A mechanism for hereditary angioedema caused by a lysine 311–to–glutamic acid substitution in plasminogen

Hereditary angioedema (HAE) is a genetic condition affecting 1 in 50 000 to 100 000 individuals.¹ Patients with HAE experience recurring episodes of soft tissue swelling involving subcutaneous tissues (hands and face), oropharyngeal mucosa, the genitals, and the gastrointestinal tract.^1-3 The underlying cause in most cases is dysregulation of the plasma kallikrein-kinin system (KKS).^2,4,5 The KKS is composed of the zymogens prekallikrein (PK) and factor XII (FXII) and the cofactor/substrate high molecular weight kininogen (HK).^6-8 PK and FXII reciprocally convert each other to the proteases plasma kallikrein (PKa) and FXIIa.^6-10 PKa cleaves HK to release the vasoactive nonapeptide bradykinin (BK).^4-8 The effects of BK are mediated through the B2 receptor, which is constitutively expressed in many tissues.^11,12 Basal BK production likely contributes to setting normal blood vessel tone and permeability,^13,14 while higher concentrations at injury sites facilitate vascular leak, tissue swelling, and pain sensation.^4-8 In HAE, soft tissue swelling from excessive BK production distinguishes the disorder from the more common histamine-driven edema associated with allergic reactions.^1-5 

The primary regulator of PKa and FXIIa is the serpin C1 inhibitor (C1-INH).^1-3,15 In most patients with HAE, plasma C1-INH activity is <50% of normal.^1-3 However, at least 10% of those with HAE have normal C1-INH activity.^2,10,16-19 Lysine or arginine substitutions for Thr³⁰⁹ in FXII were reported in patients with HAE with normal C1-INH in 2006²⁰ and have subsequently been identified in >150 families.^16-19 FXII-Thr/Arg³⁰⁹ creates a protease cleavage site that facilitates generation of a truncated FXII (ΔFXII), which accelerates reciprocal activation with PK and overwhelms the inhibitory capacity of C1-INH.^6,21 In 2018, 2 groups reported a Lys³¹¹ to glutamic acid substitution in the fibrinolytic zymogen plasminogen (Plg) in patients with HAE with normal C1-INH activity who lacked FXII-Thr³⁰⁹Lys/Arg.^22,23 Plg-Glu³¹¹ has now been reported in ∼150 patients in 33 families.^16-19 We prepared recombinant Plg-Lys³¹¹ and Plg-Glu³¹¹ and their activated plasmin (Plm) forms (Plm-Lys³¹¹ and Plm-Glu³¹¹) and tested their effects on BK production in purified protein- and blood-based systems. We conclude that Plm-Glu³¹¹ is a potent kininogenase that catalyzes BK release from human HK and the related protein low molecular weight kininogen (LK) independently of PKa and FXIIa.

Materials used were as follows: normal human plasma (Precision BioLogic); human plasmas lacking FXII, FXI, PK, or kininogen (George King Biomed); FXII, FXIIa, PK, PKa, HK, human and mouse Plgs, human and mouse Plms, and corn trypsin inhibitor (Enzyme Research Laboratory); fibrinogen and α-thrombin (Haematologic Technologies); LK, soybean trypsin inhibitor, ɛ-aminocaproic acid (ɛ-ACA; Sigma-Aldrich); tissue plasminogen activator (tPA; alteplase; Genentech); S2302 (H-d-prolyl-l-phenylalanyl-l-arginine-p-nitroanilide dihydrochloride; DiaPharma); polyclonal horseradish peroxidase–conjugated immunoglobulin (IgG) to hemagglutinin tag (HA) and rabbit anti-BK polyclonal IgG (Invitrogen); PKa inhibitor KV999272 (formerly VA999272) was previously described²⁴; and BK (Eurofins Discover X, Fremont, CA) and kallidin standards (Peptide Institute, Osaka, Japan).

