Hereditary angioedema: The plasma contact ... - Wiley Online Library

Accepted Article

DR COEN MAAS (Orcid ID : 0000-0003-4593-0976)

Article type

: Review Article

Hereditary angioedema: The plasma contact system out of control

S. DE MAAT, * Z.L.M. HOFMAN,* and C. MAAS*

*Department of Clinical Chemistry and Haematology, University Medical Center Utrecht, Utrecht, the Netherlands

Correspondence: Coen Maas, Department of Clinical Chemistry and Haematology, University Medical Center Utrecht, Utrecht University, Utrecht, the Netherlands. Tel.: +31 88 755 6513; fax: +31 88 755 5418. E-mail: [email protected]

Summary The plasma contact system contributes to thrombosis in experimental models. Even though our standard blood coagulation tests are prolonged when plasma lacks contact factors, this enzyme system appears to have a minor (if any) role in haemostasis. In this review, we will explore the clinical phenotype of C1 esterase inhibitor (C1-INH) deficiency. C1-INH is the key plasma inhibitor of the contact system enzymes and its deficiency causes hereditary angioedema (HAE). This inflammatory disorder is hallmarked by recurrent

This article has been accepted for publication and undergone full peer review but has not been through the copyediting, typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of Record. Please cite this article as doi: 10.1111/jth.14209 This article is protected by copyright. All rights reserved.

aggressive attacks of tissue swelling that occur at unpredictable locations throughout the body. Bradykinin,

Accepted Article

which is considered a byproduct of the plasma contact system during in vitro coagulation, is the main disease mediator in HAE. Surprisingly, there is little evidence for thrombotic events in HAE patients, suggesting a mechanistic uncoupling from the intrinsic pathway of coagulation. In addition, it is questionable whether a surface is responsible for contact system activation in HAE. In this review, we will discuss the clinical phenotype, disease modifiers and diagnostic challenges of HAE. We will subsequently describe the underlying biochemical mechanisms and contributing disease mediators. Furthermore, we will review three types of HAE, which are not caused by C1 esterase inhibitor deficiency. Finally, we will propose a central enzymatic axis that we hypothesize to be responsible for bradykinin production in health and disease.

Keywords Bradykinin, Factor XII, hereditary angioedema, plasma kallikrein, plasmin

Introduction Bradykinin is a well-known powerful vasoactive peptide. It triggers vascular leakage, which is needed for host defense and tissue repair. Bradykinin is liberated from its precursor high molecular-weight kininogen (HK) after interplay between the two serine proteases activated factor XII (FXIIa) and plasma kallikrein (PKa). These factors form the plasma contact system, which drives in vitro plasma coagulation in response to negatively charged particles and surfaces. A burst of enzymatic activity spontaneously occurs when these factors assemble on a negatively charged surface and activate each other through molecular scission. Next, the contact system feeds into the intrinsic pathway of coagulation via Factor XI (FXI). This reaction forms the basis for the activated partial thromboplastin time (aPTT), which is used to monitor e.g. antithrombotic therapy, coagulation factor deficiencies and lupus anticoagulant. In this in vitro context, the production of bradykinin is considered to be of minor importance. The mechanisms behind contact activation in vitro have been explored in detail. The interaction between FXIIa and FXI forms the main axis for contact-system driven coagulation. PKa acts as a catalyst for FXII activation [1], while HK acts as a cofactor by delivering the zymogens plasma prekallikrein (PK) and FXI to negatively charged

This article is protected by copyright. All rights reserved.

surfaces. The initial spark for enzymatic activity is proposed to be carried by FXII itself, which exhibits minute

Accepted Article

enzymatic activity without signs of molecular scission on western blot [2,3]. This behavior is reminiscent of trypsin, tissue- and urokinase plasminogen activator, which display similar zymogen activity. Also PK carries some intrinsic enzymatic activity, in particular when complexed to HK [4].

The confusing roles of contact activation in haemostasis Polyphosphate polymers are carried by and released by platelets and mast cells [5,6]. These can trigger contact system activation through FXII. In clotting assays, short polyphosphate polymers (60-100 mers) are ~5 times better than kaolin (by mass), while long polymers (>1000-mers) exhibit activity that is >3000-fold higher [7]. Interestingly, concentration of smaller (i.e. platelet-sized) polyphosphate polymers onto surfaces and into insoluble nanoparticles after complexation with metal ions [8] strongly promotes their FXII-dependent procoagulant activity. Divalent metal ions such as calcium are abundantly present in platelet dense granules, suggesting that intracellular complexes might form. We recently confirmed that platelets contain and release particles of concentrated polyphosphate [5]. Further investigation is needed to determine whether polyphosphate-triggered contact activation contributes to physiological haemostasis. Currently, insufficient epidemiological data is available to fully appreciate the role of the contact system in cardiovascular disease. However, there have been several interesting observations: a partial deficiency in FXII is associated with all-cause mortality [9], ischemic stroke and myocardial infarction [10]. This is unlike other clotting factors, where increased levels generally form a risk factor for cardiovascular events. Hence, it is attractive to speculate that there is a role for the contact system outside its canonical role in fibrin formation. In the next sections, we will examine the surprising clinical phenotype of a hyperactive contact system.

