Depressed activation of the lectin pathway of complement in hereditary angioedema

Authors


L. Varga, 3rd Department of Internal Medicine, Semmelweis University, Kútvölgyi 4, H-1125 Budapest, Hungary.
E-mail: lvarga@kut.sote.hu

Summary

The possibility of simultaneous measurement of the classical pathway (CP), mannan-binding lectin (MBL)–lectin pathway (LP) and alternative pathway (AP) of complement activation by the recently developed Wielisa method allowed us to investigate the in vivo significance of the C1-inhibitor (C1INH) in three complement activation pathways. Functional activity of the CP, LP and AP were measured in the sera of 68 adult patients with hereditary angioedema (HAE) and 64 healthy controls. In addition, the level of C1q, MBL, MBL-associated serine protease-2 (MASP-2), C4-, C3- and C1INH was measured by standard laboratory methods. MBL-2 genotypes were determined by polymerase chain reaction. Besides the complement alterations (low CP and C1INH activity, low C4-, C1INH concentrations), which characterize HAE, the level of MASP-2 was also lower (P = 0·0001) in patients compared with controls. Depressed LP activity was found in patients compared with controls (P = 0·0008) in homozygous carriers of the normal MBL genotype (A/A), but not in carriers of variant genotypes (A/O, O/O). Activity of CP correlated with LP in patients (Spearman's r = 0·64; P < 0·0001), but no significant correlation was found in the control group and no correlation with AP was observed. In contrast, the activity of CP and AP correlated (Spearman's r = 0·47; P < 0·0001) in healthy controls, but there was no significant correlation in the HAE patients. We conclude that the activation of LP might also occur in subjects with C1INH deficiency, which is reflected by the low MASP-2 and C4 levels.

Introduction

The complement system has an important role in innate immune defence. Three different pathways initiate the complement system: the classical pathway (CP), the alternative pathway (AP) and the lectin pathway (LP) [1]. Activation of the CP and LP is initiated by supramolecular complexes, which resemble each other. Each complex has a recognition subunit [C1q in the classical and mannose-binding lectin (MBL) in the LP] which associates with serine proteases [2]. The LP of complement can also be activated via ficolin-2 (l-ficolin) [3] and ficolin-3 (H-ficolin, or Hakata antigen) [4]. As the recognition subunits bind to activator structures, subsequent activation of the serine protease zymogen occurs. In the CP the activated C1r2C1s2 activates the next components of the system (C2, C4), thus allowing the CP C3 convertase (C4b2a) to be set [5]. In the LP only the MBL-associated serine protease-2 (MASP-2) has a clearly defined role in the cleavages of C4 and C2. MASP-1 cleaves C2 but not C4, so it might enhance complement activation triggered by lectin–MASP-2 complexes but cannot initiate activation itself [6,7]. The AP is in a state of spontaneous low-level hydrolysis of native C3 which results in a deposition of its fragments onto cell surfaces, triggering complement activation on pathogens [8]. Initiation of any of the three pathways of complement is associated with the cleavage and deposition of C3b and culminates in activation of the terminal complement pathway, which results in the formation of the terminal C5b−9 complement complex.

Experimental results indicate that CP and LP are controlled by the C1-inhibitor (C1INH) [9–11]. The clinical manifestation of the heterozygous deficiency of C1INH is hereditary angioedema (HAE) [12]. HAE is inherited as an autosomal dominant trait. In 85% of patients the amount of C1INH in the blood is decreased to 10–30% of normal (HAE type I); the other 15% of patients with HAE have antigenically normal or increased amounts of C1INH in their blood, because a non-functional protein is produced by the mutant gene (HAE type II). In the case of C1INH deficiency, unlimited spontaneous activation of the C1 occurs and results in the consumption of C4 and C2, the natural substrates of C1. Components of the classical C3 convertase, i.e. C4 and C2, are activated exceedingly in HAE; however, it does not lead to C3 consumption, as the ratio of C4bp to C4 concentrations is eight times higher in HAE patients than in healthy subjects [13].

