from various species. Note the conservation of residues in blue across species in factor XI, but not in PK or the PK/FXI predecessor. Arg184, indicated in red, appears to behave like a switch, helping maintain the zymogen conformation of factor XI (Fig. 175-8B), and forming a critical part of the factor IX binding site in the active protease.
There is convincing evidence that an additional exosite outside of the A3 domain is required for normal activation of factor IX. Loss of the A3 exosite has a more deleterious effect on cleavage of factor IX at Arg180-Val181 than at Arg145-Ala146,81,259 suggesting the A3 exosite primarily facilitates cleavage of factor IXα to factor IXaβ, and that a distinct exosite is involved in the initial interaction with factor IX, facilitating conversion to factor IXα. Work by Sinha et al. indicates this exosite is located on the protease domain.259 Interestingly, in contrast to the A3 exosite, the interaction with the protease domain exosite does not required calcium, indicating the interaction does not involve the factor IX Gla domain. Factor XIa can also activate factor XII,260 plasminogen,261 and factor XI,79,80 and cleaves HK to liberate bradykinin.262 The physiologic importance of these reactions is not clear.
Factor XI Binding to Platelets.
Unlike vitamin K-dependent coagulation proteases, factor XI lacks a Gla domain to facilitate binding to phospholipid surfaces.132,194,195 However, Greengard et al. determined that there are ~1500 binding sites for factor XI per platelet.263 Binding to platelets involves residues in the factor XI A3 domain (Figures 2B and 3)264,265 and the leucine rich repeat on the platelet GP1bα chain.266 Two critical residues for factor XI platelet binding, Ser248 and Arg250, are near a heparin binding site on A3 (Fig. 175-16).267 A CRM+ missense mutation in the A3 domain, Ser248Asn, was identified in a family with a bleeding disorder.268 Recombinant factor XI-Asn248 binds platelets with five-fold reduced affinity compared to wild-type factor XI, and other amino acid substitutions for Ser248 show similar defects.265 The original observation of 1500 factor XI binding sites was difficult to reconcile with the ~25,000 copies of GP1b per platelet. Recently, White-Adams et al. reported that factor XI binds to a platelet receptor for apolipoprotein E called ApoER2′ (LRP8) with an affinity of ~60 nM.269 There are ~2000 copies of ApoER2′ per platelet, which is in close agreement with the number of factor XI binding sites. ApoER2′ co-localizes with GP1b on platelets,270 and the factor XI binding site may require both receptors. It is not clear, if the platelet binding site on the A3 domain is required for interactions with ApoER2′, or if a different site is involved.
Platelet Binding Site on Factor XI. The locations of Ser248 and Arg250 (in black), which likely form a binding site for platelet GpIb on the factor XI A3 domain (orange) are shown relative to residues that form a heparin binding site (Ly252, Lys253, and Lys255 in blue). Ser248 forms hydrogen bonds with Asp194 and Thr249 that are probably disrupted in the hereditary mutation Ser248Asn. The adjacent A2 domain is shown in light blue, with the position of an N-linked glycan attached to Asn108 indicated in yellow and red. Reprinted with permission.
While platelets do not enhance factor XI activation in a static system,96 as once thought, they do affect factor XI and factor XIa behavior under flow.143,269 Platelets deposit in aggregates and fibrin strands form when human blood lacking factor XII is perfused over collagen at arterial shear rates. Antibodies that block factor XIa activation of factor IX prevent fibrin formation.269 It is likely that platelets play a role either in activating factor XI or localizing it to the site of fibrin formation, as fibrin does not form when blood lacking platelets is perfused over collagen. It is interesting to consider the dimeric structure of factor XI in this regard. The image of the dimer shown in Fig. 175-9B and 175-9C reveals that the two factor XI A3 domains point in opposite directions from the longitudinal plane of the molecule. This arrangement could facilitate a molecule of factor XIa or 1/2-factor XIa binding to a platelet through one A3 domain, while binding the substrate factor IX through the other. As factor XI lacks a Gla domain, the dimeric structure may represent an alternative solution to the problem of simultaneously tethering the protease to a platelet at a wound site while binding to a substrate.