Lessons from serine protease inhibitor deficiencies Most haemostatic enzymes are inhibited by serine protease inhibitors (SERPINs) to maintain physiological control over their activities. SERPIN deficiencies, either genetic or acquired, often have pathological consequences. For example, antithrombin deficiency leads to thrombosis, while α2-antiplasmin deficiency


results in a bleeding phenotype. This concept extends beyond haemostasis: deficiency in α1-antitrypsin leads

Accepted Article

to emphysema because of uncontrolled activity of neutrophil elastase, damaging lung tissue. In summary, the absence of an inhibitor generally leads to a hyperactive state of the enzyme (system) it is supposed to control.

C1 esterase inhibitor deficiency C1 esterase inhibitor (C1-INH) is encoded by the SERPING1 gene, which is located on chromosome 11 (11q12.1

[11]). This 105 kDa glycoprotein (7 N-linked and 8 O-linked glycans) is the main inhibitor of the classical complement enzymes C1r and C1s. It is also the primary inhibitor of the contact factors. For starters, it is responsible for the inhibition of 93% of FXIIa in plasma [12]. Similarly, C1-INH is responsible for 52% of the inhibition of PKa [13] and 47% for activated Factor XI (FXIa; [14]). C1-INH also modestly inhibits plasmin, but the physiological relevance of this inhibition may be limited [15]. There is functional redundancy for the activities of C1-INH: FXIIa can be partially (26%) inhibited by α2-antiplasmin and antithrombin [12], while α2macroglobulin (A2M) is responsible for 48% of the inhibition of PKa. Finally, FXIa is reportedly inhibited for ~50% by the combined activities of α2-antiplasmin and α1-antitrypsin [14]. In the presence of heparin, antithrombin takes over this role and inhibits FXIa by 42%. Based on these combined functions of C1-INH, it is reasonable to assume that congenital deficiency of C1-INH (incidence 1:50,000) results either in a complement-system related disorder or in uncontrolled intrinsic coagulation. Although there are no clear signs of a haemostatic problem in patients with C1-INH deficiency, they do develop unpredictable recurrent spontaneous histamine-independent attacks of tissue swelling that can take place at various locations throughout the body. This disease is called hereditary angioedema (HAE; OMIM # 106100). There are two types of HAE due to C1-INH deficiency; 1) quantitative (HAE-1) and 2) qualitative (HAE-2) [16]. To date, 450 known mutations in the SERPING1 gene are associated with HAE [17]. Although often familial in nature, approximately 25% of the patients have a de novo mutation in the SERPING1 gene [18]. In addition to these genetic disorders, B-cell malignancy can lead to the development of neutralizing autoantibodies against C1-INH, which result in acquired C1-INH deficiency. Furthermore, it has been reported that patients with B-cell


malignancy develop anti-idiotypic antibodies. These form complexes with self immunoglobulins on the B-cell

Accepted Article

membrane, leading to a consumption of C1q and C1-INH with angioedema as a consequence [19]. It is not needed to be fully deficient in C1-INH order to develop swelling attacks. It has been reported C1-INH levels below 38% are insufficient to effectively prevent spontaneous contact system activation [20]. Furthermore, there is little correlation between the levels and activity of C1-INH and disease activity. This motivates the search for new biomarkers in HAE [21].