Complementary DNAs for wild-type human Plg (Plg-Lys³¹¹) and Plg with glutamic acid (Plg-Glu³¹¹) or alanine (Plg-Ala³¹¹) substitutions at amino acid 311 were expressed in Expi293 cells (Thermo Fisher Scientific). Plgs were concentrated from conditioned media by binding to lysine-sepharose, purified further by ion exchange and size exclusion chromatography and converted to Plm (Plm-Lys³¹¹ and Plm-Glu³¹¹) by incubation with urokinase immobilized on sepharose at 37°C.

cDNAs encoding human HK and LK (supplemental Figure 1, available on the Blood Web site), full-length mouse HK (mHK1; supplemental Figure 2), and mouse HK lacking domain 5 and part of domain 6 (mHK3; supplemental Figure 3) in expression vector pJVCMV^9,10 were modified by adding an HA (Tyr-Pro-Tyr-Asp-Val-Pro-Asp-Tyr-Ala) tag to the C terminus. Kininogens were expressed in HEK293 cells and purified by chromatography using anti-HA IgG coupled to agarose (Thermo Fisher Scientific; Pierce Protein Biology). Preparations of recombinant kininogens are shown in supplemental Figure 4. Preparation of human FXII lacking amino acids 1 to 309 (ΔFXII) has been previously reported.¹⁰ 

All reactions were conducted in 96-well PEG-20000–coated plates in 100 μL of reaction buffer (20 mM of N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid [pH, 7.4], 100 mM of NaCl, 0.1% PEG-8000, and 10 mM of ZnCl₂) at 37°C. For all experiments involving enzyme activation or activity, the term vehicle refers to this buffer.

Plg, FXII, or PK (200 nM) was incubated with tPA (20 nM), Plm (125 nM), FXIIa (20 nM), or PKa (20 nM). In some experiments, before addition of enzymes and substrates, fibrinogen (200 μg/mL) and thrombin (50 nM) were incubated at 37°C for 20 minutes to form fibrin. Thrombin was inhibited with hirudin (250 μM), followed by addition of enzymes and substrates. At various times, aliquots were removed for activity testing. For measuring PKa or FXIIa generation by Plm, reactions were stopped with 2 μM of α2-antiplasmin for reactions with PKa or 3.3 μM of aprotinin for reactions with FXIIa. For measuring Plm generation by FXIIa and PKa, reactions were stopped with 600 nM of corn trypsin inhibitor or 10 μM of KV999272, respectively. S-2302 was added to a final concentration of 200 μM, and changes in optical density at 405 nm were measured. Results were converted to FXIIa, PKa, and Plm concentrations using control curves constructed with purified enzymes.

Fibrinogen (200 μg/mL) and thrombin (50 nM) were incubated at 37°C for 20 minutes, after which thrombin was inhibited with hirudin (250 μM). Plg (200 nM) and S-2302 (200 μM) were added with or without tPA (300 pM), and changes in optical density at 405 nm were followed.

Normal, FXII-deficient, or PK-deficient plasma was supplemented with captopril (500 μM) and Plg (600 nM), and reactions (40-μL volumes) were started with different concentrations of tPA or ΔFXII at 37°C. At various times, 3-μL reaction volumes were transferred to 12 μL of ice-cold ethanol. Samples were clarified by centrifugation at 1000g for 10 minutes. Kininogen-deficient plasma was supplemented with plasma-derived human HK (640 nM), LK (2.3 μM) or both and then treated in a similar manner to other plasmas. For purified protein assays, HK, LK, mHK1, or mHK3 (200 nM) were incubated with different concentrations of PKa or Plm in reaction buffer (150 μL) at 37°C. At various times, 10-μL volumes were transferred into 90 μL of ice-cold ethanol and placed at −80°C overnight. Samples underwent clarification by centrifugation at 10 000g for 1 hour. For all reactions, kinins were measured by enzyme-linked immunosorbent assay (ELISA; Enzo Life Sciences, Inc.). For any experiment involving BK release, the term vehicle refers to 20 mM of tris(hydroxymethyl)aminomethane (Tris)-HCl and 136 mM of NaCl (pH, 7.4).