Regulation of Factor XIa.
Factor XIa is subject to regulation by several plasma serine protease inhibitors (serpins), with C1-inhibitor (C1INH) 267,271,272 and antithrombin (AT) 273–275 probably most important. Protein Z-dependent protease inhibitor (ZPI),274–276 α1-antitrypsin,277 α2-antiplasmin,278 and protein C inhibitor279 also inhibit factor XIa. Basic amino acids in the factor XIa autolysis loop (Arg504,Lys505, Arg507 and Lys509) are determinants of serpin specificity.274 Heparin enhances serpin-mediated factor XIa inhibition through a complex mechanism involving heparin binding sites on A3 (Lys252, Lys253, and Lys255) (Fig. 175-8A, 175-9 and 175-16)267 and the catalytic domain (Lys529, Arg530, Arg532, Lys536, and Lys540) (Fig. 175-9 and 175-16).275,280 The heparin binding site on A3 facilitates inhibition through a template mechanisms where protease and serpin both bind to heparin.267 The mechanism involving the catalytic domain heparin binding site cannot be explained by this type of mechanism, and may involve charge neutralization or an allosteric effect that overcomes repulsive interactions between the serpin and basic residues on the protease domain.275 Platelets also release a soluble form of the β-amyloid precursor protein called protease nexin II, a potent factor XIa inhibitor that may regulate activity in the vicinity of platelet-rich clots.224,281–283
PK (Human Protein Reference Database 01971) is the zymogen of the trypsin-like protease α-kallikrein.284,285 It circulates in plasma at a concentration of ~580 nM (50 μg/ml),286 75-90% in a noncovalent complex with HK (see Factor XI binding to HK ).233,234 While the liver is the primary source of plasma PK,130,287 PK mRNA is present in numerous tissues,287 and PK protein has been identified in pancreatic islets and kidney tubules288 in a pattern similar to that observed for factor XI.213,214 PK is a paralog of factor XI.192–195 The human PK gene (GenBank M62345) covers ~30 kb on chromosome 4q34-q35,218 and is ~8 kb from the factor XI gene.219 As with the factor XI gene, exon 1 of the PK gene is not translated, exon 2 encodes the signal peptide, exons 3-10 the four apple domains and exons 11-15 the protease domain. PK is a kininogenase that cleaves HK to liberate bradykinin,133–135 but is not related to the large group of kininogenases called tissue-type (glandular) kallikreins.289,290
The cDNA and amino acid sequences for multiple mammalian PKs are available. Human PK is a 619 amino acid protein.130,195 While PK crystal structures are not available, it is likely the protein is organized similarly to factor XI, with four 90- to 91-amino acid apple domains and a C-terminal protease domain (Fig. 175-17).130,131,195 This is supported by the observation that PK and factor XI apple domains can be exchanged without compromising the capacity of cells to synthesize and secrete the protein.222,256 In contrast to factor XI, PK is a monomer, running as a doublet of 86 and 88-kDa bands (based on differences in glycosylation) on nonreducing gels.285,286 Cys321, which forms an interchain disulfide bond in factor XI,221,222 forms an intrachain bond with Cys326 in PK (Fig. 175-10B, 175-17, and 175-18A).195 Freeing Cys321 by changing Cys326 to another amino acid, does not result in dimeric PK.222 This is almost certainly because PK lacks the residues in the A4 domain that facilitate factor XI dimer formation (Fig. 175-10B and 175-18A).230,231 In addition, PK has an N-linked oligosaccharide in the A4 domain at Asn289. A large oligosaccharide at this location in factor XI would interfere with the dimer interface. In factor XI residue 289 is aspartic acid (Fig. 175-18A).
Human Prekallikrein - Amino Acid Sequence and Disulfide Bond Locations. Schematic diagram of the primary structure of prekallikrein (PK) as reported by Chung et al.130 The apple domains are designated A1 to A4. Cysteine residues involved in disulfide bonds are shown in yellow.195 Amino acids of the catalytic triad are shown in red. Conversion of PK to α-kallikrein involves cleavage after Arg371. An additional cleavage after Arg140 forms β-kallikrein. Amino acids highlighted by black circles are locations of mutations reported to cause PK deficiency. The amino acid substitutions are shown adjacent to the involved residues in green. Locations that may be N-linked glycosylation sites are indicated by orange diamonds. Reprinted with permission.