BOX 1. Clinical features of C1-INH-HAE Angioedema attacks. Driven by increased local vascular permeability, fluid accumulates in the deeper layers of the skin or mucosal tissues. Angioedema attacks develop over a time course of several hours, and if left untreated, last for 2-5 days. The face (example in Fig. 1), extremities, upper airways, genitals and gastrointestinal tract are often affected [22]. In general, swellings in HAE are painful, but itch is rarely reported. Facial swellings are disfiguring and swellings in hands and feet lead to temporary impaired function both interfering with patients daily activities. The most-feared complication of HAE is the development of laryngeal edema, which can rapidly leads to asphyxiation. In the era before symptomatic treatment, HAE had a mortality as high as 56% [23]. Another severe disease feature is angioedema of the bowel, which is experienced by 75 to 93% of C1-INH-HAE patients [22,23]. Abdominal attacks are accompanied by colicky pain, vomiting and diarrhea. When undiagnosed, HAE patients presenting with acute abdominal symptoms are commonly subjected to unnecessary surgical interventions [24]. Understandably, HAE has a significant impact on quality of life. Even in the absence of acute symptoms, patients suffer from its unpredictable course. Increased permeability appears restricted to the attack location, as hypotension is not a known feature of HAE. Interestingly, an attack may not be limited to a single attack location. We recently reported an analysis of attacks (n=219) in a group of 119 C1-INH-HAE patients. Out of this group, 33 patients experienced (28%) attacks took place at multiple, anatomically distinct locations. In some patients, up to five locations could be simultaneously affected [25]. Attack triggers and prodromal symptoms. Swelling attacks can be provoked by (mild) injury, mental stress,


physical exertion, menstruation, infection, dental procedures or develop spontaneously [26]. HAE patients

Accepted Article

report that they are able to recognize signs of an upcoming angioedema attack. These signs include fatigue, muscle ache and erythema marginatum, a map-like, red skin rash [27]. Erythema marginatum can already be present in newborns [28]. Disease onset and severity. On average, patients experience a first attack at the age of 11 but onset of disease can vary between infancy and late adulthood [22]. Attack frequency is highly variable; some patients experience swellings on a daily basis, while others have hardly any swelling attacks at all, even in the absence of prophylactic therapy. Symptom-free intervals of more than three decades have been described [22].

The underlying cause: complement or contact system? Based on the knowledge that C1-INH is a critical inhibitor of the complement system, initial studies focused on the vasoactive properties of C1 esterase (C1s). Observations in guinea pig models suggested that C1s had the potential to trigger vascular leakage, potentially involving a histamine-like agent [29]. This would explain why C1-INH deficiency causes angioedema. Around the same time, a female patient and her two sons with angioedema were investigated for their “inborn biochemical lesion” [30]. The investigators concluded that their serum was lacking an important inhibitor for PKa and confirmed an increased and sustained vascular leakage of these patients upon intradermal administration of PKa, explaining their disease phenotype. Some years later, it was demonstrated that C1-INH acts as the inhibitor of PKa, thereby limiting bradykinin release from HK [31]. Finally, it was demonstrated that PK and HK are consumed during HAE attacks [32]. This was found attributable to the ‘intrinsic instability’ of plasma that lacks C1-INH, leading to ‘spontaneous’ bradykinin production in a test tube setting [33]. This surprising finding revealed that the non-hemostatic properties of the contact system as a driver of bradykinin production had long been underappreciated. Additional clinical evidence that the clinical picture of C1-INH-HAE was unrelated to a dysregulated complement system was provided with the description of a human mutation in the SERPING1 gene that causes a selective defect of C1INH towards the complement factors, but not towards the contact system. This family did not develop angioedema [34]. The other way around, when blisters are experimentally induced on the skin of patients with C1-INH deficiency, the blister fluid contains uninhibited PKa [35]. This shows that PK activation follows in response to tissue injury and that, at least in absence of physiological control, the contact system enzymes are


able to escape from the systemic circulation. Together, these findings indicated that the uncontrolled activity

Accepted Article

of the plasma contact system forms the basis for pathological tissue swelling in C1-INH-HAE. This however, does not rule out the involvement of additional players that influence the disease mechanism.

Kinin formation and metabolism (Fig. 2) In plasma, cleavage of HK (110 kDa) by PK leads to the release of bradykinin (9 amino acids; RPPGFSPFR). In tissues, cleavage of low-molecular weight kininogen (LK; 68 kDa) by tissue kallikreins leads to the release of kallidin (lysyl-bradykinin; contains one additional N-terminal lysine; 10 amino acids; KRPPGFSPFR). There is only one single gene (KLKB1) that encodes PK. In contrast, there are 15 members in the tissue kallikrein family, with diverging functions. Bradykinin and kallidin are recognized by the kinin B2 receptor (B2R) that is constitutively expressed on vascular endothelial cells. Activation of this G-protein coupled receptor leads to vascular leakage through induction of endothelial cell contractility, uncoupling of endothelial cell junctions, production of nitric oxide and prostacyclin. After activation, B2R is internalized, leading to (temporary) desensitization of the tissue towards bradykinin [36]. Truncation of bradykinin and kallidin changes their functionalities. C-terminal truncation by carboxypeptidase M or –N, as well as activate thrombin-activatable fibrinolysis inhibitor, 9