Purified kininogen samples were prepared in the same manner as for the kinin ELISA. For all analytes, ionization was achieved using matrix-assisted laser desorption ionization (MALDI). Analytes were deposited with matrix on MALDI targets. Matrix 2,5-dihydroxybenzoic acid (Sigma) was purified by recrystallization from hot water. Matrix solution was prepared with recrystallized 2,5-dihydroxybenzoic acid (15 mg/dL) in a water/acetonitrile/formic acid ratio of 1:1:0.1. Matrix solution (1 μL) was spotted onto the MALDI target with 1 μL of analyte sample and allowed to dry at room temperature. Samples were analyzed in a Bruker AutoFlex (Billerica, MA) mass spectrometer. Mass spectrometry data were collected in positive ion reflectron mode with a mass/charge (m/z) range from 600 to 6000. Spectra are from an average of at least 100 laser shots per sample spot. Mass spectra were analyzed with Compass 1.4 Flex Series software (Bruker Daltonics).

Plasma HK (200 nM) with PKa (2 nM), or plasma HK (800 nM) with Plm-Lys³¹¹ or Plm-Glu³¹¹(250 nM) was incubated in reaction buffer at 37°C. At various times, aliquots were removed into nonreducing sodium dodecyl sulfate (SDS) sample buffer and size fractionated on 10% polyacrylamide gels. Gels were either stained with GelCode Blue or transferred to nitrocellulose and probed with an anti-human BK antibody. Detection was by chemiluminescence.

Two micrograms of Plg or Plm (active site inhibited with FPR-chloromethylketone) was incubated with recombinant human HK or LK (2 μg) for 30 minutes in 500 μL of Tris-buffered saline (20 mM of Tris-HCl [pH, 7.4] and 100 mM of NaCl). Kininogens were precipitated with anti-HA magnetic beads (Pierce) and washed 3 times with Tris-buffered saline/0.05% Tween-20. Proteins were eluted with SDS nonreducing sample buffer and size fractionated on 10% polyacrylamide gels. Gels were stained with GelCode Blue (Pierce).

HK was immobilized on a CM5 sensor chip using amine coupling. Binding of Plg or Plm (25-400 nM) to HK was assessed by a single-cycle assay at 25°C on a Biacore T200 device (Cytiva) in running buffer (10 mM of N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid [pH, 7.4], 150 mM of NaCl, 50 μM of EDTA, and 0.05% Tween 20). The data were fitted with a 1:1 Langmuir binding model using Biacore T200 evaluation software (Biacore AB), which was also used to calculate kinetic and affinity constants.

Full-length Glu-Plg is converted to Lys-Plg by removal of the N-terminal PAN domain (Figure 1A, left). Glu- and Lys-Plg are converted to Plm by cleavage of the Arg⁵⁶¹-Val⁵⁶² peptide bond. Because Lys-Plg is in an open conformation with a more accessible activation cleavage site, it is activated by tPA ∼10-fold faster than Glu-Plg in the absence of fibrin (Figure 1B). Recombinant Plg-Lys³¹¹, Plg-Glu³¹¹, and their Plm forms were compared with plasma Glu-Plg and Glu-Plm, as shown in Figure 1A (center and right). Plg-Ala³¹¹ (supplemental Figure 5) was prepared as a control. Although the recombinant Plgs are Glu-Plgs,^25,26 they are activated similarly to Lys-Plg (Figure 1C). This is due to glycosylation differences near the activation site in Glu-Plg expressed in HEK293 cells (isoform 1) and plasma-derived Glu-Plg (mostly isoform 2).^25,26 To account for this, Plg-Glu³¹¹ and Plg-Ala³¹¹ were compared with wild-type Plg-Lys³¹¹, rather than plasma-derived Glu-Plg, in activity assays. Plg bound to fibrin adopts an open conformation that accelerates activation by tPA (supplemental Figure 6A). Plg-Lys³¹¹ and Plg-Glu³¹¹ are activated comparably to plasma-derived Glu-Plg in the presence of fibrin (supplemental Figure 6B).