Prekallikrein Apple Domains. (A) Comparison of the PK and factor XI A4 domains. The A4 domains of PK (yellow) and factor XI (gray) are superimposed. Factor XI residues are indicated in italics. Note that the Cys321-Cys326 bond in PK causes collapse of the finger loop containing Cys321 in factor XI. Residues that are important for dimer formation in factor XIa that are not conserved in PK are shown in stick form. Black dashed lines indicate points of interaction at the dimer interface in factor XI. (B) PK A2 domain. Shown is a topology diagram of the A2 domain showing the putative HK binding pocket. Residues that likely form the epitope of monoclonal antibody PKH6 (Asn146-Lys150)131,237 that blocks HK binding are shown in stick form. A natural mutation at Gly104 (Glu) causes CRM+ PK deficiency associated with a defect in HK binding. Cleavage after Lys140 converts α-kallikrein into β-kallikrein. Reprinted with permission.
PK is anchored to cell surfaces via binding to its substrate HK. The PK binding site on HK is within a continuous stretch of 31 amino acids in the extreme C-terminus of the protein.291 A peptide spanning these residues competes with PK/HK complex formation. The HK binding site on PK is discontinuous and involves residues from the A1, A2, and A4 domains.235,292–294 Together these domains form a site that binds HK with high affinity (K d 12 nM). Structurally, the binding site probably is similar to the HK binding site of factor XI (see Factor XI binding to HK and Fig. 175-11). PK and α-kallikrein bind to HK with similar affinity, indicating that proteolytic activation of PK does not interfere with the stability of the PK/HK complex.294 With prolonged incubation, α-kallikrein is converted to β-kallikrein by cleavage within the A2 domain. This cleavage largely abolishes PK/HK complex formation, suggesting that A2 domain harbors the major HK binding site.236,295
PK is converted to active α-kallikrein by proteolytic cleavage of the Arg371-Ile372 bond, generating a 50-kDa heavy chain and a 35-kDa light chain connected by a disulfide bond between Cys364 on the A4 domain and Cys484 on the protease domain (Fig. 175-17).130,285,296 During contact activation, α-factor XIIa and β-factor XIIa activate PK.285,296 Trypsin will activate PK, and surface-dependent autoactivation has been demonstrated.297 Autocatalytic cleavage of α-kallikrein at Lys140-Ala141 generates β-kallikrein (Fig. 175-18B), a 65-kDa protease that binds poorly to HK.295,298 As reviewed in the section on the Plasma Kallikrein-Kinin System , PK may also be activated independent of contact activation on the surface of endothelial cells in a process involving the prolylcarboxypeptidase PRCP.125–127
α-Kallikrein is a kininogenase, cleaving HK to release BK (Fig. 175-1).4,5,299–301 α-Kallikrein also activates surface-bound factor XII to α- and β-XIIa,53,59,60 prourokinase to urokinase,50 plasminogen to plasmin,51 and the renin-angiotensin system by conversion of prorenin to renin.56,57 In contrast to factor XI, relatively little is known about the functional roles of the PK/α-kallikrein apple domains, beyond their importance to HK binding. Interestingly, in the absence of calcium, α-kallikrein activates factor IX with similar efficiency to factor XIa.40 However, addition of physiologic calcium concentrations reduces the K m for the reaction two-orders of magnitude for factor XIa, but has no effect on activation by α-kallikrein.40,256 As discussed under Factor IX activation by factor XIa , the reaction depends on a calcium-dependent interaction that likely involves the factor IX Gla-domain and an exosite on the factor XIa A3 domain254–257 that is not conserved in α-kallikrein (Fig. 175-15). α-Kallikrein has potent in vitro effects on leukocytes, stimulating monocytes, and inducing neutrophil aggregation, chemotaxis, and elastase release.4,5,302,303 α-Kallikrein has been identified as a major activator of plasminogen during epithelial cell apoptosis, adipocyte differentiation and stromal remodeling during mammary gland involution. In this setting, PK is localized to mast cells in the connective tissues of the mammary gland.304
Regulation of α-Kallikrein.