Neither the LP activity nor the levels of its components MBL and MASP-2 have been studied in HAE. Deficiency of MBL is common because of genetic polymorphisms. MBL genotypes A/O and O/O are associated with low serum levels of MBL and low activity of the LP pathway [14,15]. For many years, assessing the functional integrity of the complement system has been accomplished in the clinical laboratory by traditional 50% haemolytic complement (CH50) assays. Haemolytic assays based on the haemolysis of antibody-sensitized sheep erythrocytes for the CP, rabbit erythrocytes for the AP and haemolytic assays were also available for the LP [16,17]. Recently a commercial kit (Wielisa, Lund, Sweden) was developed for parallel assessment of the functional analysis of the three pathways. The enzyme-linked immunosorbent assays (ELISA) are based on specific coating [immunoglobulin M (IgM) for CP, mannan for LP and LPS for AP] in combination with a specific buffer system. The assays combine the principles of the haemolytic assay with use of a monoclonal antibody specific for neoantigen (C5b−9 complex) produced as a result of complement activation. The amount of polymerized C5b−9 (final product) is proportional to the functional activity of initial complement component to C9, which is the overall activity of the CP, LP and AP respectively. The results obtained in these complement kits showed good correlation with the results obtained in the haemolytic assays for both the CP and AP [18,19].

The primary goal of the present work was to study the in vivo significance of C1INH in the process of MBL-initiated LP activation by comparing LP activity as well as the serum concentrations of their components (MBL, MASP-2) in HAE patients and age- and gender-matched healthy controls. In addition, activities of the CP and AP as well as several other complement protein levels were also determined in the two groups and their correlations were also assessed separately in the patient and control groups.

Methods

Study subjects

Sixty-eight adult patients diagnosed for HAE [27 males and 41 females, aged 41 (30·5–49 years, median (interquartile range)] were included in the present study. The history of sudden oedematous attacks, family history and laboratory findings of low functional activity and antigen concentration of C1INH and low C4 levels established the tentative diagnosis, which was confirmed by detecting the mutation of the C1INH gene [20]. Blood samples of patients were collected between June 2002 and December 2004 and stored deep-frozen at −70°C until laboratory testing. All samples were obtained during regular control visits of patients; that is, none of them was suffering from an acute angioedematous attack. Sixty-four age- and sex-matched healthy Hungarian volunteers [40 female and 24 male aged 40 (33–49) years] served as HAE-negative controls. All the volunteers were examined and asked about any diseases. Only healthy subjects were enrolled into this study. No volunteers had clinical or laboratory signs suggestive of HAE. The study was approved by the local Ethical Commitee and all subjects gave informed consent.

Complement measurements

We applied an enzyme immunoassay kit (Wieslab) for the quantitative determination of functional classical, MBL-induced LP and AP in the serum samples. To quantify C4-, C3- and C1INH levels, radial immunodiffusion was performed using antibodies from DiaSorin (DiaSorin Inc., Stillwater, MN, USA) and a human calibator serum (Dako Denmark A/S, Glostrup, Denmark) for C3 and C4, as well as an in-house calibrator for C1INH. Serum C1q level was determined by the ELISA method [21]. The concentration of functional C1INH was measured with a C1INH enzyme immunoassay kit (Quidel, San Diego, CA, USA).