When tPA was added to normal plasma supplemented with recombinant Plg, BK generation was substantially greater with Plg-Glu³¹¹ than with Plg-Lys³¹¹ or Plg-Ala³¹¹ (Figure 1D-E). There was a threshold for inducing measurable BK in this system that required relatively high tPA concentrations (Figure 1F), likely reflecting slow tPA-catalyzed Plg activation in the absence of fibrin and the short plasma half-life of tPA. Adding thrombin to plasma to convert fibrinogen to fibrin led to measurable BK production at a 10-fold lower tPA concentration (Figure 1G), again with more BK generated with Plg-Glu³¹¹. The enhanced activity associated with the Glu³¹¹ substitution cannot be attributed specifically to loss of Lys³¹¹, because Plg-Ala³¹¹ did not create the gain-of-function phenotype noted with Plg-Glu³¹¹ (Figure 1D).

Plm activates FXII.^27-30 Increased BK generation in plasma containing Plg-Glu³¹¹ may, therefore, reflect enhanced FXII activation by Plm-Glu³¹¹.^27-30 However, Plm-Glu³¹¹ and Plm-Lys³¹¹ were comparable, relatively weak, activators of FXII (Figure 2A) and PK (Figure 2B). Furthermore, Plg-Glu³¹¹ enhanced BK generation comparably in normal (Figure 2C), PK-deficient (Figure 2D), and FXII-deficient (Figure 2E) plasmas. Rapid HK cleavage may be induced in normal plasma by adding ΔFXII, identified in patients with HAE with FXII-Lys/Arg³⁰⁹ substitutions.¹⁰ ΔFXII is activated by PKa ∼40 times faster than FXII, accelerating reciprocal activation of PK.¹⁰ Accordingly, ΔFXII-induced BK generation was blocked by the PKa inhibitor KV999272 (Figure 2F).²⁴ In contrast, BK generation induced by adding tPA to plasma containing Plg-Glu³¹¹ was unaffected by KV999272 (Figure 2G), consistent with Plm-Glu³¹¹ driving BK generation independently of PKa.

PKa can activate Plg by a fibrin-independent mechanism,^31,32 suggesting KKS activation could generate Plm that may contribute to BK formation. However, we found that PKa and FXIIa were weak activators of Plg-Lys³¹¹ and Plg-Glu³¹¹ when compared with tPA (Figure 2H). As shown in Figure 2I, ΔFXII was added to plasma to activate the KKS. Adding Plg-Lys³¹¹ or Plg-Glu³¹¹ to this system modestly increased BK production that may be slightly greater with Plg-Glu³¹¹ than Plg-Lys³¹¹. However, this would not account for the difference in magnitude of the effects of Plg-Lys³¹¹ or Plg-Glu³¹¹ in plasma treated with tPA (Figure 2C-E).

Human plasma contains 2 kininogens, HK and LK, which are encoded by alternatively spliced messenger RNAs from the Kng1 gene.^33,34 Sequential cleavage of HK by PKa after Arg³⁷¹ and Lys³⁶² releases BK (Arg³⁶³-Pro-Pro-Gly-Phe-Ser-Pro-Phe-Arg³⁷¹), whereas LK cleavage by tissue kallikreins after Arg³⁷¹ and Met³⁶¹ releases the decapeptide kallidin (Lys-BK, Lys³⁶²-Arg-Pro-Pro-Gly-Phe-Ser-Pro-Phe-Arg³⁷¹).^35,36 HK and LK each contain a disulfide bond (Cys10-Cys⁵⁹⁶ and Cys10-Cys³⁸⁹, respectively) that connects the N and C termini of the protein (Figure 3A; supplemental Figure 1).^36,37 PKa cleavage of the Arg³⁷¹-Ser³⁷² peptide bond produced a pronounced downward shift in HK migration on nonreducing SDS–polyacrylamide gel electrophoresis corresponding to conversion of HK to a linear form (Figure 3B; supplemental Figure 4). Cleavage of the Lys³⁶²-Arg³⁶³ bond then released BK, causing an additional subtler downward shift. A western blot using an anti-BK IgG was consistent with the BK sequence remaining associated with HK after 1 cleavage and then being released by the second cleavage (Figure 3B, right). LK cleavage by PKa caused a slight upward shift in migration. A higher PKa concentration was required to observe LK cleavage than that required with HK (Figure 3C).