Plasma α-kallikrein is inhibited primarily by C1-INH305,306 and α2-macroglobulin,307,308 with some contribution from AT.305 In comparison to inhibition of factor XIa, heparin has a relatively small effect on C1-INH inhibition of α-kallikrein.309 Plasminogen activator inhibitor-1 (PAI-1) 310 and protein C inhibitor279,311 also inhibit plasma α-kallikrein.
Factor XII (Human Protein Reference Database 01992) is the zymogen of the plasma serine protease α-factor XIIa.312–314 It is an 80-kDa single-chain glycoprotein that circulates in plasma at a concentration of 375 nM (30 μg/ml).314 The liver is the primary source of factor XII in plasma.315–317 The mRNA for factor XII has been detected by RT-PCR in other tissues, including placenta, but at levels at least 1000-fold lower than in liver.318 As discussed in the section on Evolution of the Contact Factors , factor XII is a paralog of HGFA.192 The organization of the factor XII and HGFA genes are similar to those of the fibrinolysis activators tissue plasminogen activator and prourokinase.319 The human factor XII gene spans ~12-kb on chromosome 5 (q33-qter), and has 14 exons.319,320
Plasma factor XII levels increase with estrogen therapy and during pregnancy.321,322 Base pairs -43 through -31 (5′-GGGCANNNTGACC-3′) of the factor XII gene promoter are similar to the consensus estrogen-responsive element (ERE).321–323 It has been postulated that an estrogen-mediated increase in factor XII triggers exacerbations in type III HAE, a condition seem almost exclusively in female patients,324,325 however, newer evidence does not support this (see Factor XII Mutations and Hereditary Angioedema type III). Similar to factor XI, hepatocyte specific expression of factor XII in mice requires the transcription factor HNF4α.220 The situation is not as clear in humans, where HNF4α reportedly antagonizes estrogen-mediated upregulation of the factor XII gene.326 A common C/T polymorphism has been identified at residue -46 of the human gene promoter.327,328 Homozygosity for -46T is associated with lower plasma factor XII levels, with a recent study reporting an average plasma level of 53 U/dl (42-67) for 46TT, 87 U/dL (77-99) for 46CT, and 132 U/dL (97-151) for 46CC.329 While lower factor XII levels were associated with increased risk for arterial thrombosis in two large studies,156,164 it is not clear that the -46TT genotype is associated with arterial disease.
Factor XII protein contains, from the N-terminus, a fibronectin type II domain (encoded by exons 3 and 4), an epidermal growth factor (EGF) domain (exon 5), a fibronectin type 1 domain (exon 6), a second EGF domain (exon 7), a kringle domain (exons 8 and 9), a unique 55-amino acid "proline-rich" region (exon 9), and a trypsin-like protease domain (exons 10-14) (Figure 175-19). Relatively little is known about these structures as they relate to function. Several regions may be involved in surface binding, including the proximal N-terminus (residues 1-28), the fibronectin type II domain (residues 39-47), the fibronectin type 1 domain (residues 134-153), the second EGF domain and the kringle domain.330–333 The fibronectin type I domain may be responsible for interactions between factor XII and misfolded proteins (see Contact Activation ).334 Two zinc binding sites have been identified in human factor XII (residues 40-44 and 78-82).335 Binding to zinc ions stabilizes intermediate structures during the activation process, accelerating activation of factor XII.335
Human Factor XII - Amino Acid Sequence and Disulfide Bond Locations. Schematic diagram of the primary structure of factor XII.314,336,337 Cysteine residues involved in disulfide bonds are shown in yellow. Amino acids of the catalytic triad are shown in red. Conversion of factor XII to α-factor XIIa involves cleavage after Arg353. Additional cleavages after Arg343 and Arg334 form β-factor XIIa. Amino acids highlighted by black circles are locations of missense or nonsense mutations associated with congenital factor XII deficiency, with the results of the mutations (amino acid substitution or stop codon) shown adjacent to the involved residues in green lettering within black circles. Mutations involving Thr309 associated with type III hereditary angioedema are shown within green circles. Locations that may be N-linked or O-linked glycosylation sites are indicated by the orange and light blue diamonds, respectively. Reprinted with permission.