Mannan-binding lectin-associated serine protease-2 ELISA

Polyclonal antibody directed against a recombinant CCP1-CCP-2-SP fragment of human MASP-2 was produced in rabbits. ELISA plates (high binding microplates from Greiner Bio-One, Frickenhausen, Germany) were coated with 100 μl 2·5 μg/ml antibodies in 0·1 M NaHCO3, pH 9·6, and incubated overnight at 4°C. Plates were washed three times after each step with 100 μl/well wash buffer [phosphate-buffered saline (PBS), 0·05% Tween-20 from Sigma-Aldrich, St. Louis, MO, USA]. In each case incubation was sustained for 1 h at 37°C. Wells were blocked with 1% bovine serum albumin (BSA) (Fluka & Riedel, Buchs, Switzerland) in PBS. Human serum samples were diluted 1/10 (v/v) in veronal buffer saline containing 0·05% Tween-20, 0·1% gelatin (Reanal Ltd, Budapest, Hungary), 1 M NaCl and 10 mM ethylendiamine tetraacetic acid, pH 7·5. The third layer of our sandwich ELISA was digoxigenin-labelled rabbit IgG anti-human MASP-2 diluted 1 : 500 (v/v) in detection buffer (PBS, 1% BSA, 0·05% Tween-20). One hundred μl horseradish peroxidase (HRP)-conjugated sheep anti-digoxigenin antibody (Fab fragments from Boehringer Mannheim, Mannheim, Germany) diluted 1 : 4000 (v/v) in detection buffer was used as the second antibody. Development was carried out with 2,2′-azino-bis 3-ethylbenzthiazoline-6-sulphonic acid (Sigma-Aldrich) at a concentration of 2·5 mg/ml in 0·1 M citrate/Na2 HPO4 buffer, pH 4·2, in the presence of 0·01% H2O2. Absorbance at 415 nm was recorded after 15–30 min incubation at room temperature. The ELISA was standardized using negative controls and dilutions of highly purified recombinant human MASP-2 CCP1-CCP2-SP fragment.

Mannan-binding lectin ELISA

HB ELISA plates (Greiner Bio-One, Frickenhausen, Germany) were coated with mouse monoclonal anti-MBL antibody (HYB131-01; AntibodyShop, Gentofte, Denmark), diluted 1 : 200 in 0·1 M NaHCO3, pH 9·6, and were incubated for 2 h at 25°C. Plates were washed in PBS containing 0·5% BSA and 0·05 Tween20. After washing, 1 : 20 diluted human serum samples and aliquots of calibrator MBL solution were incubated overnight in microwells at 4°C. Purified human MBL was kindly provided by Steffen Thiel (Denmark). After washing, the plates were incubated with 1 : 2000 diluted biotinylated anti-MBL antibody (HYB131-01; AntibodyShop) for 2 h at room temperature. Plates were washed again and HRP-conjugated streptavidin was added to each test well. Finally, tetramethyl-benzidine was added and the optical density was measured at 450 nm. The concentration of serum MBL was calculated by interpolation on the calibration curve.

Genotyping of MBL-2

Total genomic DNA was extracted from white blood cells using the method of Miller [22]. Determination of the alleles of the MBL2 gene and the regulatory promoter variants were performed by polymerase chain reaction using sequence-specific priming as described [14].

Statistical analysis

Statistical analysis was performed with spss for Windows version 13·0.1 (SPSS Inc., Chicago, IL, USA; available at: http://www.spss.com) and Prism for Windows 4·02 (GraphPad Software, San Diego, CA, USA; available at: http://www.graphpad.com) statistical software products. As many of the variables had non-Gaussian distributions we used non-parametric tests throughout the analysis. We used the Mann–Whitney U-test to compare two independent groups, Fisher's exact test to compare categorical variables and Spearman's rho to calculate correlations. Multiple linear regression analysis was performed with spss version 13·0.1 (SPSS Inc.) software. All statistical analyses were performed two-tailed and P < 0·05 was considered significant. Values presented in the text are medians (interquartile ranges), unless stated otherwise.

Results

Functional activities of the three different complement pathways and other complement parameters in subjects with hereditary C1INH deficiency and healthy controls

Activities of the CP, LP and AP and the level of individual complement proteins measured in the sera of patients were compared with that of controls and the results are shown in Table 1. The activity of CP, levels of MASP-2, C3, C4, antigenic and functional C1INH were found depressed in patients compared with controls. When antigenic C1INH was studied, HAE type II patients were not taken into consideration. The activity of the LP, AP and the levels of MBL and C1q did not differ between the two groups.