HK cleavage patterns produced by incubation with Plm-Lys³¹¹ or Plm-Glu³¹¹ are shown in Figure 3D-E, respectively. Corresponding western blots (Figure 3D-E, lower) indicate BK is released by Plm-Glu³¹¹ more rapidly than by Plm-Lys³¹¹. Perhaps Plm-Glu³¹¹ makes the second cleavage after Lys³⁶² that releases BK more efficiently than Plm-Lys³¹¹. Alternatively, Plm-Lys³¹¹ and Plm-Glu³¹¹ may cleave HK at different sites. A similar shift to that observed with PKa occurred when LK was incubated with Plm-Lys³¹¹ (Figure 3F) or Plm-Glu³¹¹ (Figure 3G). For HK and LK, additional cleavage occurred over time with Plms that were not observed with PKa, indicating PKa and Plm cleave the kininogens differently.

PKa released BK from HK at a rate 50- to 100-fold higher than from LK (Figure 4A). BK release from HK catalyzed by Plm-Lys³¹¹ was at least 50-fold slower than in reactions with PKa (Figure 4B). Interestingly, PKa and Plm-Lys³¹¹ released BK from LK at similar rates (Figure 4C). Furthermore, Plm-Lys³¹¹ released BK at a rate approximately three- to fourfold higher from LK than from HK (Figure 4D). Plm-Glu³¹¹ liberated BK at an initial rate at least 10-fold faster from HK (Figure 4B) and two- to threefold faster from LK (Figure 4C) than from Plm-Lys³¹¹. Indeed, BK release catalyzed by Plm-Glu³¹¹ was comparable for HK and LK (Figure 4E). In experiments with HK, peak BK generation roughly correlated with Plm-Glu³¹¹ concentration (Figure 4F), suggesting a stochiometric interaction rather than a Michaelis-Menten–type process. When comparing these data with results shown in Figure 3, it is again apparent that kininogen cleavage on SDS polyacylamide gels does not necessarily reflect BK release, probably because Plms are cleaving kininogens at locations other than those required to release BK.

Plasma from a person with a congenital deficiency of HK and LK (supplemental Figure 7)³⁸ was supplemented with physiologic concentrations of plasma-derived HK (640 nM), LK (2.3 μM), or both. Plg-Lys³¹¹ or Plg-Glu³¹¹ was added, followed by tPA to activate Plg. For Plg-Lys³¹¹ and Plg-Glu³¹¹, BK generation was greater in plasma containing LK than in plasma containing HK (Figure 4G). As shown in Figure 1E, BK generation was greater with Plg-Glu³¹¹ (Figure 4G, left) than with Plg-Lys³¹¹ (Figure 4G, right). As in Figure 2G, adding KV999272 to block PKa activity did not affect results (supplemental Figure 8).

The ELISA used in this study does not distinguish between BK and kallidin (Lys-BK) and may detect other novel peptides containing BK sequence. We addressed this issue with mass spectroscopy. Mass spectra for BK and kallidin standards are shown in Figure 5A. Although incubating HK without a protease did not result in detectable cleavage on SDS–polyacrylamide gel electrophoresis, a mass peak at m/z = 1200 was observed in the MALDI analysis that did not correspond to BK or kallidin. This may reflect contamination of the HK preparation with an unknown protease. Incubating HK with PKa or Plm-Glu³¹¹ generated a peak (m/z = 1060) corresponding to BK (Figure 5A). Plm-Lys³¹¹ did not produce a detectable BK peak in these experiments. Time courses of BK formation during HK incubation with PKa or Plm-Glu³¹¹ are shown in Figure 5B. The proteases generated similar patterns over time, indicating Plm-Glu³¹¹ does not form BK from an intermediate such as kallidin. Similarly, peaks representing BK were detected during LK cleavage by PKa or Plm-Glu³¹¹, with no evidence of kallidin generation (supplemental Figure 9). These data show that Plm-Glu³¹¹, like PKa, cleaves HK and LK after Arg³⁷¹ and Lys³⁶² to form BK.