Factor XII is activated by cleavage of the Arg353-Val354 peptide bond (Figure 175-19) by autoactivation or another protease (see Contact Activation ). α-Kallikrein, factor XIa, factor XIIa, plasmin, or trypsin activate factor XII, with α-kallikrein the most potent activator. Plasmin-mediated factor XII activation has been documented in patients receiving thrombolytic therapy.336 The active enzyme, α-factor XIIa, consists of a 52 kDa N-terminal heavy chain and 28 kDa C-terminal light chain (protease domain), connected by a disulfide bond between Cys340 and Cys367.337,338 Activation is dramatically enhanced when factor XII is bound to negatively charged surfaces, and by zinc ions.339–342
While a surface that facilitates factor XII participation in normal hemostasis has not been identified, there are several candidates for pathologic triggers in thrombosis models in mice. Collagen has been shown to enhance factor XII activation and platelet activation.21,139,146 Inorganic polyphosphates are released from platelet dense granules upon activation, and enhance fibrin formation through initiation of contact activation.27–29 Given that factor XII-initiated fibrin formation appears to be relatively unimportant for hemostasis, but is required for thrombus propagation in thrombosis models, factor XII may be most active on the platelet surface away from the blood vessel wall.343 Indeed, platelets have been linked to the intrinsic pathway of coagulation for over 40 years,344 with activation promoting fibrin formation in a factor XII-dependent manner.343 Polyphosphates are released from activated platelets, but not from vascular endothelial cells, and are an attractive candidate for a factor XII activator within a growing platelet-rich thrombus.27–29 Platelets from patients with Hermansky-Pudlak syndrome have reduced numbers of dense granules (the storage site of polyphosphates), and are defective in triggering factor XII-dependent fibrin clot formation.149 Furthermore, degradation of polyphosphates by phosphatases interferes with thrombus formation driven by activated platelets.149 Kannemeier et al. showed the capacity of RNA from damaged cells to induce contact activation in vitro, and that administration of RNAse reduced thrombus formation in mice.30 Similarly, a synthetic RNA homologue (polyIC) that is not degraded by plasma RNAses triggers thrombosis in a factor XII-dependent manner in mice.30 Maas et al. recently showed that factor XII is activated during interactions with misfolded proteins, like the ones in the blood of patients with amyloidosis.35 Interestingly, this interaction triggers PK activation, but not factor XI activation, suggesting specificity for a proinflammatory pathway. This is different than polyphosphate-triggered activation of factor XII, which has been shown to be both pro-coagulant and prothrombotic in mice.149 This could indicate the presence of unknown specific pathways for initiation of kinin-formation and/or coagulation.345
Factor XIIa Activity and Regulation.