Table 1.  Functional activity of the classical pathway (CP), mannan-binding lectin (MBL)–lectin pathway (LP) and alternative pathway (AP) of complement activation measured by the Wielisa method and the level of complement proteins in patients with hereditary angioedema (HAE) and healthy controls*.
 Patients with HAEControlsP
  • *

    Values presented as median (interquartile range).

  • **

    Six patients with HAE type II were excluded form the analysis; n.s., not significant; C1INHa, antigenic level of C1-inhibitor; C1INHf, functional level of C1-inhibition.

No. of subjects6864
LP (%)15·58 (1·35–52·97)30·64 (1·43–111·4)n.s.
CP (%)55·5 (23·15–80·85)93·66 (86·71–107·7)< 0·0001
AP (%)94·46 (82·25–99·79)91·88 (85·96–97·34)n.s.
MBL, mg/l2·31 (0·97–4·01)1·70 (0·43–3·47)n.s.
MASP2, mg/l0·27 (0·12–0·46)0·51 (0·26–1·08)0·0001
C1q, mg/l109·5 (94·5–128·5)103 (87–120)n.s.
C3, g/l0·92 (0·83–1·06)1·14 (0·95–1·5)< 0·0001
C4, g/l0·08 (0·05–0·13)0·22 (0·20–0·26)< 0·0001
C1INHa (%)16·0 (3·6–26·0)**115 (101–132)< 0·0001
C1INHf (%)64·1 (46·5–75)91·2 (86·5–96·1)< 0·0001

Functional activities of the three different complement pathways and other complement parameters in MBL-2 subgroups of subjects with hereditary C1INH deficiency and healthy controls

We categorized the subjects into two subgroups according to their MBL2 genotypes: subjects who were homozygous carriers of the wild-type (A) allele of MBL (MBL A/A) were separated from the heterozygous or homozygous carriers of the variant (O) allele of MBL2 (MBL A/O or O/O). Among the HAE patients and healthy controls, 40 of 68 and 33 of 64 subjects, respectively, carried the A/A genotype; the difference between the two groups was not significant (P = 0·14). In the HAE patients with MBL A/A genotype the total activity of the LP was significantly lower compared with healthy controls with the same genotype (Fig. 1a). In subjects with MBL A/O or O/O genotypes, however, no statistically significant difference was observed in the LP activity. As expected, median CP activity was very low (Fig. 1b) in both groups. When subjects with only A/A genotype were considered the difference between the two groups in the levels of other complement parameters was approximately the same as in the whole group. This was also the case in subjects with the variant MBL allele (data not shown).

Figure 1.

Activity of the lectin pathway in hereditary angioedema (HAE) patients and healthy controls according to the mannan-binding lectin-2 (MBL-2) genotype. Patients homozygous for the wild-type allele of the MBL2 gene (MBL A/A) had a significantly lower lectin pathway activity compared with healthy controls with the same genotype (a). This difference was not observed in those subjects who carried at least one mutant allele of the MBL2 gene (MBL A/O or O/O) and thus had functional deficiency of the lectin pathway of complement (b).

Correlation between the complement parameters in the group of patients and controls

The correlation pattern of patients contrasted strikingly with that of healthy controls. In patients with HAE we found a strong positive correlation between the total CP and LP activity, but no correlation was found between the activity of the CP and AP of complement. Conversely, in healthy subjects the activity of the CP and AP correlated with each other, and no association was found between the CP and LP. No correlation was found between AP and LP in either group. We also determined the correlations between total pathway activities with the other complement parameters. In the patients, CP correlated with C1q, C4, functional and antigenic C1INH (without HAE type II) and LP correlated with MBL, C1q, C4, functional and antigenic C1INH. Taking into consideration the individual complement proteins, C1q correlated with MBL and C4 only in patients, while no association was found between these complement proteins in healthy subjects. In controls we found far fewer associations. Similarly to patients, in the control group the level of MBL correlated strongly with LP and the level of C3 correlated with AP. In contrast to the patients, in healthy subjects CP correlated not only with AP but also with C3. A weak correlation was found between AP and antigenic level of C1INH in control subjects (Table 2).