removes the terminal Arg residue, resulting in Des-arg -bradykinin (or –kallidin, respectively). This product reacts with the kinin B1 receptor (B1R), which is expressed by a variety of cell types (including leukocytes and endothelial cells) at sites of inflammation in a regulated manner. Further cleavage of bradykinin and kallidin (e.g. by angiotensin converting enzyme (ACE), aminopeptidase P, dipeptidyl peptidase IV or neprilysin) degrades these peptides. As a result, they are very short-lived in plasma [37]. It is noteworthy that treatment with ACE-inhibitors extends the plasma half-life of bradykinin, which helps to explain that bradykinin-driven angioedema is seen as a side-effect of this therapy [38]. In addition, thrombolysis-associated angioedema is worsened by ACE-inhibitor use [39], and treatable by C1-INH infusion [40], indicating that this follows the same molecular axis.


Bradykinin from plasma or kallidin from tissue?

Accepted Article

There are several questions that surround the disease mechanism of HAE. Firstly, what is the exact species of vasoactive kinins that are responsible for HAE? Is it bradykinin, kallidin, or both? In general, C1-INH is a poor 3

-1 -1

inhibitor of tissue kallikreins [41], with association constants ranging from 0 (undetectable) to 2.2 x 10 M s . For each tissue kallikrein, a much more favorable inhibitor has been identified. This suggests that a lack of C1INH will have minor, if any consequences for the regulation of tissue kallikreins. By comparison, the association 4

-1 -1

constant of C1-INH for PK is 1.7 x 10 M s . Although this association constant for PK is approximately ten-fold higher than it is for tissue kallikreins [42], it is still quite poor (by comparison, the association constant of 3

-1 -1

7

-1 -1

antithrombin for thrombin in absence of heparin is 0.7-1.1 x 10 M s and 3.7 x 10 M s in the presence of heparin [43]). This explains why a partial deficiency in C1-INH can already result in the clinical phenotype of HAE. As mentioned earlier, besides C1-INH, PK can be inhibited by A2M [44]. Complexes between PK and A2M are formed during HAE attacks, as well as in sepsis [45]. This illustrates that A2M tries to take over the role of C1-INH in its absence, but apparently is not completely successful. In contrast to A2M, plasma levels of kallistatin, a liver-expressed inhibitor of tissue kallikreins, are normal in C1-INH-HAE patients [46]. This indicates that, unlike A2M, kallistatin is not a functional replacement for C1-INH. These findings together point towards a critical role for the plasma contact system, rather than tissue kallikreins, for the production of vasoactive kinins in HAE. This is corroborated by reports on elevated plasma levels of bradykinin in blood samples of HAE patients during attacks [47].

BOX 2. Disease modifiers. HAE disease severity cannot be explained by C1-INH function alone [48,49]. Patients with 0% of normal C1-INH function may experience mild, infrequent symptoms while a patient with 40% C1INH function can present with a high attack frequency. Also in our recent studies, we found that the levels of cleaved HK (cHK; reflecting bradykinin production) varied widely within two cohorts C1-INH-HAE patients [50]. In line with these findings, there is no correlation between specific SERPING1 mutations and severity of the phenotype. Attack frequencies reported by affected family members with the same mutation show no correlation [22]. These combined observations suggest disease modifiers that influence the disease severity of C1-INH-HAE. A study carried out in 258 C1-INH-HAE patients from 113 families associated the F12-46C/T


polymorphism with a 7 year delay in disease onset and lower demand for prophylactic treatment [51]. The

Accepted Article

F12-46C/T polymorphism has a lower translation efficacy resulting in decreased FXII plasma levels. The importance of FXII in the bradykinin-forming cascade make it easy to imagine that lower FXII levels are protective in HAE. It should be noted that there are FXII-independent pathways to generate bradykinin [52,53]. Since mice that are deficient in FXII still generate 50% bradykinin under baseline conditions [54], such factors deserve attention as clinical disease modifiers in HAE. On the other end of the spectrum of disease modifiers, enzymes that regulate the breakdown of bradykinin (kininases) are suspected disease modifiers. Aminopeptidase-P activity, which converts bradykinin into an inactive metabolite, correlated inversely to disease severity in C1-INH-HAE [55].