The major substrates of α-factor XIIa during contact activation are PK and factor XI. In addition, factor XIIa can activate plasminogen and factor VII, and cleaves BK from HK.53,346–348 β-Factor XIIa is generated by additional cleavages at Arg334 and Arg343 (Figure 175-19).337,338 Because it lacks most of the heavy chain, β-factor XIIa binds poorly to contact surfaces, and does not support surface-mediated activation of PK or factor XI.60 β-Factor XIIa can activate PK in the fluid phase and also activates complement factor C1.60,61,349 Factor XII and factor XIIa stimulate the proliferation of cultured human hepatoma cells (HepG2) through a mitogenic activity that may reside in the EGF-like domains.350,351 The predominant plasma inhibitor of both α- and β-factor XIIa is C1-INH, with AT and PAI-1 also having activity.310,352,353
High Molecular Weight Kininogen
HK (Human Protein Reference Database 01970) is a 120-kDa (626 amino acid) β-globulin glycoprotein that circulates in plasma at a concentration of 670 nM (80 μg/ml).104,133 The protein migrates anomalously on gel filtration (~220,000), indicating a large partial specific volume. Electron microscopy shows that HK is a complex of linked globular units.354 Low molecular weight kininogen (LK), also called α1-cysteine protease inhibitor, is a 68-kDa (409 amino acid) glycoprotein with a plasma concentration of 1.3 μM (90 μg/ml).133 HK is a member of the cystatin family of cysteine protease inhibitors. In humans, the ~27-kb kininogen gene contains 11 exons and maps to chromosome 3q27, close to genes for α2HS-glycoprotein and histidine-rich glycoprotein, two other members of the cystatin family.355–358 Through alternative mRNA splicing and use of different polyadenylation signals, the kininogen gene produces transcripts for HK (3.5 kb) and LK (1.7 kb) (Figure 175-20).356 Exons 1-9 and a portion of exon 10, encoding BK and the following 12 amino acids, are common to the mRNAs for HK and LK.359 HK mRNA contains the remainder of exon 10 encoding a unique 56-kDa C-terminus, while in the LK mRNA this part of exon 10 and its flanking 90-nucleotide sequence is spliced out, and exon 11 is used to encode a unique 4-kDa C-terminus. Hepatocytes express mRNA for HK and LK, and liver is the likely major source for both proteins in plasma,356,357 although HK and LK mRNA can be found in a variety of other tissues.318 HK antigen is present in cultured umbilical vein endothelial cells, granulocytes, renal tubular cells, skin, and in the α-granules of platelets.360–366
Kininogen Structure and Domain Functions.
Kininogens are multi-domain proteins (Fig. 175-20). HK and LK have identical heavy chains (D1-D3) and a BK-containing D4 domain. The light chain of LK is designated D5L, while HK light chain contains domains D5H and D6H. HK and LK, along with fetuins and histidine-rich glycoprotein, comprise the type 3 group of cystatins.367,368 HK and LK are the only members that retain cysteine protease inhibitory activity.367 D1, D2, and D3 are homologous to cystatin, and D2 and D3 have cysteine protease inhibitory activity.369 Numerous studies have examined properties of HK domains in the context of the whole molecule, in isolation, and as peptide fragments. HK and its cleavage products have multiple effects on blood coagulation and vascular biology.104,133,369,370 A peptide derived from D1 inhibits atrial naturetic peptide (ANP), blocking ATP-mediated diuresis and natriuresis in rats.371,372 Studies with calpain deficient mice suggest that this cysteine protease plays a key role in platelet function, regulation of outside-in signaling, clot retraction and aggregation.373,374 Platelet calpain, papain and cathepsins are inhibited by the D2 domain.369,375,376 D3 and D4 (BK) inhibit thrombin-induced platelet aggregation by interfering with α-thrombin binding to platelet GP1b and by interacting with protease-activated receptor-1, inhibiting its proteolytic activation by thrombin.377–380 Thrombostatin FM compounds, which are based on angiotensin converting enzyme-induced breakdown products of BK have also been shown to be direct thrombin inhibitors.381 Regions within D3, D4, and D5H in HK, are required for zinc-dependent binding to the surface of cells such as granulocytes, platelets, and endothelial cells that may facilitate assembly of the KKS system as described in the section on The Plasma Kallikrein-Kinin System .382–386
Kininogens. Low molecular weight (LK - top) and high molecular weight (HK - bottom) kininogen are products of a single gene containing 11 exons (shown in schematic form in the center of the diagram). Color-coding of the exons corresponds to specific domains in LK and HK. Activities of HK domains are indicated at the bottom of the diagram. The figure is designed after Colman and Schmaier.538 Reprinted with permission.