Table 2.  Correlations between complement parameters: comparisons of patients with controls.
 Patients with HAEControls
  1. Values presented as Spearman's rho (P-value). *Patients with HAE type II were excluded from the analysis. AP: alternative pathway activity; C1INHa, antigenic level of C1 inhibitor; C1INHf, functional level of C1-inhibitor; CP, classical pathway activity; LP, MBL–lectin pathway activity; MBL, mannan-binding lectin; n.s., not significant. HAE, hereditary angioedema.

CP versus LP0·644 (< 0·0001)n.s.
CP versus APn.s.0·474 (< 0·0001)
CP versus C1q0·4898 (< 0·0001)n.s.
CP versus C40·482 (< 0·0001)n.s.
CP versus C3n.s.0·436 (0·0005)
CP versus C1INHa0·500 (< 0·0001)*0·312 (0·015)
CP versus C1INHf0·378 (0·0015)n.s.
LP versus MBL0·622 (< 0·0001)0·834 (P < 0·0001)
LP versus C1q0·547 (< 0·0001)n.s.
LP versus C40·409 (0·0005)n.s.
LP versus C1INHa0·381 (0·0023)*n.s.
LP versus C1INHf0·398 (0·0008)n.s.
AP versus C30·430 (0·0002)0·463 (0·0002)
AP versus C1INHan.s.*0·311 (0·015)
C1q versus MBL0·276 (0·022)n.s.
C1q versus C40·464 (< 0·0001)n.s.

Low LP activity (in the lowest tertile, < 4·86%) was associated with low (in the lowest tertile, < 31·98%) CP activity in HAE patients, even after adjustment to C1q, C4, antigenic and functional C1INH concentration and MBL serum concentration. Patients with low LP also had a more than 10 times likelihood (1·11–98·33, 95% CI) to have low CP activity compared with those with LP in the highest or medium tertile (P = 0·040). Low LP activity was also found to be associated independently with the MBL, C1q and C1INH level (Table 3).

Table 3.  Analysis by multiple logistic regression* of the positive correlation between classical (CP) and lectin complement pathways (CP) in 68 patients with hereditary angioneurotic oedema (HAE).
VariableOdds ratio95% CI of odds ratioP
  • *

    Dependent variable: low LP activity as defined as LP activity in the lowest tertile (< 4·86%). Independent variables: low CP activity as defined as CP activity in the lowest tertile (< 31·98%). C1q, mg/l, C4, g/l, antigenic C1INH: C1INHa,%, functional C1INH: C1INHf,% MBL, mg/l.

CP low/high10·471·11–98·330·040
C1q0·950·91–0·990·034
C40·720·00–2·04 × 1090·977
C1INHa1·000·98–1·0250·704
C1INHf0·930·87–0·9980·044
MBL0·090·02–0·4440·003

Discussion

This paper is the first, to our knowledge, to point out that the activity of the MBL-induced LP is depressed − along with that of the CP − in HAE, which is known as a life-threatening condition caused by the heterozygous deficiency of the C1INH. Considering that MBL deficiency has an essential influence on total LP activity, comparison of patients and controls was undertaken by analysing results after the exclusion of subjects also carrying variant MBL-2 alleles. The relationship with depressed LP activity could not be demonstrated for the whole population, only for the wild-type MBL carriers. Among individuals with the homozygous A/A MBL genotype, patients with HAE had significantly lower LP activity compared with healthy controls.