Contactless bradykinin production Several cell types such as platelets and mast cells carry surface triggers for contact activation to contribute to thrombosis and allergic reactions [5,6]. However, there are no clear signs of thrombosis in HAE. Furthermore, by definition HAE is non-histaminergic, which argues against an important role for mast cells. Alternatively, aggregating proteins can generate surfaces for contact activation, in an amino acid sequence-independent manner [56]. This behavior is seen with amyloidogenic peptides such as amyloid β peptide, which causes Alzheimer’s disease and in vivo models confirm that contact activation contributes to this neurodegenerative inflammatory condition [57]. So far, there have been no signs of circulating amyloidogenic proteins in C1-INHHAE, except for one exception: there is evidence that C1-INH itself can generate protein aggregates upon denaturation, which can form a template for contact activation in vitro [58]. Intriguingly, these polymers have been demonstrated in the plasma of a subset of C1-INH-HAE patients, which may contribute to their disease phenotype [59]. However, this probably is not a shared feature for all forms of C1-INH-HAE. In summary, it appears as if an activating surface is missing in this form of contact system-driven human pathology. This suggests that the contact system is activated in a “contactless” manner in C1-INH-HAE. There are several interesting observations in this direction. Since the endogenous FXII-activator had remained unidentified for decades, two groups focused their attention on endothelial cell-derived PK activators: prolylcarboxypeptidase [52] and HSP90 [60]. Later on, we will explore plasmin as candidate trigger for bradykinin production in HAE.


Explanations for localization of swelling attacks

Accepted Article

The clinical phenotype of HAE attacks is difficult to understand on a molecular level. Swelling attacks are normally self-limiting, and restricted to a particular location. This leads to the question why certain tissues swell, while others do not. There are two scenarios: 1) Some tissues are more susceptible to kinins than others are. In this model, the contact system is systemically activated and bradykinin is systemically produced. This concept is supported by evidence for the systemic presence of activation products of the contact system, including cHK [50], bradykinin [61] and its metabolites in plasma samples of patients with C1-INH-HAE. Furthermore, it helps to explain the phenomenon of patients that develop swelling at multiple locations at once [62]. However, if bradykinin is produced systemically, why is swelling only seen locally? B2R is continuously expressed throughout the vasculature. Its activation results in rapid but transient vasodilatation followed by B2R downregulation. When bradykinin is systemically administered to healthy volunteers, hypotension, rather than generalized edema is observed. This argues against a role for B2R in angioedema. Hence, it is proposed that angioedema only arises when the B1R is upregulated on the vascular endothelium after initial stimulation by bradykinin or cytokines. The interaction of 9

Des-arg -bradykinin with B1R would induce prolonged vasodilatation, resulting in swelling. In other words, 9

Des-arg -bradykinin, rather than its precursor bradykinin is held responsible for edema formation. An important argument against this hypothesis is the proven clinical therapeutic efficacy of the selective B2R antagonist icatibant in C1-INH-HAE. This is corroborated in mouse studies where C1-INH-deficient mice are protected against increased vascular permeability when they also do not express B2R [63]. Finally, the absence of hypotension during HAE attacks argues against systemic bradykinin release. 2) Excessive local bradykinin production. In this model, tissues that develop swelling generate their own bradykinin. To this end, the vascular endothelium (luminal side) recruits and activate factors of the contact system in a receptor-mediated fashion, after receiving (unidentified) “danger signals” from underlying tissue. Several endothelial receptors for the contact system factors have been identified in cell biological studies (including uPAR and gc1qR). These receptors facilitate enzymatic crosstalk and bradykinin production. It was recently reported that interaction between a bone-marrow derived pool of FXII and the urokinase receptor results in signaling events that are important to neutrophil trafficking in mouse models of sterile inflammation


and wound healing [64]. Athough the in vivo evidence for receptor-mediated regulation of the contact

Accepted Article

activation in HAE is still scarce; studies like these demonstrate the potential impact of interactions of contact system factors with cell surface receptors. Insufficient control over these mechanisms may result in swelling in HAE.

BOX 3. Diagnostic challenges. There is an impressive delay in the diagnosis of HAE, even when multiple family members are affected. On average, the time to diagnosis is 8.5 years [65]. This is in part attributable to the fact that angioedema is not only seen in relation to C1-INH deficiency. Actually, it is a well-known feature of mastcell driven diseases such as allergic reactions and the skin disease chronic spontaneous urticaria. As a result, patients are often treated for suspected allergies. Only when recurrent attacks do not respond to antihistamine therapy, they are categorized as non-histaminergic angioedema. Hence, HAE patients are often mistaken for allergic patients. Similarly, the prodromal erythema marginatum is easily mistaken for urticaria [66]. In brief, the diagnostic workup of suspected C1-INH-HAE includes determination of C1-inhibitor levels, C1-inhibitor activity and C4 levels. C1-INH function will be decreased (