The D4 domains are the sources of vasoactive kinins that are released from HK and LK through proteolysis by kininogenases.135,387–389 HK is the preferred substrate for α-kallikrein, and cleavage after residues Lys362 and Arg371 generates BK (NH4-Arg-Pro-Pro-Gly-Phe-Ser-Pro-Phe-Arg). LK is cleaved mainly by tissue kallikreins liberating Lys-BK (also called kallidin). Bradykinin-free HK (HKa) contains a heavy chain (residues 1-362) identical to that of LK, and a large light chain (residues 372-626) connected to the heavy chain by a single disulfide bond (Cys10-Cys596). While HK and HKa differ by only nine amino acids, the apparent difference in molecular mass as determined by non-reducing gel electrophoresis is dramatic (120 and 90 kDa, respectively), indicating that the protein undergoes major structural changes upon kinin liberation. Under reducing conditions HKa is comprised of a heavy chain (63 kDa) and a light chain (58 kDa), the latter of which is rapidly cleaved by α-kallikrein and factor XIa to produce a stable form of 45-kDa and smaller-sized fragments.390 The two-chain form of cleaved LK that is created after kinin release is composed of a heavy chain (residues 1-362) and a light chain (residues 372-409) connected by a single disulfide bridge (Cys10-Cys389).390
BK and Lys-BK bind to, and activate the B2 receptor. Both peptides have short half-lives in plasma (< 20 seconds), undergoing degradation to inactive peptides by angiotensin converting enzyme, neutral endopeptidases and aminopeptidases. BK and Lys-BK are also cleaved by carboxypeptidases, collectively referred to as kininase I (carboxypeptidase N EC 126.96.36.199), to yield des-Arg9-BK and des-Arg10-Kallidin, which have longer half-lives than the parent peptides, and activate the B1 receptor. Kininase II, more commonly called angiotensin-converting enzyme (ACE), and neutral endopeptidase (NEP24.11, EC 188.8.131.52) inactivate BK and Lys-BK by removing the carboxy-terminal Phe-Arg dipeptide. Known effects of kinins include vasodilatation, regulation of local blood flow, and increased vascular permeability. The hypotensive effects of kinins are mediated, at least in part, by release of nitric oxide (NO) from endothelial cells. Kinins also stimulate production and release of autacoids such as prostaglandins PGI2 and PGE2. In some situations, endothelium-derived hyperpolarizing factor (EDHF) may mediate the vasodilator effects of kinins.391 Kininogen is also cleaved by factor XIIa, factor XIa, and plasmin, but the physiological relevance of these reactions is not clear.261,346,392
Two histidine/glycine-rich regions in the D5H domain of the HK light chain bind zinc and mediate the binding of HK to anionic surfaces and heparin.393–396 This surface-binding subdomain overlaps with the cell-binding epitope on D5H.386 A peptide spanning the region in D5H between His479 and His498 of HK, designated HKH20 efficiently kills Escherichia coli, Pseudomonas aeruginosa, and Enterococcus faecalis by disrupting bacterial membranes, but does not affect eukaryotic cell membranes.397 Proteolytic degradation of HK by neutrophil proteases, or a metaloproteinase from P. aeruginosa resulted in fragments with the HKH20 epitope, suggesting this region of HK is part of the innate immune system. A function for the light chain of LK (D5L) has not been clearly determined.
The D6H domain in the HK light chain contains high affinity binding sites for factor XI and PK (K ds 18 and 12 nM, respectively)398 and these proteins circulate in plasma primarily bound to HK.398,399 The factor XI and PK binding sites overlap, and are located in the extreme C-terminus of HK.291 Therefore one HK molecule can bind either a factor XI molecule or a PK molecule. The coagulant activity of HK as measured in the aPTT assay requires D5H-mediated binding of HK to artificial anionic surfaces and the D6H-mediated binding of factor XI and PK to HK.400–402 The sequence for factor XI and PK binding in HK are lacking in LK, which does not possess procoagulant activity in the aPTT assay.
Both HK and LK undergo post-translational modifications by hydroxylation and glycosylation.390 A single acceptor site for prolyl-4-hydroxylase is located at residue 365, which corresponds to residue three of the BK sequence (RPPGFSPFR). More than 50% of human kininogens are hydroxylated at Pro365. Hydroxylation does not affect the pharmacological properties of the kinins. HK and LK have three N-glycosylation sites in their common portion. In addition HK has nine O-glycosylation sites in its unique carboxy-terminus, making it a heavily glycosylated plasma protein. The specific functions of the carbohydrates are unknown; however, they likely affect solubility, half-life and secretion. HK and LK each contain 18 cysteine residues that form nine disulfide bridges. Eight bonds are in the common amino-terminal portions, while a single disulfide bond connects the extreme N-and C-termini.