We examined Plg/Plm binding to HK by surface plasmon resonance. The amounts of Plg-Glu³¹¹ and Plm-Glu³¹¹ bound to HK were greater than the amounts of Plg-Lys³¹¹ and Plm-Lys³¹¹ (Figure 6A). Plm-Glu³¹¹ also gave a stronger signal than Plm-Lys³¹¹ in coprecipitation experiments with HK (Figure 6B) and LK (Figure 6C). However, K_ds for binding of Plgs and Plms containing Lys³¹¹ and Glu³¹¹ to HK determined by surface plasmon resonance were comparable (Figure 6A), indicating different affinities do not explain the enhanced BK generation associated with Plg-Glu³¹¹.

Plg and Plm bind lysine and arginine residues on fibrin through Asp-X-Asp/Glu motifs on their kringle domains.^37,39 Plg has 5 kringle domains.^25,26 In human Plg, an Asp-X-Asp/Glu motif is present on kringles 1, 2, 4, and 5, but Lys³¹¹ disrupts the motif (Asp-X-Lys) in kringle 3. Interestingly, Plg-Glu³¹¹, which creates an Asp-X-Asp/Glu motif in kringle 3, is the standard in mammals (supplemental Figure 10). The Glu³¹¹ substitution in patients with HAE, therefore, restores the ancestral pattern. This raises the possibility that an interaction between Plg/Plm-Glu³¹¹ kringle 3 and basic amino acids on HK contributes to BK release. ε-ACA, a lysine analog used therapeutically to reduce fibrinolysis by inhibiting Plg binding to fibrin, inhibits BK release from HK by Plm-Lys³¹¹ and Plm-Glu³¹¹ (Figure 7A).⁴⁰ Both Plms, therefore, appear to bind basic residues on HK . The additional binding site on kringle 3 provided by Glu³¹¹ may alter binding orientation to facilitate cleavage of the kininogen Arg³⁷¹-Ser³⁷² and Lys³⁶²-Arg³⁶³ bonds.

The observation that Plg residue 311 is glutamic acid in most mammals, but is associated with HAE only in humans, is intriguing. Interestingly, little BK was released from full-length mHK1 (supplemental Figure 3) by either Plm-Lys^{311 or} Plm-Glu³¹¹ (Figure 7B). Furthermore, BK release from human HK and mouse HK was similar when plasma-derived human Plm (Lys³¹¹) or mouse Plm (Glu³¹¹) was used to catalyze reactions (Figure 7C). These data suggest HAE requires Glu³¹¹ in the context of human Plm, in combination with features specific to human kininogens. Mice have, in addition to mHK1, an HK form (mHK3) lacking the D5 domain and part of the D6 domain found in mHK1, as a result of alternative messenger RNA splicing (supplemental Figures 3 and 4A).⁴¹ Incubating mHK3 with human or mouse Plm produced similar results to those for mHK1 (Figure 7C-D).

The most common cause of HAE is reduced plasma C1-INH activity.^1-4,15-19 As the main regulator of PKa and FXIIa, C1-INH maintains BK generation within an appropriate range. Angioedema associated with low C1-INH is highly responsive to PKa inhibitors,^42-44 reduction of plasma PK,⁴⁵ or a FXIIa neutralizing antibody,⁴⁶ indicating dysregulated FXIIa-mediated generation of PKa is causative. Given this, the source of BK during acute episodes of angioedema in patients with low C1-INH is almost certainly HK, because LK is a poor PKa substrate. Patients with signs and symptoms of HAE, but with normal C1-INH levels, were first reported in 2000.^47,48 It is estimated that this condition, called HAE with normal C1-INH, accounts for >10% of patients with HAE and perhaps substantially more.^16-19 Mutations in genes for FXII,^10,20,49 Plg,^22,23,50,51 angiopoietin-1,⁵² and kininogen⁵³ were identified in large families and are presumed causative. Suspect mutations in genes for myoferlin⁵⁴ and heparan sulfate 3-O-sulfotransferase 6⁵⁵ have also been reported. The manifestations of HAE, therefore, may be caused by a variety of defects affecting kinin production or vascular sensitivity to kinins.