There are several possible explanations for this novel observation. One potential explanation is that the consumption of complement components consequent to C1INH deficiency, and extremely low C4 levels in particular, also leads to a depressed total MBL-induced lectin activity, similarly to CP, as C4 is a component or both pathways. The question is whether MASP-2 (the one and only MASP according to current knowledge, which has a clearly defined role in C4 cleavage) [23] also contributes to C4 consumption characteristic for HAE patients. The presumed contribution of the activation of the MBL–LP to C4 consumption would be corroborated if significantly lower C4 levels are found in MBL-sufficient rather than in MBL-deficient subjects; however, we could not detect such a difference. According to our results, MBL level is correlated closely with the activity of the MBL-induced LP, both in healthy controls and in C1INH-deficient individuals, whereas MBL and C4 levels are completely unrelated. On one hand, a remarkably close correlation exists between C1q and C4 levels in C1INH-deficient subjects. These findings suggest that activation of the MBL–LP is not a key process in C1INH deficiency, and decreased activity of this pathway in this condition is a consequence of pre-existing C4 depletion.

On the other hand, low MASP-2 levels in HAE patients compared with the controls indicate that LP is also activated in HAE. Activation of the LP may be the consequence of spontaneous activation of zymogen MASP-2 because of C1INH deficiency, and therefore an association between these and MASP-2 might create an alternative means for the latter to contribute to C4 cleavage. Information on MASP-2 is still scarce; its normal level is 0·53 mg/l in healthy individuals, according to a small-scale study [24], which is in complete agreement with the concentration we found in the healthy controls [0·51 (0·26–1·08) mg/l]. MASP-2 levels have not yet been studied in C1INH-deficient patients. MASP-2 deficiency is uncommon: only a single case has been reported to date [25]. Therefore, in our case consumption must have been the cause of reduced MASP-2 levels. As well, MASP-2 consumption should also be accompanied by the consumption of C1INH. The expected level of C1INH in HAE is below 50%, and interaction with MASP-2 might contribute to this typically depressed level. Why do MASP-2 and not C1s/C1r levels decrease in HAE? As shown by comparative data, this can be explained tentatively by the finding that MASP-2 has higher binding affinity to C1INH than C1s [10]. Of the components of the LP, this study focused on only MBL and MASP-2. This is the first report in which the level of MBL in patients with HAE was studied. Compared with controls, MBL levels were not decreased in C1INH-deficient subjects. This result is not surprising, as the levels of the corresponding components of the CP (C1q) are expected to be normal in HAE [26], as also confirmed by our results. MBL level exhibits ethnic variation [27]. A difference of approximately 1–3 mg/l on average in mean MBL serum concentrations has been reported from previous studies [27–29]. Our results were in complete agreement with these data, as the median levels of MBL were 1·7 mg/ml in healthy controls and 2·3 mg/l in subjects with C1INH deficiency. Different MBL promoter alleles exert a profound influence on serum concentrations of MBL [14], therefore we have also compared the proportions of MBL-deficient subjects; we found it to be 34% in C1INH-deficient patients and 48% in controls. These data are in agreement with the 40% prevalence published earlier [30].

In summary, it can be concluded that the MBL–LP is involved in development of the hypocomplementemic state resulting from C1INH deficiency; however, activation through C1 appears to be more important.

At the correlation analysis in groups of patients and controls, we found dramatic differences. As expected in the controls, both CP and AP were limited by the C3 concentration, while neither C1q nor C4 were found to be limiting components. In contrast, in HAE patients C1q and C4 levels limited CP, whereas no correlation was found with the C3 concentration. As for LP, this pathway in the controls was limited exclusively by the MBL levels, while in the patients with HAE, C1q, C4 and C1INH levels also contributed to the determination of the total activity of LP.

Our study failed to confirm the significance of C1INH in the regulation of the AP [31]. Nevertheless, the finding that C3 levels are slightly but significantly lower in HAE patients than in controls is still remarkable, and is in agreement with previous reports on C3 activation in HAE [32,33]. As suggested by our results, analysing the significance of the three activation pathways in comparison with the clinical parameters of HAE would be worthwhile.

Acknowledgements

This work was supported by OTKA project no. K63038. The authors are highly indebted to Mrs Márta Dóczy for excellent technical assistance.

Ancillary