Plg-Glu³¹¹ was reported in 2018 in patients in Germany by Bork et al²² and by Dewald.²³ The prevalence of the mutation is not known, but Plg-Glu³¹¹ patients have been identified in several European countries, Japan, and the United States, suggesting wide distribution.⁵¹ A relationship between fibrinolysis and kinin generation was described by Kaplan and Austen²⁸ in 1971. Oral-lingual edema after thrombolytic therapy is a well-recognized clinical example of this association.^56,57 Plm activates FXII,^29,30 while PKa activates Plg and the Plg activator prourokinase.^31,32,57-59 It has been proposed that Plg-Glu³¹¹ may have an enhanced capacity to activate the KKS.¹⁶ Hypothetically, if Plm-Glu³¹¹ activates FXII faster than Plm-Lys³¹¹, or if PKa activates Plg-Glu³¹¹ more rapidly than Plg-Lys³¹¹, a positive feedback cycle that amplifies PK activation could result, leading to excessive kinin production.

Our data suggest an alternative scenario in which Plg-Glu³¹¹ contributes to angioedema by directly catalyzing kinin release from HK and LK. HK is a component of the KKS, serving as a substrate for PKa and a cofactor for PK and FXI binding to surfaces during contact activation. LK lacks the elements for surface, PK, and FXI binding found on the HK D5 and D6 domains and is thought to function primarily as a substrate for tissue kallikreins.^60,61 Previously, several groups reported HK cleavage by Plm, although it is not clear that BK was a major product of the reaction.^31-33,62 In our analysis, both wild-type Plm-Lys³¹¹ and mutant Plm-Glu³¹¹ readily cleaved human HK and LK, but BK release was substantially greater with Plm-Glu³¹¹. Interestingly, Plm-Glu³¹¹ catalyzed BK release from HK and LK at comparable rates. Furthermore, Plm-Lys³¹¹ released BK from LK 3 to 4 times faster than from HK. Because the plasma concentration of LK is two- to fourfold greater than HK, most BK generated by either Plg-Lys³¹¹ or Plg-Glu³¹¹ may come from LK, a premise supported by our studies with supplemented kininogen-deficient plasmas.

Replacing Plg-Lys³¹¹ with glutamic acid creates a consensus binding site for lysine and arginine side chains in the kringle 3 domain, similar to those normally found on kringles 1, 2, 4, and 5.^25,26 The ability of ε-ACA to inhibit HK cleavage by Plm indicates binding to basic amino acids is important for the Plm-HK interaction, as shown previously by Kleniewski et al.^63,64 Lysine binding to kringle 3 may alter the orientation of kininogen-bound Plm relative to the kininogen Arg³⁷¹-Ser³⁷² and Lys³⁶²-Arg³⁶³ cleavage sites. Curiously, residue 311 is normally glutamic acid in mammalian Plgs. Lys³¹¹ is found only in humans and the closely related chimpanzee and gorilla (supplemental Figure 10). However, Glu³¹¹ is linked to HAE only in humans. Our data support the conclusion that it is the combination of Glu³¹¹ in human Plm and human kininogen that is pathogenic. HK and LK share common D1, D2, D3, and D4 domains. There are differences in distribution of basic amino acids in these domains between human and mouse kininogens that may underlie the different susceptibilities of the proteins for Plg-Glu³¹¹–catalyzed BK release (supplemental Figure 2).

There is a predilection for oral-lingual edema in Plg-Glu³¹¹ carriers.⁵¹ Tongue swelling occurs in 80% of symptomatic Plg-Glu³¹¹ patients and is often the only manifestation of HAE.^16-19,22,51 Edema involving the extremities, gastrointestinal tract, and larynx is less common, and the prodromal rash (erythema marginatum) associated with HAE is rare in Plg-Glu³¹¹ patients. Observations of patients with bleeding tendencies indicate that the oral cavity has relatively high intrinsic fibrinolytic activity.⁶⁵ Therefore, the predilection for oral-lingual angioedema in Plg-Glu³¹¹ patients may, at least in part, reflect normal robust Plg activation in the mouth. Angiotensin-converting enzyme inhibitors, which inhibit BK degradation, may also trigger oral-lingual angioedema. A recent case report by Wang et al⁶⁶ describes successful treatment of a patient with angiotensin-converting enzyme inhibitor–induced angioedema with the lysine analog tranexamic acid, raising the possibility that local Plm-mediated production of BK contributes to the syndrome.

Our study raises the possibility that certain therapeutic approaches commonly used to treat patients with HAE, including C1-INH supplementation and PKa inhibition, may be less effective in Plg-Glu³¹¹ patients. In a case series, Bork et al⁵¹ reported a mean angioedema episode duration of 4.3 ± 2.6 hours (201 episodes) in 13 patients with Plg-Glu³¹¹ treated with the B2 receptor antagonist icatibant, compared with 44.7 ± 28.6 hours (149 episodes) in the same patients in the absence of treatment (88% reduction in episode duration). In 12 patients receiving C1-INH infusion (74 episodes), treatment had a more modest effect (mean episode duration decreased from 48.2 ± 32.5 to 31.5 ± 8.6 hours). These results suggest icatibant is more effective than C1-INH in Plg-Glu³¹¹–associated HAE and are consistent with a relatively small contribution to angioedema from the KKS. However, such a conclusion should be considered tentative. A majority of Plg-Glu³¹¹ carriers receiving C1-INH in this study were considered to have had good responses,⁵¹ perhaps reflecting a KKS contribution to the disease process. Alternatively, C1-INH inhibits Plm and tPA,^67,68 and this could be relevant at doses used to treat angioedema. There are few data on the efficacy of PKa inhibitors in Plg-Glu³¹¹ patients. Such information will be important in determining the contributions of different BK-generating pathways that may operate in Plg-Glu³¹¹ patients. World Allergy Organization/European Academy of Allergy and Clinical Immunology 2021 guidelines calling for screening for known mutations in HAE patients with normal C1-INH will also help in this regard.

This work was supported by awards HL140025 from the National Heart, Lung, and Blood Institute, National Institutes of Health (D.G.); APP1129592 from the National Health and Medical Research Council Australia (J.C.W. and R.H.P.L.), and FL18010019 from the Australian Research Council (J.C.W. and R.H.P.L.).

Contribution: S.K.D. designed and performed assays for protease activation and bradykinin generation and contributed to writing the manuscript; S.K. and D.R.P. designed and performed mass spectroscopy analysis and contributed to writing the manuscript; M.-F.S. and B.M.M. prepared recombinant high molecular weight kininogens and conducted precipitation experiments; J.C.W. and A.J.Q. prepared recombinant plasminogen and plasmin and performed surface plasmon resonance binding experiments; E.P.F. contributed to writing the manuscript; R.H.P.L. oversaw preparation of recombinant plasminogen mutants and binding studies and contributed to writing the manuscript; and D.G. oversaw the project and writing of the manuscript.

Conflict-of-interest disclosure: D.G. receives consultancy fees from pharmaceutical companies (Anthos Therapeutics, Aronora, Bayer Pharmaceuticals, Bristol-Myers Squibb, Ionis, Janssen) with an interest in inhibition of contact activation and the kallikrein-kinin system for therapeutic purposes. E.P.F. is an employee of Kalvista Pharmaceuticals, Inc. The remaining authors declare no competing financial interests.

Correspondence: David Gailani, Vanderbilt University Medical Center, Room 4918, The Vanderbilt Clinic, 1301 Medical Center Drive, Nashville, TN 37232; e-mail: dave.gailani@vanderbilt.edu.